Mary Margaret McCabe

Platonic Conversations
Platonic
Conversations
Mary Margaret McCabe
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
# in this volume Mary Margaret McCabe 2015
The moral rights of the author have been asserted
First Edition published in 2015
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2014951693
ISBN 978–0–19–873288–4
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
For Verity
Preface
The chapters in this volume represent the work of over thirty years and deep and
abiding debts over all of that time. Since, for me, philosophy is best done by
conversation, I am deeply grateful to all those who have been willing to talk to me,
in thinking about arguments and texts and in helping me to see what matters. Some
of those debts are recorded in the notes of each chapter, and I here reiterate my
gratitude to all those mentioned in those places. But I have a more systematic debt to
acknowledge—to all the friends and colleagues and students who have enabled me to
understand whatever I have succeeded in understanding, and for having been
forgiving of the countless times where I have not and do not.
Philosophical conversation is hard, for it demands a kind of openness, thought-
fulness, and courtesy that is difficult to achieve—and I have been hugely fortunate in
finding companions who are both kindly and clever. I do not underestimate how
important the kindliness is—it allows one to try ideas without fear of ridicule or
dismissal and at the right speed—or how rare—it is easily overtaken by the fast
combat of some models of philosophical debate.
In the Philosophy Department at King’s College London, where I have been since
1990, the intellectual temper is wonderful. All this time I have benefited from the
open minds of so many friends there, and have always found people who want to
make thinking better, not to win some abstruse competition. I owe the Department
an immeasurable debt, and more particular debts to the individuals and many dear
friends who have made it up over this long time; it is a matter of huge pride to me
to have been associated with it, and to see it retain its tremendous character and
fierce autonomy through all sorts of vicissitudes: my huge thanks and my constant
admiration.
This has been especially true of the colleagues with whom I have shared our
precious Greek seminar: in chronological order, Tad Brennan, Alan Lacey, Verity
Harte, Peter Adamson, Will Rasmussen, Raphael Woolf, Fiona Leigh, Joachim
Aufderheide, and Shaul Tor; and Richard Sorabji, whose benign influence on ancient
philosophy in London has been of such significance. I would like particularly to
thank all my students, over the years both in King’s and elsewhere, who have made
me see every week how to think, who remind me over and over again how exciting
philosophy can be, and whose enthusiasms, energy, and vision have been a constant
inspiration. I would like to thank those of my friends with whom I do not share an
institution: in all sorts of strange places in the world I have met illumination from just
sitting down with them and talking: I am a very lucky woman.
My special and astonished thanks to Verity Harte and Raphael Woolf for a magical
surprise week in France in July 2014, and to those who were there—many of the
viii PREFACE
colleagues and former students I have mentioned, and Charles Brittain, Amber
Carpenter, Angie Hobbs, Nicole Ooms-Renard, and Ricardo Salles. Peter Momtchiloff
at OUP has always been a source of encouragement and support, and has borne
my chaotic attitude to deadlines; I am extremely grateful. My thanks, too, to Sally
Evans-Darby for a light touch at the copy-editing stage. The British Academy and the
Leverhulme Trust funded two periods of leave for me over the last twenty years; I am
very grateful indeed for their generosity, and for the way in which these institutions
continue to support research that is pursued for its own sake, with no requirement to
show that it has ‘impact’. My wonderful family has tolerated all this philosophy all
this time—I owe them everything, and record my gratitude and deep love here to
Martin, Kate, Mark, Poppy, Daisy, and Tom.
One of my interlocutors stands out for me as someone with whom conversation,
whether conducted standing on the staircase in the Department, or at a foreign
conference, or by email or by phone, is always just brilliant. I have known Verity
Harte since she was seventeen; she has been student, colleague, and friend, and for all
the many years since we first met I owe her a tremendous thank you. This book is for
Verity, with love.
London
2014
Acknowledgements
The chapters printed in this volume were previously published in the following
places. I am very grateful to the editors and Presses in question for their permission
to reprint here.
‘Parmenides’ dilemma’, Phronesis, 1982, 1–12. Reprinted with the kind permission
of Koninklijke Brill NV (published under the name ‘Mary Margaret Mackenzie’).
‘Heraclitus and the art of paradox’, Oxford Studies in Ancient Philosophy, 1988,
1–37. Reprinted with the kind permission of Oxford University Press (published
under the name ‘Mary Margaret Mackenzie’).
‘The moving posset stands still: Heraclitus fr. 125’ (published under the name
‘Mary Margaret Mackenzie’). Copyright # 1987 Johns Hopkins University Press.
This article was first published in the American Journal of Philology, 108 (1987),
542–51. Reprinted with the kind permission of Johns Hopkins University Press.
‘Myth, allegory and argument in Plato’, in The Language of the Cave, eds.
M. Warner and A. Barker, Apeiron, 1993, 47–67. Reprinted with the kind permission
of the editors and the editor of Apeiron (published under the name ‘Mary Margaret
Mackenzie’).
‘Unity in the Parmenides: the unity of the Parmenides’, in Form and Argument in
Late Plato, eds. C. Gill and M. M. McCabe, Oxford University Press, 1996, 5–48.
Reprinted with the kind permission of Oxford University Press.
‘Indifference readings: Plato and the Stoa on Socratic ethics’, in Classics in
Progress: Essays on Ancient Greece and Rome, ed. T. P. Wiseman, British Academy,
2002, 363–98. Reprinted with the kind permission of Oxford University Press.
‘Out of the labyrinth: Plato’s attack on consequentialism’, in Virtue, Norms and
Objectivity, ed. C. Gill, Oxford University Press, 2005, 189–215. Reprinted with the
kind permission of Oxford University Press.
‘Does your Plato bite?’, in Agonistes, eds. J. Dillon and M. Dixsaut, Ashgate, 2006,
107–21. Reprinted with the kind permission of Ashgate Publishing.
‘Is dialectic as dialectic does? The virtue of philosophical conversation’, for The
Virtuous Life in Greek Ethics, ed. B. Reis, Cambridge University Press, 2006, 70–99.
Reprinted with the kind permission of Cambridge University Press.
‘Looking inside Charmides’ cloak’, for Maieusis, ed. D. Scott, Oxford University
Press, 2007, 1–19. Reprinted with the kind permission of Oxford University Press.
‘Perceiving that we see and hear: Aristotle on Plato on judgement and reflection’,
in Perspectives on Perception, eds. M. M. McCabe and M. Textor, Ontos Verlag, 2007,
143–77. Reprinted with the kind permission of De Gruyter Publishing.
x ACKNOWLEDGEMENTS
‘Escaping one’s own notice knowing: Meno’s paradox again’, in Proceedings of the
Aristotelian Society, 2009, 233–56. Reprinted courtesy of the editor of the Aristotelian
Society: # 2009.
‘Some conversations with Plato: Aristotle, Metaphysics Z. 13-16’, for Aristotle and
the Stoics Reading Plato, eds. V. Harte, M. M. McCabe, R. W. Sharples, and
A. Sheppard, Institute of Classical Studies Supplement, 2010, 73–101. Reprinted
with the kind permission of the Institute of Classical Studies and Wiley Publishers.
‘With mirrors or without: self-perception in EE 7.12’, for The Eudemian Ethics
on the Voluntary, Friendship and Luck, ed. F. Leigh, the Keeling Colloquium, 2006,
Brill, 2012, 43–75. Reprinted with the kind permission of the editor and Koninklijke
Brill NV.
‘From the cradle to the cave: what happened to self-knowledge in the Republic?’,
for Self-Knowledge in Antiquity, ed. F. Leigh, Keeling Colloquium, 2010, forthcom-
ing. Printed with the kind permission of the editor.
Contents
1. Platonic Conversations 1
I. On Dialectic and Method

2. Heraclitus and the Art of Paradox 35
3. The Moving Posset Stands Still: Heraclitus Fr. 125 65
4. Parmenides’ Dilemma 73
5. Myth, Allegory, and Argument in Plato 83
6. Is Dialectic as Dialectic Does? 100
7. Does Your Plato Bite? 125
8. Unity in the Parmenides: The Unity of the Parmenides 138
II. On Knowledge and Virtue in Plato

9. Looking Inside Charmides’ Cloak 173
10. Escaping One’s Own Notice Knowing: Meno’s Paradox Again 190
11. From the Cradle to the Cave: What Happened to Self-Knowledge
in the Republic? 208
12. Indifference Readings: Plato and the Stoa on Socratic Ethics 228
13. Out of the Labyrinth: Plato’s Attack on Consequentialism 258
III. On Aristotle’s Conversations with Plato

14. Perceiving that We See and Hear: Aristotle on Plato on
Judgement and Reflection 283
15. Some Conversations with Plato: Aristotle, Metaphysics Z. 13–16 310
16. With Mirrors or Without? Self-Perception in Eudemian Ethics vii. 12 340
Bibliography of Works Cited 367

Index Locorum 379
General Index 393
1
Platonic Conversations
1. Paradox and Conversation

When Heraclitus says—as he may have done—‘you can’t step into the same river
twice’, he poses a paradox.1 We may walk away from him in irritation; but if we
respond at all we might find ourselves thinking, not only about rivers, but also about
logic or metaphysics. We might worry about qualification (do we step into a river by
virtue of the same banks or must the waters stay the same too? And then is it still a
river?) or contradiction (surely we can step into the same river twice? Surely we can’t
step into the same river twice? But surely not both?) or identity (what makes this the
same river?). Paradoxes, thus, have philosophical content—even if finding that
content demands a bit of thought. And they have interrogative force: the river
paradox is not explicitly a question at all, but on reflection it may become one.
Paradox has many of the features of verbal wit, so its formalities matter. After all,
it may simply fail if its verbal form is changed (consider the difference between the
paradox ‘I am lying’ and the merely empirical claim ‘she is lying’). So, as I argue, its
linguistic structure is essential to paradox: contrast the banal ‘the posset separates out
when it is not shaken’ with the interesting ‘the moving posset stands still’ (see
Chapter 3). Like irony and pun, therefore, paradox is somehow a literary form, as
well as a philosophical provocation.
But a different feature of the formality of paradox is often ignored: the way in
which it engages both the speaker and the hearer. For, like pun and irony and joke
(and the interrogative too), paradox trades not only on the formal features of what is
said but also on its conversational aspect2—on the pas de deux that occurs when the
speaker says ‘you can’t step into the same river twice’, and the hearer responds with
‘oh yes, I can’. It gets personal here—as the pronouns indicate—and provocative. But
1
DK22B91 and compare DK22B12 and 49; see Chapter 2 for the different versions of the river
fragment. Translations throughout this volume are my own, except where noted.
2
This is not true of other genres, I think. Tragedy could be solitary—pity and fear may be private
emotions for individual members of the audience, in the theatre and afterwards alone; nothing depends, for
the dramatic force, on a noticeable response by the audience. By contrast, the comic modes—pun and irony
included—often demand the active engagement of the audience at the time—to laugh at the joke, groan at
the pun. These public responses are as much a part of the comedy as the joke that is told. I have argued
(McCabe (2010)) that this is an aspect of comedy that Plato both understood and exploits; and there is
surely continuity between jokes, puns, and paradoxes.
2 PLATONIC CONVERSATIONS
what is the provocation? What happens—to the hearer or to the speaker—when

paradox strikes? And what can we say about the relation between the two interlocu-
tors that may then ensue?
The paradox itself is directly challenging: as we read it, we respond, often by
rejecting it. The first response might be ‘yes, I can step into the same river twice’. But
then if the paradox is repeated (‘no, I cannot step . . .’), and repeated (‘surely I both
can and cannot step . . .’), the respondent starts to think about the response she first
gave, and the second, and their relation to each other (‘I said that I can step into the
same river twice and that I cannot step into the same river twice’). At this point, she
realizes that there is a further problem (‘how can it be reasonable to say that I can step
into the same river twice and that I cannot step into the same river twice?’). Now her
earlier responses are themselves part of the content of the problem, and her view of
the puzzle becomes more detached, since they are the objects of later reflective
scrutiny.3 So both the first-order response (‘I can step into the same river’) and the
reflection on it (‘I said . . .’, ‘How can it be reasonable to say . . . ?’) are part of the
process of the paradox itself. The reflection is thus higher order; and this constitutes
its distance, its detachment from the first-order assertions and denials. Paradox thus
generates not only a response, but also critical reflection.
It is a marked feature of early Greek philosophy that it exploits paradox to generate
both direct philosophical principles—such as conditions on identity—and second-
order reflection on how the discursive business of philosophy gets done (or how it
might fail). Consider the resonance of another Heraclitean paradox: ‘going to the
boundaries of soul, you would never find them, although you travel every road; for it
has such a deep account’.4 This is both a gnomic remark about the complexities of the
human soul and a reflection on the remark itself.5 Indeed, paradox (like joke) often
works by self-reference, and thence by generating both first-order responses (about
the content of the paradox) and second-order ones (about what it is to utter, or to
hear, or to endorse such a tricksy remark).6
Some of this turns on what is often called the ‘dialectical context’. That context
interests me for many reasons. Dialectic exposes the logic of disagreement as well as
the content of the disagreements themselves. It considers what it is to disagree—
sometimes merely to deny what another asserts, sometimes positively to occupy a
3
This sequence could be represented thus: ‘I can step into the same river twice’; ‘I cannot step into the
same river twice’; ‘I said that {I can step into the same river twice} and that {I cannot step into the same
river twice}’; ‘How can it be reasonable [to say that {I can step into the same river twice} and that {I cannot
step into the same river twice}]?’, where the brackets show the embedding of different responses in a
reflective context.
4
DK22B45.
5
The word I translate as ‘account’ (logos) is also the word for ‘word’, for ‘statement’ or ‘speech’, or even
for ‘remark’.
6
The liar paradox is an obvious example (see the collection of paradoxes in Sainsbury (1988)). Or recall
the joke: ‘An Englishman, a Scotsman, and a penguin walk into a bar. The barman turns around and says,
“what is this, some kind of joke?”’.
PLATONIC CONVERSATIONS 3
position in philosophy. And it brings out the collaborative aspects of talking about
disagreement itself. All of these are brought out in the conversational aspect of
paradox—in the way in which paradox provokes a response from the hearer, and
by reiteration may then provoke another or a second-order turn. This kind of
argumentative movement from assertion to denial and back, or from confidence to
doubt, or from paradox to its resolution, generates reflective focus on the perspectives
of the different parties, both one’s own and the other’s. What is more, the observer,
looking in at the exchange from the outside, is likewise reflectively engaged, caught
up in, and detached from the movement of the exchange. This movement thus
engages with the content of the argument. But it also considers—as I argue in
Chapters 6 and 11—both the psychological structure and ethical stance of the
participants and even of the observers (ourselves). As a consequence, the label
‘dialectical context’ may fail to convey what is needed. To allow for the rich ethical
and psychological dimension of these philosophical pas de deux, I follow Plato’s lead
in thinking of these as conversations (in Chapter 6, I wonder how literally the idea of
dialectic as conversation should be taken, and how Plato develops what seems to be a
technical terminology from the ordinary business of discussion).
2. Three Kinds of Conversation

In fact, there are three different types of conversation under discussion in what
follows.
The first is the conversation of paradox—such as the paradoxes of Heraclitus or
the paradoxical structure of Parmenides’ poem. Here the conversation is implicit: the
provocation of paradox assumes the existence of a respondent, rather than portray-
ing one; or the respondent, as in Parmenides’ poem, is addressed but does not speak
(this device sharpens the paradox in Parmenides’ poem). Such conversation is
described in Chapters 2–4, and it bears comparison to the oddities of Plato’s
Parmenides, discussed in Chapter 8—heavily influenced, as that dialogue itself
makes clear, by the paradox-mongers from Elea.
The second is the more complex and baroque conversation of the Platonic
dialogues, where different characters talk, or fail to talk,7 to each other. The Platonic
material is variously discussed in Chapters 4–13. I take the Platonic dialogues to offer
philosophical conversation of the richest and most sophisticated kind; but I take that
richness to be a legacy from some of Plato’s predecessors, carrying its interest
forward to his successors.
The third is the conversation that occurs when later readers of Plato engage with
his ideas by direct allusion to individual dialogues. These allusions, I argue in
Chapters 12 and 14–16, differ from the commonplace critical practice, in the same
7
I discuss the failures, the ‘missing persons’ of the dialogues, in more detail in McCabe (2000).
authors, of reporting directly and by name something that some other philosopher
said, and agreeing or disagreeing with it.8 They are, rather, readings of individual
Platonic dialogues, where a complex set of ideas from some dialogue are brought into
the discussion of some particular philosophical issue by allusion (discussions such as
the status of universals (Chapter 15) or the nature of friendship (Chapter 16)). This
intertextuality, I argue, provides reflective detachment by offering a self-conscious
view of the arguments in question from a position outside the debate.
These conversations are thus represented in multifarious ways—in the direct
engagement of paradox; in the dialogues of Plato; in the Platonic conversations of
Aristotle or the Stoics. And in each case they provoke, not only direct responses, but
also reflective distance—in different ways, as I argue below. Indeed, that reflective
distance is itself the object of scrutiny (as I suggest in Chapters 5, 7, and 10). What is
it to think about a philosophical problem? What is it to make progress in reflection?
Why does it matter that we should reflect, rather than just having the answer to some
puzzling question to hand? Why should puzzlement be of any philosophical signifi-
cance at all? And what difference does it make whether we reflect in solitude or in
conversational engagement with others?
All of these questions, and others, arise as a consequence of the structures and
forms of these philosophical texts. When I reflect on a first-order conversation I can
also reflect on my reflection, and so on up the orders, generating quite general
questions about philosophical thinking. In direct paradox, such as Heraclitus’ river,
this is implicit in the nature of paradox. But Plato picks up the same feature in his
presentation of conversation. Consider two notable cases where dialogues are intro-
duced by a complex array of telling and retelling: the ornate openings of both the
Symposium and the Parmenides—what I call ‘Chinese Whispers’ in Chapter 8. This
feature of the narrative highlights the way in which conversations may be thus
embedded in other talk, where that framing talk itself may be conversational,
engaging with other interlocutors or even the reader. Reflective distance, that is to
say, may not be second order, but third or higher order; and conversation likewise.
Reflection can be a complex business.
3. Philosophy and Literature

Each of these modes and orders of conversation depends on a fine linguistic grain: on
the play and resonance of the words and their contexts. The ancient interest in verbal
tricks—such as the polished versions of oracular sayings whose ambiguity demands
interpretation9—forms the background of this mode of thought; and it is encouraged,
8
Compare and contrast Aristotle’s extended critique of the Platonic theory of forms in the peri ideôn;
see Fine (1993).
9
‘Trust in your wooden walls’ famously inspired a full public debate in Athens, Herodotus 7. 140–2.
Fontenrose (1978) argues that these ‘published’ versions differ from the more banal recorded examples of
to some degree, by the emergence of ancient culture from the oral to the literate
tradition.10
As the written word becomes an increasingly powerful mode of creating and
transmitting culture, the memorizing traditions of oral culture may change or even
perish. But at the same time, the literate tradition is itself affected by the acute
techniques of oral memory. So, at this period of Greek thought, the fine-grainedness
of literary composition will be easily noticed and remembered because the faculty of
memory is still powerfully supported and maintained.11 And moreover, in a society
where the oral tradition enhances memory and where the number of extant texts is
few, allusion in one text to another will be readily picked up in its wider context. The
emergence from orality to literacy, that is to say, promotes the importance in written
works of both literary detail and intertextuality. This, I argue in Chapters 14–16,
should make a difference to how we read Aristotle;12 it is also of fundamental
importance in thinking about how to read Plato.
If we turn from the verbal wit of Heraclitus and the metaphysical puzzles of
Parmenides’ poem (which, as I argue in Chapter 4, also depend heavily for their
philosophical point on how they are expressed) to the dialogues of Plato, it is hardly
surprising that we find the same kind of linguistic density and brilliance. But it is
sometimes supposed that the literary features of the Platonic dialogues are somehow
detachable from their philosophical content: so that either the literary finery of the
dialogues can be studied without considering their philosophical significance, or the
argumentative content can be abstracted out from how the content is expressed and
provide us with the philosophy therein.13
Leaving aside for the moment what counts as an argument,14 consider the inter-
pretative strategy that advises thus separating off the ‘literary’ from the ‘philosoph-
ical’. It is, after all, beset by two different kinds of anachronism.
Suppose we were to come to the dialogues fresh, for the first time, and in search of
their philosophical significance. The first anachronism is to suppose that we shall
see—from here and in advance of reading the dialogues themselves—two distinct
types of material: what Plato supposes to be the non-philosophical aspects of the
actual oracular sayings: the ‘published’ versions have the linguistic character of paradox and puzzle. But see
Parker (2000).
10
This is, of course, a disputed issue; but for a strong view of the importance of the transition between
oral and literate culture, see Havelock (1978); Ong (1982); and Goody (1987).
11
We might contrast the problems of short attention span and short-term recall in the context of
modern society: that we can devolve to machines the responsibility for remembering huge quantities of
information itself contributes to our needing to supplement memory in this way.
12
And see also McCabe (1994b) on how this fits Aristotle’s ‘endoxic’ method. See a recent careful
treatment of this issue by Frede (2012).
13
Some interpret Plato by leaving out the framing context of the arguments (e.g. Bostock (1986)), while
others eschew the arguments to concentrate on the literary style (e.g. Rutherford (1995)). While all of these
approaches offer illumination, my contention is that we need, somehow, to look at each dialogue as a
whole. Burnyeat’s brilliant (1990) is an exemplar of a view of a dialogue—the Theaetetus—as a whole.
14
Compare Barnes’ dismal assessment of Platonic arguments in the introduction to Barnes (1995).
writing (the ‘literary’ form of the dialogues) and what he supposes to be doing the
philosophical business (such as abstractable first-order arguments). This is to assume
that we already know, perhaps from the perspective of modern paradigms of
philosophy, what Plato counts as philosophical method—for example, that it is
abstracting arguments. But it is parochial to suppose that his ways of philosophy,
developing fresh at philosophy’s beginning, should coincide with ours, even though
we use the techniques of the modern tradition to help us understand what Plato says.
Conversely, it is a hopeless task to think that we can come to his kind of philoso-
phizing innocent of our own. Instead we have to go crabwise—piecing together with
as little prejudice as possible what he may be saying about the philosophical enter-
prise in which he is engaged. So we should start, as far as we can, agnostically about
what he thinks philosophizing is, and agnostically, therefore, about which bits of the
dialogue should qualify for philosophical scrutiny. In advance, nothing is ruled out.15
The second anachronism is to suppose that, even were we to know which bits of a
dialogue do philosophy, the philosophy they do is the instilling of beliefs. Do the
dialogues ask ‘what is true and why?’, and do they ‘say’ the answers? Or should we
rather think that more is going on here—that the dialogues seek also to engage our
minds, or to educate our souls, or to change our way of life? It may be that the
theoretical attitudes of contemporary philosophy would expect the purpose of
philosophical argument to be to change our beliefs (whether about the conclusion
of the argument or about what argument is). But Plato may begin with a broader
brief, in asking Socrates’ question ‘how best to live?’. And then what it is for the
dialogues to ‘say’ is a complex matter, not the simple business of professing doctrine.
It is often supposed that we know in advance (often by appeal to Plato’s later
interpreters, Aristotle and the neo-Platonists in particular) what he would have us
come to believe, as a result of reading the dialogues.16 That supposition takes the
dialogues to be, somehow, vehicles for conclusions, and thence for doctrines, and the
task of the interpreter to find them there. So it makes two assumptions. The first is
that because the later interpreters are somehow closer in time to Plato, they must be
closer in spirit (even though they may disagree both about whether Plato is doctrinal
and what those doctrines may be17). The second is that the later interpreters are in
15
This includes the use of myth and allegory; see Chapter 5.
16
Some look for comfort to the Seventh Letter, or to later Platonism, or to esoteric doctrines (e.g.
Szlezak (1999)). I remain sceptical that we should ever burden Plato with ‘doctrine’, and my arguments
about the form of Aristotelian writing suggest that Plato’s great pupil was highly sensitive to differences of
view between dialogues even though, true to his own method, he sometimes treats major Platonic theses as
fixed points of dispute. As I argue below, none of this supposes that Plato is constantly sceptical; but it does
suppose that the way the dialogues work is by direct philosophical engagement, rather than by allusion to
some archive of doctrinal claims. I suppose there is a separate issue here whether whatever we may piece
together of the ‘unwritten doctrines’, or of the Platonic oral tradition of philosophy, will act as a control for
what counts and what does not as the philosophical content of the dialogues since, after all, whatever those
unwritten doctrines might have been, we do not have them. If they are described as dogmata: does this
imply that they are the code-busters for what the dialogues say? In what follows, I assume not.
17
Compare Sedley (1996) on how differently later ancient interpreters may take a single dialogue.
fact engaged in the historical representation of what Plato wanted us to find in the
dialogues, rather than themselves engaged in dialectical inquiry with what they take
him to say somewhere and for some engaging reasons. This anachronism, like the
first, might underpin some distinction between philosophical content and literary
form; but it, too, begs the question of what reading the dialogues themselves may
deliver.
So, I suggest, the principles of reading should emerge from the Platonic text itself,
and the sense and significance of what Plato says should be read off from his own
emphasis and sequence of thought. Of course, it is impossible to un-think our
perspective altogether or, were we to do so, to inhabit someone else’s. But the attempt
itself may help us better to understand what is happening in the distant world of
Plato’s thought (and that better understanding may itself constitute our doing
philosophy on our own behalf). If the dialogues of Plato are heirs to the verbal
tradition of Heraclitus, it is a good interpretative maxim that Plato writes no word in
vain. Words, descriptions, characterization, arguments, progress, regress and
impasse, and overall structure may all be a part of whatever is the enterprise of the
dialogue. We are not—so I suppose—at liberty to decide in advance which parts of
any dialogue are significant for their philosophical point, and which are mere literary
artifice. Indeed, the contrast between ‘literary’ and ‘philosophical’, as it might be
applied to Plato’s dialogues, is thoroughly tendentious.
On the contrary, giving due consideration to all of the features of any dialogue may
reveal depths to its thought that are invisible if we come to it supposing, for example,
that it is the vehicle for some antecedent doctrine, or some set of ideas that we have
already labelled ‘Platonism’. This is not to say that there are no such deep theories
within the dialogues; but it is to say that we should not take for granted that any
dialogue somehow assumes them, or alludes to them, or relies on them, or endorses
them, unless we are given proper evidence by the dialogue itself to suppose that this is
so.18
4. Reading the Platonic Dialogues

The thought that Plato writes nothing in vain may go together with another: that in
writing the dialogues he engages in some kind of conversation with his readers,
because it is the reader who receives this piece of writing as a whole.19 To see this
point, think about the conversations represented in the dialogues first.
18
In Chapters 10 and 12, for example, I suggest that there are intertextual allusions that do in fact recall
conclusions from elsewhere, but that they undercut, rather than endorse, what was said in the other text.
19
Of course the brilliance of the dialogues may sometimes obscure this to us; the suspension of disbelief,
for example, when we read of Socrates’ death in the Phaedo is a consequence of its pathetic overtone (it is
not, I think, tragic, at least according to Socrates: Socrates’ choice to die rather than give up philosophizing
seems not to mar his life, even if it does end it).
The dialogues are constructed in various different ways to present discussions

between the protagonists—sometimes reported in direct speech;20 sometimes in
narration;21 sometimes in a complex web of reported narratives;22 and sometimes
even discussions imagined by the protagonists between other people who are not
present.23 But in multifarious ways, all of the dialogues represent conversation
somehow and somewhere.
Indeed, it is a consequence of this conversational feature that the dialogues often
represent discussions that we might find easily recognizable as arguments. For in the
short interrogative exchanges between Socrates or the Eleatic Stranger and their
various interlocutors, we find discussions that are connected in sequence (the
interlocutor asserts something, for example, and Socrates asks him to consider why
he did so24). In the course of the conversation, the interlocutors consider series of
claims, often inferentially connected,25 designed to progress from one claim to
another, and defended by reasons at each stage. Indeed, the very sparse aspects of
the dialogue form allow the reader to see, when the conversation is direct, just how
these discussions are structured. They are indeed arguments—not in the adversative
sense of disagreements, but in a weak sense of sequences of thought, inferentially
connected, designed to advance from one claim to another and putatively supported
by reasons. In one sense, then, Plato trades on the vernacular kind of dialogue
(portrayed, of course, in Attic tragedy or comedy26) to develop this loose conception
of what would count as an argument, as an informal sequence of thought.
In what follows, however, I consider several occasions where the form of the
dialogue intrudes, in one way or another, on the discussion or the argument that is
represented there. For example, the scene-setting of the Charmides provides us with
an account of the puzzling arguments late on in the dialogue (Chapter 9); and the
20
For example, Laches, Gorgias, Philebus.
21
Notably Republic, and compare Socrates’ extended reporting of the Protagoras, in which he com-
plains that he dislikes long speeches because he is forgetful, 334c–d.
22
Notably the openings of both the Symposium and the Parmenides. In neither case does the closing of
the dialogue quite replicate the complexity of the start—most strikingly in the Parmenides, the narrative
form of the first ten pages shifts to direct dialogue and culminates in a complex contradiction without
returning to the frame dialogue. See also the delayed reportage of the Theaetetus and the complicated
timing of the frame of the Euthydemus.
23
Sometimes these people who are not present are in fact ‘missing’—people whose theories in fact
preclude their joining a conversation at all: e.g. Protagoras who cannot sustain argument in the Theaetetus,
or Parmenides at Sophist 244b–d; see McCabe (2000).
24
The ‘why?’ question may be complex and in different orders. Compare the sequence of argument in
the Euthyphro. Socrates wants to know why actions are pious (what it is for actions to be pious: 5d), and
complains when the question is not answered in ways that are fully universalizable (6e); he wants to know
about the direction of the answering ‘because’ (10d ff.), and so challenges whether the answer gives the
explanans or the explanandum; and he inspects Euthyphro’s claim to know what piety is, and hence his
ability to answer the why question in the first place (5a–b with 15a–e).
25
See, for example, the complex arguments between Socrates and Polus at Gorgias 466–79.
26
Tragic short question and answer (stichomuthia) can sound hopelessly implausible to the modern
vernacular ear (compare, for example, Euripides, Medea 663–88), as A. E. Housman brilliantly parodied in
his (1901).
irruption of the frame dialogue into the middle of the Euthydemus demands that we
compare and contrast the different conclusions of the two Socratic episodes of that
baroque work (Chapter 13). On both these occasions, as on many others,27 the
framing conversation reflects on the conversation that is framed in just the same
way as reflective distance frames our response to paradox.28 What is more, often the
discussions in the frame are themselves sequential and responsive to reasons: they
too are arguments in this loose sense. So even if we were to think about the doing of
philosophy as a matter of abstractable arguments, those arguments should not be
restricted to the first-order content—these framing arguments of the dialogue are
second order or higher, just as they frame lower-order arguments within.
The character of the framing discussion is reflective, focusing not only on the
content of the first-order arguments, but also on the conditions for argument itself.29
It is a regular feature, for example, of the framing strategies of the dialogues that they
contain reflection on the methodology and epistemology of the conversations them-
selves (in Chapter 6 I discuss how we might think of this procedure as offering the
conditions for dialectic; and what dialectic, so understood, would be).30 For example,
Socrates’ direct discussion with Euthyphro in the eponymous dialogue is about the
right account of piety; but the framing conversation is about what it is to give an
account at all. In this respect, the conversations of the dialogues bear a strong
resemblance to the conversations I argue to be implicit in paradox, where the
reflective focus is on, for example, the logic of contradiction or the nature of
assertion. This feature of the composition of the dialogues supposes that at least
the cognitive conditions for philosophical discourse are ordered and reflective. But
once those cognitive conditions themselves come under reflection, the remit of the
frame, as I argue in Chapters 9–11, becomes very wide indeed, to include not merely
questions of logic or epistemology, but also the ethical and moral psychological
features of how we think. After all, the dialogues represent people who come to
27
Compare, for example, the puzzle of Phaedo 102a–b, where the frame breaks in, at one of the most
vexed passages of the dialogue, to announce how clear the preceding account has been.
28
The Euthyphro again: Socrates asks Euthyphro to tell him what piety is. Euthyphro responds with the
claim that piety is doing what he is doing now. Socrates objects, not to whether what he is doing is pious but
to whether this counts as what piety is; repeatedly in this dialogue Socrates insists on considering the
higher-order conditions of what it is to explain or define at the same time as he considers the substantive
question of what piety in fact is. Thus the dialogue operates both at the specific and the generic level—or, as
I suggest, at first and second order, where the reflections on what it is to explain come with the discussions
of piety as their first-order content. I discuss the nature of framing in McCabe (2014a) in the context of
framing works of art.
29
This is especially noticeable in the Euthydemus, whose frame discussion is about argument and its
principles; compare e.g. the sophists’ complaint that Socrates is tediously interested in consistency over
time, when they characterize him as an old stick-in-the-mud, a Kronos (287b), or Socrates’ worries about
regress and circularity in his own argument (291a–b), all of which appear in the framing material.
30
Compare Socrates’ meditations on his own ignorance, e.g. at Apology 21b ff., or Charmides 165b–c,
166c–e.
reflection in the midst of their ordinary lives. The narrative features of the dialogue
are not beside their philosophical point.
Suppose, then, that we take all of a dialogue to matter to its philosophical point,
including the reflective features of the frame in their narrative context. Suppose,
further, that those reflective features are generalizable—that is, whenever we discuss
some item of philosophical significance we should also take into account how we
think about it. So thinking about argument is a central part of argument—and
thinking about thinking about argument likewise. Plato’s interlocutors think about
the argument they are making;31 those who interrupt from the frame think about
how the interlocutors think about the arguments;32 the readers think about inter-
locutors and interrupters alike.33
But the conversational context takes this personally, too: it involves a kind of self-
consciousness, a comparison between what the interlocutors say and what the
interrupters concede, or between interlocutors and reader.34 This is especially acute
when the dialogues provoke aporia—impasse or bewilderment in interlocutor or
narrator or reader. For here the effect of paradox is felt again: the psychological
features of puzzlement, and the way that it provokes reflection.35
5. Reflection, Thought, and Knowledge

This should not be taken lightly. If it matters to how we think that we think about our
thinking, then these reflective attitudes may be central to philosophical method, and
thence perhaps also both to thought, on Plato’s view, and to knowledge. Perhaps
reflection requires that we are attentive to what we are doing when we tackle
argument; or perhaps, more strongly, this reflective dimension is conditional of
good thought, or reasoning. And that point might go further still—if we think in
order to understand, then it may be a condition on understanding that we recognize,
or understand, both whether we do indeed understand, and what it is to do so. This
condition of reflection, that is, may not merely require that we are aware that we are
31
Compare, for example, Critias’ willingness to dispense with all but one of his premises, Charmides
164c–e.
32
See, for example, the interruption at Euthydemus 290e ff., which considers which of the interlocutors
could have said what has just been asserted.
33
Compare, for example, the allusion to the slave-boy experiment of the Meno at Phaedo 72e ff. which
introduces a discussion of remembering, echoed by the reader’s own recollection of the Meno passage; this
affords the reader direct and self-conscious experience of the phenomenon under discussion.
34
This is sometimes a matter of our noticing some kind of bad fit. Consider, for example, the closing
episode of the Euthydemus, 304c ff., where, in the frame, Crito reports his meeting with an anonymous
critic of Socrates the day before; and it is the reader who notices and wonders about the connections and
the disconnections with Isocrates (who is certainly not mentioned by name).
35
Compare the close of the Parmenides, where a comprehensive contradiction is met by Aristotle’s
reply ‘very true’ (166c). In Chapter 4, I argue that the personal nature of (the historical) Parmenides’
dilemma is essential to its formulation; in Chapter 8, I show how paradox-mongering may work in the
eponymous dialogue.
thinking; instead, we think about thinking, where the conditions on thinking are the
same at each order. So it may be consonant with this feature of the dialogues that
Plato’s epistemology is resolutely internalist: knowing cannot be fully explained by
pointing to the external conditions of what is known. This, on some accounts of
‘Platonism’, may look surprising. The dialogue form, however, promotes the view
that thinking that we think, reflecting that we reflect, and thence knowing that we
know is conditional on knowledge at the first order. It may be no drawback that this
view has the seeds of paradox within it.
Recall the oracle who told Chairephon that no-one was wiser than Socrates.36
Socrates tried to figure out the oracle’s point37 by having conversations38 with
various pretenders to knowledge—politicians, poets, and craftsmen. He concluded
that in each case he was indeed wiser than they were: than the politicians because
they were wrong about their own wisdom; than the poets because they merely tell the
truth from divine inspiration; than the craftsmen because although they do know
some things, they are misled into thinking that as a consequence they know other
things they do not know. Socrates, by contrast, recognizes that he himself is not wise;
and this recognition in itself makes him wiser than the rest. For this argument to
show that Socrates is indeed wiser, to know your own cognitive condition must be
either a superlative kind of knowledge or a condition on any knowledge at all (or
both). It would beg the question against his opponents for Socrates merely to declare
his own wisdom the best, since that is itself the issue. So it must be that his
investigation of the oracle concludes that he is wiser because knowing that we
know—or even, in this limiting case, knowing that we do not know—is conditional
on any knowledge at all—hence the failure of the craftsmen, despite the fact that they
have knowledge of their crafts.39
But Socrates’ knowledge of his own ignorance is second order just as the reflective
features of the dialogues are: the second-order feature of the dialogues, on such a
view, would be central to the epistemology therein. Thus in Chapter 10 I argue that
the development and critique of Meno’s paradox in the Euthydemus brings out the
condition on knowledge that we cannot escape our own notice knowing. This, in
turn, may show up in the complex structure and rich language of the dialogues. They
depict various interlocutors, from all sorts of backgrounds,40 trying or failing or
occasionally succeeding in reflecting on their own cognitive states. They may also
36
Apology 21a ff.
37
The verb used is elenchein (21c1). This does not mean (as it is sometimes translated) ‘refute’, but
rather ‘investigate’. So it is not definitional of Socrates’ method of the elenchus that it is refutational; rather,
it investigates the views of the interlocutor.
38
Conversations held one by one with each of the members of the group he examines: Apology 21c5;
compare 22b4 where he asks questions of the poets, and 22e where he asks questions of himself, and
answers.
39
Apology 22d3. See McCabe (1988a).
40
Including a slave (Meno 82 ff.), a woman (Diotima in the role of questioner at Symposium 201d ff.)
and several interlocutors chosen for their pliability (e.g. Aristotle at Parmenides 137b).
show us these conversations set in a reflective context, where other interlocutors

watch, hear, consider, and think about the first-order conversation. And they may
sometimes show us ourselves, watching the conversations from the outside, and
thinking about them: so that the reflective condition is built into the way the
dialogues are set up.
This may tell us something about how the drama of the dialogues might have a
philosophical impact—an effect on how we should think about knowledge—derived
from the whole dialogue, frame and all. But it does not tell us why conversation may
matter so much: if it tells us that reflection matters to knowledge, it does not tell us
why conversation might matter to reflection. Why does conversation play such a vital
role in Plato’s account of philosophy?
6. Questions and Answers

The conversations that Plato depicts are rather particular in form: for they are often
conducted by question and answer.41 Recall Socrates’ encounters with the pretenders
to knowledge in the Apology: he asks them about what they know, challenges their
claims, wonders whether they can generalize from one claim to another.
But why is this particular conversational structure taken to be so important? A first
answer to this question, of course, reiterates that the asking and answering of
questions is itself reflective. For in thinking about the answer to a question, the
respondent needs to think—and is repeatedly seen to do so in the dialogues42—about
whether the question has indeed been answered. He43 needs to give the answer, for
sure; but he is also provoked to wonder just whether the answer is not just true, but
also a proper answer to what is asked. So, the answer is itself within the reflective
scope of the interlocutor’s response; once again, the second-order dimension is
prominent. We might think about this in terms of stance—in taking part in a
conversation, the interlocutors both occupy their own stance and take into account
the stance of the other—and when they do the latter, they include the other’s stance
somehow in the content of what they think. And we might think about it in terms of
41
Examples abound: from the richly articulated (e.g. Gorgias 488b ff.), to the thin (e.g. Parmenides 137
ff.), to the vanishing (e.g. Protagoras at Theaetetus 171c–d).
42
Notice the moments when the interlocutor starts to doubt, e.g. Callicles grumpily at Gorgias 497a ff.,
or to change their minds, e.g. Protarchus at Philebus 19c ff., or even firmly to resist Socrates’ claims, e.g.
Protarchus again at Philebus 36c.
43
Apart from Diotima (and Aspasia reported in the Menexenus), the interlocutors are all male. In fact,
however, the conversational mode of philosophy seems to me—and, as I argue for example in Chapter 16,
seemed to the ancients—to have rich moral content: it demands all sorts of virtue and all sorts of good
manners. It can be properly egalitarian; it can give anyone a voice and anyone can listen so as to hear.
However we may deplore, from a contemporary standpoint, the various kinds of elitism and exclusivity of
the Greeks, they do have something to tell us about how things might be done differently than is sometimes
the case now. In what follows, consonant with my view that the dialogues are meant to engage with readers
of any era, I use both gendered pronouns for indefinite discussion and examples.
intentional content: so when the question is answered, it is answered as an answer to

that question.44
Second, and connectedly, the conversations that Socrates has with his interlocutors
focus especially on what we might describe as accountability: in each case, the asking
and answering of questions involves the interlocutor’s being asked to give an account
of whatever it is he seeks to maintain, or an account of his entitlement to maintain
it.45 The ways of conversation, that is to say, focus not on truth so much as on
explanation: the dominant questions are ‘what is it to be thus and so?’, ‘why is
something thus and so?’, ‘how can you explain to me why things are thus and so?’.
Giving an account in these ways is taken to be essential to any claims to know.
Conversely, Socrates’ interlocutors’ failures to answer adequately support his own
claim that he is wiser only in that knows only that he knows nothing.46 For that claim
is based—as we may see from the careful sequence of thought in the Euthyphro—on
an account of what it would be to know; and that account is, once again, reflective in a
conversational mode. The second-order claim, to know that he knows nothing, is
both an instance of the explanatory function of knowledge and a condition on it. If
we seek to explain something, that explanation succeeds only if we see it to do so.47
Third, dialogues like the Euthyphro may tell us more about the nature of conver-
sation: that in focusing on accountability, it is sequential. The answer to the first
question generates a second, and the second a third. This series of questions
constrains the answers, in that they should be salient to the questions, and the next
question to the last answer.48 So this repeated demand for salience requires that the
series as a whole should be sequentially linked. Once again, this is connected to
whatever knowledge claims may be being made here—that they are not only
explanatory, but also at least connected, at most systematic. Socrates’ insistence on
the short exchange of question and answer,49 moreover, is no mere dramatic device.
Instead, it shows how these claims should be systematized: as the connected sequence
44
Compare Socrates’ complaints that his interlocutor did not answer the question he asked: e.g.
Euthyphro 6c–d; Meno 72a–b.
45
The demand for an account sometimes figures as a question such as ‘what is integrity?’ (this seems to
me to be one of the most helpful ways of rendering sôphrosunê), Charmides 159a; but it is often
accompanied by some other epistemic point, as here, that Charmides is well placed to answer this question
because he has the virtue in question (see Chapter 9). Compare Socrates’ asking Euthyphro to teach him,
since Euthyphro has knowledge about piety.
46
The exact formula is rather cagier than this, and at times there are things he claims to be certain of;
but this complexity in itself invites the reader to wonder about what it would be for someone to claim to
know, and be right.
47
My thanks to Raphael Woolf for reminding me that Aristotle takes this point, An.Post. 1.2.
48
There are times when Socrates goes off on a new tack, but these are marked: e.g. Philebus 14c, Meno
79e, which provoke a great deal of second-order discussion about whether any inquiry can be made at all.
49
The very spuriousness of his claim to be forgetful at Protagoras 334 (see n. 21), which supports his
interest in short questions and answers, calls attention to the form.
of small or simple propositions, exposing the explanatory links and avoiding the
complex theses that lead to the difficulties created by sophistry.50
Conversations by question and answer, then, build into the dialogues both the
reflective dimension and a structure (of explanation and systematicity) which are
taken at least to be good practice, at best to be conditions of coming to know or to
understand. But coming to know is tricky, even by means of the systematic practice
of question and answer.
7. The Interrogative
Meno’s paradox, I suggest in Chapter 10, asks not only about how we come to have
beliefs, or true beliefs, or knowledge—about how we ever have any answers—but also
about the process of inquiry itself; about how we come up with the questions ever, at
all.51 Conversations, seen in medias res,52 take the questions for granted—one asks
and the other answers, and the question of what makes the questioner question does
not arise. But the psychology of the interrogative is puzzling (especially in the context
of paradox): what is it that makes us puzzle, or wonder, or ask? This question,
I believe, is posed by Meno’s paradox.53
The problem is not only how we know what question to ask, but also how we are
inquisitive at all. What is it that makes us ask a question, rather than merely making a
declaration? Platonic conversations seem to canvass two quite different, but import-
antly connected, answers. The first attends to the formalities of the conversational
process; the second to their psychological counterpart.
The formal answer may merely rest on the thought that the questioner comes to
the encounter already asking; being a questioner just is the role taken by the
questioner in the process, as answering just is the role taken by the respondent. In
this, the asking and the answering of questions—the processes that take place in the
dialogues themselves—assume some kind of formal distinction between the ques-
tioner and the respondent. After all, if I ask a question, not rhetorically but in the
hope of hearing an answer, then in some respect I don’t know the answer already.
I ask to get the answer; so I assume, or I hope, that my respondent will indeed be able
to provide it. So the questioner and the answerer are distinct in respect, one might
say, of their total mental states. Compare Socrates (who knows only that he doesn’t
know, but wishes to find out) and his interlocutor who thinks he does know, both in
general and the answer to Socrates’ question, but turns out to be ignorant. The two
are obviously distinct by virtue of their different total mental states, even if neither
50
Examples abound in the Euthydemus; and compare Aristotle’s caveat e.g. at Metaphysics 1005b22.
51
This has been a matter of a great deal of important discussion between Gail Fine and Dominic Scott
in recent years, see Fine (1992, 2007, 2010) and Scott (1995, 2006).
52
Notably at Philebus 11a ff.
53
Compare and contrast Scott’s different versions of the puzzle posed by the paradox (1995).
has more first-order knowledge than the other. So, the paradigm case of such an
interaction will be a conversation between two distinct persons—portrayed, for
example, as Socrates and Euthyphro, or the Eleatic Stranger and Theaetetus.
Suppose, further, that we can make sense of the idea that I ask myself a question—a
question of a speculative sort—and expect an answer from myself. In some way,
I must expect the answer to come from some different part of myself, or from myself
differently disposed. So asking myself a question and expecting an answer from
myself presuppose something quite complex about the structure or makeup or
internal dispositions of myself. Plato seems to finesse this, at least on some occasions,
by thinking about an internal conversation as occurring between different parts of
the soul—as I argue in Chapter 6. On that account, internal question and answer will
have the characteristic of external question and answer—that the questioner and the
answerer are somehow distinct.54
Consider further the psychology of thinking as an internal dialogue:55
soc: . . . Do you call thinking what I call it?
tht: What do you call it?
soc: The talk that the soul goes through itself with itself, about the things it is
considering. I am explaining this to you, of course, without knowing. But the
soul appears to me, when it thinks, to be doing nothing but having a conversation,
itself asking questions of itself and answering them, asserting and denying. When-
ever it reaches a determinate point, whether it does so slowly or rushing more
speedily, and says the same thing [as itself ] and is no longer disagreeing [with
itself ], we call this its belief.56 So, I call believing saying, and belief something
said,57 but not spoken aloud or to another, but silently to oneself. (Theaetetus
189e–190a)
Here Socrates imagines the internal processes of the soul conforming to the sequence
of question and answer. Something similar happens at Apology 22e:
54
The metaphysics of this is, of course, complex: but it is worth noticing how commonplace these
internal conversations are in the dialogues. They may occur in contexts where the conversation is
construed as the making of a decision (recall the vivid example of Leontius at Republic 439e) and where
the literal partition of the person deciding is considered; or equally when more abstract questions are at
issue, e.g. Republic 523–5. Part of the significance of this connection is the seamlessness, for Plato, of
cognitive and ethical issues. Such cases usually suppose a real distinction between the parties; consequently
those anomalous cases where the interlocutor seems to disappear in the course of the conversation
(Philebus, or Parmenides, or Protagoras) are striking. These are missing persons (McCabe (2000)).
55
Compare Sophist 263, Philebus 38e.
56
Here I translate doxa as ‘belief ’. Others (notably Burnyeat (1990)) favour ‘judgement’, but I prefer to
keep that to translate words cognate with krinein, ‘to judge’.
57
There is a long play in this passage on the verb legein and its cognates (see n. 5). This is not, I take it,
mere carelessness on Plato’s part; but a deliberate focus, not only on the question of internal dialogue, but
also on what ‘saying’ involves (see McCabe (2014b)). Notice that ‘saying’ is here treated as the outcome of a
reflective process; to say is not merely to utter (even silently).
. . . so that I asked myself on behalf of the oracle whether I would prefer to be in the state that
I am in, neither wise in their [the craftsmen’s] wisdom nor ignorant in their ignorance, or
rather to be in both of their states. And I answered myself and the oracle that it would be best
for me to be in the state I am in.
Here Socrates is talking to himself, and answering back. But he also speaks on behalf
of the oracle, and the effect of this is to imagine the different sides of the conversation
as heard, as if from the outside and detached, by the oracle. And he seems to view the
answer provided by the examination of the pretenders to knowledge with similar
detachment (‘I answered myself and the oracle’). Thus his inner dialogue conditions
thinking in terms both of the play between each side and of the view from outside,
taken with what I have called reflective distance. The internal conversation of
Republic 7 gives us the same view—there the soul asks the questions: and perception
gives one answer, thought another. In this triangulated system, the conversation is
the focus of attention of the soul, from outside the conversation, not as a participant
from within (I argue for this in Chapter 6). The internal dialogue, thus, like the
written dialogues on the page, has both first and second-order features: it is a
discursive and reflective process between and about different points of view.
Does internalizing conversation explain how we come to ask questions at all? The
dynamic of internal questions and answers might explain both our continuing
puzzlement and its resolution; and the reflective dimension of internal dialogue
may give us an account of why resolution is needed. From the point of view outside
the conversation, after all, the dispute between the two sides needs to be settled, as a
matter of psychological comfort. The notion of an internal dialogue gives us some
account, then, of what the interrogative is, once it is established in our souls; and of
why it should be compelling, even in an external dialogue. Answering ‘why?’ is
demanding, not just by virtue of the insistence of the questioner, but as a matter of
psychological urgency. And indeed, the unhappiness of many of Socrates’ interlocu-
tors displays exactly the same kind of botheration: compare Callicles’ bad temper
(Gorgias 505c ff.) or Critias’ discomfiture (Charmides 169c) or even Euthyphro’s
hurried exit (Euthyphro 15e). The inner dialogue and the outer, that is to say, run in
parallel, each providing explanatory material to the other.
But all this starts, one might think, in the middle. While dialogue—inner and
outer—may explain both the persistence of a question and its psychological content,
it does not show why the question becomes compelling in the first place. Consider,
then, the discussion in Republic 7 of how the soul deals with the case of the three
fingers, where the middle finger is both larger than one and smaller than another.
Here puzzlement is not provoked by some kind of factitious contradiction,58 but by
58
On the contrary, in all of these contexts Socrates is very clear, in ways that the sophists of the
Euthydemus deny, that the opposite properties instantiated in some particular are always qualified in ways
that preclude contradiction. The problem, rather, is that this compresence of opposites is non-explanatory.
See McCabe (1994a).
the observation that no particular provides us with an explanation, or gives us the

causal structure, of any of the properties in question. The perceptible world itself does
not give us the answer to ‘why?’, nor to ‘what is it to be thus and so?’. On the
contrary, the puzzlement arises because the soul realizes that the particular case does
not furnish it with the explanation it seeks.
The exigent particular provokes because it fails to explain; and this provocation
resides in the world only because the soul takes it up. But the soul takes it up because
it has a complex constituency, characterized by the asking and answering of ques-
tions. So the interrogative is in part a feature of our psychological makeup, something
that comes with being a thinker, one who questions and answers, not merely a
consequence of the way the world is out there. It is also in part a consequence of
the way the world is: the world is not self-explanatory, but in need of explanation.
The explanation we then seek is not merely a matter of how we think about it, but
conditioned by the world. So the question ‘why?’ arises as a consequence of the
inconcinnity between what the soul expects and what the world provides: between us
and the world there is a vital reflective gap.
8. The Trouble with Writing

The inner dialogue works just like the conversations portrayed in the dialogues,
where sometimes characters in the dialogue, sometimes the reader herself, take the
view from outside, and the second-order perspective. But this may leave the reader in
a difficult position, twice over. First of all, even if the dialogues represent conversa-
tion, they are themselves monologous written texts. What is the relation between the
conversations that are represented and the representation itself? If conversation
matters so much, what do we learn from texts that are in themselves not conversa-
tional? The framed arguments, secondly, are often indeterminate, ending in impasse
or volte-face, or merely in the discomfiture of the interlocutor. The same may be said
of the framing arguments, too: they may not be decisive or final. By the end of any
dialogue we may be left in as much of a puzzle as the interlocutors themselves.59 How
does that help us understand what these texts are trying to say to us, or what they take
philosophizing to be, when it starts with a written text?
Consider Plato’s own discussion of the dangers of literacy. Late in the Phaedrus,
turning from the analysis of spoken words to the discussion of written texts,60
Socrates tells the story of the invention of writing by the Egyptian Theuth. Theuth
boasts to his king, Thamus, of his discovery:
This learning, your highness, will make the Egyptians wiser and more retentive in memory; for
I have found a panacea61 for memory and wisdom. (Phaedrus 274e)
59
Compare here the close of the Theaetetus or the Euthydemus.
60
The passage is about logoi in various forms; compare 274a8 and 274b9. See nn. 5 and 57.
61
Socrates makes a series of plays on pharmakon (a drug or poison); see Derrida (1972); Rowe (1985);
and Ferrari (1990).
But he receives an unexpected reply:

Theuth, you are a most ingenious man; but one man is able to make crafty inventions, another
to judge the amount of harm and benefit in them for those who will use them. And now you, as
the father of the letters, ascribe to them a capacity quite the opposite of the one they have. For
this will instil forgetfulness in the souls of those who learn it, through their failing to practise
their memory.62 For they will remember by means of trust in writing from outside, in
characters that belong to another, instead of themselves recollecting63 by themselves from
within. You have found no panacea for memory, but for forgetting. You provide the appear-
ance of wisdom for your pupils, not its truth,64 for becoming conversant with many things
without teaching, they will seem to be knowledgeable about much, while they are for the
most part ignorant and hard to be with, having become apparent-wise instead of wise.
(Phaedrus 274e–275a)
That this passage has been anticipated in the opening stages of the discussion
emphasizes its importance. In the fictional world of Socrates’ meeting with Phaedrus
(an unusual encounter with Socrates outside the city), Phaedrus offers to tell him the
main points of a speech Lysias made about love; Socrates suggests that Phaedrus may
have committed the speech to memory, but then reveals that Phaedrus is in fact
carrying the text of the speech beneath his cloak (228d). So from the beginning the
dialogue focuses on the difference between oral techniques of memory and the
facility of relying on the written word for accurate reportage. Now we are in another
fiction, the meeting between Theuth and Thamus, and looking at it, with Socrates
and Phaedrus, from the outside. It is resonant with the opening frame, but the
baroque embedding of the Egyptian story within the dialogue between Socrates
and Phaedrus calls attention to the fictional nature of the frame itself. As a conse-
quence, the reader may focus not merely on Theuth and Thamus, nor even on
Socrates and Phaedrus, but on her own distance from the action as she reads. But
when that happens, paradox threatens: Thamus suggests that we should not trust
what we read in writing; and we notice, as Socrates tells the story, that we read it in
writing. What can we trust? Are we to conclude that the dialogues themselves merely
convey the appearance of knowledge?
62
This picks up the suggestion, falsified by Phaedrus’ possession of the text of Lysias’ speech, that he
might have repeated the speech over and over until he has it off by heart; 228a–b. Socrates’ striking remark,
‘if I do not know Phaedrus, I have even forgotten myself ’, 228a, is picked up in Thamus’ insistence that the
business of learning and forgetting is something to do with the persons who remember, or fail.
63
Many are tempted to read here a reference to a Platonic ‘Doctrine of Recollection’; and some such
idea does turn up in the Phaedrus myth, 246a ff. However, I have argued above that we should not be too
quick to read the dialogues as containing mere allusions to some background set of doctrines; in Chapter 8
I offer a quite different account of how cross-references to ‘recollecting’ from the Euthydemus to the Meno
might be read.
64
Here there is a series of striking plays on the word doxa, ‘belief ’, ‘seeming’; and there is a raft of
allusions to the poem of Parmenides as well as to the Republic, notably 474 ff., 510–11.
Socrates continues:
Writing, Phaedrus, has this clever feature, and in truth it is like painting.65 For the products of
that art, too, stand as if alive; but if someone asks them a question, they preserve a high-minded
silence. The same happens with written words66 too: you might think they speak as having
some kind of intelligence, but if someone asks them a question, seeking to understand what has
been said, they always signify just one and the same thing. Every word, once written down, rolls
around all over the place, for those who understand and for those who have no concern for it
alike, and it does not know to whom it should speak and to whom not. When it is abused or
unjustly reviled, it needs its parent to help it; itself it is unable to protect itself, nor to come to
its own aid. (Phaedrus 275d–e)
Why might it matter that the written word is inflexible? If what is written is just a
truth, why should we think that this is problematic in any way at all for the reader? Is
Socrates’ claim here that any truth is somehow relative to its context,67 and that the
written word fails to carry its context with it? If p may be true in Athens in the fifth
century bc, it may not be true in London in the twenty-first century ad—lacking the
right context, shall I just miss the truth-value altogether? Is Socrates’ point really so
banal (and so easily fixed by adding the detailed specification of the context)?
We might rethink here: the passage gives us warrant for wondering not only about
words (the writing of which Theuth boasted to his king), nor about sentences or
propositions, but also about what is said embedded in a context of a different kind: a
logical context in which what is said is related to what supports it and what follows
from it. This is what, for words, amounts to self-defence (just as the question and
answer of conversation supplies logical structure and explanatory context: compare a
similar theme at Protagoras 329a). Some part of what Socrates says invites us to think
of any utterance as something that never stands alone (it is never an orphan), but
whose content is inextricably linked to a context of this logical and explanatory sort.
The risk of the written word is not that it changes its truth-value or its meaning, but
that it loses its explanatory support when it stands alone. To understand an utterance,
that is to say, is not merely to grasp its meaning, but to comprehend it within its
explanatory relations to others.68 Understanding does not come piecemeal, nor one
by one.69
65
Compare the discussion of imitation and then painting at Republic 595c ff.; see Chapter 11.
66
Logoi again. The qualification ‘written’ is implicit from the preceding speech.
67
Hence the image of location, ‘rolling around all over the place’, and the concern about who is the
interlocutor of the written word. Compare Republic 479d for the same expression used to describe sensible
particulars that are both large and small etc., but non-explanatory.
68
It is frequently observed that we do better to translate epistêmê and its cognates in terms of
‘understanding’ rather than ‘knowledge’ in the Greek context (see e.g. Burnyeat (1981)). This is surely
right, but it requires us to understand ‘understanding’ as grasping these systematic explanatory connec-
tions, not merely as understanding meaning. This complex account of logos in the Phaedrus goes to the
same point.
69
Compare and contrast what I have called the ‘episodism’ of some extreme sophists: see Chapter 7.
The paradox about writing focuses, in this respect, on the ideal virtues of conver-
sation. The written word is ‘hard to be with’; it can neither ask nor answer questions,
nor respond to an interlocutor, nor pick its discussant. It is, in this, somehow or other
inaccessible to the person who reads, in ways that direct conversations are not. And
yet the formal features of paradox, working on us as we read, have a surprising
flexibility, where—as in the river paradox—the same words, repeated, engage in an
argumentative sequence with their puzzled audience. In the Phaedrus, the puzzle
works in the reflective dimension by considering not only what is said but also how it
is said. We are not now passive readers of the encounter between Socrates and
Phaedrus, but active and critical observers, aware of both the content of what we
read and its context: its verbal form as well as its first-order content are included in
the philosophical business of the text. The paradox about writing is thus conversa-
tional after all.
9. Memory and Listening

Indeed the discussion of memory, here and elsewhere,70 ties in with the formal
features of conversation—Socrates’ insistence on short and formally simple questions
and answers. His objection from his own forgetfulness at Protagoras 334c–d must be
disingenuous; but it calls attention to the frame conversation, and brings out a
different emphasis of the paradox of writing. It is a feature of question and answer
that both parties are somehow responsive to each other. They are each accountable to
the other, so that they need to be attentive to the exact content of what the other says,
and that attentiveness is the source of any progress they may make. Conversation,
thus, is not only about speaking, but also about listening.71 And listening is hard. It
demands a kind of silent patience—which seems to be one of Socrates’ personal
virtues.72 The written word, by contrast, is ‘hard to be with’—and it cannot hear at all.
This suggests that the joint nature of the spoken word, the conversational aspect, is
somehow important to Plato’s account of the way philosophy works. Return, how-
ever, to the silent dialogue within the soul: is this the ideal, to which ordinary
conversation merely aspires; or does ordinary conversation itself have features that
are ideal for philosophy? If the silent dialogue is the ideal, then philosophy should be
a solitary enterprise, something which, by virtue of the internal dialogue, has all the
advantages of conversation (the search for explanation, for example, or the ability to
think about opposing viewpoints) without the disadvantages of having to converse
70
E.g. Phaedo 73a–e: this is at first about the ordinary process of remembering, subsequently used as an
analogue of pre-natal recollection; see n. 63. Contrast, however, Philebus 33–4 (my thanks to Joachim
Aufderheide for reminding me).
71
Notice the emphasis on the modality of hearing at Phaedrus 275a7; compare Sophist 242 (my thanks
to Daniel Vazquez for this reference); and recall Heraclitus on listening, DK22B1, 34, 50 (my thanks to
Shaul Tor).
72
See Zagzebski (1996), 173 ff, on some of the small epistemic virtues.
with someone who just might not be up to the task. This view of the solitary ideal,
indeed, might be promoted by the Republic’s suggestion that a philosophical educa-
tion results in the philosopher’s being able to see better, where better seeing is
something that is private to the person who sees. Is that right? If so, perhaps
conversation is just a second-best journey, compared with the singularity of the
philosopher’s insight into the good. And then perhaps the dialogues would aspire to
being treatises after all. Would that leave the complex structures of the dialogues as
mere artifice, to induce us to undertake the philosophical journey alone?
10. Perception and Knowledge

Plato’s apparent interest in conversation as a way to philosophize, indeed, may fit
uneasily with some of the views that are often taken to be canonical of ‘Platonism’.73
Many, both in antiquity and today, suppose that Plato has two major claims to make
in explaining knowledge’s relation to reality. The first is that knowledge is (exactly is)
the soul’s engagement with a special set of objects, namely forms, which are ultimate
and transcendent realities and not of this sensible world. The second is that the soul’s
engagement with forms is somehow or other ineffable, or direct and unmediated, or
intuitive, a sheer grasp of something that cannot be accessed any other way; this
grasp, so it is thought, is the result of a long process of education and learning, but it
transcends that process. These two views are sometimes taken even more strongly:
that the forms’ direct relation with the soul of the philosopher is what determines her
state of knowledge—if she is in contact with the forms, she knows, and if not, not.74
So—on this extreme view—just as perception is a direct, unmediated interaction
between object and sense, so knowledge is a state of mind in which the knower is
directly affected by the objects of her knowledge, in some extraordinary and unme-
diated way.75 If this is indeed how Plato regularly conceives of knowledge, it is hard
to see just how conversation might be a way to acquire it, or even a means to
developing the way to acquire it.76
Suppose, for the sake of argument (and against the grain of the more agnostic
reading of Plato I canvassed above), that we can find strongly asserted views that
count as ‘Platonism’ in the dialogues, fixed points of doctrine underlying the whole
corpus. After all, in some dialogues Socrates does indeed seem to say that it is at least
73
Compare e.g. Moravcsik (1992); Gerson (2009).
74
This is a view that is sometimes taken of Rep. 476–80, or of the elaborate imagery of Symp. 210–12.
The Phaedrus may provide a counter to the Symposium, thus interpreted.
75
We might compare Plotinus’ claim (which takes this unmediated relation seriously) that intellect is to
be identified with its objects; see e.g. Enn. V.5.1, 62–5. My thanks to Dimitrios Vasilakis.
76
There is a parallel question here about the ‘Socratic method’, which is sometimes seen as purely
negative; and compare the assessment of the ‘noble sophist’ at Sophist 230b ff.
a mark of knowledge that the soul is in contact (whatever that involves) with forms.77
Then, perhaps, the relation between the soul and the forms, often described as a kind
of perception (most often vision, but sometimes also touch), is so direct that it is not
susceptible to ordinary cognitive expression such as, for example, propositional
content.78 The soul just sees the form of the beautiful; and what it is for the soul to
see like this is for it to be affected, in a brute fashion, by the form. Knowledge would
be, as one might think, a ‘raw feel’ of the soul, analogous to what might be the ‘raw
feel’ of ordinary sensory experience.79
On such an account, knowledge would be understood in terms of the piecemeal
nature of such raw feels. After all, if perception is a brute experience, it need contain
no component that connects one experience to another.80 This would rule out, too, as
part of the same epistemic modality, second-order knowledge of what is known (for
example, knowing of some piece of knowledge that it is known), or any ordered
relations between one experience and another. It is a commonplace of some con-
temporary epistemology that knowledge could be understood as a relation to some
truth, or some proposition, or some belief, where that truth or proposition or belief is
somehow self-standing, independent, or singular.81 Why might not Plato think too
that knowledge is the piecemeal, brute, interaction of the soul with some one form in
particular, and none of the others (whichever they may be82)?
If Platonic knowledge is like this, it is hard to see how it does what it is supposed to
do. If knowledge occurs only when the philosopher contacts forms, what then is the
connection between her cognition of the forms and her cognition of the phenomenal
world?83 In the Republic the epistemic state of the philosopher explains the require-
ment that philosophers rule, so her superior cognition cannot be restricted to her
experience of the forms alone. Perhaps, instead, the forms ground her phenomenal
77
E.g. Republic 511b. Here the vocabulary of contact trades on two connected notions: one is that touch
is essential for (efficient) causation; the other is that this kind of grasp of the form is a kind of perception.
Neither causation nor perception, I suggest, require the grasping to be unmediated.
78
The question here is not whether, once we have a perceptual experience, we could then describe it
propositionally (so perception is experience + judgement: Plato does at least experiment with such a view at
Theaetetus 184b ff.). Rather, the issue is whether the perceptual experience is in itself already expressed in
ways that are somehow cognitive (for example, propositionally). I use the contrast ‘cognitive’ (for the
latter) and ‘non-cognitive’ (for the former) to capture this point; not a great deal hangs on what is involved
in cognition here, apart from perception’s expressibility as such. Nothing I say here is intended to involve
any commitment one way or another to entities called ‘propositions’.
79
I expand on Plato’s approach to this contrast in Chapter 9.
80
This, among other objections, is one of the ways in which Socrates rebuts the relativism of Protagoras
(at Theaetetus 169–86) in the context of the theory of raw perception advanced at Theaetetus 153d ff.
81
This approach is well exemplified in Gettier’s famous paper (1996), which even suggests a connection
with Plato’s Theaetetus (although, in my view, the connection is misplaced). Contrast e.g. Zagzebski (1996)
or Williamson (2000) for different critiques, both of the prioritization of beliefs and for the piecemeal
nature of knowledge claims. My thanks to Lucy Campbell, Mike Coxhead, Alexander Greenberg, Ellisif
Wasmuth, and Michael Withey for discussion.
82
The scope of the theory of forms is, as I have argued (1994), something that Plato himself calls into
question; see Parmenides 130.
83
This is the ‘greatest difficulty’ launched against the theory of forms at Parmenides 133c ff.
cognition, but are themselves grasped in an unmediated way. But then what account
of that grounding can be given? And does such a grounded cognition of the
phenomena count as knowledge, or not? If so, the knowledge of phenomenal matters
is different (because mediated) from the unmediated knowledge of forms: how? If
not, can it be shown to be more reliable or useful than the familiarity produced by the
banausic life?84
Suppose that this kind of brute cognition is the best case, the supreme epistemic
condition. If so, other epistemic conditions or activities will be somehow secondary
or derivative or even vulnerable to mistakes in ways to which the seeing of the form is
immune. The brute seeing of the form, then, would not be expressible in complex
language without mediation by other and lower kinds of cognition than knowledge
itself. As a consequence, the ineffable contact with the form is only accessible to
reflection by lower faculties and by means of discourse that is somehow vitiated by
its complexity. Whether or not knowledge is a goal impossible to reach, it is hard to
see how discourse could lead in its direction, or how the reflective pursuits of the
philosopher would make her especially able to have this kind of transcendent
experience.
It would thus be possible to have a single first-order piece of knowledge without
any reflective element, and still properly lay claim to knowledge. But then knowledge
has no intimate connection with explanation (which is not piecemeal) or with science
or skill. Is that how we should understand Plato’s view of the end of philosophy?
Further, if knowledge occurs only at the moments when the philosopher contacts
forms, and not otherwise, then not only is knowledge thoroughly etiolated, but there
is also little to say about what makes it valuable or about its role in the goodness of the
life of the person who has it, apart from its casting some ineffable light.85 For that life
is itself a phenomenal, particular matter, not lived at the level of transcendent and
ultimate reality. Yet in the Republic and elsewhere (for example, at Euthydemus 281),
wisdom is the sole condition of the life of happiness.86
This version of Platonic idealism is, I argue variously in what follows, both
unhelpful and mistaken for Plato and for Aristotle too. It is, first of all, implausible
that the final stage of the philosopher’s search, a stage reached by means of philo-
sophical conversation, would be starkly different in every way from conversation—
not only solitary, but also non-cognitive, non-discursive, even non-propositional.
The story—on the view I contest—would be that the philosopher talks and talks and
gives and receives accounts, for years and years of her educated life; and then
84
These are the questions raised by the blindness of the returning philosopher to the cave, Republic
517a.
85
I argue in Chapters 11 and 12 that wisdom is of what I call transformative value: it is of intrinsic value
in a life, and it transforms the value of what is in the life, making other things in the wise person’s life
valuable too.
86
This does not mean, as I argue in Chapters 12 and 13, that all there is to happiness is wisdom, nor that
the account of the goodness of wisdom abstracts from the quality of the wise person’s life.
suddenly one day this talk is left behind, and replaced by a new faculty, the soul’s
vision of the good, or of the beautiful—or even, if this is thoroughly piecemeal, of the
equal or the same or the large.
This account seems to me to fit uneasily with the form of the dialogues, as well as
with their political and ethical content. The direct working of the dialogues—I argue
in Chapters 9 to 11—engages the reader in a kind of reflective relation with what is
said there; and it does so, strikingly, by using the perceptual vocabulary of looking
and seeing and touching and grasping. So the continuing search is marked both by
the language of conversation and by the language of perception at the same time.
This fits with the picture of the philosopher’s ascent from the cave in the Republic: if
the final stage is a kind of mental seeing, that kind of seeing must somehow or other
have continuity with the conversations that went before.87
With this in mind, I argue that at least in the Charmides (in Chapter 9) and in the
Republic (in Chapters 6 and 10), perception is not a ‘raw feel’ at all, but, as I say,
‘cooked’. Ordinary perception, at least in these dialogues, is at the very least discur-
sive, so that perception directly reports in a full statement: ‘the rose is red’. What is
more, perception has the capacity to perceive that it perceives, so that in its discursive
content there may appear reference to the perception itself. Vision may then report ‘I
see that the rose is red’, and include a reference to the subject in its cognitive content.
Perception can, in this way, be second order and, in some way, reflective. These
passages, I argue, show how perception can be ‘twice-cooked’. So too then can its
mental analogue, knowledge.
11. Consciousness and Perception

This kind of complexity in perception is (with a little help from Brentano88) what has
encouraged the interesting view that Plato and Aristotle are here talking about
subjective consciousness, the ‘what it is like’ to have perceptual experience.89 But
subjective consciousness is not quite the focus of attention, I claim, either in Plato
(Chapters 9 and 11) or in Aristotle (Chapters 14 and 16), although they do seek to
explain what perception is from the subjective point of view. For one thing, neither
Plato nor Aristotle is here trying to tackle the problem of scepticism, how we might
know anything at all, to which the immediate privacy of conscious experience might
be a riposte. They are, instead, asking about the problem of cognitive virtue: how do
we come to understand, how do we become wise?90
87
Seeing in the frame e.g. at Republic 432d, 495a, 514a2; the eye of the soul e.g. at 518c; conversation
and dialectic e.g. at 511b–c; and synoptic dialectic (combining the two notions) at 537d–e.
88
Brentano (1973) Part II. I owe a special debt here to Mark Textor, for conversations about Brentano
and Aristotle over some years.
89
Kosman (1975); Caston (2002), (2005); and see Nagel’s seminal (1979).
90
See here the discussion in Zagzebski of just how we might conceive of knowledge as a virtue (1996);
see also Vogt (2012). The central role of learning in virtue is well captured by Burnyeat (1980).
Both philosophers regularly take reflection to be essential to improving our

cognitive capacities, to making them integrated, coherent, or even systematic.91 On
this model, they take second-order perception, I argue in Chapters 9, 11, 14, and 16,
to have a reflective function in improving our perceptual capacities, whether we think
about those in the context of sensory perception or of the idealized perception that is
associated with knowledge. This is not a matter of sheer consciousness, but of
something far more discursive. Both Plato and Aristotle, I argue, seek to explain
perception both from the point of view of its objects and from the point of view of its
subject. But their concern for the subject is not merely its subjective experiential
character. Instead they wonder about how the subject sees herself seeing; or notices
her own standpoint or perspective; or recognizes that it is she herself who sees this.
For perception here is, as I say, twice-cooked; it includes the experience of reflection,
but also its articulation and perceptually reflective dimension: including reflection on
what we ourselves perceive.
It is for this reason, I argue in Chapter 11, that perception is the sort of thing we
can improve and develop, on the view of Plato and Aristotle. Perception, both
sensory and intellectual, is normative—we can learn to see better. It is part of the
activity of perception to perceive itself, and, as a matter of perception itself, to learn
how to perceive more effectively, faster, wider, with greater sensibility.92 And when
perception is used either as an analogue for the activities of the intellect or in a broad
sense to include the activities of the intellect, it can have the same kind of cognitive
richness as for the sensory modalities, and can be susceptible to improvement in the
same kinds of ways. Perception in the intellectual sense, then, is properly associated
with cognitive virtue—for as in the case of moral virtue, habit, reflection, and practice
lead to improvement and perfection.93
Aristotle may be thought to go further still. In Chapter 14, I argue that Aristotle’s
account of perception in the de anima is continuous with his interest in the
normativity of nature, such that he can think of perceiving on the model of learning
to be good in Nicomachean Ethics 2.1: we can learn to become better perceivers. In
Chapter 16, I further suggest that he connects this notion of perceiving with his
account of what virtue-friends do ‘together’, namely perceiving as some kind of joint
enterprise, joint (subjective) perception. The rarefied friendship of ‘virtue-friends’, he
suggests in Eudemian Ethics 7.12, makes it possible for such friends jointly to see or
to hear. They might be thought of as two halves of the same soul, whose friendship
91
For Aristotle, central here is the role of demonstration; see Posterior Analytics 1.1–2.
92
There are rich examples here of how perception can improve through practice, habituation, and
reflection: for example, the speed of the eye in sport, or the acuteness of the trained musical ear—each of
which is automatic and effortless, once the training is done, but is acquired by complex means, including
discursive reflection. The kind of normativity I have in mind, that is, belongs to the faculty of sense, not the
sense-organ (it is not a matter of 20:20 vision).
93
I suggest in Chapter 11 that this view opens up for Plato the possibility of moral sensibility. This is
connected to a further question of just how far moral and intellectual virtue are distinct for Plato; but this is
for a larger project.
allows the near-perfection of joint seeing—each understands the other’s subjective

perspective so well that both can be said to share it. This high-flown idea is, I argue,
appropriate to thinking about friendships of this rare kind, in the context of practis-
ing virtue. The joint perception thus described is a long way from ideas of subjective
consciousness, joint (whatever that would mean) or otherwise.94
12. Aristotle Reading Plato

But these ideas about perception in Aristotle are, I claim in Chapters 14 and 16, the
product of Aristotle’s very close reading of Plato. It is a commonplace to point to
passages where Aristotle directly criticizes some theory that he attributes to Plato or
the Academy—notably in the complex arguments against the theory of forms, which
were live issues in the Academy in his time there.95 These passages give us evidence
that at some point Plato or his successors were actively interested in a theory of
transcendent forms, a theory that Aristotle finds ontologically expensive, logically
equivocated, and explanatorily barren.96 This is not, however, all that Aristotle found
of philosophical significance in the Platonic dialogues. Rather, as I argue in
Chapter 15 for the case of Metaphysics Z, he is careful to mount different arguments
(in this case about the universal) against what is to be found in different dialogues.97
Aristotle, I argue, has a rich and nuanced view of the dialogues, which shows up in
allusion and intertextual reference, rather than in direct citation. Thus he uses the
dialogues, I suggest, not so much as background to his discussions, but as dialectical
partners.98 This should encourage us to do likewise.
Since all these works were written when Greek culture was moving away from the
oral towards the literate culture, but retaining some of the facility of memory and
receptiveness that characterizes orality, the kind of dense and complex allusion
I describe would have been evident to Aristotle’s contemporary readers. It is not
surprising that philosophers of Aristotle’s period would see these allusions, especially
in Aristotle’s use of the distinctive language of different dialogues, in the foreground
of what he says. Aristotle—whatever we might say of his ways of writing, or of the
status of his works99—has an acute ear for allusion.
94
This point often gets lost in translation: in Chapter 16 I argue that sunaisthanesthai means what it
says—sensing together, not either joint awareness or even single consciousness. It is sometimes assumed
that in later philosophy, e.g. the Stoa, sunaisthêsis comes to mean just ‘consciousness’; I argue against this
view for the notable case of Hierocles in McCabe (2013).
95
See Fine (1993).
96
E.g. at EN 1.6; Met. 13. 4, 5.
97
In Rhetoric 1.1, as I have argued (1994b), he uses the Platonic background from two different
dialogues to give subtlety to his own refined view of the proper practices of rhetoric.
98
See Whiting (2006), (2012).
99
The old ‘lecture notes’ view seems to me to sell very short the dense composition of some of the
works—notably Aristotle’s first chapters are rendered with care. On reading Aristotle, see Burnyeat (2002).
13. Chronology?
Aristotle’s sensitivity to the differences between dialogues raises a different question,
which has long burned among Platonic scholars. How should we understand those
differences? Does Plato’s thought develop? If so, how, and how across different
dialogues?
Especially since the work of Owen and Vlastos,100 there has been dispute about
whether we should arrange the dialogues into three ‘periods’ (the ‘Socratic’, the
middle, and the late, critical period) to show how Plato’s thought changed over his
philosophical lifetime, or whether this simply misses the unity of Plato’s thought.101
Vlastos proposed that we should distinguish what is ‘Socratic’—reflecting on the
views of the historical Socrates—and what is Platonic, characterizing the grand
metaphysics of Republic or Phaedo. Owen argued that the later dialogues show us a
Plato breaking free of the metaphysics of transcendent forms, and turning himself to
a more analytic frame of philosophy. Should we then think, for example, that the
arguments against transcendent forms in the Parmenides were thought decisive by
Plato, or are somehow disingenuous? I have long declared myself an ‘inveterate
developmentalist’ (as perhaps Chapter 8, on the Parmenides, makes clear),102 but
my views on what that involves have themselves changed.
One version of the developmentalist hypothesis is that there are fixed Platonic
doctrines, represented one way or another in the dialogues of each of the three main
groups, which are somehow incompatible with each other. Since I am not convinced
that the dialogues are vehicles for doctrine, this version of developmentalism does
not recommend itself to me. But if the dialogues are—as I have suggested—more
open pieces of philosophical thinking, constructed in part so that they engage the
reader, then there is no reason at all why they should find themselves always saying
the same thing, or even urging the reader towards fixed conclusions, anticipated from
the start.
There is equally no reason, however, to think that dialogues thus understood as
philosophical conversations should be completely agnostic about any conclusion at
all, or that they may be sceptical, or indeterminate. Each of two parties to a
conversation must come to it with some antecedent views—even if those very
views come under scrutiny in the course of the conversation—and they may well,
as a result of the conversation, take some conclusions to be foreclosed, and others
encouraged. The openness of the dialogues should not be taken to mean that they are
indeterminate, any more than a live conversation between two philosophers, to be
worthwhile by Plato’s lights, need end with nothing concluded at all. The dialogues
may present views, and they may make tremendous progress both in understanding
those views and seeing why we might think they are good ones (so they may present
100
Owen e.g. (1986b); Vlastos e.g. (1991).
101 102
Compare here e.g. Kahn (1996) and Rowe (2010). See McCabe (2001).
views about views: second-order viewing). What those views may be will surely
change after further reflection, or with hindsight, or under different driving concerns,
even if that change may still not turn them into dogma. But I am not convinced at all
that the dialogues present doctrines to which the arguments are designed to lead,
failure to reach which should count as a philosophical failure. Their agenda, as
I suppose here, is in many ways what it appears to be—to think about the questions
in hand.
So, over a series of conversations, either one of the interlocutors might change
their minds. Over the course of a series of conversations, looking back to what was
agreed on some earlier occasion, any philosopher might think that agreement was
unsound, and take his earlier self to task for going too readily down some primrose
path. These critical interactions with others and one’s former self count, I believe, as
development, or at least as changing one’s mind; and this kind of openness is central
to genuine philosophical conversation. And Plato seems to me to do this, often
returning later to something that bothered him before, and changing its emphasis, or
rejecting it altogether. Two cases in point are offered below, in the reflections, as
I take them to be, of the Euthydemus on an earlier dialogue, the Meno (Chapters 10
and 12, on questions of knowledge and recollection, and questions of the nature of
value, respectively).
It is a standard view that, on the developmental ordering of the dialogues, the
Euthydemus should belong in the early group, either beside or earlier than the Meno.
And indeed if we find intertextuality between one dialogue and another, we might
find it hard to say which is reflecting on which without some other kind of evidence
or assumptions for their ordering. But to this problem, it seems to me, the contrast
between the oral and the literary traditions may come to our aid. Suppose—as
happens in the case of Euthydemus 296c–d and Meno 81b ff.103—that the intertext-
uality is asymmetrical: in the one case highly condensed, almost telegraphic, and in
the other case discursive and detailed. How are we to interpret the telegraphic
version? On one view, it is something that we hold in our minds (without knowing
either its significance or that it is significant at all) until all is revealed at some later
date. On that account, the condensed version is a coded allusion to something else; it
does not stand in an argumentative relation to the full version, but merely as an
allusion, a vague promissory note whose significance is unremarked until some point
in the future. But on a different account, the telegraphed version refers backwards to
something we have already encountered, whose detailed expression we have easily
remembered (this is where orality’s virtues may be seen). In that case, the telegraphic
condensed version refers us to what we already have in mind; that is, to the discursive
103
A different and more controversial example is the relation between the description of the power of
dialectic in the middle books of the Republic and Euthydemus 290. Here too, I have argued that the
pregnancy of the Euthydemus passage is likely to refer back to the Republic rather than anticipating it
(contra Kahn’s prolêpsis (1996); see McCabe (2001)).
one, rather than the other way around. We interpret the condensed version by virtue
of what we already know (rather than keeping it in mind for some indefinite future).
This in turn allows the two versions to stand in an argumentative relation to each
other: the back-reference may endorse or criticize; but in either case, it reflects (rather
than merely alluding to) the previous text. Plato is not, I say, designing the whole
sequence of dialogues in advance, so that the telegraphic passage looks forward to its
full explication later, but the other way around. He expects us to learn, to think and to
remember, to criticize and to evaluate, not just to decode.
This is especially to the point, I suggest, in the cases I here document, where the
telegraphic version is in fact critical of ideas and arguments offered in the discursive
one. In cases like this, the dialogues stand in reflective relations to one another, such
that we can discern a critical development across dialogues, and over Plato’s lifetime.
This is not so easy to handle, of course, since what we may find is a philosophical
problem that is severally treated (we might ask, again about the Euthydemus, whether
dialogues that seem to turn on the same sort of problem, such as the puzzle about
falsehood and the possibility of contradiction, belong chronologically with others
with the same subject matter—in this case, notably the Theaetetus and the Sophist).
Should we think that Plato develops his driving puzzles over time? Or does he return
later to worry about them all over again? One might think that this is a question
whose answer will carry little illumination. Instead, I suggest, the developmental
hypothesis should be retained only insofar as it shows up development in the
reflective content of the dialogues, rather than in some constructed history of Plato’s
absolute philosophical commitments and his doctrines over his life. It is, like the
grain of his philosophical writing, fine.
14. Philosophy and its History

If any of this is right, it asks further questions about engaging with the ancients as
part of the business of philosophy. There is, of course, a distinction between doing
philosophy and constructing a philosophy. I do not deny that Plato was a system-
atizer at times, even where I do deny that he was a dogmatist in the sense I have
outlined. But if I am right about the importance of conversation to his account of
the doing of philosophy, and if I am right in thinking that this determined how he
thought of the relation between the written dialogues and their readers, then reading
the dialogues of Plato, reading the puzzling work of his predecessors, and reading
Aristotle’s reflections on them, is, in each case, to engage in philosophical reflection,
rather than what is an equally difficult but different enterprise, studying the history of
philosophies—the histories of the constructs that may each count as a philosophy.
But the fact that these are works from more than two millennia ago makes,
I suggest, a significant difference to how they can come to engage us philosophically.
Suppose that we think that philosophy and its history are pretty much indissoluble—
that even when we read the latest volume of Mind we are still engaged with
philosophy’s past and that being engaged with the immediate past of the subject is a
part of coming to understand it or to do it—or to understand ourselves, in relation to
it.104 We might then think that this kind of approach should deal most of all with the
up-to-date—because philosophy (we might also think) moves in straight lines,
discarding broken theories and bad arguments and spurious puzzles in its march
into the future. In thinking this, we might be in the grip, perhaps, of a scientizing view
of the nature of philosophy (and also of a careless view of science). Or we might be
supposing that understanding works—as I suggested some take knowledge to do—as
the collecting of true beliefs; the larger the collection, the better. So, on all of that
account, the reflection on what is antiquated or outdated focuses on what is likely to
be false, and so is not a useful enterprise.
All of this seems to me to be full of misapprehension. Moreover, it misses some of
the brilliance of engaging with the very distant past.
Suppose that in engaging with past philosophizing we do so with a strong sense of
its antiquity. This should, at the very least, make us very much aware of just how
different the Greeks are from us.105 Of course, they may be so different that they are
inaccessible; or their distance may make the differences either invisible or uncross-
able. This is always the problem of the gulf between one society and another,
especially where the other is covered up by millennia of loss (it is a version of the
problem of the indeterminacy of translation). But even if we can find points of
contact between Plato and us, and even if we get far enough to think that we can
indeed engage in conversation with him, we shall be constantly reminded of those
differences. And some part of that reminder will be, I suggest, that the things that
bothered him may not have been the things that bother us. There are, perhaps, great
fractures in the history of philosophy that generate a completely new philosophical
landscape (one such would be the emergence of a particular kind of scepticism in the
early modern period).106 Platonic and Aristotelian epistemology may be unrecog-
nizable to us because they are not dominated by a defence against scepticism;107 Plato
is often more worried about falsehood than he may be about truth.
But we can learn to track these differences; and how we think may benefit from
exposure to these different landscapes. One particular difference might lie in the
institutionalization of philosophy in antiquity and now. Philosophy’s recent past has
been pretty technical, and embedded often in large institutions where specialization
is of central importance. As a consequence, philosophy is often seen to fall into
different sub-sections, where expertise does not travel across the boundaries (instead,
philosophers from one area may be seen as amateurs in another). This level of
specialization is of course impressive, demanding, and daunting to the amateur.
104 105
See here Savile (1996). But see Williams’ cautionary remarks (1993), ch. 1.
106
See Burnyeat (1982a).
107
For Aristotle the sceptic is a dead-end adversary (see e.g. Physics 193a3–9), although of course he
does engage with some serious deniers—e.g. the denier of the Law of Non-Contradiction in Metaphysics 4.
But it is likely to obscure places where there is an important connection, say between
ethics and logic, or between metaphysics and politics. Plato and Aristotle, by contrast,
even where they acknowledge differences across the subject (and even different
demands of specialization), nonetheless seem to suppose that one side of the subject
can usefully both inform and ground another.
This is notably so, I argue in what follows, in the connections we might make
between questions about knowledge and identity, or about logical structure on the
one hand, and questions about value on the other. One major benefit of reading
philosophy through conversations with Plato is to make licit the appeal to value when
we think about—for example—contradiction; or to virtue when we come to thinking
about knowledge. This richness of the ancients is a proper complement to the
extraordinary expertise of the moderns: it is not something to leave behind in the
philosophical museum.108,109
108
Translations from Greek or Latin are my own throughout except where otherwise indicated. The
essays are printed in a mixed order of chronology and theme, and I have largely left infelicities of the
originals, where I have noticed them, to stand.
109
My thanks to my friends and colleagues for their acute comments on this chapter: Joachim
Aufderheide, Verity Harte, Fiona Leigh, Shaul Tor, and Raphael Woolf. What remains is not their fault.
I
On Dialectic and Method
2
Heraclitus and the Art of Paradox
1. The River Argument

Everyone knows that Heraclitus was interested in rivers. Of course, he was neither a
geographer nor a cartographer—he was a philosopher. What have rivers to do with
philosophy?
There are three river ‘fragments’1 attributed to Heraclitus:
It is impossible to step into the same river twice.2 (91)
We both step and do not step, are and are not in the same rivers.3 (49a)
To those who step into the same rivers, different and different waters flow.4 (12)
Since each fragment says roughly the same thing about rivers, it has usually been
assumed that Heraclitus made just one remark about rivers, represented in a more or
1
There is a problem here, in deciding whether what remains to us of Heraclitus’ writings is a collection
of fragments of a connected whole, or rather some of a series of pithy, short statements, written rather as
they have come down to us. Certainly, while frs. 1 and 2 look like a part of an extended whole, the impact of
many of the other fragments comes from their abrupt formulation. Equally certainly, philosophers since
have used both methods of communication—for the latter, of course, the striking example is Wittgenstein.
The interpretation of Heraclitus that I offer here suggests that he wrote rather in the way of Wittgenstein,
so that while each statement is self-contained, it is connected, theoretically and in a sequence of argument,
to others. The surviving evidence does not help us to decide this issue either way.
2
Plato, Cratylus 402a and Aristotle, Metaphysics 1010a12. Commentators have objected that Plato’s
version contains a potential optative that is not archaic, cf. e.g. Kirk (1954), 372; though contrast Marcovich
(1967). Even were this test to survive the testimony of e.g. fr. 99, it remains a serious possibility that
Aristotle’s version may be derived from independent evidence; and it does not contain the offending mood.
On the defence of 91 in general, see Vlastos (1955), 339.
3
This is perhaps the most suspect of the three fragments, on independent grounds, since it comes from
a late source, the allegorist Heraclitus Homericus. It is supported, however, by its reappearance in Seneca,
Epistolae Morales 58.23: in idem flumen bis descendimus et non descendimus; though see Kahn, who attacks
the reliability of Heraclitus Homericus in general terms ((1979), 339 n. 431). Kirk is inclined to accept the
Senecan version, but baulks at eimen te kai ouk eimen on the grounds that this is an impossible existential
reading of einai: ‘an existential judgement of this sort could only be accepted for Heraclitus by those who
are content to see him through the eyes of Hegel.’ But surely here we have an entirely proper locative usage
(particularly after the locative prefix en- in the first verb), not only well attested from Homer onwards
(cf. Kahn (1973), 156 ff.) but also entirely suitable in the context.
4
Vlastos, unusually among the commentators, rejects 12 as a variant of 91 ((1955), 308). It should be
observed, however, and this objection perhaps puts 12 on a par with the others, that the Stoa, through which this
is transmitted, is a notorious distorter of ancient texts, concerned as the Hellenistic schools were to establish
their own credentials by claiming antique provenance for their own views, cf. Cicero, Academica I.15 ff.
36 HERACLITUS AND THE ART OF PARADOX
less garbled form in the three citations. The game, on this hypothesis, has been to
decide which one is the most ‘genuine’. 12 is generally preferred, where Arius
Didymus is citing Cleanthes, who may be supposed to have had access to a text of
Heraclitus (Diogenes Laertius VII. 174; Stoic interest in Heraclitus is well attested).
By contrast, the evidence of Plato and Aristotle for 91 is considered to be suspect, on
the grounds that they were unscrupulous in their abuse of received opinions.
Nevertheless, these early sources offer us two citations of 91, which might be
independent of each other. The source of 49a, on the other hand, is almost univer-
sally repudiated as a late and not particularly careful allegorist, Heraclitus Home-
ricus. This fragment, however, does have some support from Seneca, and cannot be
expelled out of hand.
So there are arguments for and against any one of these fragments, as indeed for
any of the remains of Heraclitus. Yet each river fragment is scrutinized with a less
than generous eye precisely because of the underlying assumption that only one of
them can be the real thing. That is, grounds for disbelief in any of the fragments are
relative, so that each fragment is doubted in terms of the apparently better claim of
another. But without the assumption that there was only one river fragment, there are
no absolute grounds for dismissing the testimony of any of our sources. And if we
drop that assumption, there is something to be gained by retaining all three frag-
ments as genuine sayings of Heraclitus.
Try the fragments in the order I have given them: 91 and 49a are both paradoxes,
challenges to common sense; but they are paradoxes at different stages of discourse.
The outrageous 91 directly challenges common sense; and it requires no philo-
sophical effort to understand it. It appears, of course, immediately absurd to deny us
the ability to go on getting our feet wet—so that we immediately respond ‘No, it is
possible to step into the same river twice’. Let us call this response to paradox the
doxa. This is the correlate of the paradox, the truth that the paradox denies; any
paradox has a corresponding doxa, just because seeing a paradox to be surprising or
paradoxical involves us in the judgement that it appears to be false. The doxa, then, is
the contradictory of the paradox. So if the paradox is as false as it is paradoxical, then
the paradox implies the doxa by contraposition.
In the case of Heraclitus 91, the doxa comes immediately from common sense. ‘It
is possible to step into the same river twice’ is a truth we do not usually articulate;
only the challenge of the paradox ‘it is impossible to step into the same river twice’
forces us to do so. However, the reflection that follows articulation complicates
matters. For we can justify both the doxa (for example, the constancy of the
landmarks guarantees that it is the same river we are fording) and the paradox
(watch the leaf float downstream as we try to repeat the step). This is so because the
term ‘river’ is opaque,5 and allows for this ambivalent justification. So, for different
5
I am grateful to Dorothea Frede for this term.
HERACLITUS AND THE ART OF PARADOX 37
reasons, both may be thought to be true. Instead of common-sense, unarticulated

assumptions, we now have a pair of statements both of which can be justified, but
which appear to contradict each other. At that point, we become enmeshed in a
paradox of a more formal type, the synthesis of paradox and doxa—49a: ‘we both
step and do not step, are and are not in the same rivers’.
But puzzlement does not end there: 91 has already activated reflection; and this
heightened awareness causes us to see that 49a is also paradoxical, something whose
truth is dubious. That it is justified we are reassured by the first stage of the process
(91 and its doxa). But 49a appears to assert a contradiction, so that, irrespective of the
facts of the river, we are reluctant to agree that 49a is true. So 49a, by both asserting
and denying the same thing, without qualification, is a paradox—which looks true
because of 91 and its doxa, and yet appears false because of its own countervailing
doxa, ‘it is impossible to step and not to step into the same river’. The trouble lies in
saying ‘step and do not step’, in the assertion of contradictory terms. So 49a imitates
the structure of 91, insofar as it is justified at the same time as it outrages a doxa
which the paradox causes us to see that we hold. It differs from 91 in being highly
reflective, and in deriving its paradoxical force not so much from its challenge to
common sense as from its outrage of the formal assumptions that we make when we
speak.
So far, then, the first level of paradox (91) provoked the second (49a, the
synthesis of 91 and its doxa). Is the paradox left standing? It is clear enough
from our reaction to 91 that we are uncomfortable when paradoxes survive, and,
indeed, that we use reflection to attempt to evade their toils. So does reflection give
us an escape route from 49a? Yes, it does—12. For here what was a paradox (‘we
both step and do not step into the same river’) is resolved by the judicious
qualification of the offending contradictories: we step into the same rivers, we do
not step into the same waters [we step into different waters]. By entering the
qualifiers, we arrive at a truth that is not a paradox—it gives us, rather, a complete
account of the weary stepping, qualified by the justifications offered at the first stage
(the banks remain the same; the waters change). So 12 is not paradoxical at all—it is
straightforwardly true.
So all three of the river fragments are in some way true, but for different reasons,
and at different stages of discourse. The connection between the three fragments that
I have offered shows how the fragments could be retained, and arranged in an
argumentative sequence, moving from pre-philosophical assumptions to a formal
grasp of the dangers of contradiction. From 91 to 49a we move from common sense
to reflection; at 49a we shift from material issues to formal considerations (away from
banks and water to the worry about contradiction); and at 12 we can account for and
resolve the formal difficulty in a non-paradoxical truth.
2. Other Paradoxical Arguments in Heraclitus

If all three river fragments are allowed into the Heraclitean corpus, then perhaps we
must concede that Heraclitus could argue—he was not just a man for the gnomic
utterance.6 But still, since the stages in the argument are fragmentary, then the order
of the argument remains unclear; so that the order in which I have presented it needs
defending.
The central question here is whether Heraclitus moves from the unqualified
paradox (49a) to the qualified truth (12; I have treated this as a ‘resolution’: but of
course that title is relative to its position in the argument), or the reverse, from the
qualified statement to the paradox. He could, as I have presented him, have argued
from the worry of contradiction to the relief of that worry by entering the qualifiers.
But he could equally well have performed the reverse manoeuvre familiar to the
sophists,7 and argued that because it is possible to step from the same banks into
different waters, then it is possible to step and not to step simpliciter.
This problem of interpretation is not restricted to the river fragments. For Hera-
clitus offers other paradoxes like 49a:
The road up and down is one and the same.8 (60: cf. 59, 62, 103, 125)
and qualified statements like 12:

The name of the bow is life, its work is death.9 (48: cf., among many examples, 9, 15, 88)
And, furthermore, some fragments contain both the paradoxical formula and its
qualified counterpart, thus joining these two moves of the arguments. They provide
support, therefore, for my contention that there is some argument in Heraclitus; but
they do not help us determine the direction of the inference. Thus:
Sea water is both the purest and the foulest: drinkable and life-preserving for fishes, undrink-
able and destructive for men. (61; cf. 67, 88)
Here the grammar of the sentence allows us to disjoin the paradox (‘sea water is the
purest and the foulest’) from the qualified version (‘drinkable for fishes, destructive
for men’). But we still do not know—and the fragments themselves do not tell us
6 7
Cf. here Barnes (1979), 63. Compare Plato, Euthydemus e.g. 293b ff.
8
Kirk (1954), 106, gives an account of the various ancient interpretations, most of which seek to
determine what is meant as the analogue of the road; so they assume that this fragment is a piece of
imagery, not to be taken literally. This assumption seems, both here and elsewhere, unwarranted. As Kirk
himself would agree, the fragment makes sense (or, in my view, paradox) without being treated as an image
for something else. Kahn (1973), 240, wants to have it both ways.
9
The remark relies on the ambiguity of the Greek word bios (differently accented); it can mean both
‘bow’ and ‘life’. This fragment does not commit Heraclitus to the view that names are pieces of objective
reality, although it is clear that he finds there to be a correspondence between words and the world. Cf. Kirk
(1954), 118, ‘Heraclitus was not surprised to find some real correspondence between the name and the
thing named . . .’
directly—whether Heraclitus wishes to assert the contradictories simpliciter (‘sea

water just is poison and food’10) or whether he wants to show how apparent
contradictories can be rid of their paradoxicality by entering the appropriate
qualifiers.
Suppose that, whatever the direction of the inference, Heraclitus wishes to use it to
give us some general truth. If he is asserting the contradictories simpliciter, then the
general truth will be that everything is contradictory and so indeterminate; we may
compare the position of Euthydemus in Plato’s Cratylus (386d) who claims that
everything is true of everything at the same time.11 And of course it is a long-standing
tradition that Heraclitus held an extreme theory of flux12 that bears a close resem-
blance to the Euthydeman thesis, since total flux asserts that the properties of things
are indeterminate over time. So perhaps Heraclitus wished the inference to go that
way.
There are two difficulties in saddling Heraclitus with flux or total indeterminacy.
In the first place, the surviving evidence does not support it. His cosmology has
internal problems, as we shall see; nonetheless, it does not present us with a flux-
ridden view of the world but rather one in which the elements change in a regular
cycle (cf. e.g. 31). It is of course begging the question to say that the river paradox is
an analogy for the flux of the world; and even if it were, the river is conceded to
remain stable and determinate (‘the same river’) even while the waters change.
Second, if things are indeterminate, whether over time or at a time, then nothing at
all can be asserted to be true, not even the theory itself (cf. Plato Theaetetus 181e ff.).
So the consequence, not to say the objective, of such theories of indeterminacy is
scepticism or nihilism. As Aristotle points out (Metaphysics 1005b19 ff.), if nothing is
true of anything, then no utterance can be meaningful; not even that utterance. So the
assertion of total indeterminacy amounts to the destruction of dialectic; and it is
refuted dialectically so that it is dialectically self-refuting. Now it is true that Hera-
clitus could have proposed a thesis of indeterminacy without recognizing its conse-
quences for the proposal itself, and without realizing that this commits him to self-
refutation as soon as he opens his mouth, so that this is only indirect evidence against
this interpretation of the paradoxical arguments. Nonetheless, it amounts to a prima
facie reason against interpreting them that way, since it is clear enough that Hera-
clitus wishes to assert, not deny, the possibility of dialectic, even if he allows that the
truth is generally inaccessible (the complexity of his position will be further investi-
gated in what follows).
If, on the other hand, Heraclitus infers the resolution from the paradox, then his
procedure makes good formal sense; that is, the sequence leaves us with something
sensible and consistent, not with an antinomy that threatens itself. This, I take it, is an
10
Food for structuralist interpretation here: cf., e.g. Vernant (1981), 1–15.
11
For interpretation of this passage, cf. my (1986), 132.
12
See here Wiggins (1982), 25; and Kirk ranged against Guthrie in Mourelatos (1974), 189–213.
advantage of such an interpretation. But then is it too small a mouse from the mountain?
Is the resolution of the paradoxes too trivial to explain both the many fragments that
contain these formulae and the richness of the tradition that sees Heraclitus as an
obscure and difficult thinker? I shall argue, first of all, that analysis of the formal
structure of both paradoxes and resolutions does give us some important and difficult
insights, so that this interpretation of the paradoxical arguments is to be preferred. Then
I shall show how deep the puzzling nature of Heraclitus’ remarks runs, and how this
feature of his thought relates to the destruction or the preservation of dialectic.
3. The Unity of Opposites and the Opposition of Unity

Heraclitus’ paradoxes are, in some sense, true. Whether we can or cannot step into it
twice, the river persists. Although rubbish is differently valued from different per-
spectives, it is still rubbish underneath (9). Sickness and health are parasitic on each
other, for we only value health in comparison to sickness (111). Night inevitably
follows day, and day night, unified in their constant sequence (57). In various
different ways, then, opposite properties are connected: perhaps because they inhere
in a single object, or because they form a single sequence in time, or because they are
somehow or other relative to each other. This truth about contradictory properties
may be expressed as a universal principle—the unity of opposites—and it is exem-
plified in paradoxes such as the river (49a), the road (60), or:
The road for the carding wheel is crooked and straight, it is one and the same.13 (59)
The posset stands still as it moves.14 (125)
The beginning and the end are common in the circumference of a circle.15 (103)
Any of these remarks rests upon the truth that opposites such as motion and rest,
crooked and straight, up and down are fundamentally connected, with the conse-
quence that they may be said to hold of the same subject at once. But that truth,
expressed in these unqualified examples, is paradoxical. The paradox comes about
because Heraclitus seems to insist that these contradictories just are a unity (that is,
13
I read the text given by Diels here. Diels reads gnapheiôi after Bernays; Kirk (1954), 97, retains
grapheôn from the manuscript on the grounds that the former reading refers to a machine not found in the
Greek world until the time of Archimedes. Marcovich (1967), 164, produces a counter-example from
Herodotus, l.92 (an instrument of torture so-called presumably by extension from the everyday object).
Kahn (1973), 191, suggests a complex machine wherein the wool must pass over and under a series of
rollers. But the sheer complexity of this process decreases the plausibility of the example—all we need is a
single carding roller that revolves (‘crooked’) and sends the wool out flat (‘straight’). Kahn (1973), 192
reads this as a resolved paradox to show ‘the functional unity of opposing tendencies within a purposeful
human activity’ but that is not what the fragment says. On its face the fragment is a paradox, unqualified.
14
For an extended justification of this reading of the paradox, see Chapter 3.
15
Note here the echoes, via xunon, with frs. 1, 2, and 80. If this fragment ends after kuklou, then the air
of paradox increases, cf. Kirk (1954), 113; the mention of the circumference points the way towards the
appropriate qualifiers, but does not in itself resolve the paradox.
the road up and down is one, simpliciter), without qualification. So the principle of
the unity of opposites is the basis for the paradoxes.
Conversely, the qualified statements such as:
Sickness makes health sweet and good, hunger satiety, toil rest. (111)
Donkeys prefer rubbish to gold.16 (9)
The same thing [is in us?] living and dead, waking and sleeping, young and old: for these things
changing are those, and those changing are these.17 (88)
need a different principle to explain them. For here the opposites that constitute a
unity are given vital qualifiers. Thus rubbish is good for donkeys, bad for men; this is
young now, old next year, this is hard work compared to sitting in the sun, easy
compared to rolling a boulder up a hill. The model for all these qualified remarks is
the sea water fragment (61). Here the general principle, then, is that any unity
composed of opposites is to be explained in terms of qualified opposites. Let us call
this principle the opposition of unity, where the nature and qualifications of the
opposition are what need to be stressed.
The opposition of unity, then, is the counterpart of the unity of opposites. One
asserts the fundamental connectedness of the opposites; the other stresses their
essential difference. Unity collapses opposites into each other; opposition maintains
their continued difference by showing how they are to be kept separate. And
Heraclitus gives us plenty of examples of both types of statement—even, as in the
case of 61, within the same fragment.
Heraclitus does not, however, just give us instances of the complementary prin-
ciples. He also offers us the principles stated as generalizable truths.
It is fairly uncontroversial to point to his claims for the unity of opposites:
Listening not to me but to the account it is wise to agree that all is one.18 (50)
‘The account’ (ho logos) has been variously interpreted, and fuelled many contro-
versies. It could mean some guiding principle of the cosmos, some actual real force
inherent in the world. Or, more moderately, the point might be that the cosmos
exhibits lawlike regularity, which we can describe rationally in ‘an account which
corresponds to the way things are in the world’.19 Certainly, we should not
16
NB there is a ratio missing here, men : gold. Cf. here Frankel in Mourelatos (1974), 214–28.
17
The textual difficulty here is t’eni or g’eni. The significance of the words is dubious, so that we should
guard against reading it as a strong statement of inherent properties. The point, at least, must be that
opposites are united in their temporal continuity.
18
NB the point made by Marcovich (1967), 115, that the difference in tense between akousantas and
homologein suggests some kind of inference made by the hearers. It is not clear, however, what inference it
is—from hearing the account to the necessity of agreeing? Is that an inference at all? The point would be
better taken if Heraclitus suggests that his audience infers the conclusion (‘all is one’) from the fragmented
parts of his work.
19
Here Marcovich, with shaky logic, claims that the Logos (sic) is possibly to be thought of as corporeal; he
attacks, though, the view that identifies this logos with fire, on the grounds that these are in different divisions
overlook—and perhaps we should not look beyond—the explicit association of logos

with speech and language.20 So what Heraclitus is offering us is not a piecemeal
collection of words (epea), but an interconnected, ‘common account’:
Of this account which is always men are always ignorant, both before they hear it and when
they first hear it; while everything happens according to this account, they are like people
inexperienced in everything, experiencing both the words and deeds which I go through, when
I distinguish each thing according to its nature and say how it is. But it escapes the notice of
everyone else what they do when they are awake, just as it escapes their notice what they forget
while they are asleep.21 (1)
This account being common, most men live as if they have private understanding. (2)
It is irresistible to conclude, from the conjunction of 50, 1, and 2 (invited by their

linguistic connections), that the common account he offers is some kind of principle
of unity exhibited in the world. What is more, not only is Heraclitus laying claim to
some general truth (‘all is one’), but also he says he is doing so; so that the
generalization is explicit and self-conscious. Add to that fragments such as:
They do not know how differing it agrees with itself: it is a back-turning connection like a bow
or a lyre.22 (51)
To god everything is beautiful and good and just; men take some things as just, some as unjust.
(102, cf. 23)
of Heraclitus’ philosophical enterprise. Kirk initially (1954, 43) explains logos as the formula that ‘underlay the
working of the sum of things’; but he then comes to reify it, e.g. 395, as ‘something independent’.
20
Cf. Barnes (1979), 59, ‘the noun logos picks up, in an ordinary and metaphysically unexciting way, the
verb legei; it is wasted labour to seek Heraclitus’ secret in the sense of logos’. Nussbaum (1972), 1–16,
emphasizes the connectedness of logos as opposed to the disparateness of Homeric epea.
21
The structural ambivalence of this fragment begins with the janus-faced aei in the first clause, which
may be taken either with logos or with the inexperience of man, or both, as here. The fragment once again
depends upon a series of antitheses, and on an ambiguity between the personal and the impersonal pronoun.
pantôn may refer to all men, or to all things, or to both, as I interpret it here, cf. Kahn (1973), 118 ff. The
closing sentence must, for grammatical coherence, be construed as I translate it, not as Kirk (1954), 33: ‘but
the rest of men fail to notice what they do after they wake up just as they forget what they do when asleep’,
which improbably takes epilanthanontai as parallel to lanthanei. On its own, this is a small point; its
significance will emerge further below in reflection on the careful structure of Heraclitus’ antitheses.
22
This is perhaps the most vexed of all the Heraclitean fragments. If, following Plato, Symposium 187a,
we read sumpheretai for homologeei, then the fragment echoes fr. 10 rather than fr. 50. This point does not
seriously affect the sense. The big issue, however, is whether we should read palintropos, ‘backward-
turning’, or palintonos, ‘backward-stretching’. If we read the former there may be a connection with
Parmenides, DK 28B6.9, where Heraclitus may be the Eleatic’s target. There is, as Kirk points out, ‘nothing
to choose between the two variants on the grounds of ancient testimony’, and we should not rest some
dispute between Heraclitus and Parmenides on such slim evidence. Nonetheless the reappearance of tropai
in fr. 31 to describe the changings of fire encourages me to believe that palintropos was the original here.
Perhaps also the dual function of a tropaion, a trophy—to record a past victory and ward off the enemy in
the future—may resonate here (my thanks to Rose Mary Sheldon). The epithet, whichever reading we take,
is clearly meant to point to the tension of the bow or the lyre (whether from the tension of the string or its
oscillation when it is plucked). Other passages point to the coincidence of the static and dynamic in
something, e.g. fr. 125. This fragment, like many others, opens with a paradox, ‘differing it agrees with
itself ’, and then resolves it in the second clause.
and it becomes an easy matter to infer that the common principle is just the unity of
opposites.
At this stage, of course, those who associate Heraclitus with Euthydemus, or with
Aristotle’s straw men in Metaphysics IV, point out triumphantly that this very
principle of the unity of opposites is what they find in Heraclitus: the unity of
opposites simpliciter. But this is to tell only half the tale. For Heraclitus, the principle
of unity of opposites is complemented by a reverse principle. Thus he is insistent, in
those fragments that talk about general principles, just as much upon the opposition
of unity.
Consider, for example, how Heraclitus intends to show mortal men the way in
which the account is general (1): he will do this ‘when I distinguish each thing
according to its nature’. But this discriminatory operation focuses upon opposition;
on difference, not on unity. Likewise:
The wise is one, to know the reason which rules everything through everything.23 (41)
emphasizes not so much unity as opposition (‘everything through everything’). Or in:

None of those whose accounts I have heard has arrived at this, to know that the wise is different
from everything.24 (108)
Heraclitus points us towards the difference of things, not their sameness. Indeed, in
many fragments where he appears to be talking about the principle of unity, he
emphasizes at the same time the converse principle of opposition:
23
The commentators have concentrated here on the question ‘whose reason?’; and they have inevitably
come up with different answers (cp. Kahn (1973), 171; Marcovich (1967), 452) depending on whether they
see this as an account of human aspiration or divine truth. I shall argue in what follows that to see these as
exclusive misses the point of the god : man ratio. For the reading of this fragment we need to grasp the force
of the preposition dia. Kirk gives it locative sense: ‘as in “I steer a boat through the narrows”; the whole
course of each separate thing is a result of steering’ (390). I find it hard to see how this locative ‘through’
can explain panta dia pantôn which gets its force, surely, from the ‘apart’ connections embedded in the
sumpheromenon/diapheromenon contrast of fr. 10, and thence from the unity of opposites thesis as a
whole. The discrimination idea in diapherein is perhaps derivative from the locative sense of dia by means
of a contrast between the thing located and the location (compare Zeno’s argument, DK 29B4). But then,
panta dia pantôn deliberately flouts such a contrast because the location and the thing located are described
by the same pronoun. The point must be to suggest that everything is distinct, and to do so by means of a
startling expression. Cf. Euripides, Heracles 76, where the connotations of deceit are relevant to my point
(my thanks to Kevin Lee for the reference). Cf., also, of course, Parmenides, DK 28B1.32.
24
Two issues here: (a) the reference of pantôn and (b) the sense of cechôrismenon. (a) pantôn could refer
to all things or all men. If we take the latter alternative, we might have here a renewed attack on the
cognitive ability of mankind. If we take the former, then this fragment might contain, as some commen-
tators have thought, some further evidence of the transcendence of the divine. Or, if we include both men
and things in the reference, then the connotations of both readings are preserved; with this proviso,
I translate ‘everything’. (b) cechôrismenon is generally translated as ‘separated from’ (Kirk) or ‘set apart
from’ (Kahn). Marcovich points out, however, that in Herodotus this verb has the sense of ‘be different
from’ in the passive; compare Herodotus I.140; I.172; IV.28 (LSJ s.v. II) and IIl.20; V.62 cited by Marcovich.
In each of these cases chôrizesthai is associated with the adjective (pronoun) allos; ‘is different’, then, is
justifiable here.
It is necessary to know that war is common, and justice strife, and that everything happens
according to strife and necessity. (80)
Collections: wholes and not wholes, agreeing, disagreeing, singing together, singing apart, one
from all and all from one. (10)
So even where Heraclitus lays stress upon the unity of opposites, he relies upon their
opposition for the principle to make sense. I shall call this the bivalence of the unity of
the opposites and the opposition of unity. But if Heraclitus thus emphasizes the
opposition of unity, then he is not an indeterminacy theorist, any more than he is a
strong monist. And in that case, we may prefer the reading of the paradoxical
arguments that infers the resolution from the paradox. And the accusation that
this gives us mere triviality may be rebutted: for the meshing together of the
paradoxical arguments with these general principles gives us genuine philosophical
progress—in two particular areas.
4. Contradiction and Resolution

Let us return to Aristotle. In Metaphysics IV he is defending the view that there are
some principles of metaphysics that are necessary, over-arching, and unhypothesized
(1005b11 ff.). Then he embarks, under this rubric, on a defence of the law of non-
contradiction (LNC), and its converse, the law of the excluded middle (LEM:
cf. 1008a2 ff.).
For it is impossible that the same (property) should belong and not belong to the same thing in
the same respect (and with as many other qualifications as we have already discussed as are
needed to relieve the logical difficulties); this is the most fundamental of all the principles. For
it is impossible for anyone to believe that the same thing is both true and not true, as some
suppose Heraclitus to have said. (1005b19–25)
Aristotle defends this view in the chapters that follow.

However, if LNC is the most fundamental principle of all, then he cannot dem-
onstrate it by inferring it from some other principle. Rather, he offers a dialectical
defence: the defence by refutation. This consists in showing how those (perhaps
Heraclitus; possibly Protagoras) who deny LNC cannot utter a word in defence of
their position; they cannot even state their position at all. For any utterance depends
upon LNC if it is to be determinate, let alone coherent. But if LNC is false (and if, as
Aristotle assumes, there is correspondence between what is the case and what is said
to be the case25), then nothing determinate is the case, nor can anything be men-
tioned, including the denial of LNC. So, more positively, as soon as the denier of LNC
opens his mouth, he is refuted; and by means of that refutation the law derives its
25
Aristotle, like Heraclitus, is indifferent to whether we see non-contradiction as a principle about
predicates (hupolambanein 1005b24) or properties (huparchein 1005b19). He assumes correspondence
between the two, at least at this point in the Metaphysics.
dialectical defence (1006a 12, etc.). To put Aristotle’s point another way, we only
become aware of the truth of LNC when we need to refute one who denies it.
Now Aristotle is evidently cagey in his attribution of such a position to Heraclitus.
But if we suppose that Heraclitus infers his paradoxes from the qualified statements
of opposites, then he is doing just what Aristotle shows to be self-refuting; that is, if
he asserts an indeterminate world, and correspondingly the indeterminacy of truth,
then as soon as he utters, his utterance by being determinate will be a counter-
example to the thesis itself. Thus Aristotle’s evidence would seem to confirm the view
that Heraclitus is advancing a vertiginous theory of indeterminacy by inferring the
paradoxes from the unqualified attribution of opposites to an object.
I have argued, however, that on the grounds of sense over nonsense, the alternative
view of Heraclitus’ inference is to be preferred, provided that we can see it as having
some philosophical function. And Aristotle shows us the way. If Heraclitus infers the
resolutions from the paradoxes, he achieves exactly the same effect as Aristotle does;
for by moving away from paradox and resolving the difficulties of the river and the
road, he shows how LNC works: that is, by inserting the appropriate qualifiers to
avoid what Aristotle calls ‘the logical difficulties’.
The clue lies in the psychological impact of paradox.26 After all, when we are told
that we cannot step into the same river twice, if we enter into the spirit of the game at
all, we will respond with the appropriate doxa. And so in general the effect of paradox
is provocative; paradoxes make us think, and by doing so they direct our thought in
particular determinate directions. So Heraclitus’ paradoxes startle and outrage
because they violate LNC; and the move to their resolution is motivated by our
dislike of staying strung up by paradox. Now I have argued that the paradoxes are
examples of a stark unity of opposites thesis; and that thesis on its own, as we have
seen, amounts to the denial of LNC. At the same time, however, as asserting the unity
of opposites, Heraclitus asserts the opposition of unity; and that principle is LNC. For
the opposition of unity says that opposites are a unity only in a qualified sense, and
the examples of the principle that Heraclitus gives us show how multiform are the
qualifiers that may be entered: in respect (48, 12, possibly 58), in time (88, 126), in
sense (23, 15), or relative to the perceiver (61, 9, 13, possibly 58, 67).
Now if Heraclitus’ purpose in advancing his paradoxes was to show us how LNC
works, then the business is hardly trivial. It operates at two distinct levels. First, at the
object level, the paradoxes provide us with the impetus to understand the workings
and violations of LNC, and they activate our awareness of its importance. Second, at
the theoretical level, the twin principles of the unity of opposites and the opposition
of unity give us the axiom itself: the unity of opposites shows us the necessary
connection between contradictories, and the opposition of unity shows how they
26
See my (1982).
must be qualified to avoid paradox. Here is the first area of genuine philosophical
progress without recourse to the absurdities of the indeterminacy thesis.
5. One Under Many: Individuation

When Heraclitus produces paradoxes about the objects of the phenomenal world, he
offers variants of what could later be characterized as the problem of the one and the
many (cf. e.g. 10, 41, 50). Of course, Heraclitus may have been concerned with an old
cosmological chestnut about the one and the many—how come the changing uni-
verse is materially continuous (cf. Aristotle’s diagnosis of the early thinkers, Meta-
physics 983b6 ff.)? Certainly, there are times when Heraclitus is answering—if in a
somewhat backhanded way—the questions about the stuff of the world, and its
changes, that interested the Milesians (30, 31, but see below). And so it has often
been thought that when he talks about rivers he is offering an image of flux to answer,
or indeed to confound, this same cosmological problem of stability under change.
Now I have already argued, first, that in the paradoxes Heraclitus is not so much
concerned with flux and indeterminacy as their resolution. Second, we have no
evidence to tell us that the river fragments are images for something else, particularly
if they have some philosophical impact on their own. We have already seen that they
do have such an impact in their treatment of contradiction. They also make an
important point about the sensible individuals that underlie the opposite properties.
Consider the structure of some of the paradoxes (the river; the carding wheel, 59;
the road, 60; the circle, 103; the posset, 125). In each case he shows us a single object,
characterized by contradictory properties—a road up and down, a river changing and
remaining the same, a posset moving and standing still. While the surprise of
the paradox makes us notice the imminent contradictions here, it also shows how
the subject of these predicates remains somehow fixed and stable. In particular, the
wording of the paradoxes suggests that these objects are units, ones, single items in
contrast to the plurality of their properties—each is ‘one under many’.27 Hence the
road and the journey of the carding wheel are one and the same (59, 60), and the river
remains the same for our plurality of steppings (91). That is, these remarks rest
upon a contrast between the one subject and the many predicates—or the individual
object, with its plurality of properties. Now the ‘many’ problem—the danger of
contradiction—is solved, as I have argued, by entering the qualifiers required to
satisfy LNC. However, LNC alone is not enough to secure normal discourse; and this
too is an insight that Heraclitus’ paradoxes give us.
It is a common assumption that when a sentence such as ‘the sea is poisonous’ is
true, it tells us of a property inhering in an object. That is, the object (the sea)
underlies its properties (being poisonous, being life-giving) and is not merely
27
I am grateful to N. C. Denyer for this expression in this context.
reducible to them. Individuals, that is, are not merely bundles of properties, and
indeterminate otherwise, but they are rather the stable substrate of their properties.
Reflection upon the problems of contradiction tends to confirm this assumption.
Any pair of contradictories, predicated of a single subject, must be qualified to avoid
the dangers of contradiction. But further, the qualification of the predicates relies
upon the subject. The river is the same in respect of its banks because it is this river;
and different in respect of its waters because a river is what it is. That is, the
proposition ‘the river is the same in respect of its banks’ relates the subject to the
qualifiers through the term ‘same’. So in general, paradoxes of the subject/contra-
diction type, such as 60, 61, or 125, call attention to the self-identity of the subject no
less than to the opposite predicates. And, conversely, paradoxes where there are two
subjects and a single predicate, such as ‘day and night are one’ embedded in 57, raise
the question of how two separate items can be said to be the same; so that it asks
about the self-identity of these subjects from the perspective of their presumptive
difference from each other.
So Heraclitus’ unity of opposites thesis involves opposition in terms of the
predicates or properties involved, and unity in terms of the underlying continuous
subject or object. To challenge the river is, on the one hand, to suggest that the river
embodies an impossible contradiction; and on the other to suggest that there is no
river at all. Our doxa response to the challenge resists the contradiction, on the one
hand; and the disintegration of the subject on the other: ‘Of course I can step into the
same river’ versus ‘Of course it’s the same river’. Consequently, just as the interplay
between paradox and doxa makes us aware of the workings of LNC dialectically, so
also it activates our intuition that these properties do have underlying individual
objects. So the unity of opposites suggests that there are individuals underlying
properties (not just bundles of properties); and it does so by showing how the relation
between the predicate and its qualifiers needs the primary term, the subject, for its
effective expression.
Begin with the road. If someone confronts us with the paradox: ‘the road up and
down is one and the same’, the resolution, which comes from our common-sense
assumptions, is that ‘the road up from Larisa to Athens is the same as the road down
from Athens to Larisa’.28 The resolution is provoked by the paradox, which chal-
lenges the self-identity of this one road; and the resolution reassures us of its defence
against disintegration. Like 103, then, the paradox of the road asks a question about
the self-identity of an individual item at a time. By contrast, paradoxes such as the
moving posset (125) or the river ask about the self-identity of the object over time;
and again the doxa response tells us that such objects do indeed persist as the same
individuals. And the fact that this awareness of the individual objects is brought
about in us by means of paradox and doxa urges us to think about how we know that
28
This issue is developed by Aristotle, cf. e.g. Physics 202a20 ff., and Hartman’s comments (1977), 78.
this is an individual. So the ‘one’ arm of the paradoxes focuses not upon the logic of
contradiction, but upon the metaphysical question: ‘Are these items individuals?’
When Aristotle reports Heraclitus’ approach to LNC, as we saw, he also reports the
view that Heraclitus was violating LNC. The interpretation I have offered suggests
that in part this is true, but that the function of the violation of LNC was not to assert
logical chaos, but rather to reveal the crucial role of LNC in logical order. When Plato
reports Heraclitus’ account of the items and properties in the real world, he says that
Heraclitus was committed to a theory of total flux (e.g. Theaetetus 152 ff.; cf. Cratylus
411, 439; and Aristotle’s comments, Metaphysics 987a30 ff.). Such a theory, as Plato
points out, commits Heraclitus to indeterminacy over time; and this has the particu-
lar effect of denying that there are individuals (‘nothing is a something [ti]’,
cf. Timaeus 49d, Cratylus 439d, Theaetetus 157). Once again, the later comment
is half-right; in his paradoxes Heraclitus does indeed threaten the identity of
the individual, either by synchronic indeterminacy, or by flux over time, or by the
epistemological argument that what we know is utterly private. However, once the
dialectic of paradox is correctly understood, this attack upon the individual is seen
to provoke a corresponding defence, that there are, and must be, individuals in the
real world. Once again, the axiom is demonstrated by refutation of its converse,
elenctically.
6. Up a Level: Trouble in General Theoretical Remarks

The story so far is this. Heraclitus offers us a series of examples where an individual
has opposite properties (any pair of opposites will do—up/down, good/bad, etc.).
These examples may be classified as one (the individual subject) versus many (the
opposite properties). And the contrast between the one and the many is echoed
in the contrast between the understanding the paradoxes give us, on the one hand, of
the metaphysics of individuals, and on the other, of the logic of contradiction. That
understanding is hard won through the play of paradox and common sense in the
paradoxical arguments.
So some truths about the object world are here explained by means of higher-level
principles. The unity of opposites and the opposition of unity systematize the ‘many’;
and the attention Heraclitus pays to questions of sameness and difference, motion
and rest affect our interpretation of the ‘one’. So to the extent that this is systematic,
rather than intuitive, the dominant feature is the pair (unity of opposites, opposition
of unity). Now I have argued that Heraclitus gives us examples of his pair of
principles (49a, 12, 60, 61, etc.); that he gives us the principles stated as general
truths (e.g. 10, 53, 88); and that he talks about giving the principles (1, 50). That is, his
discourse occurs both at the object level (talking about the world, the rivers, the
roads, and the possets) and at the higher, theoretical level. Further, it is apparent that
he supposes that the talk corresponds directly to the way things are—there is no hint
that what we say could be detached from the facts of the matter.29 So there is
continuity between the objects, talk about the objects, the general principles of that
talk, and the talk about the general principles. But this means that the general
principles are themselves subject to the general principles of discourse—indeed
they are self-referential. And this feature of Heraclitus’ account has tricky
consequences.
The connection between these theoretical fragments and the first-level paradoxes
begins to develop in a trio of fragments about night and day:
Hesiod was the teacher of very many: they know that he knew many things, he who did not
know day and night, for they are one.30 (57)
If there were no sun, then it would be night, so far as depends on the other stars.31 (99)
Hesiod counted some days as good, others as bad, because he did not know that the nature of
every day is the same.32 (106)
As it is generally construed, 57 (with 106) attacks Hesiod’s reputation for knowledge

(cf. 40).33 Hesiod is charged with a fundamental error: that of failing to see the unity
of a particular pair of opposites, day and night. So embedded in 57 is a first-level
paradox just like the river or the road, ‘day and night are one’. This is perhaps true if
we consider their temporal continuity; or, perhaps as in 106, their uniformity; or
perhaps because hêmera and euphronê mean the same thing (‘the gentle, kindly one’).
But Hesiod thinks, instead, that night and day are distinct; and herein lies his
mistake.
Yet this unambiguous interpretation of 57 ignores its complexity. The fragment
resounds with cognitive terminology (didaskalos, epistantai, eidenai, ouk eginôsken),
to the effect that after all, Hesiod does know many things at the same time as he is
mistaken about the nature of day and night. What is more, inspection of 99 should
give us pause. For that fragment, so far from asserting the unity of night and day,
appears to insist upon their difference (hence the counterfactual, if there were no sun).
29
See Mourelatos (1973).
30
Fr. 57 must be associated, of course, with 40 and 56 in its mention of ancient sages. My translation
sees 10 preserve, rather than avoid, the paradoxical overtones of this fragment. To avoid this, translators
have attempted to modify one or other of the verbs of cognition: epistantai is rendered ‘feel sure that’ by
Kirk, thus robbing it of its veridical content, and the danger of contradiction between eidenai and ouk
eginôsken. Bollack and Wismann (1972), 196 ff. have a better sense of paradox, and render ‘ils savent’. The
justification for reading 57 as a paradox comes also from external evidence. In 40, the sages are said to have
been taught nothing by their polymathy. If teaching is transitive, and likewise its lack, how can Hesiod be
described here as a teacher at all? That is, there is at the very least a contradiction between frs. 40 and 57.
31
It is irrelevant to my point here whether we keep the qualifier ‘for the other stars’. E.g., Marcovich and
Bollack and Wismann retain it, Kirk and Kahn delete it. Kirk (1954), 162 claims that the deletion affects the
sense; but surely the main burden of the fragment is to say that, since day and night are continuous, if there
were no sun, the night would last all the time (even if the other stars kept shining); the parenthesis,
obviously, is not crucial. Nor does its inclusion imply that the fragment has cosmological significance, pace
Kirk, and vide Marcovich (1967), 325 ff.
32
This may or may not be a separate fragment in its own right; cf. Kahn (1973), 110 and 308 n. 73.
33
Cf. here Kirk (1954), 155, who cites others.
So Heraclitus acknowledges Hesiod’s truth, that day and night are different, no
less than he denies it by asserting that they are one. But the consequence of that is
to make the structure of 57 even more paradoxical. Day and night are different
(justified doxa); day and night are the same (justified paradox). So day and night are
both the same and different (paradox about the object). Therefore, Hesiod both knows
and does not know (paradox about what we know or say about the object). These are
followed by a pair of resolutions: day and night are different in one sense, the same in
another; so that Hesiod knows one thing, but fails to know the other.
Why is the fragment directed against Hesiod in particular? Once again, we need to
focus upon the cognitive terminology. The doxa corresponding to the paradox ‘day
and night are one’ shows that everyone believes that day and night are different. But
not everyone is given credit for knowledge. Hesiod, however, is allowed to know, not
just to speak. So presumably his knowledge has some kind of justification. And that
justification must surely be the opposition of unity, the general principle that shows
the diversity of things. It follows that Hesiod both knows (by virtue of the opposition
of unity) and does not know (by virtue of the unity of opposites). So his knowledge
and his ignorance are not a mere echo of the original contrast (day and night are
different; day and night are one), but a new contrast altogether. For while the original
doxa/paradox pair are explained by the data of the phenomenal world, the know-
ledge/ignorance pair are justified by the general principles of the unity of opposites
and the opposition of unity. When it comes to knowledge, that is, as opposed to the
mere assertions that ordinary people might make about night and day, we need to
look at the explanation; and this will be found in the general principles that govern
the claim to know.
Thus the Hesiod fragment echoes the argumentative structure of the river frag-
ments. However, there is now a further dimension. While the river fragments
operated by directly provoking our own doxai in response, the Hesiod fragment
contains a paradox—an argument about objects, together with a parallel argument
about the knowledge of objects. So, by offering a further paradox about what Hesiod
does and does not know, Heraclitus shows how the paradoxes connected with
contradiction may occur both in the object and at the higher, explanatory level.
This has an interesting effect. Hesiod’s knowledge is explained by the opposition of
unity, and his ignorance by the unity of opposites; but then the paradox that he both
knows and does not know is itself governed by the unity of opposites and the
opposition of unity, just because it involves the simultaneous assertion of contra-
dictories. The twin principles of the unity of opposites and the opposition of unity,
therefore, apply no less to the pair (knowledge, ignorance) than to the ordinary
contradictions of the object world.
Now the theoretical fragments, both when they discuss general principles about
objects and their properties and when they discuss epistemology, are equally gov-
erned by the unity of opposites and the opposition of unity.
In fragments 1 and 2, Heraclitus deploys the terms ‘common’ and ‘private’ to

describe our cognitive attitudes. Men are ignorant because they do not understand
the principle that Heraclitus will tell them. Their error, as 2 tells us, lies in the
fact that they concentrate on their own private view, while they ignore the common
view that Heraclitus will expound. So they are as if asleep, and they are forgetful,
where they should be awake and remember.
To those who are awake, the order is one and common, [each of the sleepers turns aside] into
the private.34 (89)
Now in 57, Heraclitus reflected upon a particular item of knowledge—the unity and
opposition of night and day. Here, by contrast, the objects of knowledge are the
general principles he expounds; and the fragments reflect upon our ignorance of
those. So while 1 and 2 echo 57 in deploring mortal ignorance, they differ from 57 in
generality; for here we are directly concerned with knowledge of theory, of the
higher-order principles governing truth in the objective world. So in 57 Hesiod failed
to understand a particular instance of the unity of opposites; in 1 Heraclitus is talking
about understanding the unity of opposites itself.
I have argued, however, that if we compare 57 with 99, then the criticism of Hesiod
comes out ambivalent. Although he was at fault for not knowing that night and day
are one, he still knew (very many things including that) night and day are different. Is
there any such ambivalence in the general fragments?
Fragments 1 and 2 are apparently programmatic. They also contain a chain of
allusions that connect them with other remarks. These connections are made in two
separate aspects: the objective and the cognitive. The objective aspect characterizes
the objects of the real world and the principles that govern them; the cognitive aspect
deals with how men are disposed towards these objective truths. Yet while different
fragments have different emphases, it is clear enough that while Heraclitus castigates
ordinary mortals for ignorance, he does not anticipate any real cognitive gap between
knowledge and what is known. That is, he does not hint that there may be no fit at all
between the objective world and our possible knowledge of it; on the contrary, he
appears to assume correspondence between the truths that can be known, and the
subject of those truths—the objects and the principles of the phenomenal world
(cf. here 2 with 89, 115; 115 with 45 and 36; and the epistemological tone of a
fragment apparently about physics, 51).35 Perhaps as a consequence of this, both the
objective and the cognitive aspects of Heraclitus’ theory have the same features of
unity and opposition as the first-level objects such as the rivers and the roads.
34
It is debated whether this fragment contains original Heraclitean material, cf. Kirk (1954), 64; Bollack
and Wismann (1972), 262. However, as Kahn remarks, ‘I see no reason to doubt that Plutarch is rendering
Heraclitus’ thought correctly’ (1973), 104. And see also Vlastos’ defence of the fragment (1955), 344 ff.
35
Cf. here Nussbaum (1972) and Mourelatos (1973).
We may recall once again what Heraclitus said and implied about Hesiod. Because
Hesiod was both right and wrong, then he himself had the contradictory properties
(in this case knowledge and ignorance) typical of the objective world. But if Hesiod
was both right and wrong, then the programme of 1 cannot be saying that to get it
right, we have to grasp only the common view that consists in understanding the
unity among opposites. For, if the common view consists in the unity of opposites
then it consists no less, if my argument above is correct, in the opposition of unity.
The opposition of unity relies upon differences, so that when Heraclitus wishes to
explain it to us he must differentiate (‘distinguish each thing as it is’), and show how
one is from all and all from one (10; compare 88, 62). But then the differentiated
view, we may suppose, comes from the disconnected experience of men, from the
private insight of Hesiod, perhaps, that day and night are different.
So, when we come to look at the description of mortal understanding that
Heraclitus offers us, we should be wary of the easy inference that he is giving us, in
the common view and the principle of the unity of opposites, a privileged insight into
the truth. In the first place, how can we comprehend something we shall not
comprehend, any more than we can be castigated for failing to understand it before
it is presented to us? And how can Heraclitus tell us about it, if we cannot yet
understand it? How can he even tell us that we do not understand, and expect us to
understand? Or, in reverse, how can he tell us that we do not understand, when we
are indeed experienced in the things he tells us about—our woeful ignorance is no
less familiar with our own words and deeds, with the things we do when we are awake
and we forget when we are asleep. Indeed, if the account includes everything, it
includes us too—our ignorance is of what we know best, ourselves. And that offers at
least some kind of paradox:
I sought out myself.36 (101)
So because the unity of opposites is essentially connected to its converse, the

opposition of unity, the attack upon mortal cognition is tied to the admission that
we need the private view no less than the common. That is, privacy and diversity are
no less essential than unity and the common: the unifying view makes no sense
without its converse, the discriminating view. Consequently, the unity of opposites/
opposition of unity applies also in the epistemological sphere, as the contrast between
knowledge and ignorance, the common and the private, which are like waking and
sleeping, living and dying.
Most men do not understand [phronein] such things as they come across, nor do they
recognize what they have learned, but they think (imagine?) for themselves.37 (17)
36
Kahn (1973), 309 n. 84, rightly castigates the attempts of some commentators to modify the paradox
here.
37
Kahn (1973), 102 ff., has an extended treatment of the literary cross-references here. Cf. also frs. 27
and 28a. It begs the question to suppose that dokein is not veridical here, so the translation ‘imagine’ seems
to me to be loaded.
They differ from that with which they continuously associate, and what they come across daily,
this seems alien to them. (72)
Thinking [phronein] is common to all.38 (113)
In this collection of fragments, the attack on mortal men is combined with the
admission that they do have some understanding derived from their private view.
The notion of privacy is well adapted to the comparison with sleeping and waking,
since, as 1 points out, our experiences when we are asleep are private, and discon-
tinuous with our experiences when we are awake (cf. 26, 34, 73). And yet somehow
sleeping and waking, living and dying, knowledge and ignorance are crucially
connected:
Immortals mortal, mortals immortals, living the others’ death, dead in the others’ life.39
(62; cf. 77)
Then indeed to stand up and be waking guardians of the living and the dead. (63)
Those asleep are workers and contributors to what goes on in the world. (75; cf. 73)
The same . . . living and dying, the awake and the sleeping, and young and old; for these things
changing are those, and those things changing are these. (88)
For privacy contributes to what is common, offering the hopeless paradox of self-
knowledge. Compare the triad:
The soul’s reason increases itself.40 (115)
Seeking the limits of soul you would not find them, going on every road; for it has such a deep
reason. (45; cf. 101)
It is for all men to know themselves and to be sensible. (116; cf. 113)
This apparently inconsistent group suggests that while the soul has by nature
common reason, it reaches understanding only by the private activity of introspec-
tion. And that reveals the inexhaustible capacity of soul, so that it is an activity that
38
Kahn (1973), 119, takes this as some kind of expression of panpsychism, on the grounds that
otherwise it is merely a redundant restatement of 116 (though see here Vlastos (1955), 347). Fr. 116,
however, contains the vital modification that this common thought is the private exercise of knowing
ourselves. Fr. 113 then is merely a bland statement of the common nature of thinking, strikingly
contradicted e.g. by fr. 2.
39
Kahn (1973), 217 ff., points to the literary density of this fragment, and shows how it suggests not
only a comparison of the eternity of immortals with human mortality but also the reverse, the paradoxical
notion that immortals are mortal as we are immortal. Cf. here Nussbaum (1972) on the significance of this
fragment in the analysis of soul.
40
In this and the following fragment I have translated logos as ‘reason’, because Heraclitus appears to be
reflecting upon the ability of soul rather than some external account or measure of it. So here we have a
different conception of logos from 31, or 1. This could be mere imprecision on Heraclitus’ part; or it could
be deliberate word-play. On the latter view (the view of Kahn), the fragments are designed to enlarge our
understanding of the term logos; this self-referential procedure is, as it were, expansive. With this view
I would agree, since it can hardly be that the extensive word-play in Heraclitus is accidental. I would argue,
further, that the expansion results in our becoming convinced of a view, in the end, which is the reverse of
the view with which we started.
never has a determinate end. This may be connected with Heraclitus’ ‘paradox of
inquiry’:
Unless he hopes for the unhoped-for, he will not find it, since it is not to be discovered, and it is
trackless.41 (18)
So it is clear that here, no less than in the fragments directly concerned with the unity
of opposites, Heraclitus wants to show how the private view is connected to the
common view. That is, while the common view expresses unity, the private view
expresses opposition; and as with the unity of opposites in the phenomenal world,
neither aspect makes sense without the other. It follows from this that unity and
opposition are properties not just of real objects like rivers or roads, but also of minds
(Hesiod’s, for example) and of abstract, general truths, whether about objects or
about minds. The principles of the unity of opposites and the opposition of unity are
completely generalizable.
Consider the following series:
a. The river is both one and many (same and different). (from 49a)
b. The river is one in respect of its banks and many in respect of its waters.
(from 12)
c. We both know and do not know that the river is one and many. (cf. 57 and 99)
d. We know that the river is one and many in respect of our private view; we fail to
know it in respect of our view of the common. (from 1)
e. The unity of opposites and the opposition of unity are the same and not the
same (all is one and one is all). (from 10, 80, 53)
f. The unity of opposites and the opposition of unity are the same by virtue of
bivalence; they are different by virtue of their contradiction of each other. (from
67, 8, 88, etc.)
g. We both know and do not know the unity of opposites and the opposition of
unity. (from 17, 72, 18)
h. We know the unity of opposites in respect of our private view; we do not know
it by default from the common. (from 1, 2, 51, 113)
Each stage in this series is generated by a further application of the unity of opposites
and the opposition of unity, depending on whether it is a paradox (a, c, e, g—the
unity of opposites) or a resolution (b, d, f, g—the opposition of unity). Here we have,
then, an argument that is in form merely an extension of the original river sequence.
However, it contains two new features.
41
The placing of the comma affects the sense here—is anelpiston the object of elpetai or exeuresei?
I suspect the answer is ‘both’, and this makes the paradox deeper. It can of course be taken in a banal,
uncontradictory way, if the three adjectives are construed ‘unhoped-for, undiscovered, untracked’. They
may, however, have a gerundive force: ‘not to be hoped for, not to be discovered, not to be tracked’. This
reading gives a full-blown contradiction, not just a benign banality, about what happens around the next
corner. I take it in the full-blown way.
First, the argument applies not merely at the object level; it extends also to the
theory that attempts to explain the object level. Each stage of the argument consists of
a pair of statements, connected, as I have argued, inferentially. Thus there are four
stages in the series outlined here: a–b: the object level; c–d: knowledge of objects; e–f:
principles governing the object level; and g–h: knowledge of those principles. The
principles of the sequence are three: first a synthesis of contradictories; and then a
resolution based on LNC; and then a further move synthesizing its two predecessors
(either by reflecting on the knowledge of those two propositions, as at c and g, or by
reflecting on the underlying truths, as at e). There is no formal reason to suppose that
the series will come to an end, since each pair of stages can be reproduced by further
applications of the principles; although as a matter of fact it is hard to see how to
generate the next move in the sequence (‘Common knowledge is the same and not
the same as private knowledge’?). Nonetheless this formal character—which derives
after all from the fact that the unity of opposites principle is both bivalent and self-
referential—gives an air of paradox to the entire series. This air of paradox, as we
shall see, is reinforced by the following real difficulty.
Second, at each stage it becomes increasingly difficult to specify the appropriate
qualifiers to lift the paradox and produce a resolution. The object level is straight-
forward enough; here reflection upon the nature of the contradictories, and reassur-
ance from our intuitions about rivers, allows us to move from a to b with relative ease.
Once, however, we start a more reflective inference, the resolution of the paradox
becomes harder (as at c–d). Then, at the abstract level, it is extremely difficult to
specify how the contradictories are to be resolved. That is, we need to cover a fair
number of philosophical miles to see how it is that the principles of unity and
opposition are both the same and different (thus step f requires us to have under-
stood a great deal of the exposition offered by Heraclitus; and even then he says we
will not understand), and how they can be grasped by knowledge that is both
common and private. So as we move from the object level to the theoretical level,
the paradox itself, being merely a synthesis of the previous two moves, is eo ipso easy
to find. By contrast, it is hard, if not impossible, to lift the paradox by finding the
appropriate qualifiers for the resolution.
The sequence I have outlined contains two such critical points: in the move from e
to f, and in the move from g to h. The former asks how we can qualify the bivalence of
the unity of opposites and the opposition of unity; the latter asks how we can mediate
between the common and the private view. So the former is an ontological, the latter
an epistemological problem. They will remain problems just so long as Heraclitus
refuses to explain—as he does explain at the object level—how the crucial antitheses
are to be modified.
This suggests, then, that Heraclitus’ thought is structured at least in an open-ended
way, and at most in a highly paradoxical manner. For by contrast with the resolved
paradoxes of the object level, at the theoretical level the resolutions remain to be
sought or set out or explained. This hypothesis is confirmed if we look to two further
collections of fragments: one epistemological (the ‘god’s-eye view’) and the other
cosmological (the nature of fire).
7. The God’s-Eye View

A familiar Heraclitean thesis has become known as the ‘god’s-eye view’.42 Here
Heraclitus, in his criticism of mortal cognitive capacity, compares mortals unfavour-
ably with god. Just as an ape or a child is to man, so man is to god (82–3, 79). So the
god’s-eye view incorporates the unified truth that Heraclitus’ account endeavours to
put across. As with Heraclitus’ general programme, the irony may be that god is the
only one who can understand his view; nonetheless, as commentators assume,
Heraclitus still thinks that god has got it right, even if his truths and his nature are
indeed inscrutable by man (86). So:
The nature of man has no understanding, the nature of god has. (78)
To god everything is beautiful and good and just; but men take some things to be just and some
unjust. (102)
Those who speak with intelligence must hold strongly to that which is common, as a city does to
its law, and much more strongly than that. For all human laws are nourished by the single divine
law. For it rules as much as it wishes, and it is sufficient for all, and still survives.43 (114)
These fragments connect god’s view with the ‘common’; and so oppose it to the
private view of man. Further, god is actually identified with the objective unity of
opposites that the common view claims to be true:
The god, day night, winter summer, war peace, satiety hunger: it changes as when mixed with
spices it is named according to the pleasure of each.44 (67)
So in the god’s-eye view, no less than in the general descriptions of mortal cognition,
we have once again the unity of opposites thesis, and the contrast between the
common and the private, wherein god represents the common and unity, the
man’s-eye view the private and the diverse. So the god/man contrast is of a piece
with the systematic contrasts of the rest of Heraclitus’ thought.
42
Cf. here Burnyeat (1982b).
43
This fragment is highly resonant with others. There is a word-play on xun noôi xunos that ties in with
fr. 2; there is an all/one contrast that connects with 1, 10, 50, etc.; there is a man : god ratio that is echoed at
102 and 79, cf. also 32.
44
This fragment is rich in its paradoxical overtones—cf. Kahn (1973), 280: ‘as it stands the subject of
the entire second sentence is nameless; it acquires its name only from the spices with which it is mixed, and
in which the namer takes pleasure’. I have omitted pur after hokôsper in the second clause. It was supplied
originally by Diels, and has been accepted by many; but this seems to me to beg the question about the
position of fire in Heraclitus’ cosmology. Rather I follow Kahn’s excellent point so that the fragment has
continuity e.g. with 102 and 7 without the bald reiteration of some doctrine; cf. also 32 and 48. The deepest
level of paradox here lies in the suggestion that god the unnamed might change at all; compare the paradox
of cosmic fire, discussed below.
But what has gone before should make us wary of so ready an identification. I have
argued that for the unity of opposites thesis, the opposition of unity is crucial; and
I have shown how the Heraclitean view of mortal knowledge (that it fails to see the
common) requires also that it recognizes the private (cf. some of the remarks about
sense-perception, 7, 55, 101a, and perhaps 345). That is, both the ontological thesis
and the epistemological one are bivalent, since they require their converse to make
any sense at all. Likewise, the pairs of theses themselves use pairs of opposites for
their own expression. So, at the higher level the principle of the unity of opposites
holds good just the same.
But then if these other principles are bivalent, should we not say the same of the
god’s-eye view? Crucial pieces of evidence here are 53 and 62:
War is the father of all, the king of all; and he shows some as gods, some as men, he makes
some slaves and some free. (53; cf. 67 and 80)
Immortals mortals, mortals immortals, living the others’ death, dead in the others’ life.
(62; cf. 15, 21, 26)
Both these fragments suggest that far from the relation between god and man being
one of hopeless asymmetry, so that god transcends the opposites and constitutes
unity all on his own, in fact god himself is just one of a pair of opposites: man and
god. The consequence of this will be that the unity represented by god is mirrored in
the opposition represented by the man’s-eye view, so that the relation between god
and man is isomorphic: you can no more have the god’s-eye view without the man’s
than you can have the man’s-eye view without the god’s. This unexpected parasitism
of the god’s-eye view upon the man’s is a consequence of the equipollence of the
unity of opposites and the opposition of unity; if both principles are crucial, then
both the god and the man will be needed to make up the complete perspective. (And
at this stage, we may well ask whether the ‘complete perspective’ is itself balanced by
its opposite.) To this end, in the fragments cited above, Heraclitus suggests to us that
man and god are just an instance of the opposites in the world as a whole.
This has one very importance consequence. The contrast between man and god
is central, not just to Heraclitus’ view of the world, but also to his theoretical approach:
it expresses the dualism of common and private that dominates his programme. So in
45
‘The sun is the width of a human foot’. This fragment is often taken to be part of the cosmology; but
that pays no heed to the paradoxical impact of the remark. Better, we should compare Aristotle’s use of the
same example in his discussion of phantasia, and the way in which there may be a direct and obvious
conflict between the report of the senses and the relevant belief (de anima 428b3). In cases like this, that is,
there may be an immediate play-off between the paradox (the sun is a foot wide) and the doxa (the sun is
larger than the inhabited world); neither Aristotle nor Heraclitus, by reporting cases like this, are
committed to repudiating the evidence of the senses as such, or to saying that such paradoxes are
unequivocally false. Bollack and Wismann are hilarious here (1972), 68—they suggest that the view that
the sun is the width of a man’s foot can only be reached by lying on one’s back, sticking one’s foot in the air,
and then finding that it blots out the sun. This self-defeating procedure they take to be of a piece with the
rest of Heraclitus’ thought.
his account of the relation between god and man we see once again how Heraclitus’
metalanguage is itself governed by the unity of opposites. Hence the paradox:
The one wise alone wishes and does not wish to be called by the name Zeus. (32)
Here the formal structure of the fragment is the same as, for example, that of 49a,
where contradictories are asserted at once, without the resolving qualifiers (‘wishes
and does not wish’). This paradox, however, concerns not the ordinary objects of
the world, but the theory itself. And the echoes of this fragment with the others that
talk about the account or the theory are insistent. ‘The one wise’ recalls the unity
of opposites of 50; and ‘the one wise alone’ recalls the more veiled reference to
difference in 108 (‘ . . . that the wise is different from everything’). ‘The name Zeus’
may bring unity; yet ‘wishes and does not wish to be called’ imports the opposites. So
god’s attitude to the opposites is as ambivalent as man’s when man calls god by
whatever name he pleases (67). And the same isomorphism of the relation between
man and god is now apparent in 114. Here, once again, we recall the programme of
what is common, and its private converse; and we see how the divine law is subject to
the divine will—yet the divine will goes both ways, as in 32. And, finally, the notion
that the divine law ‘survives’ recalls the impassable logos of soul (45), and the paradox
of inquiry (18, cf. 22): here too the mind of god and the mind of man are alike at the
same time as they are essentially different.
The ambivalence of the god’s-eye view is to be seen also in his relations with man,
in the indirect way in which he communicates:
The lord whose oracle is at Delphi neither speaks nor remains silent but gives a sign. (93)
Nature loves to be hidden. (123)
The unseen harmony is better than the seen. (54)
Heraclitus, the man who reveals the truth, is as cryptic as the god; and the truth he
reveals is a backward-turning connection.
8. Cosmology
The way in which the puzzles of the object level extend to the theoretical level is
reflected in Heraclitus’ language. Obviously, his paradoxes at the object level turn on
pairs of opposites; at the higher level, attention is focused upon a particular set of
pairs, which express the contrasts enmeshed in Heraclitus’ theory. As we have seen,
his account of the objective world is described generally in terms of unity and
opposition. This involves two pairs of terms, which extend from the objective to
the theoretical level: one/many and same/different (objective, e.g. 49a, 60; theoretical,
e.g. 50, 51). The epistemological counterpart of these pairs is the pair common/
private, which dominates his account of mortal and divine cognition (e.g. 1, 2).
Heraclitus’ cosmology adds a fourth pair of pivotal terms: static and dynamic.
It is necessary to know that war is common, and that strife is justice, and that everything
happens according to strife and necessity. (80)
The cosmology, like the ontology and epistemology, exhibits bivalence, an emphasis
upon both the principle of unity and the principle of opposition.
This order, the same for everything, no god or man made, but it was always and is and will be
ever-living fire, kindling in measures and being quenched in measures.46 (30)
The turnings of fire: first sea, then of sea half is earth, half fiery . . . earth dissolves as sea, and is
measured up to the same amount as it was before it became earth.47 (31)
For souls it is death to become water, for water it is death to become earth, from earth water
comes into being, from water soul.48 (36)
Everything is the exchange of fire, and fire of everything, just as goods are the exchange for gold
and gold for goods. (90)
Cold things warm up, the warm cools, wet dries up, dry gets wet.49 (126)
The fairest cosmos is like a heap of random sweepings.50 (124)
At first, these fragments fit into a well-determined picture of Presocratic cosmology;

compare Anaximander, fragment 1:
. . . the indefinite . . . from which things generation comes for what is, and destruction happens
into them according to necessity. For they give justice and repayment to each other for their
injustice according to the ordinance of time.51
So, if he did not have an insane view of total flux, Heraclitus is commonly thought to
have believed, moderately, that the changes of the physical world are explained by
the judicious alterations in the masses of the elements. In this respect, it is thought,
he closely resembles his Milesian predecessors.52 The balance of power of the
elements explains the regularity of the change; and anyway, Heraclitus has shown
elsewhere that chronological sequence confers unity (e.g. 57, 88, 126). Moreover, that
one element underlies all explains the unity of the cosmos; so he offers as the
46
This ‘measure’ business is inevitably the subject of some controversy: if Heraclitus propounded a
theory of flux, was it orderly or chaotic flux? Cf. here Kirk and Guthrie in Mourelatos (1974), 189–213.
47
This translation picks up a point G. E. L. Owen used to make, that we should eschew any ontological
overtones of eis ton auton logon, and avoid supposing that logos is either some grand plan or some naturally
occurring principle. All that Heraclitus means here is that the new quantity (of earth) matches the old
quantity (of sea). Cf. 90.
48
It looks here as if Heraclitus is incorporating soul into the cosmic cycle—perhaps just because he
equates soul with the dry (e.g. at 117 and 118, cf. 126). Thus soul is the equivalent, in this sequence, of fire.
49
Cf. Stokes (1971), 92, who debunks ‘over-cosmologising’ here.
50
Cosmos means, in general, ‘order’, as I translated it in 89; and so it may be read here—or, with the
translation ‘cosmos’, as a general statement about the world.
51
I discussed the relation between this fragment and Heraclitus 80 in my (1978). In that paper I was
coyer about translating the words cognate with dikê, just because I was discussing the meaning of such
terms.
52
Cf. DK 22A1, 5, 10; Vlastos (1955) 62; Wiggins (1982). Again, Stokes (1971) is iconoclastic.
substrate the element fire, so far unused by a cosmologist.53 Thus the cosmos
embodies the principle of strife, held in equilibrium (80, 53).
Yet there is a puzzle here. First of all, if fire is the substrate (as 30 and 90 both
suggest), then what qualifies fire, as opposed to any of the other elements (which, in
31, appear to take an equal turn) to be the substrate? Then, of course, fire is rather an
odd choice, if we compare the archai of the Milesians—Thales’ water, Anaximander’s
‘indefinite’, and Anaximenes’ air. After all, all those stuffs are characteristically inert
(water, air), indeterminate (‘indefinite’, air), and homogeneous. But Heraclitus’
‘stuff ’ is the reverse, and thus perhaps has the least claim of all the elements to
underlie the rest; for fire is dynamic, determinate, and differentiated—indeed, the
notion that the all-consuming fire could underlie anything is extremely implausible.
And what is more, the very statement that fire is the underlying unity in the world is
itself equivocated. For if the fire is ever-living (30), it is also ever-changing (31). But
elemental change (36) is death. So fire by ever-changing is ever-dying; thus fire is
ever-living and ever-dying. Always the same and always different, it is a paradoxical
substrate.
Fire, by remaining the same, becomes different; the stability of the cosmos is
expressed in its changes. So the unity of opposites in fire (ever-living/ever-dying) is
modified as change between the elements. And thus the relation between stability and
change, or life and death, appears to be one of continuity over time. Life and death
are analogous to sleeping and waking (1, 62, 88, 89):
Man strikes a light for himself in the dark when his eyes are quenched, living he touches the
dead in sleep, awake he touches sleep.54 (26)
which, in turn, were crucial to understanding the contrast between public and private
in the programmatic fragments. So the cosmos (‘this order, the same for everything’)
is characterized by unity and diversity over time, which reflects itself in the contrast
between one and many (90), same and different (30, 31), and permanence and
change (living/dying, being/becoming (80), static/dynamic (88, end of 91; cf. 125,
84a)). This is all exemplified in the turnings of fire, which are both stable and
changing over time.
9. Dialectic and Metaphysics

Heraclitus’ fragments are themselves an embodiment of the unity of opposites and
the opposition of unity. For throughout the account that he gives, each thesis is
modified by an antithesis. The unity of opposites is balanced by the opposition of
53
This is, once again, a vexed issue. Stokes (1971), 102 ff., vigorously attacks the ‘substratum’ Interpret-
ation. We should bear in mind that not only do Heraclitus’ fragments support various different accounts of
fire’s position as archê, they also suggest the possibility that fire is to kinoun, e.g. 64, 66.
54
Cf. Kahn (1973), 213 ff.
unity, and thus explains the objects of the phenomenal world; and on a cosmic scale
the eternal fire dies in a stable cycle of elemental change. Likewise, from the
epistemological point of view, for man common knowledge is modified by his private
perspective, just as the god’s-eye view is counterbalanced by the man’s-eye view.
Throughout the interplay of these principles, the unity/opposites antithesis explains
both the content of each principle and the interrelations between the principles. So
the unity of opposites is both bivalent and self-referential. The result is a puzzling
sequence of enmeshed truths.
At the object level, Heraclitus exploits a gamut of opposites, none of which is given
greater prominence than the other, and from which he manufactures his paradoxes
and resolutions (e.g. night, day; just, unjust; sick, healthy). My theme so far has been
to show how Heraclitus offers us two different perspectives on the paradoxes,
corresponding to the subject on the one hand and the predicate on the other. The
solving of the threatened contradiction in the predicate shows us the workings of the
law of non-contradiction; and the dialectical effect of this is to make us see, as a
general principle, that LNC is true. At the same time, the subject that underlies the
contradiction is seen to resist disintegration, since the paradox exploits our intuition
that the subject is indeed ‘one and the same’. The effect of this is to reassure us that
there are individuals underlying the opposite properties of the phenomenal world. So
the paradoxes, at the object level, have a double metaphysical effect: first, they
establish LNC elenctically; second, they expose and confirm the assumption that
there are individuals in the world that are self-identical at a time, persist over time,
and can be re-identified—rivers, roads, and possets. These two effects, I argued,
account for Heraclitus’ use of paradox without committing him either to the absurd-
ities of flux and hopeless indeterminacy or to scepticism.
The resolved paradoxes of the object level thus have a direct argumentative effect.
They are to be understood in terms of the unity of the opposites, a principle that is
both bivalent and self-referential. As a consequence, the unity of opposites generates
higher-level paradoxes that contain a greater degree of difficulty, since the necessary
qualifiers to lift the contradictions are not readily forthcoming, certainly not without
considerable reflection. And at the higher level the puzzle is seen to turn not on
random pairs of opposites, but on a particular series of connected terms: ‘one/many’
and ‘same/different’ (together these account for the unity of opposites); ‘common/
private’ (the epistemological pair); and ‘static/dynamic’ (the cosmological pair). Why
should Heraclitus focus upon just these terms? And why should they remain puzzling
at the higher level?
The higher level is, of course, one where the explanations, as opposed to the
descriptions, of the object world are lodged. Thus it is here that the justification for
general truths must be sought. Heraclitus has exposed our commitment, on the one
hand, to LNC; and on the other to the fact that there are individuals in the real world.
The denial of LNC, like the assertion of flux, has a direct effect upon the supposed
population of the world, upon our ontology; and conversely, the assertion of LNC
and the denial of flux reassures us that there are individuals. But it leaves unanswered
an explanation of that intuition. Every stage of Heraclitus’ thought, however, insists
that we do explain the ones under the many: ontologically, the holders of properties;
cosmologically, the persistence of these items over time; and epistemologically, the
means of recognizing items beset by the private or the common view.
How are we to go about explaining the identity of individuals?
It would appear that a theory of individuation must comprise at least three things: first an
elucidation of the primitive concept of identity or sameness; second, some, however abstract,
account of what it is for something to be a substance that persists through change; and
third . . . the beginnings of some lifelike description, however schematic, of what it is for a
thinker at one time and then at another to single out the same substance as the same
substance.55
Wiggins’ account of the problems of individuation is certainly influenced by the

Aristotelian tradition of which, I suggest, Heraclitus was a precursor. But any
philosopher, whether Aristotelian or anti-Aristotelian, must in speaking of the
identity of individuals deal with the same set of questions, if only to argue that
some of them are illicit.
i) What is it to be an individual, rather than, for example, a mass, a stuff, or a
group? How does an individual have self-identity at a time? In what way is it
‘the same as itself ’?
ii) What is it for an individual to have self-identity over time? Can we explain
why this now will be the same this tomorrow?
iii) How do we know that something is a self-identical individual? This relates to
the problem, elaborated by Strawson,56 of identification and re-identification:
the assumption here is that individuation may depend upon epistemological
not ontological criteria.
Each of these questions is confronted in the fragments of Heraclitus, as we have
seen: that is, the first-level paradoxes provoke us to realize that there are self-identical
roads, persistent rivers, and identifiable days. But what it means to say that they are
self-identical, persistent, and identifiable remains unexplored while we merely insist
that ‘I can step into the same river twice’. That is, to support our bare intuitions, we
need to understand the concept of an individual; we need a theoretical account of
‘what is it to be an individual?’
That question is posed in the very lexicon of terms Heraclitus makes prominent at
the theoretical level. To discuss the identity of individuals, we need to know how
something counts as an individual; what it is for it to be self-identical; how it
persists over time; and how it can be recognized as such. Exactly these notions are
captured by Heraclitus’ key antitheses. One/many accounts for something’s being an
55 56
Wiggins (1982), 1. Strawson (1959).
individual, one not many. Same/different are the concepts that underlie self-iden-
tity—for to establish whether something is self-identical, we need to know at least
whether it is the same as itself and different from everything else (cf. the questions
about the river and night/day). Static/dynamic are the terms to describe the persist-
ence or otherwise of something over time. Common/private are the epistemological
categories that grasp items considered as a unity or as differentiated; so these are the
terms with which to explain how we can identify or re-identify an individual.
So to articulate a theory of individuals, we need to understand these terms. And
before we come to any such understanding, we must first be aware just which are the
central concepts of individuation; and second we must recognize that these concepts,
so far from being self-evident, badly need explaining (‘Is sameness a property?’;
‘What is the connection between “sameness” and its opposite “difference”?’; ‘What
counts as a “one”?’, etc.). And this, I suggest, is exactly the awareness that Heraclitus
gives us in the careful structure of his higher-level fragments.
First, he uses the key terms in his discussion of the individuals of the physical
world. In these paradoxes—the river, the road, and the posset—he emphasizes that
there are individuals, by attributing unity and opposition, sameness and difference,
motion and rest to them. Then by implicating these terms in his general theory, he
isolates and interrelates the key terms by emphasis and reiteration, and the echo of
one pair by another from fragment to fragment. He shows us how one is inseparable
from many and how the uniformity of the common view is inextricable from the
diversity of the private view. And he points out how unity is related to uniformity
(sameness) and to the common view, just as difference, plurality, and privacy appear
to be essentially linked. So the terms used at the object level are taken out of context
and abused at the theoretical level. As a result, he shows us just how puzzling they are.
He does this by challenging the relation between the two members of a pair, and by
embedding the challenge in fragments that purport to explain great philosophical
issues. The consequence of this is that, as happens in paradox, we are both convinced
of the importance of one/many, same/different, etc., and unable to see how they can
be made to make sense. Without them, both the objects of the world and our talk
about them would collapse; yet with these terms, our talk appears to be fraught with
contradiction.
Return, for a final time, to the river ‘argument’. The first stage in that argument
turned on the opacity of ‘river’, and produced a paradox about ‘the same river’. The
second move (49a) then suggested that ‘the same river’ is beset by contradictions, and
the third resolved the contradictions by pointing out that the river is the same only in
one respect; it is different (it changes) in another. The entire sequence turns on the
same/different contrast, and we emerge with some intuitive sense of the individual
river.
But if we wish to explain the focal expression ‘the same river’, we have to
understand ‘same’. To do so, we turn, not unnaturally, to the fragments that purport
to explain rather than merely describe (1, 2, 50, 80, etc.). Then we discover that
sameness is an expression of the unity of opposites (88), or an aspect of the common

view (50, 51, 114). So ‘same’, ‘one’, ‘stable’, and ‘common’ are linked by the play of
words in the theoretical fragments; and contrasted, at first appearance, with the
quartet ‘different’, ‘many’, ‘changing’, and ‘private’ (2, 72, 89). That is, the terms that
feature prominently in the puzzles about objects (the river, the circle, etc.) are pivotal
in the theoretical fragments that, in turn, claim to explain the object level. The
consequence of this, at first, is to bring out the centrality of these terms; but at last
to show how, at least within the compass of the surviving fragments, they defy clear-
cut and unequivocal explanation. For, as we have seen, once the key terms are
brought into relief and then moved from the context of objects into the context of
theory, they cannot be explained without recourse to their opposites. So, as I have
argued, sameness is meshed together with difference (51); one is entangled with
many (10, 41); the common needs the private (1, 57); and stability is involved in
perpetual change (30, 36). So once the fragments about sensible individuals have
shown us which are the terms that are crucial to understanding the paradoxes, the
fragments that tell us about understanding show us how deeply puzzling those terms
are. They leave us, that is, with a question, not an answer, and invite us to investigate
further the intuition that we do step into the same river, twice, once, or at all.
This, then, is the dialectical effect of the puzzles of the higher-level fragments: to
focus our attention upon these particular terms and to arouse our demands for an
explanation of them. The trick is done with paradoxes, and with the way in which
paradoxes invite the argumentative participation of the hearer. Heraclitus is exploit-
ing the psychology of paradox to establish, first of all, two general principles (LNC
and the intuition that there are individuals) and to raise a general metaphysical
question. In each case the effect is achieved by the challenge to our common
assumptions offered by paradoxes both singulatim and in series. By provoking
doxai, they bring to the surface and articulate general principles of discourse and
general assumptions about the world that would otherwise remain tacit and unexam-
ined. Then by leaving some areas of puzzlement unresolved, Heraclitus provokes
further enquiry, directed specifically at the explanation and understanding of the
metaphysics of individuation.57
57
Versions of this chapter were read at Brown, Cornell, Cambridge, and Harvard Universities. I am
grateful to the audiences on all those occasions for stimulating and helpful discussion and also to Julia
Annas and Jim Lesher. This chapter was first conceived when I was a Junior Fellow at the Center for
Hellenic Studies, Washington, DC; I should like to thank the then Director, Bernard Knox, the staff, and
the other Fellows, particularly Dorothea Frede, for their comments and support.
3
The Moving Posset Stands Still:
Heraclitus Fr. 125
The famous ‘posset’ fragment of Heraclitus is cited first by Theophrastus, de vertigine

caps. 9–10:1
9. ªÆØ ’ YºØªª ŒÆd ‹Æ N e ÆPe º ø Ø 76
ŒÆd K ÆÇø Ø. n ŒÆd K Æ æEÆØ Øa
ŒØı Å ŒŒºfiø B Złø ŒÆd Mæ Å ı ÆØ
e ÆPe Ł· ¼ ªaæ e ’ KÆø. ÆYØ
b F b K Bfi ŒıŒºçæÆ fi e NæÅ , F ’ K 80
fi K Ø Ø ŒÆd fiH IØ fiH, ØØ a K fiH ŒØE-
B
ŁÆØ øÇ Æ Ø Å Ø ŒÆd Ø· B Złø b
Å, e æı, ŒÆd pººÆ a ıåB K fiH KªŒ-
çºfiø ¥ ÆÆØ· ØØ Æ b ŒÆd åøæØÇ Æ a Ææ Æ
ŒÆÆ ÆæØ ŒÆd ØE e YºØªª. a ªaæ çıŒÆ 85
ŒØE ŁÆØ c ŒÅ Ø ÇÆØ ŒÆd ı Ø Øa
ÆÅ· N b c, ŒÆŁ æ HæŒºØ çÅ Ø, ŒÆd ›
10. ŒıŒg Ø ÆÆØ c ŒØ . YÅ ’ i ŒÆd B fi
ŒıŒºçæÆ fi ÆPe F’ I ØÆØ· Ø Å Ø ªaæ
Å Ææ Æ ŒÆd ŒFçÆ – Æ r ÆØ a b 90
N e ¼ªı Æ a ’ N e åÆ. ØE b
ŒÆd e Œ Ø Nºªªı Øa F· åøæÇÆØ ªaæ a
Ææ Æ ŒÆd KºÆçæa, a b ¼ø a b Œø. E b
– Æ r ÆØ ŒÆŁ æ Kº åŁÅ. 94
82–3 b Å Usener: Ø Æ Ø A Q Ald. 84 ¥ ÆÆØ Usener: ı Æ ŁÆØ A Q Ald.
86 ÇÆØ Wimmer: ¼ºº A Q Ald. 87 N b c ] N c Bernays 88 Ø ÆÆØ ] b
¥ ÆÆØ Forster 88 c add. Bernays
‘Dizziness also occurs when people look at the same thing and gaze intently. This
indeed raises a new doubt, how the same effect could come about both when the eye
moves in a circle and when it is at rest. For that the same effect be caused by opposites
1
I give the text and apparatus of Wimmer (Leipzig 1862) with some additions. I have profited from
seeing in advance R. W. Sharples’ apparatus for de vertigine (published as Sharples (2003)).
66 THE MOVING POSSET STANDS STILL
is absurd. What has been said above explains the dizziness in circular motion; while
the explanation of dizziness in staring and fixed gazing is that it is actually rest which
disintegrates things which are preserved in motion. When the eye, one part of the
brain, is still, then the other parts of the brain, being continuous with it inside the
head, stand still too. For the heavy parts, disintegrating and being separated, weigh
the brain down, and cause dizziness. i) For the things which are naturally in motion
preserve this motion for themselves (or: are preserved) and through this motion they
stay together. ii) If not, as Heraclitus says, even the posset disintegrates when it does
not move. iii) It would be possible to attribute this same phenomenon to circular
motion also; for the vortex dissipates the heavy and light when they should be
together, driving the former to the middle and the latter to the outside. Stooping
causes dizziness like this too, for it separates the heavy and the light, downwards and
upwards respectively, but they should be together, as has been said.’ (My translation:
I have enumerated the sentences that will be repeatedly cited and discussed in detail.)
We may compare [Alexander] Problemata 4.42 (Usener) alias [Aristotle] Proble-
mata inedita 2.42 (Bussemaker)2 where the same fragment of Heraclitus is referred to
thus: › b ŒıŒg, u æ ŒÆd HæŒºØ çÅ Ø, Ka Ø ÆæÅ fi , ¥ ÆÆØ. Photius’
version of the Theophrastus passage, on the other hand, while it can be helpful with
the Theophrastan argument, makes no mention of Heraclitus (Bibliotheca 526b40ff.).
First of all, what is a posset (ŒıŒ)? Possets turn up in various early sources, e.g.
Homer, Il. 11.638ff.; Od. 10. 234; Hom. Hymn to Demeter 207ff. They are, apparently,
mixtures of wine, barley groats, cheese, and herbs or drugs that remain suspended
only if they are continuously stirred. If the motion ceases, then the solids separate
out, and sink to the bottom of the liquid. To reformulate this in Aristotelian terms, to
be a posset, the mixture must move. If, contrary to its nature, it comes to rest, then it
separates out and is, we may suppose, only a posset › ø ø. It follows that for a
posset simultaneously to move and to separate out is not merely unnatural, but
impossible.
To determine what Heraclitus said about possets requires some grasp of Theo-
phrastus’ purpose in quoting him. De vertigine is a discussion of the various causes of
dizziness. It opens with an account of the ways in which motion causes dizziness
(lines 1–76 in Wimmer), on the assumption that dizziness is a phenomenon that is
contrary to nature (cf. e.g. 5). Dizziness caused by motion is said to cover a variety of
cases, such as running in a circle, shaking the head, or looking at moving wheels. In
short, dizziness is caused by all sorts of circular motion, ŒıŒºçæÆ (7, 9, 27, 29, 36,
40, etc.). Then Theophrastus encounters a puzzle (76). Dizziness also occurs when
people look fixedly at the same stationary thing. So the same effect seems to have
opposite causes, which appears absurd (this, of course, is a standard worry in Greek
thought; compare Plato Phaedo 96a ff.). Accordingly, Theophrastus goes on to
2
I am grateful to R. W. Sharples for pointing out to me the issue of the authorship of this passage.
Compare Forster (1933), 140–1, with Marcovich (1967), 15.
THE MOVING POSSET STANDS STILL 67
explain how rest can cause dizziness; and he follows this account with other examples
of opposite causes of dizziness (94 ff.). This suggests that the overall structure of the
fragment up to 94 is to offset the account of dizziness by circular motion against that
of dizziness by rest. So the passage 76–94 should give the required account of
dizziness by rest—at least if the argumentation is sound.
It is, conversely, possible that Theophrastus changes course mid-stream and
reverts to the discussion of ŒıŒºçæÆ in the midst of the discussion of rest at 88
(my iii). While considerations of the balance of the overall argument tell against this,
the wording may support it. YÅ ’ i . . . I ØÆØ means ‘it would be possible to
attribute this (to circular motion)’. Does this imply that the ascription is correct?
Theophrastus would then have returned to his original topic of circular motion.
However, this use of the optative with i turns up regularly in Theophrastus, as it
does in Aristotle, to express a question whose answer is not yet decided. So ‘it would
be possible to enquire whether without motion the heavens would be destroyed’ (Th.
Met. 6a12, cf. 8a21) does not imply that the inquiry must be answered in the
affirmative.
In iii, ÆPe F’ must refer to some phenomenon that needs explaining. Which
phenomenon? The posset? The preservation of naturally moving things? Or the fixed
gaze of the eyes? The last would merely contradict what has gone before; and the
second will not do any better. For to say that the vortex (i.e. motion) separates out the
things that ought (i.e. by nature) to be together does not explain the survival of things
that are naturally in motion. The separation of heavy and light, however, does
actually occur in the first case, that of the posset—but when the posset comes to
unnatural rest, not as it moves. If the vortex is relevant to the posset, then its
disintegration will have been explained both by rest (implied at i) and by motion.
Yet that—to suppose that the same event can have opposite explanations—is, as
Theophrastus has already said, absurd. As in the earlier case of the fixed gaze, then,
we should relinquish one of the alternative accounts; and common sense tells us, of
course, that the posset disintegrates when it comes to rest. Accordingly, Theophras-
tus goes on to cite a further case of dizziness caused by separation in rest with the
example of stooping. After all, to stay in position bent over does cause dizziness,
while the motion of touching the toes does not (cf. here Aristophanes, Eq. 1354, Th.
930 on Œ Ø as a state, not an action, of stooping). ÆPe F’ thus refers back to
the separating effect of rest.
With some sense, then, of the unity of the passage as a whole, let us return to
Heraclitus, and sentences i and ii. i appears to be hopelessly corrupt. Kirk (1954), 25
defends c ŒÅ Ø ¼ºº ŒÆd ı Ø with the gloss: ‘a contrast between
¼ºº (= ‘normally’, i.e., when they are in motion) and N b c ( = ‘otherwise’, i.e.,
when they are not in motion).’ But the contrast between ‘normally’ and ‘otherwise’
hardly suits the fact that Theophrastus is enunciating a general principle of physics,
which would operate ‘necessarily or for the most part’ on the phenomena it governs.
Marcovich obelizes ¼ºº (op. cit. 156). Wimmer suggests ÇÆØ instead, and is
followed by other editors. The point of the sentence, nonetheless, is clear from the
closing clause, namely to describe items that are naturally in motion. a) They move
necessarily or for the most part and b) their motion is crucial to the account to be
given of their nature. The latter point is made here by the emphasis on motion in
the sentence c ŒÅ Ø picked up by Øa ÆÅ. The effect of the sentence,
then, is to focus upon the motion that does the preserving (cf. 81–82).
The fragment of Heraclitus embedded in ii is affected by three textual cruces:
a) Should the first clause of ii read N b c (thus Wimmer, Kirk, Marcovich) or N
c (Bernays, Bywater, Diels)?
b) Should the verb in the second clause of ii be Ø ÆÆØ (Bernays, Wimmer) or
¥ ÆÆØ (as in [Aristotle]/[Alexander], adopted by Forster (1933))?
c) Should we insert c before ŒØ ? Bernays first suggested this (in 1848
(1885), 6), and has been followed by all the commentators bar Bollack and
Wismann in inserting it. He conjectured that the c had been displaced to the
previous clause, so he deletes c there.
The virtually unanimous answers of the commentators to questions b) and c) give
the received version of the Heraclitean fragment:
125A › ŒıŒg Ø ÆÆØ c ŒØ
The answer, however, to these textual questions cannot be isolated from the
interpretation of Theophrastus. I have argued that the entire paragraph is a coherent
account of dizziness through rest. The two answers to question a) give alternative
translations of the sentence: ‘If not, just as Heraclitus says, even the posset disinte-
grates when it is not stirred’, or, following Bernays: ‘If so, just as Heraclitus says, even
the posset disintegrates when it is not stirred.’
If so, or if not, what? What is understood in the protasis of the condition (if, that is,
we punctuate with Wimmer, and not, as Bywater, with a comma rather than a colon
before N; the argument I offer below is unaffected by this variant, since the ŒÆd of the
apodosis still demands an explanation and the stylistic considerations offered below
tell against Bywater indifferently)? My argument above suggests ‘if it is (not) through
this motion that they are kept together’, so that the protasis refers to one condition of
the principle of natural motion, namely the fact that it has a preservative effect on
natural movers.
The first version of ii, then, is: ‘If it is not through this motion that these things
are preserved . . . then even the posset disintegrates when it is not stirred.’ Now since
Theophrastus is committed to the truth of the fact that these natural movers are
preserved by this motion, we should expect this to be a counterfactual condition,
tending, by a reductio ad absurdum type of argument, to support the principle of
natural motion. But then there are three problems: first, a counterfactual appears to
require ¼ + optative in the apodosis; second, the apodosis in a counterfactual should
be false, whereas here it is true; and third, the presence of ŒÆd adds an element of
emphatic surprise to the apodosis (cf. here Denniston (1978), 316) which needs
explaining—after all, given the nature of possets, it is not surprising that they
disintegrate when they are not stirred.
The first difficulty, however, is a red herring. It is true enough that in ordinary
literary Greek a counterfactual would be expressed with optative+ ¼ (cf. Kuhner and
Gerth (1955), 573ff.). However, scrutiny of Theophrastus’ habits reveals, first of all,
that he tends to use this construction for a view to which he himself subscribes;
compare, for example, his usage throughout the Metaphysics, e.g. 5a28; 5b7; 5b26;
l0a5. Now Aristotle, when he wishes to express a counterfactual—that is, for logical
purposes, a condition with a false apodosis, useful for modus tollens and reductio ad
absurdum arguments—regularly uses N +indicative in the protasis, and the future
indicative in the apodosis (e.g. Phys. 209a23; GC 332b32 ff.; Met. 1029a5; compare an
early reductio ad absurdum in Melissus fr. 5, and an early example of something
unnatural cited in the indicative at Hdt. 3.62). This construction, however, gives way
at times to a regular condition where the apodosis merely claims that impossible
consequences do result; here we have the present indicative (e.g. Met. 1039a2). The
same construction turns up on several occasions in Theophrastus, where the para-
doxical conclusion of a counterfactual is mentioned (present tense) rather than being
actually cited (e.g. Met. 5b14; de sens. 52; de igne 4, 6). This suggests to me two
different possibilities. The first is that the construction with the future indicative,
where the particular absurdity is cited, may collapse into the construction with the
present indicative, where the existence of absurdities is mentioned; the collapse
would be the grammatical equivalent of the use/mention confusion. Second, if
the absurdity is a paradox cited in oratio recta, we should expect the original tense
to be retained (cf. the retention of the original cases in oratio recta at Th. Met. 7a15).
Both considerations, or either, may justify the view that our sentence ii expresses
more or less grammatically a counterfactual. This justifies the MSS. c in the protasis.
In that case, however, the problems devolve on the apodosis. If we add c before
ŒØ , then the apodosis turns out true. But for a counterfactual, it must be false.
What is more, that falsehood should be a striking one, since it is introduced by ŒÆd
(Professor Kirk suggests to me that this may be taken in a weak sense, ‘such as’. In a
Peripatetic context, however, this would be better expressed by some generalization,
followed by an example, probably introduced by x. The emphasis of ŒÆd, ‘even’,
seems unavoidable in our passage). So what the text needs is some striking falsehood.
Just such a falsehood is supplied by the MSS. reading, without the negative: ‘If it is
not through this motion that these things are held together, then even the posset
disintegrates when it is stirred.’
Bernays’ reading of ii fares no better than its rival. Here there are two difficulties:
first, the force of ŒÆd, and second, the nature of the argument attributed to Theo-
phrastus. ‘If it is through this motion that these things are held together, then even
the posset disintegrates when it is not stirred.’ But what, now, is the force of ‘even’? It
is standard, albeit contrary to nature, for possets to disintegrate when they are not
stirred. To achieve the emphasis, Theophrastus would need to cite some unexpected
or extreme case, such as the collapse of some regular meteorological or elemental
phenomenon. Likewise, if we punctuate with Bywater, it is hard to see how the case of
the posset can provide the heavy (emphasized) support for the principle of natural
motion that, again, an extreme case would. Furthermore, returning to Wimmer’s
punctuation, it is hard to see the grounds for Theophrastus’ apparent inference from
the truth that some things are held together by motion, to the case of the posset.
What he needs here, on the contrary, is some support for the claim that there are
natural movers as well as natural resters (compare the extent to which Aristotle feels
the need to argue this case in the Physics). So on both linguistic grounds (the presence
of ŒÆd) and logical grounds (the nature of the inference), the shifting of the negative
from the protasis to the apodosis is unsatisfactory.
This then leaves us with the unemended MSS reading, with the following trans-
lation: ‘If it is not through this motion that these things are held together, then, just as
Heraclitus says, even the posset disintegrates when it is stirred.’ Here the falsehood
(that the posset disintegrates when it is stirred) is consequent on the denial of part of
the principle of natural motion (that natural movers are held together by their
motion). The mood of the apodosis is justified by the Peripatetic construction for
introducing a paradoxical conclusion (vide the ı Ø/Ø ÆÆØ contrast through-
out the passage). Its tense is explained, on the one hand, by the use of the present to
mention paradoxes in a modus tollens argument and, on the other, by Theophrastus’
anxiety to preserve the original wording of the quotation. This gives him a good
argument by simple modus tollens: I) If natural movers are not held together by their
motion, then the posset’s disintegrating could be explained by motion. II) But the
posset’s disintegrating could not be explained by motion, as is manifest. III) So
natural movers are held together by their motion. Within this argument we find a
Heraclitean fragment that reads:
125B › ŒıŒg Ø ÆÆØ ŒØ .
So Theophrastus’ argument has been shown to be sensible and coherent; and the
Heraclitean fragment to which he alludes has been found. But could this fragment
possibly be Heraclitean? The version with which we began, 125A, was perhaps a
platitude, but it was at least true. The present version is pithier than its predecessor,
so that it has an aesthetic advantage; but it is manifestly false, if one considers possets.
Now Heraclitus was certainly addicted to paradox no less than to pith, but his
paradoxes are of the veridical type—that is, they do not assert falsehoods, whether
obvious or otherwise. Reflection on the road (‘The road up and down is one and
the same’, fr. 60) or the river (‘It is not possible to step into the same river twice’, fr.
91 or, if this is spurious, which I doubt, the alternative frs. 12 and 49a) shows
that Heraclitus is not in the business of manufacturing falsehoods (cf. also frs. 10
and 59). In that case, the fragment that Theophrastus’ text must contain cannot be
Heraclitean.
It could, however, be a modification by Theophrastus of some Heraclitean original.

Suppose that Theophrastus took some well-known Heraclitean remark, adapted it
(turned it into a falsehood) by some minor alteration, and then produced it as an
example to make his case about natural motion. His objective would be to transform
a well-worn example of natural motion, taken from the traditional philosopher of
natural motion, into a striking but nonexistent example of his thesis. The point of the
modification would thus be to make the point that much more surprising; and the
attribution to Heraclitus is retained, as is the tense of the original quotation, to
reinforce the counterfactual effect. Theophrastus’ objective, then, is to make a play on
the Heraclitean original to convince us of the truth of the principle of natural motion.
In that case, there must have been a Heraclitean original about possets that is a truth
corresponding to the falsehood 125B. What could that original have been?
One obvious candidate is 125A, with which we started. This is, of course, a true
saying about possets, and so gets by on truth-value, if not on pith. There is, however,
another candidate. The passage from [Aristotle]/[Alexander] cited above refers
clearly to the same posset fragment: › b ŒıŒg, u æ ŒÆd HæŒºØ çÅ Ø, Ka
Ø ÆæÅfi , ¥ ÆÆØ. This may be read literally, to say that if the posset is not
shaken it stands still. This awesome banality takes ¥ ÆÆØ to mean ‘it is physically at
rest’. However, the verb may as easily be read figuratively, as ‘it preserves its nature’.
If so, [Aristotle]/[Alexander]’s text is, like Theophrastus’, announcing a falsehood.
Usener, consequently, exploiting the similarity with Theophrastus, emended to
Ø ÆÆØ to give: ‘The posset, as indeed Heraclitus says, disintegrates if it is not
shaken.’ However, my Theophrastan argument suggests a different solution, namely
the deletion of c to give: ‘The posset, as indeed Heraclitus says, preserves its nature
when it is shaken.’ And then [Aristotle]/[Alexander] gives us the correct verb for the
original modified by Theophrastus, namely ¥ ÆÆØ. Its presence in [Aristotle]/
[Alexander] attests its currency in the Peripatetic tradition, and points to the
Heraclitean fragment:
125C › ŒıŒg ¥ ÆÆØ ŒØ .
Now either l25A or 125C could be the Heraclitean original—both have the merit of
being true. It is clear enough that Theophrastus has done some emending of his own
to produce the remark he needed for his argument; so on any interpretation we must
alter Theophrastus to get Heraclitus. And which alteration we prefer seems, from the
point of view of MS support, to be indifferent: given that the alteration was, as
I suggest, deliberate, then we cannot tell whether the verb has been altered or the
negative missed out. But one over-riding consideration seems to tell in favour of
125C, › ŒıŒg ¥ ÆÆØ ŒØ . That is the question of style. 125A is true,
platitudinous, and fussy in its structure. It has none of the apparent contradiction
that is so characteristic of Heraclitus’ work, none of the overtones of paradox for
which he was so famous. 125C, on the contrary, is overwhelmingly Heraclitean. It is
paradoxical, because it appears both to assert and deny the same thing. Reflection,
however, on the phenomenon that generates such a contradiction allows us to see

that this is a general truth. It is one, moreover, that reflects on the ordinary nature of
things, and does not require the interference of some cosmic stirrer to make its point.
The comment is about a common-or-garden item (in a culture where possets are
common-or-garden) and its natural state qua posset. So this fragment has a strong
affinity with the other fragments of Heraclitus which prompted the charge that he
violated the law of non-contradiction: compare frs. 10, 49a, 62, 80, 88, 126 and
perhaps most striking of all, 84a— Æ ºº IÆ ÆÆØ.
Decisive matters of style, then, tell in favour of the final version of Heraclitus’
paradox (125C). Its history, I suggest, was thus. When Theophrastus was writing the
de vertigine, the Heraclitean original was current, even well known. Theophrastus
was anxious to construct a modus tollens argument to defend his use of the principle
of natural motion. To do so, he needed a manifest falsehood, about motion—so he
turned to Heraclitus, the flux-merchant, and modified his truth, 125C, into the
falsehood 125B: › ŒıŒg Ø ÆÆØ ŒØ . Commentators, smelling something
fishy, returned this to its original truth value by inserting a negative, hence 125A: ›
ŒıŒg Ø ÆÆØ c ŒØ . But evidence from [Aristotle]/[Alexander] allows us
to recapture the full Heraclitean flavour of the original:
125C: › ŒıŒg ¥ ÆÆØ ŒØ .3
3
I have profited here and elsewhere from correspondence with Prof. M. R. D. Reeve. I should also like to
thank Dorothea Frede, Kevin Lee, and J. H. Lesher.
4
Parmenides’ Dilemma
Parmenides the Eleatic wrote a treatise that intrigued, puzzled, and confounded the
later philosophical tradition.1 In it, he argued for a strong monism: what there is
is eternal, complete, immoveable, and unvarying, one and homogeneous (DK 28B
8.3–6).2 All the rest, the world of perceptible things, is contradictory—or an illusion.
Strong monism is frighteningly radical. So Parmenides left a series of problems in
his wake, some of which have proved so recalcitrant as to be dismissed with that
counsel of despair ‘it’s a dialectical device’.3 This chapter addresses two of those
problems, and recasts the dialectical device in a mood of optimism.
1. The Rules of Dialectic

Parmenides’ poem falls into two sections. The first, the Alêtheia, argues for a
preliminary conclusion in frs. 2–7, which is further expanded by deductive argument
in 8.1–49. The second, the Doxa, occupies a further eleven lines of fr. 8, frs. 9–19,
and a considerable amount of the doxographical evidence. While the Doxa offers
descriptive material, Parmenides’ argument is apparently limited to the Alêtheia. It is
founded upon a premiss that Parmenides treats as self-evident: ‘you can think and
speak’.
Descartes’ cogito4 may be described as self-verifying.5 Provided ‘think’ is not
limited to ‘think successfully’, ‘get it right’, ‘think about an actual object’, then as
soon as the proposition ‘I think’ becomes formulated, it is self-fulfilling, just because
the proposition ‘I think’ makes sense.6
Parmenides, however, uses a second-person formulation: ‘you think’. Throughout
the Alêtheia the second-person singular, introduced by the device of the goddess’
1
The secondary literature on Parmenides is extensive: cf. bibliographies in Mourelatos (1970) and
Barnes (1979). Like many students of ancient philosophy, I have benefited most of all from the work of
G. E. L. Owen; see, for example, his classic (1974), (1975).
2 3
All references to Diels and Kranz (1968) (DK). Cf. Owen (1975), 54.
4
This is frequently associated with Parmenides’ arguments: cf. Owen (1975), 61. I argue here that
Parmenides’ position is much trickier than Descartes’.
5
Cf. Williams (1978), ch. 3.
6
‘ “I am thinking” . . . will make a true assertion whatever mode it is asserted in, publicly or merely to
myself. It can be true, of course, even if it is not asserted at all but merely if it is entertained or considered or
doubted . . .’ Williams (1978), 74.
74 PARMENIDES ’ DILEMMA
revelation to the Kouros, is insistent, with particular emphasis in fr. 2. What makes
‘you think’ self-evident? Our own reflectiveness gives us no direct evidence for the
reflectiveness of others, or even for the existence of other minds.
But Parmenides’ claim is axiomatic within a dialectical context.7 When I say to you
‘you think’, you might ignore me, or you might not even be there at all; thus my
statement would be falsified. But if the context is already dialectical, then we are
already arguing, I am speaking, and you are paying attention to what I say. So when
I say to you ‘you think’, you may either agree or disagree, in which case your very
disagreement expresses ratiocination and is thus self-refuting. Still, provided ‘think’
carries no implications of success, ‘you think’ or, to generalize, ‘you can think (or
speak)’, counts as a rule of dialectic.
To this, Parmenides adds a further premiss: ‘either esti or not-esti’ (fr. 2). So long
as esti is construed as a complete sentence, this disjunction is trivially true. And we
may shortcut the scholarly cruces here.8 For despite early ambiguities, the conclu-
sion, esti, is taken as a strong ontological claim in fr. 8: to eon is all there is, one,
homogeneous, eternal, and immoveable. The argument, therefore, is about existence,
what there is in the world.
The conclusion of strong monism is derived from the disjunctive premiss, ‘esti or
not-esti’, by the rule of dialectic—perhaps thus:
We think. That thought must have content,9 for otherwise it would be vacuous, and could not
be said to occur at all. We think about something,10 and not about nothing, which would be
nonsense. So, if the choice between ‘esti’ and ‘not-esti’ is reinterpreted as a choice between
‘something’ and ‘nothing’,11 we must take esti. What can be thought, is, and nihilism is false.
For whatever its status, something is happening, even if the event only occurs in my brain.
Moreover anything other than that ‘something’ must be ‘nothing’.12 So differentiation is
impossible, and strong monism may be thought to follow.
7
Cf. Burnyeat (1976a), (1976b). By dialectic, I mean to suggest two conditions: (a) that the process of
discourse be carried on by more than one person. There is a questioner and a respondent. The latter may
only listen; but nevertheless, dialectic supposes a more or less attentive audience. This condition is
independent of but probably generates (b) that the philosophical outcome is tentative. Its success depends
not on ‘getting the right answer’, but on the exploratory process itself. Hence Aristotle’s description of
dialectic as peirastikê, the art of trying and testing; Met. 1004b25.
8
Notably: (a) What is the subject of esti?.— cf. Owen (1975), 60, ‘What can be spoken or thought’; (b)
What is the meaning here of the ambiguous verb einai, ‘is’? cf. Kahn (1973).
9
N.B. this falls short of the claim that thought must be about a real object—the latter would not be
warranted by the rule of dialectic, which does not imply success.
10
Cf. Plato, Republic 476e, and the Greek idiom for talking nonsense, ouden legein.
11
This, of course, is dangerous—for the scope of ‘not-esti’, ‘does not exist’ seems far wider than the
scope of ‘nothing’. Both, however, may be construed as a negative existential quantifier—dangers lurk for
the construal of ‘nothing’ as a noun. Cf. Wiggins (1971).
12
In this version of Parmenides’ argument, this is the shaky move. The equivocation whereby ‘what is
not’ = ‘what does not exist’ and ‘what is not something else’, ‘what is different’ is diagnosed in Plato’s
Sophist.
PARMENIDES ’ DILEMMA 75
On this basis, without having assumed ‘real objects’ of thought, Parmenides goes on
to describe the nature of what is. He follows a programme laid down at 8.3–4: what is
is ungenerated and imperishable, whole and unique, immoveable and complete.13
This programme is followed in the sequel. Lines 5–21 refute the possibility of
generation and destruction,14 22–5 describes the temporal invariancy15 of what is,
26–33 its immobility, and finally, after a break at 34–41, 42–9 argue for its spatial
completeness and homogeneity, that it is ‘like the mass of a well-rounded sphere’.
Parmenides is indeed offering an argument for strong monism—a single homoge-
neous being is all there really is.
Dialectic is the critical factor. Thus Parmenides’ argumentative fragments adduce
‘thinking and speaking’ with the explanatory gar (2.5) supplemented by modal
operators (2.5; 6.1–2). The dialectical context is introduced by the myth of a dialogue
between the goddess and the Kouros (Parmenides himself). But this conceit recedes
into the background, and Parmenides appears to argue directly with the reader, who
thus becomes his interlocutor throughout the Alêtheia (significantly, this device
lapses in the Doxa). And the rules of dialectic are, as I have suggested, self-verifying.
But what is dialectically self-verifying may not be true. In a world where there were
no thinkers, or only one, ‘you can think’ would lose its reference, and become
vacuous or false. It would certainly no longer be true, let alone axiomatic, if there
were no-one there to assert it and no-one to listen. Under such conditions—a solitary
mind, or no minds at all—there can be no dialectic between two persons. And
Parmenides’ Alêtheia argues for just such a world, where all distinctions are vacuous,
and the predicate ‘other’ falls into disuse—so there are no other minds and no
dialectic.
Perhaps, then, Parmenides is arguing for solipsism. But solipsism is, in the first
place, not the premiss of the argument—rather, the assumption is that dialectic can
take place. What is more, solipsism generates as much unease as the denial of
dialectic. If I conduct with myself the argument that Parmenides suggests, I find
myself prey to the kind of reduction whereby not myself but my context is
disallowed—just as frightening an alternative as the conclusion to the dialectical
argument. And that now looks self-refuting, a reductio per impossibile wherein a
group of premises generates a conclusion that contradicts one of those premises. This
kind of self-refutation16 renders the argument of the Alêtheia intolerable.
The charge of peritropê has made the commentators uneasy. Some shrug accept-
ance;17 others adduce Sextus Empiricus and Wittgenstein18 and suggest that the
13
Following Owen’s reading: oulon mounogenes and êde teleion (1975), 76.
14 15
Cf. Kahn (1969). Thus Owen (1975), 63.
16
This is not the ‘dialectical self-refutation’ of Burnyeat (1976a). In that case, it is the context, outside
the argument, that does the damage; here, I argue, dialectic is incorporated into the argument as a premiss,
so that the conclusion is a straight reductio per impossibile.
17
E.g. Long (1970); Nussbaum (1979). Matson (1980) exploits this aspect of the Alêtheia.
18
Owen (1975), 67; the reference to Sextus may be particularly fertile.
ladder, once climbed, may be discarded without detriment to the argument. But the
conclusion invalidates the premiss, and vitiates the argument at its dialectical core.
Did Parmenides realize that there was a worm in his apple, a self-refutation fed and
grown fat of the emphasis of his words and his method itself? How could he miss it?
2. Alêtheia and Doxa

So the first part of Parmenides’ poem entails the rejection of the phenomena of
the physical world. The second part of the poem, the Doxa, was at least as long as the
first, if we may judge from the surviving fragments and the copious evidence of
the doxographers. It describes at length a world that has already been rejected.19 Yet
the Alêtheia provides Parmenides with an efficient weapon against his pluralistic
contemporaries; and by the same token, it refutes the cosmology of his own Doxa. If
the Alêtheia is right, the Doxa is wrong and to expound it futile. So why write the
Doxa?
Explanations of the Doxa rely upon some problematic lines for their support:
1.30–2; 6.4–9; 8.50–2; and 19.20 In these passages, Parmenides reasserts the decep-
tive nature of the Doxa on the one hand (1.30, 8.52, and perhaps 8.60) and, on the
other, offers a reason why the Kouros needs to know it (8.61)— ‘in order that no
mortal opinion may outstrip you’. The paradox of the procedure—compare the
strange proposition ‘I am now going to tell you a lie’—matches the obscurity of its
rationale.
Some commentators have suggested that the Doxa is a systematized version of
some current views of cosmology;21 others have argued that it is intended as a
paradigm of bad reasoning;22 yet others have supposed it to be a second-best view
of the world.23 The first and third of these views beg the objection that such second-
rate pluralism is merely silly, coming from a man who was primed with a first-rate
anti-pluralist argument. Why should Parmenides play by the cosmologists’ rules?
19
The problem with the Doxa may be that it uses the forbidden not-esti; or that those engaged in it
always contradict themselves; or that those engaged in it sometimes contradict themselves. (a) suggests,
perhaps, that there is something fishy about the content of the phenomenal world; (b) and (c) that we can’t
talk about the phenomenal world. That is, Parmenides may be claiming that the World of Doxa is an
illusion; or that it is contradictory. The evidence of 6.5–9 and 8.39–41 seems to weigh in favour of (b) or (c),
while 8.54 might support (a); (c) could be a version proffered by Zeno. Woodbury (1958) argues for
contradiction, while Matson’s attempt to resuscitate the Doxa exploits just this problem. I shall argue below
that the choice between illusion and contradiction is an important structural feature of Parmenides’
philosophy.
20
For commentary, see Long (1970); Mourelatos (1970); 194 ff., Owen (1975), 49 ff.
21
Cf. Long (1970) on Burnet’s Pythagorean explanation.
22
Mourelatos (1970), 260; Long (1970); and Owen (1975), 54. But would it not be more sensible to have
a paradigm of legitimate argument?
23
Cf. Long (1970), and thus, if I understand them correctly, Kirk and Raven (1966) (KR) 278 ff. It
passes my imagination, however, how unequivocally wrong views of the world can be rank-ordered in the
way such an interpretation requires.
And if he does, is he not thereby disqualified from his own ball-game? The second
account of the Doxa, the ‘dialectical24 device’, is more appealing. But how would it
function? Is a paradigmatically bad argument compelling enough for dialectical
effect? The audience, primed with Alêtheia arguments, might reject it out of hand,
and never win the dialectical prize. So perhaps the Doxa is just a bad dialectical
device. But in that case, we must still account for the doxographical evidence that
attributes the opinions of the Doxa to Parmenides himself.25
An alternative move might be to allow the cosmological implications of the Doxa,
but to reinterpret the function of the Alêtheia.26 If the Alêtheia were not an onto-
logical argument but, for example, a methodological framework, and the Doxa a
reasoned description of the world unaffected by the logical restrictions of the
Alêtheia, then perhaps the tension that otherwise stretches between the two ways
could relax. Such an interpretation may be attractive, both as a last resort in
understanding Parmenides and as a radical new stage in the early history of phil-
osophy. But to be effective, it requires a total divorce between logic and reality, which
we are bound to resist. And it cannot explain the ontological conclusions of fr. 8. So
what is the relation between the Alêtheia and the Doxa?
3. Dualism and Antithesis

The Doxa describes a cosmological system based on dual principles.27 A series of
antitheses explain cosmological and anthropological phenomena—light and dark,
left and right,28 being and not-being.29 Hence commentators have been tempted to
assimilate the Parmenidean cosmology to the Pythagorean table—one of the escape
routes from the problems we are addressing. But if that were the correct interpret-
ation, the use of antithesis should be restricted to the Doxa—and that is far from
being the case. The Alêtheia is itself antithetically contrasted with the Doxa,30 and it
begins with an antithesis, its disjunctive premiss ‘esti or not-esti’. The Doxa exploits
the same pair of terms, connected by the converse truth function, conjunction. From
the two antitheses (‘esti’ and ‘not-esti’; conjunction and disjunction), Parmenides
manipulates his options to provide three theoretically possible ways—Alêtheia, Doxa,
and the unthinkable way of not-being. Antithesis, therefore, far from being a
24
Here, I take it, the tentative nature of dialectic is being exploited. Thus Parmenides would not be
committed to the substance of the Doxa but only, perhaps, to its effect on his audience.
25
Cf. e.g. DK 28A23; 24; 33; 34; cf. Long (1970), 83. This tradition could derive from an original mistake
on Aristotle’s part that in the Doxa esti and not-esti are to be identified as the cosmological principles
fire and earth. This must be over-schematizing on Aristotle’s part. Nevertheless, his telling evidence is the
comment ‘compelled to follow appearances’ (Met. 986b31). That is Aristotelian enough, but it cannot be
explained away as part of the business of fitting the square peg of Eleaticism into the round hole of early
science.
26
Cf. Mourelatos (1970); Furth (1974).
27
Cf. Long (1970), 91; Mourelatos (1970), 132; Reinhardt (1974), 310.
28
E.g. 9.1; DK 28A37. 29
8.41, 6.8. 30
Cf. the conjunctions êmen . . . êde at 1.29–30.
Pythagorean side-effect, is fundamental to the structure of Parmenides’ entire

philosophy.
The world of Doxa is the world of dualism; correspondingly, those who wander
there are disparagingly characterized as being in two minds (di-kranoi, 6.5; postu-
lating two forms, 8.53; travelling a road that is polupeiron—‘much-experienced’?
pluralist, at least—7.3). But these dual or plural descriptions are not restricted to the
Doxa.
The proem describes the descent of the Kouros led by poluphrastoi31 horses to the
doors of Day and Night—dualism again—which are guarded by Dikê (‘Justice’). This
goddess is described as polupoinos (‘avenging’, ‘rewarding and punishing’32); she
holds the keys to the gates, keys that are amoibous (‘of retribution’33). I have argued
elsewhere34 that we should be wary of the association of dikê with ‘justice’. When dikê
describes a state of affairs, it expresses a balance of power.35 Now for there to be a
balance of power—as Anaximander realized—there must be two sides: Achilles and
Agamemnon, hot and cold, day and night. And thus for Parmenides dikê is pluralist
(polupoinos) and reciprocal (amoibos36), balancing two sides out.37 But what are
these two sides? We need a balance of power appropriate not to the pluralism of the
Doxa, but to the subsequent revelation of the goddess.
Parmenides’ remarks on method suggest just such a balance of power. At 6.9
mortals are said to follow a palintropos keleuthos, a ‘backward-turning path’. This is
the Doxa. Fr. 5, however, says: ‘It is the same (xunon) to me where I begin, for I shall
come back (palin) to the same place’. This is usually taken to declare the logical
coherence of the Alêtheia38—but, of course, it describes not coherence, but circular-
ity, which is logically vicious. And circularity characterizes the argument to establish
monism, for it is self-refuting.
So both the Alêtheia and the Doxa are backward-turning. Now the Alêtheia could
be backward-turning in the right way, the Doxa in the wrong; so that the epithet
‘backward-turning’ is in itself neutral. But in that case, any point to the resonance is
31
1.4. Should KR give ‘wise’? The word is not only pluralist, it is rare, and so striking. It may be
associated with poluphradês (LSJ) which, at Hesiod Theogony 494, describes a deceitful plan; but such
epithets may decry neither plan nor planner; cf. Od. 13.291 ff., Détienne and Vernant (1978). The -phrastos
termination signifies ‘expressed’, ‘expressible’ (I owe this to G. E. L. Owen). So poluphrastos means ‘much-
expressed’? See section 4 on the trickiness of dialectic.
32 33 34
KR, 267; and Fränkel (1970). Mourelatos (1970), 26. (1981), ch.7.
35
In Homer at Iliad 19.180, for example; or in the fragment of Anaximander.
36
Cf. Sophocles, Antigone, 1067.
37
What of dikê at 8.14, which holds being fast, and does not allow generation or destruction? On this
interpretation, we do not have some kind of ‘cosmic justice’; dikê would have a different, and suitably
argumentative, function. The previous lines have shown that generation is impossible. The argument
appears to assume that destruction has been ruled out also (cf. Kahn (1969)), although without argumen-
tative warrant. But the antithetical view of dikê would supply just such an argument—dikê rules out
destruction also, by parity of reasoning; as a matter of antithetical balance, the argument against generation
holds good also for the argument against destruction. Therefore, to eon is held bound between the two.
38
Thus it picks up the description of Alêtheia as ‘well-rounded’ (1.29) and looks forward to the analogy
of to eon with ‘a well-rounded sphere’. But cf. Jameson (1958).
lost. Alternatively, it may be that the term is intended to express, for both the Alêtheia
and the Doxa, the same sense of puzzlement or even exasperation. Could that be so?
The task of reason is to discriminate—krinein. Cognates of this word occur five
times in the extant fragments: akrita phula, 6.7; krinai de logôi, 7.5; krisis, krekritai,
8.15–16; tantia d’ekrinanto, 8.55. The first and last of these describe the ways of
mortals in the Doxa; the other three refer to early stages in the process by which the
Alêtheia is derived.
The verb krinein means, originally, to distinguish, to separate, to pick out or to
choose.39 Hence Parmenides means to say, first, that to progress along the way of
Alêtheia, we must mark off esti from not-esti (8.15–16). This basic discriminatory
move begins the journey, and it is contrasted with the failure of mortals to make any
such distinction at all (6.7). Thus the poem as a whole requires that we be discrim-
inating, judging by the argument40 the ‘contentious refutation’ (poludêrin elegchon,
7.5). So, discrimination is needed not only within the Alêtheia, but also, at the second
level, between the Alêtheia and the Doxa. And in all these contexts, the operation of
distinguishing presupposes two sides to be distinguished between—discrimination
operates on a balanced antithesis, hence the association of krisis with dikē at
8.14–15.41
However, if discrimination is the key to the Alêtheia, it jars when associated with
the ‘premises’ of the Doxa: ‘mortals distinguished opposites’, 8.55. In the Alêtheia,
krinein is sound intellectual stuff; but here it degenerates into a base process of Doxa.
This methodological bathos is given ironical point by fr. 16, apparently a physio-
logical explanation of thinking within the Doxa.42 That fragment espouses mixture,
krasis, with multiple puns on discrimination, krisis, and the monism of fr. 8. In a
more serious tone, at 8.34–41, Parmenides contrasts the superficial distinctions of
mortals with the homogeneity of what is. And here lurks paradox. Parmenides’
thinkers, those involved in dialectic, discriminate Alêtheia from Doxa. Then,
embarking on the Alêtheia, they use their powers of judgement to reach the conclu-
sion that there is only to eon; and if monism, all distinctions are vacuous. Unthinking
mortals, on the other hand, are undiscriminating just because they make the distinc-
tions the Alêtheia denies. They wander because they postulate generation and
destruction, being and not-being, and because, Cartesian-like, they suppose that
they themselves are thinking and discriminating, when all there is is to eon.43
So there is more to the resonance of krinein than mere irony. The reiteration of the
word points out the paradoxes of both ways. The argument of the Alêtheia is
backward-turning because it relies on a premiss that is falsified by the conclusion.
39
Cf. L.S.J. s.v. I, II. 40
Cf. Verdenius (1966).
41
Cf. also the legal overtones of elegchos—cross-examination or scrutiny. Pi. Nem. 8.21, Antiphon 1.7
etc.
42
Cf. Vlastos (1946).
43
This indiscriminate discrimination suggests that the problem for mortals is that they are committed
to contradiction.
Thinkers are committed to peritropê because, by discriminating, they cancel out all
distinctions, including their own identity. The indiscriminate mortals of the Doxa, on
the other hand, contrive, by avoiding the way of rationality, to preserve distinctions
between themselves, and hence their own identity.
4. The Dilemma
Is this over-interpretation? Surely not, since Parmenides chose the resonant44
medium of poetry for his work. His literary ability is deplored, and his choice of
poetry seen as a blunder.45 But poetry has one great advantage over prose, namely
that it permits, even encourages, deliberate echoes and wordplay. Thus the audience
of a poem may be exploited with verbal trickery and suggestiveness by the author,
where the reader of a piece of prose would remain immune. And the contrivance of
poetry is thus particularly suited to some philosophical jobs, notably the gadfly work
of dialectic which goads the reader into thinking for herself. By constructing an
antithetical argument, counterbalancing the Alêtheia and the Doxa in the self-
conscious literacy of a poem, Parmenides poses a dilemma.
The Alêtheia uses reason independent of sense-perception, to produce a monistic
account of reality. But rational though it may be, it denies the separate existence of
the agents of rationality, upon whose mutual discourse the argument is based. The
paradox arises, therefore, from the assimilation of the thinker into what is. This
assimilation is required by the strong monistic thesis: what is is homogeneous
and one.
So the self-refuting character of the Alêtheia is the vehicle of paradox: ‘If we can
think, we, as independent persons, cannot exist’. Balked, we might retreat to the
phainomena, reassert our own existence, and construct our world-picture from there.
The Doxa offers just such a retreat—affirm the irrational and undiscriminated
conjunction ‘esti and not-esti’ and you can reinstate the world of sense-perception.
But a price must be paid. That conjunction is indiscriminate because it is a contra-
diction and thus irrational; or because, by admitting non-existence, it lowers the
degree of reality of the phenomenal world. If we think, then we cannot exist; if we
exist, we are either irrevocably muddleheaded, or our very existence is somehow
illusory. But to reassert our existence, and then concede its illusory nature, looks like
yet another contradiction. And thus the dilemma revolves around the relation
between thinking and being. Monism requires that thinking be indiscernible from
being, and denies that there are a plurality of thinkers. That is fatal to dialectic, and
thus to the argument itself. Pluralism allows for individuation, but commits us to
the irrational. This dilemma is the relation between the Alêtheia and the Doxa.
The tension between the two is not only a necessary consequence of Parmenides’
44 45
Cf. Kahn (1979a), 3 ff. Cf. Barnes (1979), 155.
argument, it is also a consequence that he foresaw and wished to exploit.46 He subtly

expounds the conflict between what we derive by reason and what our everyday
experience requires us to believe. Both approaches to ‘reality’ demand a sacrifice—
either to preserve reason or to save the phenomena. Neither course is without its
attendant evils.47
5. Truth and the Alêtheia

There is a final difficulty.48 If Parmenides supposes that the Alêtheia is self-refuting,
how can he claim that it is true?
It may be that truth is not an epistemological but a metaphysical matter49—that is,
states of affairs are true if and only if they are the case, irrespective of whether there is
anyone there to describe them. Now Parmenides’ dilemma may readily be under-
stood in terms of such a metaphysical account of truth. The Alêtheia describes the
fact of the matter—and by doing so it rules out the possibility of there being anyone
other than that fact there to describe it. But that makes no difference to its being true,
on a metaphysical account of truth. That the argument is self-refuting in a dialectical
context does not affect the reality it describes.
The Greek notion of alêtheia, however, is not strictly metaphysical. The word is
derived from the verb lanthanein (‘to escape notice’) and the cognate abstract lêthê
(‘forgetfulness’).50 Both the prephilosophical and the paraphilosophical traditions
associate alêtheia with intellectual exercise, telling the truth.51 So the metaphysical
account of truth—facts of the matter independent of anyone to describe them—fails
to capture the epistemological tone of the Greek idea, whereby truth may be a matter
of the coherence of our beliefs. To a Greek, truth is told, by someone, to someone—
hence Parmenides’ original assumption of dialectic. There is, then, an intimate
connection between truth and the rules of dialectic.52 Correspondingly, we may
expect an epistemological contrast between the Alêtheia and the Doxa. This may
be found in the dilemmatic interpretation I suggest. The Alêtheia is the way of
rationality, the only way to make sense of our speech and thought. Arguing rationally
may lead us to a coherent, and hence true, view of the world, but it cancels the
46
Thus the ‘deceitful ordering’ of his work (8.52); hence the plurality of the roads, the ‘much-expressed’
revelation, and the balance of power. See also the later tradition (D.L. IX. 22, Simpl. de caelo 557.20) that
Parmenides proposed a double (dissên) philosophy: dissos is used of a dilemma at Sophocles Electra 644.
Thus Zeno countered absurdity (the Eleatic thesis) with even greater absurdity (pluralism) and defended
his master (Plato, Parm. 128)—and this is the procedure of dilemma.
47
So Agamemnon, in the classic dilemma at Aulis, Aesch. Ag. 211.
48
This was emphasized to me by Myles Burnyeat.
49 50
Cf. Annas (1980). See Ross (1953) on Met. 983b2. Cf. Mourelatos (1970), 64.
51
E.g. Iliad 23.361, 24.407; Thuc. 2.41–2; Aesch. Suppl. 628; Pi. Ol. 2.92.
52
This epistemological connection is apparent in every occurrence of Alêtheia and its cognates in
Parmenides. See 1.29, 30; 2.4; 8.17; 8.28; 8.39 and 8.51; and compare the attribution of literal error to
mortals at 8.54. Cf. Jameson (1958) on the relation between Alêtheia and the terms for persuasion and
conviction, pistis and peithô.
reasoners out. The Doxa, on the other hand, allows us speech, but denies us reason—
if we can speak, there are contradictions, and so no truth.
Thus Parmenides’ dilemma may be framed in either metaphysical or epistemo-
logical terms, as a question of the correspondence between what we say and what
there is; or as a question of the internal coherence of what we say. As a corollary, the
phenomenal world may be suspect because of what is in it, or because of how we talk
about it—either way, the dilemma is comprehensive. It divorces human existence
from human rationality. Existence is a metaphysical issue; and the metaphysical
interpretation of Parmenides’ work has been favoured from Plato53 onwards. Ration-
ality is, among other things, an epistemological issue; and the epistemological
interpretation of Parmenides’ philosophy is favoured by the terms Parmenides’
uses, and by the premises of the Alêtheia. Take truth either way, and the dilemma
applies.
Parmenides was neither a nihilist nor a sceptic.54 Rather, he fits in the tradition of
dialectic, the paternity of which Aristotle attributed to Zeno.55 So Parmenides is a
working philosopher, challenging our rational apparatus with the hardest antithesis
of all—the paradoxical relation of reason and perception. The Doxa is indeed a
dialectical device, but only half of it, incomplete without its complement, the self-
refutation of the Alêtheia.56
53
Cf. Sophist 243c ff. Cf. the dominance of correspondence over coherence theories of truth up to the
Hellenistic period.
54
Although such antithetical arguments became typical of the later sceptical tradition, cf. Sextus
Empiricus, Outlines of Pyrrhonism. For this reason the mention of Sextus in connection with the problem
of self-refutation in Parmenides becomes particularly telling.
55
Cf. Sextus, M 7.6.
56
I should like to thank Myles Burnyeat, Jonathan Lear, G. E. L. Owen, and Malcolm Schofield, who
have read and commented upon drafts of this chapter. The chapter as it stands is not their fault.
5
Myth, Allegory, and Argument
in Plato
1. Myth and Allegory

If the job of a philosopher is analysis, maybe she has no business telling stories; if
philosophy is about argument and truth, it should shun myth and allegory. Now
when Aristotle writes philosophy, he writes arguments. When Plato writes philoso-
phy, he writes stories, myths, allegories. Inside Plato’s stories, if we are lucky, skulk
some arguments, aimed at some unfortunate interlocutor and only indirectly pertin-
ent to our philosophical concerns. Aristotle’s arguments, by contrast, are directed
straight at his audience; they are immediate and accessible. So Aristotle is easier on
the philosophical eye. Why didn’t Plato write like Aristotle? What is wrong, though,
with philosophers telling stories?
a. Truth?
First of all, whatever a myth is, what it isn’t is straightforwardly true—nor is it
directly false. If I tell you that I caught a bus last week, then I might be telling you the
truth—and if I said I met the man in the moon, you might say I was just lying (or
mad). But if I tell you the story of Odysseus and Polyphemus, you may think I’m
telling a myth. Whatever that means to you, perhaps you won’t be worried about its
truth-value; the important question about Odysseus and Polyphemus is not whether
it really happened or not, but (perhaps) how the Greeks thought of barbarians, or
why they found Odysseus’ joke about ‘Nobody’ funny. So however outrageous,
myths still matter—perhaps they explain how we understand ourselves, often against
an alien background; their oddity may help the explanation along, rather than getting
in its way.
b. Poetry
So myths and allegories are not literally true. But if myths are stories that hide the
truth (or just miss the point altogether), has the philosopher any business using
them? Plato’s attacks on the poets suggest not. The only stories we should hear, he
suggests in the Republic (376e ff.), are those that are true (and uplifting). The trouble
with the poets is that they offer us mere imitations, at the second remove from truth
84 MYTH , ALLEGORY, AND ARGUMENT IN PLATO
(597e). So the poet either knows what he imitates or not (599a ff.). If he knows, why
not tell us; if he doesn’t know, why bother to talk at all? (cf. Osborne 1987a). The
trouble with poetry is that it has no devices for distinguishing the true from the false,
and it has no justification for the false at all. Poetry contains the trap of the soap-
opera, where we become so absorbed in the doings at Ambridge or Albert Square that
we cannot tell where fiction ends and reality begins. We are prisoners in the televisual
cave, staring at the screen and never discriminating between the truth and the lies.
c. Imagery
Perhaps, then, myths are neither true nor false, but distanced from reality by being
images of what is real and what is true. They are, that is, fashioned to suit the
inadequacies of belief, the state of mind of people enmeshed in the sensible world,
susceptible to the persuasive words of the poets and orators, who observe what is
likely at the expense of what is true. Allegory (one might think) behaves in rather the
same way. Look how Plato offers an allegory—the sun—for the intelligible world
dominated by the form of the good:
But, my lucky friends, let us leave for now the question of what the good itself is, for it seems
to me to be beyond the scope of our present attempt to arrive at what I think about that now;
instead I should like to describe something that seems to be an offspring of the good, and very
like it, if that is pleasing to you; if not, I am happy to leave it.
No, describe it, he said; you will pay your due to the parent another time.
I wish I could pay my dues, I said, and you could collect them, and not just the interest [the
offspring]. So collect this interest and offspring of the good itself. But take care lest against my
will I deceive you, giving you a false-coined account instead of the interest.
We will take care as best we can, he said; but just tell us. (Republic 506d–7a)
Here the point is that the sun is not the form of the good. The allegory works because
there is not a direct assimilation of the allegory to what it represents; instead, there is
a distancing effect between the image (the sun) and its reality (the form of the good),
a distancing effect that is, we might think, somehow fundamental to the impact of the
allegory.1 That is all very well—but images (Plato is often thought to have said) are a
bad thing, the mere appearances of the sensible world, which obscure the reality
behind and get in the way of the philosophical soul. How on earth are we to
understand them?
d. Cash-value
We might want to treat myths as crude versions (or subversions) of reality. Perhaps,
then, we can cash the myth in for its corresponding truth? But Socrates argues against
being too clever. At Phaedrus 229b ff. Phaedrus asks whether this is the spot where
1
Compare the effect of the ‘Chinese Whispers’ introductions of the Symposium and the Parmenides: a
told b; b told c; c told d . . . about this meeting between Socrates and his interlocutors. Cf. McCabe (1994a),
ch. 4.
MYTH , ALLEGORY, AND ARGUMENT IN PLATO 85
Boreas raped Oreithuia; no, not here, but a mile away, responds Socrates (is this myth
a fact?). How should we interpret the myth, asks Phaedrus; is the story true? We
could, of course, rewrite the myth as the history of a girl being swept away by the
wind. But who cares about that, says Socrates, when he still does not know himself ?
He has no time for this alien story:
If someone disbelieves in these [mythic monsters] and wants to reduce each thing to what is
likely in each case, using as it were some rustic wisdom, he will need a lot of spare time. But
I have no spare time for things like that; and this, Phaedrus, is the reason. I am not yet able, as
the Delphic saying has it, to know myself—it would seem absurd to me if while I was ignorant
about that, I investigated things that belonged to other people. So I let these matters [monster-
stories] alone and just believe the common view about them; and, as I just said, I investigate not
them, but myself, to see whether I am some kind of beast more complex than Typhon and
more bad-tempered, or a gentler and more simple creature, sharing by nature in some divine
lot, not like Typhon at all. (Phaedrus 229e–230a)
So Plato has good reason to reject the images and stories of the poets, which offer
reality at several removes. Instead perhaps he should have written like Aristotle,
eschewing myth and imagery; maybe he should have got on with the Delphic
business of knowing himself.
But he doesn’t—he writes myths—myths, what is more, of his own (no room for
the anthropologist here), and allegories and stories as well as arguments. And he
presents his stories in the terms I have described.
The myths are alien, they do not belong—hence, for example, the Phaedo’s myth is
second-hand (107d); the cosmology of the Timaeus does not belong to Socrates, but
to Timaeus; and in the Phaedrus Socrates describes the soul under the influence of
magic, and his story belongs not to Socrates, but to Stesichorus (242c ff.).2
And the myths are images, not reality. The sun is a child of the good, a likeness of
reality (Rep. 509a), similar to the good but not identical to it (509c). A charioteer with
two horses is an image of the soul (Phdr. 246a), accessible to human ignorance. And
the cosmologies of both the Phaedo and the Timaeus are ‘likely stories’ (Phd. 108c,
Tim. 29c–d). Why should we take any of them seriously?
Perhaps we shouldn’t. Perhaps we should attend to the health warnings on myth
and allegory—‘beware of what follows’. Remember the equivocal introduction to the
sun allegory: ‘take care lest against my will I deceive you, giving you a false-coined
account instead of the interest’. It is as if Plato said, ‘now I shall tell you a lie’. Are we
to believe him? Or are we to believe the lie? Can we now believe the lie? What is the
point of this sleight of hand?
One (obvious and rather dull) answer might be that Plato wanted to blandish us
into reading his arguments—so he dresses them up in literary finery to seduce us into
philosophical attention. Or perhaps his myths and allegories do what he says they
2
Cf. McCabe (1988b) on the requirement that explanations belong to their explananda.
do—they illustrate the points in his arguments that he cannot demonstrate or argue
for fully—literature supplies what philosophy lacks. The Platonic texts, on that view,
are complex vehicles for philosophical dogma, instruments of persuasion, not weap-
ons of argument. They are put together to persuade, not to demonstrate; and the
remarks the texts make about rhetoric or poetry or deceitful propaganda are ad
hominem attacks on Plato’s opponents, not taken to apply to his own procedures.
2. Argument
Where does that leave argument?
a. Truths with reasons
An argument may be thought of in Aristotelian mood, as a connected series of
propositions, where the earlier—the premises—give reasons for (they cause) the
later—the conclusion. Thus, for example,
All creatures with trunks are pachyderms
All elephants are creatures with trunks
All elephants are pachyderms.
Arguments like this (as Aristotle was the first to observe, e.g. at Prior Analytics 26b26
ff.) have two important characteristics: first, they are valid in a perfectly obvious way
(they look valid); and second, their validity can thus be formalized (All B are A; All
C are B; So all C are A). If the premises are true, the conclusion will be true; the
conclusion is true because the premises are true. So the argument is explanatory. But
what that means is that when we have the argument, we understand the conclusion—
that is what understanding is—seeing why something is so, understanding the
reasons why it is true. Now suppose that arguments like this, with true premises
and conclusion, represent facts in the world; that is why the premises are true.
Understanding, we may say, is a state of mind (not a state of affairs in the world
out there). But if an argument directly generates understanding, understanding is
itself caused by the argument’s presenting ‘truth with reasons’. So the cognitive state
is directly consequent upon the argument itself—the argument is, we may say,
transitive; and understanding requires no further cognitive or psychological content.
Understanding is, as it were, a feature of the world, and nothing particularly to do
with me the understander, except in so far as I provide the head for the understand-
ing to be in . . . Is that right? Does understanding just happen when we come face to
face with an argument?
b. Explanation
That makes it all look rather too easy. Think again about explanation, the reasons
that lie behind the truths. How does explanation work? Plato (characteristically)
approaches this issue by posing some puzzles (in the Phaedo, 96 ff.):
Puzzle i) Imagine that Socrates is smaller than Simmias by a head. How can the
head’s difference between Socrates and Simmias explain both Socrates’ smallness and
Simmias’ tallness?
Puzzle ii) Think about ‘two’. How is it that instances of two may be explained
either by the division of some whole into two parts, or by the addition of one unit to
another, so that ‘two’ seems to have two opposite explanations?
The problem here seems to be this: neither the ‘head’ nor ‘addition or division’
explain properly, because in the first case we have the same explanation for different
phenomena; in the second case different explanations for the same phenomena. Can
such accounts really be thought to explain? Surely, instead, we need some strict rules
for explaining, so that we end up with the same explanations for the same phenom-
ena, different explanations only for different phenomena. We need to be direct, to
cite what exactly fits the explananda. When we explain, then we should say, for
example, that ‘the beautiful things are beautiful by virtue of the beautiful’, ‘the large
things by virtue of largeness’, ‘two things by virtue of two’. Now the explanation of
things’ being beautiful or large will be the form of beauty or the form of large. This
gives the proper account each time, and avoids contradictory-seeming muddles. This
Socrates calls his simple-minded answer.
This account of explanation works piecemeal or one by one—each item is
explained by citing its own explanation, by citing what belongs to it (it gives the
oikeios logos, the ‘explanation that belongs’). Does that sort of explanation work? You
can see that if an explanation does not belong to the explananda, it cannot explain
them. But then if the explanation does ‘belong’, does it do any explaining? The
trouble with the simple-minded explanation is that it may turn out to be so close to
what is explained that it never explains at all—simple-minded explanations are too
thin to give us understanding.
c. Understanding
So far I have posed two questions about explaining and argument:
i. (about explanation) Is the simple-minded answer too thin, so that it doesn’t
provide any real understanding at all?
ii. (about understanding) Is there no positive psychological content in under-
standing, so that argument is just passively received?
Plato’s account of full understanding tries to answer both questions by explaining
both the systematic nature of explanation and the active role played in understanding
by the mind.
d. Systematic understanding
Both Aristotle and Plato think that it is an important part of the philosophical
method to connect things together. In the Prior Analytics it is the combination of
the two premises that generates the conclusion. In the Republic the philosopher
understands the symbiosis of the forms, analogous to the natural connection of the
phenomenal world (hence the allegory of the sun, 506 ff.). In the Theaetetus (184 ff.),
reason contrasts opposites, compares similarities and thus comes up with the com-
mon terms such as sameness and difference. In the Phaedrus and later, the best way
to do philosophy is to find systems and structures—‘collection and division’.3
Why is connectedness, system, so important for Plato? If we can contextualize our
simple-minded explanations among other explanations, we come to understand how
it is that they explain. If we see that the beautiful and the good are connected; or that
being and unity are related; or that sameness is different from difference, then we
come to understand what sameness and difference, being and unity, beauty and
goodness are. (They are, if you like, mutually explicatory.) The first condition that
Plato offers for understanding, then, is connectedness, the interrelation of one form
(or one idea or whatever) to another.
But just connecting one explanatory item to some other is still pretty thin as an
explanation. Instead, Plato offers a more stringent condition for proper explanation,
that the connections should be exhaustive. Thus proper explanations turn up when
we grasp the whole system, the entire structure—and thus can explain anything by
contextualizing it among all the others. Thus, for example, the philosopher of the
Republic sees a whole intellectual world; and is infallible in his prescriptions for the
phenomenal state.
The notion of cognitive context offered by this sort of explanation can further be
understood by explaining how it is for the best. This is a teleological account of
explanation. Plato appears to hold that if something is ordered, its order is good
order. So understanding order is understanding good order; and ordered explan-
ations thus show how everything is ordered for the best. So full understanding is
grasping the order of things, for the best.
There are, then, three conditions for understanding, increasingly stringent:
i) connectedness; ii) within a complete system; iii) where the system is organized for
the best. This gives a ‘fat’ notion of understanding by contrast with the thinness of
the simple-minded answer.
The trouble is that fat understanding seems well-nigh impossible to acquire. Think
here about ‘Socrates’ autobiography’ in the Phaedo. Socrates complains about the
inadequate theories of his predecessors, in particular their failure to show how
everything was tied together and ordered for the best (their failure to offer a
teleological explanation). He was never more disappointed than when he read the
work of Anaxagoras, who postulated a divine or cosmic Mind—and then left it
nothing to do. What Socrates hoped for was an entirely systematic and exhaustive
3
Cf. McCabe (1994a), ch. 9, on the importance of collection and division in Plato’s account of personal
identity.
account of the universe; instead what he got were some mechanisms—with a spare
part, Mind, which ended up doing nothing at all.
. . . But I was dashed, my friend, from this optimism, when I read further and discovered that
the man made no use at all of Mind, nor did he connect any reasons together with the ordering
of things, but cited airs and aithers and waters and many other silly things like that . . . I should
then gladly have become anyone’s pupil to discover such an explanation [how things are tied
together by the best and by necessity] however it might be; but I was deprived of that, and
I could neither find it myself nor learn it from anyone else . . . (Phaedo 98b–c, 99c)
Now Socrates’ irritation with Anaxagoras is not just disappointment; after all, if that
were the case, there is nothing to stop this (fictional) Socrates from simply revising
Anaxagoras’ account. But that he does not do—instead he embarks on a different
investigation which culminates in the simple-minded answer itself.
Why is Anaxagoras such bad news, and the simple-minded answer so good? What
is at issue here, I think, is the nature of scientific theorizing. Anaxagoras’ theory failed
where Socrates’ answer succeeds—in its simplicity, its economy. The great advantage
of the simple-minded answer is that it is thoroughly economical, whereas Anaxag-
oras’ account is not only expensive, but even profligate in postulating explanatory
bits and pieces that do no work at all. William of Ockham4 later forged this complaint
into a razor—‘entities are not to be multiplied beyond necessity’—which shaves off
those entities that seem otiose, redundant, extra to the theory under scrutiny; and the
razor is still plied by modern theoretical physicists. Think—from the ancient world—
about an example from Aristophanes’ Clouds (365 ff.). Socrates says that the clouds
are divinities. Does he not think that Zeus, Poseidon, and the rest are gods too? Why
should they be? Socrates responds—we have enough gods with the clouds, they
explain the rain and everything else—there is no need for any more gods than
those. Socrates’ autobiography in the Phaedo makes a similar point (as indeed
Plato does elsewhere; compare Parmenides 130e). Extra entities should do some
work in a theory; if they don’t, there is no point in having them at all. Here are
Ockham’s classical origins.
Proper understanding, therefore, needs to be systematic, teleological—and mean
with its entities. How mean should it be? That is a matter of how you define
‘necessity’; Ockham’s razor does not produce bald reductionism, but forces us,
instead, to justify the entities we want—and that is a tricky business.
e. Active mind
Moreover, fat understanding is a cognitive state, an active character of the mind, not
just something that happens to the philosopher from the outside. When we think
about systematic understanding, we may be thinking about what is understood;
Plato, like Aristotle after him, supposes that explanation gives understanding because
4
As tradition has it—but see Kneales (1962) for a more sceptical view.
explanation is systematic. Aristotle, when he talks about demonstration and under-

standing, seems to suppose that explanations (structured arguments) are not only
isomorphic with understanding (which is structured in the same way) but that they
are transitive—if an explanation is complete and I ‘have’ it, then I understand. Here,
the active psychological element in understanding is minimal. (I have reservations
about whether this is true for all of Aristotle’s works—for example, the Rhetoric
makes great play of the active participation of the audience of a speech—but it is true,
I think, for the more strictly logical or epistemological works.)5
But for Plato there is a central psychological condition on both his logic and his
epistemology. At Theaetetus 184b ff., for example, Socrates offers a refutation of the
thesis that knowledge is perception. He argues that although the senses can perceive
their special objects, they are unable to grasp ‘what is common’. It is not by means of
the senses, that is, that we say that something ‘is’ or ‘is not’, or that we make
judgements about sameness and difference, likeness and unlikeness, one and many.
Instead, those judgements are the province of the soul, working by itself. What the
soul does is to calculate and to judge; the soul is active in reasoning, and under-
standing is done by souls, not just suffered by them.
But how? How does the soul come to understand? What is it that makes the soul
think for itself—or harder still, understand itself? Return now to argument. The
English expression ‘argument’ may mean ‘giving truths with reasons’, as we saw. It
can also mean a dispute, a debate, a controversy between two (or sometimes more)
sides. Controversies progress, typically, by the vigorous giving of reasons by either
side—so Kinnock may have an argument with Major about dirty water by giving an
argument about the relation between water privatization and inefficient purification
plants. We may give arguments to win arguments—at any rate, the giving of an
argument is generally a good tactic in controversy, since the argument given per-
suades us to believe the conclusion. Moreover, no good controversy is without
argument—the ‘did, didn’t’ brigade can hardly be said to argue at all. In practical
terms, then, the two senses of ‘argument’ are connected by reasons and reasoning.
This sort of argument has a counterpart in the ancient world—dialectic.
First of all we should approach the search for understanding by considering the puzzles that
should first be posed; these consist in the opposing views that people have taken on some issue,
even including some theses that have been left unconsidered. For asking a good puzzle is useful
for getting a good answer; for the later answer is a solution of the earlier difficulties, and it is
not possible to solve them without understanding what ties them together—as perplexity
makes clear for inquiry. For when we are puzzled, we suffer just as those who are tied up
suffer—neither of us can move forward. So it is necessary to examine all the difficulties first, for
their own sakes, and because someone inquiring without asking questions first is like someone
walking without knowing the destination, or being able to recognize it when it is reached. Such
a person has no clear objective; the person who has looked at puzzles first has. Moreover,
5
But see here Burnyeat (1981).
exactly as happens in court, the person who has heard all sides of the case must be better able to
judge. (Aristotle Metaphysics 995a24–b4)
Formally, a dialectical argument represents two lines of (deductive) argument, where

either the premises of both lines are compelling but the conclusions inconsistent, or
where the premises are exhaustive but the conclusions of either line untenable. The
use of such arguments in the ancient world was widespread—from Zeno through the
sophists and Plato’s Parmenides to Aristotle and beyond into the Hellenistic period.
Now dialectical arguments (pairs of dialectical arguments) could be destructive or
constructive. They destroy when, like the arguments of the sophist Gorgias, they
promote nihilism or despair. They build when they establish a thesis by refutation (as
Aristotle and possibly Zeno do), or where they force us to grapple with serious
philosophical problems. Plato’s Parmenides, for example, presents a series of paired
and apparently exhaustive arguments which lead to the conclusion that everything
and nothing is true. The effect of such a conclusion is to force the reader to re-
examine the arguments, and to think hard about a connected set of metaphysical
questions (what is it to be the same as oneself? What is it to be here or now?). In cases
such as this, the arguments come in pairs; the dialectical effect is achieved by each of
the pair acting as a critique of the other—simply because they represent radically
different ways of thinking about identity, or space, or time. From the logical or
argumentative point of view, then, the pairing of the arguments is fundamental.
But still, why present arguments in that irritating way? Plato’s use of dialectic runs
parallel to his understanding of understanding—since it emphasizes the critical
assessment of the arguments, and demands a state of consistency in the soul—and
that corresponds to the demand that explanation should be systematic.
Aristotle suggests that a crucial feature of the process of dialectic is learning to
judge between two lines of argument. The activity of dialectic is not necessarily to
change our basic assumptions (by making us believe, for example, that monism is
true), but to make us understand the assumptions we have, by judging between the
alternative arguments, and coming to a full explanation of why we think the way we
do. The crucial thing here, surely, is judging—that is when we understand.6 This view
runs through Plato’s account of philosophical activity. The Socratic elenchus, for
example, by the dialectic of question and answer forces the interlocutor to be critical
of his beliefs. The intellectual gymnastics of the Parmenides posit an unbearable
choice between pairs of untenable hypotheses. And the Sophist claims that thinking is
silent speaking, a dialogue within the soul. So opinions are stages in that dialogue;
and their truth-value is explained by the activity of the soul. It is the soul, not the
world, who makes mistakes, by making a bad fit, a bad connection—perhaps by the
wrong association of a predicate term with a subject or by a failure in consistency.
Mistakes, if you like, are a matter of imagination, not fact; and the explanation of
6
See McCabe (1994b).
falsehood must bring with it some account of what goes on in the mind. The dialogue
in the soul is a matter of making the right connections between the stages in the
arguments that go on in our minds.
Why is it so important that dialectic goes on in the soul? When Socrates confronts
his unfortunate interlocutors, he demands that they ‘say what they believe’; but in the
end they discover that what they believe is inconsistent. Why does that matter? It
matters because consistency in the soul (consistency of our beliefs) is a primary
condition of who we are, and a necessary condition for the health of the soul. When
Socrates is defending his emphasis on consistency against the sophists’ denial that
consistency matters (e.g. at Euthydemus 287), he presses the importance of the unity
of consciousness, as opposed to the cognitive discontinuity urged upon us by
Protagoras’ account of truth. Socrates insists that we can usefully ask the question
‘who am I?’ and receive an answer in terms of the consistency of my beliefs (if we ever
arrive at such a state). Here once again the central notion is one of fit, of systematic
connections and the unity that this then brings with it. But psychological unity
matters because that is a state of psychological health and good order—so consist-
ency is something we are driven towards for ethical reasons—because of what’s in it
for us.
The drive towards consistency is something the philosopher may exploit. Socrates
was an irritating man to meet, because he forced you to face the inconsistency in your
soul. Likewise, any dialectical argument insists that we untie the knot of the puzzles
because we cannot bear to stay puzzled; and we certainly can’t move then—or even
get away from Socrates. In the same way, Plato later suggests, the sensible world
pushes us towards understanding, because it confronts us with manifest puzzles.
Think about three of your fingers. Your index finger seems both large (compared
with your little finger) and small (compared with your middle finger)—how can such
a case explain what large and small are? Or hard and soft?
The sense which is about hardness must also be about softness, so that it reports to the soul that
it feels both hard and soft. . . . So in cases like that the soul is bound to be at a loss what the sense
means by hard, if it calls the same thing soft. . . . So it is plausible to suppose that in cases like
these the soul, summoned to calculation and reason, tries to consider whether there are two
reported items or one. But if there are two, they must each be one, different from each
other . . . and if each is one, the soul will think the two separately—otherwise, it would think
them to be one. (Republic 524a–c)
The challenge posed by such apparent contradictions is resolved by the soul in the
soul; and the challenge is dynamic just because inconsistency is unbearable, a logical,
metaphysical, and ethical black hole that we must and do resist.
I say blithely ‘we’. But of course the drama of a Platonic dialogue does not seem to
be about us at all—but rather about Socrates, Theaetetus, Glaucon, Simmias, and
company. Surely it is they who search for consistency—we just watch from the
sidelines.
Think about that claim again. The characters in Plato’s philosophical stories are
fictional; they have no souls, and no deep desires to arrive at consistency. The only
real live souls involved in the Platonic dialogues are the souls of his readers, us. How
does he involve us in dialectical argument (passive as we may be)? How, if our state of
soul matters so much, can we be brought to get our souls in order? How does the
passive reader become the active dialectician?
3. Myth and Allegory Again

Now return to myth and allegory, and consider three examples:
a. The allegory of the sun in Republic 6. The sun is the source of light and growth;
in the same way, the form of the good is the cohesive force in the intelligible
world.
b. The myths of judgement in the Gorgias, Phaedo, and Republic. Virtue is its own
reward because it is the best state of soul. And into the bargain, virtue receives
the rewards in the afterlife of eternal happiness and the release from the cycles
of incarnation.
c. The creation myth of the Timaeus. The world is structured as it is because it is
the product of a benevolent heavenly craftsman who reconciled the good sense
of reason with the pig-headed workings of necessity and produced the world as
we see it to be.
There is something fishy about each of these stories. They are, remember, images;
and they belong to someone else (to Timaeus or to the old tellers of tales in contrast
with, for example, the simple-minded answer which just belongs to what it explains).
What is more, these stories discourage our immediate assent: the myths are discord-
ant with the arguments of the body of the dialogues. First, the allegory of the sun, as
we have seen, declares its own doubts in its introduction—is Plato going to tell us a
lie? Second, the myths of judgement sit uneasily in their contexts: the Gorgias myth is
a long speech presented within a dialogue that attacks the rhetoric of speechifying;
the Phaedo myth turns up at the end of a dialogue designed to show that the care of
the soul is worthwhile even though we don’t know what will happen after death—and
yet the myth tells us what will happen after death; and the Republic myth details the
way in which justice receives its rewards in an uneasy coda to the arguments of the
dialogue to show that justice is its own reward.7 In each case, the straight arguments
of the dialogue argue for the importance of knowledge and the exculpation of the
vicious; and yet the myths insist both on the moral responsibility of the soul and on
its receiving its just deserts. And, third, the myth of the Timaeus is hedged about with
doubt, since it describes a world that is only accessible to the inferior cognitive
7
For more detailed argument cf. McCabe (1981), ch. 13; Annas (1982); Sedley (1990).
activities of belief—and claims just to offer an eikôs muthos, a likely story. So do the
myths matter? Do they have any philosophical function?
We might expect them to do some explaining, to give us some understanding. But
what principles of explanation do they obey? Scientific reasoning, I suggested, is cut
short by Ockham’s razor. But myths are not. There are no principles of economy that
govern mythologizing—you can have as many chimaeras and hydras as you like, so
long as it’s a good story. And so it is with Plato’s myths—the Sun, the eschatologies
and Timaeus’ cosmology are all profligate of both entities and explanations—they are
at the opposite extreme from simplicity and the simple-minded answer. How, then,
can they be integrated at all into the arguments?
Dialectic, I suggested, provided us with two opposed lines of argument. This
dialectical relation holds, I suggest, between the myths and the formal arguments
in the dialogues. Myths and straight arguments are counterposed because they obey
radically different principles of explanation; and thus, just as with the dialectic of the
Parmenides, each is a critique of the other. Thus Plato’s myths, so far from being
confirmations of what he says in the body of the dialogues, are challenges to what is
said there; and it is by the straight arguments that they are themselves challenged. So
the combination of myth and argument is itself an argument—of the dialectical sort.
It is left to the reader to untie the knot.
Hence, I suggest, Plato’s ‘image’ imagery. Earlier I canvassed the view that myths
were inferior entities because they are images and so mere appearances. But let us
consider again Plato’s approach to the philosophical dividend from an appearance.
The trouble with particulars is that they offer conflicting appearances (cf. Phd. 74 ff.;
Rep. 523 ff.). This does not entail that they are not really there (that kind of scepticism
is resisted, e.g. at Phaedo 100a, ‘Perhaps my image is not altogether right; for I should
not agree that someone who investigates what is in words (logoi—theories) is dealing
with images any more than someone considering what is in deeds . . .’), nor does it
imply that they are only partially what they are said to be. Instead, Plato’s problem is
that they are cognitively incomplete. My index finger is not an illusion, like the
Cheshire Cat, but a puzzle; and it is for that reason that it is held to be an image of
what is real (the form of large, for example). ‘Images’ in this sense are not unreal, not
illusions, but worrying. So then if a myth is an image, that may be because it offers
simultaneously the true and the false; it is not thus liable to disappear, but instead to
threaten our understanding. Images are not self-explanatory; they are cognitively
confusing, they conflate messages, they render the person who thinks about them
unable to distinguish between what is true and what is false. Images on their own, on
such a view, have the same effect as eristic arguments—they contradict each other
and fail to provide a proper and consistent view. Provide them, however, with the
proper explanation, and the contradictions fade; the philosopher can be a king in the
sensible world because the philosopher can determine and disentangle the puzzles
presented by the conflicting evidence there. Myths, then, may be arguments—
dialectical arguments, incomplete without their other arm. And they are dialectical
arguments directed at the reader, challenges to the consistency in his soul.
4. Timaeus’ Teleology
Let us consider one example of how this might work: the likely story of the Timaeus.8
Now in this dialogue the mythical context is rather more tricky than usual. The
conversation purports to take place the day after the conversation of the Republic;
and the figures of Socrates and Timaeus are set carefully against the mythology of
earlier sages like Solon—whose own claim to know what he is doing is debunked in
comparison to the older and wiser Egyptians. But Timaeus is still an expert (is he?),
and it is he (not Socrates) who offers a cosmology.
The cosmology itself is a myth, a likely story, told instead of the truth. This
contrast between what is true and what is merely likely is set up in the epistemo-
logical premiss of Timaeus’ account, where he separates completely the two worlds—
the world of being and the world of becoming—and their respective cognitive
modes—knowledge and belief.
First of all, in my opinion, we should make the following distinctions: What always is, and has
no generation? What always becomes, but never is? The former is graspable by thought with
argument (logos—word? theory?), since it is always the same; the latter is opinable by opinion
with irrational (alogos—without argument? etc.) sense-perception, since it is subject to gener-
ation and corruption, but never really is. (Timaeus 27d)
What Timaeus himself has to offer is strictly belief, about the world of becoming;
although he persistently tries both to bridge the gap between the two areas (e.g. at
30b7–8) by a harsh juxtaposition of truth and likelihood and to widen it (as for
example in his despairing comments about reasoning at 52b). Here, as in the allegory
of the sun, the story begins with a challenge:
So if, Socrates, in so diverse an account about gods and the generation of the whole cosmos, we
become unable to produce arguments (logoi) which are altogether consistent with each other,
nor precise, do not be amazed. But if we produce nothing less than what is likely, bearing in
mind that I the speaker and you the judges have a human nature, it would be reasonable to look
no further once we have found a likely story . . . (Timaeus 29c–d)
How are we to make the arguments agree with themselves (be consistent)? How are
we to judge the likely story? The context may be mythical—but the instructions are
dialectical, opening with a fierce tension between truth and likelihood (30a–b),
knowledge and belief—and argument and myth.
Timaeus offers a cosmology; and (echoing the Phaedo) the cosmos is described in
the terms used by the early Presocratics—earth, air, fire and water; flux; and atomic
8
Compare Osborne (1987a), (1996); McCabe (1994a), ch. 6.
particles. But Timaeus also promises a teleology—to show how god (the demiurge)
made the world to reconcile the opposed principles of reason and necessity; and thus
how the world is for the best. How far does Timaeus’ cosmology answer the
complaints of the Phaedo? Is this teleology useful and economical? Or extravagant
and otiose?
If we are looking for an explanation for how everything is for the best, we might
think first about practical reasoning. Suppose that I want some chocolate cake, and
there is some chocolate cake over there. I can explain my walking over there and
stuffing my face by pointing to the telos, the chocolate cake, and explaining that I did
as I did out of desire (or greed or whatever) for that end. This is an intentionalist
model of teleology where ‘the best’ is explained in terms of somebody finding it best.
Alternatively, ‘the best’ might be explained in terms of things being best. Aristotle,
for example, explains living organisms in terms of their proper (natural) function
(e.g. at Parts of Animals I. 1). As in a plant the leaves are for photosynthesis, the roots
for gathering water, so the whole plant is ‘for being a plant’ in the right way. Or more
generally, we could say that things that are ordered are intrinsically better than things
that are chaotic—ecological balance, for example, is a good thing because it is nature
in order. Order, on such a view, is good in itself; a structure is teleologically explained
if it is understood in terms of its order. Let us call this the natural model.
Either the intentionalist model or the natural model of teleological explanation
could be used to explain both individual and universal teleology, although to use both
at once seems like carelessness. On the one hand, an individual outcome can be
explained as its agent aiming for the best; or the universe can be explained as the
product of divine intention. Or on the other hand, an organism can be ordered for
the best—and so can a whole universe. The intentionalist model then generates the
notion of a creator, a god who determines the goodness of the whole universe by
having the universe reflect his intentions. The natural model, on the other hand, may
represent the universe as a living organism (the Gaia hypothesis, for example) or as a
working mechanism (like a clock, perhaps).
But on either view the teleological explanation asks us to believe more than we are
given by the data of sense-perception. An organism’s being well ordered is a feature
of it over and above its just being an organism; a divine creator would be an entity
extra to the physical entities of the physical world. Against such a view of the universe
someone (an extreme Ockhamist) might argue that we should invent no extra
entities, that our explanations should be as thin as they can be (let us call this
character a reductionist). So the teleologist and the reductionist are at loggerheads:
the former asks us to believe in an expensive account of the world in order to
understand it; the latter goes for stringent economy and wields the razor with
enthusiasm.
The trouble about this confrontation is that it seems irreconcilable. If entities are
not to be multiplied beyond necessity, the teleologist and the reductionist disagree
about what counts as necessity, about what grounds are enough to justify an
ontological overload. And now we find just that sort of conflict turning up in Plato.
Socrates (I argued) rejected Anaxagoras for reductionist reasons; but Plato’s myths
offer an abundance of teleological theories. The eschatologies of the Gorgias, the
Phaedo, and the Republic give us a view of the world organized for the best by means
of final (or not so final) judgements—a political motif. The sun allegory illustrates a
good natural order, on the analogy of an ecological account dominated by an
overriding goodness. And the Timaeus offers us teleology under several different
descriptions.
Timaeus’ cosmos is order brought out of chaos by the divine craftsman, the
demiurge. So the cosmos is for the best both because it is the object of the demiurge’s
intentions and because it is ordered. It is also beautiful because it is a copy of a
beautiful original—the intelligible universe. But because it is the only copy of the
intelligible universe, it is exhaustive and complete, and so a perfect copy. What is
more, because this world is made of reason imposed on necessity, it is rational (and
so good); and because rational, alive—and so, good.
This, if you like, is a case of teleological overkill. The thesis that teleology is a
product of the divine aesthetic sense is overlaid with the claim that the universe is
ordered; and the claim that the universe is a perfect copy (of a perfect original) is
further supplemented by the proposal that the universe is alive (and so itself
possessed of intentions towards the best). The reductionist might argue that none
of these claims is needed; the moderate teleologist might suppose that one is enough;
but this teleological myth gives us a series of different stories about the excellence of
the universe. The myth offers us two counterpoised accounts. i) The universe may be
good because god made it. Or ii), (supposing god to be a creature of myth) The
universe is good because it is ordered as if a god made it. The question is which of
these accounts, if either, is true?
If we put the myth of the Timaeus against the background of Socrates’ complaints
about Anaxagoras, we may see how the mythical context allows us to challenge its
extravagance and its consistency. We may see at the same time that the drive towards
teleology represented by the myth is a counter-challenge to the thinness of the
reductionist account (of Socratic simple-mindedness). The same sort of contrast—
between the rich extravagance of the myths and the austerity to be found in plain
arguments elsewhere—appears in the details of Timaeus’ story. Consider an example.
The third type is the permanent space which does not admit decay but presents a place for
whatever has generation itself, grasped by the lack of perception, by some bastard reasoning,
scarcely credible. We look to it as if in dreams, and we say that everything that is must be in
some place and have some space—for that which is neither on earth nor in the sky is nothing.
Since we are at the mercy of this dream, we are unable to discriminate and speak the truth,
either about these things and what is akin to them, or about the sleepless and true nature. On
the one hand, since not even that on condition of which the image exists belongs to the image,
but the image is always carried around as a phantasm of something else, the image must
therefore come to be in something else. But on the other hand the truth comes to the rescue of
what really is, so that so long as one is one and the other the other, neither will come to be the
other, nor will they become thus one and two at the same time. (Timaeus 52a–c)
Timaeus suggests that the phenomenal world (a copy of the ideal world) is made
from three components: the stuff from which the model is made; the shape of the
original; and the copy itself, the composite of the two. But the stuff from which
the cosmos is made is itself a monster from a myth—the ‘receptacle of becoming’, the
indefinite formless ‘something’ which has no shape, no description, is not graspable
at all except as in a dream—something we cannot even mention as the substrate of
all becoming.
The Presocratics, of course, loom large here; but the early mythmakers perhaps
loom larger. Timaeus’ explanation cites something that is itself the product of
necessity, not reason, which is unmentionable and indescribable, mere stuff, inaccess-
ible to reason. What sort of nightmare mythical monster has Timaeus produced? How
can we reason about it? How can we even be persuaded that it is there?
In the Theaetetus Socrates has a dream.
The things of which there is no account are unknowable . . . while the things that have an
account are knowable. . . . It is only possible to name each one itself by itself, and no other
predicate can be attached to the name, neither that it is, nor that it is not: for that would already
be to attribute being or not being to it, which we must not do, since someone could only
mention it itself. So we must not say ‘it’ or ‘that’ or ‘each’ or ‘only’ or ‘this’ or anything else in
addition; for all those things run around and are attributed to everything, although they are
different from the things to which they are attributed, but this, if it were possible to mention it
and it had its own account (oikeios logos), should be said without all these attributes.
(Theaetetus 201c–2a)
How, Socrates asks, can I understand things that are ungraspable, inexplicable, and
unknowable? What sort of a dream is that? The arguments of the Theaetetus suggest
that these monsters fail to contribute to our knowledge; the mythical context of
Timaeus’ presentation suggests that their mythical status may threaten their necessity
as entities. And yet Timaeus describes it as a safe and simple-minded answer (49d ff.).
The same sort of extravagance turns up all over the Timaeus. For example, the
demiurge, like a divine pastry cook, makes world-soul out of slices of sameness and
difference. But, ask the Parmenides (e.g. 139b ff.) and the Sophist (compare the
ontologies of 244 ff. with the discussion of Sameness and Difference at 255 ff.), can
sameness and difference really be understood as slices of things? If I am different
from you, is that because I have a piece of difference in me, which makes me so? Do
you have one too? What is the difference (challenges Euthydemus 30la ff.) between
having a piece of difference and owning a cow? Is difference really a thing?
Or think about the forms, the models that the demiurge uses to stamp character on
the unmentionable stuff. They seem to be separate entities; but do we need them? If
Timaeus wants to argue that there are forms for every natural kind, every relation,
and every value, is his account plausible? The Parmenides argues for parsimony in
postulating forms; and the contrast between the Eleatic view represented in the
Parmenides and the extravagant teleological view offered in the Timaeus could not
be more extreme. Timaeus suggests that we can give a teleological argument if we go
for the expense of a myth. Parmenides argues that we cannot escape the reductionist
objection (if the intelligible world is just extra, is it worth having?). We should be at
least sceptical about any claim that supposes that a giraffe is made of a form (giraffe)
stamped on some stuff. Is that a likely story?
What did Plato think? That is (by now, I hope) the wrong question. Myth is a
difficult medium for Plato, because myths seem not to represent direct truths.
Arguments, on the other hand, are presented in the dialogues as incomplete without
myths. What then is the relation between myth and argument? The connection is,
I have suggested, itself argumentative. The myths and the arguments are set up as
dialectically opposed to each other, offering opposed accounts of central metaphys-
ical questions. In particular, the mythological enterprise offers ontological (and
teleological) extravagance, while the argumentative enterprise tends to a reductionist
view. Therefore the myths and the arguments are fundamentally inconsistent with
each other; the challenge is to the reader to arrive at a consistent position. The point
of the dialectical contrast is that we must judge between the rival claims—in the case
I have discussed, of the reductionist to be economical; of the teleologist to offer a
fuller and more comprehensive explanation. We do that by the process of ‘compare
and contrast’ vital to proper understanding; and we aim at a theory fully consistent
and complete—such a theory, and not some closing myth, must have the last word.
What Plato gives us, then, is arguments, and the criteria on which they are to be
judged. The judgement is up to us; we are to blame, the god is blameless.
6
Is Dialectic as Dialectic Does?
The Virtue of Philosophical
Conversation
1. A Christening?
Republic VII, it appears, is the christening ceremony for dialectic. For here, we might
say, is the moment when Plato appropriates the expression ‘dialektikê’ as a term of
art, to mark out the pinnacle of his own philosophical method. Indeed, it all seems
deliberate, even emphatically technical:
‘So, then, do you call “dialectician” the person who grasps the account of the being of each
thing? Surely you will not say that someone who has no account, to the extent that he is unable
to give that account to himself and to another, has understanding of it?’
‘How could I say so?’ he said.
‘So likewise for the good: someone who cannot distinguish the idea of the good in account
by marking it off from everything else, and who cannot get through all the tests of what he
thinks as if through a battle, nor is eager to test it according to the way things are, rather than
according to opinion, and who cannot progress through all these things without his account
collapsing—such a person, you will surely say, knows neither the good itself, nor any other
good.’ (Republic 534b3–c5)
‘So you would legislate, would you, that they should most of all receive that education
through which they would be able to ask and answer questions in the most knowledgeable way?’
‘Yes, I would so legislate—and you with me, too.’
‘So do you suppose,’ I said, ‘that dialectic lies at the top for us, like a coping-stone on our
studies, and that there is no other subject that should rightly be put higher than it, but that it
provides now the end to our inquiries into education?’ (Republic 534d8–535a1)
. . . and insistent on the name:

‘And so when someone attempts, by conversation, to arrive at what each thing itself is, by
means of reason1 without any of the senses, and does not stop until he grasps what is the good
1
The expression logos, notoriously difficult to translate, reappears in these passages to describe the
faculty by which dialectic is done (as here) and the content of what dialectic says (an ‘account’ in the
previous passages). The word dialegesthai itself is cognate with logos, of course. It is, as I shall argue,
significant that the conception of dialectic here bridges both the faculty of reason and its content. For
reasons of space I eschew here, however, further discussion of logos.
IS DIALECTIC AS DIALECTIC DOES ? 101
itself, by thought itself, then he is at the very end of the intelligible, just as then the man in the
cave came to the very end of the visible.’
‘Absolutely,’ he said.
‘Well now. Do you not call this journey “dialectic”?’ (Republic 532a5–b4)
. . . and notice the frequency of the term dialektikos in the six Stephanus pages 531–7,
compared with its complete absence elsewhere in the Republic.2 That supports both
the technicality and the claim that this is the point of its introduction as Plato’s own
term of art.3
What is more, the expression’s genealogy is marked. For Socrates has worked up to
this moment from the beginning of the discussion of the greatest learning (‘You have
often heard that the idea of the good is the greatest learning . . .’, 505a2–3) by a series
of variations on the theme of conversation, dialegesthai (a word not in fact as
commonplace in the Platonic dialogues as one might expect).
The first variation plays at 454a, where Socrates inveighs against the dangers—and
the temptations—of antilogic or disputation, the art that deceives people into think-
ing they are having a proper argument, when in fact they are only practising eristic.
‘It is a notable power, isn’t it, Glaucon, that antilogic has?’
‘Why?’
‘Because,’ I said, ‘many people seem to me to fall into it quite involuntarily, and to think that
they are not disputing, but conversing. This is because they are unable to consider their subject
by dividing into forms—instead, working just on the name of the thing, they pursue the
contradiction of what has been said, and treat each other competitively, not as in a discussion.’
(Republic 454a1–9)
Antilogic is unable to consider (episkopein) the subject under discussion by virtue of

the proper distinctions.4 Having a proper discussion, by contrast, avoids getting
caught up in sophistical difficulties (as, for example, by virtue of the dropped
qualifiers in the preceding discussion of sameness and difference of natures among
men and women) by making the right distinctions; this is done by conversation
(dialegesthai) and discussion (dialectos). ‘Conversation’ here, then, has a normative
force: not just any old talk, but conversation and discussion conducted along the
right lines and with the right precision.5
Conversation of this philosophical kind pervades the cave, too. From the very
beginning, the prisoners are imagined having a conversation even when they are tied
down (515b4): and the release and ascent from the cave resounds with the language
2
See Brandwood (1976 ad loc.).
3
We might distinguish, as Lesley Brown suggests to me, dialectic’s conception, at Meno 75d, from its
christening, here.
4
This dividing into forms (types? eidê), however it may be related to the method of collection and
division of later works (Phaedrus, Sophist, Politicus, Philebus, see Stenzel (1940); Gomez-Lobo (1977);
McCabe (2000); Silverman (2002)) is at least systematic in some way: see below.
5
It is a matter of moral character, too: see Republic 538–9.
102 IS DIALECTIC AS DIALECTIC DOES ?
of philosophical discussion—and philosophical discussion with Socrates, at that

(notice the gloomy prediction that someone who tries to release the prisoners from
their darkness would be killed, so long as he could be caught: 517a). So when the
prisoners are released they are shown one of the objects carried behind the wall,
and they are asked what it is—but they are in aporia about it, and cannot answer
(515d5–7). Their journey upwards—which Glaucon agrees to call dialectic (532b4)—
is characterized by their increasing ability to reject what they see before them (532c)
and to resist fixed assumptions (533c) in favour of being able to give an account of
what they are considering (533c).
Indeed, this feature of philosophical conversation is taken to be true, not only of
the prisoners as they emerge, but also of Socrates and Glaucon themselves. The
discussion of antilogic, of course, was provoked by the position they had both
reached in the argument; and repeatedly the conditions associated with dialegesthai
apply to the interlocutors themselves. At 528a Socrates exhorts Glaucon to conduct
his conversation with the right sort of people: with those who
‘accept that in these subjects an organ of each person’s soul is purified and rekindled—an organ
which is destroyed and blinded by other pursuits—whose preservation is more important than
that of a thousand eyes, since only by this organ is truth seen.’ (527d7–e3)
He should do this, indeed, for his own sake: and it is for his own sake that he should
ask questions and answer them: this is how the arguments should be made (528a).
The process of conversation, indeed, starts to work in exactly the prescribed way. At
532d Glaucon vacillates6 in his response to what Socrates says; and then, when he
asks Socrates to tell him the whole story about dialectic, Socrates strikingly warns him
that he is not yet ready to follow that far (533a). Glaucon needs a bit more
philosophical conversation before he may see the good: and he should have known
it, for Socrates had already told him that they might only see the good through an
analogy (506e).
The emphasis on philosophical conversation throughout the central books of the
Republic might, then, make us hardly hesitate, at least on a careful second reading,
when—in advance of the formal account of dialectic itself—Socrates announces that
it is the ‘power of conversation’, hê tou dialegesthai dunamis, that will allow us to
touch, or to view,7 the topmost intelligible part of the line (511b4; 511c5). Conver-
sation, we might then think, proceeds to a conclusion; and it does so by overturning,
or confirming, the assumptions upon which its earlier stages rested (511b, 533c). It is
an old question just how that is supposed to work.
6
Notice the way this works: he thinks it is hard to accept what Socrates says, and hard to deny it. The
terms of assent and dissent will reappear in the account of the silent dialogue; see further below. On the
importance of aporia see e.g. Matthews (1999) and Politis (2004b).
7
I shall return to these perceptual metaphors in what follows. I here translate theôreisthai as ‘view’; the
word is notoriously tricky, however—see e.g. Nightingale (2004).
The features of conversation urged in the setting of the Republic are reinforced in
the description of the divided line: that conversation allows us to treat our assump-
tions with proper caution; and that it is, somehow, a method of proceeding in
philosophical inquiry. In particular, we might think, five aspects of the conception
of conversation would be indispensable to the philosopher as he moves up the line:
• a logical aspect: that the question and the answer represent two sides of a case,
and the imagined conversation takes place as the two points of view play off
against each other. This play has a compulsive side: these kinds of opposition
demand resolution.8
• a psychological aspect: the philosopher remains agnostic, suspends judgement
about which side of the case he proposes to take while he considers the matter;
his agnostic stance is a sense of puzzlement, of aporia, in his soul, and the
considering is something he does in his soul: e.g. at 524e5.9
• a sequential aspect: conversations are conducted in such a way that the answer is
relevant to the question, and the next question to that answer. The notion of a
conversation, that is, has an order, a proper sequence built in (compare Socrates’
repeated insistence on doing things in the right order, e.g. at 527b, 528d, 535a).10
• an epistemological aspect: the philosopher takes a synoptic view of both sides of
the case at once: he both entertains the opposed views and considers their
relative merits.11 The synoptic view, that is, is reflective, or second order; and
it has both sides of the conversation within its scope (this is exemplified by
Socrates’ own reflective procedures, for example at 529a–b; and see the ringing
claim at 537c, that the dialectician is someone with a synoptic view).
• normativity: you can do this kind of conversation well or badly; or fail to do it at
all (see e.g. 525d, 527d, 528a, 531e–532a, 538–9).
These conversations may be conducted, indeed, not just between two parties to a
philosophical debate, but within the soul of the speculating philosopher. Notice a
passage to which I shall return: at 523b ff. Socrates describes the way in which the
8
We may readily see why, from Glaucon’s vacillation at 532d: he cannot both assent and dissent, on
pain of contradiction. See Politis (2004a) on the nature of aporematic argument, and Chapter 7 of this
volume on one of Plato’s detailed accounts of the nature of contradiction. The element of compulsion is
emphatic in the discussion of philosophical progress: 515e1, 525d6, 526b1–2, 529d1.
9
Compare the muddle the released prisoner gets into when he is forced to evaluate the impressions he
gets at different stages in his ascent, along with a principle of interpretation (the latter impressions are more
significant than the former) at 515d. The idea that puzzlement may be philosophically productive, of
course, lies behind Socrates’ disavowal of knowledge—which is not to be confused with scepticism: see e.g.
Ap. 20–3.
10
There is an obvious normativity here: consider an exchange of views in which neither party pays
attention to what the other says: here the two parties may talk at cross-purposes and, in extreme cases, fail
to have any conversation at all.
11
Compare the reflections on the drawings of Daedalus at 529e, and on the absurdity of their claims to
give clues to the truth. Compare also Socrates’ brusqueness with Glaucon’s suggestion of a shortcut at
532d–533a.
reportings of the senses cause the soul to be at a loss about what perception is saying,
and to consider (episkopein again, 524b4–5) whether the senses report one thing or
two. Soul asks questions of itself (524e5–6); by this means the soul will be turned
towards, and come to touch, being and truth (525b), and will have a discussion
(525d5–6, 526a2) of the numbers themselves.
This internal conversation, we might further think, fits with Plato’s account of
thinking as the ‘soul’s silent dialogue’ in Theaetetus and Philebus (Tht. 189–90; Phlb.
38c ff.; and see Soph. 263e).12 Does that comparison allow us to answer the old
question, how philosophical conversation is to reach a conclusion? For just as the
soul, in silent dialogue with itself, is able to come to a single view (it ‘says the same
thing’, Tht. 190a3), so the philosopher engaged in the conversation of dialectic—
whether with herself or another—might be able to resolve the question at the end of
the discussion—by virtue, perhaps, of her synoptic view. Something, then, about her
epistemological state, or her epistemic capacity, will explain the way that philosoph-
ical conversations get somewhere, and do not merely continue to vacillate, as
Glaucon does, between two different points of view. That is what we might think,
at any rate, if we read the soul’s silent dialogue back13 into the Republic’s account of
dialectic. But the Republic has something else up its sleeve.
2. The Form of the Good

Quite right, too, you might say. For after all, the soul’s silent dialogue has only a
limited account to give about just why I might reach a conclusion, or why its outcome
might constitute knowledge—this is one of its problems as a model for epistemol-
ogy.14 More broadly, it is hard to see why a philosophical methodology analogous to
conversation would ever appeal, since there is no guarantee that—as such—it would
produce truth as its output. After all, why should I not ask myself all sorts of
questions and deliberate on the answers—and end up, as Socrates insisted, with a
consistent set of beliefs—and still be comprehensively wrong? How can Socrates’ sort
of inquiry provide itself with independent support? Meno pressed the point: he
demanded that Socrates show just how we might begin in our inquiries after
knowledge and truth; and how if we ever reach the end we are to know that the
end is what we have reached (Meno 80d).15
In the face of this kind of difficulty, the Republic’s account of dialectic has what we
might think of as an appealing answer: the form of the good. Consider again just how
12
See Frede (1989); Burnyeat (1990); Dixsaut (1998); McCabe (2000).
13
There is, I fear, an issue of the chronology of Plato’s dialogues here; I ignore it here, save to assume
that Theaetetus, Sophist, and Philebus were written (perhaps considerably) later than the Republic.
14
Indeed, the soul’s silent dialogue is not offered as a model of knowledge, for this very reason: instead,
it gives us either a view of the psychology of thinking or of the mechanics of judging. Still, it is normative:
there are good and bad judgements or decisions.
15
On this see especially Scott (1995), (2006).
Socrates describes the end of the dialectical journey (even though what he says is still
hedged about by disclaimers to Glaucon): that the soul suddenly sees the form of the
good (517c); or it touches the unhypothesized beginning (511b). The form of the
good, of course, is the source of everything good and fine; as a consequence of seeing
it, the dialectician is able to give an account of everything; then to get clear all the
connections between things, and thus to have a complete account of all the things he
treated as provisional on his way up the line (see 516b). The form of the good, on this
account, answers the second limb of Meno’s paradox by suggesting that while the
process up the line may be conducted by conversation and investigation, its verifica-
tion is provided by the dialectician’s view of the good. For the good is independent of
him, and it provides an independent justification of his knowledge; consequently, it
allows the rest of his system to be objectively confirmed.
A view of Republic VI and VII has long been popular that takes what happens at
the top of the line to be explained in terms of some special epistemic access the soul
has to the form of the good.16 Consider, for example, Cornford’s remark: ‘The
apprehension of the Good is rather to be thought of as a revelation which can only
follow upon a long intellectual training . . . ,’17 which he follows up by talking of the
philosopher’s ‘immediate knowledge of the Good’.18 Or:
The backwards regress is said, vaguely, to end in apprehension of an unhypothesised begin-
ning: the form of the good. Plato does not elaborate, but since the Good is the first principle,
there must be nothing more basic in terms of which the Good can be explained or defined.
Knowledge of it will have to consist in some sort of intuition.19
More recently, Nightingale emphasizes the connections between what occurs at the
top of the line and the sexual imagery of both Republic (e.g. at 490a–b) and
Symposium (210e, which connects the visual and the sexual imagery): ‘both meta-
phors portray the apprehension of truth as a receptive activity’.20
This view has many variants, and many different accounts of how we should cash
Plato’s metaphors. But fundamentally it supposes that the relation between the
philosopher and the form of the good is precisely not given by the process of
philosophical conversation. Instead, by contrast, the good is thought to affect the
16
But N.B. the detailed and persuasive dissent of Fine (1990) to the common view that in the Republic
someone’s state of mind is determined by the objects they encounter. In what follows I recant what I said in
McCabe (2000), ch. 7, that the form of the good acts on the passive mind of the philosopher.
17
Cornford (1941), 208; and see Cornford (1965).
18
Cornford (1941), 208 n. 2. Cornford is brisk, however, in dismissing Neoplatonist interpretations of
the Parmenides that press the idea of a mystical union between the intellect and its objects ((1939), 131–3).
Cross and Woozley are more reluctant, but still concede that ‘presumably the hypothetical method has to
be supplemented in the end by intuition’, (1964), 252. Robinson characterized this as the ‘intuition-theory’
of the upward path (1953), 172–7; but he is inclined to gloss intuition in terms of confirmation, rather than
as an unmediated grasp of what is known.
19
Heinaman (2003), 377, here rightly emphasizes the epistemological problem with which the intuition
theory is meant to deal.
20
Nightingale (2004), 116; and see her programmatic remarks at 109–14.
soul. In particular, this is often taken to be the point of the perceptual metaphors
throughout these pages of the Republic:21 that the good is quasi-perceived by the soul
(whether by quasi-touch or quasi-sight). When perception happens, moreover, on
this view, the perceiver is acted upon by the object perceived: the agent, as it were, of
perception is the object, the perceiver the patient.22 So it would be the soul’s passivity
to the form of the good—or, to put it a different way, the directness of the presentation
of the form of the good to the soul—that would provide the verification at the end of
the dialectical process.
The claim that these presentations are direct, that they are somehow or other ‘raw
feels’, could make two quite different points, however. On the one hand we may wish
to emphasize the rawness of the feel, the unmediated affection of perception by its
object. In such a case the verification is provided by the causal relation between object
and perceiver: the fact that the object just does act on the perceiver (and that when it
does so the perceiver actually perceives the object as it is) is what guarantees the truth
of the perception. On such an account, perception must make no contribution, do no
work, of its own. This rawness, then, carries a realist or objective assumption along
with it—that as a matter of fact there is an object that rawly determines this feel. On
the other hand, the emphasis on a raw feel may be rather on the feel: on the fact that a
raw feel is subjective, so much in the private experience of the feeler that it is
inaccessible to anyone else. Here too, perceptual experience is unmediated in some
way; if there is something that it is like for me to see red, that is somehow directly
available to me—and, more to the point, not to you. That gives my feel a claim to
truth, but none to objectivity.
If the form of the good is to perform a role as the unmediated source of truth at the
end of the dialectical process, then, it had better do so raw, rather than by virtue of
the fact that it is felt. The privacy of a subjective feel cannot, at least, perform this role
in the account of knowledge in the Republic—for the very issue in that account is how
the objective realities of the world (sensible or transcendent) are rendered accessible
to the soul. In what follows, therefore, I shall consider whether Plato postulates events
that are in the objective sense intellectually raw, and not whether he supposes that
what is intellectually raw is privately felt by its subject.
If the intellect receives it directly, the form of the good would be a foundation twice
over: it is the foundation, the source, of the goodness in the world; and it is the
foundation of the dialectician’s knowledge. It is not the dialectician who must verify
21
Both tactile and visual imagery is used: seeing—511a1, 516a5, 517c1, 519c10, 519d2, 520c4–5, 526e1,
526e4, 527e3; touching—511b4, 511b7, 525c1; and the imagery of unmediated contact, grasping: 524e1,
529d5, 529e5, 532b1.
22
That this could be a Platonic view is confirmed by the theory of perception offered in the discussion
of Protagorean relativism at Tht. 154–5; on the complex issues that arise about that passage and its sequel
see Cooper (1970), Frede (1987), Burnyeat (1990). Notice, also, the role the language of perception and
sight may play in the account of recollection at Men. 81c6 to exempt the process of prenatal learning from a
vicious regress.
his views, but the form of the good that does it for him by appearing to him directly,
unmediated and objective. Meno’s second limb may be thought to have been shod.
The price of the shoes, many have thought, may be too high: for it is fixed at the
theory of forms (and we may be too parsimonious to pay it). It looks, furthermore, like
an extreme version of that theory: where the highest flights of knowledge are explained
just by the ineffable—by what is ‘beyond being in authority and power’ (509b9–10)
and by the direct action of the ineffable on the mind of the philosopher (that it is his
mind then has little to do with it). Thence there is a second price, concealed in the
account of dialectic itself. For it seems to commit Plato to two rather different views of
the relation between philosophical conversation and what happens at the top of the line.
First, he may need to make a psychological claim, that:
i) philosophical conversation will as a matter of fact open the ‘eye’ of the soul to
the forms.23
Should we find this persuasive, in the absence of any actual cases? And even if it is
persuasive, it suggests that the progress up the line, all the way to and including the top,
is somehow psychologically continuous. The construal of the soul’s view of the form
of the good that I have just elaborated, however, requires discontinuity, since it requires
the verification of an inquiry to be different in causal structure from the inquiry itself.
Second, on this interpretation, Plato would claim that:
ii) philosophical conversation is replaced, at the summit of the line, by the

affection of the soul by the form of the good.
This discontinuity may itself be problematic, for it suggests that at the very top of the
line, cognition is determined by its objects. We might object that any account of the
divided line that has cognitive states determined by their objects interferes with
the thought that philosophers will (in the ideal situation) be better kings than
those who find the lower reaches of the line or the cave familiar. After all, if the
philosopher-king is to be any use to us on his return to the cave, he had better include
the objects of the phenomenal world in the scope of his knowledge;24 his superior
cognition cannot be restricted to the world outside the cave if he is to be a king.
Somehow it is his cognitive state that makes him better than us at dealing even with
the sensible world; so his state of mind, his cognitive state, cannot just be determined
by the objects he encounters.25
These two claims invite quite different construals of the nature of philosophical
progress: the first that it is continuous from inquiry to verification, the second
23
E.g. at 527d. See Frede (1999). How should we cash the thought at 518c that the eye moves with the
whole body, or that the whole soul moves with the eye of the soul? I shall return to this below, }6.
24
Or, better, his understanding: I shall return to this issue below, }5. See e.g. Burnyeat (1981), Nehamas
(2004).
25
The exact significance of this issue is disputed; but see Annas (1981); Frede (1999); Fine (2003).
discontinuous. This, as well as general considerations about Plato’s account of the

philosopher-king, might invite us to wonder whether the role of the objects at
different stages of the line is more complex, in each case, than a direct encounter,
from the topmost section down.
Objections to the thought that the good is just directly experienced by the soul may
be reinforced by the descent down the line. For here Socrates repeatedly speaks of the
soul’s activity of contrasting and comparing, of establishing a systematic set of
connections between different parts of knowledge (511b, 517c, and compare 531d,
537c7).26 That activity, for sure, bears its similarity to philosophical conversation on
its linguistic face: especially in the way in which the soul negotiates the tension
between assent and dissent by ‘syllogizing’ (516b, 517c). This may put further
pressure on the idea that what happens at the top of the line is somehow or other
quite different in kind from what happens elsewhere: and it puts that pressure by
virtue of the emphatic interest in the setting as a whole on the discursive ways of
philosophical conversation.
So: are we easily convinced of the thought that the ways of the intellect might alter
sharply as we come to the experience of the good? Must we agree that the form of the
good only appears to us after we have done some philosophical talking? Is this all
just too easy to wrap up in the language of seeing of a mystical sort? Let me put the
point in a different way: if the end of dialectic is something essentially non-discursive,
quasi-experiential, why take the trouble to call this dialectic at all? Is the christening
of dialectic just misconceived?
Well, perhaps christenings are like this anyway: it doesn’t matter, if I call my
daughter Poppy, that she doesn’t end up pink and crinkly. Proper names (and so
christenings) behave in rather more rigid ways than definite descriptions, and we
should not be perturbed that Plato chooses this route to coming up with a term of art
for his very own philosophical method. Its name (like my daughter’s) makes some
obeisance to its ancestry (in dialectic’s case, Socrates’ ways of inquiry; in Poppy’s
case, her grandmother). So, many have supposed, the christening of dialectic is not
meant to give us a description, just to mark out, with a privileged name, the best
possible way of doing philosophy. Dialectic, as Robinson remarked and as others
have regularly repeated after him,27 is the name Plato uses at any time for his best
philosophical method.
Is that right? It may be true of what happens after the Republic (although I have my
doubts about that28); but it is hard to defend for the Republic itself, for two reasons.
26 27
On the mathematization of this see Burnyeat (2000). Robinson (1953), 70.
28
See McCabe (2000). Outside the Republic the expression dialektikos turns up in Phaedrus (266c,
276e), Sophist (253d–e), Politicus (285d, 287a), and Philebus (17a) in (otherwise notorious) contexts where
we might reasonably understand it as a term of art, directed at a knowing readership. It also appears in
three other dialogues: Meno (75d) where it is used as a condition of philosophical conversation; Cratylus
(390c, 398d) where it reminds us that the dialectician is the expert in asking and answering questions, and
has the dialectician operating as the overseer of other skills; and Euthydemus (290) in a context that appears
First, the eventual use of the expression dialektikê is carefully anticipated in the
conditions for philosophical conversation, for dialegesthai. It is hard to comprehend
why the outcome of all this fuss is a name that has no real resonance for the method
to which it applies. Second, if philosophical conversation is not essential to dialectic
as a whole, but only to its preliminaries, the verbal connection between dialectic as
conversation and dialectic as whatever happens at the end of the philosopher’s
endeavour merely serves to emphasize an uncomfortable tension within the meth-
odology described in the central images of the Republic. Must we just accept that this
is an unfortunate feature of Plato’s account of philosophy?
That would be a counsel of despair. Instead, I shall make three suggestions to invite
the conclusion that philosophical conversation is indeed essential to Plato’s meth-
odology, through and through. My first suggestion (}3 below) revisits the question
whether the deliverances of perception are merely passive and thence non-discursive.
I argue, instead, that the evidence of the Republic—especially of those passages where
Socrates and Glaucon seem to be speaking non-metaphorically—supports the view
that perception, too, is a part of the internal conversation of the soul; and that it fulfils
this role by delivering reports that have propositional content: so its reports are not
unmediated.
My second suggestion (}4), consequently, denies that at the top of the line the soul
should be imagined as the merely passive recipient of an unmediated ‘seeing’ of the
form of the good. Not only should we understand ordinary seeing in a discursive
manner, I argue, but also the notion of the ‘spectacle’ of the good is itself discursive,
because it is synoptic and second order.
Third (}5), I suggest that when the form of the good operates as the cause of
everything else, the soul sees it as the cause: and this special ‘seeing as’ ensures the
content of the soul’s understanding, its reflectiveness, and its claims to unity. This
move, moreover, insists on the realism of what is thus understood. So it avoids the
complaint that dialectic is inconclusive without having the verification of dialectic’s
conclusions done by something intellectually raw. The Republic’s model of dialectic,
therefore, incorporates a rich account of the psychology of the philosopher with its
objective validation by the form of the good.
This, in turn, explains just why the philosopher’s state of soul is virtue: the
completion of the philosophical journey demands not only the right object of his
knowledge, but also the right and ordered state of soul. In closing, therefore (}6),
I offer a speculation on an issue beyond the scope of this chapter. Why is dialectic
focused on the good? It is not, surely, merely of instrumental value, not merely to
provide the philosopher with the ability to make accurate choices (and so to achieve
deliberately designed to recall the Republic (see McCabe (2001)). All of this evidence is consistent with the
Republic’s offering us a christening: a christening with the conception of dialectic as philosophical
conversation firmly in mind, and where the technicalities of dialectic are worked out more carefully (but
still, I claim, as conversation of some kind) in later works, Phaedrus, Sophist, Politicus, and Philebus.
the best result every time); such a role for the good would be tendentious. Instead,
Plato claims that the philosopher knows the sovereign good; and this knowledge is
transformative of his life just because it is the source of value for him. Understanding
the good, therefore, incorporates both the psychological and the metaphysical con-
ditions for dialectic: the role of philosophical conversation is essential.
3. Seeing Fingers
The central books of the Republic are much taken up with images and metaphors—
and it is the cashing of these metaphors that causes so much trouble.29 Pervasive is
the use of verbs of seeing, looking, and grasping to describe the activities of the
mind;30 but is the implication of this terminology that the mind works (or works at
the crucial topmost stage of the line) as if it were passively perceiving?
One passage in the midst of the extended images of sun, divided line, and cave,
however, seems to be largely non-metaphorical: the contrast between perception and
intellection amplified at 523a–525b. This passage shows, I shall argue, that so far
from perception being a rawly direct affection of the perceiver by the sensible object,
the deliverances of perception are propositional—even complexly so—in content.31
As a consequence, the parallel between perception and intellection urged throughout
the middle books of the Republic may not demand a non-propositional account of
intellection either; no more than perception is intellect rawly affected by its objects.
This, I shall argue further, extends the significance of philosophical conversation for
Plato’s account of dialectic.
Socrates’ argument proceeds by considering what study ‘drags thought to being’
(523a2–3). Some things ‘in perceptions’ do not call the intellect to reflection (epis-
kepsis) because they are adequately judged by perception.32 This happens when the
perception does not ‘result in’ (ekbainei) the opposite perception at the same rime.
But there are things that do ‘order thought to reflect’ (episkepsasthai) because
perception ‘makes nothing sound’ (523b3–4). For when there is the opposite per-
ception at the same time, the mind is called to reflection; in these cases perception
‘shows33 one thing no more than its opposite’.
29
Burnyeat (2000), 55, is surely right to suggest that whether a given expression is metaphorical in these
passages of the Republic may well be a matter of perspective, in the sense that something may look
metaphorical when viewed from an early stage in one’s philosophical (mathematical) education, but turn
out to be literal when viewed from a more advanced stage of philosophical understanding. However, the
interpreter still faces the challenge of working out which aspens of an apparent metaphor are salient.
30 31
See n. 21. See Scott (1995), 83.
32
Perception makes a judgement in the sense of a settled decision between two options, perhaps, or a
discrimination already adequately done (so the expression ‘judgement’ here is importantly not merely
equivalent to ‘belief ’ or ‘opinion’).
33
The same verb is used of what Socrates is doing here in the dialogue: showing Glaucon what he
means, 523a5. See also 524a3.
What on earth could Socrates mean? Glaucon attempts a gloss: surely Socrates is
talking about ‘objects appearing at a distance, and shadow images’? Socrates demurs
and insists that perceptual mistakes or illusions are not at issue here. This little
exchange invites us to be careful about what follows; to wonder how—other than by
mistake or illusion—perception fails.
Socrates begins with an example (523c10 ff.). Consider three fingers, one large, one
small, one in the middle. Each appears equally a finger, and sight does not ‘signify34
at the same time’ (523d5) that the finger is the opposite of a finger; so the soul is not
impelled to ask of them ‘what is a finger?’. Contrariwise, sight does not ‘see
adequately’ their largeness and smallness, and cannot discriminate35 which one lies
in the middle. The same account can be given of the other senses. They show such
things deficiently36 just because the same sense reports on opposite properties (large
and small, hard and soft, sweet and bitter); thus the same sense ‘announces to the
soul that the same thing is hard and soft as it perceives’ (524a3–4).
What exactly happens when perception ‘announces to the soul that the same
thing37 is hard and soft as it perceives’?38 Neither the context nor the terminology
34
At Cra. 436, ‘signifying’ seems to apply to the announcing of a name; this would fit with my option
(d) below, that all perception does is label an experience.
35
‘It makes no difference to sight’, e2. Does this mean that sight cannot discriminate at all (which
would be decisive for my option (d), below)? Or is the point that sight cannot make the discriminations
that reason can do; but that instead, it proffers something ‘mixed up together’?
36
There is no claim here that the properties of the objects in question are deficient (that, e.g., the finger
that is seen in the middle is not really in the middle)—the point made is an epistemological one, about what
sight says to the soul.
37
In addition to the problems of interpretation discussed below, there is another: what is ‘the same
thing’ about which perception makes its announcement? Perhaps perception announces something about
some individual particular object—a finger, or a stick, or a man. In that case the announcement is imagined
to be a banal utterance about the ordinary phenomena of the sensible world. Or perhaps perception is here
imagined to be making a claim about what it is to be {such and such a thing}. This will allow its
announcement to be more philosophically loaded: what a finger is, it can tell us easily; but what hard or
soft is, it confuses. It has been suggested (not least, on the basis of the apparent contradiction attributed to
sensation, ‘the heavy is the light’, 524a9–10; see Fine (1993), 56–61; Irwin (1995), 157–62; but White
(1992)) that perception is here answering a question that has already been put to it by the soul: ‘What is a
finger?’ ‘What is the heavy?’. Because perception is limited to sensible properties, it can only answer in
those terms. So it might say: ‘This is (what it is to be a) finger’, or ‘This (e.g. being one kilogram in weight)
is what it is to be heavy.’ This works for explaining what finger is; but with individual perceptual properties,
such as heavy or soft, no sensible property (being one kilogram in weight) can offer an answer that is other
than confused (being one kilogram in weight might just as well explain what it is to be light). (On this
loaded model, perception is conceived as a sight-lover, telling us, for example, what beauty is: see 475e ff.
However, there is not, I think, evidence in the text that Plato supposes all perceptual reports to be like those
of the sight-lovers.) I have no space for further discussion of this point here; but perhaps what follows
below may contribute. For if perception is imagined to make a report as a result of a question it has already
been asked (about what something or other is), and so as already engaged in a philosophical discussion with
the soul, then the second loaded model may be plausible. If, however, the conversation begins after
perception makes its announcement, it is hard to see that the announcement itself is already engaged
with questions about what things are. In that case, as I shall urge below, we should prefer the banal account
of the announcement perception makes.
38
The loaded reading described above might prefer that perception ‘announces to the soul that hard
and soft are the same thing’. The Greek suggests, however, that ‘hard and soft’ are in the predicate position.
allow us yet to choose between the following glosses for what perception is imagined
to say:
a. perception announces that it perceives the same thing as hard and soft; per-
ception says: ‘I perceive that the same thing is hard and soft.’39
b. perception announces, as it perceives, that the same thing is hard and soft: ‘This
(same) thing is both hard and soft.’
c. perception announces, as it perceives, of the same thing, that it is both hard and
soft. Here perception seems to say two things: ‘This is hard’, ‘This is soft.’
d. perception announces, as it perceives, of the same thing, that it is both hard and
soft. Here perception may merely label its raw affection: ‘hard!’, ‘soft!’.40
The first of these possibilities is ruled out by the argument as it proceeds. Self-
consciousness is something perception manifestly lacks: this occurs only when the
soul begins to puzzle. Still the matter is not clear. Does sensation merely label what it
feels (d)? Or does it utter something like a proposition ((b) or (c))? And if the latter, is
the proposition simple (c) or complex (b)? The latter question turns further on how
to interpret what perception is imagined to say: does it say of the same thing (de re)
both that it is hard and that it is soft; or does it explicitly attribute both properties to
the same thing, saying that the same thing is both hard and soft (a de dicto reading of
524a)? Of course, if this last is what happens, there is undoubtedly more than raw
labelling going on; can the text be further disambiguated between the de re and the de
dicto readings?
The terminology of the passage so far is, I think, insufficient to ground a deci-
sion.41 Scrutiny of the sequel, however, may allow us to do so. Once perception
announces to the soul (whatever it does announce), the soul is necessarily42 puzzled
and asks itself: ‘What does perception itself signify by hard, since it says that the same
thing is soft too; and likewise for the perception of light and the perception of heavy,
what is light and heavy, if it signifies that the heavy is light and the light heavy?’
(524a6–10). One might be forgiven for supposing that Socrates here imputes a
contradiction to perception (it says ‘the heavy is light’); but perhaps this would be
39
So ‘as it perceives’ is in the scope of what perception is imagined to utter. The word order is
inconclusive.
40
Compare the differing views of Cooper (1970) and Burnyeat (1976c) on Tht. 184–6; and also Frede
(1987).
41
The terminology in the passage for what perception does is varied: it ‘shows this no more than its
opposite’, 523c2–3; it does not signify that the finger is the opposite of finger, 523d5–6; it reports perceiving
that the same thing is both hard and soft, 524a3–4; it says that the same thing (the hard, or what is hard) is
soft, 524a8; it signifies that the heavy is light and the light heavy, 524a10; sight sees the large and the small as
something confused, 524c4.
42
This may show too much, on any reading; does everyone puzzle about the deliverances of the senses?
On the loaded reading, however, it might come out true; for then the philosophical conversation has
already begun—the soul’s puzzlement is inexorable. For other reasons, however, I think this account of
what is happening is not persuasive.
better construed transparently,43 or otherwise as stating something less logically

worrisome (they say ‘what is heavy is (also) light’ where each of the predicate
terms is understood as incomplete: ‘heavy compared to a mouse’, ‘light compared
to an elephant’). On either account, however, the deliverance of perception seems to
be both discursive (perception is said to say something, as well as to signify—the
latter might be just labelling, but not so, surely, the former in this context?) and
complex. For the expressions ‘heavy’ and ‘light’ are used of the same thing; and they
appear (as the last two clauses of the passage show) somehow in the same announce-
ment. The announcement itself is what generates the soul’s puzzlement. The contrast
between perception and the soul is urged here, therefore, in just one dimension: the
affect of puzzlement (suffered by the soul), or its lack (in perception, which just
reports). That affect is what causes the soul to formulate a question, ‘what does
perception signify by the heavy?’.
Glaucon responds, and amplifies the collection of metaphors used here to describe
the interchange between perception and the soul. The deliverances of perception are
announcements, messages, utterances44 (524a3; the language is picked up again at
524b5); and Glaucon says that they are puzzling, and require consideration. The
messages are puzzling, but perception itself is not puzzled. The puzzlement is felt,
instead, in the soul, for it is the soul that needs to embark on consideration. Then the
significance of the difference between what perception does and what the soul does is
the difference in, if you like, cognitive stance, not in the discursive character of what
each faculty says. The cognitive stance of perception is mere reportage—reportage
that includes judgement (523b2); whereas the soul, appropriately enough, has all
sorts of other stances, including the affect of puzzlement, and the ability to have a
conversation with its own parts. So perception just delivers the message that this
thing is hard and soft; the soul reacts by being puzzled, and then summons calcula-
tion and thought to think about the message that has been delivered (524b4–5).
When perception reports, soul invites reason to consider whether ‘each of the
things announced’45 is one or two (524b5). If they are two, then each (of the two) will
be one; and in that case, each will be separate from (non-identical to) the other—for
otherwise they would not be conceived as two, but as one. But as a matter of fact,
perception presents, for example, large and small, not as separate, but as mixed up
together (sunkechumenon, 524c4). It is this confusion of two separate things that soul
43
They say, of the heavy thing, that it is light. Compare the amplification of this problem following the
discussion of allodoxia at Tht. 188 ff. This expression might be thought to support the loaded interpretation
(they offer an account of what heavy is that is also an account of what light is); but the language at this point
is not decisive in favour of that interpretation.
44
The Greek is hermêneiai, whose cognate verb is often translated as ‘interpret’. However, Plato often
uses this group of words to describe an articulation or an utterance; e.g. at Tht. 209a; likewise at Soph. 246e
the verb describes Theaetetus’ taking the role of the reformed giants (he is supposed merely to express their
view, not to do any more on his own behalf ).
45
That is, each of the things that figures in the announcement: reason does not check whether there is
one announcement or two.
is supposed to sort out, and to see as distinct (524c7), quite to the contrary of the
deliverance of perception. What is it for perception to muddle something up? Here
the issue cannot be that perception provides the soul with two messages, which
the soul then mixes up together and needs to separate out. On the contrary, the
mixing up is done by perception: it reports as one large and small, hard and soft. In
that case the report of perception must be not only discursive, but also complex.
The soul succeeds in seeing each property as one (hence the predicatives at b10–c1),
while perception sees large and small not as separate, but as something mixed up.
The complex structure of perception’s deliverances falls out from this directly. What
perception actually says is, ‘this (same) thing is both hard and soft’ (option (b)
above).46
So each report of perception has propositional content: seeing is seeing as, and
even bare perception reports a complex situation. That complexity is what causes
soul to wake thought up (524d). Without it, we should be left with the dull series of
distinct pieces of information: ‘this is a finger’, ‘that is a chair’; and without waking
thought up, we should be left in confusion: ‘this chair is both hard and soft.’ What
perception does is complex, but banal: the questions arise in the soul after its report.
The deliverances of the senses, then, are not represented here as providing purely
empirical foundations for thought, foundations different in kind and in mode of
access from the comparisons and puzzles forced on us by the conflicted appearances.
Instead, the contrast between sight and what the soul does is offered in terms of the
non-reflective attitude of sight (it merely reports the conflict), and the puzzlement
the soul feels when it realizes there is a conflict—that puzzlement is what drags it to
think. Socrates’ point is then not that the soul operates in a quite different, because
discursive, medium from perception: indeed, that would render incoherent his
account of the failings of perception (and leave insignificant the warning that he is
not going to speak of dim and failing sight). Instead, the soul is provoked—puzzled—
by the deliverances of perception, and so has something to do: to resolve the puzzle.
It is the puzzle itself then—both the puzzle as it appears in the report of perception
(a logical aspect) and the sense of puzzle in the soul (a psychological aspect)—which
generates the move up the line towards the full account of what largeness is.47
The interaction between perception and the soul, despite the appearance that it is
described here in non-metaphorical terms, is dominated by the metaphor of conver-
sation. The conversation is initiated when perception delivers a message to the soul: a
message about the way the world is. That message puzzles the soul, and causes it to
try to find out what is meant. The soul asks the mind some questions; and the
46
If this is a report, not an answer to a question, it is, in the terms I suggested in n. 37, banal, not
philosophically loaded: the philosophical issues are raised after perception delivers its report. It follows,
I think, that we should prefer an interpretation that has the subject of perception’s report as one of the
ordinary individuals of the perceptual world.
47
See Politis (2004a), (2004b).
dialogue between soul and mind is directed towards questions such as ‘what is the
large?’. So the report of perception leads the soul to the consideration of being: to
consideration of what it is to be one, or large, or soft. But the soul is led by perception
as discursive, by ‘seeing’ as ‘seeing as’.
Nonetheless, perception’s role is to connect the soul with the way the world is.
Even though it does not, as I have argued, deliver unmediated, raw foundations for
the ratiocinations of the soul, but reports something that the soul finds puzzling,
perception is not thereby represented as mistaken. On the contrary, as Socrates’
correction of Glaucon’s suggestion (at 523b: ‘surely we are talking about things seen
dimly, or about illusions?’) makes clear, perception is still thought of as delivering
reports that are somehow true of the way the world is. Perception’s contact with
reality is the starting point of the conversation in the soul.
The account of the conversation that takes place within the soul has several points
of contact with the conversation that is represented in the dialogue itself. Socrates
anticipates this parallel at the outset:
‘I shall try’, I said, ‘to show how it seems to me. When I distinguish, on my own behalf, between
those things which lead in the direction we said, and those which do not, then you must view
them with me, and agree or disagree, so that we may see this more clearly too, if indeed it is as
I pronounce it to be.’ (523a5–8)
The verbs used to describe the report of perception and the soul’s conversation are
used by Socrates to describe what he is doing for Glaucon: he is showing Glaucon
what he (Socrates) thinks (523a5, compare 523e7); he invites Glaucon to join with
him in viewing or seeing (523a7, 523a10, compare 524e10); he delivers an oracular
message (523a8, compare 524b1); they are both invited to see more clearly what the
message means (523a8, compare 524a7); and part of that clarification will come
through agreement or disagreement with what they each say (523a7, compare 524e5).
So the philosophical conversation represented by the Republic itself is analogous to
the conversation that takes place in the soul, when it puzzles about the report of
perception. This analogy presses the question again: how comprehensive is the
conversation of dialectic?
4. The Soul’s View (of the Good)

If perception is discursive, then its role, I claim, cannot be to provide foundations in
some special, unmediated way.48 For what perception reports is already mediated by
the propositional nature of its utterance. How, then, will sense-perception serve as an
48
One might argue that the fact that perception reports the way the world is (save when the organ fails,
or is confronted by an illusion) is sufficient for its report to be foundational. In that case, perception’s claim
to being foundational is not a matter of its being unmediated, but a matter of its being true. I have seen this
matter differently (at McCabe (2000), ch. 7); but see Frede (1999).
analogue for the quasi-perception of the soul when it reaches the form of the good, or
the physical eye for the eye of the mind (e.g. 518c–e)? How can the form of the good
be foundational for knowledge if it is not grasped in the special manner we might
attribute to quasi-perception, or to intuition, or to something otherwise intellectually
raw?
When perception makes reports, such as ‘this (finger) is large and small’, the soul,
reacting to its own puzzlement at the conflict, separates out the two confused claims,
reflects on them and thinks about just how to resolve them and explain the terms
used. This dialogue in the soul has a crucial second-order element; for the soul’s
puzzlement is a result of its noticing a problem with perception’s report; and the
questions it asks of thought are reflective on the structure and grounds for percep-
tion’s message.49 The soul’s questions, that is, include several other (first-order)
claims in their scope. Throughout the imagined dialogue, the role of the soul is to
take the detached stance of considering the reports and the answers to its questions,
as it were from outside. This detachment, therefore, is reflective and second order.50
We might say the same of the soul’s silent dialogue:
socrates: So by ‘thinking’ do you mean the same thing as I do?
theaetetus: What do you mean by it?
socrates: When the soul goes through a dialogue itself with itself about something it is
considering. Of course it is in ignorance that I tell you this, but this is how it seems to
me. As the soul thinks, it seems to me to do nothing but have a conversation with
itself, asking itself questions and answering them, agreeing and disagreeing. When it
comes to something definite, whether after a slow process or swift as a flash, and it
now says the same thing and doesn’t dissent, we call this its belief. So I call this saying
believing, and the speech I describe I call belief, not speech to another nor aloud, but
in silence and to itself. (Tht. 189e–190a)
The silent dialogue seems to represent ordinary thinking;51 but its structure bears
significant similarities to the dialogue of the soul in the Republic in two important
respects.
First: in the silent dialogue, as in the Republic, there are two parties to discussion: at
times this is described as the soul ‘talking to itself ’, at times as though different parts
of the soul take opposite roles, one part assenting, the other dissenting.52 But the soul
takes a third role: of ‘wanting to reach a judgement’ (Phlb. 38c5–7); of ‘considering’
(Tht. 189e7). In that role, the soul seems to have an overview of the whole discussion,
49
This is most noticeable for a de dicto reading of perception’s report.
50
See McCabe (2000), ch. 8, on ‘detachment’ in this sense.
51
For discussion, see Dixsaut (1998); Sedley (2003), ch. 1, (2004); Long (2004).
52
For my purposes here it is indifferent whether this divides the soul or simply illustrates the soul’s
taking different positions over time. Notice, again, the theme of assent and dissent, e.g. at 523a7, 525d–e,
529a, 533b.
and to assess what is going on, since the outcome is represented, in both texts, not as
the soul merely plumping for one side or the other, but as reaching a thought-out
judgement. The forming of that judgement comes about through reflection, through
higher-order thought about the dialogue that is going on.
Second: the final stage of the silent dialogue, the judgement that is reached, is
represented in the Theaetetus as the soul ‘saying the same thing’ (190a3); this occurs
when a belief has been reached (or, if the discussion is held aloud, a statement, logos,
Phlb. 38e353). At that stage there is no longer a discussion, no longer any differences
of view within the soul—instead, in the metaphor suggested by the Philebus account,
the journey comes to an end. The finality of this point is expressed in the silent
dialogue by there being a single, unified view in the soul: this is the belief.
If we ask just how these accounts of what are, after all, pretty ordinary cognitive
episodes—deciding what is under the tree over there, or whether Theaetetus is
Theodorus, or the beautiful ugly—bear on the grand business of dialectic in the
Republic, these two features provide two points of contact. First, thinking seems to
have an essential second-order element: and that, as I have suggested, is central to the
Republic’s dialectic. But the second feature, the unity of the soul’s belief at the end of
its dialogue, may assist with understanding the point of the analogy between the final
stages of dialectic and perception.
The soul sees the good after it is puzzled, as a result of reflection on that
puzzlement. For as it puzzles, the soul asks, for example, ‘what is the one itself ?’
(525a1); and this question leads it to the contemplation of what is (525a2), or to
the grasp of being (525b3) and thence to its grasp of the unhypothesized beginning—
the form of the good.54 The return of the language of ‘learning’ at this point in the
discussion (525a3, 525b1 ff.) points to the connection between this part of the
philosopher’s activity and the ‘greatest learning’ of the form of the good, at 505a.
But the analogy between what the soul does at this final point, and what sight is said
to do, is still at work, too:
‘But work it out from what has been said before,’ I said. ‘For if the one is adequately seen itself
by itself, or grasped by any other sense, then it would not be something that drags us towards
being, just as we said for the case of the finger. Yet if some sort of opposite to it is always seen at
the same time, so that it appears no less one than also its opposite, then this will demand
someone to judge it, whose soul will be compelled to puzzle and inquire, moving thought
within itself, and asking what the one itself is. In this way learning about the one is one of the
things that drag and turn us towards the contemplation of being.’ (524d8–525a3)
Sight, no less than thought, may provide the soul with something ‘itself by itself ’; the
contrast between sight and thought in this passage is urged, to repeat, not in terms of
53
See n. 1: the expression logos is normative here.
54
There are those who say that the form of the good and the unhypothesized beginning are not the
same; in what follows it will become clear that, and why, I think they are.
the objects, but in terms of the puzzling nature, or otherwise, of what perception
reports.55 Soul or thought grasps the ‘itself by itself ’ in the cases where perception
reports a puzzle. But perception reports both the uncontroversial cases (the finger)
and the controversial ones (hard and soft); so Plato’s point cannot here be that
thought is veridical just when it is analogous to the passivity of perception to the
uncontroversial cases—since, on such an assumption, perception would be passive to
the controversial cases, too. Passivity is not the point. Instead, this sequence of
argument turns on how thought is able to produce uncontroversial cases (the one
itself by itself ) from controversial ones, deciphering perception’s messages and
turning them from the coded versions, which contain an opposition, into uncoded
accounts of something ‘itself by itself ’. The summit of the line, therefore, demands
that what the soul says be said in clear.
So perception proper is discursive, and this culminating intellection may be no less
so. The intellectual event that occurs at the top of the line may differ from its
dialectical precursor, then, not in kind (e.g. as intuition as opposed to reasoning)
but in terms laid down by the conditions of dialectic. First, it differs in psychological
structure. At the end of the silent dialogue, the soul ‘says the same thing’. At the
end of the process of dialectic, likewise, the soul says the same thing; its view is no
longer puzzled, vacillating, bothered by conflict. In this, it is like perception that is
untroubled, when it delivers its message, by any sense of conflict. In perception’s case,
this is because it is unable to puzzle; in the case of the intellect, the puzzles have been
dispelled by the process of dialectic.
Second, the intellect’s view at the top of the line differs from the preceding process
of dialectic in logical structure. For there are no logical conflicts at the top of the line.
Does this imply that quasi-perception presents the soul with something simple and
unmediated by virtue of its simplicity?
Not so: perception is not simple, and no more is intellection. Socrates suggests that
what happens at the summit of the philosopher’s emergence from the cave is a
spectacle (e.g. at 516b or in the (proleptic) account of the analogy between the form
of the good and the sun at 50956). Again, if we were still to suppose that perception is
unmediated and direct, we might think that the language of spectating here insists
that the observer is passive, merely the recipient of what is borne by his mind’s eye.
But the notion of a spectacle may have a different set of connotations.
To spectate may be an activity, to survey the spectacle, to take it in, integrate it, see
how it all fits together into a whole (notably at 508 and 517).57 This kind of surveying,
after all, is the task of the dialectician after he has reached the pinnacle (537c–d); it is
directly connected to the dialectic of conversation just because it is reflective, engaged
55
Cf. Fine (2003) on the significance of this for the earlier discussion of the contrast between knowledge
and belief, at 476–80.
56
See 511c8, 516a9, 517b4, 518c10, 525a1, 525c2, 526c6, 532c6; and also 517d5, 529b3.
57
See Burnyeat (2000).
in the second-order consideration of what is before it. This, in turn, is connected to

the pervasiveness of the synoptic view; it looks over, looks through and through, it
‘syllogizes’ (516b9, 517c1, 531d2, 537c and see, for an example, 528d7). The soul’s
survey is thus definitive, because it is complete and, in just this sense, final (511c2,
517b8, 532b2, 535a1). This completeness of its vision both mirrors and replaces the
way in which perception produces its message whole. The soul’s vision, on this
account, is unitary, and unconflicted—a synoptic view, not simple or raw.
Once again, the metaphor of conversation is doing important work here. This final
stage, where the soul views the whole, imitates the close of a discussion, the moment
in the silent dialogue when a belief is finally reached. But the final stage of the
discussion can only integrate the whole if the discussion itself is somehow conducted
in sequence, connectedly. I suggested that the model of philosophical conversation
trades on just such a notion of the sequence of a conversation; it is explicit in these
pages of the Republic as the giving and taking of reasons (531e4, 533c2, 534b); as the
asking and answering of questions (515d5, 524e6, 526a, 528a5, 534d9); as the way in
which puzzlement provokes a question and demands an answer—as the entire
process of turning the soul towards being. This is what I called the sequential feature
of conversation; and that sequence is integrated into the final view of the soul.
5. The Good as Cause

But this hasn’t said enough, you might reasonably complain, about the good, or about
how this final spectacle explains the philosopher’s grasp of the truth. At an earlier
stage of dialectic the philosopher faces the kind of difficulty urged against Socrates
and his method of question and answer. For when someone practises what the
Republic describes as dianoia, he posits principles that are treated as self-evident,
and that are coherent with the rest of the system derived from them (510c5–d3). In
such a case, there is no independent verification of the system; instead, it is entirely
possible that the entire structure of principles and consequences, albeit sequential
and coherent, is comprehensively wrong. In the account of the divided line, this
difficulty is apparently addressed by the claim that the person practising dialectic
treats the hypotheses of dianoia as mere starting points, until he can arrive at the
unhypothesized beginning. By touching this, he is able to verify everything that
depends on it (511b3–c2). Somehow, therefore, the philosopher’s touching of the
unhypothesized beginning, the form of the good, provides the verification demanded
by Meno: it allows him to know that his conclusion is the right answer to the question
with which he began.58
How does the ascent to the form of the good offer such verification? In the first
place, the move from the system of thought based on hypotheses to the system based
58
See discussion of this issue in Scott (1995), chs. 1 and 2.
on the good may be characterized as a move away from what is true just by virtue of
its coherence with its system, to what is true by virtue of some fact of the matter,
independent of the mind of the philosopher. This realist dimension to the philo-
sopher’s knowledge is crucial to explaining just why the top of the line is unhypothe-
sized (it is real); and in showing how there would be just one true system that
knowledge knows, or understanding understands (reality is, one might reasonably
think, unique). But how does Plato make a bridge between the hypothesized system
and the one true one?
Here again the version of the process of dialectic against which I have been arguing
has an answer: the form of the good is the single and simple item that is responsible
for the verification of all the rest. It does so, that account would insist once again, by
being apprehended in a quite different manner from everything else: grasped raw and
unmediated. But if this is the only way of bridging the divide between hypotheses and
the unhypothesized beginning, we are back to the problem with which I began, of a
fracture within the method of dialectic itself.
I have argued, however, that Plato’s model of perception in these pages of the
Republic is discursive, not raw and unmediated; and that intellection should be
understood analogously. And, indeed, the form of the good cannot be strictly simple;
for it is not only the object of intellection, but the cause of everything else, and the
source of its intelligibility.59 Consider the discussion of the light of the sun, which
yokes together the faculty of sight and the power to be seen (507e ff.) and is itself seen
by what it explains (508b). Likewise, the light of truth that comes from the form of
the good, and that explains the intelligibility of what is known (508d), is itself the
object of knowledge, the greatest learning (505a). The cause of knowledge and truth
and what it causes are tied together, and they form (as does the sun for the sensible
world) a complete whole.
But the form of the good, of course, also represents something more: it is ‘beyond
being in authority and power’ (509b9–10). This remark, famously, started a tradition
of discussions: is the good somehow mysterious, mystical, ineffable? To suppose so,
I submit, would take us, once again, away from the carefully wrought discussion of
dialectic in the Republic. Recall that, so far, the view of dialectic as philosophical
conversation presses both the sequential nature of what is discussed and the synoptic
view taken of it by the philosopher. From the psychological point of view, this higher
perspective, this second-order thinking is what allows us to understand the complete
system, to take an entirely synoptic view. But there is a further demand put on such a
sequence by the conversational model: that it be conducted by the giving and taking
of reasons. We may understand this from the objective point of view in terms of the
causal structure of what is known. The good is the cause of everything else; and we
know just when we grasp that it is so: when we ‘see’ the good as the cause, and of the
59
I am grateful to Vasilis Politis and Christopher Rowe for discussion about what follows.
things it causes. But that grasp can be neither non-discursive nor of something
simple: the cause offers knowledge when we see just what it is the cause of; and
knowing the cause is itself complete (on pain of causal explanation becoming
regressive), and so authoritative and final. Such an account is also reflective; for we
know not only that the cause is a cause but also that our knowledge is a consequence
of seeing the causal structure of things. For such knowledge, the second-order
perspective is essential.
Think again about the second limb of Meno’s paradox: how do we know that we
know? The Republic’s dialectic is interested both in the psychological aspect of
coming to know60 and in the systematic, objective nature of what is known. We
might rephrase Meno’s question, therefore, as one about understanding: how do we
understand that we understand?61 If we think of dialectic as conversation, both the
psychological and objective aspects come into view. In its psychological aspect,
dialectic takes a synoptic, systematic, and reflective view of what it concerns. In its
objective aspect, that systematicity is assured by the explanatory structure of caus-
ation, itself the object of dialectic’s study.
This objective aspect is brought out strongly by the role of perception in Plato’s
account of dialectic. For while perception is not here construed as raw or unmedi-
ated, it does still have a peculiar feature of veridicality. Perception is, in the way that
thought is not, unpuzzled, because perception makes its reports wholesale; bur it is
not thereby false. On the contrary (recall Socrates’ caution with Glaucon’s talk of
perceptual errors), perception does indeed connect with the world as it is (507d–
508b). When I see, even when my seeing is complex and discursive, my seeing is
somehow veridical: seeing a finger, for example, as large and small is still a case of
seeing it the way it is—even if that is, on reflection, confused and hard to understand.
But the psychological aspect of that seeing is captured by the way in which seeing
seems to cross a perceptual threshold: ‘Yes, yes, I see it!’. In the same way, we use
perceptual or physical metaphors in English to capture the crossing of the cognitive
threshold when we suddenly understand something we failed to understand before—
we ‘see’ it, we ‘grasp’ it, we ‘get’ it. From these moments, we are no longer puzzled or
confused—but quite sure of what it is that we have understood.
In that case, there is an exact parallel between the untroubled stance of perception
when it delivers its report to the soul and soul’s untroubled grasp of the truth when,
indeed, it does see the unhypothesized beginning. The essential feature of the
analogy, however, is not something about how perception occurs but—as we have
already seen—in the cognitive stance adopted by perception: it does not puzzle, it
60
I avoid calling this ‘subjective’ to avoid raising a series of epistemological issues that would be out of
place here. See }2 on the two ways of understanding the rawness of a feel.
61
‘Understand that we understand’ looks clumsy; but it makes the point that a part of understanding is
the second-order element, understanding that we understand. On understanding, see, notably, Burnyeat
(1981); Nehamas (2004).
simply states. When the soul reaches the final stage of dialectic, likewise, what falls
away is its puzzlement, not its discursive grasp of what it understands. So those
moments when we step across the cognitive threshold are not individual events of
raw perception or quasi-perception—on the contrary, that sense that we suddenly
understand arrives when we see how things fit together, how they are all explained,
and when we see, further, that we see it; this is why the form of the good is beyond
being in authority and power. The language of perception, construed thus discur-
sively and reflectively, offers Socrates the analogy he needs for the crossing of the
dialectician from the process of inquiry to the state of understanding. The process of
dialectic is thus continuous up to, and including, its end; for it culminates—as, in a
way, it began—in the untroubled satisfaction of ‘I see!’. And it does so when what is
seen is the explanatory structure of the whole.
6. Knowledge, Understanding, and the Good

You might still complain that there is just not enough about the good in all this: not
enough, first, to account for the normativity of dialectic; and not enough to show why
the choice of the form of the good as the end of philosophical inquiry is anything but
tendentious. Even if Plato can show how to do good dialectic, can he show how
dialectic is good? The question is a huge one, and well beyond the scope of this
chapter. I offer, instead, a brief speculation about how philosophical conversation
might figure in its answer.
The paradigm of philosophical conversation is normative: it is something that can
be done well, or badly. If I am right about its structure, this normativity bears both on
the psychological conditions for dialectic and the objective ones: dialectic reaches the
truth out there in the world just if and just when the soul of the dialectician is in the
right (discursive and synoptic) state for the truth with which it is presented. So for
dialectic to be done well it is not enough for it to issue in a reliable and defensible
grasp of the truth; for then any method might do, just so long as this end is reached.
Nor, however, is our cognition determined simply by the conversations we have, nor
the subjective decisions we may make about conflicted points of view. Instead, just as
in cases of perception, conflicting appearances invite us to think—not from a
standing start, but from something already fully cognitive; and when we reach the
end, the end is not a singular event, a vision to which we are passive, but a synoptic
view of what is completely informed by the good. As we move up and down through
the intelligible world, the interrelation between what is intelligible and our cognition
of it is uniformly dialectical. This outcome of philosophical inquiry is rightly seen as
understanding, rather than knowledge: for in this way the state of mind of the
knower is fully engaged with what is known.62
62
Burnyeat (2000), 70—as the philosopher progresses through and up the disciplines, he gets a more
and more integrated vision. ‘Someone who has achieved that integrated vision has not only assimilated a
If this is right, then the conversational model is essential to the way in which
dialectic hopes to explain understanding. But it still has the appearance of something
uniformly epistemological: does it tell us anything more substantial about the good,
or about virtue? That seems to be what Glaucon is promised (528a).63 If we know
what it is to do good dialectic, do we also see that doing dialectic is good?
One answer to this question might be that if knowledge is of the good, our
knowledge always provides us with the right thing to do. Having knowledge of the
good, this answer declares, allows us to be regularly successful (certainly always more
successful than without it). The advantage of dialectic, that is to say, is an instru-
mental one—that it uniformly offers the right answer to ethical questions; and thus
supplies happiness (and virtue too, perhaps).64
This instrumental account of the importance of dialectic is one that will explain
why philosophers should rule (they will always make the best decisions for the state);
but it fails to account for why I should care for knowledge or why understanding
should be what I search for. After all, if my state is ruled by a philosopher-king, his
decisions on my behalf will be as good as my own would be; so there is no need for
me to trouble myself with learning mathematics or with the hard business of
philosophical examination; I can simply enjoy the benefits of my ruler’s decisions
on my behalf. All I need to know, for happiness, is someone who knows. Once again,
this account of the role of the good in my intellectual life is too thin to account for the
complexity of Plato’s claims for dialectic; notably, it fails to show why the psycho-
logical conditions on dialectic are anything more than a pious hope for the
philosophical life.
Instead, I suggest, we should attend to the way in which the process of dialectic
transforms the soul of the dialectician,65 and to the way in which this transformation
is related to the good.66 Suppose we find ourselves clambering out of the cave by
means of dialectic; we might think that each higher step is better than its predecessor,
just in the sense that it is nearer to the point when we are actually wise (see 531d).
The situation is different after we have seen the good, after wisdom and virtue
become the settled state of our souls (if that can ever happen). For from that point
everything we do is informed by our understanding. This is not an instrumental
claim about how our virtue, or our knowledge, makes us better at getting goods.
vast amount of mathematics. They have assimilated it as a structured whole. And for Plato, assimilation
means that your soul takes on the structure of the abstract realm you study.’
63
Notice Socrates’ rejection of the instrumental account of the value of learning at 527a–b.
64
This is the tip of a large interpretative iceberg: the nature of Platonic eudaimonism. This should not,
in my view, be construed in a thinly consequential way—or even fatly so; but see different approaches by
Irwin (1995); Penner and Rowe (2004).
65
Burnyeat (2000), 56, ‘For Plato, the important task of ruling is not day-to-day decision-making, but
establishing and maintaining good structures, both institutional and psychological.’
66
See here Nightingale (2004), 116.
Instead, the point is that this state of soul makes valuable the features of our lives,67
makes them into good lives; for then the good operates as an informing cause.
When our souls understand and are wise, then, the ethical structure of our lives is
determined by that wisdom; and the sight of the good at the top of the line is thus
transformative. The eye of the soul turns the whole soul with it (518e); after that
point, the value of everything is determined by virtue; one thing is better than another
in this sense (and in contradistinction to the way in which things can be better on the
way up the line). Ethical sight, on this view, makes things (actually, really) valuable in
the life of the person who sees them, and the virtue of wisdom thus transforms the life
of the person who sees the good. That transformation occurs by virtue of the bridge
philosophical conversation builds between the soul of the dialectician and the reality
of the good.
This, then, is what it is for the form of the good to be reached by dialectic. The
form of the good on this account is the sovereign good, the source of all value.68 Its
role as the sovereign good in the life of the dialectician is for it to make value in his
soul and his life; and that happens by virtue of his state of soul. For the good to be
sovereign for him, that is to say, he must satisfy both the objective and the psycho-
logical conditions for dialectic. Without that, what seems valuable is an illusion, a
dream of the good:
‘So, then, do you call “dialectician” the person who grasps the account of the being of each
thing? Surely you will not say that someone who has no account, to the extent that he is unable
to give that account to himself and to another, has understanding of it?’
‘How could I say so?’ he said.
‘So likewise for the good: someone who cannot distinguish the idea of the good in account
by marking it off from everything else, and who cannot get through all the tests of what he
thinks as if through a battle, nor is eager to test it according to the way things are, rather than
according to opinion, and who cannot progress through all these things without his account
collapsing—such a person you will surely say, knows neither the good itself, nor any other
good. And if perchance he gets hold of some kind of image of it, he does so by opinion, not by
knowledge; and living his present life in dream and sleeping, before he ever wakes up here, he
arrives in Hades and sleeps at last.’ (534b3–d1)69
67
This is the point made at Euthyd. 281; for the virtuous man, poverty is better than wealth for the
vicious. Cf. Burnyeat (2003) and Chapter 12.
68
On this issue, see Korsgaard (1983) (who describes this as the intrinsic good); Williams (2003);
Burnyeat (2003); Broadie (2005); and Chapter 12.
69
It is a pleasure to me to offer this chapter to my dear friend Dorothea Frede (in the hope that the
pleasure will not be too deplorably mixed, for her). It was first delivered at the Leventis Conference on the
form of the good, organized by Terry Penner, March 2005. I am immensely grateful to the participants in
that conference for their comments and suggestions, especially to Terry Penner, Tim Chappell (my
commentator), Lesley Brown, Chris Gill, Thomas Johansen, Vasilis Politis, Christopher Rowe, George
Rudebusch, and Dory Scaltsas; also to Jonathan Lear, John Cleary, Owen Gower, Alex Long, and Jimmy
Doyle, to an anonymous commentator for CUP, and also, especially, to Verity Harte.
7
Does Your Plato Bite?
1. Reading Plato Whole

Why should we bother with ancient philosophy, at all? Worse still, why should we
bother with Plato, at all? There are, you might say, at least three strikes against him.
The first is that he has been dead for two thousand, three hundred and fifty-odd
years. Philosophy may not progress very fast, or even very predictably—but Plato’s
antiquity may be thought excessive. Surely we have come further than this by now?
The second is that he just doesn’t write philosophy properly. Consider Frege:
What are called the humanities are closer to poetry, and are therefore less scientific, than the
exact sciences, which are drier in proportion to being more exact; for exact science is directed
toward truth and truth alone.1
Tough luck for Plato then. For he writes with style—and that brings with it the sort of
linguistic curlicues that analysis, or exactitude, or scientific explanation should
deplore. And he writes in stories: not only intense dramatic pieces such as the
account in the Phaedo of the death of Socrates, but also irony and slapstick (you
may recall the episode in the Charmides where everyone wants to make a place on the
bench beside himself for the beautiful young Charmides; everyone budges up, and
the one at the end falls off ). What does the telling of stories—the telling of stories
with curlicues—have to do with philosophy, or with the clarity desiderated by
modern analytic techniques?
The third strike against Plato is his bad arguments. Jonathan Barnes recently
complained that:
. . . Plato’s philosophical views are mostly false, and for the most part they are evidently false;
his arguments are mostly bad, and for the most part they are evidently bad.2
Perhaps Plato’s only real value for the modern logician—or the modern metaphys-
ician, come to that—is the reassurance offered by his colossal mistakes. If we fear
being caught out, we may be consoled by doing some catching out for ourselves.
If Plato’s arguments are bad, should they interest us? And whose arguments are
they, anyway? Remember that in Plato’s dialogues there is no Plato, only a motley
1 2
Frege (1997), 330. Barnes (1995), xvi.
126 DOES YOUR PLATO BITE ?
crew of Socrates and his friends, along with some rather more remote (and dismal)
protagonists such as the Eleatic Stranger. It looks—when we read the dialogues—as
though Socrates (and the Eleatic Stranger) get the good parts, the stronger argu-
ments, the fatter theories; while the collection of friends, opponents, arrogant orators,
or dismal stooges are saddled with the poverty of untenable theories, crashingly
mistaken concessions, or, worst of all, empty agreement with Socrates—‘yes, Socra-
tes’, ‘certainly, Socrates’, ‘it seems so to me, Socrates’—the dead punctuation in the
dialogues that the translator struggles to turn into verisimilitude. This appearance
has at some time or other tempted many of us to suppose that Socrates just is Plato.
Then, the bits and pieces of the dialogue that surround what Socrates says are either
mere decoration or a device for putting a different point of view, soon to be
demolished by the words of the master.
Of a piece with this might be Plato’s shockingly prejudicial presentation of
Socrates’ opponents: especially his professional opponents, the sophists. They are
poor losers (Critias, sometimes), weak defenders of their own positions (Gorgias), or
simply sophists in our pejorative sense: peddlers of a crass competitive ethic by
means of even more crass argumentation. That ethic—the sophists are supposed to
believe—justifies the argument: if all that matters is winning, any argument, on any
rules, will do. The point, on this account, of representing the sophists thus is to make
us see that they are the bad guys. This, too, makes us see that Socrates is the good guy;
accordingly, perhaps, these dramatic representations make us readier to swallow the
(sometimes thoroughly suspect) views that Socrates feeds us.
This Plato, you might agree, is desperately annoying—not only because he
manipulates us towards his philosophical conclusions, but also because he does so
by poor logical means. Accordingly, the temptation to read him this way has been
firmly resisted in recent times. Instead we have been encouraged towards a more
catholic approach, more sceptical towards Socrates and more tolerant towards his
interlocutors. Who Plato is—and what he means to say—is thus less easy to read off
from the dialogues. To interpret him, therefore, we must read him whole, tackle the
arguments in context, attend to the detailed settings in which his characters speak.
Only then, we might think, can the philosophical sense of a dialogue be found.
This, however, may revivify the Fregean objection that not only are there too many
curlicues here for immediate clarity, but that ultimate clarity is missing too. If we are
to read Plato whole, how on earth are we to understand what he means?
2. My Dog, My Father
I invite you to think about a case in point. In the Euthydemus two desperate
charlatans—the twin sophists, Euthydemus and Dionysodorus—show off their argu-
mentative skills at the expense of two young men, Cleinias and Ctesippus. Apparently
starkly juxtaposed, Socrates mounts a series of arguments on the subject of wisdom as
the intrinsic good. No real confrontation appears here. Instead, the dialogue seems
DOES YOUR PLATO BITE ? 127
merely to offer the two cases for our inspection—Socrates on the one hand, the
sophists on the other—and to invite us to choose Socrates. This is not least because of
the dreadful arguments with which the sophists seem to be saddled.
Consider what seems to be one of the very worst. It comes late in the dialogue; the
sophist Dionysodorus is asking the questions, and Ctesippus, newly bitten by the bug
of sophistry, is answering.
‘Tell me,’ said Dionysodorus, ‘do you have a dog?’
‘Yes, indeed,’ said Ctesippus, ‘a dreadful creature.’
‘Does your dog have puppies?’
‘Yes, all as bad as he is.’
‘So your dog is the father of them?’
‘Yes, indeed—I saw him covering the bitch.’
‘Well then, the dog is yours, isn’t it?’
‘Certainly.’
‘So, since he is a father, and yours, surely the dog is your father and you are the brother of
puppies?’ (298d–e)
Ctesippus does not answer.

You will, of course, remember the exchange in The Revenge of the Pink Panther:
‘Does your dog bite?’ ‘No.’ ‘Hello little dog.’ . . . wurrawurrawurra . . . ‘Ouch—you said
your dog did not bite!’ ‘That is not my dog.’ The conversation between Dionysodorus
and Ctesippus has the same flavour of daft inconsequence—is there any more to
Plato’s contribution on canine relations than there is to Inspector Clouseau’s version?
The standard move is to say a) that the dog argument is an outrageous fallacy
and b) that it only goes to show what frightful charlatans sophists are. The fallacy is
outrageous—this view will say—just as a matter of self-evidence: you can just see that
there is something wrong with it. In that case, anyone who uses such an argument
must be up to no good. Proper serious philosophers will have nothing to do with it;
and proper serious philosophers will give the diagnosis of the fallacy itself short
shrift. It’s a bad argument; it’s obviously a bad argument; and Dionysodorus is
obviously a bad man: why bother?
Why is it such a bad argument? Why is it so obviously a fallacy? Well, the
conclusion is false, as the dialogue itself makes manifest: however unruly Ctesippus
may be, he is clearly not a mongrel (or even a Cynic). And yet the premises seem
reasonable: within the dialogue’s fiction, there is no problem in supposing that
Ctesippus has a dog, or in supposing that his dog is fertile. So the problem must lie
in the fit between premises and conclusion: and that should lead us to suspect the
structure of the argument itself. It is that suspicion that grounds the charge of fallacy;
for it is the validity of the argument that is in question, the sneakiness of the sophists
in proceeding from what is evidently true to what is evidently false. But where,
exactly, does the sneakiness lie?
Start by conceding an historical point: that this argument antedates any formal
classifications of fallacy. So interpretation should strive not to depend on what came
after. Still, you might think that the problem lies in the language (while resisting, of
course, the temptation to suppose the problem therefore trivial). Perhaps we have
here a sleight of word. There is a difference between saying that the dog is yours and
saying that he is your father; only by ignoring that difference may we infer from the
claims that the dog is a father and the dog is yours that the dog is your father. What
sort of a difference would that be, though?
Try this: if ‘this dog is small’ and ‘this dog is smelly’ then you can add the words
up: ‘this is a small smelly dog.’ You can’t do quite the same simple sum of ‘the dog is a
father’ and ‘the dog is yours’ to get ‘the dog is your father’. But why not, exactly? Or
try this: perhaps the sophist pulls the thing off by silence—while Ctesippus has
acknowledged that the dog is the father of the puppies, Dionysodorus treats this as
the admission that the dog is a father simpliciter. The trick is turned, in that case, by
simply suppressing the qualifiers in Ctesippus’ admission—by just failing to mention,
that is, whose father we are talking about. But why should you add the qualifications?
Must you always do so?
Is all this just a matter of speech or silence? Aristotle, amid a pretty impenetrable
analysis (at Sophistical Refutations 166b), says not—and we might be tempted to
agree. The argument seems, after all, to tell us something about the way the world is
(that Ctesippus’ dog is his father); and his sense of insult, like our sense of logical
outrage, comes from something we believe about the world (that Ctesippus’ dog is
not his father), not merely about how we talk about it. So is it a feature of the world
that ‘yours’ does not name a property, in the way that ‘smelly’ does? Or a feature of
the world that whenever you mention a parent you had better mention their
offspring, too? The solution to the fallacy had better be principled—otherwise we
may ward off the dog only to be bitten by a larger animal altogether. What sort of
principles would we have to adopt to protect ourselves from either beast?
3. Gross Contradictions
The dog, however, is not alone. Consider first the two arguments about family
relations that immediately precede the dog:
‘And Patrocles,’ Dionysodorus said, ‘is your brother?’
‘We have the same mother but not the same fathers’, said I [Socrates].
‘So he is both your brother and not your brother?’
‘We do not have the same father—his father is Chairedemus, mine is Sophroniscus.’
‘So Sophroniscus is a father and so is Chairedemus?’
‘Yes—the former mine, the latter his.’
‘So Chairedemus is other than the father?’
‘Than mine.’
‘But is he a father if he is other than a father? Or are you the same as a stone?’
‘No.’
‘So you are different from a stone.’
‘I fear lest you will show me to be the same; but I think I am not.’
‘So you are different from a stone?’
‘Yes.’
‘But being different from a stone, you are just not a stone, and being different from gold, you
are just not gold?’
‘True.’
‘Chairedemus too, being different from a father, will be not a father.’ (Euthydemus 297e–298a)
At this, Ctesippus takes over as the questioner:

‘But,’ said Ctesippus, taking up the argument, ‘won’t exactly the same happen to your father?
Isn’t he different from my father?’
‘Far from it,’ said Euthydemus.
‘Oh? Then is he the same?’
‘The same, indeed.’
‘I’d never concede that. But is he just my father, or also the father of everyone else as well?’
‘Of everyone else—or would you say that the same man is both a father and not a father?’
‘I did think so,’ said Ctesippus.
‘Oh really? Do you think it possible for what is gold not to be gold? Or for what is a man not to
be a man?’
‘But perhaps you are not tying flax to flax, as the saying has it; for it’s a terrible thing to say, that
your father is the father of all.’ (Euthydemus 298b–c)
Ctesippus’ father must be the father of everyone, indeed of every creature too; so that
Ctesippus is the brother of goats and guppies and puppies, and his father must then
be a goat and a fish and a dog.
Now this is all pretty bad news for Ctesippus. He is, after all, a member of the
young aristocracy, and in love with the well-connected Cleinias, who has fallen victim
to the sophists. Ctesippus is self-conscious in front of his beloved, and anxious to
appear at his best (notice his vainglory at 285c–d). An argument that reduces his own
family to goats and puppies does not augur well for the future of his relationship with
Cleinias (remember the insults that Achilles and Agamemnon exchanged on the
subjects of each other’s bestial ancestry in the Iliad, e.g. I. 225). And from the very
beginning the sophists have invited us to think of them as all-in wrestlers, as
competitors whose objective is always to win, to defeat their interlocutors, to humili-
ate those who dare to trade arguments with them (273c ff.). Indeed, as they boast
early on, whatever their opponent says, they will defeat him in argument (275e). But
now the price of that defeat becomes clear: if argument is about competing, and
winning, Ctesippus just lost hands down.
So is that Plato’s point? He is not merely engaged in presenting to us a collection of
bad arguments, uttered by bad men, we might now suppose: instead, is the purpose of
this dismal display to represent to us just how arid competitive argument is? If that is
his purpose, is it achieved by the representation itself ? As with the obviousness of the
fallacies, are we supposed just to see that this is an unsatisfactory way of doing
philosophy (or of doing anything with the remotest claim to intellectual signifi-
cance)? Equally, are we supposed just to see that Socrates does it right? Does Socrates
do it right? And is there anywhere here an account to be given of why any of this
should be so?
Return to the arguments about family relations. Once again, you might think they
exploit something linguistic: the terms of sameness and difference or otherness,
notably, if not that old friend to the ancient philosopher, the Greek verb ‘to be’. So
in the first argument: if Chairedemus is other than some father, then he is other than
a father, then he is not a father. (If you have your logical wits about you, you might
wonder whether the problem here is the quantifiers—if Chairedemus is other than
some father, is he other than any father?—but suspend judgement for a while.) The
second is the mirror of the first: ‘if you are a father you are the same as a father; so you
are the same as any father’. In both arguments, the work is done by similar moves,
which are supported by something rather more general. In the first argument: are you
the same as a stone? No. So you are different from a stone? Yes. So:
‘But being different from a stone, you are just not a stone, and being different from gold, you
are just not gold.’ (Euthydemus 298a)
In the second argument, is it possible for the same person who is a father not to be a
father? Surely not, because:
‘Do you think it possible for what is gold not to be gold? Or for what is a man not to be a man?’
(Euthydemus 298c)
In all these cases, two things seem to happen. First, to the verb: ‘same’ and ‘different’
are taken to be convertible with ‘is’ and ‘is not’, and thereafter to operate as a simple
identity sign or its negation. That’s sometimes fair enough: Ringo Starr is the same as
Richard Starkey because Ringo Starr just is Richard Starkey. But sometimes the
conversion is downright dangerous, concealing as it may a difference between
identity and similarity. However, I shall forgo the delights of the verb ‘to be’ and
enjoy, instead, the second thing that seems to happen here. Consider what comes
after the ‘is’ or the ‘is not’: Chairedemus is not a father. The complement of the verb,
‘a father’, is taken to be a predicate (‘a father’), not a definite description (‘the father
of Socrates’). And this then seems to warrant treating the predicate naked, as it were:
as a one-place predicate: ‘father-full-stop’ rather than ‘father of someone or other’.
Hence the comparison with stone and gold: ‘being different from a stone, you are just
not a stone’ is treated as analogous to ‘being different from the father of Socrates, he
is different from a father, so he is just not a father’. You might have thought that
fatherhood is a relation; but here you are discouraged from that thought. Instead,
fatherhood is taken to be, if you like, a property the subject has, or fails to have. So
family relations are here being re-described in terms of respects in which members of
the family have some property or other (fatherhood, brotherhood, and so on). Again,
there is something plausible here: after all, if Martin is the father of Poppy, Martin is a
father; it just doesn’t work the other way around: if Socrates is not the father of
Poppy, it doesn’t follow that Socrates is not a father at all.
So what? Why bother detailing all the many vices of these arguments? The point,
I take it, is not to tell us something about our common ancestry: although it might be
congenial to Socrates so to undermine a grandiose appeal to noble birth. But we
would do well to notice that this set of arguments pursues a common path: to
derelativize all sorts of expressions (from ‘same’ and ‘different’ to ‘father’ and
‘brother’), and produce them as one-place predicate terms. ‘Different from the father
of Socrates’, ‘same as the father of Euthydemus’ are reduced to ‘is not a father’ and ‘is
a father’, simpliciter. And in the process, these terms come to fall under a general
principle of non-contradiction that is exemplified in the second argument (it is not
possible for gold not to be gold, 298c4–5). The principle is something like this:
Nothing is both thus and so and not thus and so.
Or, removing that dangerous ‘nothing’,
It is not the case that anything, if it is thus and so, is also not thus and so.
You might find this plausible: try some counter-examples. The most familiar principle
of non-contradiction worries about contradictory propositions and outlaws some-
thing like this:
Tony Blair is the Prime Minister of the UK today and Tony Blair is not the Prime
Minister of the UK today.
It outlaws, that is to say, sentences that appear to affirm and deny the same thing at
the same time. Quite right too (if we refuse to worry about recherché counter-
examples from quantum mechanics or arguments from fallibilism). An analogous
principle, then, should govern contradictory properties. The sophists’ principle is
such a one; and it stops us from echoing Heraclitus:
Seawater is healthy and poisonous . . . (DK22B61, in part)
The road up and down is one and the same. (DK22B60)
The beginning and the end are the same . . . (DK22B103, in part)
These are paradoxes, even for a sophist. Their paradoxical nature—and our imme-
diate response to them (‘oh surely not!’)—is explained by their violation of this
principle of non-contradiction. So—we might now say—the principle on which the
sophists rely is a respectable one.
Yet behind it trots the disreputable dog. Suppose the sophists’ principle treats ‘is
the father [of puppies]’ as somehow analogous to ‘is gold’, and ‘is not the father [of
Ctesippus]’ as analogous to ‘is not gold’. Then it will be unconcerned by a move from
a qualified claim (‘this dog is the father of puppies’, ‘Chairedemus is not the father of
Socrates’) to an unqualified one (‘this dog is the father’, ‘Chairedemus is not the
father’). Rather, it will proscribe qualified complex sentences such as ‘this dog is the
father of the puppies but not of Ctesippus’ and ‘Chairedemus is the father of
Patrocles but not of Socrates’ on the grounds that they contain a contradiction (‘is
and is not the father’). In the normal run of things, this would indeed be argumen-
tative mischief, positively Megarian in its malice, with qualifiers dropping like hairs
from the head of a bald man. But here the argument is entirely legitimate, just
because both Ctesippus and Socrates go along with the principle and its application
to sentences describing sameness and difference. Their suspicious responses, how-
ever, underline that this principle is indeed what is at issue. What else should they
have done?
For a start, they should have noticed the principle before: for it has a long and
dubious history in the Euthydemus, not only within the arguments, but also in the
dramatic commentary upon them. At the outset Dionysodorus promised that what-
ever answers young Cleinias gives to whichever brother, he will be refuted (275e). So
Euthydemus induces him to say that it is the ignorant who learn, while Dionysodorus
persuades him that it is the wise (276a–277c). The young man is bewildered and
embarrassed, because in concluding that those who learn are both wise and not-wise
he seems to have entangled himself in a contradiction, irrespective of how his various
conclusions have been either arrived at or qualified. For reasons that will become
clear, I shall call the principle that underlies Cleinias’ misery the gross principle of
non-contradiction.
The gross principle of non-contradiction, by the lead-up to the dog argument, has
become an explicit premiss. But we may well ask ourselves what sort of a premiss it is:
is it—as we were tempted to say of the dog argument—a claim about language? Or is
it, as the formula that outlaws simultaneous assertion and denial, a principle of logic?
Or does it tell us something about the world—something, that is, about the political
status of Tony Blair or the medicinal constitution of seawater? The sophists, it seems,
deploy it as a part of their method: not only is it their victims’ horror of contradiction
that allows the sophists to win in argument, but also the gross principle seems to
dictate the rules of engagement. How we do what passes for philosophy among the
sophists is determined by gross constraints. So, in a parody of Socrates’ frequent
insistence that people who talk to him shouldn’t give him long speeches (he can’t
concentrate for that long, he maintains) the sophists refuse to allow their interlocu-
tors to add qualifications to the answers they give (296b). By this means, for example,
they conclude that if Socrates always knows all the things he knows, then he always
knows all things (296c–d).
This argument is thoroughly annoying; and it is disturbing, too. For it draws
attention to how thoroughgoing the gross principle is: it seems to govern not only
canine pedigrees, but also anything we may say in conversation. Socrates goes along
with it—and that is what irritates us. But if this gross principle determines what
counts as a reasonable answer to a question, how on earth are we to resist becoming a
victim of sophistic dirty tricks?
4. Refined Contradictions
Aristotle, of course, is your man for principles of non-contradiction. He says:
It is impossible that the same thing should belong and not belong to the same thing, at the same
time and in the same respects (we should assume to be added all those additions which are
needed for the logical difficulties). (Metaphysics 1005b19–22)
What does he mean? Translated out of Aristotelese, the principle maintains that
something cannot both have and not have the same property, at the same time and in
the same respects, and under any other relevant qualification. It is this principle that
Aristotle undertakes to defend from attack; because, he says, it is the most stable and
fixed of all. And it looks a pretty good one. It stops me, first, from saying that this
great horse, Sir Desmond, is both grey and not grey, here and now, on his nose.
Second, the principle looks universal: anything we say—or so common sense, that
notorious deceiver, might encourage us to suppose—relies on some such assumption.
So when I insist that Sir Desmond is grey here and now on his nose, I mean also to
deny that Sir Desmond is not-grey, here and now, on his nose.
Now the sophists had earlier seemed to disagree, claiming that every statement is
true, and that contradiction is impossible (283e–288a). That seemed to throw open
the possibility that the two sentences ‘Sir Desmond is grey, here and now, on his
nose’ and ‘Sir Desmond is not-grey, here and now, on his nose’, if uttered at all, could
both be true, without contradiction. By the time of the dog, however, they have
abandoned that position, in favour of the gross principle of non-contradiction:
Nothing is both thus and so and not thus and so.
How come such a principle lands us in the mess of the dog argument?
Notice what Aristotle would add to the formula:
Nothing is both thus and so and not thus and so at the same time and in the same
respects (we should assume to be added all those additions which are needed for
the logical difficulties).
The additions, the extra bits, constitute what I shall call a refined principle of non-
contradiction; and they clean up the dog, twice.
In the first argument, the sophist infers from the claim that Ctesippus’ father is a
father, via the gross principle of non-contradiction (no father could be not a father)
via the claim that Ctesippus’ father is the father of all, that Ctesippus’ father is
indistinguishable from Euthydemus’ and Dionysodorus’ father. Here refinement
might allow the claim that this man is a father relative to Ctesippus, but not to
Euthydemus and Dionysodorus: the sophists’ conclusion will not go through. In the
second argument, the sophist infers from the fact that the dog is the father of puppies,
and Ctesippus’, the claim that the dog is the father of Ctesippus. Refinement would
insist that the dog is a father just of the puppies; or else that ‘yours’ is always filled out
by a qualification: again, the sophistic conclusion will not go through.
Now this refined approach is clearly one that Socrates prefers. For when Socrates
leads the argument, he especially relies on refinement: for example, when he argues
that health and wealth are good in some circumstances (when wisdom leads them)
and bad in others (when folly prevails) (281c–e). So there are two principles of non-
contradiction at stake in the Euthydemus—the gross and the refined. And they are in
competition: to subscribe to the refined principle just is to deny that a gross account
is sound; and to prefer the gross is to suppose that the refinements are immaterial.
But there is more: as the progress of the sophistic arguments makes clear, these
principles are somehow fundamental—since they govern assertion or denial, they
govern everything we say. However, if they are both thought to be basic, how do we
choose between them?
Pragmatism, notice, may not be enough (to say that refinement is more hygienic,
and so preferable). Why must we refine? How can we show that the sophists’
arguments are wrong? Aristotle’s throwaway line, ‘all those additions which are
needed for the logical difficulties’, looks dead easy. But is it? Trying to diagnose the
dog reveals that neither what went wrong, nor its righting, are perspicuous.
The situation may be more dangerous than first appears, indeed. If a principle is so
basic that it governs everything we say, how can it be defended? Not, after all, by
producing something more basic, so not by deductive argument: if every assertion we
make is governed by the principle itself, we just beg the question when we argue in its
favour. Aristotle suggests that we offer a dialectical defence: we should seek to show,
that is, that our opponent, in the very denial of the basic principle we claim, turns out
to be committed to it (Metaphysics 1006a11–12).
But where is the dialectic here? The sophists’ arguments reduce their interlocutors
repeatedly to silence; and they do this by exploiting the interlocutor’s sensitivity to
contradiction. The same, you might think, is true of Socrates’ normal procedures.
After all, many of the conversations he has with others reduce them to silence, to
embarrassed gaping, and to the horrified perception of the inconsistency of their own
beliefs. The same happens here: for he himself describes how silly he feels when
he finds himself unable to escape the labyrinth of his own argument (293a). In fact,
in the Euthydemus alone, there is no direct confrontation between the gross
principle of the sophists and the refined principle that Socrates prefers. Instead,
the dramatic commentary makes a great play of the alienation of each party from the
other. And by the end of the dialogue Socrates is reduced, by the triumphalism of the
sophists, to the sort of muttering in corners that Callicles so despised. This dialogue is
supposed to be a protreptic to philosophy, an exhortation to dialectic: and yet, more
than most other dialogues, there is no manifest dialectic between the protagonists. So
does the triumph of the sophists amount to a victory for gross contradictions? (So at
least they take it, 303b.) And if it does, are we to be reduced to the logical disasters
of the dog?
5. Philosophical Conversations
Socrates repeatedly insists that philosophy is best done by conversation: by the
encounter between one person and another, and the giving and taking of reasons
between them. What is so special about philosophical conversation? What would it
be to give and take reasons in this ideal way? Irrespective of what the reasons are, we
can say something at first about who does the giving and taking: because we can see
from this dialogue what would count as the failure of such a conversation. The
sophists fail to do philosophy because it doesn’t matter what the person they are
talking to says: whatever he says, he will be refuted. The sophists, that is to say, treat
the opinions of the person to whom they talk as indifferent—whatever the person
says, he will end up saying the opposite too, and will then be reduced to silence (so
Socrates insists to the end, 303e). Socrates produces silence too, albeit by different
means: for he reveals, in his unfortunate interlocutor and sometimes also in himself,
inconsistency, regress, and circular arguments (291b). Having a philosophical con-
versation with Socrates, that is to say, is as dumbfounding as talking with sophists.
What is more, since Socrates and the sophists follow different basic principles, they
seem unable to communicate at all. But in seeing the dialectic break down between
them, we see also what we need to prevent the silence. For what is needed here is
difference of opinion: for a dialectical encounter to get going, that is to say, one party
should not (or not yet) be convinced of what the other maintains. It is, that is to say,
some kind of requirement on dialectic that the parties do not agree, are not silenced
by each other or reduced to being the other’s clone (as seems to happen to Ctesippus
as the sophistry unfolds). And this failure to agree brings a refinement with it: that
this proposition is believed by Ctesippus (say), but not, or not yet, by Socrates.
Suppose, now, that there is a dialectical discussion of the gross law of non-
contradiction. If its exponent puts the gross principle forward, he expects his
interlocutor to agree. But if he supposes that there can be any dialectical exchange,
then he must allow that his interlocutor may not agree, or that he himself may not
agree with his interlocutor. In such an admission he concedes refinement: for he
concedes that such and such a proposition may be believed by himself, but not
believed by his opponent. In that manner, the gross principle may be dialectically
refuted.
But how extensive would such a refutation be? May it just be that the gross
exponent should concede differences of opinion, but assert that grossness pervades
everything else? Is a dialectical refinement merely a linguistic quirk, and dialectic the
exception that proves the gross rule?
Consider, first of all, the metaphysical consequences of this sort of dialectic. If
there is a dialectical encounter between two people, then that encounter reflects a
relation of difference between the interlocutors; and it does so in showing up a
relation between one belief and one person, or another belief and the other. Those
relations, however, are real—that is to say, aspects of the real world, not merely words
on the page or in the head. In dialectic, beliefs are related to persons; for dialectic to
work, beliefs need to be thus indexed. What is more, this seems quite true: the best
philosophical conversations take place just when one party does not—or not yet—
agree with, accept the reasons of, the other. This failure to agree tells us something
about the parties themselves: that they observe and respect their differences, because
those differences are not merely about words, but about the way the world is.
But if dialectic is like this, it underpins a refined contradiction. For there to be
dialectic, we must refine; but if we are doing dialectic, we have already refined. Gross
principles of non-contradiction will not survive. How might Plato show us that? Not,
to repeat, as a confrontation between Socrates and the sophists: indeed, this is just
what that confrontation does not show. On the contrary, it shows us the process
where disagreement is expressly squashed. Instead, Plato shows us dialectic by being
annoying . . .
Go back—maybe with some relief—to the dog. That argument is presented
without the commentary I have offered it: all this discussion about principles of
non-contradiction is, you might say, lamentably absent from the particularities of
this text. And yet I have been making the claim that the argument actually focuses
attention on a piece of philosophical logic: on the nature and justification of prin-
ciples that themselves govern argument. How did that ever come into it? It can only
do so if there is more to this text than representation. For the encounters between
Socrates and the sophists may not be just histories of what happened (or even of what
might have happened). Instead, they are contrived descriptions of encounters that are
pure fiction. Why? Well, here is why: because they are set up in such a way as to force
us to consider what has gone wrong. The response we have to the dog just is to say
that it is a fallacy: it just is to complain that the conclusion has come out wrong, that
there must be something fishy about the argument. The very ostentation of the
fallacy, its startling absurdity, is what provokes us to reject it. And in saying,
vigorously, that this argument must be wrong somehow, we refuse to agree. That
refusal in turn constitutes the dialectical context for the refutation of the principle
that underpins it.
Think about it this way. There are some modes of speech that exploit the absurd:
and in doing so they directly provoke disagreement. Fallacy is one such mode;
paradox another (think of the irritant qualities of ‘you can’t step into the same
river twice’). Some jokes are another (I forbear to illustrate). But such provocations
are littered throughout the Euthydemus. Socrates and his followers fail, ostenta-
tiously, to defend the refined principle of non-contradiction from the sophists.
They fail to do so notably when it comes to discussing the very relations that
underpin disagreement: think about how, unless sameness and difference are refined,
we cannot account for the possibility of conversation—of different people talking
about the same thing. But our sense that these sophistic ways of talking are absurd is
based not on some equally contested linguistic propriety, but on a view about the way
the world is. For the world does not exhibit the blank, relation-free quality that the
sophists demand—the quality that allows them to take over their acolytes completely:
no pupil of the sophists is his own man. That the world is not like that is exemplified
by our own response of disbelief to the arguments the sophists produce. Those
arguments are not dealt with by Socrates who is, after all, not real. Instead, the dog
is left to growl at you, who are. The dialectical defence of Socratic logic, that is to say,
takes place not between Socrates and the sophists, but between Plato and his reader.
It takes place only if the dialogues are read as a whole; and it takes place only if we
take seriously their worrying character. Does your Plato bite? Maybe not. But then he
is not my Plato.3
3
It is a pleasure and an honour to offer this chapter to Denis O’Brien, in longstanding gratitude. It was
my inaugural lecture; and Denis was my first philosophy teacher, who practised the methods of Socrates
impeccably, and showed me how the subject should be done: open-mindedly and with care (that
I constantly fail to live up to his example is not his fault, but mine). It is appropriate, too, that
I acknowledge here another great philosophical debt: to all the members of the Department of Philosophy
at King’s College London, who make philosophical conversations a brilliant experience.
8
Unity in the Parmenides: The Unity
of the Parmenides
1. Chinese Whispers
The Parmenides is a dialogue about what there is. It asks two different kinds of
question about what there is. First, how do we count what there is—how many things
are there? Second, how is what there is arranged, so that it is susceptible to reason—
how are we to explain what there is? The relation between what there is and reason,
therefore, is expected to be well formed, structured, and thus unified; indeed, a large
part of the work is an intricate discussion of just what unity is. Yet the Parmenides is
all too often treated as ill formed and fragmented. It seems, in the first place, to be not
one but two dialogues: the first, a vigorous attack on the theory of forms which is at
least comprehensible, even if its exact philosophical purpose is difficult to discern; the
second, a thoroughly peculiar discussion of ‘the one’, better left either to Neoplaton-
ists or to enthusiastic analytic philosophers. In practice, even if not in principle,
scholars deal with one or other of these ‘dialogues’; and the overall purpose of the
Parmenides—if indeed there is one—is lost to view.
This consequence, however, is hardly unexpected; for the dialogue does contain
two dialogues, severely kept apart. The first part is a conversational, vivid account of
the encounter between the young Socrates, the venerable Parmenides, and his
sidekick Zeno, introduced by an elaborate ‘Chinese Whispers’: the story is told
directly to the reader by Cephalus, who had it from Adeimantus’ stepbrother
Antiphon, who memorized the story as he heard it from Pythodorus, the follower
of Zeno, who was himself present when the encounter took place.1 Socrates and
Parmenides debate the merits or otherwise of Socrates’ theory of forms as a system-
atic explanation of what there is. Socrates is reduced to aporia when Parmenides
suggests that either asserting or denying the theory has impossible consequences. The
second part is introduced by Parmenides as a ‘gymnastic session’ for Socrates, to
train him for dealing with definitions of the beautiful and the just and the good. What
1
Chinese Whispers, for those who have never played it, is the children’s game where the players stand
in a line, and the first player whispers a message to the second, who whispers it to the third, and so on until
the message reaches the end of the line. At each stage, the whispered message becomes distorted; the point
of the game is to see just how different the final version of the message is from its original.
UNITY IN THE PARMENIDES 139
follows is formally a dialogue about ‘the one’ (between Parmenides and Aristotle),2
but a dialogue where the respondent offers little but monosyllabic agreement, in
marked contrast to the Socratic appearance of the first part. It ends Socratically
enough, however, with a terminal aporia: ‘Let it be said, then, that, as it seems,
whether the one is or is not, in every respect and in every way both in relation to
themselves and each other the one and the others are and are not, appear and do not
appear’ (166c). To this Aristotle meekly—and surely surprisingly—responds, ‘very
true’. There is no return to the frame, no closure of the Chinese Whispers, and no
final coordination of the second part with the first. Why is this dialogue about unity
shaped in this disunited way?
The Chinese Whispers device turns up elsewhere.3 In the Symposium, for example,
it serves to unify the motley collection of speeches that are made at Agathon’s dinner-
party, closing with Aristodemus’ account of Alcibiades’ drunken interruption and
Socrates’ sober discussion through the night. In the Symposium, the Chinese Whis-
pers effect encloses the whole work, so that the dialogue appears distanced from the
reader, remote and separate, in that it is transmitted through a series of people of
which we are not members. Thus the philosophical content of the Symposium seems
presented to the reader to view, not to engage with; to accept, not to dispute.4
The frame of the Parmenides, by contrast, is never closed. When the terminal
aporia is reached, we are never reminded that this is a story told to one person and
then to another; instead the fictional setting has disappeared altogether. Why?
A preliminary answer might be that here the open frame suggests a series of
narrators of which there are indefinitely many members, and of which any reader
may well be one. As we first hear how the story of the encounter between Socrates
and the men from Elea was told by one person to the next, so we ourselves are but
another stage in its transmission. There is nothing more privileged in Cephalus’
hearing of the story than there is in ours—the distance, if you will, imposed by the
Chinese Whispers in this dialogue decreases, rather than increases, our awareness of
the distortion to which a told tale is vulnerable. We might just as well have been there
ourselves. This dialogue, then, is one that we join, in which we dispute, where we too
take an active part in the argument.
You may object that this is a thoroughly fanciful way of thinking about the
structure of this odd dialogue, and of explaining its three striking literary features:
its disunity, its frame of Chinese Whispers, and the open-endedness of that frame.
You may object further that merely to affirm that the literary style of the dialogue
2
What is the relation between this young Aristotle (subsequently one of the Thirty, 127d) and Aristotle
himself? Cornford wryly observes that this young man is chosen exactly because he will give Parmenides no
trouble; but ‘of all Plato’s pupils Aristotle would have been the least likely to give no trouble’, (1939), 109.
3
Cf. the use of a background conversation between Phaedo and Echecrates to locate the conversation
before Socrates’ death in the Phaedo, 57a–59c, 102a–b, 118a.
4
Cf. the authoritative nature of the figure of Diotima in Socrates’ speech; and contrast it with the
trembling of father Parmenides before the race at Parmenides 136e.
140 UNITY IN THE PARMENIDES
engages the reader in its dialectic tells us nothing very important about the content of
the dialogue itself, only something (either trivial or over-optimistic, depending on
your point of view) about how we should adopt a critical stance towards it. You might
say that we always adopt an active dialectical stance towards a Platonic dialogue
(including the Symposium) if we take it seriously as philosophy; we never (well,
hardly ever) just accept what Plato has to offer.
In what follows, I shall suggest that the reader is put in a specially dialectical
position in this dialogue so that the open frame of this dialogue is particularly telling.
The dialogue progresses, I shall argue, from the discussion of views that are particular
to the protagonists to a discussion that is entirely general; and it does so in three
stages—the Socratic debate of the first part, the discussion of the first hypothesis, ‘if
one is’, and the discussion of the second hypothesis, ‘if one is not’. Each stage reflects
on the theoretical conditions of its predecessor; each stage of the dialogue is of a
higher order of abstraction than its predecessor. Just as the story is narrated by one
person who is reporting what was said by another, and he by another, so the
arguments of the dialogue are ordered, each reflecting on its predecessor; the frame
provides an analogy for the structure of the arguments. The dialogue is then unified
by its topic: the whole discussion is an investigation of what there is, demonstrating
that any ontology, any account of what there is, counts individuals. The discussion of
individuals, of ones, unifies the Parmenides.
2. Dramatis Personae
The story begins not with Parmenides, but with Zeno. Pythodorus comes upon
Socrates listening to Zeno read his book out loud. When Zeno finishes, Socrates
asks him to repeat the first hypothesis of the first argument: ‘If the things that are, are
many, they must be both like and unlike. But that is impossible. Unlike things cannot
be like, nor can like things be unlike’ (127e). Socrates interprets Zeno to mean that
since it is impossible for unlike things to be like and like things to be unlike, it is
impossible that the things that are can be many (127e); later (128b ff.) this argument
is connected with the Eleatic claim that ‘the all is one’.5 This claim is, on Socrates’
interpretation, strongly monistic—all there is is a single individual, just one (it rules
out, for example, the possibility that Socrates might be one among many, as it rules
out the possibility that Socrates might be one with many parts, 129aff.).6
Socrates, on the other hand, does allow that there are many, and that they may be
both like and unlike. For he supposes there to be ‘some form of likeness, itself by itself
and something else opposite to it, unlikeness itself ’ (129a), by participating in which
sensible particulars are both like and unlike, without absurdity. But ‘if someone were
5
It is Socrates who takes Zeno’s argument in this way; Zeno himself is cagier, saying only that he plans
to show that pluralism is ‘even more absurd’ than monism, 128d.
6
This thesis comes under attack at Sophist 244b–245d.
to show that the likes themselves7 become unlike, or the unlikes like, I should think it
a monstrosity’.
There are, it seems, two points of dispute between Socrates and the Eleatic. First,
Zeno makes a point about contradiction: sensible particulars are absurd because they
are characterized by opposites. Socrates denies that this constitutes absurdity, but
concedes that if forms had opposite characters8 they would be absurd (129c). Second,
Zeno makes a point about plurality: sensible particulars are unacceptably plural, just
because they are characterized by opposites. Socrates allows particulars to be plural in
all sorts of ways and still to retain some unity (Socrates is one man among many, he
has many parts and still remains one); but he refuses to admit that forms can be
plural at all, so that they cannot ‘mingle and separate among themselves’ (129e).9
Why, first of all, should anyone be held up for a minute by Zeno’s puzzles? One
diagnosis of his argument might be simply that he has dropped the qualifiers that
complete ‘like’ and ‘unlike’. He may, for example, have argued thus: ‘This crocodile is
like that alligator, and it is unlike the elephant’s child. So the crocodile is both like
and unlike. So this like is unlike. But that is a contradiction. So there is no crocodile,
just the Eleatic One.’ If that is his argument, Socrates has a ready riposte (cf. 129a ff.);
he can easily say that the crocodile is like in one respect, unlike in another, and leave
Zeno standing. He needs no forms for that. Yet Socrates does offer forms as the
answer to Zeno’s puzzle. Why?
Notice, first of all, that Zeno does not conclude, as a sceptic might, that the world is
full of contradictions so that we cannot understand it. Instead, he points to the
contrast between the contradictory nature of plurality and the blissful purity of the
One; and Socrates takes him to conclude that there is only the One. This is a radical
ontological argument: just because the world is confusingly plural, it is not there at
all. Correspondingly, Socrates consoles him by showing how plurality is not so bad
after all—provided we postulate forms. So what follows is already focused not on
some factitious contradiction but on the difficulties faced by plural or complex
entities, difficulties apparently solved by having pure ones as well.
Instead of monism, therefore, Socrates offers a complex ontology. There are
particulars, which are one and many, and there are forms, which are each just one,
simple, and ‘separate, themselves by themselves’. Just so, in the Phaedo, Socrates
supposed there to be particulars, which are ones that are also many, composite, and
7
How much hangs on the plural here (cf. the anxieties that commentators display about the plural ‘the
equals themselves’ at Phaedo 74c)? Nothing, in my view; Socrates clearly intends to contrast the particulars
that are both like and unlike and the form of likeness; the use of a plural expression for the form may be
mere carelessness after the plural for the particulars, and not meant to imply (at least before the Third Man
arguments) that the form of likeness is itself many.
8
I.e. opposite characters from the ones that they themselves explain; e.g. the form of likeness cannot be
unlike.
9
This suggests that forms may not have any opposites: that is, the form of likeness may not be unlike;
but nor may it be equal, or unequal, or large, or small. The forms, on this account, are to be free from
interpredication.
therefore perishable, and forms, each of which is just one, simple, incomposite, and
imperishable. Both forms and particulars, therefore, are in some way one: each is
countable as one because each has some principle of unity. They are items in
Socrates’ ontology just because they are individuals.
The contrast between them can be understood in terms of the properties they have.
First, particulars are many because they possess an indeterminate number of prop-
erties, all of which could—at least on some account—be described as parts of the
particular. This will be true, Plato has suggested elsewhere, whether the property in
question is common-or-garden (green, smelly), or relational (as large as the goose
next door), or evaluative (thoroughly unappealing).10 Second, particulars are espe-
cially many because they are subject to the compresence of relational and evaluative
opposites (if the goose is larger than the frog, there is something somewhere—
perhaps some gander—than which it is smaller, and so on).11 Forms, conversely,
are meant to explain the compresence of opposites, and so cannot suffer from it. It
seems, moreover, that they cannot suffer from any pluralization at all—they are just
one, in contradistinction to the indefiniteness of any particular.12 A form is, thus,
what I call an austere individual (just one), and a particular is a generous individual
(one with many properties). Socrates includes both austere and generous individuals
in his ontology; Zeno, by contrast, who supposes that generosity can only amount to
hopeless profligacy, so that plural particulars have no unity at all, includes only
austere individuals—and only one of them, to boot.
In answer to Zeno, then, Socrates suggests that particulars are doubly parasitic on
forms. First, they depend on forms for the relations and values that qualify them (that
is, this stick is equal to that stick because of the form equal, in which they both
participate). Second, they depend on forms to save the pluralist’s ontology from
degenerating into chaos. Socrates seems to suppose that to protect us from Zeno’s
inference, that if particulars are like this then monism must be true, we must postulate
some entities of the austere type. Socrates defends manifest plurality from the austere
One of Elea only at the price of offering some austerity of his own—the forms. So long
as the austere individuals are basic, generous individuals may be admitted too. His
thesis, then, in response to the Eleatics, is about how many things there are (there are
more than one), about their nature (some things, particulars, tolerate being many;
other things, forms, are simply one, just like the One of Elea), and about how they
are related or structured (austere individuals are basic, generous ones parasitic).
10
I call this Plato’s thesis of natural inherence: he supposes that relations are real, property-like features
of the relata (e.g. this frog has the property large-as-the-goose-next-door), just as values are real features of
what is valued (not dependent on the existence or view of some valuer). See e.g. Euthyphro 10e–11a, Lysis
217a–218c, Phaedo 102c–103c.
11
There are various ways of interpreting the compresence of opposites exemplified here: perhaps the
point is rather that all phenomena are irretrievably perspectival, or maybe relative in some other way. See,
for a different interpretation, White (1992).
12
This claim is particularly clear in the Phaedo, 78b–79e, where the forms must be incomposite to be
invulnerable to destruction.
Throughout, the debate turns on things that are somehow or other (or fail somehow
or other to be) one. This dispute sets the agenda for the rest of the dialogue.
Now Parmenides takes over and launches, notoriously, an attack on Socrates’
theory of forms. Just as the dispute between Socrates and Zeno turned on three
separate issues in ontology (how many individuals are there? what is their nature?
how are they related?), so Parmenides’ challenge to Socrates follows the agenda laid
down by his pupil. Parmenides attacks the theory of forms on three fronts. (1) In
terms of the number of entities it proposes, he questions the scope of the theory
(130a–d).13 (2) Then he attacks the nature of the forms, given the explanatory
purpose they are supposed to serve (130e–133b). (3) Finally he points to their
remoteness from particulars and thus attacks their status as the individuals basic to
the structure of Socrates’ ontology (133b–134e).
(1) Consider, first of all, the number of entities there are. When Socrates first
outlines his theory, he offers his serious commitment to forms of values (beautiful,
just, etc.) and forms of relations (like, one, many).14 He is puzzled, however, whether
there are forms of man or fire; and repudiates forms for ridiculous things like mud or
hair. On reflection, he gives the following account of his restrictions on the theory.
The ridiculous things, he says, ‘are just as we see them; it would be nothing short of
absurd to suppose that the form for these is something. And yet I have often been
troubled whether I should not say the same thing about everything. And then when
I come to a standstill there, I retreat, fearing lest I fall into a pit of nonsense and am
destroyed’ (130d).
This argument is conventionally understood as a plea for teleology; Socrates denies
that the ridiculous things have forms just because their forms would be ridiculous
too, and ill suited to the teleological structure of the realm of forms (suggested, for
example, at Republic 505 ff.). But that is not what Socrates says; and it does not
explain why he should hesitate about fire (which is, after all, thoroughly respectable
in the teleology of Timaeus). Socrates’ argument is, instead, based on two principles:
one a constraint on ontology, the other a consideration about theory. The constraint
is one of parsimony. If the ridiculous things ‘are just as we see them’, then there is no
need for forms for them (since forms are called in to explain vexed cases such as like
and unlike, so exploited by Zeno—cases where there is a compresence of opposites).
The consideration, on the other hand, is one of universalizability. If Socrates puts
forward a theory to explain things, then the theory ought to apply universally (he
should ‘say the same thing about everything’), or not at all. If the theory is restricted
13
Moravcsik (1992), 55 ff. says that the Greeks were not interested in the question of counting, only in
determining the nature of the items in their ontologies; for they were not, Moravcsik claims, worried about
reductionism. I think that the Parmenides provides the evidence against this. Indeed if the Eleatics were
strong monists, one explanation for their taking that view would be a concern for reductionism.
14
That one and many are to be treated as relations is clear from the opening discussion of Socrates as
one man among many, as a man of many parts, etc., 129a.
(for example, to values and relations), Socrates needs some account of just why the
realm of forms is so limited. Socrates’ ‘pit of nonsense’ comes from the tension
between these two demands, which are, or should be, felt by any ontology. On the one
hand, my commitment to extra entities should be as mean as possible; on the other
hand, my theory should be as general as possible. The second demand enlarges my
ontology, the first shrinks it—the challenge is to come out with an ontology of just
the right size.
(2) The problems about counting forms are compounded by the puzzles that
Parmenides raises about their nature as explanations of the particulars. Within this
section there are three different phases of argument:
a. The puzzles about participation (131a–e) show that if particulars participate in
forms, the relation of participation can only be understood in such a way as to
render the forms themselves plural—either wholes with parts, or ones among
many. If Plato supposes that relations are real, property-like features of the
relata, then being one among many will render a form a plurality, and thus
vulnerable to Zeno’s complaints.
b. The Third Man arguments (132a–b, 132d–133a) show that any form is one
among many forms that must be invoked to explain some group of particulars.
These arguments might turn on the question whether any form has the
character it imparts; but, even if this is denied, the second Third Man argument
can be used to show that any form, which should be austerely single, is plural
just if it stands in any relation—for instance, of likeness—to its particulars.15
Structured ontology of the sort Socrates proposes is in trouble.
c. (132b–d.) Even if forms are thoughts, they will be composite; and, if forms are
thoughts, they will either be prior to what they explain (and so real but
vulnerable to the other objections) or posterior to what they explain (because
they are thoughts of something), in which case they cannot be explanations. On
this view, forms may fail either to be austere individuals or to be basic.
(3) In his final argument, Parmenides suggests that the forms that Socrates offers
are so remote from us that they cannot be known by us (indeed, he suggests that the
nature of forms is such that becoming known would itself interfere with their
nature).16 But, if this is so, and if their nature is as fraught with difficulties as the
earlier arguments have suggested, we may wonder whether they exist at all (135a).
This suggestion (that the unknowability of the forms might imply their non-exist-
ence) rests first on general assumptions about theory. If forms are theoretical entities,
postulated to deal with the problems of knowledge, and if they fail to deal with those
problems, we might reasonably attack their existence—once again, on the grounds of
parsimony. For we need only the theoretical entities that do any work; any others can
15 16
Cf. the analysis of this argument by Schofield (1996). See my (1986).
be done away with.17 But Socrates is in more trouble than this. For the argument
about the knowability of forms is preceded by an argument about the relations
between forms and particulars, which suggests that while form may be related to
form (e.g. mastery to slavery) and particular to particular (master to slave), there is
no cross-over from form to particular (the slave is not the slave of mastery). The
forms, that is, are totally separate from the particulars. This affects their knowability,
certainly; but it also affects the status of forms as the basic individuals in Socrates’
ontology. We might think about it like this. Socrates suggested that the world is saved
from Zenonian chaos by having in it basic entities that are austere individuals. Now it
transpires that there are after all two separate worlds—the world of particulars, which
does not have forms in it, and the world of forms, which does not have particulars in
it. In the world of particulars, therefore, there are no basic, austere individuals—those
are all in the world of forms. If they are supposed to save the world of particulars,
they cannot do so by being totally separate. Yet the demand that they be austere
seems to suggest that they must be separate. Socrates is stuck.
Parmenides closes his attack on Socrates’ theory of forms with a dilemma:
‘But these difficulties, and many others as well, must confront the forms, if there are these
very ideas of the things that are and if someone defines each form as itself something. So the
hearer is puzzled and wonders whether either there are no such things, or whether if indeed
there are, they are bound to be unknowable by human nature. And the person who puts
forward these difficulties seems to have something to say and, as we said just now, to be
amazingly hard to persuade otherwise. It is only a man of outstanding nature who will be able
to understand how any type of thing is, and how being is itself by itself; and it will be the task of
someone even more amazing to discover it and to be able to teach all these things to someone
else who is sufficiently well versed in these problems.’
‘I admit it, Parmenides,’ said Socrates. ‘I entirely agree with what you say.’
‘But then again, Socrates,’ said Parmenides, ‘if someone does not allow that there are forms
of things that are, because he has an eye to all the difficulties we have just raised and others of
this sort, nor defines the form of each one as something, he will have nowhere to turn his
thought, since he does not allow that the same idea for each of the things that are is always; and
in this way he will utterly destroy the power of dialectic.’ (134e9–135c2)
Parmenides offers Socrates a choice. If he postulates forms, he confronts all the

difficulties that have been outlined in the early arguments. If, to avoid those difficul-
ties, he refuses to admit that there are forms, he loses the power of dialectic. But it is
Parmenides who urges that without forms we lose the power of dialectic; and
Parmenides is the Eleatic who does not believe in forms and is the master of Zeno
17
The same argument does not apply, of course, to entities that we already suppose to be real and
independent of our theories; there are just as many long-toed sloths as there are long-toed sloths in the
world, irrespective of our theorizing. Socrates postulates his forms to be real; if the postulate is accepted,
then there are just as many forms as there are—but parsimony should still bear on whether we accept the
postulate.
the arch-dialectician. It is Socrates, the exponent of the theory of forms, who

concedes that the theory makes no contribution to the explanation of particulars.
Only a man of ‘outstanding nature’ could escape this dilemma.
The entire debate so far is presented as the clash of two theories (Eleatic monism
and the complex theory of Socrates) in the debate between three vividly represented
characters. The dilemma, however, marks a change, both in the tone of the discussion
and in its assumptions. The dialogue will shortly abandon the richly personal
characterizations of the first part; and the argument will abandon the assumptions
that those characters made. Parmenides begins his dilemma with a disjunctive
premiss—either there are forms or there are not. If there are forms, they fail, it
seems, to help either our understanding of particulars or our understanding of the
realm of the forms itself. Zeno’s arguments, then, may prevail; our talk about
particulars will become contradictory, our talk about forms vacuous. If, on the
other hand, there are no forms, there is no dialectic. Each disjunct, then, implies
that there is no dialectic—or at least there is no dialectic outside the limits of strong
monism. But that conclusion is intolerable. Why?
Not, surely, because life is intolerable without forms (Parmenides, for one, could
not have conceded that), but rather because life is intolerable without dialectic. The
protagonists have that much in common; and they have a concern for dialectic in
common also with the man of outstanding nature—and with the reader. We must,
then, find our way out of the dilemma—but to rescue dialectic, not forms. If that
should be done by dialectical means, it may provide, if not a deductive, then a
pragmatic defence of dialectic. That is what is promised in the sequel.
3. Gymnastic and Philosophy

Parmenides now promises Socrates a gymnastic session, to train him to grasp the
truth. This training will echo procedures of Zeno,
‘ . . . with this exception. I enjoyed hearing you say to him that you were not prepared to think
about visible things, nor to confine your wanderings to them, but rather to those things which
someone might grasp in argument and where he might suppose there to be Forms. That is
because I think it easy to show that in the sensible world the same things are like and unlike or
anything else at all.’
‘Quite right,’ he said. ‘But it is necessary to do this as well, for each thing hypothesized to be,
not just to consider what follows from the hypothesis, but also if the same thing is hypothesized
not to be, if you want to be better trained.’ (135c–136a)
Up to this point in the dialogue, theories and dramatis personae have been inextric-
ably connected; every theory under discussion has belonged to someone.18 It is often
18
Indeed, the question about who owns what theory has taken up some space—see e.g. the discussion of
the extent to which Zeno believes in monism, 128b–e.
assumed that the subject of the second part, ‘the one’, belongs to someone too.
Perhaps Parmenides is discussing his own One. Or perhaps (more plausibly) they are
considering one of Socrates’ forms, the form of unity. After all, discussion is
promised, not of visible objects, but of those things we ‘might grasp in argument
and where someone might suppose there to be forms’ (135e). Thus, at various
stages of the argument Parmenides’ terminology suggests the theory of forms (e.g.
‘the one itself ’, ‘unity itself ’, tou henos autou, 137b2). More significantly perhaps, the
argument treats its subject both as an object (‘unity itself ’) and as what imparts unity
(the property ‘unity’); such a dual role, it is often thought, could only be performed
by a form. Is this a plausible account of the gymnastic session?
Parmenides gives a long and exhausting treatment of two hypotheses: ‘if one is’
and ‘if one is not’. Each hypothesis is treated from two different perspectives: ‘if one
is, what is true of it?’, ‘if one is, what is true of the others?’, ‘if one is not, what is
true of it?’, and ‘if one is not, what is true of the others?’, generating four stages in the
argument.19 Each stage is then treated in two movements (or three, in the case of
the first), to reach either a negative conclusion, that nothing can be said of what is
hypothesized, or a positive one, that anything can be said of what is hypothesized.
Both negative and positive conclusions are absurd. The negative ones conclude (as
Parmenides’ dilemma had) that what has been (exhaustively) mentioned is unmen-
tionable. The positive ones conclude (as Zeno’s paradoxes had) that what has
been hypothesized is contradictory. The four stages and their movements may be
schematized:
First hypothesis, ‘if one is’:
Stage I. ‘If one is, what is true of it?’
movement (a) 137c–142a (negative)
movement (b) 142b–155e (positive)
movement (c) coda 155e–157b.
Stage II. ‘If one is, what is true of the others?’
movement (a) 157b–159a (positive)
movement (b) 159b–160b (negative).
Second hypothesis, ‘if one is not’:
Stage III. ‘If one is not, what is true of it?’
movement (a) 160b–163b (positive)
movement (b) 163b–164b (negative).
Stage IV. ‘If one is not, what is true of the others?’
movement (a) 164b–165d (positive)
movement (b) 165e–166c (negative).
19
The antinomies have been carefully discussed in recent years. See esp. Cornford (1939); Ryle (1965),
97–147; Owen (1986b); Schofield (1977); Meinwald (1991).
The entire exercise culminates in the terminal aporia, re-emphasized by the fact that
the asymmetrical ordering within the stages allows the entire argument both to start
and end with a negative movement.
Now if all this is about the Eleatic One, the first stage might go to show that,
whether it is understood in a minimalist, severe way (there is only one One, so that it
is not even mentionable) or an excessively liberal way (if there is the One, then
anything can be said of it), it is not a reasonable postulate. But then the second stage
confuses matters by introducing something that no Eleatic would accept (‘the others’,
whatever they are); so the Eleatic One cannot be under attack here. Likewise, the
third and fourth stages could sometimes be taken to offer support for the austere
Eleatic One, sometimes to undermine it, and sometimes to be quite tangential to
Eleatic concerns.
A similar tale might be told of a Platonizing interpretation of the One. If the form
of unity is under attack here, the first two stages may repeat the dilemmas of the first
part, by showing that it is either so austere as to be unmentionable, or so generous as
to be contradictory; and then by showing that, if this is so, anything else in our
ontology will be similarly infected. But then the first and fourth stages, which deny
that the one is, may be said to support the form of unity by reductio ad absurdum of
its denial. The result is another impasse for the theory of forms, and another disaster
for dialectic; the only significant difference between the first part and the second is
that the second is duller by far than the first.
Prima facie, then, we might wonder why Parmenides should persist in maintaining
a discredited thesis before the training has even begun. Moreover, when Parmenides
compliments Socrates on resisting Zeno’s paradoxes—‘you were not prepared to
think about visible things, nor to confine your wanderings to them, but rather to
those things which someone might grasp in argument and where he might suppose
there to be forms’ (135e)—he does not imply that the forms are the topic of the
second part. Rather, he stipulates an abstract discussion, such as might lead to
the postulation of the forms. The first part of the dialogue has an explicit ontological
commitment to the contrast between forms and particulars. Now, however, in a way
that is appropriate to an abstract discussion, Parmenides explicitly eschews any
interest in the particulars. While someone could still take the forms to be the
appropriate focus of rational discussion, father Parmenides himself did not suppose
that to be a necessary condition of rationality or of abstraction. We are to have, then,
a different type of investigation from physics, which concerns itself with real sensible
objects, and from the discussion of the forms, where the ontological content is
already determined. The argument that follows is an abstract thought experiment,
free of ontological commitment (hence, for example, the repeated suggestion that
we should take the one ‘itself by itself in thought’, for instance, at 143b7; compare
158c2, 165a).
This, of course, does not imply that the gymnastic session offers no serious
philosophical contest at all, or that it is merely the fitness exercises that will limber
Socrates up for the grown-up job of rescuing the theory of forms. Socrates will be
completely trained when he has gone through an extraordinary task, a race in which
Parmenides will be the aged racehorse trembling before the fray (137a). But training
is for the sake of the race; and so the gymnastic, which is also a race, may be valuable
in its own right. Doing dialectic may be itself worthwhile; the practice of philosophy,
no less than its theories or its conclusions, may be the objective.20 In that case,
perhaps the significant difference between the first part of the Parmenides and the
second is not that one takes dialectic seriously and the other does not; but that one is
(variously) ontologically committed, the other commitment-free.
The dispute between Socrates and the Eleatics was about ontology. How many
entities are there? What is the nature of the entities there are? How are they related?
But these questions were framed in a particular way. For it was assumed, by Eleatics
and Socrates alike, that any entity can be counted (each entity counts as one) and that
any entity thus has some kind of unity (each entity is somehow one). The dispute was
about individuals. The Eleatics supposed there to be only one—the austere One.
Socrates allowed many individuals, of two types—austere individuals, forms, and
generous individuals, particulars. But the argument, as it has proceeded, has cast
doubt on the specification of either sort of individual. The austere forms are prob-
lematic both by virtue of their nature (can there be austere individuals like this?) and
by virtue of their very existence (if forms are like this, do we need them in our
ontology? if forms are like this, can they be the basis of a complex ontology?). But
then Socrates is back with Zeno’s problem, namely how to explain, without recourse
to the forms, the generous individuals of the phenomenal world (if there are no
forms, can we avoid monism? if there are no forms, are not the many hopelessly
many?). Now what is needed is an account of ‘what is an individual?’ as a general
question, leaving open the issue of which entities qualify. That general investigation,
I argue, is on hand in the second part of the Parmenides; it continues in the
Theaetetus and the Timaeus and is brought to a satisfactory conclusion in the Sophist.
The agenda from here onwards, that is, is not the viability of the theory of forms, but
the nature of individuation. ‘What is one?’ is the question that demands an answer
first.
This general enquiry, moreover, is exactly what we might expect from a dialogue
composed in the way this one is. Because the end of the dialogue is open, the Chinese
Whispers effect, I suggested, so far from distancing the reader from the issues
contained therein, has the opposite effect, of making you, as you read, just one
among the many recipients of the arguments. In that case, that at least some part
of the dialogue should be free from the ontological commitments of Platonism is
20
Compare the way that Socratic dialogues tempt the interlocutors to suppose that philosophy has a
product, and then show them that they were doing what matters all along, e.g. Charmides 167a ff., where it
is suggested that sôphrosunê may be the process of enquiry and investigation, with the Politicus’ suggestion
that life in the golden age would be less fun without dialectic and philosophy, 272b ff.
entirely appropriate. Likewise, the juxtaposition of the first, Socratic,21 part of the
dialogue with the characterless conversation of the second suggests a radical contrast
between the two. The first part of the dialogue is theoretically committed: it repre-
sents two specific ontologies, held by specific persons, under attack. In the second
part, the persons have effectively disappeared, leaving behind just the arguments. The
arguments, then, are of general application, to any ontology that postulates ‘the one’.
4. If One is, What is True of It?

The first hypothesis is: ‘if one is’. In the first stage, this is treated in two movements
with a coda.22 The first movement ends in the conclusion that the one cannot even be
mentioned, let alone known or understood; the second, from showing that of a series
of pairs of predicates both of each pair apply to the one (indeed, that any predicate at
all may be applied to it in any respect, 155c), infers that we can mention, perceive,
believe, and know the one. Both conclusions are absurd. The first violates the law of
the excluded middle; and it seems to be self-refuting because it rules out speech
altogether (the conclusion claims that we cannot even mention the hypothesis we
have been discussing). The second, on the contrary, admits everything: we can say
whatever we like about the one, without restriction (in particular, without let or
hindrance from the law of non-contradiction).
The first movement begins by denying parts to the one:
If one is, then the one cannot at all be many. So it must have no part, nor be a whole. For a part
is a part of a whole, and a whole is that from which no part is missing. Either way, the one
would be made of parts, whether it were a whole or had parts. Either way, then, the one would
be many and not one. But it must not be many but itself one. So it is not a whole and has no
parts.23 (137c–d)
This argument operates on a restricted reading of ‘one’ in the hypothesis ‘if one is’.
All the deductions that follow (so it has no beginning, middle, or end; it is indefinite,
formless, in no place, and neither moving nor at rest) suppose that for something to
be one rules out any pluralization at all; and that is taken—as the arguments show
exhaustively—to outlaw ascribing any predicates, or attributing any relations or any
properties, to this one. If it is just one, nothing is true of it, and nothing can be said of
it at all.
Conversely, in the second movement, anything goes.
If one is, it cannot be and not participate in being. So the being of the one is not the same as the
one, or else this would not be the being of that, nor would that, the one, share in this; instead, it
21
Meinwald (1992) is good on this.
22
For reasons of space I have omitted detailed discussion of the coda here, which contains arguments
that are both interesting and important in their own right.
23
My translations generally omit Aristotle’s responses.
would be the same thing to say ‘one is’ as to say ‘one one’. But the hypothesis is not ‘if one one’
nor its consequences, but ‘if one is’. So ‘is’ signifies something different from ‘one’. So, when
someone says ‘if one is’, they mean just that the one partakes in being. But then the one is the
sort of thing that has parts. For if the ‘is’ is said of the one that is and the ‘one’ is said of what is
one, and the being and the one are not the same, but belong to the thing which we have
hypothesized, namely the one being, then it, being one, must be the whole, and being and one
must be its parts. But each of those parts is not merely a part, but a part of the whole. So
whatever is one is whole and has parts. Now each of the parts of what is one, the one and the
being, will not be lacking: the one will have being [as a part], and being will have one [as a part].
And each of these parts in turn has being and one, and the least part comes to be out of two
parts, and so on forever—whatever is a part itself always has these parts, so that one always has
being and being always has one. So, necessarily, it keeps becoming two and is no longer one.
Thus, what is one is indefinite in number. (142b–143a)
Here, Parmenides’ strategy is quite different. He appeals to the semantic difference

between ‘one’ and ‘is’ to generate a one with parts; he assumes, that is, that any
ascription of a predicate (including, perhaps, ‘existence’?) to the one signifies some
real future of the one. In that case, to say that the one is pluralizes the one; likewise, to
say that one part of the one is pluralizes its parts. Upon that assumption is built the
rest of the second movement of the first hypothesis, and the contradictions that
result.
On reflection, these contrasting assumptions are familiar. The first, that the one is
just one, so that nothing at all can be said of it, is the demand that ‘one’ be treated
austerely. The second, that ‘if one is’ legitimizes any predication to ‘one’, in particular
the predication of opposites, is generous in its ascription of properties. But the
contrast between the generous and the austere so far has reflected the contrast
between particulars (which are generously endowed with properties of all sorts)
and forms (which austerely are just what they are). The difference, that is, between
generous individuals and austere ones was a difference in ontological level. But here,
that difference has disappeared, and the (same) one is treated first austerely, then
generously. Why? This question should, perhaps, be approached through another.
What is this one we are dealing with here?
The Greek expression ‘the one’ (to hen) suffers from multiple ambiguity. Consider
some English parallels. If I say, ‘The grey won the race’, I use ‘the grey’ as a singular
referring term for an individual particular. If I say, ‘The pink of the painting clashes
with the red of the stained-glass window’, ‘the pink’ and ‘the red’ describe individual
properties. ‘Beige is a dismal colour’, on the other hand, speaks quite generally of a
colour, wherever it occurs, while ‘Black is beautiful’ uses ‘black’ to classify individual
particulars. So ‘the one’ may be a particular one or a group of ones; or it may be the
unity of some particular or the unity of a collection of particulars. There is a risk, as a
consequence, that when Plato starts to talk about ‘if one is’, and if he is talking about
individuals, he may confuse ‘an individual’ with ‘what it is to be an individual’.
Consider a slightly different ambiguity, reflecting a difference in argumentative
strategy. ‘The one’, if it is a class term, could describe a class whose members are
already determinate (‘these ones’, like ‘these elephants’), or one of the members of
that class (‘this one’, ‘some one’); or it could describe a class whose membership is
still undecided (here the expression ‘the one’ contains a variable, ‘some one’, ‘any
one’). In the second case, this may be because we do not know what the values of the
variable are—we do not know which items in the world actually fit the bill; this is an
empirical matter. Or it may be because we do not know the conditions for the
predicate to apply. This, I take it, may be a much deeper, theoretical matter. And if
the conditions for being one are still unclear, or disputed, then we must read ‘the one’
as indeterminate, ‘any one’; for without knowing the conditions for membership, the
empirical question cannot be settled anyway.
Thus, ‘if one is’ cashes out as ‘suppose any individual . . .’. That is, the ‘one’ which
is the subject of all the arguments may be understood to be any individual, abstractly
conceived. The arguments that follow in the first hypothesis consider what will follow
from supposing there to be some individual; the arguments of the second hypothesis
consider what will follow from denying that there is some individual. In each case,
the individual is viewed in two different ways: on the generous view, the individual is
defined in terms of its properties; on the austere view, the individual is seen to be
what it is in isolation from any properties it might have. This contrast is the basis of
the alternating movements within each hypothesis. The arguments of the first
hypothesis then amount to an examination of what it is to be an individual; the
arguments of the second reflect on the first by denying it.
5. Parts and Properties

The contrast between the generous and the austere conceptions of the individual
could be understood as a puzzle about priority.
Austerely, the individual per se is prior to the specification of its properties and can
be specified as such. Take a trivial example. This blackbird before me has a grubby
beak. Having a grubby beak happened to the blackbird, which was already digging
holes in the lawn. The blackbird came first. And once we see the same blackbird
among a flock, we would readily assume that this blackbird was an individual before
it became one among the twenty-four for the pie. The individual (as Aristotle would
agree) comes before its accidents—before its various and changing properties, affec-
tions, and relations. (‘Comes before’ of course means, to Aristotle at least, that it
antedates any particular accident, not all of them; nonetheless, generally understood,
the individual blackbird comes first.) Now suppose that this principle is generalized,
in the austere manner. If what is one (an individual) comes first, and is basic, then we
should be able to understand it as one on its own, and not in terms of its properties,
affections, or relations. To reach what is basic (by this reasoning, which both Socrates
and Zeno endorsed) we have to grasp what is just one—an austere individual.
Generously, all there is to being an individual is the having of properties, so that

the having of properties comes first. After all, the blackbird cannot even be thought of
without some properties, even if it seems to antedate this particular change (from
digging up the lawn to appearing in a pie). Perhaps, then, the individual blackbird is
some collection of properties (whether essential or accidental is a further, and later,
question). In that case, there should be no question of some ‘just one’, since any one
is a collection of features. This generous view supposes that understanding one is
understanding it in terms of the properties it has. And it may follow from this that its
unity is one among the properties it has.
These approaches to ‘being one’ are quite different; and they seem to be incom-
patible, since the austere view denies any appeal to the properties of what is one,
while the generous view supposes that there is nothing but properties. Moreover, the
austere view treats the individual as primary, while the generous view must explain
the individual by virtue of the property of being an individual that it has. But then
once ‘some individual’ is postulated, a further issue of priority arises: what is the
relation between ‘some individual’, and ‘what it is to be an individual’? Thus, suppose
we have some individual: is it just an individual, from which we can grasp what it is to
be an individual (so that the individual has priority), or is it an individual because it is
characterized by ‘what it is to be an individual’? In that case, how are we to
understand that character? Is it a property, like any other? In that case the property
is prior, and we are committed to some generous view of individuation. If not, then
we may retain the austere view, and the individual and ‘what it is to be an individual’
are one and the same. However, as the second part of the Parmenides shows
graphically, neither escapes paradox. Dialectic, in the end, seems to be in worse
trouble than before.
The hypothesis is about ‘one’; and in both movements the attack is focused on
individual entities—any individual entities. The first movement hypothesizes what is
‘just one’ and considers how its nature as ‘just one’ precludes its partition. So, if it is,
it cannot be one (because that will be an additional property), nor can it be whole; so
it is indefinite. Such an item cannot even be said to be one at all; no individual,
defined in this austere way, can be a definite individual, and so an individual is no
individual at all. Contrariwise, in the second movement, the one that is allowed to be
turns out not to be one at all, but any number, and indeed anything you like and its
opposite as well. Such an item is no more definite (no easier to mention, describe, or
know) than the austere one of the first movement.
What exactly is the assumption on which all this depends? One account of
what has gone wrong here is that Plato has muddled identity and predication.
When we say:
i) this paper is the very one I threw away last week,
we are using ‘is’ in a different sense from when we say:
ii) this paper is half-witted.
In the first case, the ‘is’ gives us the equation of identity, so that the items
mentioned are indiscernible: they are the same item, just one entity after all.24
Thus, if (i) is true, then if this paper is about the Parmenides, so will the paper that
I threw away last week be about the Parmenides. In (ii) we have the ‘is’ of predication,
which attributes a property to an object in such a way that Leibniz’s Law does not
hold: even if this paper is half-witted, it does not exhaust all the half-wittedness there
is, or prevent several other papers from being so too; nor, if they are half-witted too,
will this imply that they are the same paper as this one. I cannot say ‘half-witted’
when I mean ‘this paper’, or vice versa, and come up with the same truths.
This contrast—between identity and predication—is a tricky one to see, let alone
to manipulate. The problem may lurk in ‘is’ (which, it could be argued, may be
ambiguous between the ‘is’ of identity and the ‘is’ of predication);25 and it is
exacerbated by the oddity of expressions such as ‘the grey’ or ‘the one’, which may
signify both an object (on either side of an identity equation) or a property (on one
side of a predication). Has Plato fallen into the trap of confusing the two—or failing
to employ them differently in appropriate circumstances? (Or, more tendentiously,
are we supposed to spot this as the deliberate mistake?)
Suppose there is just one item to consider, then maybe its being one constitutes
exactly what it is; so to say ‘what is is one’ or ‘the one is one’ is not to attribute a
property to what is, but to state its identity. If being one is what it is for this one to be,
what is is identical with what is one, and not more than one. To think about this from
a grammatical point of view, the second ‘is’ of ‘what is is one’ is an ‘is’ of identity, not
predication.
Plato’s mistake in the first stage might be that he treats identity statements as
predications and imagines that the predicate attributes a property to the subject over
and above its identity. The first movement goes wrong because it supposes that the
one must have no additional properties; hence, it is denied even identity claims
(because they seem to import extra entities). The second movement goes wrong in
supposing that the one’s being one attributes a property to the one, and so turns it
into many.
Or perhaps Plato thought all ‘is’ statements were identity statements. The first
movement, on this view, rightly denies what is one any identity but austere unity, but
then ends up not being able to talk about the one, since talk needs predication. The
second movement, conversely, claims that when we say many things about the one,
we are making many identity statements. But then—Plato infers, mistakenly, since
identity statements do not relate two items that are numerically distinct—there are
many things, not just the one we first thought of.
24
The expressions that refer to them, that is, are intersubstitutable—except, of course, in opaque
contexts, e.g. modal contexts or the various devices of direct and indirect speech.
25
But see e.g. Mates (1979) who denies any such ambiguity.
So perhaps Plato did not understand ‘is’. He had not grasped that the identity
relation is reflexive, and that it associates one item with itself, not with something
else. Nor had he understood that predication is not subject to the stringencies
of Leibniz’s Law, but that it relates objects and properties, relations, affections
in multifarious ways without impairing the original object’s identity at all. He
may have been, quite generally, in a muddle about ‘is’ and used both senses
indiscriminately.
But the difficulty may lie deeper than this. Identity statements themselves require
some grip on the notions of singularity and individuation. Consider the following
sentences:
i. Polly is ultramarine.
ii. Polly is a parrot.
iii. Postumus is the world’s noisiest parrot.
The first sentence gives us a predication. The second sentence may sort Polly into the
correct natural kind, and it may allow us to infer:
iv. Some parrot is ultramarine.
But it will not allow us to conclude:
v. Postumus is ultramarine,
or
vi. Postumus is Polly,
or
vii. Polly is the world’s noisiest parrot.
While sentences connecting singular descriptions (such as iii) may be subject to
Leibniz’s Law, sentences expressing essence or kind (such as ii), while more
fruitful than ordinary predications (such as i) in inferential content, are none-
theless not full-blown identity statements. Even the conjunction of i and v will
not justify vi, ‘Postumus is Polly’. However, to understand the structure of these
inferences, we must already understand quite a lot about the identity of indi-
viduals. If what constitutes the identity of individuals is under critical scrutiny,
then the contrast between identity statements and predications may itself be
tendentious.
Perhaps Plato’s trouble was not ‘is’ but what came after it. Plato may be making
some assumption about the metaphysics of properties and their possessors, and not
about the grammar of the verb ‘to be’. In earlier dialogues, Plato supposed that any
properties of particulars, including relations and values, are features of the particular
in such a way as to make them composite (plural) entities. When, earlier in this
dialogue, Socrates faces difficulties with the theory of forms, his demand that they
resist interpredication—any predication at all—seems to be based on the assumption
that they are simple entities with no plural properties at all. So there is a choice
between something with properties (whose properties are parts of it) and something
without properties (which may then be a simple entity). Both alternatives suppose
that properties are parts in such a way that the thing that has them is pluralized by
them. ‘Plural characters pluralize a thing.’26 What has properties may be (nothing
over and above) the sum of its properties; contrariwise, for something to be simple
and one, it must have no properties at all. Once again, lacking some independent
account of what makes something one, we cannot deny that it is the sum of its
properties, nor that it is devoid of properties altogether.
The pluralizing assumption is threatening in two quite different ways, as the first
and second movements of the first stage show. Consider the parallel treatments of
sameness and difference.
The negative movement concludes at 139e that the one is not the same as itself, nor
as another, nor different from itself, nor from another. The paradox, evidently
enough, lies in the denial of self-identity and of difference from others; and the
force of that paradox is felt particularly for an individual subject. If the nature of the
one is one (see, for instance, ‘otherwise it would not be that which is one’, 139c1; ‘by
virtue of being one it will not be different’, 139c6), its nature is different from the
nature of either same or different; so that the nature of one can neither explain nor
cause the sameness of the one to itself nor its difference from others. So, just one, it is
not the same as itself nor different from the others.
Here Parmenides considers the nature of the one, and derives uncomfortable
consequences for the one from its having that nature. But it could be argued, once
again, that the argument contains a gross mistake in its failure to distinguish between
individual natures and universal natures (whether properties or relations, formal or
material). Innocuously and trivially, some individual might be identical to its (indi-
vidual) nature. Mozart just is what it is to be Mozart—and his nature includes
musical brilliance and an unpredictable temper, as well as self-identity and difference
from everyone who is not Mozart. But this identity relation does not imply that
anyone who is musically brilliant must be Mozart, any more than it implies that
anyone who is self-identical must be Mozart. Nor does it suggest that Mozart is a
property that may be distributed among many participants, or that self-identity must
be limited to a single individual, namely Mozart. Individual natures and universal
natures are fish from quite different kettles.
Once again, however, recall that the argument begins from the bare hypothesis
‘suppose any individual’, and consider what follows from that hypothesis alone. The
first movement, I have suggested, works on austere principles. So, ‘if the one is
different from itself, it will be different from the one and no longer one’. If the one is
considered as such, just what is one, then ‘itself ’ and ‘one’ are intersubstitutable; the
inference to ‘it will be different from the one’ is sound. But if it is just one, then its
being different from itself could result in nothing other than its no longer being one.
26
Owen’s slogan (1986b), 90.
This argument, which produces the truism that the one (or indeed anything else)
cannot be different from itself, is the model for what follows.
‘If the one is the same as the different, it would be the different, and not itself ’
(139b7). If the one is identical to something other than itself, it will be other than
itself (and that is absurd). This will be true whether ‘the different’ means merely some
different item (which will be different as such) or ‘what it is to be different’, which, if
we are considering the one as such, will not be identical to the one. On the
assumption that we are considering the one as such, the argument is sound, indif-
ferent to the slide between something and its nature. The conclusion—‘So it cannot
be the same as the different nor different from itself ’ (139c)—is unexceptionable.
‘And so long as it is one, it cannot be different from the different; for it is not
appropriate for one to be different from anything, but only appropriate for the
different, and nothing else, to be different. So by virtue of being one, it will not be
different. But if it is different by virtue of being one, then it is not different by virtue of
being itself, and if not by virtue of itself, then it will not be different. While it is
different in no way, then it is different from nothing’ (139c–d). Here, the same
assumption seems to be in play. The one as such is not different from anything (not,
that is, qua one, only qua different). But this one we are talking about is just one, and
not (by virtue of being just one, or by virtue of itself) different from anything. Why
not? Because (as before) Parmenides treats ‘being different’ as an additional property
of this one; and because, on the austere hypothesis, the one has no such properties.
The problem here, then, is not that the one is assumed to be ‘what it is to be one’, or
that Plato conflates a particular with a universal; rather, he produces the paradox by
relying on the assumption that relations (especially the relations governing identity
such as sameness and difference) are real features of the relata, and thus pluralize. But
the one, as such, cannot be pluralized; and so it is precluded from all relations of
identity.
The one as such, then, is not different (from anything), nor is it what it is to be
different, nor does it have the nature of what it is to be different. That follows from
the austere hypothesis. But then the one, by parity of reasoning, cannot be the same
(as anything), nor is it what it is to be the same, nor does it have the nature of what it
is to be the same. ‘Then again, nor will it be the same as itself. For the nature of being
one is not the nature of being the same. After all, if something becomes the same as
something, it does not thereby become one; when it becomes the same as many, it
must become many, not one. But if the one and the same were no different, whenever
something became same, it would always become one, and whenever one, then same.
So if the one were the same as itself, it will not be one with itself; and thus while it is
one it will not be one’ (139d–e). This argument works, again, without falling foul of
the particular universal fallacy. Suppose something is just one; and suppose further
that for anything else (other than ‘one’) to be true of this one would pluralize it, and
make it not one at all. In that case, the only way in which the one could be the same as
anything would be if to be one were the same thing as to be the same. But that is not
so (there may be something that is the same as something that is many). So we cannot
substitute ‘same’ for ‘one’; nor can we attribute sameness to the one (without
pluralizing it). So the one is not the same as anything, either. ‘So it is impossible
for the one to be either different from the different or the same as itself. So the one is
neither different nor the same, whether in relation to itself or to anything else’ (139e).
Parmenides presses the paradox that even the relations that explain something’s
being one pluralize it. So no austere one is either the same or different (to anything,
from anything); and, contrariwise, anything that has the properties of sameness and
difference cannot be austerely one. This is not, of course, to muddle identity and
predication, but to adopt a special stance towards the consideration of identity. If, as
I have suggested, these manoeuvres take place to explain ‘the individual’ by consid-
ering individuals in the abstract, then the nature of the individual will be just what is
under consideration, and inseparable from the individual hypothesized. So, either
‘the one has the nature of the one’ is a legitimate identity statement, so that there is
no difference between the one and its nature; or there is such a difference, and ‘the
one has the nature of the one’ will turn out to attribute a property to the one, and be
illegitimate. From the argument about the denial of parts, any attribution of self-
identity or difference from others to the one will result in a violation of its austerity,
and must be disallowed.
The second, positive movement makes the opposite assumption from the first: that
the one is generously endowed with properties—indeed, that the one just is its
properties, multifarious though they turn out to be. This produces both truism and
paradox: the one ‘must, therefore, be the same as itself and different from itself, and
likewise the same as and different from the others’.
Consider the truism and the paradox that the one is the same as and different from
the others. The truism comes first: ‘But if something is different from something else,
it will be different from something different’ (146d). Applause—some relations are
symmetrical. So, ‘whatever is not one is different from the one and the one from
those which are not one’. Fair enough; the negative marks the difference. ‘So the one
is different from the others.’ Is that true, too? That depends on what these others are.
The others may be items that are not (are not identical to) this one we are discussing;
or they may be items that do not have the character of one (and are for that reason
non-identical with the one we are discussing). ‘Difference’, that is, may mean non-
identity or it may mean unlikeness (especially of quality). The argument might be
harmless (although the premises may be false): if x is F and y is not-F then x is not y
(where x and y are individuals, F a property). So, if the one has the character of one,
and the others have the character of not-one, one and the others will not be identical.
Or the argument could be dangerous: if x is not y, then any character that x has, y
does not have. So, if the others are other than (this) one, and the one is thus (for
example) different, the others cannot be different, because they cannot share
characters with the one. Or the argument could be formal: if x is not y, then y is not x.
So if we take the one and some non-identical others, the one and the others are non-
identical. The conclusion is taken as true (which suggests either the formal or the
harmless readings); is the dangerous reading implied in the paradox that follows?
‘Same and different are opposites, and the same cannot ever be in the different or
the different in the same’ (146d). Here sameness and difference are properties, which
may inhere in something else, but which exclude each other. ‘And the different
cannot be in anything for any length of time, or else it would be the same; so the
different is never in anything, whether it be one or not.’ It turns out that difference is
so extreme that it cannot even inhere in something else, at the risk of becoming its
opposite. Why not? Why should difference be different (instead of making its object
different)? ‘So it is not by virtue of the different that what is one is different from
what is not one’ (146e). If the one is to be different from something else, that must be
explained by some character it has. But the character is so slippery that it cannot do
that explaining. ‘However, the one and the others are not different by virtue of
themselves either, if they don’t share in difference.’ So they cannot be different at all.
Suppose, as the generous hypothesis does, that anything just is its properties; and,
further, that the collection of properties includes the relations of identity. On this
view, the identity of something is determined by (some of) its properties. How, on
that account, are we to understand what makes x not y? Only by virtue of its
character; in particular (why else?) by virtue of the character that determines its
identity. Conversely, the non-identity of something to something else can only be
explained by some character that it has; indeed (once again on the generous hypoth-
esis), the character that it has will be all there is to its difference from other things.
But that character just is what it is to be different; and so the character itself cannot
stay fixed, but slips away in constant flux. In that case, there is nothing stable about
the non-identity of the one and the others—indeed, it collapses altogether.
The conclusion may go through if we concede the generous hypothesis in all its
complexity. The claim (to recapitulate) of the generous hypothesis is that being ‘this’
or ‘it’ just is being a collection of properties; and that those properties account for the
identification of the ‘it’ in question. Indeed, the ‘it’ is an illusion, a mere way of
speaking about this collection;27 properties are all there is. Nothing ‘has’ a property;
instead, the property occurs in a collection. But, on that view, it is hardly surprising
that the property displays its own character; the appearance of the character is all
there is to the property. In that case, the identity of the collection of properties will
depend on the properties themselves—but not on difference (which keeps slipping
away).
Ryle might have complained that Plato fails to distinguish between different sorts
of property (or concept or predicate), namely between the entirely general features of
27
As Plato saw: Theaetetus 156–7; see Fine’s analysis (1996).
the world and the specific features that are limited to particular classifications. Plato,
that is, should contrast predicates (and their corresponding properties) such as
‘exists’, ‘same’, ‘single’ with predicates such as ‘is yellow’ or ‘flies’.28 In particular,
he is wrong to ignore the asymmetry between objects and properties; and wrong
again to suppose that any property (such as being one or being the same) is just like
any other (such as being purple or being heavy). But that complaint begs the
question. Unless we already know what constitutes a proper distinction between
object and property, by capturing the contrast between the properties that identify
and individuate (which should not pluralize and cannot be parcelled out) and those
that are accidents, extra features of some objects, we cannot deploy that distinction to
effect an account of individuation. Parmenides throws down the gauntlet: we must
come up with an account of identity and individuation that avoids these paradoxes or
else we must allow the One of Elea to win the day.
6. What If the One is Not?

Socrates, Parmenides, and Zeno all suppose that ontology counts individuals; they all
see individuals as basic in the structure of reality. But once Parmenides starts to
investigate just what it is to be an individual, he finds that neither of the two
specifications offered by Socrates can be made to make sense—neither austere nor
generous individuals give ontology a rational basis. Perhaps, then, we should dis-
pense with them altogether? The third and fourth movements ask this question by
supposing that ‘the one is not’. Once again, the argument produces aporia.
The third stage begins by asking what is meant by the hypothesis ‘if one is not’:
Well then, surely we should consider next what must follow if the one is not?—We must.—
What would the hypothesis itself be, if one is not?29 Does it differ at all from this, if not one is
not?—It differs, surely.—Does it just differ, or does it mean something quite opposite to say ‘if
not one is not’ and to say ‘if one is not’?—Quite opposite.—What if someone were to say ‘if
largeness is not’ or ‘if smallness is not’, or anything else like that, in each case it would be clear
that he speaks of what is not as something different?30—Certainly.—And so now too it would
be clear that he speaks of what is not as something different from the others, when he says one,
if it is not, and we know what he is saying.—We do.—First, then, he says it of something
knowable; second, of something different from the others, when he says one, whether he
28
See Ryle (1965), 130 ff., whose list of general properties is longer; in my (1994) I argue that Plato has a
quite specific, and short, list of properties in mind, namely the megista genê of the Sophist.
29
Why does Plato change his wording, from ‘if the one is not’ to the indefinite ‘if one is not’? Cornford
(1939), n. 219, emphasizes that there is a difference between the two expressions. I concur; and suppose
that this points to the entirely general nature of the whole discussion.
30
The construal of this sentence is tricky; I take all the verbs of speaking to have the personal subject
‘someone’, followed by indirect speech: literally, ‘he says that what is not [is] different’. The main verb of
the indirect speech is missing, however, and the point of the argument is better captured by the English
locution ‘he speaks of x as . . .’, since this emphasizes Parmenides’ point that the subject of the discourse
remains comprehensible no matter what predicates are attributed to it.
attributes being to it or not. For it is understood no less, what it is that is said not to be and that
it differs from others. (160b–d)
The first movement is positive (160b–163b). It anticipates the discussion of the

Sophist about not-being and argues that despite the negative ‘not’, this one (that is
not)31 is knowable and identifiable—in particular, identifiable as what is not the
others. That allows sameness and difference to be attributed to it, as well as all sorts of
other relations to all sorts of things. If it is thought to be different from the others, it is
different because of its difference from them, and not just their difference from it. So
difference must be a property of this one; and it is determinate in other ways (it is
‘that’, ‘something’, related to ‘these’, etc., 160e), if it is to be mentionable at all. So it
has likeness and unlikeness, equality, largeness, and smallness; and even thus a share
in being, motion, rest, and change.
This argument works, again, on generous assumptions (for example, the insistence
that difference, or being, are properties that this one which is not has) to a generous
conclusion. But its important feature for my purposes now is not its generosity, but
the assumption it makes about the status of this ‘one’. Take the sentence ‘if one is
not’, and focus on the predicate, ‘is not’. This seems to say something about the
subject, and it implies that we have knowledge of the subject. It also implies that the
subject has some kind of identity. For it is different from the others, and it is itself
mentionable as ‘that’ or ‘something’. But saying this begins to characterize the
subject, to endow it with properties, to turn it into a mentionable individual. So,
once the subject ‘one’ is enmeshed with a predicate, even so unpromising a predicate
as ‘is not’, it becomes a viable subject for speech (even if it does turn out to have a
hopelessly generous endowment of properties after all). Even in the worst case, when
the subject of a subject-predicate pair is said not to be, it retains its being and its
identity, both in relation to itself and in relation to everything else.
Take the sentence ‘if one is not’ and focus on the subject term, however, and the
result is quite different. The negative movement (163b–164b) supposes that the
negative ‘is not’ deprives the one of any character at all; it cannot partake in being,
or in change, motion, size, likeness, sameness, or difference. In that case, there can be
no knowledge or perception or opinion of it; it has no character at all. This reverses
the moves of the positive movement and, by stripping all the possible properties of
the one away (courtesy of ‘is not’), reduces this one to nothing. The negative here
fences off the subject from any predication; and such an austere subject is not
mentionable at all.
Now compare the two movements of the third stage. The first movement argued
that any successful subject-predicate sentence has a subject that retains its individual
identity even when it is said not to exist. The second movement, on the other hand,
31
Again, the way this is put is significant: from the formula ‘if one is not’ at 160b7, the word order is
changed to ‘one if [it] is not’ at 160c6, so that ‘one’ is outside the grammatical scope of ‘if ’.
shows that when the subject of a sentence is denied any identity, then the subject-
predicate sentence fails. So for talk to be meaningful, it must be attached to identi-
fiable subjects.
The fourth stage builds on the third when it asks ‘if the one is not, what is true of
the others?’ Now it transpires that these identifiable subjects must be individuals. In
the first movement, again positive (164b–e), the others are shown to be other (how
else would they be the others?) and different from each other. Their difference, if the
one is not, is multiform, or indefinite in number (even for the smallest of them); they
are masses. But then:
As masses of this sort the others will be other than each other, if they are other while there is no
one.—Certainly.—Therefore, they will be many masses, and each one will appear to be one,
even though it is not, since there is no one.—That is so.—And then they will seem to have
number, since each seems to be one, there being many of them. (164d)
Parmenides argues that these others, characterized as indefinite masses, will have
identity (they will be differentiated from each other) so they must appear one, and
numbered, and even and odd, in their relations with each other. What has identity,
that is, must at least appear to be countable; and if so, then it must appear to be an
individual. But these appearances will not be enough, after all, to save the others from
incoherence. For the identity of some mass, relative to some other as it is, does not
have any internal structure (165a–b):
And so, even if it has a limit relative to some other mass, itself relative to itself it has no
beginning or middle or end.—How?—Because when someone takes some part of them in
thought to be beginning or middle or end, there will always appear some other beginning in
front of the beginning, and after the end always some other end left behind, and in the middle
there will be other more middling middles, and smaller, since it is impossible to take each of
them as one, there being no one.—Very true.—So I think that all being which someone takes in
thought must be chopped up and broken in pieces; for it would always be taken to be a mass
without a one. (165b–c)
From this, Parmenides takes it to follow that anything can seem to be true of these
indeterminate entities: they are both the same and different from themselves and
each other, both in contact with each other and not, both moving and at rest, both
coming into and being destroyed, and neither. In short, of such entities anything at
all is true (and our speech turns out meaningless).
Alternatively, according to the second, and final, movement (165e–166b), if the
one is not, there is no many either. For none of the many can be one; and then each is
nothing, and there is no many at all, either in reality or appearance. Anything, that is,
must be an individual before it is anything else; but if the one is not, nothing is an
individual. Why not? The conclusion would not, of course, follow, if ‘the one is not’
meant that some particular one is not (for example, the Eleatic One); and indeed it
might not be thought to follow if there is no form of one. It would, however, follow if
‘the one is not’ is true for any one. For that implies that there are no ones; and that in
turn implies that no member of a plurality can be a one. Parmenides concludes that
without ones, there is nothing.
This pair of arguments began with the hypothesis that the one is not. This is taken
to mean that there are no ones, no genuine, paid-up individuals to correspond to the
subjects of sentences. But then, Parmenides argues, there must be bogus individuals
instead; for anything we talk about, if our talk is to be successful, must appear to be
one, even if in fact it is not. Even bogus individuals, however, will not save the day, for
they lack the proper identity that must underlie their being ‘grasped in thought’. So
both thought and speech need genuine individuals, with genuine identity.32 Individ-
uals are basic to thought and speech—without them reason perishes.
7. Constructing Metaphysics
The first stage of the second part of the Parmenides contains three canonical
assumptions that hold the remaining sequences of inference together:
1. Any individual may be understood in either the austere or the generous mode.
In the negative movements, the one we are dealing with is austere, just one; so it
has no parts. In the positive movements, the one we are dealing with is a
collection of properties, so generously conceived that it may have any property
at all, as well as its opposite.
2. The identity of the one, or its nature, may consist in its being (austerely) just
one; in that case it is nothing but itself (which implies in the end that it is not
even that). Or the one may be one because it has the property of unity
(generously distributed over all ones); in that case, it is anything at all.
3. Thus, being one may be a matter of being identical with the one (that is, with
some individual already specified), or a matter of having the property of unity
(which belongs, thus, to any specifiable individual).
The foundational arguments for the first, negative sequence exploit the austere
account of individuation; from the inadequacy of that account all the other paradoxes
flow, concluding with the denial of being. The puzzles here, then, are not about
existence (as is often thought), but about being one, being an individual. If an
individual has the austere nature of this one, then it cannot enter any relations at
all with anything, whether itself or anything else. And from this conclusion flows
another—that such an individual has no affections (pathêmata, 141d4) at all, and
therefore no contact with being.
32
Schofield (1977), 156, comments: ‘this appearance of unity is only an appearance, with the conse-
quence that the individuation of non-unitary things is radically unstable, so unstable that talk of individ-
uals does not properly gain a purchase’. I take Parmenides’ overall point to be stronger—that no talk gains a
purchase unless it is talk of individuals.
The second movement generously allows that the one has parts, and supposes that
being one, and being, are universal properties of whatever is one (‘never lacking’,
144b2, ‘distributed’, 144b1, ‘parcelled out’, 144b5). From these (premises of the
generous account of individuation), the remaining arguments are drawn. The argu-
ments of the second movement are more complex than the first; but they echo the
strategy of the first movement. They begin from the generous hypothesis that
individuation is a matter of properties, and identification a matter of ascribing
predicates, and from its assumption that we can understand ‘one’ and ‘being’ as
properties, universally spread out. From this are derived secondary relations between
the one and anything else, and thereafter other affections, both at a time and over
time. The conclusion is grossly paradoxical: this one is, was, will be, became,
becomes, and will become anything at all; and we can know it as such because we
can say it to be such.
The same structure appears in the remaining stages of the gymnastic session. The
third movement of the first stage discusses the identity of the one over time; it uses
both negative and positive arguments to show that becoming cannot be explained on
either a generous or an austere view of the individual, so that becoming takes place in
an absurd ‘suddenly’ which has, it seems, no relation to whatever individual it is that
might underlie any change. The coda reinforces the paradoxes of the first two
movements by making us worry about the identity of an individual through time;
together the first three movements constitute an exhaustive attack on all conceptions
offered so far of any individual (any ‘one’), either at a time or over time.
In the second stage, Parmenides tackles the relation between the one and ‘the
others’: ‘if one is, what is true of the others?’. Just as the first stage treated ‘the one’ as
needing determination by the arguments themselves, so we understand what is
meant by ‘the others’ as the argument progresses.
The positive movement (157b–159b) argues that while the others are not the one,
they partake in the one because they are wholes with parts (as anything other than
the one must be). The parts of these wholes, likewise, are each one by virtue of
participating in one (but not by being the one). So the others, which participate in the
one, will be many and indefinite (in contradistinction to the one); and thereafter
prone to all opposite affections.
The trick is turned here by two different moves. First, Parmenides assumes that for
something (some individual, emphasized at 158a) to be one, it must have the
property of unity. Being one (as in the second movement of the first hypothesis) is
spread out and divided up over all individuals. Second, he supposes that without such
a property these others will be indeterminate and indefinite, and have no features of
their own apart from their indistinct plurality. Even if they are hypothesized as
individuals, their individual nature can only be explained by their having the
property of unity. For something is either one, or many (partaking in one)—or it is
nothing at all (158b4).
The negative movement, conversely, starts by assuming that the one and the others
are totally distinct and separate; and that they exhaust all there is (159b). The one is
completely one (as in the first movement of the first hypothesis). The others, then,
can have no share in this one; nor can they be many (otherwise they would share in
the one). So there is no number among the others, no likeness or unlikeness (for
possessing both qualities would make them two, not indefinite); indeed, they turn out
to be unqualifiable and unmentionable, just as the one was in the negative movement
of the first stage.
This argument exploits its positive predecessor by making the point that the one,
austerely conceived, is not (after all) a property spread all about, but just one. If the
others are defined as other than any such one (here the use of ‘one’ as a variable starts
to matter), then they cannot have any plurality either; and on that count they, like the
one, turn out absurd. The conclusion insists once again that a proper account of
individuation (not existence or anything else, but individuation) is vital for any
attempt at dialectic, or any talk at all, to get off the ground. Yet the clear contrast
between this treatment of the one and the treatment of one as a property in the
preceding positive movement makes discourse impossible. The dilemma is complete.
In response, then, we might find ourselves denying that there are individuals at all.
The third and fourth stages set out the price for that. If we try to deny that there are
individuals, the very denial imports individuals as its subject. Otherwise, speech fails
altogether. Even if there are no individuals, our discourse will import bogus individ-
uals instead, even if to do so threatens rational speech. Without them, speech is
impossible.
This entire sequence of arguments and paradoxes, therefore, treats the discussion
of individuation as foundational. But that treatment is developed in two ways. In the
first two stages, where the hypothesis supposes that there are individuals, Parmenides
offers us a collection of puzzles about how individuals are to be conceived and
understood. In the last two stages, he attempts a hypothesis that denies that there
are any individuals. This shows that, after all, individuals are necessary for discourse;
individuals are basic.
So the gymnastic session is not a haphazard collection of philosophical puzzles (as
some have supposed), but a connected whole. The connections, however, might be
understood in an ordered way, since each new phase in the discussion investigates
the failings of its predecessor. What is more, this order relates the second part of the
Parmenides to the first.
Recall that the first part of the dialogue offered a confrontation between two
committed ontologies: the austere monism of Elea and the more liberal, structured
ontology of Socrates. Strong monism (as even Zeno allows) is a nasty dose to swallow;
it offers a minimal ontology, full rationality, but little scope for common sense.
Socrates saves common sense by arguing that so long as we have some austere
individuals as the basis of our ontology, we may admit generous individuals too.
But that theory came to grief at Parmenides’ hands: by the end of the first part of the
dialogue, neither of the ontologies canvassed by the protagonists had retained its
persuasive power. Two fundamental questions, however, had been raised about
individuals: first, how are we to determine the conditions of individuation (and
identity)? Second, how are we to show how an ontology, composed of individuals,
is structured? A further, more general, issue was canvassed too: what principles are
we to say govern any theory? In particular, this was asked of ontology. How far is any
ontology shaped and structured by the twin demands on theory—that it should be
universalizable, and that it should be parsimonious? Reflection may show that these
questions dominate the arguments of the gymnastic session, as Parmenides turns
away from a committed ontology to answering these higher-order questions.
I have argued that the first two stages of the gymnastic session replicate some of the
assumptions of the committed ontologies of the first part. That is to say, in the first
and second stages, where the hypothesis is positive, ‘if one is’, the arguments
investigate the plausibility of the two conceptions of the individual—the austere,
simple account, that an individual is just one, and the generous, complex account,
that an individual is one and many—relied on by Socrates and Zeno. Neither
account, considered now quite generally and outside the context of some particular
ontology, allows its individuals to be both accessible to reason (or even speech) and
free from contradiction. Not only are the theories of Zeno, Parmenides, and Socrates
at risk here, however, but so is any theory that postulates individuals and cannot
provide an account of how they are to be determined.
At this stage, the argument once again reflects on itself, by speculating on the
consequences of denying that there are individuals. The third and fourth stages of the
gymnastic session, therefore, are of a higher order than the first and second, just
because they investigate what might be true if the assumption that there are individ-
uals is denied. If the first and second stages treat the existence of individuals as
necessary for ontology (as the committed ontologies of the first part had done), the
third and fourth stages offer a higher level of abstraction by doing away with all
individuals. This sequence obeys the constraints on theory outlined at 130 ff.; they try
to practise ontological parsimony at the same time as they tackle universalizable
principles of ontology. One might say that the conclusion is positive (that individuals
are after all basic to ontology)—and this assumption is one that Plato continues to
maintain (compare, for example, the demand at Sophist 262d ff. that every sentence
should be ‘of something’, that is that its subject term should successfully refer). Or
one might say that this conclusion forms a final dilemma—the first hypothesis shows
that it is impossible that there be ones; the second that it is impossible that there be
no ones—impossible at least so long as dialectic is to survive.
8. The Unity of the Parmenides

In the transmission of Chinese Whispers, the message may be distorted or distanced
from its original version; but each transformation is itself an interpretation of the
original. In a parallel way, if my reconstruction is right, the arguments in the

Parmenides are constructed in layers just because each stage reflects on its predeces-
sor. In the first part, the protagonists have firm ontological commitments, Parmeni-
des and Zeno to monism, Socrates to the theory of forms. Each fixes his ontology by
specifying how many individuals he will tolerate, and why. But first Socrates’ dispute
with Zeno and then Parmenides’ attack on Socrates point to the difficulties either of
monism or of Socrates’ complex dualism; at the same time, the aporia that closes the
first part questions whether any rational metaphysics can survive without these
ontological commitments. Consequently, the first hypothesis of the second part
examines how we may understand what individuals are, and concludes with puzzle-
ment. Then the second hypothesis supposes that there are no individuals, and shows
how then speech and reasoning are impossible. The terminal aporia claims that,
whether we postulate individuals or not, dialectic seems impossible (and even this
conclusion, established as it is by dialectical means, seems self-refuting). It might be
thought to drive us either towards nihilism (if reason tells us anything, it tells us that
there is nothing) or towards unreason (of course, there are things, so reason tells us
nothing). Does the Parmenides have such a dismal conclusion?
Consider, first, the two aporiai. It is an important feature of Plato’s representation
of aporia that it is not merely a logical feature of an argument; it is felt by the person
whose beliefs are under scrutiny. Aporia, that is, is a psychological state, belonging to
some person in the dialogue. Appropriately, therefore, the first aporia of the Par-
menides, which closes the personal, Socratic section of the dialogue, is felt by the
character Socrates.33 For it is Socrates who is committed to the theory of forms, and
Socrates who supposes that, with the theory of forms, the puzzles that Zeno proposes
may he avoided. Socrates, therefore, is bound to be impaled on the horns of
Parmenides’ dilemma; it is Socrates who supposes that there is no dialectic without
forms, and Socrates who must concede that his theory falls into the difficulties that
Parmenides has outlined. So the dilemma fits the character of the dialogue in which it
appears. In the second part, by contrast, the vivid characterization has faded, and the
encounter is impersonal and arid. This fits the arguments; the assumptions made by
the second part of the dialogue are not part of the theoretical baggage of the
protagonists. Instead, the arguments are quite general, for they apply to any theory
that either postulates individuals or denies that there are individuals, and they
enquire whether either postulate is tenable. In that case, the terminal aporia is not
felt by a character in the dialogue, because the argument is now so general that it
applies to any theory held by any person at any time; the feeling can be had by
anyone. The characterization of the second part is minimal because it is general; and
the terminal aporia, although it parallels the aporia of the first part, is no longer ad
hominem. If the dialectic of the second part is general, its target is not the characters
33
Cf. e.g. Meno, 79 ff.
in the dialogue, but instead anyone who may encounter the arguments: it is addressed
to the reader.
Consider, second, the way the frame of this dialogue is strangely open; we are
never returned to the dramatic context so elaborately portrayed at the beginning of
the dialogue. Instead we are left with the bleakness of the terminal aporia, and
Aristotle’s glum ‘Very true’. Now it could be argued that this open-endedness does
actually conceal the inference we should make (such as, ‘Oh yes, clearly nihilism is
true’); more plausibly, however, it represents the way the argument is itself open,
unresolved beyond the terminal aporia itself. The series investigates two particular
theories, then investigates any theory that postulates individuals, and then investi-
gates the consequence of denying individuals. Just as each stage reflects on its
predecessor, so the final aporia is itself susceptible to further reflection. In particular,
it invites reflection on the assumptions it has employed. Each phase of the second
part has called attention to the two characterizations of the individual on offer, the
generous and the austere. The very shape of the terminal aporia, drawing attention as
it does to the treatment of these characterizations as exclusive and exhaustive, invites
us to consider whether there is any way to be forced between them. What is more, it
invited Plato to do so too. For he revisits the puzzle of the relation between austere
individuals and hopelessly generous ones in the dialogues that follow the Parmenides.
The Theaetetus contrasts bundles of properties (discussed at 152 ff.) with simple
elements (201 ff.), the Timaeus contrasts the complexity of the world-soul, 35a ff.,
with the unmentionable nature of the receptacle of becoming, 49a ff.; and in the
Sophist the discussions of being and not-being, 237–49, alternate between austere and
generous treatments of both what is and what is not. Finally, in the Sophist (254 ff.)
and in the Philebus, Plato revises his account of relations as real features of the relata.
As a consequence, he is able to offer an account of individuation that determines
individuals by their relations to each other, without generating the hopeless plurality
characteristic of the generous view, and without supposing that these entities, to be
individuals at all, need to be utterly austere.
The Parmenides is not, after all, a haphazard collection of arguments, but a unified
whole, in three respects. First, the literary construction of the dialogue, indeterminate
though it initially appears, parallels the layered way the arguments proceed. Second,
the separate stages of the argument are ordered because each stage reflects on its
predecessor. Third, the dialogue is unified by its topic. The metaphysical question
that underlies each stage is: ‘what is it to be an individual?’. Thus, the subject of the
puzzles of the second part is not existence (as many have supposed), but individu-
ation, just as the dispute between Socrates and Zeno was about just how many basic
individuals we need, and the debate between Socrates and Parmenides tackled
Socrates’ specification of his basic individuals. Socrates, his theory of forms under
attack, worried about some general conditions for theory: that theories should be as
general as possible, and that they should be parsimonious with their entities. The
tension between the demand of universalizability and parsimony is evident
throughout the dialogue. Plato uses the first encounter between Socrates and his
Eleatic opponents not only to scrutinize his own theory, but also to examine the
conditions that govern any theory. In particular, he suggests that any theory must be
underpinned by some account of its basic entities. In the second part, as a conse-
quence, Plato conducts an investigation of the general principles for any entity to be
basic: and that is for any entity to count as one entity in someone’s ontology. The
Parmenides is unified by the way in which it advances towards more and more
general metaphysical principles; and it is unified also by the basic assumption that is
gradually revealed throughout—that any theory must be based on some ontology,
and that any ontology must be based on a theory of individuals. At the object level,
the dialogue ends in impasse; but the principles it enunciates at the formal level offer
the means to progress beyond.34
34
Some of the material in this chapter (in }} 4–5) appeared in my (1994); I am grateful to Princeton
University Press for permission to reproduce that material, and see (1994), chs. 3 and 4, for a more detailed
discussion of some of the arguments treated cavalierly here. For comments on this chapter and on earlier
versions of it, I am particularly grateful to John Cooper, Nick Denyer, Dorothea Frede, Christopher Gill,
Keith Hossack, Geoffrey Lloyd, Malcolm Schofield, and an anonymous reader for Oxford University Press.
II
On Knowledge and Virtue
in Plato
9
Looking Inside Charmides’ Cloak:
Seeing Others and Oneself in
Plato’s Charmides
1. Critias’ Proposal
The Charmides offers a series of examples of the untranslatable Greek virtue sôphro-
sunê (so I shall refrain from translating it: some offer ‘temperance’ or ‘self-control’),
and, among others, an influential definition of it, as knowledge of oneself. One of the
reasons, it seems, why self-knowledge comes to grief as an account of a virtue—as it
does by the end of the dialogue—is the repeated assimilation of knowledge and self-
knowledge to perception and self-perception. This assimilation, I shall claim, is both
deliberate and tricky: I hope to offer some explanation of its point.
After a good deal of horseplay—some of it apparently pretty smutty—Socrates
embarks on a discussion of sôphrosunê, first with young Charmides, and then with
his mentor, Critias.1 Thereafter almost half of the dialogue is taken up with the
refutation of Critias’ passionate declaration that sôphrosunê is knowing oneself
(164c–d, 165b4). That is glossed as the claim that sôphrosunê is the knowledge of
itself; and that is what comes under critical scrutiny (165c7). (I shall call the first
‘knowledge of the self ’, the second ‘self-knowledge’.) Socrates wonders first whether
self-knowledge is possible; and second whether, even if it is possible, it is any good to
us (if not, it can hardly, he suggests, count as a virtue). I shall ignore the question
whether it is any good, and consider instead the arguments that are designed to show
that self-knowledge is impossible or, if not, thoroughly odd.
Oddity seems, at first, to be to its credit: when Socrates objects that self- knowledge
is unlike any other knowledge in not having a subject matter other than itself, Critias
responds that this is exactly its distinguishing mark. Critias insists that sôphrosunê is
the only knowledge that is of itself and all the other knowledges2 (166c2–3, 166e5–6).
For:
1
The dramatic tone of the dialogue is intensified by the record of the (extreme) political activities of this
Critias. Critias was a member of the Thirty in 404/3, against whom Socrates’ civil disobedience is recorded
at Ap. 32c; and Charmides was at least implicated in their activities. See Nails (2002), 108–13, 90–4.
2
I translate epistêmê as ‘knowledge’, and tolerate rebarbative expressions like ‘the other knowledges’, in
order not to beg the question from the outset about the nature of knowledge (is it, e.g., piecemeal or
174 LOOKING INSIDE CHARMIDES ’ CLOAK
I. sôphrosunê is knowledge of itself (it is reflexive) and of other knowledges (it is

second order).
II. sôphrosunê is not of anything else (other than itself and the other knowledges)
(it is just second order).
It follows from these two theses that there is a knowledge which is not of anything
else but of itself and of the other knowledges, and of ignorance too (167b11–c2). In
what follows next, the focus of attention is on the relation between knowledge and its
objects:3 on the ‘of ’ relation. What is it for knowledge to be ‘of ’ its object? And how
does that reflect on the possibility of either a reflexive cognitive state or a higher-
order one?
Socrates offers a triad of arguments against Critias’ proposal: I shall call them,
collectively, the Relations Argument:
1. (167c–168a) Reflexivity is apparently impossible in other cases (similar to
knowledge): perception, desire, and belief.
2. (168b–d) The ‘of ’ relation expressed by the thought ‘this knowledge has the
power (dunamis)4 of being of something’ is (in some central cases: larger,
double) irreflexive.
3. (168d–169a) Thence it seems implausible that perception should perceive itself
(168e9–10); perception may only perceive its own special objects.5
The first phase of the argument assimilates perception to some other states6 such as
emotion and belief—states we might readily describe as psychological.7 The second
assimilates it to other, non-psychological relations. Perception occupies a pivotal role,
therefore: for it allows Socrates to claim that the features of non-psychological
wholesale? Is it craft-like or virtue-like, or both?). On this, and on whether we should understand

knowledge as ‘understanding’, see of course Burnyeat (1981); Nehamas (2004); and also Lyons (1963)
and Bailey (2006).
3
I use ‘object’ to refer to the intentional object of knowledge, belief, etc.; what the ontological status of
such objects is, I leave open.
4
This expression (coupled as it later is with ousia) may seem to anticipate the argument of Rp. 476–480,
especially 477c–e. In what follows, I argue that the dunamis/ousia characterization of the relation between
knowledge and its objects is rendered problematic in the Charmides; I suspend judgement here on how that
may affect our reading of Rep. V.
5
Aristotle is undoubtedly reading this passage at de an. 425b12–20: notice the close correlation between
Chrm. 168d9–e1 and de an. 425b17–20.
6
It is unclear whether we should speak here of states or events. The argument focuses on the relation
between psychological subject and object, which suggests that we should be thinking of psychological
states; but the resulting view of perception may well be one that sees it in terms of coming to be in that
relation, and so as an event rather than a state (compare the theory of perception as an event at Tht. 154 ff.).
In what follows, I speak of ‘states’ except where the terminology of ‘events’ is obviously more felicitous;
since this occurs at 167, where the event is a second-order seeing, I think the dual terminology does not
interfere with the point I make.
7
In the sequel, it becomes unclear just how we should classify the difference between the psychological
and the non-psychological; I use the expression ‘psychological’ in what follows immediately with a health
warning.
LOOKING INSIDE CHARMIDES ’ CLOAK 175
relations also characterize cognition. This turns out to damage the possibility of both
reflexive and higher-order cognitive states. I shall wonder just how, in the economy of
the dialogue, this conclusion forces us to modify our assumptions: about perception,
about self-perception, and about self-knowledge.
2. The Relations Argument

Consider the Relations Argument in more detail. Critias claims there is a knowledge
that is of itself and the other knowledges, but not of the objects of the other
knowledges.
In the first phase of the argument, Socrates gets Critias to concede that perceptions
must be at least of their special objects: that is, sight must be at least of colour, hearing
at least of sound. So, in general, no perception could be only of itself and the other
perceptions. The same, Socrates argues, is true for other psychological states: no
desire, for example, is of itself and of the other desires, but not of pleasure; and the
same for wishes, love, and fear. Cognitive states, too:
. . . surely there is no belief which is of itself and the other beliefs, but not of the things the other
beliefs believe? (168a3–4)
So it is absurd to suppose that:

. . . there is a knowledge which is such as to be of no learning but it is of itself and the other
knowledges. (168a6–8)
Socrates’ argument so far claims that Critias’ original restrictive thesis (that there is
some knowledge that is only of itself and the other knowledges) is implausible: all
these psychological states must be at least of their special objects (whatever those may
be). The next phases of the Relations Argument, however, go further.
For the second phase Critias agrees that:
Knowledge itself 8 is of something, and it has some such power (dunamis) as this, to be of
something. (168b2–3)
Socrates takes this claim (about the ‘power to be of something’) to capture an

asymmetric relation. Consequently, he takes licence to construe it on the model of
some non-psychological relations, notably larger than and double. Suppose there is
something (Gerald the giraffe, let us say) that is larger than itself and the other larger
things, but not larger than the things than which the other larger things are larger. If
Gerald is larger than himself, himself will be smaller than him. If, further, he is larger
than the other larger things, but not larger than the things they are larger than, he will
8
This is reading autê, as proposed by Shorey (1907), not hautê as given by Burnet. Van der Ben (1985),
55, rightly, I think, glosses ‘in general’; and compares Rep. 438c7, Prm. 134a3.
be both larger and not larger than them.9 Or double: suppose something is double
itself and the other doubles, but not double the things of which they are double. Then
it will be its own half; and its half will be double its own halves, as well.10
Socrates concludes: ‘Whatever has its own power (dunamis) towards itself it will
also have the being (ousia) towards which the power (dunamis) is’ (168d1–2). This
both summarizes the argument so far, generalized across all relations, and provides
the springboard for the third phase, on perception. But what does it mean? It
manifests, perhaps, two assumptions of the argument. First, it reiterates the suppos-
ition that the relations under consideration are asymmetric ones: the power of one
relatum is correlated to (‘of ’) the being of the other. Second, in such cases there is not
only something we can say about the relatum that has the power, there is also
something in particular we can say about the other relatum: that it has the being
correlate to the power. Problems then arise when—as here—the relatum that has the
power is identical to the relatum that has the being: for then one thing will display
opposite properties, the correlate features (the power and the being) of an asymmet-
ric relation.
Notice, however, that more has happened here. First of all, Socrates moves easily
from claims about perception, emotion, and cognition in the first phase, to claims
9
Socrates asks: ‘So if we were to find something larger, which is a) larger in relation to itself and the
other larger things, but b) not larger in relation to the things in relation to which the other larges are larger,
then this would surely happen to it, that c) if it is larger than itself it will be smaller than itself, too. Is that
not so?’ (168b10–c2). The argument can be construed the easy way: if Gerald is larger than himself (we
should add, as Aristotle would exhort, the qualifications that avoid the logical difficulties: Metaph.
1005b20–2), then the himself than whom he is larger is smaller than himself: so he is both larger and
smaller than himself. This moves directly from (a) to (c). Or it can be construed the hard way: if Gerald is
larger than himself and the other larger things, but not larger than the things than which the other larger
things are large, then the other larges will be larger than Gerald; so he will be both larger than them and
smaller than them, although they are (ex hypothesi) smaller than him. This moves from (a) via (b) to (c),
and takes (c) to be demonstrated both by reflexive considerations and considerations of transitivity and its
failure. The easy way leaves (b) out of consideration altogether; why, if not to offer us the hard way, is (b)
still in the text? Since it is, contrariwise, two points are being made about these relations at once: about their
reflexivity and about their transitivity. In what follows, I shall suggest that both characters of relations are
important in the discussion of self-perception.
10
Socrates’ question (168c4–7) might be construed in two slightly different ways: ‘And so if something
is double all the other doubles and itself, then I suppose it will be double in relation to both itself, its half,
and the others. Double, after all, is (exactly) of a half.’ Or: ‘And so if something is double all the other
doubles and itself, then I suppose it will be double while both itself and the others are halves. Double, after
all, is (exactly) of a half.’ Both the double’s relation to itself and its relations to ‘the others’ are expressly
included here: so we might expect the resulting puzzle to turn once again on both reflexivity and
transitivity. In the case of double, unlike the case of larger, the relation between this double and all the
other doubles should indeed be intransitive (if a double has a half, the half of its half will not be half of the
original double). But then the puzzle here seems quite different from that of the preceding move about
larger: the preceding puzzle works because larger is transitive, this one because double is not. The passage is
highly elliptical, however, so that it is unclear just where the puzzle is supposed to arise. Here is an attempt:
at least, if something is double itself, itself will be its half, and so both half and double at once. Then also if
something is its own half, it, qua half, will also stand in the same relation to its double as its own halves
stand in to the same thing, the double. But any double is exactly double its half; so neither the relation
between the double and itself nor the relation between the double and the other doubles makes sense.
about larger and double in the second, and supposes that we can reasonably gener-
alize across them. One might complain that the generalization is spurious: the
relations in question—marked by ‘of ’ or ‘than’—are only apparently similar, their
differences masked by an over-generous use of the Greek genitive.11 But that
complaint might be theory-laden. For it may rest on the supposition that the
examples of the first phase are quite different in kind, and so in their relations,
from the second—because the first examples are psychological, the second not.
Perhaps Socrates, by contrast, makes here no unargued distinction between psycho-
logical or mental states and (merely) physical ones; instead, he offers considerations
about some relations (e.g. ‘larger than’) as a way of accounting for others (e.g.
‘perceives’). (It is we who give the latter a problematic metaphysical status, by calling
them ‘psychological’ or ‘mental’, and then wondering what sort of property those
terms describe.)
Second, this phase of the argument is carefully structured to turn on two features
of the (asymmetric) relations under scrutiny: on whether they are reflexive, and on
whether they are transitive (where, intuitively, some are, e.g., larger; and some are
not, e.g., double). Critias’ proposal was that there is a knowledge that is both of itself
and of the other knowledges, but not of the objects of the other knowledges.
Translating this into talk of relations, Socrates takes Critias to be committed to
saying that self-knowledge is an asymmetric relation that is both reflexive (it applies
to itself) and intransitive (if the knowledge is of the other knowledges, it is not of
whatever the other knowledges are of ).12 But this, Socrates suggests, is absurd, on
both counts. Both because it insists on reflexivity and because it denies transitivity,13
Critias’ original account of self-knowledge fails.
In the third phase of the argument, the general principle about relations is applied
to perception; but now the focus has narrowed, leaving transitivity out of view
to consider the issue of reflexivity alone. Socrates makes a strong claim about
perception:
Hearing is of nothing but of sound. (168d3)
Sight sees nothing that is not coloured. (168d10)
So perceptions have special objects, to which they are exactly correlated: hearing is
just of sound, sight just of colour. Given that a perception occurs, on this view, its
11
The objective genitive in ‘sight is of colour’ and the comparative genitive in ‘the giraffe is larger than
the gerbil’.
12
So, if a is knowledge of b; and b is knowledge of c; still, a is not knowledge of c. If these relations
between knowledges are understood as ordered, with a the highest, this means the content of the first-order
knowledge is not included in the content of any higher-order one. Socrates later goes to considerable
lengths to point out the absurdities of this thought, 170a ff.
13
That transitivity, as well as reflexivity, is at issue is brought out by the detailed treatment of the two
relations larger and double, the first of which is (intuitively) transitive, but disallowed as such by Critias’
proposal; the second of which must be intransitive, but whose intransitivity seems not to protect it from
absurdity either; see n. 9.
special object is both necessary and sufficient for its content: the being of its object
determines the power (the nature) of the perception itself. But this is a puzzle for self-
perception: if sight sees itself, it must see itself coloured; and if sight is not coloured, it
cannot see itself.
What exactly is the puzzle? Perhaps it turns on the special objects; perhaps
Socrates simply needs an empirical claim (as a matter of fact, sight is not coloured14)
to generate his conclusion. But that seems just too quick for all that has happened in
the argument; this conclusion was available as soon as the first phase said that the
special objects are necessary for perception. Instead, I suggest, the model of sight as a
relation persists, and the puzzle derives from the principle that the power (in this
case, sight) and its object (colour) are asymmetrically related. Then it is a metaphys-
ical mistake to say that sight sees itself, since that would be to conflate distinct relata:
the power with its object, hearing with a sound, and sight with colour. This gives
Socrates his conclusions, that:
i. Relations such as size cannot be reflexive (168e5–6).
ii. Reflexive powers (dunameis) are implausible in cases such as perception and
motion15 (but perhaps not in other cases) (168e9–169a1).
Where does this leave us? More significantly, where does this leave self-knowledge?
If the argument goes through, it concludes that none of these relations—whether
we would call them psychological or not—is reflexive; so Critias’ thesis fails. Why
might we care if it does? Critias has proposed that self-knowledge is both reflexive
and higher order, but not transitive: that is, it does not have as its objects both first-
order knowledge and the objects of that first-order knowledge, but just the first-order
knowledge, simpliciter. In developing his counter-argument, Socrates focuses both on
reflexivity and on higher-order thought; but he denies Critias’ thesis on the basis of
reflexivity alone. This leaves open two questions, which may well be distinct:
i. Does the denial of self-knowledge imply that there can be no knowledge of the
self ? (Why should we care?)
ii. Does the denial of self-knowledge imply that higher-order knowledge (or, in
general, higher-order thought) is impossible too? (Why should we care?)
I shall suggest that the real cost of the refutation of Critias is that knowledge is
denied its higher-order dimension (so the question of just how all these relations
might be transitive is important). I shall further suggest that the argument is set out
in the elaborate way that it is, and in the context it is, to show how we might avoid
14
Once again this dimension of the argument bears comparison with its Aristotelian commentary in de
an. III. 2; it is, of course, a vexed question whether Aristotle thinks that sight is, or is not, literally coloured.
See, notably, Burnyeat (1992); Sorabji (1992).
15
The inclusion of motion at this point is rather a surprise, we might think. It brings out, however, that
perception is construed in this argument as a brute causal relation; below I suggest that this is the refutand.
paying this price: and why it matters that we do avoid it. En passant, this will deliver a
conclusion about how we should read the dialogue: by taking the details of the
dramatic context into account when we try to figure out just what the arguments say.
3. The Structure of a Rebuttal

Socrates has argued that some ordinary relations (large, double) and perception,
emotion, and cognition are asymmetric and therefore irreflexive, fatally to Critias’
account. And he has suggested that there is something problematic about Critias’
claim that these same relations are all intransitive: especially, this will mean that the
content of higher-order knowledge will be just lower-order knowledge, but not the
content of the lower-order knowledge.16 Is anything about self-knowledge to be
salvaged from Critias’ proposal? Three broad responses suggest themselves:
1) Capitulation: the argument works, and so much the worse for self-knowledge
and higher-order thought.
2) The argument goes through for relations such as large and half; perception is
such a relation, so for perception as well. The argument does not go through for
cognition (emotion is another matter, tricky in all sorts of ways; I shall not
discuss it in detail here).
3) The argument goes through for relations such as large and half, but not for
psychological relations/states.
Well, perhaps capitulation is just bad news for Critias, whose thesis about sôphro-
sunê is made to collapse. But it would not be so good for Socrates, either, to lose the
possibility of higher-order thought. When he reflects on how the discussion is to be
conducted, Socrates insists on the importance—both argumentative and ethical—of
questioning whether we know what we think we know; and he characterizes this as
inquiring into oneself (tracking oneself down):
How could you think that—however much I test you—I do so for any other reason than
I would give for tracking myself down, asking what I am saying, fearing lest I may escape my
own notice thinking that I know something, but not knowing it. And so now I claim to be
doing just this: looking at the argument most of all for my own sake, and perhaps also for that
of my friends. Or don’t you think this is a common good for just about everyone, that each of
the things that are should become clear, as it is? (166c–d)
In Socrates’ version of Critias’ thesis, likewise, this inquiry into oneself is connected
with self-knowledge, and the ability to test others: but now with the apparent
restriction that self-knowledge is just knowing what one does and does not know.
16
This is the main focus of attention in later arguments, 169d–171c, which I do not discuss in detail
here.
So the person who is sôphrôn alone will know himself and will be able to test what he knows
and what he does not, and will have the power to look for what others might know and think
that they know, and again what they think they know, but don’t; no-one else can do this. And
being sôphrôn and sôphrosunê and knowing oneself are just this: knowing what one knows and
doesn’t know. Is this what you are saying? (167a)
Compare and contrast the two passages. Socrates’ version is heavily ordered—he
inquires into whether he knows what he thinks he knows—and unrestricted as to the
content of any one of these orders. Critias’ version, on the contrary, insists that
higher-order thinking is intransitive (we are allowed to know that we know some-
thing, but not its content), and so turns out to be either impossible or absurd.
Socrates is committed to showing, therefore, both how we know what we know
and why it might matter. Indeed, we ourselves are implicated: if Critias’ version
prevails, what are we doing thinking about what Plato thinks Socrates might think?
Could we ever know what it is that Socrates might think, if higher-order thought is
blocked in Critias’ way? And would we care?
4. Rethinking Perception
The argument needs to be resolved, then, either in favour of 2) or in favour of 3):
whatever we may say about reflexivity, we had better save something of higher-order
thought, of reflection. To do this, we need to think again about perception (as
I suggested, the pivot of the argument) and how it works.
On 2), perceptions fail to be reflexive just because the relation between perceiver
and perceived conforms to the model of the non-psychological relations. It is, as
I shall call it without prejudice, brutish: an exclusive relation between perceiver and
object with no room for mediation or indirection. On such an account, there will be
scope for claiming that perception is quite different in kind from cognition, so that
perception falls, where cognition does not, to the Relations Argument.
On 3), the divide may come in a different place: between what is psychological and
what is not. On this account, perception would have higher-order capacity; it would
be, as I shall call it, civilized, since its closest affinity is with cognition and the other
psychological states. It remains to be seen whether the civilization of perception
could resist the Relations Argument, and rescue any account of self-knowledge or of
self-perception.
Should we opt for response 2) to the Relations Argument (and brutish perception)
or for response 3) (and civilized perception)?
The Relations Argument begins:
s: ‘Well, see how odd is what we are trying to say, my friend; for if you look for the same
thing in other cases, you will believe, I think, that this is impossible.’
c: ‘How is this? What cases?’
s: ‘Cases like these: for consider whether you believe that there is a sight which is not a sight of
the things other sights are sights of, but is sight of itself and the other sights and likewise of
the non-sights; and although it is a sight, it sees no colour, but sees itself and the other
sights. Do you believe there is any such thing?’ (167c–d, my italics)
Socrates says to Critias: ‘see how odd is what we are trying to say’ (167c4); and
follows up his remark with a discussion, in the first instance, of sight.17 Seeing, here,
is both first order (we are talking about seeing colours) and higher order (it has as its
content something about sight). At the end of the Relations Argument, Socrates says:
s: ‘And sight, I suppose, my excellent friend, if it sees itself, must have some colour; for
something colourless, sight could never see.’
c: ‘It could not.’
s: ‘So do you see, Critias, that for the things we have gone through, some seem to us to be
altogether impossible; and others are hardly credible as having their own power in relation
to themselves? For largeness and multitude and such things are altogether impossible,
aren’t they?’ (168d–e; my italics)
Here too, the discussion of sight in the argument is juxtaposed with an exhortation to
see in the frame dialogue.18 The sight that is under discussion sees colours; the sight
urged in the frame has the conclusion of the argument itself in its scope. Does this
double use of sight have any significance?
The framing of the discussion of sight with instances of seeing as a second-order
psychological event19 is, first of all, ostentatious, placed strikingly at the opening and
the close of an argument about seeing. It is troubling, too: for the framed argument
undermines the possibility of second-order seeing. Are we to say this is a mere
literary accident (even were we to have an account to give of what that might be)? Are
we to say that the frame is outside the argument, irrelevant to it, merely the gilding
around the real picture? The onus probandi, perhaps, is the other way around:
without a methodical way of demarcating the gilded frame from the arguments
within, and some grounds for supposing it irrelevant, we should prefer to start by
supposing that all of the dialogue somehow or other counts towards its point. We
should not, therefore, dismiss out of hand the double use of sight here.
In that case, it seems that the dialogue itself directs us towards an account of
perception as civilized, and wards off the brute. For the simple argument that
Socrates seemed to offer (that since sight is not coloured, it cannot see itself) is
undermined by a conception of sight broad enough to include higher-order attitudes.
And in that case the brutish view—which denies the possibility of higher-order
perception by insisting on the brute relation between perception and its object—is
ruled out.
17
ide at 167c4 appears emphatically at the beginning of the sentence.
18
Horas at 168e3 runs directly on from idêi at e1, where the latter is part of a move within the argument.
19
See n. 5; that this is an event is indicated by the imperative at 167c4.
Well, an objection might run, perhaps the double use of sight reveals only an
ambiguity in ‘sight’, or in ‘perception’ in general: sometimes it is used narrowly of
sense-perception, when it may be brutish, and sometimes of perceptions of a more
general, civilized, second-order kind. This passage, then, is but an instance of that
general contrast; if the seeing of the frame is civilized, the seeing attacked in the
argument is not. (And in this case the relation between frame and framed has little
impact on the arguments themselves.) But this objection may assume too much. For
it assumes that Plato must begin with, must indeed already have fixed on, a brutish
account of perception (sense-perception), and that he extends it to a metaphorical,
civilized use without thinking that the structure of the latter has any bearing on the
nature of the former. On that assumption this passage (again) is a literary accident
(and response 2) to the Relations Argument is the right one).
We need not suppose, however, that ‘perception’ here is equivocated; nor does the
composition of the passage, and its ostentatious double use of sight, encourage us to
do so (there is no literary accident here). Perhaps, instead, Plato starts from a broad
conception of perception, whose nature and delineations he is here trying to make
clear. His careful composition, then, is part of his examination of the difference
between a civilized and a brutish view of perception. Thus the frame invites us to
think of perception in general, and sight in particular, as capable of being higher
order, capable of having in its scope both the content of lower-order seeings and
those seeings themselves. The structure of the Relations Argument then allows him to
suggest that in fact perception may be civilized right through; all perceptions, on such
a view, will be broadly cognitive, whatever their order: ‘perception’ is not equivocated
at all. If that is what the Relations Argument offers, then the argument must be
understood in terms of the third option I offered above; and in that case the crucial
contrast here is not between the brute and the civilized, but between those states that
are psychological and those that are not.
Quite right, we of the twenty-first century might think: for the discussion of these
higher-order states—we might think—is exactly appropriate for cases—perception,
desire, knowledge—where we want to insist that there is something that it is like for
me to perceive.20 In just such cases, that is, the higher-order state (perceiving that we
perceive) is designed to capture the phenomenon of consciousness, awareness, the
state that is peculiarly a part of human psychology.21 Is this what Plato has in mind?
I think not: for these arguments about self-perception and self-knowledge do not give
him an account of consciousness, but instead a very different account of reflection.22
20
Nagel (1979).
21
This, e.g., is often taken to be the issue in the parallel passage of de anima; see e.g. Kosman (1975) on
‘perceiving that we perceive’ as a description of consciousness.
22
It would be a mistake, of course, to think that Plato’s lack of interest in the issues that provoke
contemporary discussions of consciousness (issues often dominated by the problem of scepticism) renders
what he does say about higher-order knowledge and perception redundant. It is a broader question than
5. Two Dramatic Incidents

Go back, once again, to the gilded frame: to those parts of the dialogue that are often
supposed to be irrelevant to the arguments contained therein, and to the elaborate
horseplay of the introduction. Two episodes are striking, once you start to read the
dialogue with, as I shall say, foresight (that is, at least twice).23 They are striking
enough, moreover, to ward off the conclusion that the frame drama has no philo-
sophical bearing on the arguments within.
Incident i. Charmides’ introspection
The dialogue is named after the beautiful young man, Charmides, who is apparently
sôphrôn in extraordinary measure—and under several different definitions. He is
well born and respectable (157e ff.). He is modest and unassuming (158c–d). And he
comes recommended (if such that be) by his mentor Critias, who insists that he is the
most sôphrôn of his generation (157d). Socrates seems impressed, and invites Char-
mides to supply him with an account of sôphrosunê for discussion. For:
s: ‘This is how I think it best to look at it. For it is clear that if sôphrosunê is present to you, you
will have some belief about it. For it is necessary, I suppose, that if it is in you, it will present
you with some perception24 from which some belief might be yours about it, what
sôphrosunê is and what sort of thing it is. Do you not think so?’
c: ‘I do think so.’ (158e–159a)
The same motif reappears a bit later:

‘So once again, Charmides,’ I said, ‘turn your mind to the matter even more, and look into
yourself, thinking about what sôphrosunê, by being present to you, makes you be like; and what
it is like, that it makes you so; working all these things out, tell me well and bravely what it
appears to you to be.’ (160d–e)
With foresight, this is all pretty odd. First of all, is the point a general one (anything
that is ‘present to us’ presents us with a perception from which we get an opinion), or
particular (when sôphrosunê is present to us, it presents us with a perception from
the one that occupies me here exactly what he supposes higher-order knowledge does for us; this is the
question that the last part of the Charmides seeks to address.
23
This is not the place to defend this approach in detail; it requires merely that the dialogues were
written carefully, to be read and reread, so that resonances whose significance appears only later in a
dialogue should not be disregarded.
24
aisthêsis: does this mean the faculty to perceive, or what is perceived? Opsis, like ‘sight’ in English, can
be both subjective and objective; and see Alc. 133a. For aisthêsis compare de an. 417a3–6 and Burnyeat’s
comment (2002), n. 32. If aisthêsis is the faculty of perception, and if it is then to be followed by a belief
(‘from which some belief ’), does that imply that the faculty of perception is brutish? Not necessarily: the
mode of perception’s presentation may be different from the mode of judgement, but still involve more
than merely a raw feel: see e.g. the account of judgement given by the soul’s silent dialogue, and also the
description of the conversation in the soul at Rep. 523 ff., where perception proffers a report; see Chapter 6.
which we get an opinion), or somewhere in between (some things, virtues for

example, when present in us, present us with a perception etc.)?
If it is general, why would we think it true? We might, of course, think that my
experiences are presented to me immediately; would we say the same of any other
inner states? Would Socrates say so, the Socrates who later worries lest his own
ignorance might escape his notice? Or is his thought that all my moral states are
transparent to me? If I am wicked, will my wickedness be ineluctably borne in on me?
Or is sôphrosunê in particular bound to present itself to its possessor? If sôphrosunê is
self-knowledge, this might be plausible—so long as this is not Critias’ model of self-
knowledge. For that takes it that our psychological states are each related to their
special objects intimately and exclusively: how then could Charmides’ inner inspec-
tion form the basis for a belief?
Incident ii. Looking inside Charmides’ cloak
Earlier in the dialogue, as Charmides approaches and tries to sit down, everyone who
is already seated tries to make space for him on the bench beside themselves; so one
person has to stand up, and another falls off the end of the bench. Socrates seems to
rise above the crude slapstick, remarking rather sourly that he will be impressed just
so long as Charmides has a beautiful soul (154e). But now when Charmides
approaches:
Then, my friend, I was straightaway at a loss, and my earlier boldness, in supposing that it
would be easy to have a conversation with him, was cut off; for when Critias said that it was
I who knew the drug [sc. a leaf to cure Charmides’ headache, which Socrates pretends to
have25], he looked at me with his eyes in the most extraordinary way, and made to ask a
question; and everyone in the palaestra flooded right round us. Then, my fine friend, I saw the
things inside his cloak; and I was on fire; and I was beside myself,26 and I thought that Cydias
was so wise about love, when he advised someone besotted by a beautiful young man, ‘to take
care lest he should come as a fawn up to a lion, and become his breakfast.’ (155c–d)
Three things, it is often assumed, happen to Socrates, all to show that he has
sôphrosunê, self-control (after a struggle): first, he sees Charmides close up, and
realizes that he is even prettier than he had seemed from afar; second, Charmides
‘looks at him with his eyes’; and third, Socrates sees inside Charmides’ cloak: Critias
25
The early action of the dialogue turns on a subterfuge suggested by Critias and agreed by Socrates:
that Socrates should pretend to have a cure for the headache from which Charmides is suffering. The leaf is
a fiction; but Socrates elaborates the subterfuge by saying that the leaf cannot be effective without some
incantations: and those just turn out to be dialectical discussion. The fact that Socrates seems to be lying
about the leaf is strikingly contrasted with what one might suppose to be the good faith of his emphasis on
philosophical discussion, and its bona fide connection with the priest-king Zalmoxis (see section 7). His
pretence is later alluded to when he suggests that the person with the skill of sôphrosunê should be able to
tell pretenders from those who are sincere (173a ff.). The leaf episode has the ambivalence that also seems
to characterize Socrates’ looking inside Charmides’ cloak.
26
En emautou: literally, ‘I was no longer inside my own [something or other]’. LSJ suggest ‘in my own
house’; the vagueness of the expression is telling, on the interpretation I offer here.
had promised that Charmides was even more beautiful naked, and Socrates has
helped himself to a view.27
But we might pause to notice that what Socrates has a view of is not fully explicit. It
is Plato’s readers who suppose that what Socrates sees are Charmides’ genitalia; what
Socrates says is that he saw what was inside Charmides’ cloak. But ‘being inside
Charmides’ cloak’ is transitive (as the Relations Argument may later allow us to
understand); it includes not only the most exciting parts of Charmides’ anatomy, but
his soul too, the part that Socrates says he is interested in. Our assumption that what
Socrates sees inside the cloak is physical rests on our further assumption that the
objects of sight are just what is physical; and the short passage describing Socrates’
discomfiture is emphatic that what we are dealing with here is perception—in
particular, sight. But is that Socrates’ assumption?
Compare the two episodes. They are, on scrutiny, pretty striking. Both turn on
something about perception (of a sort): Charmides’ perception is somehow ‘inner’; in
Socrates’ case it is vague just how far inside he sees. The language of sight is overlaid
by some other talk about ‘looking’:28 Charmides is to look inside himself; and
Socrates is overthrown by Charmides’ look. Both episodes, moreover, are somehow
about being (or failing to be) sôphrôn (when what that means is still undecided).
Again with foresight, they reflect on whether either Socrates or Charmides has the
virtue of self-knowledge, and on what it would be for them to have it.
But the perception they have (or are invited to have) should not be explained in
terms of the brutish claim, which isolates perceptions and their objects from other
reflective states. For both perceptions, on one account, turn out false: since Char-
mides is not in fact an exemplar of sôphrosunê. He does not know himself; he has no
understanding of whatever his own virtue might be; and the other descriptors of the
virtue (nobility, good manners) are left to one side: after all, they perform no
explanatory role in the psychology of virtue (this is emphasized by the injunction
to Charmides to look inside). So if either case of ‘looking inside’—Charmides’ or
Socrates’—is a perception of sôphrosunê (rather than, e.g., just an attempt at
perceiving, a looking-but-not-seeing), it is a mistaken one. If the perception is
mistaken, it cannot easily be understood as a direct relation between perceiver and
some object, since what is perceived is somehow or other false, or not there. The
relation between perception and what is perceived is here therefore a complicated
one, and so in some measure indirect. Neither case can be, straightforwardly,
brutish.29
27
Cf. e.g. Schmid (1998), 8, 91.
28
Alc. 132c ff. (and see Brunschwig (1973) and Denyer (2001) ad loc.) offers a different account of the
connection between sight, looking, and self-perception.
29
In part this point is about intentionality. If perception is intentional, it can be mistaken; but it cannot
then be brutish.
6. The Civilizing of Perception

The double use of sight at 167–9 together with these mistaken perceptions at the very
outset of the dialogue invite us to reject the brutish view of perception and, with it,
the bracketing of perception with the non-psychological relations like large and
double. Instead, we might press the connection made between perception and the
other psychological states: perhaps, that is, perception is here conceived as analogous
to belief and desire. After all, the Relations Argument itself is bracketed by two
examples of perception with higher-order content. Perception, then, like belief, is
somehow civilized.
What does it mean, then, for the object to be presented to perception, as Socrates
puts it to Charmides? Presentations are themselves cognitive (notice that at 169a
what is presented is unconvincingness); when the presentation is made to perception,
it is made by an appearance. That is why Charmides is encouraged to tell Socrates
how—on introspection—sôphrosunê appears to him to be (160e1); and how what he
sees at the end of the argument (168e4) is something that appears impossible.
Perception, then, is closely connected to appearance: to the way in which things
appear for the perceiver (the language of objects determining the cognitive state is
displaced).30 And appearance, we should notice, is both somehow partial (perception
is not the same thing as reflection or belief) and also somehow compelling: this is how
it really seems. That perception of this kind can be wrong is part of its seeming; but
that it seems so with such compulsion is why it seems so real.
How, then, is perception distinct from belief? The difference lies in part in the
power it has (we may call this its attitude) to its objects. Belief does believing towards
its objects: I believe that the moon is made of blue cheese. Desire, equally, does
desiring towards its objects: I wish that I had finished this paper, I wish he were
mine—I wish that the moon were made of blue cheese. By parity, perception does
perceiving towards its objects. What does that mean? Well, on the evidence of what
Socrates says to Charmides, it reports the objects as they are presented. We could
describe this in terms of how the object is seen (I see an argument as valid; I see
Charmides as having an intemperate soul), or in terms of the content of the report (I
see that the argument is valid, I see that Charmides has an intemperate soul). The
report, therefore, is a complex one, taking in the whole of the presentation; it forms
the basis for a belief; and the belief in turn can come under the scrutiny of reason or
thought.31 These attitudes are civilized throughout.
30
Plato is not worried here, I think, about the subjective nature of appearances (and so, e.g., about the
problem of explaining the subjective in terms of objective facts of the matter). Instead, he is interested in
the perspective imposed by appearance, and thence by the partialness of it: on this see the argument of Phd.
74 ff., where the problem of appearance is not that things appear thus and so (but aren’t, really), but rather
that they appear thus and so, and are; and also appear differently, and are.
31
That perception reports to thought is a theme of the discussion of thought at Rep. 523–5.
On such an account, furthermore, the gap we might postulate between sense-

perception and introspection will narrow: they are both, somehow, of how things
potently seen (it is not the case that inner seeing is somehow metaphorical, or that the
canonical case is what we would call physical32). But this narrowing of the gap,
legitimated by a civilized account of perception, will further resist the thought that
either perception or introspection is somehow a matter of the perceiver being purely
affected by the object perceived (this is the model of the Relations Argument; and it is
rejected, I am suggesting, by the frame dialogue). When Charmides looks inside
himself, and when Socrates looks inside Charmides’ cloak, what is in there provides
them with the faculty to perceive; but it does not constitute the seeing. The soul is no
more a theatre than is the outside world.
In that case, then, there is both continuity and discontinuity between the attitude
that is perception and the attitude that is belief. They are continuous by being
complex in content, civilized not brutish, susceptible to higher-order attitudes of
the same type. They are discontinuous, because the attitude of perceiving and
reporting one’s appearance is quite different from the belief based upon it. This
ensures that the two cognitive states neither collapse into each other nor are
completely detached from each other: there may be, just as Plato describes, a complex
set of relations between belief and perception, inquiry and reflection, even desire: just
as we find instantiated in the frame dialogue. And it is the frame dialogue that allows
us to see that this is so.
7. Self-Knowledge and Knowledge of the Self

Return now to the dialectical structure of the dialogue. The Relations Argument can
be resisted if perception is not brutish; and the civilized model of perception is to be
found in the assumptions of the frame dialogue. What is more, if perception is
civilized, there is no reason why it could not be higher order: no reason why I should
not both perceive something and perceive that I perceive it, without risk that the
higher-order attitude turns out either vacuous or absurd. For the content of the first-
order attitude may be included in the content of the higher-order one; there is
nothing about complex content that demands the intransitivity of relations like
double or large.
We might, however, find there to be something rather unexpected about this
account of perceiving that we perceive. If Socrates does indeed see something
about the inner Charmides, what he sees is wrong. When Charmides offers an
opinion derived from his inner looking, that opinion too is false. So neither seeing
is privileged; and inner seeing is not private, either, since Socrates sees inside
32
See section 7 on the holism of Zalmoxis; I am grateful to Malcolm Schofield for making me see that
holism would encourage the view that there is no standardly dualist view of the relation between the
physical and the psychological in play here.
Charmides’ cloak (rather than his own). If—on this account—we perceive that we
perceive, the higher-order perception is not going to generate anything at all like our
own idea of consciousness.33 Higher-order attitudes, then, must be doing some work
here other than expressing the incontrovertible view of our own inner states, or an
awareness that we are feeling, thinking, perceiving thus or so.
Consciousness, I think, has never been the issue here. Instead, the Relations
Argument was interested in reflexivity, and in how knowledge could be self-know-
ledge. But that interest, of course, was the result of a muddle: for the original project
to explain knowledge of the self was replaced by a discussion of self-knowledge,
which was revealed in turn to be impossibly reflexive.34 The muddle is explicit in the
text. For Socrates expressly (169e) puts aside any objection to treating knowledge of
the self and self-knowledge as the same; his recusatio serves to emphasize the
possibility that they are not. Now perhaps the reason for this crabwise approach to
knowledge of the self may turn out to be just what is left over from the Relations
Argument: the possibility of higher-order thought. For, I suggest, the target all along
has been the connection not between self-knowledge and reflexivity (for this is
rejected by the Relations Argument), but between knowledge of the self and
higher-order thought.35 That connection has been left untouched: but its importance
is emphasized in the frame dialogue. The account we must give of the argument
about self-knowledge, therefore, will be incomplete unless we read it within its frame:
for it is the frame that supplies its interpretation.
If the dialogue offers an account of perception as civilized, and also suggests that
there are significant analogies between perception and knowledge, then knowledge,
too, may be construed in a civilized manner. If knowledge is civilized, it is to be
understood not as a brute relation between a power and its objects, but as an attitude
with complex content; and this allows for the possibility of a higher-order knowledge,
which can embed the content of the first-order attitude in its content (content is not
intransitive). What, then, are we to say of knowledge of the self ? It should be no
longer brutishly reflexive, for sure, and so no longer paradoxical in the manner the
Relations Argument urged. Instead, because knowledge may be both higher order
and reflective, we may know what we know and do not know without this kind of
knowledge being empty or absurd. This reflective attitude may still be construed as
33
Or, better, our ideas of consciousness: see e.g. Tye (1995), ch. 1, on the contrast between phenomenal
consciousness (what it is like to feel pain, etc.) and the sort of consciousness that occurs via introspection of
my own inner (mental) states—some kind of self-consciousness, perhaps, as in Locke, Essay, II. 1. 19, or as
discussed by Ryle (1949), ch. 1. Neither phenomenal consciousness nor self-consciousness can be Plato’s
targets here; the first because it is private, the second because it is exclusively mine. There is a great deal
more to be said about just how Plato treats these issues, but not here.
34
The reference of ‘self ’ in reflexive expressions is always tricky: a sign on the tables in the University
Library in Cambridge announces them to be ‘self-clearing’.
35
This is why transitivity is one of the features of relations in which Socrates is interested in the
Relations Argument.
knowledge of knowledge (by virtue of its higher-order component): how exactly is it

knowledge of the self?
That knowledge of the self may be construed as knowledge of knowledge is
congenial to an intellectualist account of Socratic inquiry. Especially, this reflective
knowledge will include other knowledges within its scope: it will, therefore, be a
suitable condition for a complex investigation into what one does and does not
know—this kind of knowledge, therefore, will have a holistic cast. That thought, in
turn, would be welcome to one of the heroes of the dialogue: the Thracian priest-king
Zalmoxis, who has a holistic approach to medicine (you can’t fix someone’s headache
without fixing their whole soul), and who supposes, further, that your soul is the
source of value for the rest of you. Socrates reports him as having said:
As you should not try to cure the eyes without the head, nor the head without the body,
likewise you should not try to cure the body without the soul. But this is the reason why the
Greek doctors fail to cure many diseases, because they care nothing for the whole, which
should be cared for, since while the whole is not well, it is impossible for the part to be so. For
everything arises from the soul: all good and evil for the body and the whole person; and it
flows out thence as if from the head to the eyes. (156e)
But the combination of Socratic higher-order thought and Zalmoxis’ holism has a
rather uneasy consequence. For notice the Heraclitean tone of Socrates’ account of
inquiry at 166d1: he is searching for himself, but there is no guarantee he will find
him.36 Fair enough: for if knowledge is holistic, it may seem that it needs to be
acquired as a whole; and that, perhaps, is an extraordinarily hard task, if not an
impossible one. Further, if knowledge of the self is knowledge of knowledge, the self
that is sought may be so complex that the search will never end: the self as a whole is
elusive, just because knowledge itself is (well-nigh) impossible to complete (and
Socrates is dead). Plato would reject as spurious any puzzle that makes the self
elusive on the basis of reflexivity;37 he is right, I think, to observe that the real
problem of the elusive self is that reflection on its limits never ends.38
36
Compare the journeying metaphor in Heraclitus: ‘I sought out myself ’ (DK22B101), or ‘Going to the
limits of the soul, you would not find them, even if you travelled every road; it has such a deep account’
(DK22B45).
37
See e.g. Hume (1978), 252.
38
It is an honour, a pleasure, and a matter of considerable trepidation to offer this chapter to Myles
Burnyeat, an incomparable reader of Plato. I am grateful to various audiences (at King’s College London, at
the Moral Sciences Club in Cambridge, and at Rice University) and to various individuals (especially
Simon Blackburn, Nick Denyer, Hilary Mackie, Hugh Mellor, Vasilis Politis, Malcolm Schofield, Dominic
Scott, Frisbee Sheffield, Nick Smith, and Harvey Yunis) for their comments, critical and otherwise. My
thanks, especially, to my colleagues Verity Harte and Peter Adamson for a discussion of the Charmides
which has gone on for several years. I am also extremely grateful to the Leverhulme Trust, who funded the
Major Research Fellowship during the tenure of which I wrote this chapter.
10
Escaping One’s Own Notice
Knowing: Meno’s Paradox Again
1. Meno’s Paradox: The State of Play

Should I apologize for coming back yet again to something that has puzzled me for
forty years, back to Meno’s paradox? My excuse is that the puzzle itself is a good one;
and Plato’s presentation of it cleverer, and more troubling, than is often thought.
The recent literature is (rightly) dominated by Gail Fine1 and Dominic Scott;2 part
of my purpose is to get clear just why I am still unsatisfied by what they say. Consider
these dominant claims in each of their accounts:
a. Fine: the theory of recollection, answering the paradox, does not take us to
have latent knowledge all the time after we are born; but that it explains how we
‘have and rely on’ the true beliefs with which we can embark on, and conduct,
inquiry.3
I ask: What is it thus to ‘rely on’ a true belief ? Is such reliance enough to explain how
we come to inquire?
b. Scott: what really matters about Meno’s paradox is not how we begin inquiry,
but how we finish: the issue of real interest is discovery. The theory of recollection
responds to the paradox with an account of the knowledge on which discovery
depends.4
I ask: How does the elaborate treatment of the paradox fare, if Plato’s primary
interest is discovery?
To explain inquiry we do indeed need to know how we progress with it; and how
we might be able to understand its end, the discovery.5 But to explain inquiry we need
also to understand how searching was what we started to do, especially if under-
standing is where we end.6 After all, I may have all sorts of beliefs about all sorts of
1 2
Fine (1992), (2007), (2010). Scott (1995), (2006).
3 4
Fine (2010). Scott (2006), 70 ff.
5
Especially if we have an antecedent commitment to some principle that knowledge is always built on
knowledge: some ‘foreknowledge principle’ is certainly at issue here. See Scott (2006), 84 ff.; and Fine (2007).
6
On this, see Nehamas (1985).
ESCAPING ONE ’ S OWN NOTICE KNOWING 191
things; but I shall inquire only when I think there is a question to ask. Part of what
needs explaining, that is to say, is the interrogative feature of inquiry, the way in
which, about some belief that I hold, I take it to pose some question. Merely having
the belief is not enough to account for the interrogative.
There is material in the way Plato presents the paradox that supports both Fine’s
interest in true beliefs and Scott’s interest in discovery. But that material, taken
together, shows up something important here that is often missed. For there is
something complicated going on from the very start. So I shall argue that all of the
presentation of the paradox is important, since it points to two different aspects of
the puzzle(s) about inquiry—what I shall (without epistemological prejudice) call the
external aspect, that which concerns itself with the objects of inquiry; and what I shall
call (ditto) the internal aspect, focusing attention on the states of mind of the person
who does the inquiring.7 If this is right, does the theory of recollection account for
both aspects? That this is where Plato turned his critical eye is, I shall argue, made
manifest by the theme and variations on Meno’s puzzle that we find in the Euthy-
demus. Central to the Euthydemus’ discussion is Socrates’ astonishment that he
might ‘escape his own notice’ knowing, or being wise.8 I shall wonder what sort of
a condition on knowing, or being wise, this might suggest, and invite the conclusion
that Plato is rejecting the account of knowledge urged on Socrates by the sophists: an
account of knowledge that relies on the external aspect.
2. Preliminaries to the Paradox

What is it that makes the paradox puzzling? I shall take it to be a constraint on its
interpretation that it poses some—at least prima facie—puzzle. The crucial passage
is this:
Meno: (1) And in what way will you seek, Socrates, that which9 you do not know at all what it
is? What sort of one of the things that you do not know will you put up and then seek? (2) Or,
no matter how much you come across it, how will you know that this is what you didn’t know?
Soc: I understand what you want to say, Meno. Do you see about this that you are fishing up10
an eristic11 argument; that, indeed, it is not possible for a man to seek either what12 he knows
or what he does not know? (1) For he would not seek what he knows—for he knows it, and for
7
I hope not to bring in here any particular theoretical assumptions, or to beg questions about
externalism and internalism. But on the contemporary debate, see e.g. Kornblith, ed. (2001).
8
This expression is picked up by Aristotle, An.Po. 2.19, 99b22–27.
9
Throughout Meno uses the construction of pronoun (‘that’, touto) + relative clause (at 80d5, 6, 7,
and 8); my clumping translation is intended to bring this feature out. Socrates’ version is more elliptical;
see n. 12.
10
See Scott (2006), 78, n. 5, on the metaphors here.
11
This is commonly translated as ‘contentious’; in what follows I suggest that there is an express
connection being made here with sophistic argument.
12
Throughout this speech Socrates elides the antecedent into the relative pronouns at 80e3, 4, and 5.
This allows for the possibility that the last clause is not a relative clause (‘he does not know that which he
192 ESCAPING ONE ’ S OWN NOTICE KNOWING
such a person there is no need of seeking—(2) nor what he does not know—for he does not
know what he will seek. (Meno 80d5–e5, my numbering)
Is there just one puzzle here, in two different versions—Meno’s (loose) and
Socrates’ (formalized)? But neither version replicates the other. Meno misses Socra-
tes’ point that we don’t inquire into what we know, while Socrates misses Meno’s
interest in what happens when we reach what we seek (so this may turn on differing
interests in inquiry—the beginning of the search—and discovery—the end). The two
versions generate a composite puzzle, with three limbs:
1. (Meno 1 and Socrates 2) No inquiry into what we do not know.
2. (Socrates 1) No inquiry into what we do know.
3. (Meno 2) No discovery of what we do not know.
For completeness we may supply a fourth:
4. [No discovery of what we do know.]13
This larger puzzle has pretensions to being exhaustive, via Socrates’ premise: for
anything we might seek, either we know it or not. And thence it has claims to formal
validity; but perhaps it is not after all well formed.
Consider the shape and structure of the puzzle overall. First of all, does it risk
equivocation on ‘know’? Suppose that ‘to know’ means ‘to have in mind’ (as seems
plausible for Meno’s first limb). If, when I don’t know something, I don’t have it in
mind at all, then it seems entirely plausible to deny that I can inquire into what
I don’t know; for I cannot even articulate the first question. But it does not seem at all
plausible also to deny that I have no need to inquire into what I do know, if that just
involves my having it in mind. Contrariwise, suppose that ‘know’ means ‘really really
know’ (as seems plausible for Socrates’ first limb). Then it sounds right to say that
I shan’t need to inquire into what I really really know; but quite wrong to infer that
I shan’t inquire into what I do not really really know.
To fix this, do we need to say something about what we might think of as the
puzzle’s theory of knowledge: that knowing is ‘an-all-or-nothing’ affair? That would
fix the logic; but it would demand further and better particulars from our epistem-
ology. Why would we think that knowledge is ‘all-or-nothing’? Would that be
because we construe knowledge as acquaintance? Or because we think that know-
ledge is holistic, perhaps, so that we either know everything or nothing?14 In any case,
how are we to explain not only how we reach knowledge, but also how we set out to
acquire it—how we come to ask the first question, whence comes the interrogative?
will seek’) but an indirect question (‘he does not know what he will seek’, reporting the direct question:
‘what shall I seek?’).
13
The theory of recollection may be designed to deny this limb.
14
Acquaintance is at issue at Meno 71b; arguably, a holistic conception of knowledge awaits with the
insistence on explanation or reasoning at 98a.
What is more, the detail of the way the puzzle is put might discourage some ‘all-or-
nothing’ account. For although Meno’s way of putting the puzzle might invite a
model of knowledge as access to some object, or some totality, that way of thinking
about things is, I shall suggest, starkly contrasted with Socrates’. That calls attention
to the difference between what Meno represents as extremes (knowing absolutely
versus not knowing at all) and what Socrates reduces to a bland disjunction. Does it
matter how the puzzle is put?
3. Reading the Puzzle

Why are there two versions of the puzzle, Meno’s and Socrates’? Was Plato just in a
muddle (and thought them roughly the same)? Does what Plato thought matter, just
so long as we can recreate a good puzzle from the mess? I shall say that he wasn’t in a
muddle at all, and that the good puzzle is his; although it is not clear to me that he has
a definitive reply.
Suppose—if only for charity—that the differences between the two versions do
matter, that they were put like this a-purpose. Then there would be two different
puzzles:
Meno 1 You cannot inquire into that which you don’t know at all.
Meno 2 You cannot discover that which you didn’t know in the first place.
This first puzzle is missing a limb (about inquiring into what you do know) that is
present in the second. The second:
Socrates 1 You cannot inquire into what you know.
Socrates 2 You cannot inquire into what you do not know.
misses a limb about discovery (although there is one such present in the first).
Now consider the three limbs that seem at first concerned with the beginnings,
with inquiry, Meno 1 and Socrates 1 and 2. Meno 1 trades on the idea that to inquire
I must be able, somehow, to specify what I am inquiring about: without that, how
could this unknown object of inquiry be present to the mind of the inquirer at all?15
For this to be puzzling at all, the stipulation that it is unknown must be that it is not
even present under the description that the inquirer does not know it.16 So the
(impossible) inquiry has narrow scope:
Meno 1 You cannot inquire into [that] which you don’t know at all.
15
Fine (2007), 340, n. 23, suggests that the qualification ‘at all’ merely modifies the knowledge (we don’t
know it, although we could be in all sorts of other states of mind about it) and does not suggest a ‘cognitive
blank’. But it is the cognitive blank—or perhaps the psychological blank, if the interrogative is the
problem—which poses the puzzle, whatever its solution.
16
This is the point, I take it, of ‘that which you do not know at all what it is?’ at 80d6; and of Meno’s
suggestion that we can’t even ‘put up’ what we don’t know for inquiry, prothemenos at 80d7.
where the intentional object of the inquiry is marked by [ . . . ]. Meno’s first limb, that
is to say, focuses on the object of inquiry, not on the description under which we
(don’t) know it.17
Socrates’ first limb, on the other hand, works differently.
Socrates 1 You cannot inquire into what you know.
If this had the same structure and scope, the expression ‘what you know’ would not
be part of the description of the inquiry (‘what’ is elliptical for ‘that which’ and the
relative clause would be outside the intentional scope, on the model of Meno 1). But
that would make little sense of Socrates’ explanation for why we don’t inquire into
it—‘for he18 knows it, and for such a person there is no need of seeking’. Here ‘for
such a person’ allows us to express his reasoning—that since this is something he
knows, he has no need to inquire. But if it is his reasoning, it must be something he
has in mind when he thinks about what he should seek.19 So for Socrates the
intentional scope of the inquiry includes the claim that the thing in question is
something the inquirer already knows. Thus:
Socrates 1 You cannot inquire into [what you know].
Part of the content of the reasoning (not to inquire) is reflection on the agent’s own
cognitive state.20 This limb requires, then, not merely external but also internal
conditions to make its puzzle felt.
Meno 2 is not replicated in Socrates’ formulation of the puzzle. Asking ‘no matter
how much you come across it, how will you know that this is what you didn’t know?’,
Meno wonders about discovery: even if we come across the object, how are we to
recognize it, since we did not have it in mind in the first place? Suppose that the
specification of what you seek were in your mind while you look, so that your search
would be described by you as ‘looking for this thing which I do not know’. In that
case, there would be nothing puzzling about the discovery (unless we add a great deal
more complex psychology). I could say to myself, ‘I wonder what is the road to
Larissa’ and find it, and say ‘Oh yes, this is the road to Larissa, just what I was looking
for’. But that interpretation of Meno 2 eliminates any initial puzzle: and that militates
against the interpretation. Instead, Meno 2, about discovery, assumes that the object
of the search is completely absent from the mind of the searcher. All ‘coming across
things’, then, will be serendipitous; no discovery can be the recognition of the object
17 That this is Meno’s focus is confirmed by the grammatical structure of his version of the puzzle: the
object is expressed in the pronoun ‘that’; see n. 9.

18
The pronouns are gendered to follow the Greek.
19
That this is Socrates’ focus is confirmed by the grammatical structure of his version, where the object
of the knowing and the inquiring is expressed either by the relative clause (without the antecedent
pronoun), ‘what you know’ or by an indirect question, ‘he does not know what he will seek’; see n. 12.
20
This reasoning is done by the agent himself: hence the expression ‘for such a person’ expresses the
thought that this is a reason for him.
of my search, because the object itself can’t figure in the description of the search.
This, on Meno’s account, seems to make it not discovery at all. So the intention has
narrow scope:
Meno 2 You cannot discover [that] which you didn’t know in the first place.
Meno again focuses on the object, not on its description as something I do not know.
What about Socrates 2, the limb that corresponds to Meno 1? Socrates says: ‘he
does not know what he will seek’, and seems thus to reverse Meno’s point (‘in what
way will you seek that which you do not know at all what it is?’). Socrates’ interest,
that is, seems to be in how the inquirer puts the matter to himself,21 in circumstances
where the object of his search is not something he knows: his problem is knowing
what he is looking for, not looking for what he knows. So Socrates 2 again has broad
intentional scope:
Socrates 2 You cannot inquire into [what you do not know].
What would be Socrates’ correlate of Meno 2, if Socrates’ interest is in how
I represent my search to myself? Socrates should press how I understand what
I am doing—how I describe my success in finding the thing I sought (if, per
impossibile, I could even start to do so). So the puzzle would arise because I could
not know that this was what I didn’t know, and it focuses on how I have any kind of
grasp of what I do and do not know. But if the stipulation about what I do or do not
know appears in the formulation of the inquirer’s intentions, on Socrates’ version of
the puzzle my grasp of any object I might discover must be second order: grasp of
what I do or do not know, as such.22
If that is right, then Meno’s puzzle and Socrates’ puzzle are significantly different,
however we may gloss the sense of ‘know’. Meno attends to the external conditions of
knowledge, to how the object of inquiry comes into my purview at all. Socrates
attends more particularly to how my own cognitive grip on the object of inquiry can
figure in the inquiry itself. But in that case, we should perhaps refrain from assuming
that somehow Socrates revives and replaces Meno’s account: instead, the two ver-
sions may complement each other, neither sufficient without the other. That would
make sense, perhaps, of Socrates’ earlier remark, that he is keen for them to be joint
inquirers into virtue. And it would prefigure the discussion of learning and recollec-
tion with two separate conditions: an external one (about the object of inquiry) and
an internal one (about the reflective cognitive state of the inquirer; on how something
21
‘For he does not know what he will seek’ could be construed as an ellipse (‘he does not know that
thing which he will seek’), which would allow the intentional object to have narrow scope, or as an indirect
question (‘he does not know what he will seek’, reporting that he does not know ‘what shall I seek?’), which
would give the impossible intention wide scope. The easier reading is the latter.
22
Notice Socrates’ connective remark, ‘I understand (manthanô) what you want to say, Meno’, 80e1:
both relevant to the discussion in hand, which will turn into a denial that ‘learning’ is as we normally
conceive it, and second order.
can be an object of inquiry for him—hence the formula at 80e4). Meno’s version
insists on the external conditions, Socrates’ on the internal ones.
How does this fit with the odd lacunae in each version of the paradox? Meno
mentions discovery, Socrates does not; Socrates mentions inquiry into what we do
know, while Meno does not. The composite puzzle seems, in terms of the contrast
between inquiry and discovery, to be asymmetrical.23 If, on the other hand, the
difference between the two versions is designed to bring out the contrast between the
external features of inquiry (emphasized by Meno’s interest in what we do not know,
in the object of inquiry) and the internal ones (emphasized by Socrates’ interest in
our own reflective condition), it is symmetrical. How might that help with under-
standing how the puzzle of the paradox might be solved?
4. Recollection and the Slave

If all of that is the problem, the paradox turns out to be a pretty comprehensive one:
not just Meno’s paradox, or just Socrates’, but Plato’s. If Plato’s paradox offers a
contrast between external and internal conditions on inquiry and discovery, it calls
attention to some genuine difficulties both for inquiry and discovery. For inquiry, the
problem may be not so much its process once it has begun, but how it might ever
start. How come we ever start to ask questions? This has both an external dimension
(how do we get hold of the object of an inquiry, when the inquiry has not yet begun?)
and an internal one (what is it to be puzzled, to be inquisitive, to be in an inquiring
frame of mind?). Likewise, discovery might need to account not only for what we
discover, but also for our own understanding that it has been discovered. Both
dimensions of Plato’s puzzle, I suggest, are important: even if the puzzle allows for
some inquiry once it has started, it still demands an account of both its beginning and
its end, in terms of both external and internal conditions.
How, then, does recollection fare as the answer? Well, recollection is a story, and
there is a lot of stuff about priests and priestesses, but it does have some promising
features. Remember remembering at Phaedo 72ff. Here Socrates’ elaborate descrip-
tion of the phenomenon of recollection24 reminds us that it has both a psychological
component and what we might think of as a factive one. The psychological part is
the feeling of remembering—its emotional, affective, phenomenological side (this
includes, I take it, both the business of being reminded by something of something
else quite different, and the nagging feeling such as Simmias evinces of not quite
being able to bring something to mind). The factive bit is where we get it right;
23
Even if one disagrees, as I do, with Scott’s verdict on the problem of inquiry itself; and even if one
agrees with Scott, as I do, that the problem of discovery is an important one for Plato.
24
Cebes talks about recollection in the context of a geometrical example: the reader recalls the Meno,
but Simmias ostentatiously forgets: we are invited to notice our own experience of remembering just when
the conditions for doing so are discussed.
remembering is of something we knew before, it captures the earlier memory right, or

it fails to be recollection at all. Both aspects of recollection are crucial to it; they
correspond, loosely at least, to the twin demands of Plato’s paradox—to both an
internal and an external dimension to the business of knowing and coming to know.
But, still, this is analogy, or worse still, myth—recollection does not, we might
reasonably complain, provide us with much epistemological cash.
Consider the discussion that Socrates has with Meno’s slave at Meno 82b–86c: a
discussion that is supposed, somehow, to exemplify what recollection promises, and
that gives a running commentary on the slave’s cognitive states. Does the investiga-
tion of the slave boy show how recollection solves either the problem of inquiry or the
problem of discovery?
Perhaps it does so by telling us that the slave’s inquiry can begin and proceed by
means of true beliefs; and by telling us that recollection is somehow or other their
source. Recollection, thus, in its factive aspect, gives us security in our investigations
(it cannot explain how we could ever be comprehensively wrong); and it allows us,
thereby, some content to the inquiry as we proceed. But the factive element is not
enough to explain why the slave is bothered when he gets it wrong (84a), or how he
may be anxious to extricate himself from his own muddle (84c).25 What is needed for
that—and what the example of the slave illustrates—is that the process of inquiry
goes on in the mind of the inquirer. Without these pressing doubts, worries,
questions, and inquiries, no matter how many true beliefs will be inert. That point,
indeed, is represented by26 the figure of Socrates, and his repeated questions not only
to the slave but also to Meno about how far he is allowing the slave to learn for
himself (at 84a–d; 85b–86c). This brings out not so much the factive side of
recollection as its phenomenal aspect. In seeking to explain how inquiry gets going
at all and how, once begun, it might come to an end, we need to say something about
the states of mind of the knower or the inquirer; without doing so, Plato’s paradox
still demands an answer (85c).27
In that case, the slave’s ordeal shows us, in a vivid way, the contrast between
external and internal conditions on knowledge: the contrast between what the boy
can see out there drawn in the sand and his own state of mind, as it moves from
confidence to puzzlement to progress in understanding something more about
triangles. If the point of the theory of recollection is to respond to the paradox by
insisting that we need both external and internal conditions on knowledge, or even
milder sorts of cognition, then the interrogation of the slave works pretty well to
25
Contrast Fine (2007), 342.
26
This may answer the complaint that Socrates’ protestations do not let him off the charge of leading
the witness.
27
However, if discovery is so hard that the gap between merely thinking or believing and knowing may
be a huge one, one that very few can cross, then it seems too airily resolved by the suggestion that asking the
slave the same questions over and over will tie his opinions down. Instead, the problem of discovery must
be accounted for in more complex ways than the theory of recollection suggests, as I argue further below.
show that; and it does so by means of the contrast between the slave’s answers to
Socrates’ questions and Socrates’ commentary on the slave’s state of mind. The
interrogation of the slave, that is, supports the suggestion that Plato’s paradox is
designed to call attention to both the internal and external conditions on knowledge;
and that the theory of recollection is designed to meet both.
5. Recollection and Reflection

A running theme of the Meno is how we are able confidently to distinguish
knowledge—knowledge fully fledged, really-got-there knowledge—from belief (espe-
cially when we get to the road to Larissa, 97a ff.). This question centres upon the end
of inquiry—on the nature of discovery, and what makes that knowledge. One answer
to it might be an externalist one: whether or not we can successfully distinguish it, it
is knowledge just if it is true and arrived at by reliable processes. Those processes
need not be reflective processes; the externalist knower could be un-self-conscious
about it without losing his claim to know. If this is what is going on here (if, that is,
Meno’s version of the paradox is all we need), then recollection’s elaborate features
seem lamentably beside the point.28
A different view would be that an externalist account just misses what is interesting
about knowledge—that it is, somehow or other, self-intimating. Knowing something
just is a different, a manifestly different, cognitive state from believing something; so
that knowledge is borne in on the knower—when she knows, in some sense or other
she knows that she knows. The theory of recollection might fit the demand for a
phenomenology of knowing, for a mark that knowing is what we have actually
managed. But there is, of course, at the centre of epistemology both ancient and
modern a dispute about how that works. What is it to know that I know; and why
would it matter? Is it explained by my having a feeling about it, whether that is a
feeling of puzzlement, or of satisfaction at the end of inquiry? Or does knowledge
need stricter conditions than that?
If Plato’s paradox—as a complex whole—asks about this feature of knowledge,
then perhaps recollection is designed to explain what it is to know that we know.29
But one might complain that this won’t work—because recollection seems to trade
on two different kinds of knowing (a self-conscious phenomenal sort—recollection—
and some real first-order knowing—what is recollected) whose difference the label of
‘recollection’ tends to obscure. Or one might complain—as many have done—that
recollection is otherwise regressive: not only must it explain how we learned before
28
One view of recollection is that it contrasts ‘conscious’ and ‘unconscious’ levels of knowledge; see
Fine (2007), 348. But this does not explain what it would be to have knowledge that is conscious, or
whether knowledge that is unconscious would count as knowledge at all.
29
I concede to Scott (1995), (2006) that discovery is important here; but I still maintain that inquiry
matters too. The point of the paradox, however, on the interpretation I offer here, is not restricted to either
inquiry or discovery.
what we only remember now, but it also needs to show us how we knew that we knew
before we now know that we know it. If this is resolved—as I long thought,30 but no
longer do—by supposing that there is an externalist account of the first principles of
knowledge (so that knowing that you know does not go all the way up the orders, but
is blocked by some unmediated knowing of something), then other problems follow
for Plato (for example, a thoroughgoing externalist epistemology causes notorious
problems in the Republic).31 If any of that is worth consideration—by us, or by
Plato—perhaps we might need to rethink what is going on when we reach the end of
inquiry, and know that we know.
At the same time, we need to explain what it would be to start and then to conduct
an inquiry—perhaps by means of true beliefs on which we rely. The problem here is
both how I might rely on such true beliefs and that inquiry cannot progress any old
how, as a matter of occurrent beliefs. In asking about asking (not just about
believing), Plato’s paradox raises a puzzle about the intentionality of inquiry; and
here, as in the problem of discovery, an externalist answer may not suffice. From the
start, Meno’s worry about inquiry is reflective:
For in truth I am numb in my soul and in my mouth [clearly not, at the second order], and I do
not have an answer to give you. And yet I have given many speeches about virtue, in front of
many people, and extremely well, as it seemed to me at the time; but now I am not able to say at
all32 what it is. (80a8–b4)
Even though Meno’s remarks are pervaded by talk about how he feels about the
phenomenology of aporia (and clearly enough there is in all of this something about
how all these cognitive states feel), the phenomenology does not exhaust the content
of these states. On the contrary, they are fully explicit, second-order reflections on the
failure of the first-order states to be true, or defensible, or even explicable. For true
beliefs, then, to underpin inquiry, the status of those beliefs must itself be under
scrutiny. But then how does inquiry start? How do we ever reach the stage of
explicitly relying on or wondering about beliefs we hold true? How would the theory
of recollection help with that?
6. Entr’acte
I have offered three suggestions so far:
• In Plato’s paradox there is a thoroughgoing contrast between the external
component of inquiry and discovery (the ‘object’ that is sought and found)
and the internal cognitive state of the person doing the seeking.
30 31
McCabe (1994a), 53 ff. See Chapter 6.
32
This expression recurs at 80d6: I take it that the echo is designed to connect the two passages, and to
make the earlier puzzling as a consequence of the paradox.
• The theory of recollection may respond to the demands of the paradox by

offering a contrast between the factive features of remembering (what the recall
is of ) and the phenomenal features (what it feels like to recall). In particular, the
phenomenology of recollection may respond to the puzzle about the interroga-
tive at the beginning of inquiry, as well as the sense of discovery at the end.
• However, these phenomenal features may not be enough to explain the reflect-
iveness demanded for both inquiry and discovery; for part of what the paradox
asks for, in both cases, is a higher-order, reflective dimension to our cognitive
processes.
All of this suggests that the very complexity of the paradox is central to its
seriousness. Perhaps then the Meno’s response to it, the theory of recollection, is a
serious answer to the puzzle, somehow or other, and correspondingly complex. But
Socrates complains that at least Meno’s version of the puzzle is eristic: does this mean
we should give it any head-room?
In answer to this question, I shall consider the locus classicus of Plato’s account of
eristic, the Euthydemus, with two preliminaries. First, there is plenty in the Euthydemus
that has affinities with other dialogues commonly designated ‘late’, especially the
Theaetetus and the Sophist, as well as the Parmenides. Although I do not wish to engage
in the battle about developmentalism here, it at least seems reasonable to suppose that
the Euthydemus was written (or at least supposed to be read) after the Meno. Second,
there is other evidence that the Euthydemus is peculiarly critical of the Meno (e.g. the
discussion of the things that are good ‘themselves in themselves’ at 278–81).33
In particular, I shall suggest that the Euthydemus offers a critical view of the eristic
features of the Meno paradox, and of the theory of recollection. There are three
crucial passages:
• 275d–278e: In the first sophistic engagement, Euthydemus and Dionysodorus
get young Cleinias to agree both that it is the wise—those who know—that learn,
and that it is those who are not wise or do not know who learn.
• 283e–288a: In the second sophistic engagement, the sophists give an account of
truth that precludes falsehood and contradiction.
• 293b–297b: In the third sophistic engagement, the sophists argue that Socrates
knows everything, always (if they so desire). It is on this passage that
I concentrate now.
7. Knowing Everything
At 293b Euthydemus gives Socrates a choice. Should he teach Socrates the knowledge
he has long sought, or should he show Socrates that he (Socrates) had this knowledge
33
See Chapter 12.
all along? Socrates professes himself amazed: certainly, he would like to be shown
that he already has the knowledge—after all, that would be far easier than learning it,
old as he is (293b).34 Thus the sophist invites Socrates to suppose that knowledge is
either present all along or acquired by learning; and Socrates seems to agree.35 There
follow three brisk arguments to the conclusions that Socrates knows everything and
that he knows everything all along, always:
Argument 1 (293b–d) If (as he agrees) there is something he knows, then Socrates
is a knower. Socrates qualifies: of the thing that he knows. It makes no difference, says
the sophist: if Socrates is a knower, then he knows everything. No, no, says Socrates,
there are many things that he does not know. But then he is not a knower. Socrates
qualifies again, of the things he does not know. No matter, insists the sophist, he is
still not a knower, whereas before he conceded that he was a knower. But he cannot
be both a knower and not a knower (via a principle of non-contradiction, 293b); so
either he knows everything, or nothing. If, then, he concedes that he knows some-
thing, then he must know everything (293d6).
Socrates appears to be smitten:
As the saying goes, a hit, a palpable hit! So then how do I know the knowledge that we seek?
Since indeed it is impossible that the same thing is and is not, if I know one thing, then I know
everything (for I could not be both knowing and unknowing at the same time). But if I know
everything, then I have that very knowledge which we seek. Is that what you are saying, and is
this your piece of wisdom? (293d2–5)
Argument 2 (293e–294e) Amid some squabbling about self-refutation (itself a

reflective relation), they all agree that everyone—just if they know one thing—
knows everything. Socrates now leads the questioning (so that the argument that
follows commits the sophists themselves to its conclusion36). If there is nothing that
the sophists do not know, was this always so? Is it just now that they know everything
(the opening discussion describes how they have learned new skills at 271–2) or have
they always done so? They reply that they have always done so. So they knew
everything since childhood—indeed, as soon as they were born?37 They agree that
they did—and Socrates and Ctesippus find this claim incredible.
Once again the argument is embedded in some wrangling about how to proceed;
and Euthydemus boasts that so long as Socrates is prepared to answer his questions,
34
The theme of Socrates’ antiquity, and whether he can learn anything at all, is thematic in the
Euthydemus from 272c ff.
35
Compare what Socrates says about the slave at Meno 85d9–10.
36
It is rare in the dialogue that the sophists are the respondents; usually it is the interlocutors who are
committed to the so-called sophisms.
37
At 294d9 euthus genomenoi recalls the other recollection passage, Phaedo 75b10, where we are said to
use our senses as soon as we are born. If there is a reference here to two major recollection passages, we may
suppose that the reference is one we should take up. Notice, too, that Phaedo 75e talks about the
equivalence of recollection and ‘what we call learning’. The question of nomenclature has been at issue
in the Euthydemus from the beginning, e.g. at 277d ff.
he will show that Socrates agrees with these astonishing things too. Socrates responds
that he would be delighted to be refuted in respect of these matters:
For if I have escaped my own notice38 being wise,39 and you demonstrate that I know everything
and always did, what greater piece of luck could I have for my whole life? (295a5–7)
Argument 3 (296a–d) Now Socrates agrees that when he knows, he knows with
something (his soul), and always with that same something; so since he knows
something, he always knows something, and always everything.40 Just as before,
the sophist proceeds by disallowing the importance of the qualifications of Socrates’
concessions, qualifications that reflect each time on the scope and duration of the
knowledge he does concede he has. From the sophists’ point of view, those are
irrelevant to an account of what Socrates’ knows; but without them, Euthydemus
takes himself to be allowed the following conclusion:
So indeed it is always that you have agreed that you know with that with which you know,
whether it’s whenever you know or however you like; for you have agreed always to know
everything at once. So clearly you knew when you were a child; and when you were born, and
when you were conceived; and before you were born, even before the heavens and the earth
came into being, you knew everything, since you always know. And, by Zeus, you yourself will
always know, and know everything, if I please. (296c–d)
Once again, the argument is punctuated by an increasingly tense interchange

between the sophists and Socrates and Ctesippus; and it is followed by a unique
collapse of the partnership of the sophists—Socrates asks them whether they know
that good men are unjust. The sophists are already committed to saying that
everything is true; so they should agree, and have no qualms at the manifest falsity
of what they are invited here to know.41 But Dionysodorus now hesitates, inclined to
deny it; and Euthydemus loses his temper, complaining that Dionysodorus will
commit himself to the implication that he is both a knower and not a knower at
once (297a). So neither Euthydemus nor Dionysodorus know what they know; they
merely agree to everything (or try to do so). This global acceptance is itself a mark of
what is going wrong here. But why should we think it matters?
38
The question about self-knowledge was set up at the end of the previous Socratic episode, 292d, where
the discussants find it hard to account for the content of reflexive knowledge.
39
The shift of terminology, from what I have translated as ‘knowledge’, epistêmê (with reservations:
better for many of these contexts would be ‘understanding’) to ‘wisdom’, sophia, is revealing; I shall return
below to the connection between these discussions and an account of virtue. The sophistic version is found
in their ‘piece of wisdom’ at 293d8.
40
Whatever else falls to the sophistic arguments, this surely caricatures a holistic account.
41
There is a parallel here with the way in which recollection may or may not figure in the elenchus: if we
recollect what we learn, are we always right? See Scott (1995), ch. 1, and the question of how the slave-boy
comes to make mistakes; Scott is surely right to suppose that one of the things that is of interest in the Meno
is the nature of critical reflection.
8. The Euthydemus and the Meno

First of all, notice how strong and how particular are the echoes of the Meno. Notice,
too, how they are distorted—perhaps by the sophistic context, or perhaps by the
critical relation between the Meno and the Euthydemus.
• The account of knowledge the sophists press on Socrates supposes that knowing
one thing implies knowing everything; compare Meno 81d. But what are we to
make of this in the context of sophistic argument?
• The shift they rely on between knowing something and being a knower is one
they have exploited already, in the first exchange with Cleinias back at 275 ff.:
this, translated by Socrates in terms of wisdom, recalls to us the Meno’s terms of
engagement: an account of virtue.
• The opening set of puzzles in the Euthydemus about knowing and learning
begins with the question: ‘Who learns, the wise or the ignorant?’ (275d) and
continues to the claim that both those who know and those who do not, learn
and do not learn. The terms of Socrates’ version of the paradox have been in the
background of the dialogue all along.
• The question of how knowledge is acquired (is it there all along, or learned de
novo?) picks up the paradox (and the Meno as a whole) especially in the context
of Socrates’ search, in the Euthydemus, for some knowledge.
• The conclusion—that we know always, before we are born, before even the
world was created—is phrased in Meno language (compare 81c–d): our know-
ledge is pre-natal, universal (knowledge of everything), and presented in a
grandiose cosmological context (it was, before the heavens and earth came
into being).
So are we to suppose that what we have in the Euthydemus just is the theory of
recollection? Is this now the province of sophists? Socrates’ response to the sophists’
promise unnervingly inverts his own lazy argument from the Meno (81d, 86b–c): if
they can show him that he knows what he seeks all along, he won’t need to bother to
learn it. On that choice, that is, knowledge is either such that it escapes our notice, or
it is learned. Socrates, worried about the effort of learning, goes for having had it all
along, without noticing it; and so will end up preferring having it pointed out to him
(that it had escaped his notice) to the labour of learning. How does this set of
arguments bear comparison with what he avows in the Meno, the hard struggle of
finding out? What is going on, when the Euthydemus asks us both to recall and to
recoil from the Meno’s account of learning?
9. Knowing, Learning, and Discovery

At the centre of the Euthydemus lies a dialectical exchange in which the sophists force
their interlocutors to agree that there is no such thing as falsehood; and that it is
impossible to contradict someone else (283–8). Their argument suggests an account

of truth in which statements are true just if they say the things they are ‘about’;
otherwise they do not say anything at all. This account of truth, if it is to imply both
that every statement is true and that contradiction is impossible, has truths as
piecemeal utterances, cognitive episodes if you like, which are disconnected from
each other and discrete. It precludes, therefore, truth-functional relations between
truths; and precludes also reflective relations between them (no truth can meaning-
fully fall in the scope of any other: Socrates’ version of the Meno paradox will
disappear).42 That directly rules out the significance of a reflective notion of know-
ledge: no claim to know can be falsified (because nothing is false) so that the higher-
order claim that I know what I know is—if it can make sense at all—empty. I may
know everything, since I can just keep knowing one thing after another, but there is
no room for the thought that I know that I know—or, more to the point perhaps,
when I know that I don’t. This account of knowledge is firmly external.
If we compare and contrast this with Socrates’ version of the Meno paradox, we
may see what is at stake here. Suppose we take the sophistic (i.e. forced by the
sophists on Socrates) account of knowledge to be objectionable, barren, a repudiation
of the difficult process of learning. Then, conversely, we might see that we would
want to include, as conditions on knowledge, those very things the sophists deny, or
claim to be unimportant:
• Knowledge is not merely of truths; knowledge can account for its field, how its
truths are collected together. Knowing everything is more than just knowing one
thing after another.
• Knowledge is reflective: it involves both first-order cognition and second know-
ledge of the first.
• This kind of reflectiveness, however, is not satisfied by a mere feeling that
I know. When I know that I know, my higher-order knowledge is not merely
a phenomenal concomitant of the first-order cognition, but an active reflective
stance. For:
• The reflective stance shows up in the intentional scope of what is known, at the
higher order:
○ knowing that I know (so knowing something about my own cognitive
condition)
○ knowing what I know (knowing the field of my cognition)
○ being able to give an account of my knowledge, to respond in question and
answer (knowing the explanatory structure of what I know).43
42
On the sophists on truth and falsehood, see Denyer (1991), McCabe (1998), (2000), ch. 2.
43
These conditions for knowledge are elaborated, and then come to grief, in the preceding Socratic
episode of the Euthydemus, 288d–293a.
• This kind of reflectiveness requires effort; it can’t just be delivered by means of

a dodgy sophistic argument (contrary to the claims of the sophists’ three
Arguments).
• And yet the surprise with which Socrates greets the possibility that he might
escape his own notice knowing suggests something more: that (in fact) it is an
absurd suggestion that we might escape our own notice knowing. Instead, then,
perhaps knowing that we know is a necessary condition of knowledge (or, better,
of understanding).
Now all of those conditions for knowledge might make us think quite hard about
just what the Meno might seek, and the Euthydemus may (ostentatiously) miss. And
the cross-references between the two dialogues might encourage us to think that
there is a critical relation between them. After all, the Euthydemus seems to give us an
exhaustive account of what it might be to call an argument eristic (as Socrates
complains of Meno’s version of the paradox, 80e2). And the shocking prospect of
Socrates’ taking the easy route to knowledge in the Euthydemus reminds us of his
own complaint that the eristic argument encourages us to be lazy (81d6). But now
these echoes from one dialogue to the other focus our attention on the deficiencies of
an external account of knowledge, just because they involve a conception of inquiry
and discovery without reflective features. The lazy view of learning trades on an
externalist account, and invites scrutiny of what it would (and would not) be to
escape one’s own notice knowing.
Suppose, on the one hand, that knowing that I know is a condition for my knowing
at all; and suppose, further, that knowledge can come relatively easily: that there are
plenty of things that I know. In that case, we might think that the knowing-that-I-
know condition must be a low one—mere awareness that I know, nothing very
complicated (and in that case, we might further think, the higher-order knowledge is
rather a different beast than the lower: perhaps, on this account, the knowing-that-I-
know principle would be equivocated).44 This, we might think, the theory of recol-
lection could supply.
Suppose, on the other hand, that the knowing-that-I-know condition is a demand-
ing one, in which knowing that I know is tied to a difficult process of learning at the
first order: the sort of learning Socrates says he still struggles with, and the sort of
thing the sophists could not possibly provide. And suppose, further, that the higher-
order knowledge is knowledge of the same sort as the lower (‘know’ is not said in
many ways, even if at many levels). Then the higher-order condition will be as
difficult as the lower: the lazy arguments reflect a serious danger. But laziness is
wrong: the interest in inquiry is important because knowledge is something we aim
for, try hard to achieve, aspire towards. This aim will be satisfied not merely with a
44
I stick here to the laborious ‘knowing-that-I-know’ to avoid collapsing these puzzles into the
problems about knowledge involved in modern discussions of a ‘KK’ principle; see e.g. Williamson (2000).
feeling, not merely with the phenomenon of remembering, but with fully explicit
reflection. This has, for sure, considerable affinity with the process of becoming wise
and so virtuous; and that is a consequence of its internal condition. Indeed, if
knowing demands knowing what I know, it might be hard to give a satisfactory
account of higher-order knowing without a further condition that the lower-order
knowing is somehow or other complete (that its field is determinate, even if not
exhaustive). But this may lead us to think that knowledge is somehow holistic; and we
might then wonder just how much of a difference there might be between knowledge
and virtue. The demands of the Meno, on this account, would be met by an internalist
response to the paradox.
But this needs to be moderated. If the sophistic outcome is to be avoided, then
Socrates should not be counted as a knower just if the sophists please (296d). If the
condition on knowledge were merely internal, then we might think that knowing
could happen just when the sophists want it to happen, or just when someone had
knowing in mind. Of course this condition on knowing is by far too thin; and the
effect of the complex Meno paradox is to insist that an external condition—the
factive aspect of recollection—is necessary too. So knowledge, on this account,
demands the right objects (external) and the right state of mind (internal) before it
can evade the sophists’ trap.
10. The Meno Paradox Again

Suppose that I am right that the Euthydemus offers a critique of the Meno paradox;
and right in suggesting that we should read the latter as a rejection of a strictly
externalist account of knowledge. What should we say, now, about inquiry and
discovery? Does the contrast between the external and the internal aspects of the
puzzle help us with them? Where does recollection stand?
First, inquiry: how does this deeper account of what is involved in knowing (that it
has both internal and external conditions) help us with the problem of inquiry, of
asking a question to which we don’t know the answer? If we inquire because we ‘have
and rely on’ true beliefs, how do we come to have them, and what is it to rely on
them? This seems to involve both external and internal conditions, too. Suppose that
the beliefs are themselves acquired by external means (by empirical means, for
example); my relying on them must surely itself be a reflective feature of my
inquiring. And the paradox, if it bites at all, demands an account of how that reliance
could be achieved, without itself having been acquired by inquiry. So inquiry itself
looks regressive, if the paradox is to be answered at all. Does recollection stop the
regress? The Euthydemus suggests that it does not; for recollection as it is caricatured
in that dialogue can escape my notice, and thus fail as an internal condition of how
I come to inquire.
Second, discovery: the internal condition on knowing, that we know that we know,
seems right for discovery. However discovery may come about, perhaps a discovery
only counts as such when we realize that a discovery is what it is. So this condition
shows us what it would be to know, by insisting that part of it is to know that we
know. But there is, of course, a sting in this tail. Suppose that the choice Socrates is
offered by the sophists is a genuine one. Either there is some knowledge of which we
are just aware: part of its being knowledge is its self-intimating character. This will
accord with the phenomenology of the theory of recollection; and will not demand of
it that it account for the cognitive content of higher-order knowledge. But then the
answer to the part of the Meno paradox that concerns discovery will be objectionable:
this isn’t knowing that you have found what you sought, but feeling it. The phenom-
enology of recollection, in that case, will be inadequate to answer the puzzle: and the
sophists’ objection will be telling. Or, if we think that higher-order knowledge is
knowledge just as the first-order kind is, but with different (broader) content, then
doesn’t the paradox bite even deeper? If then we ask how we know that we know, in
the case where what we know is that we know that we know, the account will be
obviously regressive. This, indeed, may be the point of Socrates’ failure to deal with
discovery in his version of the paradox. Does Plato have an answer to that? Or did the
paradox puzzle him still?45
45
I am extremely grateful to audiences in Oxford, Pittsburgh, the University of Western Ontario, and
the Aristotelian Society for discussion, comments, and kindness; and especially grateful to Gail Fine,
Dominic Scott, Verity Harte, Raphael Woolf, Peter Adamson, and Margaret Atkins for many brilliant years
of discussion of Meno’s paradox. I should like to thank the Leverhulme Trust for the Major Research
Fellowship, during the tenure of which I did a good deal of the work for this chapter.
11
From the Cradle to the Cave: What
Happened to Self-Knowledge in the
Republic?
1. Two Oddities in the Republic

Some bits of the Republic are odder than others. Some are just odd, in themselves;
others acquire their oddity when we reflect on their place in the dialogue as a whole
or on their connections with other Platonic works. Start with two: the story of the
prisoners’ view of their own shadows in the cave; and the tale of the ‘amazing sophist’
in book ten, who holds up a mirror and creates a world. Their oddity, I shall argue,
calls attention to something that matters to Plato: the role we should give to our view
of ourselves in our account of knowledge. ‘Ourselves’ is what causes the trouble here;
but to deal with that we need to rethink—I shall suggest Plato to suggest—what it is
to have a view. In thinking about that, I shall wonder about three connected notions:
self-perception, self-reference, and self-knowledge; and I shall argue that Plato poses
puzzles about all of those notions to make his readers see better how knowledge
should be understood.
The central books of the Republic offer some kind of contrast between the objects
of knowledge and belief; and it is a common enough view that this contrast is both
exhaustive and exclusive.1 It is then a short step to supposing that the state of mind
that is knowledge (511d) is determined by its objects; knowledge, on that view, is
passive, a state of being affected by the reality of the intelligible.2 If the objects
themselves are then distinct from each other, knowledge may be correspondingly
simple or even piecemeal.
But that account of knowledge poses severe difficulties for the Republic itself. If
knowledge is like this, we know just if we encounter the right objects. How is that
encounter made to happen? The Republic suggests that we come to know because we
have been educated; how can education ensure such an encounter with reality? And
the Republic suggests that we come to know because we have become virtuous: how
1
See here most recently e.g. Gerson (2009), ch. 3, but contra e.g. Fine (2003).
2
Hence the expression tettera tauta pathêmata en têi psuchêi gignomena, 511d7.
FROM THE CRADLE TO THE CAVE 209
does virtue ensure an encounter with reality? What account is to be given of how
education, or virtue, put us in the right place at the right time to be affected by the
intelligibles? What are we to say, in all of this, about what goes on in the knower in
the process of coming to know?
In what follows, by way of an excursus in the direction of Hierocles the Stoic,
I shall suggest that the oddities on which I shall focus give us the beginnings of an
answer to this last question. For, I shall claim, they reveal two features of the Republic
account of coming to know that are located in the subject, rather than the object, of
knowledge. The first feature is a condition of self-knowledge: that it is essential to
knowledge that the knower knows herself to know. The second is an account of what
this self-knowledge would include: namely a perspective on what is known, where
that is understood as systematic, seen from a wide view of what is intelligible. This
complex account of knowledge, I think, shows how knowledge can be developed by
education; and it shows, too, how the conditions of knowledge allow the objects of
knowledge to extend more broadly than individual forms—this makes it at least
possible that the philosopher, returning to the cave, could have understanding of the
world around her.
2. The Cradle
First, the cradle: Hierocles the Stoic thought that the ability to perceive ourselves is
something with which we are born, something that comes with the cradle.3
. . . animals perceive their own parts and . . . this happens to them from the very beginning.
(Hierocles, Elements of Ethics 45 ff).4
He thus seeks to explain our basic adult nature (phusis) by thinking about the
capacities we must have at birth. For, he argues, we are born able to perceive not
only what is outside ourselves, but also (and equally) what is within.5 That this is so is
shown by the fact that animals know how to use their limbs—for locomotion and
defence—so that they must be capable (from birth) of proprioception, the immediate
perception of the inner disposition of their bodily parts.
3 4
See here Brunschwig (1986). Ramelli/Konstan (2009).
5
For this argument to work, the perceptions of what is outside and those of what is within must be
‘perception’ of the same sort. Suppose we take the perception of what is outside to be objective perception;
that is to say, perception that is determined primarily by what it is of; then so too will inner perception be
objective. Or, if we think of all perception primarily in terms of perceptual awareness, then we may likewise
think of proprioception as a kind of awareness. But the use of ‘proprioception’ is tricky: it regularly refers to
the information received by inner perception, and to the reception of that information, and then often also
to our awareness of that information (even when this is very low-grade awareness of our physical
dispositions); see here e.g. discussion in O’Shaughnessy (2003). In what follows I seek a middle ground
between objective perception and subjective awareness, namely the subject’s own perspective on her
perceiving.
210 FROM THE CRADLE TO THE CAVE
Proprioception seems thus to be explained on the model of our perception of

external objects: so as objective perception, rather than subjective consciousness. Still,
proprioception is also somehow perception of the perceiving subject—it is, under
some description, reflexive; it is self-perception.6 But there is a problem here (I shall
call it the problem of reflexivity): surely the agent of some action cannot also be the
patient, as such? Surely there are no self-movers, strictly speaking?7 We might insist
that if something is said to move itself, or act on itself, it does not do so in respect of
the same parts of itself. If perception is such an action, then how can we self-perceive,
in anything like a strict sense? (Plato might ask the same question: see Charmides 167
ff.;8 Republic 430 ff.)
Proprioception, however, does not fall to that objection, at least on Hierocles’
account—for here what is perceived is some part of the subject, where the perceived
part and the perceiving part are not the same. Since the same account applies to the
perception of ‘the things outside’ and of the things that are ‘one’s own’, this suggests
that proprioception is the grasp of some inner object of perception: one’s bodily
parts, or one’s disposition. Indeed, one can show that proprioception occurs by
seeing that animals ‘grasp’ that they have the equipment to defend themselves against
predators: so the internal and the external are parallel, and both seem to be cases of
objective perception. So Hierocles escapes the problem of reflexivity by indirection.
But then why is proprioception so important, since the same task could be done by
ordinary perception?9
In the case of proprioception, by contrast to ordinary perception, we have also an
ineliminable reference to the subject: the animal sees its own parts as its own: for that
is the basis for its self-defence. This kind of self-perception is limited, as befits its role
in the cradle where our moral development may start; but it is designed to give to the
young of any species some mediated grasp of themselves as the subjects of their
cognitive processes. For Hierocles’ cradle argument is the basis for his general theory
of oikeiôsis, which is an injunction to broaden one’s sense of oneself. This normative
theory10 is incompatible with treating proprioception as merely a matter of subjective
consciousness; it does admit, however, of thinking about a richer version of subjec-
tive perception, where the perspective of the subject is in view. For, Hierocles
famously insists, we should extend our perception of our own parts outwards,
coming to consider as our own (as parts of ourselves, or as ourselves, simpliciter)11
first our nearest and dearest and then our neighbours and fellow citizens, as far out as
the furthest Mysian. It is only when this process of oikeiôsis is complete that the Stoic
6 7 8
See Chapter 16. See Furley’s classic (1978). See Chapter 9.
9
My thanks to David Papineau for discussion here; this is a promissory note for a deeper account of
proprioception at a later date.
10
As I argue in McCabe (2013) and in Chapter 16.
11
Do I perceive myself and others as parts of me (as mine) or as being me (as me)? Does this give me an
account of myself as a perceiving subject to found my moral development? I have discussed the different
versions of oikeiôsis in McCabe (2005) and (2013).
sage may be thought to have reached virtue and self-knowledge. Self-perception,

thus, is to be understood on a normative continuum from the cradle on. Our
capacities and our limitations from birth determine what, or who, we should become.
3. Shadows in the Cave

Turn back now to Plato, and the cave and its prisoners. At the beginning of Republic
7 Socrates says:
Next compare our own natures to the following effect in respect of education and lack of
education. (514a1–2)
What follows seeks to explain something about ‘our natures’, in the context of the
discussion of education (cf. 515c5); and the cave starts with childhood (514a5). The
cave uses a thought experiment to describe human nature (‘us’) in general,12 and our
potential for education. So the cave matches the running theme of the central books
of the Republic: the development of the philosopher through education. Conse-
quently, we might expect this extended image to give us an account in detail both
of our basic natures before education starts13 and about how we might develop
therefrom.
Socrates’ account is heavily structured, in four marked sections:
i) the prisoners in the cave (514a1–515c3);
ii) what happens when the prisoner is released and turned around
(515c4–516c3);
iii) the reflections of the returning philosopher on his previous state (516c4–e2);
iv) what would happen, were he to return to the cave (516e3–517a7).14
So while the first section describes the prisoner’s natural state, the second, third,
and fourth explain what it is for him to be educated.15 Just as the second and fourth
sections complement each other (they describe, respectively, the journey out of and
12
Notice the odd word order at 514a2–3: ‘For see men as in an underground cave-dwelling . . .’. The
generic ‘men’ here again suggests that what follows is a description of human nature, rather than just
introducing the story—‘imagine some men . . .’—and then giving an account of what happens in some
cases—e.g. where men are banded together in a society. The repeated vocabulary of seeing (which some
translators misleadingly render ‘imagine’) brings the vocabulary of the frame and the framed close
together.
13
Cf. the reference to aphrosunê at 515c5.
14
There is a question about just how, and with what detailed correspondence, the cave fits with the line;
517a8 ff. Are we to say that the prisoners are in a state of eikasia (510a, 511e)? Whatever that would
involve, we must surely attribute something cognitively complex to them: the point is not that they are
merely confronted by shadows, but rather how they process them (otherwise we could not say that there is
anyone in this state exclusively, let alone that we are like them). The prisoners fail to (are not in a position
to) answer questions about what things are, both because of the limitations of the objects of their cognition
and because of their cognitive constraints. On this, see Harte (2007).
15
Still, the state of the dialectician at the end of his education is also described in terms of his nature,
537c; compare 424a.
the journey back to the cave), so also there are connections between the first and the
third sections—in particular, between the released prisoner’s reflections on himself
and his erstwhile companions in the third section and his earlier view of himself and
them in the first.
4. The Prisoners’ Cognition

So what is the prisoners’ natural state? Their physical situation—bound so that they
can move neither their limbs nor their necks, stuck facing the wall on which the
shadows are cast—explains what they see (514b2): from the outset the focus is on
their perceptual range.16 Socrates then describes four distinct cognitive17 conditions
of the prisoners:
i. how they see themselves and each other (515a5–8);
ii. how they talk to each other and name what they see (515b4–5);
iii. how they attribute sounds to the shadows (515b7–9);
iv. how they think that the shadows are the truth (515c1–2).
The second condition is introduced as parallel (tauton touto, 515b2) to the first.18
Consequently, the last three conditions, which elaborate how the prisoners relate to
the shadows of the things carried past behind them, allow us to fill out the first, how
the prisoners relate to the shadows of their companions and themselves. But this first
condition, I shall suggest, renders the entire account both complex and severely
puzzling. The puzzle, I shall suggest, is no mistake.
Consider, then, the last three conditions:
Second condition: So if they were able to have conversations with each other, don’t you think
that they would think themselves to be naming these things passing by, the things they see, as
being? (515b4–5)19
Perhaps what they see—the shadows of artefacts cast on the wall before them—gives
the prisoners the conditions for reference: these things (on this reading of the text)
16
514b2, 515a6, 515b5, etc.; reinforced and marked by the language of vision in the frame dialogue, as
Socrates and Glaucon continue the discussion, e.g. at 514a2, 514b5, 514b8–9.
17
Does ‘cognitive’ need defending? If perception were raw, unmediated, and non-complex, it might; but
I say that perception is not raw here or in at least some other passages either. When we see, we see that this
is thus and so, or we see this as thus and so: more below and compare Charm. 167 ff.; Rep. 523–5 (an
exception may be the theory of perception in the Theaetetus; but the standing and role of that theory in the
dialogue are problematic). See Chapters 6 and 9.
18
‘What of the things carried past?’ (515b2) is elliptical, picking up the construction at 515a5.
19
Translation and text are vexed. ¯N s ØÆºªŁÆØ x ’ r æe Iºº ºı , P ÆF Æ ªB fi i a Z Æ
ÆP f OÇØ – æ ›æfiH; [Slings] ¯N s ØÆºªŁÆØ x ’ r æe Iºº ºı , P ÆF Æ ªB fi i a
Z Æ ÆP f ÇØ OÇØ – æ ›æfiH; [Harte]. Harte’s extremely persuasive version of this reads it
without the qualifier ‘as being’ (2007). She argues that they would think themselves to be naming the things
that are when they mention the things they see. Whether or not one follows that account of it, Socrates
seems here not to draw a sharp contrast between the structure of seeing and the structure of thinking [of
something as]; so here perception may have broad cognitive content, as I argue further below.
are what there is; this is their reality.20 At least, the prisoner can name what he sees
(and somehow take himself to be doing so)—he may say ‘that is a bison’ (or perhaps
just ‘bison’). And he should be able to identify it: ‘a bison, not an okapi’ (that is some
part of the competitions held with his companions: ‘an okapi next’). Furthermore,
that reality somehow exhausts what there is for them to talk about—so that when
they hear sounds, they attribute those sounds to the shadows before them:
Third condition: What if the prison also had an echo in the facing wall? Then when one of the
carriers said something, do you think the prisoners would think what sounded was anything
but the passing shadow? (515b7–9)
The shadow moves and the echo sounds—and the prisoner says ‘the bison bellowed’,
‘the okapi squeaked’. And these echoes and shadows somehow exhaust the truth for
the prisoners; this is all there is to reality for them. What are shadows in fact are not
shadows for the prisoners, but the real things for them—these are the bison and the
okapi of their world:
Fourth condition: Altogether, then, I said, such men would think that the truth is nothing but
the shadows of the artefacts. (515c1–2)21
Perhaps the prisoner may say ‘These are all the bison there are’ or ‘what a complex
world I inhabit’. But notice just how much is captured by these conditions. The
prisoners perceive what is projected in sight and sound by the wall and they seem to
talk about it in complex ways. Their talk is even reflective to a degree (‘they think
themselves to be naming’; ‘the truth is nothing but . . .’) and its content is exhaustively
explained by their perception of the shadows before them.
Suppose, then, that the cave describes the bare cognitive equipment that comes to
us by nature (as if in the cradle), especially how what we say is related to what we
see.22 That relation is determined by two aspects of the prisoners’ situation: that the
objects they see are somehow ersatz or derivative; and that their point of view is fixed
and so limited to what they behold before them. Consequently their (our) cognitive
state is somehow or other deeply wrong, mistaken, inadequate, badly in need of the
education the Republic may provide.
20
Perhaps they also give them something more like the answer to a ‘what is x?’ question, like the lovers
of sights and sounds (as Harte’s account would allow). Maybe the prisoner can say ‘you ask what is a bison?
That is what a bison is . . .’. If so, the first condition might have him asking ‘who am I, really?’ and
answering ‘I am that’, pointing to the shadows on the wall.
21
This condition is very strongly framed—prima facie, it does not allow a restriction on the scope of
what they see on the wall of the cave (for example, that the image only tells us about their moral truths and
concepts). They do not, according to the image, have any different sort of access to ‘themselves’.
22
Of course it is hardly cradle-like to suppose we speak when we are born. So someone might complain
that the role of the cave is much more limited than Hierocles’ cradle. In reply, notice the claims to
exhaustiveness that pervade the passage: that sits ill with supposing that this is an illustration of something
much more limited than the experience of the prisoners as a whole (e.g. their moral experience). Consider
the repeated phrases for identity (‘nothing but’ vel sim.) at 515a6, b8, c1.
5. The Prisoners are ‘Like Us’

The prisoners are, Socrates says, like us (515a5). And this likeness is explained by23
the first condition: for they can see themselves and each other only by virtue of the
shadows they throw on the wall opposite:
First condition: For have such people, first of all, seen anything of themselves or each other, do
you think, save the shadows cast by the fire on the wall of the cave in front of them? How could
they have, he said, if they have been compelled to hold their heads fixed throughout their lives?
(515a5–b1)
This opening account of the prisoner’s position at 515a5–8 is elaborately put: literally
‘such people, first of all, of themselves and each other do you think anything have
seen but the shadows?’. This calls attention both to the subjects of the seeing and to
the reflexive nature of what is seen, at the same time as it points to how the seeing is
in fact an identification: they see in the shadows something ‘of themselves’.24 How
does this work? And how does it manifest itself in what the prisoners say (on the
analogy with the later conditions)?
Imagine what happens when the prisoner sees his shadow or that of his compan-
ion. As in the later conditions, he will name what he sees: he says, for example,
‘Socrates’ or ‘Alcibiades’. But suppose he is in fact Socrates, and he is naming the
shadow that happens to be cast by his own head. Perhaps he says ‘Socrates’ and in
fact names himself: but in doing so, he just names ‘Socrates’, and makes no claim to
be seeing ‘himself ’. Then his seeing himself will be de re: in fact the prisoner has a
seeing of himself, but that fact will not be present to the mind of the prisoner. Or
perhaps he sees the shadow, and sees it as somehow his own, and the naming as
naming himself; then his seeing himself will be de dicto (he will have in mind not only
that this is Socrates, but also that this is himself: ‘that is me’; ‘I am Socrates’).
Which?25
If de re, why should this be significant as the first thing he sees, or the condition
that explains the prisoners’ likeness to us? His own shadow would be but one among
many. And if de re, why would this ground the talk among the prisoners?26 After all,
the prisoners don’t just identify and recognize the shadows in front of them (they
don’t just itemize things on a list); they are imagined to have conversations with each
other about the things they see, conversations with complex linguistic content—
reference and attribution—and competitions with each other where one or the other
of them will be acknowledged as the winner.
In all this talk, each prisoner needs to be able to say of his companion just what he
would say of the other shadows in front of him. So he should be able not only to
23
Notice gar at 515a5.
24
Something similar is to be said of the second condition, on Harte’s view.
25
I am grateful to Doukas Kapantais for discussion.
26
Notice the repeated pronouns: allêlous at b4; autous b8; allêlôn 516c9.
recognize that his companion is there, but also to identify him, to be able to say:
‘Alcibiades’; or ‘this [pointing to the shadow] is Alcibiades’. And he should be able to
make attributions, such as ‘Alcibiades is speaking’, or ‘Alcibiades is dumb’, or
‘Alcibiades won the competition’. So whatever else we say of the content of what
they say to each other, that conversation is predicated on their being able to recognize
that there is someone to have the conversation, and the competition, with: they must
be able to make that identification. But their information is restricted to what the
shadows provide: the shadows are their reality and their truth. So when the prisoner
says ‘this is Alcibiades’, he refers to the shadow before him, even when the shadow is
in fact cast by Alcibiades himself, sitting three seats away, himself in chains.
The same will hold of the things the prisoners say about their own shadows. But
their attribution here needs to be set in the context of the lives they live. If they are to
compete, there must be a way of their saying that they themselves have won the
competition, of saying not ‘Socrates won!’, but ‘I won!’. So they see the shadows as
themselves; and their attribution is de dicto. Each might start out in the competition
by saying, of the shadow in front of them, ‘that [pointing to the shadow] is me’. As a
consequence, the ‘seeing’ that the prisoners do of themselves and each other is
cognitively rich, forming as it does the condition for talk between them.27 It can
have expression in some such sentence as ‘this is Socrates’, ‘this is Alcibiades’; and
in further sentences: ‘that is Socrates, again.’ This is about identifying, and re-
identifying, themselves and each other: the first thing they see (it may be first in
the order of explanation, not necessarily in time) is a precondition for their life in the
speluncar community they inhabit.
6. Self-Reference and Self-Perception

Theirs is, however, a parlous condition—as Glaucon anticipates (515a4). Identifying
another could be a matter of christening that shadow, in the first instance; and thence
of recognizing it, once christened, again. So, Alcibiades sees Socrates’ shadow for the
first time, and declares—or is told—that this is Socrates; and thereafter re-identifies
that shadow with the same expression, ‘this is Socrates’, and takes that shadow-
Socrates to speak and to win the competition. But what are we to say of the
corresponding case, where I claim victory and identify myself, where that identifica-
tion is somehow de dicto? If I articulated what I see, I would say something like ‘this is
me’. How do I get the referent, the ‘me’, the first time around? What is it for ‘this’ to
be limited to the shadow world before me, and then to be said to be ‘me’? And what is
it for this limitation to be explained by Glaucon’s next remark, that the prisoners have
had their gaze fixed forwards throughout their lives? Somehow or other the limita-
tions of the prisoners’ perception of themselves are accounted for by their point of
27
There is a great deal to be said about why talking is so fundamental, in Plato’s view; see Chapters 1
and 6.
view—straight forwards—while that point of view still allows them to get hold of
the referent, ‘me’. And somehow or other those limitations need to be limited: for the
prisoner, surely, gets something right about it when, pointing to his own shadow, he
says ‘this is me’. For—in claiming success in the competition, for example—he rightly
denies that Alcibiades was the winner. How can this work?28
If the shadows are all there is to the prisoners’ reality, there is no more to my
identification of myself than what I say about my own shadow. I can’t say of my own
shadow that it is a shadow caused by me, any more than I can say of the shadow I call
‘that bison’ that it is caused by a bison figurine (or that the figurine is a copy of
something else). I have no other source of access to ‘me’ than the shadow itself: I see it
as me, not as my shadow. And that just looks seriously peculiar for the special case of
‘me’.
Consider a different way of thinking about ‘me’ and ‘I’. When I use the expression
‘I’, it might be said, I cannot be wrong in my self-reference—in referring to myself
I am immune to misidentification.29 Perhaps the linguistic role of ‘I’ just does this
trick:
I claim that there’s no more to understanding a token of ‘I’, whether as speaker or hearer, than
being able to apply to the token the rule: English speakers should use ‘I’ to refer to themselves
as themselves.30
Such a thought might be apposite to the cave, which is repeatedly about the relation
between experience and speech. And an ‘immunity to error through misidentifica-
tion’ might be thought to be something basic to our relations with others and the
world (something, perhaps, that turns up with the cradle). This ‘I’ to which I refer
must be the ‘I’ who does the referring; and from that point flow all my other self-
attributions (an ‘I’ who looks within myself must, equally, be the ‘I’ who does the
looking). Successful self-reference is—so this thought goes—somehow natural or
basic to language. But just this self-reference is what seems to go wrong in the cave.
The prisoners’ situation is, however, complex: on the one hand there is the object
in view (the shadow, and the causal relations the shadow in fact has to other things in
the world, although those relations are obscured from the prisoner); on the other
there is the prisoners’ view of that object. So while self-reference might be immune to
error by misidentification, if self-reference is explained by self-perception—as it is in
the cave—that may bring error along with it. Indeed, there may be a whole tangle of
issues here. On the one hand, if I see myself, I seem to run foul of the paradox of
reflexivity: how can I see myself, when myself is doing the seeing? On the other hand,
if what I see, in seeing myself, is in fact only my shadow, even though I don’t know it,
it looks as though my self-identification has gone wrong. In both cases, the paradox
28
Again, see Harte (2007) for parallel issues about the rest of the prisoners’ talk.
29
See Shoemaker (1968) and compare discussion of de re misidentification in Pryor (1999).
30
Sainsbury (2011).
of reflexivity is tackled by indirection: either I perceive myself by perceiving my parts,

or I perceive myself by perceiving my image. Even Hierocles acknowledges that
perceiving myself is not a direct reflexive, but shows up instead in the limited
perception of my bodily parts. Likewise, for the prisoner, self-perception comes out
peculiar, just because both his view and the object in view are somehow
compromised.
7. Does it Matter?
Yet however the prisoner’s view of himself works, it seems to matter. For when the
philosopher returns to the cave, he is in a position to see his erstwhile companions,
and to recall and reflect on his own life in the cave, and theirs (516c4–5). When he
does this, moreover, his old life suffers by comparison: for he thinks himself fortunate
in the change in his situation, the others he left behind pitiable (516c6). If there is
meant to be a careful parallel between the philosopher’s position and that of the
prisoner, this moral view of his own life and that of others should have as its
counterpart the prisoner’s view of (or about) himself, impaired as it is by his being
tied down in the cave. Some part, that is to say, of the comparison between the
prisoner and the philosopher in the first and third sections of this short passage
involves the differences between their views of themselves. And the prisoners are ‘like
us’ (515a5): what they see there is somehow analogous to what we see right now—a
shadow on the wall, an image in a book, some other indirect sight of ourselves.31
Someone may object: If the prisoners have such an anomalous view of themselves,
why is the sequel not focused on resolving it? The description of the ascent from the
cave seems to attend most of all to the objects the prisoner sees, where those objects
are expressly outside him; even outside the cave the images and reflections he sees are
images of men in general, not of himself in particular (516a). Starting with the
prisoner’s view of himself, so the objector may insist, is merely a matter of artistic
detail, and that is why the prisoner’s self-seeing is dropped once he ascends from his
seat before the shadowy wall. So self-perception and, on the return, self-knowledge
become redundant to real enlightenment.32 Instead—on this account—the ascent
from the cave offers to the philosopher the increasing objectification of his know-
ledge, with the result that he will have, at the end of his journey, no interest in what is
personal at all, including himself; and that is what makes him good at being a king.
Self-knowledge, then, is quietly dropped.
A different view might be that the dropping is loud, if not clangorous; and that the
sheer paradoxicality of the prisoners’ situation is designed to make us respond. That
31
Hence see section 4: the repeated language of vision in the frame dialogue. This will become
significant later, in Book X, when the discussion of seeing oneself is repeated in the context of a discussion
of artistic imitation.
32
Burnyeat (1997), 239–40, and Brunschwig (1999), (2003) take differing positions about this.
response may demand, however, a deal of revision of our assumptions, not only
about what constitutes the self, but also about what it is to see or know it. Both issues,
I say, are expressly invited by the anomaly of the first thing the prisoners see—
themselves—and by the radical change when the philosopher comes back down into
the cave, and reflects on how his life compares to those he left behind. Here we
cannot imagine the philosopher simply occupying an impersonal perspective (he
explicitly compares their lives to his own, 516c5); nor can we imagine his stance being
one of detachment, since he thinks about his own life in terms of its happiness, its
eudaimonia. What happens, then, when the philosopher-king thinks at this stage
about himself ?
Well, perhaps the point is that what the prisoner sees is a self determined by his own
culture—the philosopher, by contrast, has a view of himself and his erstwhile com-
panions that is enlightened, and right.33 On that account, what the prisoner gets wrong
is the object of his view: and so perhaps what needs to happen when the philosopher
descends is that he repudiates this earlier self, supposes that this was not himself at all.
When he comes back down, on such an account, the philosopher says ‘this is not me’,
in response to his earlier self ’s identification of his own shadow, ‘this is me’.
But this seems to miss something important: the prisoner’s problem is not merely
that his visual object is wrong, but that the wrongness of the object is explained by
how he views it. The explanation (at 515a9–b1) of what goes wrong in the prisoner’s
view of himself is not said to be ‘oh well, that’s only a shadow’; nor is it, as it is in the
case of the other objects the prisoners see, that this shadow is cast by an ersatz object.
On the contrary, this shadow is indeed cast by the speaker who might say ‘this is me’;
and there is some sense—as we consider this image from the outside—that he is right,
along with a rather worrying sense that he just must be wrong. The difference
between the prisoners and the returning philosophers is not that we have a new
self, but that we have a new view when we come down: we can see the causal
structures of things, even whole lives, rather than just looking straight ahead at
shadows. On that new view, acquired outside the cave, we might revise what we
said before; for when, as a philosopher, I see myself, I see myself better, and perhaps
discount some features (grey and flat, for example) as belonging to the shadow, not to
my self. But I cannot also deny, on my return, that my own shadow was indeed cast
by myself: my earlier view was still, somehow, right. What is it to see myself better in
this way?
8. The ‘Amazing Sophist’ of Book X

My second oddity occurs in Book X, when Socrates is talking about making things—
about craftsmen, about poets, and about the things they make—under the general
33
On this, see Burnyeat (1997).
heading of imitation, mimêsis. This passage, like many others before it, has often been
construed as including a further explication of how the status of the objects of sight,
or of understanding, determines the mental state of the subject (is it knowledge? Is it
belief? Are we to look to what it is of, and read the mental state off from that?). At the
same time, many have argued that this account of Plato’s epistemology is too thin to
do justice to the complex things he says about states of mind and their structure;34 yet
the strangeness of this passage has, I think, been underestimated.
Socrates has been talking about his ‘customary method’ of positing a form for each
named plurality, and applied this to artefacts and their templates (596b). The
craftsman for each artefact, he says, looks to the form, the idea, in making whatever
he constructs—for without that he would be unable to make anything at all (596b).
All this seems to be about making (copies of the template) and looking (at the
template to do the making). Accordingly, there are three kinds of craftsmen—
painters, craftsmen, and god—each of whom makes things, of different orders of
reality (596e–597e). Central to this arrangement is the contrast between the painter
and the craftsman, a contrast in part explained by the difference between appearance
and reality. And somehow all of this is to be deployed to explain why tragedy is
deceptive: like a painting, it provides us only with a view (of whatever it is we see
when we watch a play) from one direction. So what is it that is wrong with what we
see? Is it that the actors are dressing up? Or is the problem also something about
where we are sitting in the theatre?35
That the latter, our point of view, is essential to Plato’s account is suggested by a
shock in the middle of the discussion of the couches and the tables and their makers.
In advance of the entrance of god into the classification (at 597b), Socrates suddenly
seems to change tack:
. . . But look at this craftsman, too, and consider what you would call him.
Which one?
The one who makes everything, everything which the individual craftsmen make. (596b10–c1)
In the context of the discussion of the idea of the couch and the table, we might be
forgiven for expecting this ‘one who makes everything’ to be someone supernatural,
the creator of all the templates of the arts and crafts. This expectation is encouraged
by Glaucon’s response:
Someone clever you describe, and a wonderful man. (596c3)
This translation preserves the Greek word order: in Glaucon’s response the noun
‘man’ seems bathetic—surely no man makes everything? Surely Socrates is talking
about god, wondrously ingenious?
34
E.g. Annas (1981); Fine (2003); Harte (2007).
35
For some responses, see Harte (2006), (2010).
I haven’t said that yet, but you will soon say what you say even more. For this same craftsman is
not only able to make all furniture, but also he makes all the things that grow from the earth, all
the animals . . . (596c4–7)
God, surely?
. . . and himself. And in addition to these things, he fashions the earth and the sky and the gods
and everything in the heavens and in Hades beneath the earth. (596c7–9)
We may now begin to hesitate: although everything Socrates says may remind us of,
for example, the divine causation of the form of the good (509b) and the world
outside the cave (516a), he seems to suggest that this amazing creator fashions also
himself. Is that what we would expect, even of god? The Baron Munchausen effect of
this kind of self-causation should worry us more than somewhat (especially if we
recall the cautious discussion of the problem of reflexives in book 4, 430–1). And the
marvel-making features of this craftsman’s activity echoes, surely, the way that
someone makes the ersatz objects that are paraded before the fire within the cave
(514b). Glaucon, at any rate, begins to be suspicious, for (again with the odd delayed
word order):
You are talking, he said, of a quite wonderful sophist (596d1).
Suspicion might easily fall on the expression ‘sophist’—but here, surely, its evaluative
content is unclear, until Socrates begins to clarify:
Are you not convinced? I said. And tell me, do you think that there could not at all be such a
craftsman, or do you think that there could be someone who makes everything—in one way,
even if in some other way he does not? Or do you not perceive that you yourself would be able
to make these things, in one way?
And what, he asked, would that way be? (596d2–6)
Socrates has disambiguated: we are not talking about god, only Glaucon:
Nothing difficult, I said, but what is fashioned often and quickly—most quickly of all,
I suppose, if you were to take a mirror and carry it around with you everywhere. For quickly
you would make the sun and the things in the heavens, and quickly the earth, and yourself and
all the other animals and furniture and plants and all the other things I mentioned right now.
Yes, indeed—I would make them appear, but not to be in truth. (596d7–e4)
An amazing sophist indeed: someone who carts a mirror around, and who looks at
himself in it, and in that way creates himself. The short excursus into mirrors is à
propos, Socrates says. Why?36
My earlier objector would say that this confirms the thought that the prisoners
only see themselves when they are at the lowest grade of existence, stuck with their
36
Again there is a great deal else in play here, notably in the talk about mirrors; this is taken up—and,
I say, dropped—by Aristotle in his discussions of the friend as another self (see Chapter 16).
heads fixed looking at the shadows. The mirror, perhaps, is sneakier, because it
moves around with us, so that whichever way we turn, we see our own mirror images
in it, but self-perception is still something limited to low-grade cognition. But—in
riposte to the objector—the earlier puzzles should not leave us now: if the mirror in
which I gaze upon myself makes me, what is the relation between the me that is made
and the I who does the looking, or, indeed, the I who makes myself by carrying a
mirror around? The mirror invites the same sort of difficulty about reflexives posed
in the cave, and asks the same questions with renewed emphasis. What is the relation
between an image of myself and myself, given that the self who sees the image is just
what the image is of? How can I say, whether looking in the mirror or at the wall of
the cave, ‘this is me’? In cases of self-perception and self-knowledge, how am I to sort
out the difference between subject and object, when they are the same me?
The puzzle is intensified by the vertiginous composition of the passage. Until very
late on, we may find ourselves unsure whether we are talking about a divinity or a
charlatan; and the moment where our doubts crystallize is the point where Socrates
suggests that this amazing craftsman makes himself. He does it, it transpires, by seeing
himself: so the self-perception that seemed to drop from view and from significance
in book VII here again takes on a central role. What exactly is the point of bringing it
back with such attention now?
Both in the cave and in the mirrors, the demand to disentangle subject and object
is a double one. First, we are not just asked to classify what we see, or to determine
how we see in terms of sight’s objects, but we are invited to consider the seeing subject
in the special case where what we see is ourselves. Second, the puzzle is a direct
challenge: we, the readers, are invited to think about the seeing subject in an extended
image of ourselves as we read: the prisoners are ‘like us’.37 Is the point here that the
subject of self-perception is created by its context—so, cashing out the image, by the
society in which the individual lives?38 What, then, is going on when we connect
what we are told by Socrates about ourselves, with our own position as we read? And
what, consequently, are we to say about the deficiency of this mirrored view of
ourselves? Is Socrates telling us that we need to jettison our interest in ourselves
altogether to reach the pure truth of the world on reality? Or should we recognize
that the image is a sham, imposed by others, something we should reject when we
properly see who we really are? How are we to achieve that?
Notice the similarity between the mirrors and the shadows. In both cases, the
ordinary objects of sight are somehow or other derivative: we see the shadow of a
bison, the mirror image of an okapi. In each case the image derives from some
independent original, and we might well suppose that in changing the object of our
sight, its veracity would be improved: all I need to get the right view of a bison is to
37
This is part of the puzzle (‘like who?’): as such, I suggest, it ensures that the puzzle provokes a broad
range of people, including the readership of the Republic itself.
38
See here e.g. Wilberding (2004).
look at a real bison, not its shadow. However, the suggestion that mirrors and
shadows may provide us with a view of ourselves changes the emphasis of the
account; and it does so by having these self-perceptions perform a prominent
role in the two passages. For in self-perception the relation between the image and
the original is tricky, since the original is not so much the proper object of our
perception, but rather the subject—who is present, all along. As object, conversely,
the subject must be seen indirectly—whether partially (as in Hierocles’ account of
proprioception) or derivatively (in a mirror or a shadow). To get at the truth about
ourselves, then, we do not merely have to adjust the object of our inspection, but also
figure out how to access the subject. What we see, when we see ourselves in a mirror
or as a shadow, is indeed ourselves, even if there is some kind of deficiency in the view
of ourselves it provides. The mirrors and the shadows shift focus towards how we
view, and away from the image in view. Essential to that shift is the reference to
ourselves—a reference, as I have argued, which appears in the content of what we say
of our viewing: the prisoners and the mirror-carrier see their images as themselves.
9. A View in the Round

Consider what happens next in Republic X. The issue is still who the imitator is
(597b2).39 The painter—like god—is making imitations: imitations of the creations
of the craftsmen (598a1), portrayed not as they are but as they appear. And that
contrast, between how things are and how they appear, is now explained:
. . . a couch, whether one looks at it from the side, or from in front or from anywhere else, does
it differ at all from itself? Or does it not indeed differ from itself, but appears different [sc. from
different viewpoints], and the same for everything else?
That’s right, he said. It appears, but does not actually differ at all. (598a7–10)
But the sequel discusses how the painter can imitate not a couch but a carpenter—
and so on for the other craftsmen:
For example, the painter, we say, will paint a cobbler or a carpenter, or one of the other
craftsmen, for us, although he [the painter] knows nothing of these crafts. But nonetheless, if
the painter is a good one, then by drawing the carpenter and exhibiting him from afar, he will
deceive children and stupid people into believing that he is a real carpenter. (598b8–c4)
This shift is once again surprising, away from the objects that appear, and towards
their creators. Painting here imitates what appears by imitating the point of view of
the painted craftsman. In the painting, the craftsman is frozen in his view of his
39
Even if the painter is a mirror-carrier, so that the focus seems to shift back towards the object of our
view (596e ff.). This explains Socrates’ interest in the intentions of the makers (god, for example, is a pretty
determined character, 597c–d), and his conclusion that if the tragedian provides images that are at the
third remove from ‘nature’ (597e3), then the tragedian, like the other imitators, is the third remove from
the king and from nature (597e).
object: the pictured craftsman does not see what he is pictured as making from all
sorts of points of view, but only from the point of view in which he is depicted. Like
the prisoners in the cave fixed at the neck, he can see only in the direction in which
his gaze is depicted. The contrast between appearance and reality, thus, is offered not
merely in terms of the objects of the craftsman’s inspection, but in terms the view of
the craftsman himself, as he is depicted. (Again the doubleness of the account should
not escape us: for we too see speakers frozen in their points of view, as we look at the
dialogue from the outside.)
This, as 598a made clear, is exactly what we are meant to get from the idea of an
appearance: something not seen from more than one point of view. That does not
make the point of view false; but it makes it seriously deficient. The fixity of the point
of view is the same one we are offered when we see ourselves in a mirror, for the point
of view is determined and exhausted by the angle of the mirror. That does not mean
that this is not, somehow or other, an image of me: but it makes that image
inappropriate as the source of knowledge about me (see 598c ff.). For I cannot see
myself, as it were, in the round, only straight-on, just as the depicted carpenter can
only fix his painted gaze on one side of the table, depicted as it is, flat. The self-image
of the prisoners, likewise, is as flat as my view of myself in the mirror. This limitation
of my image is the source of the paradox of ‘this is me’, said of my shadow first
encountered in the cave.
10. Education
So what should we do about education? The prisoners are vulnerable to a double
failure—they are focused on the wrong objects, and in the wrong way, from a
perspective that is fixed and flat. They fail, as a consequence, in the first stage of
their cognitive progress: for they fail even to see themselves right.
The reason for this is partly a matter of their point of view. Both shadows and
mirrors let us down: they are mono-perspectival, and thus account for the subject’s
failure of self-perception. If I see my shadow, or my reflection, I can get it right in
saying that this is me; but I can only do so if I have access to other views of myself,
other ways of seeing that will allow me to get knowledge of me—not just knowledge
of that flat image that happens, indeed, to be an image of me. The fault, then, lies
partly in the subject of self-perception: without a view in the round, an extensive self-
inspection, he must fail, in some way, to identify himself on the wall of the cave, even
insofar as he is right that the shadow is indeed somehow him. For what he sees is not
himself, as a whole, but some image of himself; and he sees it from only one of many
possible points of view. His self-perception is indirect and it is partial; and the
indirection explains both what he gets right and what he gets wrong. (There still
lurks, thus, the problem of reflexivity.)
But as the prisoner emerges from the cave, he is not merely exposed to
different objects; he takes different views of what he sees (he looks round and back,
e.g. 515d–e). The philosopher’s culminating view of the sun (and of the form of the
good) is described not only as a seeing, but also as a contemplation (theôria, e.g. at
517d4) and a viewing (theasthai, e.g. 516b6). This vision has been thematic since the
very beginning of the dialogue, where Socrates’ vision of the Thracian procession
(327a ff.) is thoroughly discursive. The story of the ascent describes how the
philosopher sees the artificial sources of the shadows (and, it seems, recognizes
them as such, 515c4 ff.) and then the shadows of their natural originals, and then
the originals themselves—by being able, in the first instance, to turn his neck round
(515c7). The ascent is punctuated by periods of blindness, when his point of view fails
altogether; and the descent likewise has him fail to see at all, once back in the
darkness of the cave. But each of these phases of the life of the philosopher press
hard, not so much on the objects of his perceptions (are they ersatz or somehow
real?) as on a contrast between the fixed point of view of the prisoner and the
improvement of the sight of the philosopher, described in terms of how his vision,
as he escapes, is more and more in the round, more and more, as I shall say,
stereoscopic.
That is why, in his practice of dialectic, the philosopher is imagined taking up
different points of view; and that is why, once he has reached the summit of his
ascent, he has become a ‘synoptic dialectician’ (537c7). Within the imagery of the
cave, he still sees, but he sees better. His achievement is not merely a matter of
standing out there in the sunlight—it is a matter of how he has an all-round view of
things.40 When he returns, therefore, he sees himself and his erstwhile companions in
the round: and this allows him to compare favourably his own life with the lives of
those still imprisoned—it gives him a view of his whole life. It does not falsify the
remark he may have made when he was still tied down—of the shadow before him,
‘that’s me’—but it allows him a full account of what it is to be himself, and to live the
life he has now discovered.
But the difference between how the prisoner sees and how the philosopher sees is
supposed to illustrate the difference between their general cognitive states. The
prisoner’s single perspective illustrates the piecemeal nature of his cognition; the
stereoscopic vision of the philosopher shows how he knows or understands. While it
is true that the prisoner and the emerged philosopher have different objects of
cognition, they differ also in the view they take of those objects. Perception is like
knowledge in giving us some kind of genuine view of the things we see: but it is unlike
it in giving us that view from a single perspective. Knowledge has some special objects
of its own; but it also has a special way of seeing: the stereoscopic way. This is not an
40
There is an important issue here—for expansion on a different occasion—about how all this figures in
moral vision and moral transformation. At least, this material presses the idea that there must be an
internal, as well as an external, component in the development of moral vision; but see Chapters 6 and 10.
impersonal view, nor need it be a neutral one; but it is one that sees things all the way
round.41
How, now, does this stereoscopic account of the philosopher’s view explain what
we might recognize as self-knowledge? Self-knowledge is acquired, on this account,
not from a direct seeing (of one’s shadow, or one’s reflection) but by a concerted
attempt to see oneself as a whole. Self-knowledge is not here understood as mirror-
recognition (even though the prisoner does somehow recognize his shadow as his
own), but as some deep understanding of myself, of who I am and the life I live. The
contrast between these two views of myself explains just why it is that the lives of
the prisoners are so limited; and why their views of themselves are so puzzling. The
prisoners escape the paradox of reflexivity only by virtue of making a mistake (their
shadows are not themselves). By seeing himself stereoscopically the philosopher, by
contrast, avoids the risk that in seeing himself reflected back he somehow creates
himself in his own mirror. For the object of his vision is not known by a mirror-
identification (or, then, by the first-person authority of introspection, or by proprio-
ception, or by the immediate awareness of consciousness), but as a complex whole, by
being seen from all around. And in seeing myself from all around, I am able to grasp
not only the object of self-knowledge, but, and more importantly, its subject: the self
who views, all around. The stereoscopic view explains both the object in view and the
subject.
11. The Normativity of Perception

The cave is echoed in the taxonomy of imitation and the descriptions of the world of
the amazing sophist. And both passages, I have argued, are set up to puzzle the
reader, and to invite reflection on the problems of reflexivity, self-perception, and
self-knowledge, from the point of view of both object and subject. I conclude from
this that both self-knowledge and reflexivity matter to the Republic’s account of
knowledge.
This has the consequence, first, that knowledge has internal conditions, conditions
specifying in some respect the disposition of the subject, as well as external ones (it
cannot merely be true that the objects of knowledge determine it). Second, the
perspectival account of knowledge supposes knowledge to be complex, if not holistic,
in its content. And third, it shows (via the problem of reflexivity) that the subject of
knowledge must be understood in complex ways: this may allow there to be some
logical space for how the subject of knowledge may change and develop in the course
of education. All three consequences should press us to think of knowledge here as
epistemically rich: as understanding, rather than a piecemeal assemblage of justified
41
Seeing may still provide some kind of information, and it may still be a suitable analogue for
knowledge. But the conditions on seeing are complex: i) it is not merely the direct affection of the subject
by the object; ii) it has complex content; iii) it is essentially perspectival.
true beliefs, where ‘understanding’ tells us as much about the understander as about
what is understood.
But at the same time the puzzles of reflexivity are not easy to shift. This self that
sees is somehow the same one as the object of its seeing. If this is not to fall to the
puzzle of reflexivity (the agent should not also be the patient, as such), we may find a
different problem looming: what is the self that is the object of self-knowledge?
Remember Hume:
For my part, when I enter most intimately into what I call myself, I always stumble on some
particular perception or other, or heat or cold, light or shade, love or hatred, pain or pleasure.
I never catch myself at any time without a perception, and never can observe anything but the
perception . . . the mind is a kind of theatre, where several perceptions successively make their
appearance; pass, re-pass, glide away and mingle in an infinite variety of postures and
situations. . . . The comparison of the theatre must not mislead us. They are the successive
perceptions only that constitute the mind. (Treatise of Human Nature, I.iv.6, Selby-Bigge/
Nidditch 251–63)
Hume insists that introspection only delivers ‘particular perceptions’, no view of ‘the
self ’; the self is a convenient fiction. Does Platonic stereoscopy have an answer to
Hume’s objection? If there is a general problem about the self ’s being the object of its
own knowledge, that problem remains even if knowledge is glossed as something
broader in compass than individual events of cognition. For why should we think
that the collection of points of view that are possessed by the synoptic dialectician are
in any sense held together as focused either on, or by, some kind of metaphysical
unity? Perhaps in focusing upon the view, rather than upon the object in view, I am
attributing to Plato a vanishing conception of the self and an impersonal philoso-
pher-king?42
The pervasive analogy between perception (especially vision) and knowledge may
help here. Right from the lowest stage of the cave, perception has cognitive content,
as I have argued (‘that is me’). So my perception of myself, even at this lowest stage, is
not merely a raw feel, not merely unmediated input from outside, but capable of
complex expression (hence the strongly linguistic features of the conditions of the
cave). But this cognitive content is somehow improved as the prisoner escapes and
climbs out of the cave. That improvement—as I have argued—is not merely a matter
of the philosopher’s changing the objects in view, but of enriching her points of view.
And the two processes are interconnected. For at each stage her understanding both
of what she sees and how she sees it is reflective: she sees that she has a point of view,
or that she had a different one back then—her view of herself as she once was is
essential to her understanding where she stands. By the process of ascent, and the
connected process of reflection, she comes to see better and better, and that
42
This view of the philosopher-king has seemed right to some; see McCabe (2013) for discussion of the
impersonal view.
improvement is connected to her better view of herself.43 So on the stereoscopic

model, perception is normative.44
But the ascent and the return cash this normativity in ethical terms. What is it that
self-perception seeks to see, in the cave? What is it that the prisoner seeks to find out
in the ascent and in the return? The ascent and the return are—as it is often
observed—centred on the life of the philosopher. The project of education is not to
reach a kind of scientific knowledge that would deliver a count of the entities in the
world, including the entity that has that knowledge, even if such a count is available
to the philosopher. Instead, what she seeks is a kind of moral understanding that will
make coherent the role of value in her own life, and that will thus make of her life an
integrated whole. This will be self-understanding in the sense urged on us by the
Delphic oracle; but it is continuous with the elementary attempts at self-perception of
the prisoner in the cave.
So there is here no separation of the moral from the intellectual; on the contrary, as
the context of the Republic makes clear, knowing and understanding are thoroughly
imbued with value. In coming to understand, the philosopher becomes wise. That
involves self-knowledge and self-perception in the sense that wisdom engages
the subject of cognition, as well as its objects. The puzzle of reflexivity is resolved
by taking self-knowledge, or self-perception, to be a stereoscopic matter, multi-
perspectival, reflective and broad in scope. This is understanding; we see and we
are seen, in ways that include the perspective of the subject, in the round. But it is also
moral understanding: essential to the value of a life.45 It is hardly surprising that it
takes so long to come out of the cave; and hardly surprising that to do so we reach a
state that is good in itself, as Glaucon had demanded: understanding of this sort is
virtue—supremely difficult, but a very good thing.46
43
See Burnyeat (1980) on how this complex of reflection and habit is deployed by Aristotle to explain
learning to be good.
44
Normative in the sense that the faculty of perception can be improved (not in the sense that I may
once have aspired to 20:20 vision).
45
See Chapter 12.
46
This chapter has benefited from discussion with many people and many audiences. My thanks to my
hosts and my interlocutors at Hamburg, Western Ontario, Cornell, Toronto, Athens, Prague, Oslo, London
(at the 2010 Keeling Colloquium), Sao Paolo, and Dublin. As always, I have benefited from advice and
discussion with colleagues, especially Charles Brittain, Verity Harte, and Fiona Leigh.
12
Indifference Readings: Plato and
the Stoa on Socratic Ethics
1. Reading Plato
There was a time when to read Plato was to do away with reading.1 Instead,
philosophers sought to analyse Plato’s arguments, to discover within the infinite
variety of his prose the single sense,2 to find out ‘what Plato meant’ and how he
meant us to believe it, too.3 This approach has been replaced, more recently, by
something more inclusive. The objective, for sure, is still to find out what Plato
meant, but it is no longer supposed that this can be encapsulated in the abstract
formalization of those parts of the dialogues that might be designated ‘arguments’, in
contrast to those parts that were once thought to be merely literary, and so philo-
sophically dispensable. Consequently, nowadays greater attention is paid to the way
in which we should read the dialogues—reading them as a whole, and without
prejudice as to which bits matter and which do not.4
But this is a dangerous business, and one in which the boundaries between sense
and nonsense may be difficult to determine. Is this approach going to take us so far
away from the concerns of analysis that there is no common ground between the old
practices and the new? Are we to say—for example—that any reading of a dialogue
may legitimately be described as ‘what Plato meant’? In what follows I offer a case
study. I shall suggest that Plato himself may be used as a guide to reading Plato, and
that such a guide does indeed lead to a philosophical destination of which the analytic
tradition might approve. For a comparison of two short and markedly different
passages, one from the Meno and the other from the Euthydemus, shows us, I shall
argue, the second ‘reading’ the first.
One explanation, of course, of differences between different dialogues might be a
developmental one. Here we might invoke a contrast between Socrates and Plato.
1
This time may itself be a legend; but there have been recent masters of the analytic method:
G. E. L. Owen, outstandingly.
2
This highly analytic approach is what we find at the very beginnings of the analytic tradition of
philosophy, e.g. in Frege (1997).
3
This point is made, for the Euthydemus, by Sprague (1962), 10, who argues that Socrates’ interest in
‘the things denoted by words’ is indicated by his ‘freedom in the use of synonyms’.
4
Examples are now too numerous to list, but mention should be made at least of Burnyeat (1990).
INDIFFERENCE READINGS 229
The historical Socrates—it is commonly thought5—may be represented in Plato’s

earlier, so-called ‘Socratic’ dialogues:6 dialogues in which the ethical theorizing
coheres around the denial of weakness of the will and the insistence that virtue is
knowledge. By contrast, this view maintains, Plato’s middle period ethical theory
moves away from this by posing a new and complex moral psychology, which allows
for ethical conflict and denies any simple equivalence between virtue and knowledge.
Corresponding to these chronological contrasts, the difference I shall find between
the Meno and the Euthydemus may be one between a Socratic view (the Meno) and
its Platonic replacement in the Euthydemus.7
This account, however, seems not to meet the present case: in particular because
both my two passages are significantly indeterminate. The indeterminacy of the
Meno, I shall argue, is teased out by the Euthydemus; it is this feature that,
I suggest, should encourage us to see the latter as a ‘reading’ of the former.8
This invites us to a rich conception of ‘reading’. Consider, first of all, a single
Platonic dialogue. An impoverished read would run straight through from beginning
to end, once and for all (this gives a view of the Platonic dialogue as detective fiction:
once you know who did it, you don’t need to read it again; once you know the answer
to the question, there is no need to reread). This, of course, would be a silly way to
read a philosophical work. But on a richer conception, reading is something you
do again and again, a business that attends closely to allusion, to anticipation and to
echo.9 Now if some later passage in a Platonic dialogue alters one’s view of an earlier
one in the same dialogue, this alteration is itself a part of understanding the earlier
passage. So to read Plato, we read him again and again, and no less back to front than
front to back (this is to treat the dialogues as complex wholes). And we may want to
say something similar about the relations between dialogues,10 and between com-
parable passages in different dialogues: these may not be the simple relations
5
See here, for example, Vlastos (1983), (1991).
6
The view that there is a significant difference, even amounting to an inconsistency, between dialogues
characterized as early or middle or late is increasingly coming under attack (cf. Rowe (1995); Kahn (1996);
Annas (1999)). I confess myself, however, an inveterate developmentalist: I find it entirely congenial to
suppose that Plato changed his mind as he thought about a given issue more deeply, and that different
dialogues allow us to see these changes of mind. This is compatible with a different claim, upon which
I focus here: that dialogues may be related in other more complex and reflective ways.
7
Without prejudice to the question of Plato’s development, many would deny that there is a significant
chronological difference between the Meno and the Euthydemus. In what follows, I shall at least claim that
the Euthydemus shows affinities with the Republic—although I think it was written, or fits best, with
dialogues that are later still. I shall also, however, insist that in terms of the passage I discuss, the earlier
dialogue may not be obviously Socratic after all—merely indeterminate. (A strong developmentalist view
would agree that the Meno is not Socratic, but transitional.)
8
Even that thought, of course, makes a developmental claim: that the Euthydemus is in some sense
‘after’ the Meno.
9
Of course, to those close to an oral tradition this sort of reading will be congenial; compare here e.g.
Havelock (1963).
10
This is a view put quite generally by Kahn’s ‘proleptic’ account of the development of Plato’s thought
(n. 6). I am not convinced that this view, which seems to invite us to a strongly unitarian account of Plato’s
work, does justice to Plato’s talent for self-criticism (and his ability to change his mind) (see McCabe
230 INDIFFERENCE READINGS
between, for example, earlier and later, but instead something more reflective. I shall
argue that we should understand my two passages in terms of one reading the other.
This allows us to see how the view apparently put forward in the Meno might
successfully be criticized, and how the unclarities of the Meno might be resolved in
a quite different way than at first appears possible.
This process, moreover, is a dialectical one. For the Euthydemus itself offers a
conclusion that is vague. This vagueness proved fruitful for Plato’s ancient inter-
preters, themselves engaged in the project of reading Plato. Once again, recent
scholarship has enlarged earlier strategies: this time by turning its attention
decisively towards the later tradition of ancient philosophy.11 As a consequence,
considerable attention has been paid to the ways in which later philosophers used
their predecessors. After all, a great deal of the philosophical business of the post-
Aristotelian period was taken up with interpretation—whether as a matter of
philosophical engagement with earlier thought or as a matter of preserving the
canon, especially of the Socratic tradition. For the figure of Socrates looms large in
later antiquity, both as a philosopher and as the exemplar of how philosophy
consoles us for mortality—of how the philosophical life should be lived. As such,
he was of particular interest to the Stoics; so here I pursue one strand of early Stoic
ethics, which itself provides us with a reading of my two Platonic passages. The
complex interplay that develops between Socrates and Plato, Zeno and Aristo,
provides a view of how a detailed reading of the original texts may bear philo-
sophical fruit.
2. Accounting for Goods

They [sc. the Stoics] say that goods are the virtues, intelligence, justice, courage, moderation
and the rest; evils are their opposites: folly, injustice and the rest. Neither good nor evil are the
things which neither benefit nor harm, such as life, health, pleasure, beauty, strength, reputa-
tion, noble birth; and their opposites death, sickness, pain, shame, weakness, poverty, lack of
reputation, low birth and the things related to those . . . these12 are not goods, but indifferents of
the class of ‘preferred’. (Diogenes Laertius VII. 102)
(2001)). But Kahn is surely right to insist that the relation between dialogues is not merely one of linear
development.
11
This process has been made much easier by Long and Sedley (1987), whose translations of Hellenistic
sources I have used throughout this chapter.
12
Does Diogenes’ ‘these’ refer back to the list of apparent goods—i.e. life, health, pleasure, beauty,
strength, reputation, noble birth? Or does ‘these’ describe the whole list—life, health, pleasure, beauty,
strength, reputation, noble birth and death, sickness, pain, shame, weakness, poverty, lack of reputation,
low birth, etc.? His suggestion that ‘these’ are preferred indifferents suggests the former; but I shall argue
that the version of the Stoic theory that was directly influenced by Plato should allow anything on the whole
list to be a preferred indifferent, depending on the value it derives from wisdom.
Accordingly, after it had been satisfactorily established that only what is right is good,13
and only what is wrong is bad, they [the Stoics] wanted there still to be some difference
between those things which were of no importance to the happy life nor the miserable
one, so that of these some had positive value, some negative, and some neither. (Cicero, de
finibus III.50)
When the early Stoics formulated their theory of adiaphora, ethical indifferents,
they had, it seems, been reading Plato.14 In particular, they seem to have been
reading the Euthydemus and the Meno.15 So we might compare the following two
passages:
soc: Health, we say, and strength and beauty and wealth, indeed. For surely we say that these
things and things like them are beneficial?
m: Yes.
soc: But we sometimes say that the very same things16 do harm, too—or would you disagree?
m: I would not.
soc: Consider, then, what is it that when it leads these things, they benefit us, and when it
does not, they harm us?17 Isn’t it the case that when correct use leads them they benefit,
when it does not, they harm?18
m: Certainly.
soc: Now, let us also consider the qualities of the soul.19 You say, don’t you, that there is such
a thing as self-control, and justice, and courage, and quickness to learn, and memory, and
magnificence and everything like that?20
13
I use ‘right’ here to translate honestum (compare Rackham’s ‘morality’) to pick up the influence of
this sort of theory in modern neo-Kantian accounts; compare e.g. Rawls (1972). My argument will suggest,
however, that not all the early Stoics were proto-Kantians. See here Cooper (1996) and Schneewind (1996).
14
This chapter was in part provoked by reading Long (1996) who treats the Euthydemus as an
important antecedent of the Zenonian theory of the indifferents. My account of the history of the
Euthydemus passage differs from Long’s in two vital respects: first in terms of the analysis of Socrates’
conclusion, and second in the way in which Zeno took that conclusion up. In part this chapter was also
inspired by hearing Bernard Williams on Plato on the intrinsic good (2003).
15
This is frequently pointed out; see e.g. Annas (1994), Long (1996), and Striker’s suggestive analysis of
the Stoics as readers of Plato with a wider scope than merely the Socratic material (1996).
16
Back here to an old controversy: here presumably ‘the same things’ are the same types, not the same
tokens. This contrasts with other passages in Plato where he argues from the compresence of opposites in
some token (this stick, for example, is equal to that one, unequal to some stone; Phaedo 74b ff.) to the
existence of a form that does not suffer compresence (equality itself is never inequality).
17
Socrates makes two assumptions: first that there must be some account to be given of when the same
things harm and when they benefit; and second that this account is to be given in terms of what ‘leads’
them. The vocabulary of ‘leading’ will become important.
18
It is unclear whether the leader makes health nice; or whether it makes it productive.
19
This translation of ta kata tên psuchên is warranted by the realistic cast of the next sentence.
20
‘You say that there is such a thing as . . .’ makes no claim for the transcendence of these qualities of
soul (as is claimed for a similar remark at Phaedo 74), merely for their reality. The symmetry of this remark
with the opening of the argument, 87e5, suggests that at this stage self-control, etc. are treated as analogous
to health and wealth.
m: I do.
soc: Consider, then—if you suppose any one of these not to be knowledge, but to be other
than knowledge, surely it sometimes harms and sometimes benefits?21 Courage, for
example, if courage is not wisdom but some sort of recklessness—surely whenever a man
is reckless without intelligence he is harmed, and whenever with intelligence, he is
benefited?
m: Yes.
soc: And isn’t the same true of self-control and quickness to learn—whatever is learned and
organized with intelligence is beneficial, whatever without intelligence, harmful?
m: Absolutely.
soc: In short, all the endeavours of the soul and all its endurances end in happiness22 when
wisdom is the leader, but the reverse when ignorance is the leader.23
m: It seems so.
soc: Therefore if virtue is one of the qualities of soul, and if virtue is necessarily beneficial, then
virtue must be wisdom, since all these qualities of the soul are not beneficial nor harmful in
themselves, but they become harmful or beneficial depending on whether wisdom or folly
is added to them.24 And according to this argument, since virtue is indeed beneficial,25 it must
be some kind of wisdom. (Meno 87e6–88d3)
21
Socrates shifts here from supposing that correct use should lead health, etc. (so that correct use
and health, etc. are non-identical) to wondering whether self-control, etc. should be identical to
knowledge. In the thought-experiment that follows, non-identity is assumed, so that the next sen-
tences revert (ostentatiously?) to a more instrumental account: courage with intelligence and courage
without it.
22
This is a strikingly consequential claim: wisdom leads the sort of ‘endeavours of the soul’ that end
in happiness. The thought seems to be that happiness is a consequence that is other than the means to
securing it; this thought is often allied with the view that there is nothing morally or ethically
significant about just these means (if there were some other, more economical way of getting happiness
than being led by wisdom, that would be preferable). So I shall characterize consequentialism as the
view that moral/ethical value is conferred by the end pursued, and not by any features of the pursuit
itself. It is worth noticing that on some accounts of consequentialism this treats the end as having
intrinsic worth, and thence value as residing principally in states of affairs. See here Williams (1985), 76.
Myles Burnyeat suggests to me in conversation that here ‘ending in’ should be construed inclusively,
as ‘the endeavours of the soul when accompanied by wisdom are actually happy’ (compare, perhaps,
Theaetetus 173b, where boys end up as men). If, as I suggest below, the Euthydemus passage
provides us with a reading of the Meno, then we may be intended to rethink the implications of
teleutan.
23
Socrates returns to the metaphor of ‘leading’: but now the suggestion seems to be that what is led is
not one of the virtues, but the effects of the virtues (virtuous behaviour, perhaps, or its results).
24
Here Socrates allies the identity claim about virtue and wisdom to a consequential claim about the
ethical quality of the ends. Nothing so far suggests that the quality is determined by the virtue that produces
the ends.
25
That virtue is good was conceded at 87d. The emphatic ‘virtue is indeed beneficial’, like the earlier
‘virtue is necessarily beneficial’ (88c5), contrasts the case of virtue with the supposition about self-control in
the thought-experiment, that it may be both beneficial and harmful.
‘So, by Zeus, is there any benefit in our other possessions without reason and wisdom?26 Does
it profit a man to have many possessions and to do many things27 if he has no intelligence, or
rather to have and do fewer things, but with intelligence?28 Consider the matter thus. Surely if
he were to do less, he would make fewer mistakes, and if he were to make fewer mistakes he
would do less badly; and if he were to do less badly he would be less wretched?’ ‘Certainly’, he
said. ‘Would a man be more likely to do fewer things when he is poor or when he is rich?’
‘When he is poor.’ ‘When he is weak or when he is strong?’ ‘When he is weak.’ ‘When he is
respected or when he is without respect?’ ‘When he is without respect.’ Would he do fewer
things when he is brave and self-controlled29 or when he is cowardly?’ ‘When he is cowardly.’30
‘And when he is lazy more than when he is busy?’ He agreed. ‘And when he is slow rather than
fast, and short-sighted and dull of hearing rather than sharp-sighted and sharp-eared?’ We
agreed all these things with each other. ‘In short, Cleinias,’ I said, ‘it seems probable that as for
all the things which we said at first were goods,31 the argument32 is not about this—how they
are by nature goods themselves by themselves—but it seems that matters stand thus: if
ignorance leads them, they are greater evils33 than their opposites, to the extent that they are
26
There has been an earlier discussion of the relation between wisdom and good fortune that has
affinities with the Meno; Socrates has concluded at 280a6 that wisdom makes men lucky because wisdom
cannot make mistakes.
27
‘To be busy’, perhaps: the expression does not imply that there are specific products of this activity.
28
Reading ê mallon oliga noun echôn with BT, and resisting Iamblichus’ deletion of noun echôn
(Protrepticus V, 26.1 (Pistelli)). For Iamblichus’ paraphrase of the Euthydemus argument is designed to
show us the instrumental value of wisdom; so he rewrites Plato’s words thus: ‘what use is it to possess many
things and do many things, rather than a few, if it is without intelligence?’ Iamblichus thus reorders the first
three words of the phrase as mallon ê oliga. The MS tradition of the Platonic text, however, reads ê mallon
oliga and invites the concluding participial noun echôn, symmetrically with the previous clause. Gifford
(1905), 24 ad loc., following Iamblichus, is wrong to maintain that ‘in the following argument there is no
place for an antithesis between noun echôn and noun mê echôn but only between polla and oliga’. Were this
to be the case, Socrates would have no warrant for his conclusion about wisdom itself, only for some kind of
claim about its necessity for success: and on any account of what follows he needs at least to show wisdom
to be both necessary and sufficient for success or happiness. The next stage of the argument, it is true,
focuses on the case of the ignorant man, for it is designed to show that none of the putative goods is a good
at all without intelligence, but that with intelligence, the goods we may have are greater, more good than
they would otherwise have been. So in the argument that follows, the contrast is repeatedly made between
different (the one apparently negative, the other apparently positive) characters or situations of the agent
(wealth/poverty, strength/weakness, etc.). The conclusion is that only the intelligence of the agent is
significant. Hawtrey (1981), 85, claims that the addition of noun echôn would ‘merely add an extra
complication’. Caizzi (1996) does not discuss the issue, but follows Burnet. Irwin (1995), 362 n. 12, also
follows the MS.
29
Here, as earlier in the discussion, the virtues are treated on a par with other states of a person that
might initially be thought of as unqualifiedly good (health, wealth, sharp sight). Despite the parallel with
the Meno, Badham (1865) and Gifford (1905) delete kai sôphrôn; Hawtrey (1981) retains it.
30
Cleinias seems to have been betrayed by the sequence of argument here: why should the cowardly
person be more quietist than the brave?
31
Socrates alludes here to various lists in the argument that went before: first, wealth, health, beauty,
good reputation and power, self-control, justice and courage, wisdom and good luck (279a–c); and then in
the argument immediately preceding my quotation, wealth, strength, reputation, courage and self-control,
business, quickness of foot and eye, sharpness of hearing (281b–d). There is something strange about both
lists, in comparison to the Meno list, as I shall suggest further.
32
Sprague’s reading of logos here is ‘correct account’ in Cooper (1997); cf. Hawtrey (1981), 89.
33
The comparatives here emphasize the relativity of these values; relative, that is, to the context or the
character of the agent. This gives rise to the compresence of opposites claim; see section 4.
better able to serve a bad leader; but if intelligence and wisdom lead them, they are greater
goods; but themselves by themselves neither sort of thing is worth anything.’ ‘It seems to be
exactly as you say,’ he said. ‘So what follows for us from what has been said? Surely it is that
nothing else is either good or bad, but these things alone are so—wisdom is good and
ignorance is bad.’ (Euthydemus 281b–e)
The Euthydemus passage is usually construed as a variant of the Meno, on the

grounds that both passages seem to argue for the view that wisdom (knowledge) is
necessary and sufficient for happiness: ‘whatever is done with intelligence is benefi-
cial, whatever without it, harmful’.34
One view of the argument common to both passages would take Socrates to be
asking about wisdom’s utility. Wisdom itself seems to be a good because it is
instrumental to the goodness of all the other goods: hence the conclusion of the
Meno that wisdom is ‘the beneficial’ (89a2). For it seems obvious, in the first place,
that health, wealth, and so on are both beneficial and harmful. If correct use ‘leads
them’ they are beneficial; if not, they are harmful. Next allow that self-control, justice,
and so on are qualities of soul. Suppose that some quality of soul (courage, say) is not
knowledge: in that case, it both harms and benefits. For it harms when it operates
without wisdom, benefits when it operates with it. So for all the endeavours of soul, if
wisdom is the leader they end in happiness, and if ignorance is the leader they end in
the reverse. Qualities of soul, therefore, are not beneficial in themselves, but if and
only if wisdom leads them. But virtue must be beneficial, so (contrary to our earlier
supposition that some quality of the soul cannot be knowledge) virtue must be
knowledge.
Socrates reaches his conclusion, apparently, by supposing that all the other
putative goods considered will harm or benefit according to whether they are ‘led’
by wisdom or ignorance: thus, for example, the reckless person rushing towards
danger in ignorance will come to grief (88b4). So throughout the Meno passage value
seems to be construed in terms of whether something has good or bad results, and
harm and benefit are construed in terms of consequential goods, in terms of whether
they ‘end in happiness’. Accordingly, knowledge will be both necessary and sufficient
for happiness in a purely instrumental sense: we shall end up happy if and only if we
are knowledgeable. How might that conclusion be plausible? Perhaps knowledge is
sufficient for good results because whenever we know what will be a good result, we
always pursue it (and, if knowledge has executive skill, we get it, too).35 And perhaps
knowledge is necessary for good results because ignorance reliably results in disaster
(ignorance completely lacks executive skill). And that thought may be reflected in the
positive conclusion of the Euthydemus passage, that wisdom is the only good,
ignorance the only evil.
34
Assuming, simplistically, of course, that benefit and harm are exclusive and exhaustive.
35
On the assumption, perhaps, that akrasia is impossible.
So if wisdom is a good thing just because it produces good things or just because it
has excellent, happy results, wisdom is instrumentally good. The value it has,
therefore, derives from its consequences, which are understood to be desirable in
themselves (whatever they might be: many theories of this kind need some kind of
hedonism to make it clear that there are consequences of this sort36). Suppose, now,
that something other than wisdom turned out to be an equally reliable instrument to
those same ends (having a fairy godmother, perhaps, or being attached to a well-oiled
pleasure-machine): in that case, there would be nothing to choose between one’s own
wisdom and the convenient interventions of one’s fairy godmother or the switch gear
of the machine. There would, that is to say, be no condition on the best outcome that
it should be achieved through one’s own efforts or endeavours, just so long as an
equally efficient means to the end turns up. If wisdom is instrumental in this sense,
Socrates’ argument is classically consequentialist.37
Yet the Meno argument, albeit clear in structure, is murky in sense. What is it to
‘end in happiness’? Does this describe some extrinsic result of some process, the
cheerful counterpart of its ending in tears? Or might the end be somehow internal to
the process—as Aristotle would after all allow?38 What is it for something to be
‘beneficial’? Does what is beneficial have good results extrinsic to itself, or may it
simply be valuable, in itself? When Socrates asks whether there can be a good
separated from knowledge, then supposes that courage may be other than knowledge,
and concludes that virtue must be knowledge, what does he mean? Does he mean that
we are virtuous if and only if we are knowledgeable, so that virtue and knowledge are
coextensive? Or is there a stronger metaphysics here: does he mean that somehow
virtue is the same thing as knowledge?39 What is it, further, for wisdom to ‘lead’ the
endeavours of soul? Perhaps Socrates’ point is that somehow wisdom or knowledge
controls the means to some good end: so the process of leading is, again, to be
construed instrumentally; but is the metaphor of ‘leading’ as simple as that? We
might demand a reread: the Meno may not be a safe seat for the consequentialist,
after all.40
The Euthydemus, at first glance, might reassure. For that seems, too, to deploy a
consequentialist analysis: wisdom is exactly what produces success, ignorance is
36
See the many utilitarian appeals to hedonism, most famously of all in Bentham.
37
See here e.g. Scheffler (1982), ch. 1.
38
Compare and contrast: I go to the dentist to have healthy teeth; the good of having healthy teeth is
extrinsic to, independent of, the nasty business of going to the dentist, even if having healthy teeth demands
dentistry (could I achieve the teeth without the dentist, I would). But the good of learning about Plato is
intrinsic to, inseparable from, the process of reading him: I cannot do one without the other, nor would
I want to do so. Compare here e.g. Moore (1903), Korsgaard (1983), Rabinowicz and Ronnow-Rasmussen
(2000).
39
Evidently this would need to claim that virtue is more than (some entity over and above?) the
collection of virtuous people, and likewise for knowledge.
40
We may note that what follows, 88d4–89a3, is also indeterminate: does it claim that the soul is the
best guide to what is good, or the source of its goodness? As we shall see, the ‘leading/guiding’ metaphor is a
complex one.
exactly what produces failure; and their value derives from what they produce or fail
to produce.41 So both texts seem to attempt an answer to a question such as ‘what
makes us happy?’—where the notion of ‘what makes us . . .’ seems to be an instru-
mental one. What is more, the two texts are remarkably alike, both verbally and
structurally. The Meno first suggests that it is the leadership of correct use that
produces benefit, and then argues that if the qualities of soul are not knowledge
(epistêmê 88b2) or reason (phronêsis 88b4) then benefit turns up just when intelli-
gence (nous) is present (88b5).42 Then, 88c2, ‘when reason (phronêsis) leads’, it ends
in happiness. This is taken to imply that virtue is reason, and that ‘the things of the
soul’ are not beneficial in themselves, unless reason is added (88c). It is, thus, the
leadership of reason that makes things beneficial (i.e. that makes things end in
goods). So ‘for a man everything else needs to be connected to the soul; and
everything in his soul itself needs to be connected to reason, if he is going to get
goods;43 and on that argument reason is the beneficial’ (88e). The Euthydemus, too,
makes its point via a discussion of correct use, and it has a similarly catholic
vocabulary. Wisdom (sophia) either makes people lucky or makes good luck
unnecessary (280a6, b2–3); correct use is explained in terms of knowledge (epistêmê
281a3), so that it is knowledge that produces good luck as well as faring well (281b3);
no possession is any use without reason and wisdom (phronêsis, sophia, 281b6); men
can do no good without intelligence (nous 281b7); and the conclusion is phrased in
terms of a contrast between wisdom (sophia) and ignorance (amathia 281e5). The
Meno relies heavily on the metaphor of ‘leading’, whether the leader is knowledge or
correct use (88a3, 88c2, 88e1); so too does the Euthydemus (281b1, 281d6, d744). But
both passages amplify this metaphor by talk of the possession or the presence of
wisdom, reason, or intelligence (Meno 88b7–8, c7, Euthydemus 280b2, 281b7). In
each the terminology of value attends heavily to ‘the beneficial and the harmful’. And
both passages conclude with some reflection on what fails to be good itself by itself
(although the Euthydemus, notably, goes further in making a claim about what
succeeds).
So perhaps we should see both passages as variants of each other, expressing
roughly the same point (about the instrumental use of knowledge or wisdom) in
roughly the same language. The similarity between the two passages is to be
explained by the thought that there is a basic thesis that underlies them both;45
41
On Platonic consequentialism in general, see Irwin (1995). If my hypothesis—that the Meno is reread
by the Euthydemus—is correct, then the Meno may be less consequentialist than it seems.
42
Is the point of the rather elaborate set-up to make us contrast an instrumental view (I did it with a
blunt instrument) with something more internal (I did it with malice aforethought)?
43
This clause is ambiguous: it may mean that if everything is connected in this way, then they will be
good; or it may mean (taking tôi anthrôpôi with this clause) that everything must be connected in this way
if goods are going to belong to him (an analogous expression is to be found at Euthyd. 279a3). I shall
suggest below that the indeterminacy of this point is part of the focus of the Euthydemus passage.
44
Here the metaphor is made richer by talk of the servants of the leader.
45
So Hawtrey (1981), 88: ‘the unity of aretê and its identity with knowledge or wisdom.’
and this thesis just happens to be phrased in similar terminology in each case. What
Plato lacks in technical vocabulary—we might suppose—he makes up for in
metaphor.
Perhaps, in that case, we should see both passages as expressions of a ‘Socratic’
account of the relation between knowledge and happiness. Knowledge is the only
reliable instrument to happiness (however happiness is explicated46) and happiness
only arises when we have the reliable instrument to it: knowledge is necessary and
sufficient for happiness. Just so, it might be said, Socrates recommends the ‘measur-
ing art’ in the Protagoras (352–7):47 for it is exactly the art that can measure pleasure.
So—to return to the Euthydemus and the Meno—we might suppose that their
similarity is a mark of this Socratism, of a body of ethical views represented in the
arguments of Plato’s early dialogues.48
3. Consequences and Value

Two thoughts about the Euthydemus, however, might make a prima facie case against
such an account, an account that supposes that the literary connection between the
passages merely emphasizes some doctrine they are thought to hold in common. If,
first, there is a doctrine about the relation between virtue and knowledge lurking
behind both passages, it is a doctrine we might most easily associate with a Socratic
view of action and motivation (such as we seem to find in the Protagoras). But the
view of the conflicted soul to be found in the Republic—standardly thought to be
characteristically Platonic—tells a more complicated story, both about the nature of
motivation and about the relation between knowledge and the ends we in fact pursue.
So if the Euthydemus has any affinity with the Republic,49 or with dialogues later than
the Republic,50 then its use of the Meno’s material is hardly likely to be straightfor-
ward. In the second place, the Euthydemus itself invites scepticism about facile
consequentialism. In the later companion piece to our passage (288d–293a),51
Socrates and his companions52 find themselves trying to specify what science it is
that makes its possessor happy, and are caught in a regress. Every science, they
suggest, shows us how to get goods, but not how to use them: and it is only proper use
46
Again, questions of the relation between hedonism and eudaimonism intrude here when we think in
consequential mode.
47
Of course this passage of the Protagoras may need rereading, too—as many of those puzzled by its
hedonism have allowed.
48
See Vlastos (1991). This account of the influence of Socrates on Plato has, of course, been vividly
contested recently, notably by Kahn (1996) and Annas (1999).
49
Cf. e.g. Narcy (1984), 183; Burnyeat (2002), n. 46; also Hawtrey (1981), 127–8.
50
As I have argued, (1998). But see the more conventional view of Hawtrey (1981), 3–11.
51
The companion piece is the other Socratic episode of the dialogue; each of the Socratic passages is
flanked by two sophistic episodes, giving five episodes of argument altogether.
52
There is considerable by-play about just which companion says what, 291a; this enhances the irony of
the Socratic episodes considered as a pair.
that will give us happiness. In the end, Socrates announces that they have been on a
wild goose chase—or running through a labyrinth where each time they think they
have reached the end of their journey, they find themselves back at the beginning
again (291b–c).53 But the regress affecting their reasoning comes about because they
are unable to specify the object of their search, to say what it is that provides and
brings about happiness (291b6). This failure, in turn, is not so much an ignorance of
the means, but a failure to understand just what happiness is. If, then, the earlier
passage is a reprise of the Meno, and if the Meno argument is construed in a
consequentialist manner, then the later Socratic episode of the Euthydemus directly
undercuts the earlier, and in so doing undermines its predecessor in the Meno.54 Yet
my two passages have remarkable similarities. So what then is the relation between
them after all?
The Euthydemus passage is set in an elaborate context, introduced by a heavily
ironized account of what we might hold to be goods,55 by whose presence we count
ourselves to be doing well (278e–280a). It proceeds to the claim that wisdom (sophia)
includes good luck, because wisdom protects us from making mistakes, and then
suggests (‘we reached the conclusion, I know not how’ 280b1) that if you have
wisdom, you don’t need good luck anyway. The argument then restarts at 280b,
with exactly the Meno’s point: goods that are merely ‘present to us’ are useless unless
they are actually beneficial; and they will not be beneficial without correct use (280b–
281b). So knowledge (epistêmê) does not just provide good luck, it provides doing
well (eupragia), for none of our possessions are of any use without reason and
wisdom. There follows a short argument about relative benefits and harms (281b–
c). Then Socrates concludes with our passage: none of the putative goods is worth
anything in itself; but wisdom is the good, ignorance the bad. On a cursory reading,
we might indeed conclude that the two arguments are the same.
This apparent congruence of the two passages, however, may mask a more
complex relation. If the Meno asks ‘what makes us happy?’, we could read the
question instrumentally: what promotes this end, happiness?56 Or the question
may be differently understood: suppose we treat ‘happiness’ as a place-holder for
53
The complaints of regress or circularity commonly beset consequentialist reasoning: if you suppose
that some specified set, or type, of goods are just good, just desirable, in themselves, what makes them so?
An appeal to virtue, or character, of function is designed to pre-empt those complaints by focusing on a
richer account of the agent, rather than on setting his ends independently (see here e.g. Anscombe (1958),
Foot (1978)). The danger is that ‘virtue ethics’ often seems rather vague (although see Hursthouse (1999)).
Here Socrates points up the worry about specifying the end of our endeavours by showing that the same
kind of regressive feature besets arguments about such endeavours.
54
Point taken by Annas (1994). Cf. here Striker (1996b), 318, who takes the two Socratic episodes to be
roughly continuous. I think this is not so and that the second episode treats some of the assumptions of the
first in different, and critical, ways.
55
Notice, for example, the cagey way in which the virtues are introduced at 279b4 ff., or the alacrity
with which Socrates—Socrates?—suggests that wealth is universally acknowledged to be a good, without
himself endorsing the universal view, 279a7.
56
This seems to be how Sprague takes the passage (1962), 9–10.
whatever we shall take to be valuable in a life,57 then ‘what makes us happy’ may
inquire into the nature, and the source, of value itself: what is happiness, how is it
constituted?58 If we ask the Meno this different question, ‘what makes something
valuable?’, we might expect two different kinds of answer. On the one hand,
something may be instrumentally valuable, because it produces happiness; on the
other, something may be non-instrumentally valuable—because it is the end to which
the instrumental values lead. The argument itself seems to sustain the view that
wisdom is instrumentally valuable (and since it is both necessary and sufficient for
happiness, wisdom may be the only instrumental value) but not that it is non-
instrumentally valuable (there is nothing here, for example, to show that wisdom is
in fact constitutive of happiness, or to show that it is valuable anyway, irrespective of
its relation to happiness). The value of wisdom, on that account, is derivative from its
consequences, and in this sense is a consequential value. What is more, although this
argument seems to show that the possession of those consequences requires wisdom,
it seems to do nothing to show that the value of those consequences derives from
wisdom (for while health and wealth do in fact convey benefit because of wisdom, the
value of the benefit they convey derives from their consequences). In this respect,
therefore, the Meno seems again to have affinities with modern consequentialist
arguments, which explain the good in terms of states of affairs, and are (or are in
their simpler forms59) indifferent to how those states of affairs are brought about, just
so long as they are brought about in the most efficient and prolific way. There is
nothing, on such an account, to choose between two different means to some desired
end except by virtue of their efficiency. So, for the Meno argument, were it to turn out
false that wisdom is good at producing the right kind of outcome, there would be no
residual reason to be wise. What is more, and considerations of efficiency aside, the
Meno should be indifferent in particular to whether those states of affairs are brought
about by the agent who enjoys them, or not brought about by him but merely enjoyed
by him. If it were to turn out that being born at the rising of the dog-star is a more
efficient way of getting happiness than being wise, then we should desert the
philosophy schools altogether. Consequentialism is not, as the modern debates put
it, agent-relative. According to the consequentialist, I should value wisdom because it
is good at producing the desired consequences, not because there is something
important about the fact that this wisdom belongs to me, or because there is
something pertinent to my happiness in the fact that I myself am its instrument.60
So a consequentialist view is inclined towards instrumentalism, too: any instrument
is valuable insofar as it produces the consequences we desire.
57
Ancient eudaimonism is especially about lives: how best to live? See Williams (1985).
58
This point is taken by Canto (1987), 132 and 276n88.
59
But notice here modern attempts to patch consequentialism up: e.g. Scheffler (1982).
60
To make this point we need to distinguish between whether I am the agent of my own happiness and
whether I am its patient. The consequentialist view does not, of course, insist that the happiness is not
enjoyed by the person in question: but this is merely, if you like, a locative claim.
So what makes the goods that constitute success valuable? That question invites an
answer, not about the means to success, or even about the necessary and sufficient
conditions of happiness, but rather about the source of value itself.61 (If I want to play
at Robinson Crusoe, and this aeroplane is both necessary and sufficient to get me to
the requisite tropical island, the aeroplane has value derivative from the value of the
consequences it provides. When I arrive at my island, only to be smitten with a grim
sense of the meaninglessness of this desert existence, my misery will not be explained
by virtue of the aeroplane that got me there, or merely by virtue of the fact that the
aeroplane has flown away.) And that question presses on the consequentialist, who
needs to provide a robust account of the goods that constitute our ends (which is why
he often finds hedonism so attractive). The Euthydemus’ argument suggests that this
question, ‘what makes these goods valuable?’, is the one we should ask.62
Consider, first, an argument that appears in the Euthydemus without any coun-
terpart in the Meno. At 280e6 Socrates suggests that if someone uses something
badly, he is worse off than if he had left it alone. This introduces a relative consid-
eration, which is amplified at 281b–c. Is someone better off with more possessions
but no wisdom, or with wisdom and few goods? The sequence of argument that
follows seems to espouse a kind of quietism: we are better off doing less if we are
stupid, because we can make fewer mistakes, and the fewer mistakes we make, the less
wretched we shall be. So far, so consequential. It then follows that the person who
lacks the putative goods (wealth, strength, reputation, even courage) will do better
than the person who has them, but no wisdom. It is taken to follow from this that ‘if
ignorance leads them, they are greater evils than their opposites, to the extent that
they are better able to serve a bad leader; but if intelligence and wisdom lead them,
they are greater goods; but themselves by themselves neither sort of thing is worth
anything’ (281d). Now the focus of attention seems to have changed. For we are not
now thinking about how beneficial or harmful these things are (things we originally
thought were uncontroversial goods), or how far they will reliably issue in the
happiness we pursue. Instead, we are asked to consider their value in themselves:
and Socrates insists that they have none, but that their value is derivative from what
leads them: if that be wisdom, they are good, if that be ignorance, evil.
How might this be thought to follow from the case of the blind man who is saved
from the consequences of his own folly by being unable to see what he is doing
(281d1)? The argument does its work, as the conclusion makes clear, by inviting a
comparison between the wise man who has access to few assets,63 and the foolish one
61
Cf. Korsgaard (1983).
62
Indeed, the puzzle of the second ‘Socratic’ episode—the labyrinthine way in which the ends of our
actions recede from us as we pursue them—makes this point too.
63
Rejecting Iamblichus’ reading of 281b7; see n. 28. Annas (1999), 40 ff., follows Iamblichus, but reads
the point as ‘that without wisdom you are actually better off with conventional evils’, since you will be less
able to abuse what you have, less able wrongly to exploit what are conventionally thought of as advantages.
Consequently, the value of putative goods is dependent on the life into which they are incorporated. I find
who has access to many. How are we to assess, or to compute, the relative value of the
two situations (the question of such computation is one that the consequentialist
keenly asks, as Plato himself knew, cf. Protagoras 351 ff.)? After all, even if the foolish
man squanders most of his wealth, there might be enough in his pocket to weigh
against the small change of the wise man. Even if he squanders it all, is he any worse
off than the wise man who started with nothing and ends with the same amount? If
wisdom is valuable instrumentally, and has no scope for action, it will be no more
and no less valuable than folly, which is instrumentally useless. Why then should
wisdom and folly make the difference in cases like this? The answer Socrates supplies
only follows if wisdom and folly are now conceived not as the instruments to
maximizing our enjoyment of goods, but as themselves the source of their value,
and also as valuable in themselves. Then if the foolish man has no matter how much
more loot than the wise one, but the very goodness of the loot depends on the wise
man’s wisdom, its evil on the foolish man’s folly, then no matter how little he has,
even nothing at all, the wise man comes off best.
The Euthydemus alerts us to this conclusion by the way it began its inquiry into the
nature of ‘doing well’. It is absurd, Socrates suggests, to ask whether all men want to
‘do well’—for of course everyone wants that. Socrates’ use of the famously opaque
expression ‘doing well’ sets the agenda for what follows, not by telling us just what it
means, but rather by suggesting some scepticism about what it might be taken to
mean. For if we all want to do well, how exactly will this happen? Is it by having many
goods? This question is even more absurd, Socrates says, for it is obviously so. Now
Socrates offers a list of putative goods, starting with wealth, health, and beautiful
bodies, continuing to good birth and political power, including more dubiously self-
control, justice, and courage, then wisdom and finally, after a long flourish, good
luck. The list should give us pause.64 Is this Socrates, his old disreputable self (notice
the repeated issue of Socrates’ aged obstinacy, e.g. at 272c, 287b), who is extolling
the importance of wealth? Or Socrates, who turns some of his best irony on the
importance of an impressive genealogy (here of Cleinias at 275a–b, in the same
breath as describing the fear lest he be corrupted), advocating the best life of the élite?
Or Socrates who thinks that the importance of wisdom is the way it secures good
luck?65
Annas’ overall interpretation sympathetic, as will become clear. However, Iamblichus’ reading allows the
conclusion that the problem is merely an executive one—of being properly guided in correct use. This is
too weak to support the further conclusion that there is something—wisdom—which is good ‘itself by
itself ’. Contrariwise, the thought that the wise man is better off with nothing than the ignorant man with a
great deal does give warrant to the conclusion: as, I shall suggest, the Stoic readers of this passage saw.
64
Even as we read it for the first time; of course the conclusion of the first phase of the argument at
280d–e justifies our unease.
65
Here, 280a, we might suppose that Socrates’ interlocutors mean something rather different by the
connection between good luck and wisdom than Socrates does. Once again the text is ironical, in the sense
that it invites the reader to see the gap between what Socrates says and what his interlocutors understand.
Consider the conclusion again. Socrates claims, first, that the value of the items on
his original list depends on the value of what ‘leads’ them (that claim, of course, is
compatible with a consequentialist analysis, that they are only valuable, e.g., in this
particular assemblage, intelligently arranged): so they are not valuable ‘themselves by
themselves’. So far, so compatible with the Meno, and so far in the Meno’s words. But
he suggests, second, that the items on the original list have no value, while wisdom,
which ‘leads them’, is the only good. This claim seems markedly different from the
Meno’s offering: that ‘all the qualities of the soul are neither beneficial nor harmful in
themselves’ (88c6), and that wisdom is ‘the beneficial’ (89a1–2). For the Meno still
allows for a consequentialist account, still may be asking the instrumental question:
‘how can we do best?’. The Euthydemus, by contrast, in asking ‘which things are
goods (after all)?’, undermines the list of putative goods,66 and attacks the conse-
quentialist view that our ends can easily be specified. As a result, it shifts our
attention to the deeper—metaphysical—question: ‘what is the source of value?’.
4. ‘Good Itself by Itself ’

So does the Euthydemus, so far from confirming the consequentialist view of the
Meno, actually make a quite different point? And if it does, why does Plato choose
later to reuse the Meno’s terminology and argumentative structure? Had he just run
out of new ideas—was he forced thus to recycle the old? On the contrary, I suggest
that the similarity between the two passages has a direct philosophical purpose. After
all, the Meno passage is short and condensed; and it is a matter of interpretation to
elicit from it the consequentialist view I have outlined: for example, by construing the
assumption that virtue must be beneficial instrumentally to generate the conclusion
that virtue is knowledge; or, as I suggested above, in taking the relation between
virtue and knowledge as a material equivalence rather than as some stronger identity
claim.67 In reading the Meno passage, that is to say, we need a bit more help than the
bare text gives us. That help is forthcoming, I suggest, from the Euthydemus, which
invites us to rethink the instrumental account of the Meno, and to think in a different
way about the nature of value. To invite this rereading, Plato uses similar vocabulary
66
Both by the ironical way in which they are discussed, and through the accusation of circularity of the
second Socratic episode.
67
Consider the following, permitted within a consequentialist analysis. Suppose that virtue is maxi-
mizing pleasure for myself. Knowledge is necessary and sufficient for virtue just because whenever I know
what will be pleasant, I grab it, and because I can only grab a pleasure (pleasures being the elusive things
they are) when I know what it is. So I am virtuous if and only if I am knowledgeable—but what it is to be
virtuous is defined in terms of grabbing, even of enjoying pleasures; what it is to be knowledgeable in terms
of spotting the pleasures that are out there to be grabbed. Knowledge and virtue are here materially
equivalent. Contrariwise, I might think that virtue is not so much a felicific maximizer, but a state of soul,
some quality of the person who has it over and above its executive advantages. In that case, I might
suppose, for example, that virtue has the same psychological dimensions as knowledge, so that virtue and
knowledge are in fact two different descriptions of the same thing: the best state of soul. In that case, my
identity claim is stronger than a material equivalence.
and argumentative structure: but not from either intellectual poverty or from parsi-
mony, but as an explicit reflection of one passage on the other. This kind of
reflectiveness is itself a mark of a philosophical approach: if, in rereading each
passage, we compare them and contrast, we may by that very process be brought
to ask the higher-order question: ‘how are we to think about the source of value?’.
Even so, the Euthydemus’ conclusion is vague.68 It could mean one of two things:69
a) nothing has any value at all except wisdom. Wisdom is the only valuable thing.
[I shall call this the exclusive view.]70
b) wisdom is the only thing that has value ‘itself by itself ’; all the other valuable
items are so derivatively from wisdom. [I shall call this the derivative view.]71
Either goes beyond any sort of consequentialist or instrumentalist account. a) takes
Socrates’ remark at Euthydemus 281e4–5 literally (as ‘nothing is good or bad at all
except wisdom and ignorance’). b) takes it to be an ellipse (for ‘nothing else is either
good or bad in itself except wisdom and ignorance, which are good and bad
themselves by themselves’, which allows for the possibility that other things may
be good, but not themselves by themselves).72 In both cases Socrates’ claim is not the
negative one we find in the Meno (‘nothing on the list is good or bad on its own’,
which neither implies that wisdom is the only good, nor even that wisdom is good
itself by itself;73 after all, there may be nothing that is just good), but rather a positive
claim about what value really is: wisdom is the only good (itself by itself ).74
This conclusion is derived from the thought that things other than wisdom—the
putative goods, wealth, health, and so forth—may be characterized in opposite ways,
since they may either harm or benefit. In general these putative goods suffer from
what is often described as the ‘compresence of opposites’: because an individual case
68
In my opinion, deliberately so.
69
See Vlastos (1991). Vlastos, however, takes the view that the text is both determinate in its meaning
(which I deny, since I suppose that the ellipse is deliberate) and moderate in its sentiment (and thus
compatible with the Meno’s view).
70
This could be expressed, although it is not apparently so in this passage, as a contrast between an
objective view (wisdom is the only thing that is really, genuinely, valuable) and a subjective one (all the
other things seem valuable, but in fact they are not). That contrast could be emphasized by the way in
which the putative list is set up in terms of who believes such and such an item to be valuable (although
I myself think that Socrates’ strategy there is quite different).
71
This view is entirely compatible with an objectivist account of value: both what is valuable itself by
itself and what is valuable derivatively from it may be objectively so.
72
Cf. here Vlastos’ remarks (1991) and Long’s disagreement (1996), 26.
73
What may be an identity claim at 89a1, ‘according to this argument wisdom would be the beneficial’,
does not imply that either. The beneficial nature of wisdom here may be its being a necessary and even
sufficient condition for all other valuables—this still allows for the possibility that its value may derive from
its relation to them.
74
What should we say about the unclarity, or the elliptical nature, of Socrates’ conclusion? In the
context of the elaborate composition of the Euthydemus, and especially with a view to the critique of some
of the assumptions of the first Socratic episode in the second, I am of the view that the unclarity is
deliberate: to expose, as the argument progresses, the different assumptions we must make along the way.
of wealth may harm or benefit, wealth in general both harms and benefits.75 This may
then prompt us to look for what is valuable in itself—or ‘itself by itself ’, as Plato often
expresses it.76 Why would we embark on that search? We may be driven to find what
is valuable in itself because we suppose that wealth, and the other putative goods,
because they both harm and benefit, lose their claim to being truly, properly, good. If
that is why we search for what is good in itself, we may incline towards an exclusive
view of this argument (we may be searching, that is to say, for perfection). Or we may
suppose that the fact that wealth, health, and the others suffer from the compresence
of opposites implies that they do not explain, by themselves, why we value them. In
that case we may search for an explanation, or a source, of their value: and this will
incline us towards the derivative view of this argument.
But in either case, what is required from knowledge or wisdom is not so much a
calculus of what to do in some particular situation (as the consequentialist reading of
the Meno suggests) but rather an account of what it is for any of the objects of choice
to be valuable at all. If wisdom is what is good itself by itself, then—whether
exclusively or derivatively—it explains value. It does that, as the final argument of
the Euthydemus makes clear, not pragmatically77 (by merely allowing us to under-
stand the messy chaos of the world and its values), but in reality, by being the source
of the value itself. For only thus could Plato justify the argument that so puzzled
Iamblichus.
Does it profit a man to have many possessions and to do many things if he has no intelligence,
or rather to have and do fewer things, but with intelligence? Consider the matter thus. Surely if
he were to do less, he would make fewer mistakes, and if he were to make fewer mistakes he
would do less badly; and if he were to do less badly he would be less wretched? (281b7–c3)
Consider what is happening here. One construal of this suggests that the argument
turns on the idea that the stupid man makes mistakes, reliably and predictably, by
virtue of his stupidity. Making mistakes is a bad thing, because it is harmful; and so it
results in the stupid man doing badly and being unhappy. So we are to compare the
idiot millionaire who has huge opportunities for following his misguided principles
with the ignorant pauper who has none; just because the millionaire has executive
capacity, he has the scope to make a lifetime of mistakes, from which poverty protects
his counterpart. The wise man is a pauper, too; he too has few assets, but the wisdom
to use them well: so he is in for a lifetime of doing well, and better off than either of
his ignorant counterparts. But of course the assets that each of these characters is
75
This argument form is familiar from elsewhere in Plato. It is worth noting that in the case of value
Plato moves from the claim that ethical tokens may be either valuable or harmful to the claim that the type
is both valuable and harmful. For other examples, such as relations, he suggests that the token suffers from
compresence directly: Cebes is taller than Socrates and smaller than Simmias, so Cebes is both tall and
small (Plato does not, of course, make the mistake of supposing this to be a contradiction). Cf. here e.g.
Phaedo 102b ff.
76 77
E.g. Phaedo 78d. See Ruben (1991).
using or abusing are themselves members of the list of putative goods. So we are to
imagine each character with their putative goods: the millionaire has a million units
of wealth; the pauper and the wise man only one. The millionaire will abuse all
million; the ignorant pauper will abuse his one; and the wise man will benefit from
his singularity. The millionaire, then, has an immense capacity to make mistakes, the
pauper hardly any. But why then should we prefer to be one rather than the other? If
a mistake is simply a failure to get an asset (as the consequential view suggests), there
is nothing to choose between them; but if a mistake is itself a harm, then clearly the
millionaire is by far the worse off. Likewise if the choice between the millionaire and
the wise man is simply that the millionaire repeatedly fails to get things right, where
the wise man does get things right, but only on his one, singular occasion, we might
not suppose that the wise man is markedly better off than the millionaire. If, on the
other hand, the mistakes are themselves harms (rather than mere failures to get
goods), then the choice is an easy one: the millionaire is the worst off, the pauper
next, and the wise man is the best off of all. But this will only be the result of the
calculation if the mistakes are themselves items of value or the reverse; and this in
turn will only be a reasonable assumption if we suppose not that wisdom confers
executive power, but that it confers value itself. By virtue of a man’s rational
capacities, that is to say, his actions are valued: wisdom is a good itself by itself
because it is the source of value.
Why then should we say that the stupid man is better off with fewer assets because
he has ‘fewer servants of a wicked leader’? What exactly are we to make of the
metaphor of leadership? The consequential view is made tempting by what we might
call the executive image: the leader is the person who gives the orders, the designer of
the plan, the artificer of the cunning scheme she hands down to her subordinates to
secure those goods that are already and independently valued. But there is a different
image of leadership, which we might call regal. The king organizes his court, decides
who shall be in favour and who shall be cast into the outer darkness; it is the king who
appoints one courtier to be master of the bedchamber, and denies another any access
to his royal person at all. There are no independent goods, on this view: the value is
all within the structure of the court. We may imagine here, also, two different images
of a mistake. A mistake, on the executive model, may be simply a bust machine, a
broken instrument for achieving the ends set before the leader, an incompetent
executive in the owner’s employ. Still on this view the ends are pre-set; the instru-
ments are valuable insofar as they achieve them. Or, on the regal model, we may
imagine the disappointment of the old career courtier whose value becomes less and
less as he is passed over for the younger and more ambitious men. Here the ends are
inextricable from the value conferred by the king himself: the courtier’s life depends
for its value on the king, and not on the ends it achieves.
The conclusion of the Euthydemus turns away from the executive, and favours the
regal model of leadership. And in doing so, it shifts away from consequentialism. It
still insists, however, that the source of the value of other things (what ‘leads them’)—
not now construed as the end of some action, independently specified—is itself
valuable. And that value is intrinsic: that is to say, its value is not dependent on
anything else.78 That might allow two further claims for the Euthydemus argument:
c) what explains value is the source of value;79
d) what is the source of value is intrinsically valuable;
claims, once again, which are missing from the apparently consequentialist account
of the Meno.80 One purpose of the similarities and differences between the two
passages may be to invite reflection on just this point.
In the Meno, Socrates’ argument can go through on the basis of two sorts of
consideration: of extrinsic value, which is dependent on something else (hence, e.g.,
the implication at 87d that no good is separate from knowledge: a claim that does not
imply that knowledge is separately, that is intrinsically, good); and of instrumental
value, value that is derived from the achievement of something else valuable. These
considerations are not the same. Take, for example, health: to achieve, or to use in the
most effective way, a healthy constitution, we need wisdom: health, therefore, is
actually valuable for us only in the presence of something else, namely wisdom;
health is extrinsically valuable.81 But wisdom, conversely, may be valuable because of
this instrumental contribution it makes to something else: the wisdom that provides
health and makes it useful is itself valuable by virtue of the good thing it provides—
namely health. In the Euthydemus the situation is different: here (if I read the
conclusion aright) the contrast is between items of extrinsic value, derivative from
something else, and intrinsic value, a solitary item that is valuable in itself. And here
the extrinsically valuable depends for its value on the intrinsically valuable; but the
intrinsically valuable is valuable itself by itself.
78
Again, it seems to be a commonplace of Plato’s account of explanation that what is extrinsically thus
and so is explained by what is intrinsically thus and so, especially when it comes to evaluative properties
(see Moore (1903), Korsgaard (1983)). So intrinsic value will be independent, extrinsic value dependent on
something else for its value; and intrinsic value will thus be, at least from the explanatory point of view,
prior. The same pattern seems to be urged for other cases where Plato postulates a form; see here the claims
about the ‘separation’ of forms at Parmenides 129a ff.
79
Just as what explains equality is the source of equality, and the same for largeness, etc.: see Phaedo
100d. As elsewhere in Plato, this is a realist account of explanation: the point made in the Euthydemus is
not about how we see value, but rather about the way value is actually constructed and derived, out there in
the world.
80
Again, one may see this by reflecting on the conclusion of the Meno’s argument. If the argument
shows that wisdom is the beneficial, it seems to have done so by showing that all other benefits depend on
it; but its beneficial nature may equally depend on them, since its beneficial nature derives from its capacity
to collect other goods together. This mutual dependency is not vicious, but it does rule out the claim that
wisdom is intrinsically valuable. The Euthydemus, on the other hand, is committed to the claim that
wisdom alone is valuable itself by itself. In doing so it takes wisdom to be intrinsically valuable; and then its
value cannot be derivative from the consequences it may contrive.
81
Of course, that there are extrinsic values (values dependent on something else to be valuable) does not
imply that there are intrinsic ones: there may be nothing that is independently valuable; instead everything
may be meshed together in mutual dependence.
This conclusion has two important consequences for the account of value to be
found in these pages of the Euthydemus. First, the relation between what is valuable
itself by itself and what is valuable by virtue of it is an explanatory relation between
two real things. This explanatory relation supports an answer to the question, ‘why
do we value this?’. Second, at least on the derivative view, goods are thus arranged
hierarchically: what is intrinsically good is so absolutely, what is extrinsically good is
so relatively to the intrinsic good.82 But—because the one explains the other—they
are valuable on the same scale. In cases such as this, therefore, the expression ‘good’ is
not used ambiguously: these are goods of the same sort, even when one transcends
the other. This will be true whichever account of the conclusion of the Euthydemus’
argument we take. On the exclusive reading, since nothing else than wisdom is
intrinsically good, nothing else is good (in the same sense) at all; on the derivative
reading, wisdom is intrinsically good, the other things good (in the same sense, but)
relative to it.83
So, I suggest, the Euthydemus prompts us to reread, or to rethink, the equivalent
argument in the Meno, specifically by the linguistic and argumentative similarities
between the two passages. The purpose of the Euthydemus’ complex allusiveness may
merely be to re-examine the condensed argument of the Meno. Or it may be to
provide a Platonic revision of a Socratic account of wisdom and happiness. But, if my
analysis of the differences between the appearances of the Meno and the account of
the good to be found in the Euthydemus is accurate, then the point of the allusion is
to invite a deeper consideration of the problems of consequentialism. The Meno lacks
a good account of just how value is to be located in the agent, and remains unclear
about whether this importance of the agent in ethical theory is itself merely executive.
The Euthydemus, by contrast, invites us to think of the agent as central to the account
of the intrinsic good, just because it is the agent who is wise or knowledgeable, and
because wisdom and knowledge are the sources of value itself. The difference
between them may express not so much the inadequacy of the earlier text and its
later revision, but the opacity of the Meno, elucidated in the Euthydemus. Conse-
quently, the rereading of the Meno by the Euthydemus may be not so much a
corrective as a provocation to reflect.
5. Stoic Indifferents
Recall, now, the Stoic theory of the ethical indifferents. This theory may indeed have
Socratic antecedents—but it is the Socrates of the Euthydemus who provides them.
For it is the Euthydemus’ account of the intrinsic good (rather than the theory that
seems to be presented in the Meno) that seems to lie behind the Stoic contrast
82
This relativity does not make the extrinsic good subjective, even if the intrinsic good is somehow
cognitive, like wisdom or knowledge.
83
Relativity does not introduce equivocation.
between virtue and the indifferents. Only the Euthydemus directly makes a strong
claim about what is good itself by itself (compare Cicero’s report, ‘only what is right is
good’). But now here is the problem. If this passage is what influenced the Stoics, it
appears to be, after all, inconsistent with what they claimed about the relation
between what is good and what is indifferent. For that relation is often construed
as one between two quite different sorts of value: moral value, on the one hand—
virtue—and the non-moral value of health, wealth, and reputation on the other.
I shall call this evaluative dualism.
They expressed this thesis by restricting ‘good’ to what is morally excellent and ‘bad’ to the
opposite of this, and termed everything which makes no difference to happiness or unhappi-
ness ‘indifferent’.84
This is often compared to the Kantian distinction85 between those actions that are
categorically enjoined, and those that are pressed on us by mere hypothetical
imperatives: the former are unconditioned, the latter conditional on our desires for
their result. And the dualism of this distinction is reflected in the difference between
the sources of the two values: on the one hand the hypothetical imperative is derived
from the practicality of desire, and on the other
. . . there is an imperative which, without being based on, and conditioned by, any further
purpose to be attained by a certain line of conduct, enjoins this conduct immediately. This
imperative is categorical. It is concerned, not with the matter of the action and its presumed
results, but with its form and with the principles from which it follows; and what is essentially
good in the action consists in the mental disposition, let the consequences be what they may.
This imperative may be called the imperative of morality. (Kant, Groundwork of the Meta-
physic of Morals, 4: 416–43)86
One way of understanding this contrast may be in terms of what we want to do, and
what we feel we ought to do. And this contrast itself is often glossed in terms of two
quite different psychological drives—on the one hand egoism, the demands of the
greedy self, and on the other altruism, the demands of others upon us.87 Even if we
leave on one side the question whether this contrast is itself spurious or grossly
simplistic,88 we may find ourselves dubious about its ancient antecedents.89
First of all, it is not to be found in whatever relation between what is good ‘itself by
itself ’ and the putative goods in either the Meno or the Euthydemus. For whether we
have here an exclusive theory (what is good itself by itself is the only good) or a
derivative one (all other goods are derivative from what is good itself by itself ), the
84
Long and Sedley (1987), vol. I, 357.
85
Kant himself was apparently heavily influenced by the Stoics, cf. here Schneewind (1996).
86
Trans. H. J. Paton (1964).
87
Kant himself is not so crude; see e.g. Critique of Practical Reason and compare Korsgaard’s defence of
Kant (1996).
88
This issue, of course, is a central one in ethics, and the literature too massive to cite in detail here.
89
Consider here MacIntyre (1981).
good itself by itself and the putative goods, if they are good at all, must be good in the
same sense—if there is, after all, to be some explanatory relation between the two.
Secondly, and consequently, if the Stoic theory is dualist—if the Stoics are com-
mitted to an account of value that quite separates the values attached to virtue and
morality from the prudential values associated with common lists of goods—then
should we say either that the Euthydemus is not what influenced them, or that they
misread the arguments they found in the Platonic text? I shall argue that we should
prefer a third option: that the earliest Stoics were indeed influenced by Plato, but that
they were not evaluative dualists at all.90
It is undeniable that the earliest Stoics were readers of Plato. Zeno was a pupil of
both Crates the Cynic and the leaders of the Academy, Xenocrates and Polemo
(D.L. VII. 1); and Aristo was a pupil of Zeno, until he became heretical.91 What is
more, the influence of the Euthydemus itself hangs heavy on the later doxographical
tradition. Diogenes Laertius cites the Euthydemus’ hostility to the ‘sticky arguments’
(glischrologia) of the sophists (II. 30) and later (II. 32) mentions Socrates’ late
learning of the lyre—an activity of which Socrates makes a great deal in the
Euthydemus.92 In between he outlines what he takes to be Socrates’ ethical principles;
and the context, once again, suggests that these principles are taken from the
Euthydemus:
He said, too, that there is only one good, knowledge (epistêmê), and only one evil, ignorance
(amathia). Wealth and noble birth have nothing high-and-mighty about them,93 but quite the
reverse, evil. (II. 31)
Just as Socrates’ influence is characterized by Diogenes in Euthydeman mode, we

may see, I suggest, the same influence at work on Zeno, and on Aristo as he espoused
the heresy that caused the split between him and his master. The issue throughout is
the theory of indifferents.
90
Here contra Long’s explicit claim (1996), 32. It may be that later Stoicism, more concerned with an
account of the nature of the physical world, could arrive at some kind of dualism by separating the
imperatives derived from nature as a whole and those derived from human needs and drives. Such an
account, however, would run the risk of imposing on the Stoics a psychological dualism to which they were
opposed. The strength of the connection between developed Stoic ethics and Stoic physics is downplayed
by Annas (1993), but not by Striker (1996a). See also Frede (1999).
91
See here Schofield (1984), on the question from whom the schism was made.
92
Diogenes misses its ironic significance, however. At 272c Socrates mentions his lessons with Connus
and suggests that they are liable to make both him and his teacher ridiculous. Diogenes, however, reports
that Socrates declared there to be nothing absurd in someone learning what they do not know. He evidently
garbles Socrates’ music lessons with the first sophistic episode; but that he is thinking of the Euthydemus is
undeniable.
93
Diogenes uses the word semnon here to describe what wealth lacks: I take it that it is no coincidence
that the inclusion of wealth in the list of goods at 279b6 is something that it doesn’t need a high-and mighty
person (semnos anêr) to provide. It is easy to see how Diogenes, reading the conclusion of the argument
back into the disingenuous list, would come up with the claim he makes about wealth. That he uses the
same word signifies, once again, that his account is derived from the Euthydemus, and not from the Meno;
contrast the bland terminology of Meno 87e.
First of all, there is some dispute about the origin of the theory: Diogenes suggests
(VII. 37) that Aristo ‘brought in indifference’. But both Cicero (de finibus III. 51) and
Stobaeus attribute the coinage of the contrast between ‘preferred’ and ‘dispreferred’
indifferents to Zeno; and Sextus (Math. 11. 64) suggests that Aristo’s heresy was to
reject Zeno’s contrast. That evidence might suggest it was Zeno who first proposed
the contrast between goods and indifferents, but that he qualified the indifferents in a
way rejected by his breakaway pupil Aristo. But there are two separate issues here:
first, the postulate that what seems to the common view to be an item of value is in
fact indifferent; second, the attitude that we should therefore cultivate towards these
items: indifference. It is this attitude of indifference that Aristo recommends, and in
this sense that he ‘brought in indifference’. Now he could not do that without
subscribing to a strong account of the indifferents, that is, that they are strictly
indifferent (hence Diogenes’ report, ‘for without exception things indifferent as
between virtue and vice have no difference at all’).94 But his objective seems to
have been more prescriptive, to claim that the end is living ‘indifferently disposed
towards the things which are in between virtue and vice’ (D.L. VII. 160). This may
have been a recommendation to living haphazardly, doing whatever occurred (cf.
Cicero Fin. IV. 79); or it may have been an account of what virtue in fact is.95
A different account of what the sources give us would be not that Aristo identified
indifference and virtue,96 but that he had a two-fold account of the end: we should
live virtuously; and as far as indifferent things are concerned, we should cultivate
indifference towards them. Such an attitude towards indifferent things, indeed, might
be a part of what it is to be virtuous, or even a consequence of the intrinsic goodness
of virtue; but what it is to be virtuous is independently explained—as the intrinsic
good of wisdom.97 If that is right, Aristo is exempt from the charge of vicious
circularity, because he supposes that the account of virtue is primary and that virtue
is the primary good. But because he takes this primary good to be the solitary good,
too, then there are no extrinsic goods dependent upon it. In that case, the attitude of
indifference is how we should live, in consequence of the nature of virtue.
If this was Aristo’s view, it fits well the exclusive reading of the Euthydemus
argument:
a) nothing has any value at all except wisdom; wisdom is the only valuable thing.
94
Supposing, realistically, that it is the nature of the things out there that determines our attitudes to
them: hence, if indifference is appropriate, that must be because the indifferents are indifferent. This sort of
realism is commonplace, of course, in ancient ethical theory.
95
Here cf. Striker (1996a), who sees this as the focus of Plutarch’s objection (de comm.not. 1071F–
1072A) that Aristo’s theory is viciously circular.
96
Cicero seems to suggest both that Aristo said that the highest good was living indifferently (e.g. at
Acad. II. 130) and that he said that living virtuously was the end (Fin. IV. 43) without drawing Plutarch’s
conclusion, that he identified living indifferently and living virtuously.
97
My suspicion is that the doxographers conflated these two aspects of the end just because they did not
always acknowledge that Aristo’s heterodoxy was not his theory of value, but his recommendation to
cultivate indifference.
Aristo’s argument to that conclusion (reported by Sextus, Math. 11. 64–7) is based on
the thought that any ‘preferred indifferent’ may under different circumstances turn
out to be dispreferred—so that nothing is unconditionally preferred. He infers that
nothing is preferred at all.98 And this argument may be what we find in the
Euthydemus: for on the exclusive view Socrates argues that if none of his list of
putative goods are valuable unconditionally, then they are not valuable at all (com-
pare Cicero, Academica II. 130).
But all the same, it is not obvious that Aristo’s position is well founded. If nothing
but virtue is unconditionally good, he seems to suggest, then everything but virtue is
indifferent, and should be treated with indifference.99 The best life, therefore, will be
the life of indifference towards ‘everything which is in between virtue and vice’
(D.L. VII. 160) on the grounds that these in-between things have no distinction
between them, but are exactly on a par with each other. We might complain that the
notion of what is ‘in between’ merely indicates that such things are context-relative. It
neither implies that what is in between has no value at all nor suggests that the
appropriate attitude to these items is indifference. But Cicero (Fin. IV. 69) suggests
that Aristo’s focus of attention was on the practical issues involved in dealing with
things that are only valuable in some contexts. And if what we pursue is thus variable,
Cicero has Aristo suggest, our choices should be made haphazard, with no expect-
ation of the fixity of value of what we choose.100 And this suggestion too has its
Platonic origins: if whatever is not good itself by itself is unknowable, dangerously
unreliable from the point of view of cognition, perhaps there is no reasonable
approach to adopt towards the practical life but the throwing up of hands. It is to
answer that question (once again, a question provoked by the Platonic background),
I suggest, that Zeno developed the theory of preferred and dispreferred indifferents.
Zeno proposes that while there is a contrast between what is good and what is
indifferent, there are still grounds for preference among the indifferents (that is to
say, the practical conduct of our lives still has some scope, contrary to Aristo’s view).
Consider two passages, Cicero, de finibus III. 51 (a passage that follows the citation at
the beginning of this chapter) and Diogenes Laertius VII. 102–3.
But among those things which are valuable, in some cases there is sufficient cause why we
should prefer them to other things (as in the case of health, unimpaired perception, freedom
from pain, glory, wealth and other such like101) and in other cases not so;102 and likewise
98
Hence the objection of the orthodox Stoics that this makes life unlivable, Cicero de finibus III. 50. It
is significant that this argument turns, as will Zeno’s and as did Socrates’, on how we treat the compresence
of opposites in cases of value: any indifferent is no more preferred (in some circumstances) than it is
dispreferred. The dispute between the contributors to the debate concerns what they suppose to follow
from this.
99
Cf. Cic. Acad. II. 130; this confirms the suggestion that Aristo’s coinage was not the category of
indifferents, but his description of the attitude that he recommends towards them, adiaphoria.
100
Cf. Ioppolo (1980).
101
This strategy of rejecting, or qualifying, a list of putative goods is true to Euthyd. 278e ff.
102
There is a danger of both over- and under-translation here: Rackham gives ‘while others are not of
this nature’, which pushes the interpreter towards evaluative dualism. I have under-translated: better might
among those things which are not worthy of value, in some cases there is sufficient reason why
we should reject them, and in some cases not so.
These things [sc. e.g. health and wealth] are not goods, but indifferents of the class of preferred:
for just as it is the special quality (idion) of what is hot to heat, not to chill, so it is the special
quality of the good to benefit, not to harm. But neither wealth nor health benefits any more
than it harms; therefore they are not good.103
The first passage is directly associated with Zeno (it precedes Cicero’s account of
Zeno’s coinage of ‘preferred’ and ‘dispreferred’). The second has the air of a
definition (hence the expression ‘of the class of ’ kat’eidos), and so may belong to
the originator of the theory of indifferents—and so perhaps to Zeno himself. And we
may notice that there is some similarity between the two passages: for both speak of
the theory of indifferents in terms of an issue about causation or explanation. Cicero
suggests that there is some cause, or perhaps justification, of the preferability of some
indifferents; Diogenes points to the contrast between things that do have causal
efficacy (the hot heats) and those that do not (health does not [sc. necessarily]
benefit). And in this respect the argument is true to the strategy adopted by Socrates
in the Euthydemus, on the derivative reading:
b) wisdom is the only thing that has value itself by itself; all the other valuable
items are so derivatively from wisdom.
If Socrates’ conclusion is that wisdom alone is the good itself by itself, he reaches that
conclusion on the basis of an argument to show that wisdom alone is the source of
value. Since everything other than wisdom no more harms than benefits, wisdom is
the cause of the value of anything else. The same point may be made by the Stoic
analogy with the hot heating. At first reading, the Diogenes passage may simply insist
on the thought that the good must be unconditioned: so what is good always has
beneficial effects. But that is, I suggest, to make too little of the point about special
qualities (idia) and their causal properties. If the special quality of the hot is to heat
(and never to chill), then anything that may either heat or chill is not the hot;
likewise, whatever may either benefit or harm cannot be the good (so that wealth,
e.g., cannot be the good). But the good is what makes other things good (it is what has
that special quality), just as the hot makes other things hot; so whatever is made good
by the good is not in itself such as to make other things good; therefore it is not the
good. Now that account of the causal powers of the good bears some resemblance to
the causal powers of what is good itself by itself in the Euthydemus (and the
resemblance is marked by the reappearance of an argument from the compresence
be ‘other things are not like this’. The point may be to remind us of the account of two different sorts of
indifferent, cf. Diogenes Laertius VII. 104–5, and thus to tell us about what is preferred by nature (this
tends towards dualism); or to suggest that the grounds for preference may vary (which tends towards the
derivative view which, I shall shortly suggest, should be attributed to Zeno).
103
Notice here another version of the argument from the compresence of opposites.
of opposites). It gives Zeno an argument, not about two sorts of value, but rather
about the explanatory relations between what is good itself by itself and whatever
derives its goodness from that. Zeno makes, in short, a derivative reading of the
Euthydemus’ conclusion.
That may be confirmed by a feature of the theory of indifferents that is unques-
tionably Zenonian. If among indifferent things there are distinctions of value, these
are to be described in terms of the distinction between what is ‘preferred’ (proêgme-
non)104 and what is ‘dispreferred’ (apoproêgmenon).105 It was Zeno who coined the
expressions ‘proêgmenon’ and ‘apoproêgmenon’, as Cicero elaborately explains:
Just as in a royal court—Zeno says—no-one says that the king is so to speak promoted to his
honour (here ‘promoted’ translates proêgmenon); but the term is used of those who hold some
office which comes close to the primacy of the king, so that they are second; so also in a life it is
not those things which are of the first rank, but those which are in second place which are
described as proêgmena, that is to say ‘promoted’.106
Why does Zeno settle on this coinage in particular?107 If Aristo is right, as Cicero
dryly observes, the whole of life would be thrown into confusion (exactly, I suppose,
what Aristo intended) (Fin. III. 50). So there must be some way of making choice
rational, and avoiding Aristo’s conclusion (that neque ullum sapientiae munus aut
opus inveniretur, cum inter res eas quae ad vitam degendam pertinerent nihil omnino
interesset neque ullum dilectum adhiberi oporteret108). It is this objective, according to
Cicero, which provoked the Zenonian theory of what is preferred and what is
dispreferred.109 But once again, this needs to be carefully understood if it is not to
miss entirely the point of Cato’s objection to Aristo, as Cicero presents it. If Aristo’s
heresy can be understood as a reading of the Euthydemus, then he takes Socrates’
point to be that only wisdom has value; and he misses the possibility that the intrinsic
value of wisdom may explain the extrinsic value of everything else. But then there
had better be something about Zeno’s response that counters the thought that
104
This translation, which is now standard, is appropriate to the Zenonian origin of this theory, but
only, as I shall suggest, if we take the expression ‘preferred’ to allude not to some aspect of choice, but to the
notion of promotion. Zeno concerns himself not with our desires and preferences, but with the arrange-
ment or hierarchy of value. ‘Preference’ is thus a slight archaism, but entirely appropriate to the origins of
the metaphor.
105
Long and Sedley’s translation (1987) captures what I take to be the rebarbative features of Zeno’s
coinage.
106
Stobaeus cites the same comparison, 2.84.2.
107
Long (1996), n. 63 compares this coinage with Stobaeus’ account of what it is for something (e.g.
health) to be choosable in itself, 2.82.20–83.4. This does not, I believe, account for the strangeness of the
expression proêgmenon.
108
‘ . . . and no function or task for wisdom could be found, since there would be no difference at all
between the things that concern the living of life, and no choice between them would have to be made’ (trs.
Long and Sedley (1987), 58I).
109
A point here about chronology: if Zeno first proposed the theory of indifferents, and then Aristo
took it over in a heretical manner, then Zeno’s account of preference will itself be a response to Aristo. See
Schofield (1984).
non-absolute value is arbitrarily disposed, or else non-existent. And—to reread the

Euthydemus—Zeno needs to do that by explaining just what is the explanatory
(causal) relation between what is good itself by itself and what is not. It will not be
sufficient simply to observe that the value of what is valuable relative to context is
valuable in a secondary way. And it will not be enough for that, either, simply to
reinstate the importance of practical reason for what is relatively valuable, as an
alternative to cultivating indifference. Instead, Zeno needs to settle a metaphysical
issue: to show how it is that the real, objective values of things are themselves
explained by what is valuable itself by itself. For Zeno supposes that life should be
lived according to some rational disposition; and he needs to show that this rational
disposition matches some rational order out there in the world.
His development of the contrast between proêgmena indifferents and apoproêg-
mena indifferents is, I suggest, designed to fulfil this purpose (that is confirmed by
Stobaeus, who defines a preferred indifferent as what we ‘select on the basis of a
preferential reason’,110 2. 84. 18–85. 11). Recall that the relation between wisdom and
the other goods in the Euthydemus, a relation that I have described as a relation of
source to derivative, was described as wisdom leading the other valuables (this
expression was also used in the Meno). The same relation is invoked in a rather
more complex context: to justify the claim that wisdom turns fewer goods into
greater, ignorance turns greater goods into evils.
. . . if ignorance leads them,111 they are greater evils than their opposites,112 insofar as they are
better able to serve a bad leader; but if intelligence and wisdom lead them, they are greater
goods. (Euthydemus 281d6–8)113
Suppose someone has a collection of putative goods, but is ignorant; then their
ignorance will turn those goods bad, and will do their possessor more harm than
good. Contrariwise, if someone has wisdom and yet only a few of the putative goods,
those goods will do their possessor more good than we might otherwise expect: for
wisdom will make them valuable. So wisdom and ignorance determine the real
benefit of possessions and assets. The putative goods are not goods at all without
the presence of the source of their value; and the source of their value explains not
only how we may compare one extrinsic value with another, but also how we may
explain their context relativity. (Might we find this plausible? We might if we had not
110
Trs. Long and Sedley (1987), 58E.
111
‘them’: the antecedent here is the list of putative goods, the source of whose value is here under
discussion. Plato’s discussion up to this point has thoroughly undermined the thought that the original list
of putative goods has any objective standing. At this stage in the argument, that is to say, there is no
suggestion that these items already have some value, which is increased by wisdom; on the contrary, their
value is entirely determined by wisdom.
112
The ‘opposites’ are the items that are formally opposite to the original list of putative goods: so they
are putative evils (e.g. sickness, poverty). Here the opposition is objective (sickness is objectively opposite to
health) but not yet evaluated objectively.
113
Cf. Long (1996), n. 59, on this.
a consequentialist but an agent-relative bias, in which the source of value is, some-
how, the agent, whose happiness is necessarily connected to what he himself brings
about.114 Even so, Socrates’ theory is an extreme rationalist or intellectualist one.)
This might press the derivative reading of the Euthydemus argument: the items
from the list of putative goods that are given objective value derive their value from
wisdom. Now Plato explains any possible relation between derivative (extrinsic)
goods and the intrinsic good in terms of a metaphor of leader and led, of master
and servants, of king and courtiers.115 So, expressly, the intrinsic evil (ignorance) is
conceived as the leader, the extrinsic evils as the led, and conversely, the good itself by
itself will be the master, the derivative goods the servants. But to cash this metaphor
is to reveal further detail of the theory itself. Recall the distinction between what
I called the executive and the regal accounts of leadership. In the regal account, all
goods that are not the king himself are extrinsically good. It is the extrinsic goods that
serve the intrinsic, and not vice versa; this image, that is to say, fits ill with an
instrumental or consequential account of the relation between wisdom and the assets
that might constitute the happy life.116 Second, both the power and the purposes here
belong to the leader—the led serve the leader, not the other way around. But the
king’s purposes (within the metaphor, at least) are not directed at some other,
differently defined goods, for there are none such. This gives the leader (wisdom,
or virtue) the sort of teleological priority we should not expect from a consequen-
tialist account. There is, then, an explanatory priority in Plato’s image, which Zeno
will come to exploit. For the servants serve because they have a master, the led are led
because the leader leads them:117 it is the intrinsic good that explains the extrinsic
goods, not the other way around.
Now consider the similarity between Plato’s idiom and how Zeno is reported to
have explained his use of the expressions ‘proêgmenon’ and ‘apoproêgmenon’. His
coinage of ‘proêgmenon’ may strike us, after all, as initially surprising if his point is
merely to emphasize that one asset is preferable (either by nature or by virtue of
circumstance) to another (why not say, simply, ‘choosable’, or selectable: cf.
D.L. VII. 105?). But it may after all reflect a conceptual connection between Zeno’s
proêgmenon (‘brought out’, ‘led out’) and Plato’s hêgêtai (‘lead’).118
114
An example might be the Kantian notion of the good will. For such a Kantian approach, made more
generous about the scope of the good will than the Groundwork might be thought to allow, see Korsgaard
(1983). A different example is to be found in Aristotle’s account of the function of man in E.N.I.
115
Hence the interest in kingship in the second Socratic episode, 291b.
116
This inconcinnity appears even if we concede that extrinsic goods may be final; for even then we are
unlikely to suppose that intrinsic goods are instrumental. Again cf. Korsgaard (1983).
117
This is familiar from Euthyphro 10a ff.
118
The idiom of leading is common to both the Euthydemus and the Meno. In the Meno, I suggested,
the argument did not imply that wisdom had intrinsic value of its own (but merely an instrumental value,
cf. Meno 89a1), although it insisted that it is necessary for the achievement of happiness. This is consistent
with the (different) idea that wisdom might lead from in front, might be the pre-eminent value, rather than
its source. However, if the Euthydemus alters the terms of this debate (as I take to be the case, and as I take
to be the point of the positive claim in its conclusion), then the relation of leading to led may be a transitive
None of the goods is preferred because they have the greatest value.119 But what is preferred,
having the second place and value, is in some way adjacent to the nature of the goods. For in
the court, the king is not one of those who have preferment,120 but those people have
preferment who are ranked after him. (Stobaeus 2.85.11)121
Just as in a royal court—Zeno says—no-one says that the king is so to speak promoted to his
honour (here ‘promoted’122 translates proêgmenon); but the term is used of those who hold
some office which comes close to the primacy of the king, so that they are second; so also in a
life it is not those things which are of the first rank, but those which are in second place which
are described as proêgmena, that is to say ‘promoted’. (Cicero de Finibus III. 52)
Zeno’s analogy of the king may just be intended to bring out the contrast between the
king and even the highest rank of courtiers—the king is always at the top, the
courtiers always and necessarily below, the point merely one of hierarchy. But Cicero,
I suggest, gives us better elaboration of the point, when he translates proêgmenon into
Latin, productum. The crucial thing about the first rank of courtiers in a court is not
so much that they are, however high-ranking, always behind the king, but that the
explanation for that fact is that it is the king who promotes them; so, as Cicero
observes, no promotion can turn them into the king themselves, any more than the
king got to be king by rising up the courtierly ladder. Equally, the disappointed
courtier is ‘dispreferred’, that is demoted and sent away, because he is sent further
from the king’s presence and the high rank that proximity confers.123 In a court, the
king is the source of value; in Zeno’s account, wisdom is the source of value for the
indifferents; and in offering that gloss, he fits the theory offered by the Euthydemus
very well. Zeno, thus, turns away from consequentialism, towards a view of moral
value in which the agent is primary, both in importance and in explanation; and he
turns, also, away from any idea that there might be two separate types of value in
play. When Zeno read Plato he did so with care: evaluative dualism is as out of place
in the early Stoa as it is in Plato.
I began by asking about how we should read Plato (section 1). I suggested that the
significance of this question might appear in consideration of a particular case: two
passages in the Euthydemus and the Meno on the subject of the explanation of value,
which appear to be closely related (section 2). This close relation between the two
rather than an intransitive one: wisdom does not lead from in front, but by actively bringing on what comes
behind.
119
This claim, that the goods have hê megistê axia, is, we may note, compared to the derivative axia of
the preferred indifferents: once again this suggestion that they differ by degree is incompatible with
evaluative dualism. It is, I think, worth noticing also that this notion of the highest value, which is
paralleled by Diogenes’ account of the goods that are ‘at the summit’, has strong Platonic echoes (cf. e.g.
Rep. 505 ff.) and may be intended to construe the good auto kath’hauto.
120
This expression is meant to bring out the sense in which proêgmenon is construed, in Zeno’s
derivation, as a passive. Notice, again, the slight archaism in ‘preferred’.
121
The explicit connection between this analogy and Zeno’s coinage is made earlier in Stobaeus’ text.
122
And ‘promoted’ also translates productum.
123
Here the derivation of apoproêgmenon shows up its locative features, and in this respect it fits Zeno’s
image of the hot heating: the hot heats what is closest to it.
passages, I suggested, could be understood in terms of one ‘reading’ the other—where

one is alluded to in the other, and thereby becomes the subject of critical reflection
(section 3). I argued that the Euthydemus goes further than the Meno in insisting that
wisdom is the intrinsic good: the good, that is, which is itself the source of value.
Indeed, the careful relations between the two arguments suggest a complex reflection
upon, and rejection of, consequentialist accounts of value (section 4).
This case of Plato reading Plato seems to have inspired some central debates in
early Stoicism about the nature of the good. Nonetheless, the scope of the Euthyde-
mus’ contention about the good remains vague; and that is why, when the early Stoics
come to read this passage in Plato, they differ in their interpretations. That difference,
I claimed (section 5), is the source of the disagreement between Zeno and Aristo
about the nature of ‘ethical indifferents’, since different readings of the Euthydemus’
argument produce different views about whether there are any other items of value
than the good itself by itself. But in either case, the Euthydemus does not encourage
what I have called evaluative dualism: the view that there may be two quite different
structures of value. For if what is good itself by itself is the source of value for
something else—as I argued Zeno to suppose—then the value they have will be of the
same kind. Neither Zeno nor Aristo, therefore, were evaluative dualists. Furthermore,
if we understand the Stoics as having been engaged in reading Plato, we may further
understand just why they elaborated the theory of indifferents in the language and
terms they did.
So how did the Stoics read Plato? The relation between the Meno and the
Euthydemus, I suggested, is one that encourages critical reflection on the nature
and the metaphysics of value. This feature of the Platonic texts was felt by his Stoic
readers. For Zeno and Aristo are not simply influenced by some doctrines they
happen to find in the Euthydemus and the Meno. It is true that Stoic ethical theory
would be an ally of the attack on consequentialism in the Euthydemus. But the
agenda for the debate between Zeno and Aristo is set by the Euthydemus, which
asks about the source of value. And the conduct of that debate about the source of
value is itself informed by the very indeterminacy of the Euthydemus’ answer. So the
Stoic reading of Plato is a rich one; and this is exactly, I submit, what Plato intended
to invite.124
124
I should like to thank Malcolm Schofield for his criticisms and encouragement.
13
Out of the Labyrinth: Plato’s Attack
on Consequentialism
What is the relation between Plato’s account of virtue and his defence of ethical
objectivity? Some views of Platonic virtue may relegate it to a role severely subor-
dinate to objective value. In what follows I reflect on Plato’s attack on consequen-
tialism in the Euthydemus and consider how that dialogue, read as an integrated
whole, may provide a view of value inimical to consequentialism. From this view,
I argue, flows an account of the nature of value to which virtue is central; but the
metaphysics of that account explains objectivity too. This, in turn, may illuminate
our modern concerns: for it provides a distinctive starting point for the metaphysics
of value, and hence for any debate with the consequentialist.
1. Consequentialism and Socratic Intellectualism

‘Surely (said Socrates) all of us want to do well?’ His question is notoriously suspect,
but its sentiment has often been endorsed in the history of moral philosophy. Leave
to one side (if you can) the question of whether its self-regarding aspects should
make us uncomfortable.1 Focus instead on how easily, when we think of doing well,
we imagine it as some optimal state of affairs, the turning up of which will constitute
our happiness.2 One thought may bring another: that our ethical task is to discover
the means to this end, so that (all other moral things being equal—of course they
ain’t, often) any means is good just so long as it leads to this end.3 On such an analysis
(and barring some other, independent, moral imperatives), means are evaluated just
by their effectiveness. There is nothing morally significant in wondering whether I
should bring this state of affairs about or whether I should leave it to someone else to
benefit me or whether I may rely on luck.
This is what I shall take consequentialism to be: a theory that rejects the inde-
pendent importance of the agent in reaching ethical outcomes in favour of the value
1
As I was revising this chapter, I was also learning from Penner and Rowe’s interpretation of the Lysis
(2004), who confront this issue, in the context of Socratic eudaimonism, head-on.
2
This is a claim about what we really want: a psychological eudaimonism.
3
An ethical eudaimonism; see Irwin (1995); Annas (1999). But see Penner (2005) on how Socratic
‘ultra-realism’ makes no room for ethical eudaimonism.
OUT OF THE LABYRINTH 259
of some state of affairs as the outcome itself. It does not matter, that is, how the best
state of affairs (whether for me, or overall) is brought about, just so long as I (or we
overall) get it. Consequentialism is, of course, capable of far subtler specification than
this.4 But it seems, nonetheless, to make two connected claims about the nature of
value:
(1) Value is to be found in ends, and not in the means to them (provided that the
distinction between means and ends can be satisfactorily maintained). So,
notoriously, without some other assumption acting as a restraint on what may
be done to achieve the end, any means is justifiable by its end. Then, value
flows from the end to the means, so that the means is instrumentally, but not
independently, valuable.
(2) Value is agent-independent (where the agent is specified as the person who
brings about the means to the end); even if it is enjoyed by the agent, it is not
enjoyed because she is the agent.
It is commonplace to suppose that this sort of theory was offered by Plato’s
Socrates—the character we find in those dialogues is often thought to have been
written early in Plato’s career.5 Perhaps Plato’s Socrates was (at some point) com-
mitted to the view that happiness is all we pursue, and that virtue is the instrument to
it.6 Consider, for example, the argument in the Protagoras that seems to deny the
possibility of akrasia (351–8). Here, on the assumption that there is a single scale of
value (whether hedonist or otherwise eudaimonist), Socrates denies that someone
could know the better, and be able to do it, but nonetheless do the worse. Instead, he
maintains, actions that look akratic are mere executive failures—failures to exercise
the proper measuring skill, to compute where value lies.
This is an intellectualist thesis:7 failure occurs just when the person who fails is
ignorant. And it has consequentialist roots in taking value to come simply from the
consequences of action.8 There is no independent account to be given of what it
would be to be weak, or tempted, or wicked, since all the ethical content is taken up in
4
See e.g. Scheffler (1982), (1988). In particular, the specification of a ‘state of affairs’ is vague. The
refined consequentialist may include good, even moral, states of the agent in the specification of the end;
but he will still, I think, construe the end independently of how the agent arrives at it. More hangs on this
than I have space for here, where I argue that for Plato the state of the agent (or the person) cannot be
contingent, or optional, in accounts of value; and I take this to be something the consequentialist cannot
allow.
5
I shall not discuss the ‘Socratic question’ here (but see the vigorous recent debate in, e.g., Vlastos
(1991); Kahn (1996); McCabe (2001); Annas and Rowe (2002)), except to continue to insist that there are
fruitful ways of understanding Plato as reflecting later on his own earlier views: see Chapters 10 and 12.
6 7
See Irwin (1995), ch. 4, for some discussion. See Penner (2005).
8
See e.g. Gorgias 467c–468e on the derivation of the value of the means from their ends. The moral
psychology is simple and intellectualist—the desire for good is universal; what we actually do is determined
by what we know or believe about the goods before us. The metaphysics is simple too: there really is
something (objectively) good before us; our problem lies in recognizing it. See Penner (2005).
260 OUT OF THE LABYRINTH
the ends whose value is to be computed for right action. If happiness is some end-
state of our actions and affections, and virtue is its instrument, then not only virtue
but also the very engagement of the agent seems contingent upon virtue’s achieving a
successful outcome.9 If there were some other means of getting happiness, virtue
should be ousted from the sovereign place it seems to occupy in Plato’s ethical
theories.10
We might come to virtue’s rescue by construing Plato as an ‘evaluative dualist’.11
He might suppose that there are two different (incommensurable, possibly conflict-
ing) kinds of value, probably with two different sources; in particular, he might
contrast the values expressed in the ends of action with those expressed in the virtue
of the agent. So Plato might sharply contrast the value to be sought in the pursuit of
happiness and the values to be expected from the good, virtuous agent.12 The thought
is an old one; what follows will suggest that evaluative dualism of this kind does not
characterize Platonic ethical theory, at least at some crucial stages of his philosophical
career.13
2. Platonic Objectivity and the Moral Phenomena

Socrates’ argument in the Protagoras may be a travesty of what might be meant by
the Socratic slogan ‘virtue is happiness’; and it is surely to traduce the Protagoras as a
whole to take this argument out of context and produce a theory of such minuscule
moral muscle.14 Nonetheless, critics have often applauded Plato’s theory, as it
supposedly moved away from the heavy influence of his master,15 for providing a
view of virtue and happiness in which the agent is central—the account in the
Republic of virtue as psychic harmony.16 But not so fast.
Although the Republic’s moral psychology focuses close attention on states of
soul—and even identifies the harmonious state of soul with happiness—it has its
weaknesses when it comes to thinking about the position of the ethical agent. Those
weaknesses become evident in Plato’s metaphysics of value: in his account of the
9 10
See Nussbaum (1986), ch. l. See e.g. Gorgias 469–81; Republic 4, passim.
11
See Chapter 12.
12
This used to be expressed as a contrast between the prudential and the moral: e.g. Vlastos (1991), ch.
8. One great benefit of recent discussions of ancient and modern ethical theory, however, has been to thaw
the frozen grip of those two categories of value: see Williams (1985); Annas (1993).
13
Whatever we should say about Plato’s development, I (continue to) find a significant difference
between the Euthydemus and what one might see in, e.g., Meno or Lysis. I take that difference to be the
result of reflection; but this view is the source of some fruitful (for me, at any rate) dispute between myself
and Penner and Rowe (2004).
14
Socrates may here use hedonism as a marker for the unacceptability of this moral psychology.
Nonetheless, the denial of akrasia in the Protagoras is a powerful challenge to any subsequent moral
psychology.
15
This is Vlastos’ view (1991), and has been mine. Here is an attempt to qualify this view.
16
Penner and Rowe (2004) take this to be the sole (significant) area of development and change in
Plato’s thought.
objective good. The good is explained in terms of some transcendent nature, ‘beyond
being’; and it is utterly impersonal.17 That may be fine and dandy. It makes the good
objective, alright, and confirms Plato’s view of value (goodness, beauty, and so forth)
as a real, out there property of the things we value. In the mode of the Euthyphro
puzzle,18 we value them because they are really valuable; they are not merely valuable
because we value them. But this view, too, tends to diminish the moral agent. If the
good is impersonal, and if this impersonal good determines happiness, the state that
is happiness is expressly detached from the person who achieves it. So if happiness is
defined in terms of objective value, but virtue is a character of persons, then virtue
turns out, once again, to occupy a subordinate position. Platonic happiness seems to
have nothing to do with me.
This may generate a difficulty for any account of objective value. To suppose that
value is objective may conveniently suggest that we can be right about it, that disputes
can be settled and conflicts resolved in its light.19 But that supposition may cause
some trouble—as the example of the Protagoras makes clear—when it comes to
moral psychology. For objective value, it might be thought, is what causes our pursuit
of whatever is valuable. In that case, our failures to succeed in our search for value
may only be ascribed to failures in cognition; and in that case (it might be thought)
there is no room for a richer account of the psyche that does the pursuing. When we
end up not pursuing what is objectively valuable, is this just because we didn’t know
what we were doing? In that case, is there any virtue in the pursuit, or any vice in its
avoidance? If value is just out there, can anything interesting, from an ethical point of
view, be said about in here?20
Well, why should we care about the role of the moral agent in an account of value?
Why should it make any difference to the way I feel about some state of affairs that
comes about that I was somehow responsible for it doing so? Does it make me
happier if my happiness was my own doing? Perhaps it does; and perhaps in any case
this is the wrong question to answer. For it may come at the business from the wrong
end. The starting point should be, some say, moral anthropology.21 We do in fact
think that it makes a difference to the value of a state of affairs how, and with
whose connivance, it came about. The moral emotions of regret and remorse, for
example, as well as the positive sense of pride in an achievement, cannot be explained
without citing the agent in stating the value of what she makes happen. So, if
consequentialism fails to include the agent in its explanation of value, it fails to
17
See e.g. Kraut (1973), Annas (1981), ch. 10; for a range of interpretations of the idea of Good in the
Republic, Reale and Scolnicov (2002).
18
Euthyphro 10a–11b.
19
I am not sure how far this would mean that there are no ethical norms, as suggested by Penner (2005):
does the existence of norms require the possibility that we might be blamed for mistakes?
20
To render this difficulty acute we need some extra assumptions: for example, that real value is
transparent to our true desires (e.g. Gorgias 466e–470e); again, see Penner (2005).
21
See e.g. Anscombe (1958); Williams (1985).
account for the moral phenomena. And that is a serious blow to the explanation
itself.
That criticism is characteristically Aristotelian—as, many suppose, is moving the
moral agent to a central position in moral explanation.22 Aristotle insists that in
ethics as elsewhere we should begin with the phenomena;23 any account that leaves
them out is correspondingly deficient. So much the worse for Plato, whose starting
points are not the phenomena at all. Socrates himself makes a virtue of disagreeing
with the many, and all the wise are suspect—except for Socrates himself, who knows
nothing. Does Plato have any response to the displacement of virtue in the structure
of value? And should he care whether the agent is part of the characterization of
value? Is there any interest, for Plato, in the rejection of consequentialism?
This question will go hand in hand, then, with another, about the methods of
ethics. What are the starting points for ethical inquiry? Someone might deny that
moral anthropology is of any significance: what people actually think may make no
difference to how they ought to think. So (someone might say) the collection of data
about people’s ethical beliefs can only tell us about those beliefs; it is inevitably banal,
useless for the formulation of ethical starting points.24 No explanation is to be found
in a collection of the explananda.25 In that case, where should we begin? Consider-
ation of the Platonic approach might make the question more pressing: for Plato
often implies that the deliverances of moral anthropology are inadequate.26 But then
what will ground our ethical methodology? If ethics is to provide its own starting
points, is it ineluctably circular? Or if its starting points come from elsewhere, can
those starting points carry any weight, or provide the basis for an explanation (rather
than merely a restatement of the evidence)? To this I shall return.
3. Good Luck or Good Management?

Two passages in the Euthydemus, I argue, give us grounds to suppose that Plato is
interested in the rejection of consequentialism. The dialogue is divided into five
episodes, framed by an elaborate narrative. Three of the five are dominated by the
sophists; the second and the fourth are led by Socrates, talking first to the young man
Cleinias and then, in an interruption of the frame narrative, to Crito. In the interplay
of the Socratic episodes with each other and with their frame lies a quite different
22
See e.g. Nussbaum (1986), chs. 9–12; Hursthouse (1999).
23
See EN 1045b2–6. Even if the phenomena of physics were obvious (are they? Are they raw data? Or
information already processed by judgement?), we might still be puzzled about what counts as a phenom-
enon for ethics. An action? A belief? See Barnes (1981); Nussbaum (1986); and Owen (1986a).
24
See Nussbaum (1986), ch. 8.
25
Even a ‘simpleminded answer’ insists on more: ‘the beautifuls are beautiful by virtue of the beautiful’,
Hippias Major 287c–289d, Phaedo 100e.
26
Inadequate does not mean non-existent or false: this does not commit Plato either to the Socratic
fallacy or to extreme other-worldliness. See Geach (1960); Burnyeat (1977).
moral theory, located in an argument that consequentialism cannot be rationally

defended.
The first Socratic episode begins with the eudaimonist commonplace with which
I began: ‘surely all of us want to do well?’. When Cleinias agrees, Socrates asks a
trickier question: ‘since we all want to do well, how are we to do well?’. The ambiguity
of that question often escapes notice: is Socrates wondering how we can bring it
about that we do well? Or is he asking a deeper question: what is it to do well? The
sequel disambiguates: although his interlocutors think that Socrates is talking about
executive skills, Socrates himself is trying to understand the nature of happiness. In
replacing his interlocutors’ assumptions with his own, he shows up the dangers of an
executive model of virtue; and he replaces—or so I shall suggest—his interlocutors’
consequentialism with his own account of value. The argument turns on explaining
happiness; it invokes, therefore, Socrates’ account of the metaphysics of value: of the
direction in which value flows.
How are we to do well? Surely, by having many goods—such as wealth, health, a
good reputation, virtue, and wisdom—and good luck (279c). And yet good luck
seems to be supplied by wisdom—since expertise reliably controls success. So it is
wisdom that makes people do well; and there is no need to ask for good luck too
(280d). Cleinias agrees to this instrumentalist thesis: wisdom is what supplies the
goods, the accumulation of which constitutes happiness. But now Socrates demurs.
Having lots of good things is only happiness if the good things actually benefit us; and
they only do that if we use them—indeed, if we use them correctly (280e). Incorrectly
used, conversely, we are better off without them. Correct use is supplied by nothing
but wisdom, because it is wisdom that supplies good fortune and success (281b).
Then Socrates produces a surprise.27 If we are not wise, we are better off poor, weak,
inactive, and out of the public eye. And in that case, the original list of goods (wealth,
power, and so on) turn out not to be good, themselves in themselves; instead, their
value depends on whether they are led by wisdom or by ignorance. Only wisdom, on
the contrary, is good (itself by itself ); and only ignorance is bad (itself by itself )
(281e). Indeed, wisdom is worth any sacrifice, and it can be acquired by teaching. So
we should philosophize (282d).
This argument seems to end up at a very different place on the ethical map from its
starting point. And it reaches its destination by two marked changes of gear. First
(280e), against the initial thought that doing well is ‘having many goods belonging to
us’, Socrates insists that possession is not enough: at least our goods need to be ruled
by wisdom. Second (281b), the very conception of a list of goods is called into
27
‘So, by Zeus, is there any benefit to arise from anything else without intelligence and wisdom?’
(281b5). This question marks the shift in Socrates’ argument from a discussion of how we are to realize the
benefit of such goods as we have to the question of what makes them good at all. Here, I think, the views
expressed in the Euthydemus diverge from those of Lysis 220 or Meno 87–8. See Chapter 12.
question by the claim that there is, after all, only one good ‘itself by itself ’: wisdom.
So consider the route of the argument again.28
At 279a Socrates starts with wealth, moves on to health and a good reputation, and
only then (with some hesitation) mentions the virtues, and then wisdom itself. En
passant he toys with the relation between wisdom and good luck. Socrates does that?
This is the Socrates who is old and unattractive (he uses the sons of Crito as bait to
induce the sophists to take him on, 272c–d), slow to learn and absurd (272c), the
object of the sophists’ contempt (273d), and lacking in the elaborate lineage of the
young Cleinias (275a). This is the familiar Socrates, who listens to his daimonion at
inconvenient times (272e), and who seems to care as little for the slings and arrows as
he does for the good fortune of others—he is interested only in wisdom and turning
others to philosophy. But in that case, his list of goods is offered in a thoroughly
disingenuous manner. Why?
Well, this is not merely a list of goods: it is from the outset a discussion of the
grounds for their acceptance—or their rejection. Thus, for the first premiss, we all
think (that we want to do well). In the second, it is obvious (that doing well is having
goods). In the third, even someone not very important will get the point of wealth,
and health; while the assets of public reputation are just (apparently) obvious. Once
Socrates moves on to disputed cases, however, the ground shifts: Cleinias should
answer not what everyone thinks, or what is obvious, but what he thinks himself
(278b6–7).29 Having persuaded Cleinias to take sides in the case of the disputed
goods (the virtues, 279b4–c2), Socrates then discovers him blank and amazed in the
face of yet another proposition which, he says, everyone believes: that wisdom is good
luck (279d6).
The Platonizing reader might feel blank and amazed too. How can this be Socrates’
view? ‘Good luck’ fits well enough in the vulgar list of putative goods (health, wealth,
and so on). Cleinias’ easy assent to its inclusion trades, we might think, on the idea
that it is obviously a good thing to be lucky; in that sense, good fortune is on a par
with the assets of wealth or good looks. Call this passive (good) luck: the merely
passive state of having good things happen to you (it has, of course, a bad
counterpart—passive bad luck, disasters, and catastrophes). In this sort of luck the
agent is not actively involved and takes no deliberate or conscious part in the lucky
outcome.30 Passive good luck is like the other assets by virtue of its contingency, and
it is a general descriptor of them all: being wealthy, being beautiful, being well
28
Other interpretations of the argument see no such radical revision towards its close: see Vlastos
(1991), ch. 8; Irwin (1995), ch. 4; Long (1996), 23–7; Annas (1999), 40–2; Gill (2002); but see e.g. Burnyeat
(2003) and Broadie (2005) on how the good ‘itself by itself ’ comes out as a radical postulate.
29
On the demand for sincerity, see McCabe (2004).
30
The lottery winner who chooses to buy a ticket does participate in his own good fortune. But his
action is, from a statistical point of view, so blindingly daft, that it only takes its sense from the outside
chance of a win, and not from the rational engagement of the agent.
connected may indeed be matters of pure luck.31 The luck is ‘pure’, that is to say, just
because it does not come subject to its beneficiary’s control: the causal arrangements
for beauty, for example, come from the genes, not the active participation of the
person with them.32
Socrates resists this view—as we might expect. He first reformulates good luck as
success.
‘Wisdom everywhere makes men have good luck. For wisdom does not miss the mark, but
necessarily hits it, and does well. Otherwise it would no longer be wisdom.’ We agreed in the
end, I do not know how,33 that in summary this is so: whenever wisdom is present, to
whomsoever it is present, he does not need good luck in addition. (280a6–b3)
This move trades on the notion of a craft. It is the nature of a craft to make sure that
what the craftsman does hits the mark;34 in doing so, he provides good luck—success,
nothing to do with lucky accidents. Now if good luck is what you want, you want it
no matter what its source and irrespective of your own involvement in its turning up:
you want passive good luck. Success, however, is a different beast: to be successful
seems to require agency. This active luck is both a matter of causation and a matter of
responsibility. Causation: the craft controls its outcome; so the craft is both successful
and reliable. If I am a craftsman, therefore, I have a reliable means to success because
I control what happens, and need not wait for fickle fortune to produce good things
for me. Responsibility: if I am not a craftsman, I can still go to one for help: provided
his advertisement is reliable, then by entrusting my affairs to him, I can ensure that
I share in his success.35 So active luck can be delegated; it is not crucial that I be the
craftsman, so long as I exploit a man who can. The location of the skill, we might
therefore say, is indifferent to success, just so long as the skill is exercised on my
getting the goods. Those goods, however, are valuable independently of the skill that
provides them: the value of the consequences of my (or my colleague’s) exercise of
skill is unconditional upon that skill.
But now Socrates rethinks and reformulates, again. The final phase of the argu-
ment emphasizes the (mental) state of the agent; and it reconsiders ‘doing well’.36 ‘Is
31
There is a major debate about the significance of ethical contingency in Greek thought; see e.g.
Williams (1976); Nagel (1976); and Nussbaum (1986). If the present passage finesses the role of luck in
happiness, Socrates’ point is to rule out the importance of luck for our explanations of happiness.
32
A vigorous entrepreneur might not agree, of course, that wealth is outside the control of its owner; the
plastic surgeon might say the same about beauty. I leave it to my reader to supply the likely response of
Plato’s Socrates to such counter-examples.
33
A reminder to the reader to wonder about the structure and sequence of the argument.
34
That Plato is deliberately forcing our attention towards the scope of the expression eutuchia is
reinforced by his use here of the idiom of hitting/missing the mark (hamartanein, tunchanein, 280a7–8;
tunchanein provides the root verb for eutuchia). This vocabulary runs right through the passage (281cl)
and is, I think, no mere dead metaphor.
35
This is taken up in the discussion of teaching at 282c.
36
Does Plato just muddle a vital distinction between doing well = acting virtuously and doing well =
succeeding? On the contrary—his careful exposure of the breadth and scope of these expressions is
there any benefit in any other possessions without prudence and wisdom?’ This now
is the agent’s own prudence and wisdom: borrowing it won’t do. Does a man37 with
many assets and activities but no intelligence do better than a man with few assets but
with intelligence?38 Surely, a man does not get any benefit from many assets or
activities if he has no intelligence, while he gets more benefit from few assets if he has
intelligence. Of course, when a man does less, he makes fewer mistakes, and so he
does less badly, and is less wretched. He is more likely to do less if he is poor and
powerless, weak and cowardly, lazy and slow. So (281d–e):
‘In short, Cleinias,’ I said, ‘it seems probable that as for all the things which we said at first were
goods, the argument is not about this—how they are by nature goods themselves by
themselves—but it seems that matters stand thus: if ignorance leads them, they are greater
evils than their opposites, to the extent that they are better able to serve a bad leader; but if
intelligence and wisdom lead them, they are greater goods; but themselves by themselves
neither sort of thing is worth anything.’
‘It seems to be exactly as you say,’ he said.
‘So what follows for us from what has been said? Surely it is that nothing else is either good or
bad, but these things alone are so . . . wisdom is good and ignorance is bad.’
The putative goods—wealth, strength, a powerful reputation39—are in fact greater

evils than their opposites if they are led by ignorance, and greater goods than their
opposites if led by wisdom: they are not worth anything ‘themselves by themselves’.
Wisdom is the only good, ignorance the only evil; everything else is neither. And, to
reiterate, this is the happy person’s own wisdom, the miserable person’s own ignor-
ance; delegation won’t do.
The sense of the final clauses of this argument has been disputed.40 But Socrates
cannot (on pain of contradicting himself )41 mean that wisdom is the only good at all;
he must mean that it is the only good itself by itself. In so doing he exposes the
relation between wisdom and ignorance on the one hand, and the other goods on
the other; and claims that this relation is not an executive one at all (as we might have
supposed). The other goods are not ranked below wisdom because wisdom is the
instrument to their acquisition, but because wisdom is the source of their value.42
Someone ignorant could easily be wealthy; but they would not be happy. For poverty
designed to clarify, not to obscure, the complexity of ‘doing well’, of ‘good luck’, and, later, of ‘wisdom’ and
‘knowledge’.
37
The gender is in Plato’s Greek: for ease of reading and no other reason I retain the masculine pronoun
when directly adapting what Plato says.
38
The text at 281b8 is disputed; I follow Burnet and reject Iamblichus’ version, which omits ‘ . . . but
with intelligence’. I defend this reading in Chapter 12.
39
281b–e recalls the initial list of putative goods, including the virtues of courage and sōphrosunē. The
argument later revises these earlier assumptions; as a whole it is thus self-critical.
40
See n. 28.
41
281d6 allows the putative goods to be comparatively good, under the leadership of wisdom: they are
not completely valueless. Aristo and Zeno fell out over this: see Chapter 12.
42
Burnyeat (2003) makes the same claim for Apology 30b.
and wealth benefit or harm depending on the presence or absence of wisdom.43

Likewise, someone prudent or brave may easily be ignorant; and their courage or
prudence will benefit them only when made valuable by wisdom.44 Wisdom, there-
fore, is the source of the goodness of everything else. Its unconditional (itself by itself )
nature marks its metaphysical function: it is ‘the good’.45
What does that mean? Early stages of the argument have established that wisdom
is necessary for doing well; its final stages seem to ask about wisdom’s sufficiency.46
How could someone who is wise and poor be better off than the rich man who is
ignorant? Even if the ignorant man is unable to exploit his wealth, how is the wise
pauper any better off? Socrates’ conclusion does not rely on what he has argued for
earlier: that wealth cannot be exploited without wisdom. On the contrary, wealth has
no value at all without wisdom; wise poverty is better. So wisdom is not merely our
ability to use what we have; it turns what we have from a disadvantage into an asset,
while ignorance renders null the assets we thought we enjoyed.47 Indeed, we can
count nothing as either an asset or a disadvantage unless we know how wisdom is
related to it; then, anything that is accompanied by wisdom will (somehow) turn out
to be an asset, anything without it, a disadvantage. So if wisdom is the source of value,
which things are in fact rendered good by it may be radically revised: we cannot
presume, unless we ourselves are wise, that avoiding the rack or being brave will be
on that list.48 We must—as Socrates urges Cleinias to do—revise completely any
43
Socrates’ comparative claim is crucial to understanding this point, but it is tricky to interpret. For his
conclusion, he need not show that nothing but wisdom is good; but he must show that nothing but wisdom
is good itself by itself. He does so by arguing that everything else depends for its value upon wisdom. So the
ignorant man with wealth is worse off than the ignorant man without it; and the wise man with poverty is
better off than the ignorant man with wealth. What of the wise man with wealth? The argument does not
deny (and for symmetry should maintain) that the wise man with wealth is better off than the wise man
without. So wealth may have value: but not independently of wisdom. Wealth may have, for example, some
potential, which is only realized when accompanied by wisdom (wealth is good when led by wisdom:
material facts may be transformed by wisdom into something valuable). And on that account wisdom
retains its status as the only good (itself by itself ). So is the wise poor man happy? Yes—he is better off than
the ignorant rich man; but he may not be as happy as the wise rich man. So wisdom seems to have some
value, as well as conferring it (I return to this below). Socrates need not deny that, or insist that wisdom
exhausts all value, provided that wisdom is the only source of value (wisdom is in this sense the only good
itself by itself ). Wisdom is sufficient, therefore, for some value, and some happiness, on its own; and for
complete happiness wisdom is necessary, because it is the only source of value.
44
This may come as a surprise: what of the ‘Socratic’ claim that the virtues are a unity?
45
Korsgaard (1983): the intrinsic good. She distinguishes, on the one hand, between goods that are final
and those that are instrumental to those ends; and, on the other, between goods that are independently
valuable (intrinsically good) and those whose value depends on something else (extrinsically good). See
also Williams (2003). During the Exeter conference, this specialized use of the expression ‘intrinsic’ caused
some dispute; I have tried to gloss it in what follows.
46
Irwin (1995), ch. 4.
47
Thinking we enjoy something is insufficient for that thing to be a component of our happiness; see
Gorgias 466–70.
48
See n. 43. If wealth accompanied by wisdom is more valuable than poverty accompanied by wisdom,
but wisdom is the only source of goodness, wealth will not have some residual value of its own. Wisdom,
Socrates’ argument implies, does not render everything other than itself indifferent but causes distinctions
in value.
received ethical opinions we might have had. We cannot assume in advance that the
avoidance of pain is of any significance—any more than we can assume that it is not.
By parity of reasoning, we should not presume that courage or prudence will be good
come what may: they too depend on wisdom for their value. The ethical phenomena
are overturned.
Consider the structure of explanation exploited here. Socrates’ point is not that
wisdom is the best end we may have, an end that trumps or transcends any other; the
relationship of wisdom to doing well is not based on a scale of relative value. Nor is
his point merely that wisdom is the only good we can have on its own, while the
others are dependent for their occurrence upon it. For here the sovereignty of
wisdom49 is a matter not of the extremity of its goodness, but of the role it plays in
making other things good. We understand value, that is to say, not in terms of an
ordering of ends, but in terms of its causal structure. That structure shows that
everything else depends for value on wisdom. It also shows that wisdom always
confers some value: for even the man with no apparent assets at all (poor, incapable,
disenfranchised) will be better off than someone who is ignorant, no matter how
fortunate they may seem. Wisdom is the sole source of value, necessary for any
goodness and sufficient for some.50
This causal structure declares consequentialism to be false, for two reasons. First,
in respect of the analysis of value: the source of value is a state of the wise man;
happiness cannot be specified merely as the acquisition of a collection of goods.
Second, in respect of the relation between means and ends: the value of wisdom does
not lie in its ability to supply some end-state other than wisdom itself; wisdom is not
an executive skill. No more, however, does wisdom seem to be the end to which
everything else is a means: instead, its role is to supply the value to the things from
which a life is constructed. It has the effect, therefore, of providing value (by being its
source) and of unifying it (by being its only source). It is in this sense we might say
that it is the intrinsic good; without it there is no goodness at all.
So is wisdom then agent-centred? Even this characterization may fail to capture
the extremity of Socrates’ point. Wisdom is not, on this account, something like a
disposition to act in certain ways (for then the value of the disposition would have to
derive, at least in part, from those ways of acting), so it is not valuable as a character
of agents who do such acts. How then does wisdom work as a source of value? And
how will this give us any help with the problem of how best to live? The consequen-
tialist, rising from the last punch, may ask this question with some insistence: if a
49
Cf. Broadie (2005). Vlastos’ treatment of the sovereignty of virtue takes it to indicate virtue’s absolute
preferability; see e.g. (1991), 209–14, 228, n. 94; this is not, on the interpretation canvassed here, what the
Euthydemus maintains.
50
It might be objected that this claim is fatally weak: surely, if wisdom is not sufficient for all goods but
only for some, there must be some other, independent, source of goodness. This objection underestimates
the force of Plato’s objectivity about value: while wealth may have some features that would render it useful
only under the leadership of wisdom, those features are not its value or its goodness.
story told about agents, or about states of mind, is to replace a story told about ends,
it had better work to explain how best to live, especially if the received opinions have
been thrown away.
4. One Must Philosophize

The first practical consequence of the claim that wisdom is the only good itself by
itself seems to be this: if wisdom is the sole source of goodness, and wisdom is
acquired by philosophy, then one must philosophize. To reach this conclusion is the
point of the dialogue: to turn young Cleinias to philosophy. It is facilely reached at
the end of the first Socratic episode,51 and reappears at the beginning of the second
(289d). But then the argument seems to run into trouble; and it directly questions
whether (or how) this good ‘itself by itself ’ can support an account of value. The
injunction to philosophize, moreover, exposes an unclarity in the conclusions
reached so far: if wisdom is the source of value, is it also valuable? Is wisdom an
end, worth having, in itself ?
The second Socratic episode focuses our attention on the differences between
treating ‘the good’ as the source of value and treating it as an end. Socrates (surpris-
ingly)52 relies on the assumptions of the early stages of the first Socratic episode. If
philosophy is the acquisition of knowledge, then surely knowledge is only worth
having if it is correctly used? But for correct use, we need knowledge, as agreed at
281a; so do we need the kind of knowledge that will show us both how to make and
how to use what we have made? Then do we need a knowledge that will show us how
to make knowledge and how to use it? What knowledge would that be? Socrates and
Cleinias spend some time trying to figure it out; but they end up stuck:
We were completely absurd: like children chasing larks, we kept thinking that we had just
about caught each of the knowledges; but they kept slipping through our fingers. Why should
I elaborate? When we reached the kingly art, and inquired whether that was what provides and
creates happiness, then we seemed to tumble into a labyrinth, thinking we were at the end, but
then it became clear that we had come round the corner and back to the beginning of our
inquiry, and still as far off as we were at the beginning. (291b)
51
If my account of the argument in 281b–d is correct, Socrates is disingenuous in his reprise regarding
the putative goods; hence the tense of the claim that this is how ‘it seemed to us’ at 282a2.
52
Given, that is, the striking shift of perspective by 281d–e. But this surprise, I claim, is part of Plato’s
strategy to expose the difficulties of a consequentialist account of value. So the new phase of argument is
introduced at 288d by anticipating that it will run into difficulties (Socrates expects that his fate in the
argument will force the sophists to take pity on him); and with some worries about whether they have the
right starting place (Socrates asks Cleinias to remind him). In the event, although Socrates begins
(288d6–7) with the injunction to philosophize, he replicates the argument of the previous episode only
as far as the penultimate stage, missing the significant final phase (281b–e). It is a result of this omission
that the present argument falls to a regress. Notice also the heavy irony in which this exchange is encased:
Socrates’ act of innocent ignorance with the sophists at 287–8, and his wide-eyed admiration of the
speechwriters at 289e.
And a labyrinth indeed it seems to be. If knowledge is one of the things whose value
must be supplied by knowledge, the analysis is either circular (the labyrinth) or
regressive (chasing larks)—the strategy to explain value by appeal to what is good
itself by itself becomes hopeless when the good itself by itself demands explanation in
turn.
They continued the search, nevertheless, but ended with a similar difficulty.
Suppose that the kingly craft is what accounts for the correct use of other things.
What exactly is it that the kingly craft does, when it does that? What is its product? If
the kingly craft is a good thing, then it must have a good product. But it was agreed
that the only good is knowledge;53 so that must be the product of the kingly craft—
making the citizens wise and good. In respect of which of the (many) knowledges
canvassed so far will they be wise? To escape the labyrinth, it must be knowledge of
itself: but still, what is its product? Making other people good? But in respect of what
will they be good? This falls foul of the Corinthian move—it vainly repeats the same
point over and over.54 Socrates is in despair.
We should be struck by the way in which this new phase of the argument is
conducted in terms of knowledge and skill, and eschews mention of wisdom (phron-
êsis, sophia) until the discussion of the dispositions and capacities of the citizens, at
292b. It is, therefore, in sharp contrast with the conclusion of the first Socratic
episode.55 For the effect of Socrates’ present emphasis on knowledge and skills is to
push us back towards the early, instrumental phases of the first episode, to thinking
about value as a means-end structure; and it invites us to ignore what we had come to
understand by 281d. Knowledge here, that is to say, may have either instrumental
value (as a means to some end which is happiness) or value as the end to which
instrumental knowledge may be a means: it does not figure as the source of value.
Once this construal of knowledge as an end runs into trouble, however, we are invited
by the frame to reconsider our position.
By this stage of the dialogue, then, we have been given two quite different models
of ‘the good’. It may be the source of value (as the first episode maintained) or it may
be the most valuable thing (as the second episode presumes). The puzzles of the
labyrinth, in fact, may cause trouble for either model. If we focus on the source of
value, it unravels in opposite directions, and exposes the complexity of what it is to be
good ‘itself by itself ’. If wisdom is the one thing that is good itself by itself so that
everything else has its value by virtue of it, then the relation between what is good
53
292bl–2: ‘But Cleinias and I somehow agreed with each other that the good was nothing but some [i.e.
some particular] knowledge.’ Of course what was agreed was that wisdom is the only good (itself by itself );
and this was earlier taken to mean that wisdom is the source of value, rather than an, or the, end. This
revision of the conclusion of 281d–e is noticeably cagey; and it is put into the elaborate frame of Crito’s
interruption.
54
See Pi N. 7.104–6: to plough the same furrow over and over is bewildering, as if idly babbling ‘Zeus’
Corinth’ to children. ‘Bewilderment’ is described by Pindar as aporia. Socrates would approve.
55
The detailed allusiveness of the frame of the argument suggests that this is deliberate.
itself by itself and what derives value from it is an explanatory or causal one.56 But if
the good itself by itself needs to be explained by its relation to something else, a
regress is inevitable—unless there is some definitive way in which the good itself by
itself explains (or causes) value.57
Contrariwise, the good might be definitive, if it is the sole (the dominant) end. But
if the value of wisdom thus lies in its end-like qualities (it is definitively good, good
without qualification), it is not clear how that will account for the value of anything
else. Further, if we suppose that the regress stops with something definitively good
(the regress affects the ends of practical reason, not of explanation: the end provides a
definitional stop), what is good about it? What is it for it to be good? Its goodness
can’t be explained in terms of its products (for then it is not the definitive end); but
how can we account for it without?
The labyrinth argument, therefore, offers a critique of the argument of the first
Socratic episode. For it suggests that our conception of the good itself by itself may
invite one or other of two difficulties: a regress, or an unexplained definitional stop. If
the function of the good itself by itself is to constitute an end that, by its very finality,
will explain pursuit and action, it had better not be regressive. But such ends have the
air of unexplained definitional stops (the Corinthian move). If, on the other hand, the
function of the good itself by itself is to explain value, it is hard to see how its finality
is explanatory (rather than just a refusal to explain). For explanations, then, regresses
may seem more expansive, and thus more promising. But they may then be circular,
or labyrinthine; in that case, they may not explain at all. The tension between these
two difficulties in understanding the good itself by itself reflects a deeper problem:
when the good itself by itself is the source of value, does it perform this role by being
an end, or somehow otherwise?
Furthermore, this unravelling of explanation (good as the source of value) and
pursuit (good as the end) exposes a gap between knowledge and wisdom. For the first
episode wisdom needs to be the definitive source of good, and maybe also to be
definitively good. But knowledge may not be definitive like that: twice. If knowledge
is like a craft, then it may need to be subordinated to some other craft. And if
knowledge is like a craft, it derives both its value and its content from something
else—namely the thing of which it is the knowledge (its product). Knowledge—thus
instrumentally conceived—fails to have explanatory power in itself. So then is
knowledge distinct from wisdom? Is knowledge distinct from wisdom here? If it is,
how exactly does wisdom constitute a definitive good, itself by itself ?
Return to the rejection of consequentialism. In the first Socratic episode the
wisdom of the conclusion is the wisdom of the person living the life under consid-
eration: and as such it defies consequentialist analysis. For my life it is my wisdom:
56
These explanations, in the structure of value, are existentially committed: for they cite real features of
the world in the explanans. See Sedley (1998); McCabe (2000), ch. 6.
57
This is a familiar problem for Plato: see Parmenides 130–5.
and that performs the role of the good itself by itself for me. Knowledge, on the other
hand, need not be mine to be ethically significant to me—it just needs to be effective
in my life, to provide me with active luck (some philosopher-king would do the trick,
as Socrates and Cleinias ostentatiously concede at 290b–e).58 But knowledge is lost in
the labyrinth, if knowledge is construed according to the two conditions of a
consequentialist analysis: as deriving its value from something else (its end) and as
not necessarily mine (borrowed knowledge is useful to the acquisition of ends, but
not as the source of their value).59 But if not knowledge, how does Plato offer us a
wisdom that escapes the labyrinth?
This difficulty reflects another, and demands its solution in terms of the overall
project of the dialogue. Socrates wants to show Cleinias a (serious) protreptic to
philosophy. It seems, moreover, that he has provided one: if wisdom is the only good
itself by itself, then we should philosophize, just if that is the process by which
wisdom is acquired. But this may be too simplistic, especially in view of the puzzles
posed by the labyrinth. Is philosophizing merely a means to the end of wisdom? And
if wisdom is then end-like, does it lead us into the labyrinth? How else might the
doing of philosophy relate to the doing well that is wisdom? If Cleinias is to be
encouraged to philosophize, what exactly is in it for him?
5. Who’s Who?
It is at the point of impasse that the dialogue takes a new direction. The conversation
between Socrates and Cleinias—about exactly which knowledge is the object of
philosophy—had been going as such conversations usually do, when suddenly the
worm turns. The hitherto passive and bashful Cleinias delivers himself of some
opinions. Speechmaking is not the answer, he says, nor is generalship. Generals,
after all, hand their catch over to others to deal with; likewise geometers, astron-
omers, and calculators all hand their product over to the dialecticians. Cleinias seems
to have sneaked a look at the Republic while Socrates was arguing with the sophists
about contradiction; and even Crito is astonished (290e–291a):
crito: What are you saying, Socrates: did that youth utter those words?
socrates: Do you think he didn’t, Crito?
crito: No, I certainly do not. For I think that if he did, he would not need educating at the
hands of anyone, whether Euthydemus or any other man.
58
There is a clear reference here to the Republic which the argument exploits. In the Republic, the
happiness of the citizens can be provided by the knower (the philosopher-king) to whom they delegate
their decisions; here, I say, Socrates rejects that view of the centre of ethical causation.
59
The wisdom is owned by whoever lives the life in question—the wisdom is mine because this life
belongs to me. There is nothing special here, as far as I can see, about a first-personal perspective, other
than its expression of ownership. There is still, however, a conception of the self here, albeit not one built
on Cartesian introspection. See further on ancient and modern conceptions of self Taylor (1989); Gill
(1996b), especially introduction; Sorabji (1999).
socrates: Well, by Zeus, perhaps it was Ctesippus who said this, and I have forgotten?
crito: What sort of Ctesippus?
socrates: Well I know for sure that it was not Euthydemus nor Dionysodorus who said that.
But, my good Crito, maybe it was one of the superior beings who was present and uttered
these words? For I know very well that I heard them.
crito: Yes, by Zeus, Socrates—it seems to me that it was one of the superior beings,
and very much so. And did you find what you were looking for, or not?
It is tempting to read this little exchange in code. Who really said all this stuff about
knowledge and dialectic? Why is Crito sceptical about its author? Is it because it was a
clever thing to say? Or is it because it displays unusual cooperation with the Socratic
enterprise—so it couldn’t be Ctesippus or the brothers either? Who does Crito think
said it? Does he think Socrates said it (a superior being)? Or does Plato make Crito
imply that Plato said it? And why should any of that matter so much? (Should we
read Plato’s dialogue as a code?)
These questions devolve onto another—why should this be an extraordinary thing
to say, anyway? First, the content: the reference to dialectic and its advantages over
astronomy seems to be an elaborate appeal to the high theory of the Republic. Second,
the argument: Cleinias here argues with vigour, supported by reasons and on his own
behalf. Crito’s insistence that it could not have been Cleinias suggests that there is
something special—even divine—about being able to argue vigorously, on your own
behalf, and with grounds. But that ability does not seem to be productive (Cleinias
just forces renewed aporia): instead, this kind of reasoning seems to be (divinely)
valuable in itself, and certainly what education is for (since if he has it, Cleinias
doesn’t need educating). Does that make this reasoning an end, without entering the
labyrinth? Does this somehow, whether as an end or otherwise, account for its being
the source of value?
How you reason is here understood to be internal to who you are (‘what sort of
Ctesippus?’).60 This thought is not new in the dialogue. Consider, for example, the
earlier attack by the sophists on the importance of consistency (285d–286b). When
Socrates complains that what they say is inconsistent with what they said before, they
reply that what they said before has no bearing on what they say now: if Socrates
insists that it does, that is because he is an old Cronos (287b). Socrates’ character-
ization of intellectual positions repeatedly cashes issues in rationality (consistency,
contradiction, the differential commitment to some point of view) in terms of who
someone is. Someone may tell the truth over time (like Proteus at 288b–c); observe
consistency over time (like Cronos); or insist on saying what they really think (as
Socrates advises Cleinias at the beginning of the first Socratic episode). These are no
mere myths or metaphors: on the contrary, the justification of consistency, of telling
60
See Chapter 10.
the truth, and of saying what you really think is delivered in terms of the person who
does so, the person whose identity is conceived as reason. Reason—or wisdom—is
both developed and deployed through philosophical discussion. It is not something
that can be delegated or borrowed; in a philosophical conversation, both parties are
actively engaged just so long as they speak for themselves (recall Socrates’ insistence
on this in the discussion of the putative goods). And it is not something that derives
its value from its product or its effects: if reasoning well is an aspect of divinity,61 it
must be itself a part of the good.
So why should Cleinias philosophize? How does his philosophizing make any
difference to how he best should live? The process of argument, on Socrates’ account
in this dialogue, is one of establishing the internal coherence and consistency of
someone’s view: and this is sharply differentiated from sophistic argument, in which
it does not matter what you say (you will be refuted, whatever). But coherence and
consistency, once established, are not the means to wisdom; they are constituents or
conditions of wisdom.62 And if wisdom is the good itself by itself, it will give value to
its conditions not instrumentally, but as a matter of its own constitution. It is
necessary to philosophize, therefore, because philosophizing is the development of
wisdom; both its value and its conditions are dependent on wisdom.
Philosophizing, what is more, is the process in which each interlocutor is himself
engaged, on his own behalf (it is not the consequence of luck, whether active or
passive). In philosophizing—so the Euthydemus contends—the philosopher develops
(or changes or integrates) himself (thus, the protreptic works on Cleinias by 290b).63
In so doing, the philosopher proceeds towards wisdom at the centre of the best life,
the source of its value. So is wisdom valuable in itself? And is it by being valuable in
itself that it confers value on everything else? How far does this account of the
centrality of reason move away from an account of value determined by the relation
of means to ends?
Certainly, wisdom itself, once developed (if it is ever developed),64 equally imparts
value to other things: nothing here entails that there are no other goods, or no other
ends than wisdom. So it does not follow that the happy life just is philosophizing, or
that philosophy exhausts all the dependent goods. But it does show that the relation
between philosophy and wisdom will transform the life of the philosopher. It will
transform, of course, the list of dependent goods (for that list is still up for grabs); and
it will transform the philosopher herself. It is thus part of a structure of value that
centres on the person who may live the happy life; wisdom makes that life coherent
and integrated, at the same time as it identifies the liver of the life. Then wisdom
61
See Sedley (1999).
62
Notice, for example, the connection between consistency and the ancient divinity, Cronos, at 287d.
63
The theme of personal identity is pervasive; McCabe (1998).
64
This is a thorny issue: does Plato need to insist that there ever is or will be a wise man if he is to make
this account of wisdom make sense? See Gerson (2003).
seems to have two roles: it is, for the individual, the source of happiness in her life;
and it is itself a good thing, a state of soul incontrovertibly worth having. Wisdom is
both the source of value and an end; and so an end like no other.
There is, therefore, a dialectical relation between the two Socratic episodes of the
Euthydemus. The first concludes by representing wisdom as the source of value, and
places little emphasis on its function as an end. The second, by contrast, begins by
treating knowledge as an end, and claims that to be incoherent. The way out of the
labyrinth is found by postulating the wisdom of individual persons (if it is attainable)
as what renders things valuable, since this is what structures a person and a life. But
then wisdom is also valuable: not as its own product, but in itself. In that case,
wisdom’s role as an end is sharply demarcated from that played by the putative goods
with which Socrates began. The effect of this dialectic is to show just how the claim
that wisdom is ‘the good itself by itself ’ needs to be interpreted; and thence to
distinguish the source of value from valued ends. That distinction is a weapon in
Plato’s armoury against consequentialism.
6. The Moral Phenomena and Objectivity

Only if we read the sequence of arguments in the Euthydemus in this dialectical way
will a distinction emerge between wisdom as the source of value and wisdom as
valuable itself. But such a dialectical reading is pressed upon us by the frame
arguments, whose complex interplay allows us to understand the structure of the
dialogue as a whole.65 What does it give us, as a moral theory? What does it say to the
consequentialist? And what—to revert to the questions with which I began—does it
tell us about objectivity in ethics?
We might think that consequentialist explanation is inadequate because it fails to
give sufficient weight to the agent in ethical explanation. And we might think this, as
I suggested, just as a matter of the moral phenomena: no ethical explanation will
satisfy that fails to give a full account of what it is like for the agent to be involved
in the moral life. But is this Plato’s point? His arguments here, if I have them right,
are not primarily focused on the agent as such. Instead, they insist that the central
location of value should be the person. This gives no commitment (one way or
another) to whether the best life for a person is one that characterizes him as an
agent, as a doer of deeds, rather than as a thinker of thoughts. This is a far cry from
consequentialism, because it locates all primary value in the person; and its explan-
ation of value does not specify how that should be translated into action, or how it
views the accumulation of ends. As Cleinias is brought to see, Socrates maintains that
we should not pre-empt the theory by assumptions about the moral phenomena. The
65
See Gill (2002); Penner and Rowe (2004).
life of the wise man may be active; or it may be quiet—whichever will depend on how,
as a matter of fact, wisdom organizes its value in a life.
All this should force us to rethink what is meant by Platonic intellectualism.
A common view of the intellectualist focus of Plato’s ethics takes it to be based on
something like a craft: since value is an objective feature of the world, and is
computable, then what we need for happiness is the craft that allows us to make
successful ethical computations: a measuring skill. This sort of claim, I suggested,
may be what we find in arguments such as the Protagoras’ denial of akrasia; and these
arguments proffer an intellectualist ethic as a strategy for arriving at the most
possible goods. For they suppose that there are no countervailing psychological
drives other than knowledge (so that failure is ascribed to ignorance). So knowing
what course of action maximizes the good is sufficient to motivate us to do that
action; and, as I argued above, this makes tangential the role of the agent herself in
the ethical life (since reliable luck, or reliable and clever friends, would suffice). On
such a consequentialist account, as I argued, virtue occupies a subordinate position in
the explanation of value.
The arguments from the Euthydemus we have considered, however, take a differ-
ent view. They are intellectualist because they suppose that wisdom is the source of
value, rather than the source of our acquiring value. Wisdom, then, is central to
ethical explanation. But that centrality, I argued, puts the person in a central position,
and it supposes that value flows from there to the exercise of rationality: philosophy.
So far, so intellectualist. But does this now move Plato right away from any claim he
might have to ethical objectivity? If the good is not in the known (such as the form of
the good) but in the knower, how can it be objective?
Of course if the virtue that is wisdom is a state of mind, it does not follow that
its value is merely mind-dependent. On the contrary, in fact: for wisdom is a real
disposition of the person: it has character over time, and coherence at a time (indeed,
the sophistic sections of this dialogue are dedicated to examining these metaphysical
assumptions).66 What is more, wisdom’s content is (I take Plato consistently to
suggest)67 what corresponds to reality out there: knowledge is of what is, how it is.
The good itself by itself, in both these ways, is real, even if it is a real feature of persons,
and as such it is a suitable object for metaphysical study. This real feature of persons,
moreover, will be, in this very specialized sense, virtue, and as such it is a suitable
object for ethical study. Plato’s metaphysics of morals, that is to say, rests on his
account of the person who is wise; and that person is who the philosopher is
becoming. Plato is, notwithstanding his silence about the form of the good in this
dialogue,68 committed to the thought that ethics and metaphysics are mutually
66
McCabe (1998).
67
See e.g. Republic 476–80; Sophist 243–5—despite what I think are radical changes of mind.
68
We need not suppose that the form of the good is the only way in which objectivity about value might
be expressed by Plato.
dependent; and that inspires his account of the objective good. Virtue is determined
by the nature of things.
What does that then mean for Plato’s moral theory? What we find here is not, as
I have argued, a theory that begins from the moral phenomena; and as such it would
find little assistance from the prevailing wind nowadays. Plato’s methodology does
not start from the middle of things;69 and it is a consequence of this that the theory
that eventuates is so startling. It is a theory that eschews the thought that while
wisdom may explain happiness, there is some other source of the imperatives of
morality (it is not, then, an example of evaluative dualism). No more, as I have
suggested, is this theory based on the great metaphysical structures of the Republic—
whether the Euthydemus anticipates the Republic or corrects it or is merely ignorant
of it, there is no form of the good here. But that does not mean that in this dialogue
Plato’s moral theory starts from nowhere. On the contrary, it starts from some
striking (and strikingly persuasive) claims about the identity of persons; and con-
tinues with some insights into how that may be related to the proper practice of
philosophy. The starting points, that is, are not the moral phenomena; they are
metaphysical theory. The Euthydemus does not suppose that there are lines of
demarcation already drawn between the study of value and the study of what is
real. This appeals to objectivity most of all.
Suppose we grant Plato all of this: what then does his theory say to a modern
account of consequentialism? We might start, perhaps, with the labyrinth; and recall
how Plato gives structure to the ways in which the good may function in ethical
accounts. For his arguments, I have suggested, separate the good as the source of
value from the good as an end, insisting that end-like explanations are insufficient for
explanation. In doing this, they make the claim that in reaching ethical explanations
we should look not to the value of some good external to the ethical agent, but
internal to him: to some state of the agent that in itself explains value. Nay more:
Plato insists that the intrinsic good is not a state of an agent, as such, but rather a state
of persons—leaving undetermined whether what constitutes happiness does in fact
arise from agency at all.
This is, you might complain, a thoroughly rarefied conception of the source of
value, which begins from a highly abstract conception of what the person is, the
idealized possessor of wisdom. In that case, such a response to the consequentialist
approach to value and action may seem to be entirely beside the point; a theory that
just fails to start where it must, from things as they seem to us from amid the hurly-
burly. But why, Plato might respond, should the highly abstract not figure in the
determination of value? Why, he might object, has it become a matter of doctrine that
we should start from the deliverances of moral anthropology?
69
Williams (1985).
Plato does not—we should remind ourselves—proceed from what he says about
wisdom to a view of the best life that must be ascetic, or quietist. Rather, he argues
that whatever actually turns out to be the best life has its source in wisdom; and that
state of mind is one we do not yet occupy. In that case, what he has to offer here is not
actual suggestions for the content of the best life; rather, he offers us some account of
how to go about thinking about it. This approach (and not a pre-existing prejudice
against the goods of the vulgar conception) is what makes for the abstractness of the
account. But does this in turn mean—as the advocates of moral anthropology insist—
that ethical inquiry, conducted in such an abstract way, is thoroughly limited, unable
to engage with any of our own realities? If our realities are limited to the phenomena
of ethical experience, they may well be inimical to an abstract approach. But it is not
obvious that this tells against abstraction, except for the broad sceptic. For the
phenomena, by virtue of their very particular nature, are vulnerable to the objection
that they are mere collections of subjective appearances, with no application beyond
the fact that they are believed, or acted upon by someone, somewhere.
Plato offers us a different view, one that responds both to a complaint that ethical
theory should not be self-absorbed and to the objection that it should provide more
than a collection of subjective opinions. For his account of the origins of value relies
not on anthropology or market research, but on theses from elsewhere in the
philosophical spectrum—from metaphysics and epistemology. He suggests that
there are answers to be found to the structure of ethical explanation from considering
broad philosophical questions: what counts as a person? How is it that a person may
be related to their wisdom, to their actions, to their own experience of happiness? If
the origin of value is wisdom, what constitutes wisdom? How are wisdom and
knowledge to be separated from belief and confusion? How are they to give us access
to the truth? This, I submit, provides us with an approach to the study of value that
confronts the issues of subjectivity and objectivity where they belong. For this is not a
separate puzzle among the moral phenomena of ethics, but one to which ethics,
metaphysics, and epistemology may jointly contribute when we try to understand
just how to go about asking ‘how best to live?’.
Plato does not espouse subjectivism; but he supposes that real value depends on
the person who is involved in it. This implies in turn that any view of value that
dissociates it and makes it independent from the person involved, is just wrong; so
that consequentialism must be false. Plato attacks consequentialism, that is to say, on
the grounds of ethical metaphysics—and not, as other opponents of consequential-
ism do, on the grounds of the moral phenomena, or moral anthropology; or yet on
the basis of some other a priori truth that generates evaluative dualism.70 Conse-
quentialism gives us the illusion that in talking about ends, we speak of facts; and it
70
The conclusion of the argument of Euthydemus 281 denies pluralism, and so dualism.
hopes to ward off the vagueness of virtue theory (let alone the austerity of categorical
imperatives) by offering something like objectivity. Plato has a response: objectivity
may not be just out there, in what we pursue; that does not preclude value’s being
real, or its being ours.71
71
I am grateful to many people who have given me advice: notably Zsuzsanna Balogh, Sarah Broadie,
Peter Gallagher, David Galloway, Christopher Gill, Michael Lacewing, Terry Penner, Anthony Price,
Patrick Riordan, Christopher Rowe, Janice Thomas, and Ann Whittle. They should not be held responsible
for the consequences of their advice.
III
On Aristotle’s Conversations
with Plato
14
Perceiving that We See and Hear:
Aristotle on Plato on Judgement
and Reflection
1. Perceiving that We See and Hear: The

Opening Gambit
I begin with a short, and famously vexed,1 passage from Aristotle’s de anima:
Since we perceive that we see and hear, it is necessary either to perceive by sight that one sees2
or by some other (sense). But the same sense will be of the sight and of its underlying colour.
Then either there will be two senses of the same object or the same sense will be of itself.
Further, if indeed the perception of sight is other [sc. than sight] then either the perceptions will
go to infinity3 or some sense will perceive itself. So we should assume the latter in the first place.
But this involves a puzzle: for if perceiving by sight is seeing, and what is seen is colour or
what has colour, then if some sight is going to see the seer,4 the first seer will have colour. Well,
then, it is clear5 that perceiving by sight is not one.6 After all, when we do not see, it is by sight
that we judge both the darkness and the light, but not in the same way.7
1
See, for example, Hicks (1907) ad loc.; Ross (1961) ad loc.; Hamlyn (1968) ad loc.; Kosman (1975);
Osborne (1983); Caston (2002), (2004); Sisko (2004); Johansen (2005). I translate the text of Ross’ OCT
(1956) except at 425b19, where I follow his (1961). In what follows, I have tried to avoid elephantiasis of the
footnotes by only mentioning points from the commentators that directly affect my argument.
2
The Greek shifts from ‘we perceive that we see . . .’ at 425b12 to a third-person formula (‘to perceive
that one sees’ at 425b13) which allows the construal that it is sight that perceives that it sees.
3
Kosman (1975) argues that the infinite regress implies that second-order perception must be a necessary
condition of first-order perception; so too does Caston (2002), (2004). Johansen disagrees, but argues that his
interpretation (that this is an account of an ‘inner sense’) goes through anyway (2005). More below, }3.
4
Ross (1956) prints horan here, twice, with less authority than the majority of MSS, which have horôn;
contrariwise he prints the latter in his (1961). Support for the latter is supplied by horôn at 425b22,
provided we can understand the role of the intervening sentence. Ross makes the proviso that in 19 the
reference is to the faculty of sight, seeing, and to the absurdity of the faculty’s being coloured, while in 22
Aristotle must be talking about the organ’s being coloured. See Caston (2002), (2004) and Johansen (2005)
for different views of this issue.
5
Osborne (1983), 402 n. 7 takes this to introduce a new point; below, I disagree.
6
This is baldly put; what it does not say is that perception is ‘said in many ways’, vel sim. (compare and
contrast the discussion of actuality and potentiality in perception, 417a9 ff., 426a23–5, or the use of
perception as an example of homonymy at Topics 1.15; and see Burnyeat (2002)): this need not be the claim
that the expression ‘perception’ is ambiguous, rather than that the phenomenon of perception is not simple
or uniform (compare the different ways of denying unity at Metaphysics 1017a3 ff.).
7
This must mean ‘not in the same way as we see colours’, and not that we see light in a different way
from seeing dark; see Ross (1961), ad loc.
284 PERCEIVING THAT WE SEE AND HEAR
And again, that which sees is coloured8 in a way; for each sense-organ is receptive of what
is perceptible without the matter. This is why even when the sense-objects are no longer there,
there are (still) perceptions and appearances in the sense-organs. (Aristotle, de anima
425b12–25)
Aristotle’s opening gambit, ‘Since we perceive that we see and hear, it is necessary
that we perceive that we see either by sight or by some other (sense)’, gives the
appearance of being a fresh start—although that appearance may mislead. The
previous chapter was about whether we have just the five senses; but now, or so it
seems, Aristotle has shifted his attention to a different issue. If we read Aristotle as
I was brought up to do, we might be unsurprised by his lacunose style (surely, we
reassure ourselves, these are just lecture notes?9). And—however new the point—
does Aristotle take it as just obvious that we perceive that we see and hear? The
obvious is where he tends to begin, for new points and even some old ones.10 Is it
obvious that we perceive that we see and hear? What is it to perceive that we see and
hear? I shall wonder about the continuity, or otherwise, between what Aristotle
thinks is obvious, and what we do; and I shall wonder just what he is advancing
here as an explanation of perceiving that we see and hear. En route I shall ask a bit
more about Aristotle’s method, and his ‘style’.
2. The Snares of Consciousness

So: ‘Since we perceive that we see and hear’ . . . do we? And when do we? When we see
and hear, do we always perceive that we do? Or just sometimes?11 It has been
8
The perfect tense here may recall the discussions of change, process, and completion in de anima 2,
especially chapter 5.
9
See, variously, Hamlyn (1968), 121; Kahn (1979b), 50; Osborne (1983), 401; Barnes (1995), 11 ff.;
Burnyeat (2002).
10
See e.g. Top. I. 1, E.N. 7. 1. One possible explanation of the shift from the first-person plural here to
the impersonal formulations of second-order perception might be that ‘we perceive that we see’ is taken to
be one of the phenomena (as also the first-person plural krinomen at 425b21) from which we should begin
our investigation.
11
There is nothing about the grammar of the present tense of ‘we perceive’ that determines the matter
(and see }9 on a parallel passage in E.N. 9.9): both in English and in Greek the present tense can be
aspectual: continuous (‘I breathe’); iterative (‘I get up in the morning’); expressive of a capacity (‘I play
tennis’, ‘I solve crosswords’); and conative (‘I fight off the Persians’). Kosman (1975) argues that Aristotle
here is offering an account of perceptual awareness; Caston (2002), (2004) that for Aristotle perception is
always complex, and reflexive; Johansen (2005) that although the evidence to show that we always perceive
that we see and hear is disputable, Aristotle here supposes that the complexity of perception needs to be
explained by an inner sense. Connected here is the debate about what Aristotle might mean by a ‘common
sense’ (see e.g. Kahn (1979b), Osborne (1983)) e.g. at de somno 455a12 ff.: ‘ . . . There is also a common
faculty associated with them all, whereby one perceives (aisthanetai) that one sees and hears (for it is not by
sight that one perceives that one sees; and one judges (krinei) and is capable of judging that sweet is
different from white not by taste, or by sight, nor by a combination of the two, but by some part which is
common to all the sense organs; for there is one sense-faculty, and one paramount sense-organ, but the
mode of its sensitivity varies with each class of sensible objects, e.g. sound and colour) . . .’ (trs. Hett (1964)
modified).
PERCEIVING THAT WE SEE AND HEAR 285
argued12 that this passage does indeed start from the obvious: the obvious of the
subjective side of perception. When I see, there is something that it is like for me to
see—and this ‘something that it is like for me’ is itself perception.13 This conscious
side of perception14 is a regular concomitant of ordinary acts of perceiving. So if
I smell the baking bread, there is something that it is like for me to smell it; if I see the
vivid pink scarf, there is something that it is like for me to see it; and so forth.
I am sitting at my desk in the late afternoon, and the smell of buttered toast wafts
through the door. When the smell hits me, it seems somehow or other to me, there is
something that it is like for me to perceive the delicious aroma. My consciousness of
the smell is not an inference from the smell (the inference is my muttering, as I leap
from my hard chair, ‘Oh, good, teatime!’), but rather just the subjective side of the
perceptual event, my side of the smelling that goes on: coextensive with the drift of
buttered toast, but importantly my end of the business.15
Phenomenal consciousness might just be this subjectivity that marks me out as a
perceiver, rather than as merely an object in the way of the buttered-toast-smell as it
wafts. When we perceive, being conscious may be an irreducible and primitive
feature of perceiving. Or consciousness may be intrinsic to perceiving, but susceptible
to analysis or explanation (maybe I smell the toast, and at the same time I have
some consciousness of the smelling, some perspective on it). However we account for
the feeling involved in perceiving, it is distinct from something more reflective: the
turning of my attention to what I perceive. I am sitting at my desk, and the argument
I am trying to construct is stuck; and I think about what time it is and turn my
attention to the smell of buttered toast that my ennui makes compelling to me. Or
someone asks me: ‘is that buttered toast or burned sausages?’ and I think about it
(and am disappointed that it is sausages, not toast). This sort of reflection—reflecting
on my smelling of sausages and toast—could be construed as something perceptual
(although it need not: do I perceive my smelling of the sausages, or do I just think
about it?). It may, further, mark a difference between the noticed smell of sausages
and some unnoticed one, and so might be described in terms of consciousness.
Nonetheless, reflection of this perceptual sort can be quite complicated (deliberate
sniffing and attending); and it is not intrinsic to the first-order perception, which may
frequently occur without it. So one of the broad lines of demarcation between
phenomenal consciousness and reflective consciousness may be this: phenomenal
consciousness should be a regular feature of ordinary perception, while reflection
may be irregular, piecemeal, not necessary at all to the perceptions I ordinarily have.
12 13
Kosman (1975). Nagel’s famous formula (1979).
14
I shall call this what it is like to see, hear, etc., ‘phenomenal consciousness’, following e.g. Tye (1995).
On the various senses of ‘consciousness’, see Lycan (1996), ch. 1.
15
Caston (e.g. 2002, 2004) takes Aristotle to account for consciousness in terms of the complex mental
content of perception.
If Aristotle is indeed talking about phenomenal consciousness here, however, it

might be a huge and exciting moment in the history of philosophy: the moment when
it was first understood that there is something peculiar, special, strange about
perception, from the subject’s point of view.16 Aristotle’s predecessors—so this
moment would be described—missed what perception feels like and concentrated
instead on how it comes about (notably by the affection of the perceiver by the
perceived), while Aristotle, the forerunner of modernity, saw perception right.17 Is
that correct? And is Aristotle, by contrast with his predecessors, taking (what we
might call) phenomenal consciousness to be one of his phenomena? Or is he talking
about something more reflective?18 In what follows I shall suggest some reasons why
reflectiveness may indeed be his focus of attention.19 I shall also suggest some reasons
why we might not think that so dull.
If this is about phenomenal consciousness, Aristotle must, I think, at least commit
himself to the following conditions:20
• regularity: being (phenomenally) conscious of an act of perception is a regular,
even if not a constant, feature of individual events of seeing, hearing, etc.21 This
just is what it is for perception to be conscious. So, here, ‘since we perceive that
we see and hear’ would translate as ‘since we regularly perceive that we see and
hear . . .’ or, in Aristotelian, ‘since we always or for the most part perceive that we
see and hear . . .’.
• ‘what it is like for me’, part 1: when I perceive, there is a subjective or affective
aspect of that perception, which distinguishes between this being a case of
perception and its being a case where the perceptual quality of one thing is
somehow directly transferred to something else.22
• ‘what it is like for me’, part 2: this subjective aspect is special to me. So
phenomenal consciousness is (somehow or other) indexed to the subject who
has it. Thus, here, ‘since we perceive that we see and hear’ would parse as ‘since
I perceive that I see and hear, and you perceive that you see and hear . . .’,23 etc.
Are these three features of phenomenal consciousness in Aristotle’s sights here?24
In what follows I argue that the regularity condition may not be met; and that this
16
Aristotle himself may be making the claim to innovation at 426a20 ff.; and compare 427a21 ff.
17
This is not Aristotle’s own account of his difference from his predecessors: see 427a17 ff.
18
See e.g. Kahn (1979b).
19
Osborne (1983) takes a slightly different view of reflectiveness.
20
On consciousness in general, see e.g. Tye (1995); Lycan (1996); Papineau (2002).
21
There are obvious tricky cases: just how conscious am I of the smell of buttered toast while I am
concentrating on writing this sentence?
22
Here see the debate about spiritualism versus literalism, especially between Burnyeat (e.g. 1992) and
Sorabji (e.g. 1992); and compare e.g. Caston (2005).
23
And you don’t perceive (in the relevant way) that I see or hear, or vice versa.
24
I do not deny that Aristotle may sometimes be interested in something like consciousness; see, e.g.,
the cases where he contrasts alterations that ‘escape notice’ with those that do not (e.g. at Phys.
244b10–245a11, Sens. 437a26–9), discussed by both Caston (2002), e.g. 757, and Johansen (2005), 265;
may cause us to have some suspicions about how to understand Aristotle’s interest in
what it is like for me to perceive.
3. The Argument to ‘Self-Perception’

The construal of the argument is contested: here is just one account of it.25
1) We perceive that we see and hear (425b12).
2) We perceive that we see and hear either by sight26 or by some other sense
(425b13).
What follows is usually interpreted as an argument to resolve what seems to be the
choice here (perception of sight is done either by sight or by some other sense) in
favour of sight.27 This seems to be broadly right;28 but it does so by means of an
interim conclusion, that when we perceive that we see or hear, it is a case of self-
perception: of the perception’s perceiving itself.29
3) Sub-argument 1 (425b13–15): In either case (sc. when I perceive that I see),30
there will be the same sense of the sight and of the (first-order) colour.31
4) So either there are two perceptions of the same object;
5) Or [there is one, and] perception perceives itself.
6) Sub-argument 2 (425b15–17): If indeed the perception of the seeing is done by
some other sense than sight, either there will be a regress of perceptions (which
is impossible);
7) Or the perception will be of itself.
or compare Burnyeat’s account of perception (1992). But in these passages he uses the verb lanthanein to
make his point and not the ‘perceiving that I see and hear’ locution.
25
Four notable, and different, detailed accounts are Kosman (1975); Osborne (1983); Caston (2002);
and Johansen (2005). In what follows my interpretation comes closest, in different ways, to those of
Johansen and Osborne.
26
Johansen (2005), especially 246 ff., 252 ff., argues that this must refer to the faculty of sight; Caston
(2002), (2004) that we are talking here about individual ‘activities’. The Greek does not determine the
matter, either for perception or individual sense-modalities: both aesthêsis and e.g. opsis can refer to the
faculty or the activity.
27
Osborne disagrees (1983).
28
When Aristotle says, at 425b20, that ‘perceiving by sight’ is not one, this seems to follow from the
argument about self-perception; it suggests, therefore, that self-perception happens somehow or other by
sight (or whatever the appropriate sense-modality).
29
Notice the emphatic position of autê hautês at 425b15, and of autê tis estai hautês at 425b16. The
reflexive, ‘self ’, refers back to the perception, not to its subject: self-perception is perception of perception.
30
Osborne (1983) suggests that this applies to both options; not so e.g. Kosman (1975) or Hamlyn
(1968), ad loc.).
31
Why should the second-order perception include the first-order object? Osborne rightly observes that
this is an easy consequence of Aristotle’s account of how (at least first-order) seeing and its object are
actualized at once, in the sense-organ itself. Perceiving the first-order subject must be perceiving it as
seeing; so it will be perceived as actualized along with its object; see also Johansen (2005), 243.
8) So, when we perceive that we see and hear, this is self-perception.32

Sub-argument 1 focuses on the claim that any higher-order perception must have
as its object not only the first-order perception, but also the first-order objects. Sub-
argument 2 focuses on the claim that any higher-order perception must itself be
either perceptible or perceived. Together they seek to show, twice over, that percep-
tion perceives itself. So either all or some perception is self-perception.33 Aristotle
takes that conclusion to imply34 that higher-order perception is done by the same
sense-modality:
9) [So when we perceive that we see, we perceive that we see by sight: and the
same, mutatis mutandis, for the other sense-modalities.]
The two sub-arguments exploit two assumptions, which themselves come under
scrutiny in the sequel:
i) manifest at 3, about the content of the perceptions: if there is a higher-order
perception of a first-order seeing, the higher-order perception will include the
content of the first-order one.
Think about how it would be if this were false. When I perceive that I perceive, my
first-order perception is both subject (of its own object) and object (of the higher-
order perception). In perceiving that I perceive, however, I perceive at least the
subjective aspect of the first-order perception (otherwise, it would not be perceiving
that I perceive). Suppose, nonetheless, that I do not perceive what I perceive. My first-
order perception’s being the subject of perception is explained by its being somehow
altered by its object, on Aristotle’s view. So if the higher-order perception were to
perceive the lower, without perceiving what it perceives, it could not perceive it as
perceiving, would not perceive that it perceives. So it is plausible to suppose that the
higher-order perception in some way includes the content of the first-order one. If
the higher-order could be perception by sight, perhaps Aristotle invites us to think
about this as though the first-order seeing is somehow transparent, so that the
colour it sees will be perceptible at the second order.35 If seeing sees its proper object,
higher-order perception perceives right through to the first-order object.36 Or if the
32
This is a gloss of ‘So we should assume the latter in the first place’: the interim conclusion, on the
interpretation I offer here, from the thought that we perceive that we see and hear.
33
That this is problematic drives the dialectic of what follows. For the faculties of perception cannot be
constantly self-actualizing (417a3–5: see Burnyeat (2002), 38).
34
Aristotle does not make 9) explicit; but he begins the next sequence with the assumption that we
should be talking more generally about sight, 425b18. This may be consistent with the de somno passage
(see n. 11) if both passages aim to deny that a sense-modality of a single order may both perceive and
perceive that it perceives.
35
Kosman’s objection (‘why from the fact that we see . . . that we see, should it follow that we also
see . . . that which sees, and thus that . . . that which sees is coloured?’) rests on the prior assumption that it is
phenomenal consciousness that is at stake here; the same may be said, perhaps, of Hamlyn (1968), 121.
36
This claim is itself as theory-laden as the ‘phenomenal consciousness’ interpretation, of course: for
Aristotle, ‘seeing through’ would be understood in terms of perception as alteration of some kind.
first-order perception is understood in terms of change, then the change caused by

the first-order object is somehow transitive to the second. (This is an assumption,
then, about the objects of perception).
ii) clearest in the second strategy, about counting the perceptions: if there is a
higher-order perception of a first-order one, will we have two perceptions here?
And if two, then perhaps three (if I perceive that I perceive that I see), or more.37
But this pluralizing is a bad thing (it doesn’t explain, after all, what the perceiving
that I see and hear is): so we should block it at the first stage, and insist on there
being just one, so that we may say that perception perceives itself.38 Call this
(whatever it is of ) the reducing assumption:39 even if perception is ordered, it is not
thereby merely plural; instead, the ordering of perception amounts to self-
perception.
The reducing assumption appears twice: in the choice of self-perception over there
being two perceptions of the same object (at step 4); and over there being a regress of
perceptions (at step 6). The symmetry of the two arguments suggests that steps 4 and
6 somehow deploy the same assumption: but the nature of the regress in step 6 is
vigorously disputed.40 Is Aristotle’s claim that any perception (of any order) is also
actually perceived—so that of any perception there is a higher-order perception? This
would invite a regress, unless it is blocked by some self-perception. Or is the claim
that any perception is also perceivable—so that of any perception there may be a
higher-order perception? This—it has been suggested41—may not invite a regress at
all, but would just peter out when the higher-order perceiving happens to be done (so
it supports no inference to self-perception). If the latter (perception is perceivable) is
too weak to force the regress, then we may prefer to find the former here (perception
is perceived)—after all, Aristotle does indeed conclude that perceiving that we see
and hear, since it cannot be regressive, implies that perception perceives itself.
But if his point is that perception is perceived, then self-perception will accompany
all (or most) perceptions. That favours an interpretation of the argument as a whole
that has it talk about the regularities of phenomenal consciousness. It favours, also,
the view that the topic here is some individual event or activity of perception, rather
37
If the arguments are parallel, as I suggest they are composed, the point about an infinite regress
matches the point about there being two perceptions here—so pluralizing, rather than an actual infinite
regress, may be Aristotle’s real target here.
38
But: just one what? Perception in general? Sense-faculty? Activity of perception? Of course, see
Barker (1981), the discussion about faculties is itself parasitic on discussion about individual events, at least
later in the chapter (e.g. 425b26 ff., 426a27 ff.); equally, the faculties are supposed to explain what happens
in the individual cases; hence, perhaps, the instrumental datives e.g. at 425b13, 18.
39
That the reducing assumption is one focus of Aristotle’s attention is indicated by the apparent
tensions in what he says: if higher-order perception is self-perception, perception is somehow just one; but
in that case, also, somehow perceiving by sight is not one. If Aristotle writes with care, this tension is part of
his argumentative strategy, not just a gap in his argument.
40
Notably by Kosman (1975); Caston (2002); (2004); and Johansen (2005).
41
Kosman (1975); and in his response to Johansen (2005).
than the faculty of perception, or the individual faculties of sight and hearing.42 For if
any given event of perception is perceived, then (generalizing) all perceptions will be
perceived; and the inference to self-perception will generate an account of phenom-
enal consciousness. If, on the other hand, the chapter is more concerned with
faculties of perception, it may allow the mere possibility of perception’s being
perceived; and then we may need to account otherwise for this argument’s purpose.43
The suggestion that Aristotle is here concerned with actual individual perceptions
sits uneasily with the place of this argument in de anima 3—immediately after a
discussion about how many sense-faculties we have, and before an extended, and
parallel, account of the faculties of imagination and thought.44 Further, both modi-
fications of the puzzle in the immediate sequel turn on cases where the first-order
perception is somehow missing (judging that it is dark, and retaining sense-impres-
sions after their object has gone). These cases are taken to show that ‘perception is
not one’. But that cannot then mean that individual actual perceptions are complex
in the required sense, since in the cases in question that complexity is expressly
missing. Instead, it implies that the faculty of perception in question is multiform:
that is, we may call various different things ‘perception’, including first-order and
second-order perceptual events. The generalization that is made here, therefore, is
about perception and its nature, not about all actual perceptions.
This, in turn, may incline us to read the dominant argument of the passage as
attending to the nature of the faculties of perception, multiform though they may
turn out to be. But then how does the regress work? There will be, of course,
individual actualizations of the faculties of perception; and of each of these we can
say not that it is perceived, but that it is perceivable (at least this, after all, is warranted
by the premise ‘since we perceive that we see and hear’, however it is parsed). Indeed,
that is a natural point to make in the context of the view, explicitly recalled in this
passage (425b22), that perception is somehow an affection by the perceived as such
(supported by the thought that first-order perceptions are either transparent, or
transitive, to higher-order ones: see 426a2). For perception comes about by the
affection of the perceiver by the object; however queer that affection may be, it is
still a change of some kind in the perceiver. As such, it is quite reasonable to suppose
that, for example, what is ‘coloured in some way’ is perceivable in some way (just as
what is coloured is perceivable: Aristotle certainly does not maintain that everything
coloured is perceived). That is, the very affection that is perception is perceivable; and
if so, there must be a faculty that perceives it. But that faculty, whatever its sense-
42
Caston (2002), (2004).
43
See Johansen (2005), 245 ff., on how we should construe the claim that ‘if some sight is going to see the
seer, the first seer will have colour’ at 425b19, reading horôn twice; and, differently, Caston (2004), 524 ff.,
who compares the Charmides, on which more below.
44
The datives at 425b13 and 20 to describe perceiving ‘by sight’ are most naturally read as instrumental
or explanatory: and hence as referring to the faculty, not the act. Compare the point made at 426a13 that
seeing is the activity of sight (opsis).
modality, will be affected by its object when it perceives, and so be perceptible in its
turn.45 If this chain of perceptible things is not to stretch to infinity—if, that is to say,
it is to be explicable by positing faculties of perception—then some faculty of
perception must be able to perceive itself.
So the regress may work, I argue, on either construal of ‘since we perceive that we
see and hear’: whether always or sometimes. Either way, the regress needs to be
blocked by a reflexive: some perception perceives itself. That may be just as well,
perhaps, to sustain the thrust of the opening chapter of the book, that there are only
five senses. But it is problematic for Aristotle, severely exercised as he is by reflexivity.
If an action is reflexive, and strictly so, it seems to be cause and caused at once—and
that makes a nonsense of causation.46 If a faculty is reflexive, it both explains its
actualizations and itself is one of its actualizations (and so needs explaining). Yet
Aristotle posits sense-faculties to explain our ability to perceive; if those faculties are
somehow reflexive, are they both explanantia and explananda at once?47
If we are to avoid the risk of regress, some account must be given of self-perception
that is not so strictly reflexive as to be impossible. And yet the argument as Aristotle
presents it may seem to emphasize that strictness. For there is, at first at least, nothing
about the self here; and nothing, therefore, about the subject of the perceiving as such
or about what makes (any of ) these perceptions mine.48 Instead, the argument is
resolutely impersonal, marshalled around the two relata in perception: the object of
perception (strategy 1) and the perception itself (strategy 2). Is this self-perception
what we might call consciousness?
4. A Puzzle
Whatever it is, Aristotle now confronts it with a puzzle—but a puzzle too condensed
for clarity. Aristotle provides responses to it that consider both reduction (at lines
20–2) and transparency (at 22–5): but the point of the responses remains disputed.
We might be forgiven for complaining about gappy lecture notes, once again.
The puzzle runs like this:
If perceiving by sight is seeing, and what is seen is colour or what has colour, then if some sight
is going to see the seer, the first seer will have colour.
45
The interpretation I suggest, then, is a hybrid: it worries both about a regress of faculties and about
the nature of individual events of perception. Quite right too—for Aristotle talks of faculties to explain the
events.
46
Compare de anima 417a2 ff., or the worries about self-motion at Physics VII. 5.
47
Johansen is surely right to compare Third Man Arguments of the Parmenides (2005), 244 ff. and n.
16, which support two different regresses—one where the higher-order perception is the same sense-
modality, the other where it is different.
48
The argument, notably, does not rest on any claims about the first-person features of this self-
perception: the two personal verbs (at 425b12 and 425b21) seem to introduce, first, the phenomena to be
explained, where the explanation notably lacks any indexing to persons; and, second, some additional data
about our responses to light and dark.
Is the point that it is manifestly silly to think the seer is coloured? The puzzle would
then be a parasite of transparency. Is it a satisfactory riposte to say, merely, that ‘that
which sees is coloured in a way’—as Aristotle does four lines later? Or is the puzzle
focused, as the preceding argument suggests, on self-perception and the problem of
reflexivity? It might, thus, turn on the tension between reducing the perceptions we
have here and making some sense of the paradoxes of the reflexive (how could
something act on itself, without qualification?). But how reasonable a response to that
is what Aristotle says next, that ‘this makes it clear that perceiving by sight is not
one’? And how, anyway, does the puzzle fit into the economy of the chapter as a
whole, or into its place in the de anima? How do the puzzle and the responses to it
modify what Aristotle has said about self-perception?
One might think that both responses—that ‘perceiving by sight is not one’ and that
‘that which sees is coloured in a way’—rather fudge the issue. If the puzzle is how
self-perception can occur, Aristotle’s solutions might seem to turn somehow on
denying that this is self-perception in the strictest sense, and allowing merely that
there are some loose ways in which perception perceives itself, which are not
vulnerable to the puzzle. Thus, if perception by sight is ‘not one’, then perhaps
there are two quite different sorts of perception involved in our perceiving (sort 1)
that we see (sort 2).49 The result, then, would be a quasi-self-perception (sort 1
perceives sort 2) fit to meet the reducing assumption only by sleight of word. Or if it
is somehow objectionable that the primary seer is coloured, again in some strict way,
the transparency assumption could still be met by conceding that the seer is coloured
in some way that is other than strict. I shall suggest that the qualifications Aristotle
enters in answer to the puzzle are less disappointing.
5. The History of the Puzzle

Aristotle’s puzzle, however, is not original to him.50 In Plato’s Charmides (167–9),
Socrates offers a tricky argument that there can be no perception of perception, and
thence no perception that is ‘of itself ’ (167d; 168d–e).51 From this he infers the
implausibility of Critias’ claim that the virtue of sôphrosunê is in fact self-knowledge:
and of his further claim that self-knowledge is reflexive: knowledge of knowledge,
knowledge of itself.52
49
Perhaps sort 1 sees the perceiver as perceiving, but sort 2 sees its object by virtue of a sheer affection of
the sense-organ. On this see e.g. Caston (1998), 280.
50
This is regularly observed, e.g. Sorabji (1979), 49, n. 23; Caston (2002), (2004), 524; Sisko (2004);
Johansen (2005), 248. My contention is that Aristotle is engaged not just with the short argument of
Charm. 167b–169a, but also with the projects of the dialogue as a whole.
51
The argument as a whole is about knowledge, to which perception comes in as an analogue. It began,
at 165c, by discussing knowledge of oneself; then ostentatiously shifted to a discussion of knowledge of itself
(166c3). See Chapter 9.
52
The marked impersonal formulae in the Socratic argument are mirrored in Aristotle’s response.
But this is rather a queer thing for Socrates to do. After all, he himself has a high
stake in some sort of knowledge of knowledge, or at least in higher-order cognitive
attitudes about knowledge. How else could he claim that he is wiser than others in
knowing that he does not know? How else could he explain the reflective stance
provided by the Socratic elenchus, the investigation of what people claim to know
(expressly brought to mind at 166d)? His argument’s trajectory needs to be redrawn:
especially, it needs to take account of the connections between knowledge and virtue.
The problem (one of the problems) here is how both knowledge and perception
are construed in terms of a direct causal relation between subject and object: a
relation that is both asymmetrical and intransitive.53 If it is asymmetrical, then
there can be no reflexive relation between subject and object; and if it is intransitive
(and still directly causal), there can be little to be made of the content of a higher-
order attitude to a lower-order state of mind. So perhaps Plato’s cunning plan in the
Charmides is to expose this impoverished account of the subject-object relation, for
knowledge at least; and to suggest that to have both higher-order and reflective
features, it must be understood as a far richer cognitive state. Knowledge, then,
needs to be something like understanding, or wisdom; it needs to be reflective and
broad based. Its reflexivity, then, will be modified and explained as reflection, where
knowledge is not of itself, stricto sensu, but rather higher-order knowledge has lower-
order knowledge within its content. This suggestion affects knowledge’s analogue,
perception, too: either perception needs to be understood in a similarly rich way (as
highly cognitive, reflective, unlike a raw54 relation between subject and object); or it
ceases to be an analogue for knowledge.
There is a repeated use of perception as a reflective attitude in the Charmides,
which suggests that it is richer than raw. But these suggestions appear in the frame
dialogue: in the discussions between Socrates, Charmides, and Critias about how the
discussion should proceed.55 By contrast, the direct argument, discussing both
hearing and sight, comes twice to a halt. On the first occasion, Critias agrees with
Socrates that there is no sight:
a: . . . which is not a sight of the things other sights are sights of, but is sight of itself
and the other sights and likewise of the non-sights, and which although it is a
sight, it sees no colour, but sees itself and the other sights. (167c8–d2)
On the second occasion, they agree that:
53
Asymmetrical—if x perceives y, then x is the perceiver, y the perceived, and as such the relation is not
reversible; intransitive—when x is the perceiver and y the perceived that exhausts the perception relation,
and x cannot see through y to z. Socrates’ argument is designed to show that if perception is like this, it
cannot illuminate knowledge. Aristotle, on the account I give below, goes further—if perception is like this,
it cannot account properly for the richness and complexity of perception.
54
‘Raw’ may be about the feel of it (I smell the buttered toast in some irreducibly subjective way) or
about its causal structure (the buttered toast somehow impinges on my sensation directly). The subject-
object relation in which Socrates is interested involves the second kind of rawness.
55
Notably at 159a, 167c, 168e.
b: . . . if it is going to see itself, [sight] must have some colour; for something
colourless, sight could never see. (168d9–11)
Critias and Socrates, then, find it hard to fit their search for reflexivity (manifest in A)
to their conviction that there must be some kind of transparency condition (if sight
sees sight seeing, it sees also the colour it sees). As a consequence, either perception is
not a proper analogue for knowledge, or it must be re-described. But nothing here
suggests that higher-order perception—if it can be made to make sense at all—is
intrinsic to, or necessary for, lower-order perception. So if we perceive that we
perceive, on this account, this is not phenomenal consciousness. On the contrary,
if perception can be re-described to account for the perception of perception, it
should provide an analogue for wisdom—for that hard-won virtue that is the goal of
philosophy. To perceive that we perceive may, therefore, be a hard task, not the
natural occurrence of consciousness.
Aristotle offers a direct allusion to the Charmides.56 Compare, to A:
a* But the same sense will be of the sight and of its underlying colour
(425b13–14);
and to B, the last two clauses of the puzzle:
b* . . . then if some sight57 is going to see the seer, the first seer will have colour
(425b17–20).58
Like Socrates and Critias, Aristotle denies that there can be any seeing that is not
seeing a colour. Unlike Socrates, who purports to solve the dilemma at this stage of
the dialogue by denying that there can be any perception of perception, any self-
perception,59 Aristotle characteristically goes with the phenomena: we do perceive
that we see colours and hear sounds. He infers from this that there is self-perception;
and from this that perceiving by sight is ‘not one’.60
56
Notice that Aristotle and Plato both start with the same two example senses—sight and hearing.
57
This construal, if there is an allusion to the Charmides going on here, is explained by opsis tis at
Charmides 167c8.
58
That A* is an allusion explains several things: the thoroughly condensed form of Aristotle’s remark;
the difficulty in identifying its subject (does this refer to both of the options canvassed in the previous
sentence, or just one?); the emphatic ‘the same’ at 425b13, where the Platonic background shows that
denying transitivity is absurd. In B*, the quotation is very close: notably, the verb forms (‘will see’, opsetai);
the word order of the first clauses; in the second clauses, the emphatic position of ‘colour’ (chrôma) at
Charmides 168d10 and de an. 425b19; and the shift from Plato’s ‘it must have’ to Aristotle’s ‘it will have’.
By ‘allusion’ I mean that Aristotle intends his reader (some of his readers) to notice the connection, and
that the dialectical relation with Plato is important to the development of Aristotle’s point. Surely (some
of ) Aristotle’s readers were, like him, careful readers of Plato (see Halliwell (2006))?
59
This solution, however, may be what Socrates needs to avoid, in the outcome of the dialogue as a
whole; he must at least be able to give some account of higher-order knowledge.
60
So the Charmides is in the background of Strategy 1 of the argument: Aristotle assumes what Critias
denies, that higher-order sight is transitive.
Aristotle’s puzzle is, I suggest, a reading of the Charmides’ argument, a reading of it

that takes its conclusions not to be compelling, but problematic (just, I would say, as
Plato set it up). I shall suggest that we have good reasons to ask what else Aristotle
shares with the Charmides: in particular, the question whether perception is, or is
not, analogous to knowledge, in circumstances where what is required of knowledge
is that it be reflective. For—to repeat—what is not at issue, in the argument Aristotle
inherits, is phenomenal consciousness.61 If that were what it is to perceive that we
perceive, it would not even begin to illuminate what Socrates wants to say about
knowledge, about reflection, and about philosophical inquiry; nor would it bear any
comparison with the context in which Socrates raises these issues—the discussion of
virtue.
Someone might complain that to suppose this makes a difference to how we read
Aristotle is to suppose that Aristotle reads Plato carefully and, as it were, dialogue by
dialogue. On the contrary—the objection would run—Aristotle takes Plato on as a
series of doctrines (about forms, about the good, and about perception) with which
he engages, and which he rejects.62 So Aristotle could not care less about the overall
structure of the aporia in the Charmides, just so long as its elements can be exploited
in puzzles of his own. His engagement with the (many and the) wise is satisfied just
by having in mind some version of some view that Plato might have held; neither
the historical nor the textual accuracy of the views in question matters a bit.63 So
Aristotle’s allusion to the Charmides would not show that he shares that dialogue’s
background assumptions or its dialectical structure: the form of words would tell us
nothing about the shape and structure of Aristotle’s own argument.
This objection seems to me to make a broad assumption both about how the
dialogues are written and about how they were read in the Academy within Plato’s
lifetime, which is unjustified. Plato himself does not give us grounds to make any
such assumption, or to suppose that some bits of the dialogues are dispensable, others
not. If his silence on this matter was—as I see no reason to doubt—deliberate, then
Aristotle may not have been brought up to think the dialogues merely vehicles for
doctrine, whose elaborate dialectical and dialogical structure is of no philosophical
interest or importance at all. Quite the contrary: if Aristotle was taught by Plato how
to read a Platonic dialogue, he was surely better at reading it than we are. And if the
detail of the dialogue matters to the argument, we had better not suppose in advance
that Aristotle did not see that, or exploit it to his own advantage. So we cannot
assume that Aristotle read the Charmides but ignored its reflections on the reflective
character of knowledge when he comes to his own discussion of self-perception. No
61
Contra Kosman (1975), 517.
62
This is sometimes manifestly the case: see e.g. Fine (1993); but my claim is that direct arguments
against Plato do not exhaust Aristotle’s interactions with the dialogues.
63
See the account of the starting points for dialectic in Metaphysics B1 and Topics 1.1, 10, 11.
more can we assume that allusions thus detailed to Platonic originals would be lost
on his audience.
But now this allows to Aristotle the choice that Socrates puts before us. If
perception is analogous to knowledge,64 and if it is conceived as a strict (asymmet-
rical, intransitive) relation, then there can be no reflexive perception (no perception
of itself ) and no reflexive knowledge. But knowledge (or, better, understanding)
needs to be if not reflexive in this strict sense, then reflective. For understanding needs
to have a higher-order reflective dimension, where the first-order knowledge is its
intentional object (if I know that I know, then part of the content of my higher-order
knowledge is the content of the lower-order one; another part of its content is that
I know that lower-order content). So either perception is not analogous to know-
ledge, or perception is not the strict relation conceived by Socrates’ argument. The
Charmides read as a whole, I claim, suggests the latter; although it is easily read as
proffering the former. Moreover Aristotle, in his readings of the Charmides, takes as
his starting point the latter—that perception is not a strict, exclusively first-order,
relation, but is capable of higher-order dimensions. This, I suggest, is what he means
by starting: ‘Since we perceive that we see and hear . . .’ At the start of the chapter he is
already dialectically engaged with the Charmides: and it is from this engagement that
he takes himself to be entitled to conclude that ‘perceiving by sight is not one’. I shall
suggest that his engagement with the Charmides runs deep—to the ethical base on
which Socrates’ argument is set.
6. Aristotle’s Responses to the Puzzle

Aristotle resists the aporia to which Socrates and Critias seem to be reduced. He does
so by supposing that self-perception is proof against the apparent puzzle, that ‘if
some sight is going to see the seer, the first seer will have colour’. His two responses to
the puzzle, therefore, should amplify just what he takes self-perception to be.
Well, then, it is clear that perceiving by sight is not one.65 After all, when we do not see, it is by
sight that we judge [krinomen]66 both the darkness and the light, but not in the same way.
(425b20–2)
And again, that which sees is coloured in a way; for each sense-organ is receptive of what is
perceptible without the matter. (425b22–4)
The two responses seem pretty haphazard, at first—yet more evidence for gaps in
Aristotle’s notes: how exactly does the second follow on from the first? And how does
the first respond to the problem of self-perception?
64
For a subtle account of how this works in the peculiar 3.7, see Osborne (1998).
65
See n. 6: this is not about linguistic issues (ambiguity), but about the real structure of things. Notice
the later claim, 426a15 ff., that although the actuality of the perceived and the perceiver is one, their being is
not: by the next stage of the chapter Aristotle is talking about the real complexity of perceptual events
(compare de sensu 449a5–20), even if that complexity gives rise to linguistic error (426a26).
66
This expression appears also in the parallel passage from the de somno: see n. 11.
First, the first: to avoid the aporia, we need to say that ‘perceiving by sight is not
one’. This is confirmed by the point about what happens when we don’t see (because
it is dark) but nevertheless judge that it is dark. This, like the perceiving of light, is
done by sight, even though it is not seeing in the way that we see colour. How does
Aristotle’s point work?
Consider the contrast between cases when we are blindfolded, so that it seems
dark, and cases when we are able to see, but see nothing, so that we judge that it is
dark. In the second case, unlike the first, the judgement is made—or so Aristotle
suggests—by virtue of the sense-modality of sight: we survey, as it were, the nothing
that is before us, and judge ‘Cripes, it’s pitch-black’. The same happens for light,
Aristotle suggests (light, after all, is not a special object of sight): we are seeing, and
there are things that we see,67 so that we judge that it is light—we say ‘Aargh, that’s
bright!’. In both cases (darkness and light) we are said to ‘perceive by sight’, even
though in neither case do we do so exactly by perceiving sight’s special objects. The
expression ‘perceiving by sight’, therefore, includes this kind of judgement in its
scope. And because it does so, perceiving by sight is not narrowly restricted to the
direct perception of sight’s special objects; instead, we might say that the general
category of things that come under the description ‘perception by sight’ is ‘not one’.68
Aristotle’s point, I have argued, should be not that individual perceptions are
complex, but that perception, the faculty, is multiform: but what exactly does that
mean? And how are these observations about perceptual judgement à propos?69
Maybe Aristotle distinguishes between sight proper (first-order seeing of the special
objects of sight: my seeing this fuchsia pink, for example) and sight in some
derivative sense—sight improper, where sensible judgements take the place of direct
vision. On such an account, he only partly assimilates perceptual judgement to
perception, allowing that the former can be called perception, but only in some
etiolated way. This might be the point of his saying that perception is ‘not one’: he
means us to understand that the expression ‘perception’ can refer to both perception
proper, and perception improper, but not in the same way. ‘Perception by sight’ on
that account is ambiguous, it is ‘not one’ (and self-perception is then only of the
‘quasi’ sort).70
67
This elliptical sentence should, I think, read something like this: ‘when we are not seeing, it is by sight
that we judge the darkness; in just the same way, we judge light (when we are seeing) by sight, but in neither
case do we perceive in the same way as we perceive colours.’ We see both darkness and light in the same
judgemental way (but Johansen (2005), 250, n. 27).
68
The de somno (n. 11) limits the point: perception by sight is not an ungoverned plurality, but held
together by the ‘common faculty’ of perception. See Kosman (1975), 518, on understanding perception in
terms of the whole organism.
69
This question, in my view, tells against Caston’s view (à la Brentano) that the chapter shows that
conscious perception is a complex of the perceived object and our reflexive consciousness of it: in the case
of darkness the judgement occurs in the absence of an object, albeit the presence of our ability to see.
Instead the comparison with light and darkness amplifies the different and independent ways in which we
may be said to ‘perceive by sight’ and decouples perceiving that we perceive from first-order perceiving.
70
See }4.
A passage from the previous book of the de anima might give us pause, however:
Now sight is both of the visible and of the invisible (for darkness cannot be seen, but sight
judges that too), and again of what is too bright (this too cannot be seen, but in a different way
from darkness). (de anima 422a20–3)
Here Aristotle imagines a case where we can see by virtue of the fact that our faculties
are working perfectly well; but we see nothing—so we judge that it is dark. He follows
that up with a different point about light from the one he makes in 3.2, suggesting
that excessive brightness is destructive of the faculty’s ability to see. The passage
turns, then, on the contrast between cases when sight works, and cases where it does
not; in the former, we should say even-handedly that sight is ‘both of the visible and
the invisible’: seeing the invisible—judging that it is dark—seems to count as sight in
a parallel way to seeing the visible,71 and to be properly contrasted with cases where
the sense-faculty is not working at all. The first clause of this sentence, therefore, does
not suggest that sight’s operation in some cases of what cannot be seen is somehow
etiolated, or metaphorical, or that the cases where sight manifestly judges are not
proper cases of sight at all. Nor, indeed, would such a solution do justice to Aristotle’s
earlier argument in 3.2, which insists that there is self-perception. For if this is the
explanation of that phenomenon, it is only self-perception by virtue of an ambiguity.
If ‘perception’ is ambiguous between the proper and the improper sorts, and both
sorts are in play in perceiving that we perceive, we need to avoid the sophistical
difficulties, and say that this is a case where I perceive-improperly that I perceive-
properly; and that is not self-perception at all (‘quasi-’ won’t help).
Instead, perhaps, we should read the suggestion that ‘perception by sight is not
one’ rather differently: as a warning that perception works in several different ways—
a warning that Socrates and Critias would do well to heed. But if this is to avoid the
complaint that it will only explain quasi-self-perception, Aristotle needs to show at
the same time that perception is somehow unified.
Suppose he means the comparison with the perception of dark and light to suggest
that perception generally involves judgement (as it obviously does in the case of dark
and light). Suppose, further, that he intends not to draw a contrast between percep-
tion-proper and perception-improper, but to show us that what holds of the most
complicated cases holds also of the cases we might think simple, or perception-
proper. I see this fuchsia pink before me. This might be quite basic, a seeing of fuchsia
pink. But how does it relate to cases when I see that this is fuchsia pink; or when I see
that this fuchsia pink is distinct from that burnt umber; or, in darkness, where I see
no fuchsia pink at all? Perhaps—by virtue of Aristotle’s argument that the last case is
indeed a perception by sight—we might come to agree that all of these cases are
perception by sight; and that although they differ in various ways (for example, in the
case of darkness there is no first-order seeing going on; and in the case of perceiving
71
This, surely, is the effect of ‘sight judges that too’, krinei de kai touto hê opsis, 422a21.
that I see, there is also second-order perceiving going on), they all count as genuine
examples of perceiving by sight.72 Then, they are all indisputably perceptions proper
(‘perception by sight’ is not equivocated) even though they differ in important ways.
Now, so far from this separating off the basic case (I see fuchsia pink) from the rest,
including the higher-order ones, the basic case turns out to have after all the essential
features of the higher-order ones: they are all to be understood in terms of perceptual
judgement, even if they vary widely in the details of their content.
Perception, when we perceive that we see, has a complex object, as Aristotle’s
argument brings out (both the first-order object and the first-order subject: I perceive
that I see fuchsia pink). But then perhaps the basic case works in a parallel way.
When I see fuchsia pink, I see that it is fuchsia pink: the basic case can look as much
like a judgement as the case where I contrast something fuchsia with something
burnt umber. Instead, that is, of the perceptions that are clearly judgements (com-
parisons, for example, or higher-order perceptions) being perceptions-improper, we
might say that they are all both proper and complex—and eschew the thought that
there is something proper only about cases of the apparently basic kind. In that case,
on the account that Aristotle offers here, we should count as ‘perception by sight’ all
sorts of things, including the perception that I see; and we should do so by virtue of
the fact that perception has judgemental content from the lowest level up.73
This, however, makes perception multiform, since types of perception can differ
both in their content and in their order. Aristotle’s amplification of ‘perception by
sight is not one’ makes it clear just how.
First, he compares (by the train of thought that runs from 425b19–20 to 425b21–2)
the business of higher-order perception with the judgement involved in cases like
seeing darkness. Both count as perceiving by sight; but they differ in order. He also
compares seeing darkness with seeing a colour. Both count as perceiving by sight, but
they differ in content because their objects are formally distinct. And yet all three
(perceiving that I see, seeing darkness, seeing a colour) count somehow as perceiving
by sight. If they are to be comparable at all, then first-order sight of colour needs to
be, I have suggested, cognitive in content in something of the same way as seeing
darkness and as perceiving that I see. But if they are distinct, as Aristotle also urges,
there is a formality to their distinctness—they may be different in order; or different
in their relations to the special objects of the sense-modality in question—and still
count as perception by sight. I will see colour under different conditions than
I perceive darkness by sight; and I may see colour without perceiving that I do so:
‘perception by sight is not one’.
72
They may also have a phenomenology, awareness (see Burnyeat (1992)). This does not imply that
perceptual awareness is what is described by ‘perceiving that we see and hear’.
73
If perception in humans is continuous with that of animals, are we then to say that animals judge? Or
are we to allow that perception changes its content as it goes up the scala naturae? On this, see e.g. Burnyeat
(1992); Sorabji (1993), part I.
Second, if perception by sight is a case of being affected by the object of perception,

but this allows for cognitive content too, then we shall need to modify the sense in
which we may say that the eye is coloured.74 The basic example needs to be made
more sophisticated, and to allow that its content is cognitive (hence the modifications
suggested at 425b23–5) while still retaining the central thesis, that sight in the first
place occurs when the eye is affected by colour. This may be the point of Aristotle’s
modification: ‘that which sees is coloured in a way.’
If this account of perception is the outcome of Aristotle’s argument about per-
ceiving that we see, it fits well into the chapter as a whole, which brings out two
aspects of perception thus understood. On the one hand, it occurs when the perceiver
and the perceived are actually unified (425b26–426a26): the realist cast of Aristotle’s
account of perception is not damaged by insisting on perception’s cognitive features.
On the other hand, what it is to perceive at least includes cases that are readily
admitted to be perceptual judgements (for example, seeing that it is dark, 425b21;
perceiving that something is both sweet and white without assimilating the two
features, 426b12–21). It is for this reason that a large part of the rest of the chapter
is concerned with understanding just how a perceptual judgement can be both
complex and unified75—the dual condition on its being as much perception as the
first-order, basic case, which is a single event at a single time.76 Thus the perception
of a concord or harmony is a single ratio (426a27–b7); the judgement of two sensibles
as different must be done by a single faculty (426b20–1); and judgement may
sometimes use the same element twice (as the point may appear twice in our
definition of a line: 427a9–14). If perception is cognitive, that is to say, it is unified
in judgement, not merely by the physical unity of the object with the subject, but also
by the conceptual, cognitive unity of the perception itself.
Where, now, does that leave my original question about whether this chapter is
about phenomenal consciousness? The burden of Aristotle’s argument here is not to
show the way in which the subject is aware, when it perceives, that it perceives.
Instead, he emphasizes the features of perception that will distance it from phenom-
enal consciousness: its judgemental, cognitive content. He also implies—in insisting
that perception by sight is not one—that we should not expect that the different
conditions of perception he describes in this chapter—perceiving that we see; judging
difference—all occur in every instance of perception, or even just normally. Instead,
74
Hence, ‘that which sees is coloured in a way’, 425b22. Does this imply a literalist account of
perception? No—all we need, both for Aristotle’s puzzles to get underway and for his solution to be
consistent with what he says elsewhere, is for perception to be sufficiently the same, whatever its level (so
not raw at the first level, cooked at all the others).
75
The unity and complexity of perception are reiterated in the later stages of the chapter: the discussion
of the perception of harmony (426b29–427a6: see Barker (1981)); the account of judgement done by one
sense (426b8–14) and of judgement across senses (426b14–29); the puzzle about how this can be explained
in terms of opposite motions (426b29–427a6); and the resolution in terms of actuality and potentiality
(426a6–16).
76
See 426b28 ff. on the impossibility of some perceptual event’s being two opposite movements.
perceiving that we see, the case that drives the argument against the singularity of
perceiving by sight, should not be understood as a regular concomitant of first-order
perception: it is not phenomenal consciousness. What, then, is going on here?
7. Some More Plato

The question of the relation between perception and judgement is not original to
Aristotle: it comes, instead, from the other Platonic dialogue in the background of
these chapters of the de anima, the Theaetetus.77 There Socrates examines a theory of
perception (a theory associated with Protagoras, to whom I shall return) which is a
raw relation between object and subject; and suggests that it cannot explain all of our
attitudes or mental states, such as our judgement of the common terms (such as
‘same’ and ‘different’ or ‘good’ and ‘bad’).
So . . . In respect of being, and that they [sc. perceptible things] are, and their opposition to each
other, and, again, the being of this opposition, the soul itself, rising up78 and comparing them,
attempts to make a judgement (krinein) for us. (Theaetetus 186b)
Socrates suggests that if perception is construed as a strict relation between object

and subject, judgement will be done by the soul. The judgement itself is comparative,
a weighing up of the relations between perceptible things, and reasoning about those
relations themselves. When the soul makes a judgement, however complex—as a
famous later passage in the Theaetetus shows—it is involved in a ‘silent dialogue’. It
scrutinizes opposing views put to it (by reason, or by perception79), listens to each
side, and itself comes to a decision: this is a belief (Theaetetus 190a) or a judgement
(Philebus 38c).80
This account of the soul’s activity has a striking feature: its detachment. As the
dialogue goes on in the soul, the soul itself stands aloof from the opposed views; only
after consideration (Theaetetus 190a2–3) does it come to its judgement. So while the
perception delivers reports, the soul considers them as if from the outside;81 this
77
Verbal echoes include the introduction of Plato’s (term of art?) koina (from 425a6: compare Theaet.
185b8); both philosophers’ interest in sight and hearing as their example senses (compare Theaet. 184b10
ff.); and the shift from sight to the question of perceiving by taste, e.g. something salty, Theaet. 185b10, de
anima 426b5. There are structural similarities, too: de an. 2.12 and the discussion of phantasia in 3.3.
tackles the theory of perception; 3.1 tackles the unity of the soul and the common terms; and the theme of
falsehood, which becomes more and more important in the early chapters of de anima 3, has the same
emerging significance in the Theaetetus. This background is frequently observed; it often explains the
lacunosity of these difficult Aristotelian chapters.
78
Compare the importance of the soul’s view of things outside the cave at Republic 516b.
79
Compare Philebus 38c ff.; Republic 523–5; Sophist 263e.
80
The ‘silent dialogue’ is designed, surely, to show the structure of the central case, even if some
judgements may be far more exiguous.
81
Notice the odd echoes in the two models of the mind offered later in the Theaetetus: the wax tablet
(191c ff.), which has to be held up and manipulated by someone other than it (notice 191d4–7); and the
aviary (197c ff.), outside which there is the birdcatcher, in whose head most of the business of mistaking
must occur.
difference in stance separates perception from judgement. According to the Theae-

tetus, then, if any judging goes on, it is not perception that does it: perception would
be a simple phenomenon, and self-perception would be ruled out.82 The soul’s
activity, contrariwise, is something that philosophy promotes: there is a direct
connection, that is to say, between the reflective stance of the soul when it judges,
and the progress of the philosopher towards god (Theaetetus 176b).
Aristotle dissents, up to a point. For him, perception is more than a simple raw
relation, since it includes judgement, in various different ways. But it is not a mere
collection of different phenomena, either, since it includes self-perception in some
proper sense. This further suggests that the business of perceptual reflection is a
continuum of cognitive activities, based on first-order perception of the special
objects of a sense, but capable of greater and less reflective capacity. So in some
measure the argument revises the reducing assumption with which it began by
enlarging the scope of ‘perception by sight’. But if—in making this claim—Aristotle
picks up not only the terminology of the Theaetetus, but its train of thought as well,
then the judgement in question, now claimed to be part of perception, should exhibit
the properties Socrates there allocated to the soul. In particular, the perception that is
self-perception should have the quality of detachment Socrates wants for the soul: it
is by this very detachment that reflectiveness (soul’s judgement for the Theaetetus,
perception’s judgement for the de anima) is ensured.
Where then does that leave the second response to the puzzle: that ‘that which sees
is coloured in a way’? For this, in its turn, invites the regress to run: is Aristotle’s
account of self-perception sufficient to block it? The arguments for self-perception
imagine that the first-order object (the colour, for sight) is perceptible both by the
first-order and the second-order perception; the higher-order perception by sight
sees through to the special object itself. But if the higher-order perception is reflective,
this transparency may be modified; perhaps instead, the higher-order perception is
thought of as somehow looking at the first-order perception, because that is ‘coloured
in a way’. This ‘looking at’ might fit the thought that what the higher-order
perception does is judging: for judging, like looking at something, supposes that the
judge, or the looker, is somehow detached from what it sees.83 Still, on this account, it
is perception; and can itself be the object of reflection. Reflection, however, is broad
enough to include itself. Judgements, after all, can be self-referential without doing
violence to the thought that nothing can be cause and caused, or explanation and
explained, at once. If self-perception is a judgement, the regress can be blocked.
But the two responses to the puzzle are now seen to be connected, in amplifying
the perceiving aspect (rather than the object aspect) of higher-order perception:
higher-order perception considers, looks at, inspects, the lower order and its content.
82
Much of the language for the soul’s detachment is perceptual: looking at, surveying, inspecting, etc.
This should make one hesitate before committing Plato to a raw theory of perception.
83
Thus 425b23 talks about the sense-organs, where what is looked at (somehow or other) resides.
Higher-order perception is detached. But it is still perception: far from the raw
conception of perception offered both by Critias in the Charmides and by Socrates
in the Theaetetus, Aristotle supposes that these perceptual judgements are themselves
still part of the business of perception. It is a rich account, one which, by building
detachment in, makes perception thoroughly cognitive.
8. Detachment and Teleology

It does more. Compare another passage where Aristotle speaks of perceiving that we
perceive:84
. . . and if the one who sees perceives85 that he sees, the one who hears perceives that he hears,
the one who walks perceives that he walks, and similarly in the other cases there is something
that perceives that we are in activity, so that if we perceive it perceives that we perceive, and if
we think it perceives that we think; and if perceiving that we perceive or think is perceiving that
we exist (for as we said, existing is perceiving or thinking); and if perceiving that one is alive is
pleasant in itself . . . (Ethica Nicomachea 1170a29–b1, trs. Rowe)86
This is part of Aristotle’s argument to show that the self-sufficient person (the person
of supreme virtue) still needs friends: not, for sure, friends for instrumental reasons
(since the self-sufficient person has all she needs), but friends who contribute to her
sense of living the best life, and who thus enhance the pleasure she has in the life she
lives. The full life she lives is determined by perception and thought (they are the
actualizations of her nature), and so it is good and pleasant in itself, an object of
everyone’s desire. In that case, the self-sufficient person gets pleasure from perceiving
that she perceives, or that she walks, or that she lives; and she enhances that
perception by joining with her friend in its contemplation.
This argument rests its account of the self-sufficient person’s need for friends on a
thesis about what living the best life is: the full actualization of our capacities. But this
full actualization—as Aristotle argues both here and in the parallel passage in the
Eudemian Ethics—is something we enjoy contemplating when we have it; it is
something, therefore, that we desire to contemplate—on our own, or with our
friends. Perception, as it is here construed, both marks the full actualization of our
capacities (it is not merely passive, or a raw event impinging on our sense-organs)
and reflects on that full actualization: that is what happens when we perceive that we
perceive (we perceive that we perceive, that we exist, and that this life is a good one).
But as such, it is an object of desire and aspiration, something that involves the self-
sufficient person in a wide spectrum of ethical activity (including the cultivation of
84
This passage is regularly cited in the discussion of de an. 3.2, but its context underestimated. Compare
Eudemian Ethics, 1244b25–34, which I discuss in more detail in Chapter 16.
85
Notice here that the present tense has the same variable aspect as noted above, n. 11.
86
See Johansen (2005) 264, n. 58, on the text.
friends, and the avoidance of certain distractions). So perception is a manifestation of

human excellence, and perceiving it a source of pleasure. And in that case, the
perception of perception simply cannot be the consciousness that necessarily follows,
or is intrinsic to, our first-order perception of special sensibles. Perceiving that we
perceive, on the contrary, is a normative ideal, something we endeavour to do,
something we practise, habituate ourselves to, and take steps to achieve. But if
perception that we perceive is thus teleological, it cannot be mere phenomenal
consciousness.
‘Since we perceive that we see and hear . . .’ in de anima 3.2 does not, I conclude,
describe phenomenal consciousness, either. For here too the discussion is enmeshed
in a larger context, in which Aristotle lays out an account of natural function that
underpins his ethics. If there is indeed continuity between what he says in the ethical
works and his account of the nature of the soul, we should take the de anima account
of self-perception to have similar normative features. Self-perception is not a neces-
sary or a regular concomitant of first-order perceptions (this is one consequence of
the argument that perception by sight is not one). It is not, either, a mark of what
things are like for me, of the subjective side of perception, or of its specialness to the
perceiver herself. For the argument turns not on questions about how I feel about my
perceptions, or yet about how my perceptions are ineradicably mine not yours, but
on the reflective relation described by ‘self-perception’. In thus elaborating my
perceptual judgements, the argument fails, that is to say, the tests for phenomenal
consciousness I set for it at the outset. Instead, for Aristotle, higher-order perception
is reflective and detached: looking at, and adjudicating, what the first-order presents
to it, as if from the outside.
9. The First Person?

In this Aristotle differs sharply from a lurking shadow in this text, Protagoras, once
again from the Theaetetus.87 Protagoras’ relativist account of perception, of truth and
of knowledge, one might be forgiven for supposing, is of nothing but consciousness.
Protagoras—as Plato represents him in the Theaetetus—starts with perception, and
insists that my perceptions are incorrigible by you; then extends this claim about
perception to all appearances; and extends that to a claim about truth and the way
things are: ‘as things seem for me, so they are for me’. This is universalized (always: as
things seem for me, so they are for me) but ineliminably relativized (always: as things
seem for me so they are for me). In Protagorean relativism, that is, what things are like
for me is a permanent (and exhaustive—but that is another issue) feature of my
account of truth and the way the world is; and the same goes for you, too. Protagoras
87
Notice his presence at de anima 427b3; and compare Metaphysics 1009a6 ff.
cannot accommodate reflective detachment; for Protagoras, all my phenomena are

ineluctably mine.88
Aristotle rejects both this account of what it is to perceive and any connected
account of what it is to perceive that we see and hear: his account of appearances is
designed to show how appearances themselves may be subject to reflection and
correction.89 But the argument of 3.2, nonetheless, allows him to include these
reflective judgements in the purview of perception (and so discussed further under
the heading of phantasia in 3.3) rather than attributing them to some separate
rational activity standing outside the perceptual process. Instead of consciousness,
he proffers an account of perceptual judgement as indeed a part of perception; not
regular or necessary for individual events of perception, but essential nonetheless to
our understanding of what it is to perceive, of the collection of cognitive phenomena
described by expressions like ‘perceiving by sight’. On such an account perception is
both one and not one: each in a way. It is not one, at least because there is a contrast
to be drawn between first-order perceptions (where the object and the subject are
actualized together) and second-order ones—which may not even occur when a first-
order perception takes place (first-order perception is only perceptible). It is one,
nonetheless, because even these second-order perceptual judgements are perceptions:
they are genuinely reflexive, in that they are perception properly so-called of
perception.
The connection with the Nicomachean Ethics, on the other hand, allows us to see
just where (and how radically) Aristotle offers this account of perception to replace
Protagorean relativism. Perceiving that we perceive is not phenomenal conscious-
ness, because higher-order perception is not necessary for (or intrinsic to) perceiving
in a first-order way. But the account of perceiving that we perceive that does issue
from the chapter is not focused on the person who perceives—quite the contrary, the
argument of 425b13–17 is notably impersonal. I suggested that the first-person
formulations with which the chapter began mark their status as the phenomena to
be described—where what catches Aristotle’s interest is the ordering of perception,
not its indexing to the person who perceives. However, the role of Protagoras in the
background to this chapter, and the significance of perceiving that we perceive in the
arguments of EN 9.9, may give a better account of why Aristotle begins with what
happens to us. For when I consider the development and flourishing of my faculties,
I do so in terms of my own life (the puzzle about friendship, as Aristotle sets it up,
88
The ‘for me’ business could express the privileged view, the view especially from here, from my end of
things (and so be congenial to claims about the incorrigibility of our perceptions or appearances by others);
or it could be designed to show that a given perception is mine (not yours) and so an inalienable part of my
world view.
89
We say ‘it appears to be a man’ just when there is some doubt about the matter (428a14 ff., echoing
the Philebus). And our judgements may be corrections of what appears to us: the sun appears to be a foot
wide, but we believe it to be greater than the inhabited world (428b3–4).
rests on why, if I am self-sufficient, I should have any room for others in my life). It is
essential, then, to understanding higher-order perception that we think about it in
the context of a life belonging to a determinate person. Perceiving that I perceive,
therefore, is essentially indexed to me; but this indexing comes from the normativity
of perception, not from its connections to consciousness.
In this Aristotle is true to his Platonic antecedents: for both in the Charmides and
even in the Theaetetus Plato is attending to the nature not just of cognition, but also
of virtue and wisdom. In the Charmides self-knowledge (and its analogue, self-
perception) is of interest because it may explain how best to live, and it may explain
what the virtue is that is called sôphrosunê, self-control, or temperance. In the
Theaetetus, the discussion of perception and judgement comes at the end of a long
engagement with Protagoras and Heraclitus, whose views of truth and reality are
contrasted with the reflective life of the philosopher (172a–177b). That reflection, in
turn, is contrasted with the state of mind of the clever politician, who is unable to see
the way things really are (e.g. 176e4–5, 177a1). In Plato, no less than in Aristotle,
perception has a normative cast.
10. A Philosophical Antique?

What exactly would be in this for Aristotle? What would be in this for us? How far
does Aristotle’s claim, as I have construed it, simply fall to the complaint that these
are just ancient lecture notes, of no significance to us now? Why, oh why, didn’t
Aristotle just see that consciousness is really important, much more important than
this assimilation of perceptual judgement to perception proper? If Aristotle fails to
see that there is something that it is like for me to perceive, and so fails (on the
account I have given) to explain the difference between my smelling the garlic and
the butter smelling of garlic, so much the worse, either for Aristotle or for my
interpretation of him.
In response, I offer only a gesture, in the direction of Aristotle’s broader agenda.
The current debate about this passage takes Aristotle to be interested here in some
kind of account of the nature of perceptual awareness, whether that be by virtue of
the subjective feel of perception (Kosman), the intentional structure of consciousness
(Caston), or the inner sense that explains it (Johansen). But this debate takes its
agenda from the discussion of the mechanics of perception in earlier chapters, not
only in de anima 2.12, but also earlier (2.5). However, from the beginning of de
anima 3, the argument has been about the number and nature of the faculties of soul;
and it is no surprise that the book continues with a discussion of the queerness of the
faculty that delivers phantasia, as well as the complex nature of thought and
knowledge (3.4, 3.5). But this talk about faculties is connected to the talk about
functions in the ethical works (and indeed it is one of the most significant features of
the de anima that it has this kind of continuity with both the ethical and the
metaphysical works).90 Both the de anima and the Platonic texts to which it is heir
have the same presumption that questions about epistemology cannot be divorced
from questions of ethics: indeed, to ignore this feature of the Platonic background
risks missing the point of the complexity of de anima 3.2 altogether.
Although perception by sight is ‘not one’, I have argued that nonetheless in de
anima 3.2 higher-order perception is rendered continuous with the natural activity of
perceiving first-order sensibles, by still qualifying as perception. This continuity
allows Aristotle a naturalist account of perceptual judgement, wherein the activity
of judgement is itself part of man’s natural function. Judging, at least at the perceptual
level, is something we do by nature. It is not immune to error, for sure; but it is
something that can be explained by reference to the essence of man, and thence as
something we can be good at. His claim, then, is not so much that we perceive that we
perceive every time we perceive (not, as I have argued, an account of regular
phenomenal consciousness, so not ‘always or for the most part’ in this sense) but
rather an account of how we may develop our natures to exercise cognitive function
of this kind—given the right development, then, man ‘always or for the most part’
makes perceptual judgements well: and he can come to do so by reflection on
perception.
Well, that may still look like an antique, until we recall the wider role Aristotle has
for judgement, and for good judgement, in his account of how best to live. Consider
the following:
But since rhetoric is for the sake of the giving of judgements (krisis) (the hearers judge between
one counsellor and another, and a legal verdict is a giving of judgement) rhetoric should look
not only towards the judgement, that it should be demonstrative and convincing, but also to
make both himself and the judge disposed in a certain way. (Rhetoric 1377b20–7)
Being a good judge is a matter of how you are disposed, and how your dispositions
are developed. This is true in both ethical matters and elsewhere: compare the
account of dialectical development:
Therefore one should have surveyed all the difficulties beforehand, both for the reasons we
have stated and because people who inquire without first stating the difficulties are like those
who do not know where they have to go; besides a man does not otherwise know whether he
has found what he is looking for or not; for the end is not clear to such a man, while to him who
has first discussed the difficulties it is clear. Further he who has heard all the contending
arguments, as if they were parties to a case, must be in a better position for judging.
(Metaphysics 995a33–b3)
Aristotle supposes, then, that judgement is common to perception and to thought:
90
Compare the notoriously difficult 3.5 with the discussion of theôria in E.N. X. 7, and the account of
the unmoved mover in Metaphysics XII. 9.
Knowing and thinking seem to be like perception in a certain way (for in both of these [sc. too]
the soul makes a judgement and recognizes something of the things that are). (de anima
427a19–21)
Second, he supposes that judgement in general is normative. To understand how best

to live, we look to the good person; and the good person exercises good judgement
(E. N. 1099a23, 1113a4 ff., 1143a6 ff.). This is something the good person does both
as a matter of understanding and rationality (cf. E.N. 1141b9 ff.) and as a matter of
developing her own nature best (cf. Politics 1.2, justice is a virtue of judgement,
1253a38; Rhetoric I. 1–2). But good judgement in the ethical sphere is not simply
analogous to the higher-order judgements of perception: just as perceptual judge-
ment is continuous with first-order perception, so too the judgement we need for the
best life is continuous with the good judgements we may make in perception. We can
see all sorts of things: valid arguments (Prior Analytics 24b24); the right thing to do
(E.N. 1143b5); first principles (E.N. 1098b4); the quality of a life (E.N. 1100a33). And
we can learn to see better, to be good at seeing these things.91 Aristotle neither
supposes that there are two quite different things going on here (judgements of value,
perhaps, as opposed to judgements of fact) nor that the development of our cognitive
skills and abilities are separate, the ethical on the one side, the matter-of-fact on the
other. Instead he imagines that the life of a person, and human nature, is seamless.92
Is he wrong?
11. Coda: Aristotle Reading Plato

Return, finally then, to the old thought that Aristotle’s works are often just jottings
for lectures, or the notes taken by a sharp student; and that this explains his
rebarbative style. I have suggested that in fact this short passage from the de
anima, apparently an extreme example of the gappiness of which he is so often
accused, may be viewed quite differently if we read it as itself a complex reading of
two Platonic dialogues.
First of all, these chapters of the de anima follow an agenda set by the Theaetetus.
The Theaetetus raises three fat questions about perception (and, I think, remains
unclear about just how far any of these questions are to be solved). The first is
whether perception is to be understood as a raw mechanism, wherein the agent
perceives just when affected by the patient. The second is how perceptions are
somehow brought together to form a judgement that would include the ‘common
terms’ as well as the objects of a special sense. The third is how perception can be
mistaken: how can perception deliver falsehood? All of these questions are central to
91
Pericles is thought to be practically wise because he can see (theôrein) what is good for himself and for
others (E.N. 1140b8 ff.); and practical wisdom is a virtue or excellence, at which we can improve.
92
The Protagoras figure in the background is dismissed because since relativism rules out higher-order
judgement, it rules out the possibility of a proper life, too.
Aristotle’s discussion in de anima 3; and they are put in terms that make it quite clear
that Aristotle is actually reflecting on the dialogue itself, not on doctrines we might
think to have been abstracted from it. But the Theaetetus does not include a
discussion of perceiving that I perceive; Aristotle gets that, as I have argued, from
the Charmides. He puts it here because he supposes—what apparently the Theaetetus
does not allow—that perception is not as simple as the Theaetetus allows.
But the Charmides, read as a whole, takes perception to be just as complex as
Aristotle would like: in the Charmides for example, Socrates might be charged with
looking inside Charmides’ soul to see his virtue. Perceiving, in the Charmides, is a
complex matter, not restricted to the raw feels of sensation, and careless, equally, of
the problems of phenomenal consciousness. Aristotle introduces the Charmides to
the Theaetetus; and perhaps in so doing he too reads the Charmides whole, realizing
that Plato was before him in taking the business of perception to be about our broad
cognitive processes. For Aristotle’s text, the interplay between these two Platonic
texts and his own arguments is responsible for the lacunae; this is not a set of lecture
notes, but a complex reading of Plato.93
93
Many people have discussed the topics of this chapter with me, in various different contexts. I am
grateful to Melita Brownrigg for a discussion about the phainomena; to Hamid Hejawi for a discussion
about the professionalization of perception; to Peter Adamson, Verity Harte, Ursula Coope, Frisbee
Sheffield, and the other members of the KCL Thursday seminar on this text; to those participating in
discussion at the conference, especially to Richard Sorabji and David Wiggins; to Peter Adamson and to
Peter Baumann for their written comments; and to Mark Textor for extensive discussion. My special
thanks to Alan Lacey, for his comments on this chapter as well as for his sapience always in discussion of
ancient texts, and for his shining example of the most virtuous way to do philosophy—with honesty,
gentleness, and an undeviating desire for the truth. I am also very grateful to the Leverhulme Trust, for the
Major Research Fellowship during the tenure of which I wrote this chapter and co-edited the volume in
which it first appeared.
15
Some Conversations with Plato:
Aristotle, Metaphysics Z. 13–16
1. The Notoriety of Z. 13–16

A notorious section of Metaphysics Z offers a fresh start (1038b1–2) and a discussion
of the candidate for ousia1 so far left unaccounted: the universal. That discussion
runs all the way to the end of Z. 16, where a double conclusion seems to be reached:
That none of those things said universally is ousia and that there is no ousia from ousiai is
clear. (1041a3–5)2
But it is disputed just how Aristotle reaches this conclusion; and disputed just how far
the conclusion is consistent with whatever else Aristotle wishes to conclude about
ousia (since some of the features of universality seem to belong to Aristotelian form).
I shall endeavour to sidestep the question of Aristotle’s overall conclusions about
ousia in favour of a more limited question about how he proceeds in these four
chapters—although perhaps the limited answers may range unexpectedly wide.
I shall make six suggestions:
i. The apparently sequential argument of the chapters is broken into seven
separate arguments.
ii. Those arguments are dialectically constructed: that is, they are constructed as
disputes between two different, or opposed, points of view.
iii. Each of the arguments contains at least one point of view that is represented in
some Platonic dialogue; and these representations are a deliberate feature of
them.
iv. But in this respect each of the arguments is self-contained as well as separate,
each tackling a different Platonic view, neither implied by nor implying the
view of the preceding argument.
1
Cravenly, perhaps, I leave this crucial term untranslated. A traditional translation of ousia in the
Metaphysics is ‘substance’. I have not transliterated Greek in the footnotes, since most of these points are
about the language and the text; but I have still transliterated ousia throughout.
2
Translations are mine, from Jaeger’s OCT unless otherwise noted; texts are from Metaphysics Z, unless
otherwise noted.
SOME CONVERSATIONS WITH PLATO 311
v. There is no canonical ‘Plato’ here, but rather a series of representations of what

Plato says in different dialogues: nothing here requires us to suppose that all
seven Platonic points of view add up to some single coherent (or supposedly
coherent) set of doctrines, or to suppose that Aristotle thought they did.
vi It remains, however, at least indeterminate just how far any of the opposing
views should be ascribed to Aristotle in propria persona, since the dialectical
context dominates the entire sequence of conversations. Often positions that
seem to be uttered by the author of the work are best understood as proffered
by one of the parties to the dialectic, and limited thereto.
The overall sequence of conversations, nonetheless, provides a continuous, if
complex, examination of the constraints on postulating universals; to this extent at
least the arguments have a positive outcome. In conclusion, I shall wonder where this
leaves us in understanding how Aristotle thought Plato should be read.
2. The Divisions of the Arguments

The sequence of the text is carefully articulated by claims about what some people
think and about what follows for them, if they think that. The different stages of the
sequence itself are set out by prefatory remarks such as ‘and it is evident also in the
following way’ (1039a3) which both connect what follows with what went before and
mark it as distinct. On the basis of various such devices of demarcation, seven stages
of the argument may be separated out:
Section 1 (Z. 13: 1038b6–1039a3) 1038b6 opens with the view ‘of some’ about
causation (aition):
The universal also is thought by some to be a cause most of all, and to be a principle; therefore
let us proceed to the discussion of this point also. (1038b4–6)
It proceeds through a series of different arguments against this view, and seems to
have a conclusion in:
To those looking at the matter from these considerations, it is clear that no universal attribute
is ousia; (1038b34–5)
with the coda:

and that no common predicate indicates a this something but rather a such. Otherwise, many
difficulties follow, including the third man. (1038b35–39a3)
Section 2 (Z. 13: 1039a3–23): the discussion then moves on:

It is evident also in the following way—
and runs to the end of Z. 13, with:

And what is said will be more evident from what follows. (1039a22–3)
312 SOME CONVERSATIONS WITH PLATO
Section 3 (Z. 14: 1039a24–b19): the next chapter (Z. 14) begins:
It is clear from these very things what is the consequence also for those who both say that the
ideas are separate ousiai and make form from the genus and the differentiae. (1039a24–6)
It then runs through an extended argument against this composite position, and
concludes:
If, then, it is impossible that things should be so, it is evident that there are not forms of sensible
things in the way in which some say there are. (1039b17–19)
Section 4 (Z. 15: 1039b20–1040a7): the opening of Z. 15 is more abrupt, picking up

on claims made earlier in Z:
Since ousia is different in the cases of the composite thing and of the account . . . (1039b20)
Through an interim conclusion:

. . . it is evident that there can be neither definition nor demonstration of sensible individuals.
(1040a1–2)
and then finally:

Therefore whenever one of those who are busy at definitions would define any individual, he
must not be ignorant that his definition may always be overthrown; for it is not possible to
define such things. (1040a5–7)
Section 5 (Z. 15: 1040a8–b4): here once more the parties to the discussion come into
view:
Nor is it possible to define any idea. For the idea is, so they say, an individual, and
separate . . . (1040a8–9)
and remain noticeable throughout:

As has been said, then, it escapes their notice that it is impossible to define in the case of eternal
things, especially those which are single, such as the sun or the moon. For they make a mistake
not only . . . (1040a28–30)
until the concluding passage:

Why does no one of them produce a definition of an idea? It would become clear, if they tried,
that what has now been said is true. (1040b2–4)
Section 6: now (Z. 16, 1040b5–27) the debate seems rather more generalized:
It is clear that even of the things that are thought to be ousiai . . . (1040b5)
to a further conclusion that ends:

Further, that which is one could not be in many places at the same time, but that which is
common is present in many places at the same time; so that it is evident that no universal exists
separately apart from the individuals. (1040b25–7)
Section 7 is brief (Z. 16, 1040b27–41a5), from:

But those who posit the forms, in one respect are right . . . (1040b27–8)
to the end of Z. 16, which seems to provide a more general conclusion:

So that none of the things said universally is ousia nor is there an ousia of ousiai, is evident.
(1041a3–5)
The articulation of these chapters is made by means of their dialectical structure: for
the lines of demarcation are provided by the introduction of different views about the
subject in hand, and concluded by claims about what is clear, after consideration of
those views. We might compare the advertisement for this kind of procedure in
Metaphysics B1:
So it is necessary to have looked at (tetheôrêkenai) all the puzzles first, both for these reasons
[sc. to avoid puzzles, aporiai, by stating them well, 995a28] and because those who inquire
without first puzzling are like those who do not know whither they are going; and besides a
man does not then know whether he has found what he seeks or not. For the end is not clear to
him, whereas to the man who has puzzled beforehand the end is clear. (995a33–b2)
or in the Topics (100a30ff.) which is optimistic about how dialectic contributes to the
understanding of starting points:
When we are able to go through the puzzles on both sides we shall more easily see clearly both
what is true and what is false. (Topics I. 2, 101a34–6)
Here in Metaphysics Z Aristotle repeatedly offers summative remarks about the

parties to the dispute, so that we are indeed considering the question ‘from both
sides’. As a consequence, each of these sections can be read as a separate conversation
between two parties whose accounts of the role of the universal are somehow at odds
with each other. This has the further effect of giving us, his readers, a reflective view
of the positions in the debate. That reflectiveness is reinforced by Aristotle’s regular
use, at the end of stretches of argument, of ‘it is clear that’, ‘it is plain that’, and
similar expressions.3 The effect of these Aristotelian tropes is at least to engage
the reader critically in the claim being made—so that the argument may be seen
from the reader’s perspective (it is clear to us, as we read). There is, however, more to
the dialectical structure of the passage, as I shall suggest, so that learning to see clearly
is a more complicated business than at first it might seem.
These four chapters of Metaphysics Z are, unfortunately for my purposes, some of
the most contentious in the corpus. My object in what follows is to expose their
dialectical structure. To do so, I have perforce been cavalier in the interpretation of
3
Frequent markers are ‘it is clear’ (çÆæe, 1038b34, 1039a24, 1040b5, 18) and ‘it is evident’ (Bº,
1039a3, 11, 22, 1039b18, 1040a1, 1040b3, 26, 1041a5): on the role of these expressions in dialectic, see
Owen (1986a); Nussbaum (1982).
individual arguments: I hope, merely, to offer a plausible account.4 Under the

constraints of space, furthermore, I shall focus my attention on the first two conver-
sations, leaving the last five ever more sketchily filled in; for all of this I beg the
indulgence of my reader.
3. The First Conversation: Structure

Why should the universal have a claim to be substance? Well,
The universal also is thought by some to be a cause most of all, and to be a principle. (1038b7)
These ‘some’ provide the starting point for the discussion that follows. Against them,
it seems to be impossible that any of the things said universally should be ousia. (1038b8–9)
in favour of which is ranged a series of objections to what the ‘some’ say (at
1038b9–15, 1038b15–16, 1038b23–29, 1038b29–30), ending with consideration of
how things end up if the ‘some’ are right (1038b30–34). This then issues in a
summative agreement:
is ousia. (1038b34–5)
supporting the earlier denial (1038b8–9) of the title of ousia to the universal.
But the argument itself is elliptical, and contentious. It has two features that might
incline us to think it especially dialectical. From the outset, first, it ranges two
opposed positions against one another. One of these (1038b7–8: call this the position
of A5) makes two claims: that the universal is (most of all) a cause; and that it is a
principle.6 These two claims are taken to imply that the universal is ousia, since the
following counter-argument (1038b9ff.) focuses on that, and not on the point about
causation.7 Conversely, the counter-argument starts from how things seem (1038b8);
it is not merely a series of objections to A’s position, but rather itself a detailed
position (it includes, for example, the claim, with which A concurs, that the ousia is
peculiar to the thing whose ousia it is, 1038b10: call this the position of B).
As the argument proceeds, second, we see some development, at least in the
position of A. The claim that the universal is ousia is elaborated as committing
A to the further claim that the universal is the ousia of ousiai: but this elaboration is
4
There is an excellent and comprehensive account of all the arguments of Z, and various interpret-
ations, by Heinaman at <http://www.archelogos.com>.
5
I refer to the protagonists of the arguments with a masculine pronoun, observing the conventions of
Aristotle’s culture, not our own.
6
This is suggested by ºØ Æ: so that the second claim, about the principle, is inferred from the most-
of-all-ness of the universal’s being a cause. ŒÆ at 1038b7 thus does not merely offer a hendiadys (this gives
point to Jaeger’s punctuation).
7
The opening claim about causation could be a pretext for a much more general argument against the
universal; but that reading would ignore the very specific features of the view of A.
derived by virtue of more that is said by A, at 1038b16–23. This dialectical structure,

indeed, rescues the argument from what seems, at first reading, to be hopeless
ignoratio elenchi (why, after all, should A be vexed for a minute about claims about
the universal’s being ousia, if what he said in the first place was that the universal is a
cause? Why, in the later exchanges, could he not forestall B’s objections by insisting
that all he claims is that the universal is a cause? If he says it is a cause, must he
thereby be committed to its being an ousia?). That, in turn, may allow us to see how
the interchange might play out—between someone who has views about what it is for
the universal to be a principle, but who is forced to modify those views in the course
of the argument; and someone who supposes that this involves A in a commitment,
among other things, to an absurdly over-populated count of entities.
So A starts out with two connected claims: that the universal is most of all a cause,
and that the universal is a principle. Suppose that A, in talking of causes, is talking of
real causes (whatever he says causation involves): so he is here talking about his
ontology, about ousia. Suppose, also, that any ontology is structured by the claim that
ousiai are prior to other things (this is the burden of Z. 1). A, in talking of principles,
asserts the priority of the universal, in some respect. Then A might be held8 also to
the claim that the universal is ousia. B denies this:
For9 it seems to be impossible that any of the things said universally should be ousia.
(1038b8–9)
The first stage of B’s counter-argument rests on claims that seem to be represented as
straightforwardly true:
First of all, the ousia of each thing is what is peculiar to that thing,10 which does not belong to
anything else; but the universal is common, since that is called universal which naturally
belongs to more than one thing. (1038b9–12)
At any rate, this contrast—between the peculiarity of the ousia of something and the
commonality of the universal—is taken to be agreed by A. Indeed, if the basis of A’s
interest in the universal is an interest in causation, he may well be interested in the
regularity of things, and thence in the commonality of the universal. But this contrast
between the common and the peculiar now prompts B to ask:
Of which thing then will this be the ousia? Either of everything or of nothing. But it is
impossible for it to be the ousia of everything; and if it is to be the ousia of one thing, this
8
This is not a direct inference: for example, A might suppose that there is something else prior to these
causes and principles, which would thence have a better claim to being ousia than theirs.
9
Does this particle align Aristotle (who is included in the subject of the previous clause, ‘wherefore let
us continue with discussion of this’) with the people to whom this seems impossible? Or does it mark a
continuation of the embedded discussion between A and B?
10
On issues of the text throughout, see especially Ross (1953); and Frede and Patzig (1988). Again for
reasons of space I have avoided detailed discussion of the many textual cruces in these passages.
one will be the others also;11 for things whose ousia is one and whose essence is one are
themselves also one. (1038b12–15)
Here the first element of the agreement between A and B (that the ousia of each thing
is peculiar to it) is restated as the claim that ‘things whose ousia is one and whose
essence is one are themselves also one’. But this then tells against the universal’s claim
to be ousia. Consider the things of which it is the common universal: either it is the
ousia of all of them (in which case they collapse into one); or of some of them (in
which case some are unwarrantably excluded); or of none (in which case the claim
that the universal is their ousia is subverted).
To this B adds another objection:
Again, ousia is said to be that which is not predicated of something that underlies; but the
universal is always said of something that underlies.12 (1038b15–16)
This objection is puzzling in the context of what A has actually offered: for nothing in
the opening claim about causation and principles mentions predication at all. Of
course, it may be implicit in the thought that the universal is common, that it is
predicated of a subject in this way. Or A may be imagined to have amplified his
position further to include something about how things are said. In what follows, it is
clear that A does indeed develop and alter his position; I suggest that here too B’s
argument assumes the moves of the opponent. And the point of his objection is clear:
if the universal is thus said of something else, then it cannot have the priority that
A originally claimed for it in saying that the universal is a principle.
Now the argument takes a turn,13 and A appears to modify his position:14
But perhaps, while it is not possible that the universal is ousia in the way in which the essence
is, it can be present15 in this, e.g. animal can be present in man and horse? (1038b16–18)
A seems here to be addressing the objection that the universal is common while
substance is peculiar to that of which it is the substance, by limiting the role of the
universal. For he wonders whether one might deny that the universal is the essence,
while retaining a role for it as present in the things to which it belongs (for example,
11
On the interpretation of this clause, see Burnyeat et al. (1979), 131, and G. Hughes’ paper therein,
107–26; also Woods (1967).
12
In this sentence the verb º ª ÆØ is repeated, but has quite different roles in the two clauses: in the first
it makes some theoretical claim about what is said about ousia; in the second it reports actual usage.
13
Heinaman (n. 4) calls this ‘one of the most difficult arguments in Z’; the grammar is condensed, the
text unsure, and the structure of the argument horrid to disentangle.
14
At 1038b16 Iºº’ pæÆ marks a response by the first party to the debate to the objection that has just
been made, and is marked by its exploratory tone. Although it is clear that A now ‘speaks’, it is not clear
where his speech ends. See Burnyeat et al. Notes (1979), 126 ff., and Gill (2001) who has A’s sequence of
thought extend to 1038b23.
15
Reading Kı æåØ with Ross Aristotle (1953), against Kı æåØ favoured by Jaeger and Frede/Patzig
Metaphysik Z (1988).
animal in man or horse).16 His move, that is to say, is to give an account of the
relation between the universal and the things it is designed to explain. This is either a
modification of his original claims or a specification of them: and it is not inconsist-
ent with them, at least prima facie. After all, it is entirely compatible with the
universal’s being a causal principle that it might do its causing by virtue of its
presence in things; and compatible also with A’s opening claim that the universal
may not exhaust what we should say, either about causation or about principles.
Nonetheless, it seems indeed to be a modification, since his opening position was at
least treated by B as one of full generality.
The next sentence is disputed:
Then it is evident that there is17 an account of it. (1038b18)
Reading the MSS text, this sentence should be attributed to A. In that case, it may
respond to B’s objection about predication at 1038b15–16, so that B’s emphatic talk
about what is said meets A’s response in terms of a logos, an account or a definition.18
The objection about predication had the universal’s being said of something, and
thus being secondary to whatever it is said of. Here, by contrast, A may suggest that
there will be an account of it,19 even if it is only present in things. In that case, this
remark represents in part the rationale for his modification: that he can say that the
universal is in things, without entirely sacrificing its explanatory role.
To this B responds in the terms A has offered, but harking back to his earlier
complaint about the scope of A’s position:20
It makes no difference even if there is not an account of everything that is in the ousia; for even
so this [sc. the universal] will be the ousia of something, as man is of the man in which it is
present, so that the same will happen again; for it, e.g. animal,21 will be the ousia of that, in
which form22 it is present as something which is peculiar to it. (1038b18–23)
B suggests that even if A narrows the scope of his claim about the universal, he will be
vulnerable to the objection that the universal, which is common, will need to be
16
The example is not redundant at 1038b23, as Ross suggests, whether or not it is an analogy rather
than an example.
17
Here Jaeger reads ÆØ, so as an argument against what was maintained in the previous sentence. But
the MSS have Ø (as both Ross Aristotle (1953) and Frede/Patzig Metaphysik Z (1988) print), which
suggests that this remark runs on from the previous sentence.
18
Heinaman has ‘definition’ passim; but if this is a response to B’s previous objection, ºª is here the
correlate of º ª ÆØ at 1038b15–16.
19
The grammar of Øe º ª ÆØ I at 1038b15 is mirrored in Ø ÆP F ºª at 1038b17.
20
Taking the sentence at 1038b19, ØÆç æØ PŁb, to correspond to A’s revision at 1038b16, Iºº pæÆ
o ø b. This makes the minimal assumption that A’s contribution ends at 1038b18, in the interrogative
(there are considerable difficulties in supposing that A’s contribution extends as far as 1038b23, despite
Gill’s (2001) emendation of Ø b ŒÆd), and then all of 1038b19–20 must belong to B.
21
Ross Aristotle (1953) deletes this example on the grounds that it disturbs the sentence structure;
Frede/Patzig Metaphysik Z (1988) likewise.
22
Does YØ belong? Ross Aristotle (1953) deletes it. If I am right about the antecedents of this passage,
the presence of YØ can be explained either as Aristotle’s debt to Plato, or as a scribe’s noticing that debt.
peculiar to the things of which it is the ousia. Once the universal is admitted as ousia
even in a limited number of cases, it will always seem to resist the earlier suggestion
that something and its ousia are one (1038b14–15). Indeed, B may be suggesting,
even A’s restriction of his thesis will not help him. ‘The same will happen again’:
either this means that A risks a regress (there will be more and more ousiai even in
the limited number of cases); or he just risks the previous argument (1038b8–15) all
over again.
Now B adds a further sub-argument, concentrating on the effect of A’s theory, even
in its restricted form, on the individual particular.
And further it is impossible and absurd that the this something and the ousia, if it is from23
things, should not be from ousiai nor from what is a this something but from a such; for that
which is not ousia, i.e. the such, will then be prior to the ousia and to the this something.
Which is impossible; for neither in account nor in time nor in generation24 is it possible for
affections to be prior to the ousia; for then they would be separable from it. (1038b23–29)
This argument is followed by another:

Again the ousia will be in Socrates too [sc. as well as in man], so that it will be the ousia of two
things.25 (1038b29–30)
The form of B’s objection could be construed as a dilemma. Think about the
individual particular, in the context of A’s revised claim that the universal is in
things, albeit not as an essence. What would it be for individual particulars to have
the universal in them? If the universal is not ousia, but rather, for example, a quality
(a ‘such’), then what is in the particular will be posterior to it. But if we suppose that
the individual particular is somehow or other ‘from’ the universal, this will be absurd
(since what is prior will be ‘from’ what is posterior). If, on the other hand, the
universal is ousia, and this includes generic universals, such as animal or even man,
then in the individual particular there will be two ousiai (animal, perhaps, and man;
or animal and Socrates?). And this—the argument supposes—would be absurd.
Once again, the structure of the argument is extremely dense: while it is still
focused on A’s revision,26 its formal structure is unclear. Suppose, however, that
this is a dialectical encounter between A and B, then the background question
pressed by B, after the revision, may be a question about the ontology of the
universal: what sort of thing is it? And how, whatever sort of thing it is, can it be
understood in relation to the things of which it is a cause or a principle? Either it is
23
‘From’ is tricky. In this context it reads as a claim about composition; but it could be about causation,
as A could have taken it at the beginning (he would then owe us an account of what he means by
causation).
24
Frede/Patzig Metaphysik Z (1988) have ªØ, ‘in knowledge’.
25
Among the questions raised by this sentence: what is the subject of either clause? How does the ousia
fit in? How does any construal of the first clause imply the second?
26
Notice Kı æåØ repeated at 1038b29.
not an ousia (but something from a different category): in that case it falls to the
priority objection—the individual would thus have as its cause or principle things
that are posterior to it. Or it is an ousia: in which case it falls to the objection that the
individual particular will either contain or be explained by several ousiai.27
However the argument works, three things are relatively clear. First, it focuses
attention on A’s revised position, examining exhaustively what A must say of the
universal’s ontological status (is it an ousia or not?). Second, it does so in the context
of A’s original claim about principles; for the sub-argument at 1038b23–29 assumes
that A cannot concede that the universal is posterior to the individual particular.
Third, however one construes the sub-argument at 1038b29–30, it represents as
absurd—that is, as absurd from the point of view of either A or B—that the universal
should somehow or other proliferate entities (hence the emphatic ‘two’ at 1038b30,
following up the implications of the opening argument at 1038b8–15). To these
features of the argument I shall return.
B seems to take it from this that A concedes:28 at any rate the conversation now
seems to have the air of a conclusion:
In general it turns out that if man is ousia and likewise the things that are said like that, then
none of the things in the account is the ousia of anything, nor exists separately from them, or in
anything else; I mean,29 for example, that there is no animal besides the particular animals, nor
any other of the things in the account. (1038b30–34)
These remarks confirm the focus of attention throughout: on the relation between the
universals espoused as principles by A and the things of which they are the principles.
This relation is sometimes considered in terms of the status of the universal—in
which case the expression regularly used is that the universal is the ousia of the
particular. Or it is considered in terms of the consequent nature of the particular—in
which case the relation is regularly described as the particular’s being from ousiôn.
These relations are apparently treated as converses of each other, and as reflecting,
simply, the relatum under scrutiny—sometimes the universal (e.g. at 1038b20–3),
sometimes the particular (e.g. at 1038b23–7): they do not, that is to say, have at this
point a great deal more content than the relation itself. Either version of the relation,
however, causes trouble to A if he insists that the universal is separate.30
27
This last sub-argument is horribly terse; see the note of despair sounded by Frede/Patzig Metaphysik
Z (1988), ad loc.
28
Pace Hughes’ version of this, in Burnyeat et al. Notes (1979), 107 ff.
29
Is this Aristotle speaking? See section 8.
30
If this is right—the ‘of ’ and ‘from’ are just the converses of each other in respect of the relation
between something and its ousia—then we might expect different ways in which these relations are glossed.
This is, in fact, exactly what seems to happen in these chapters; which is why the ‘from’ sometimes
expresses a causal relation (as perhaps at 1038b23–7), sometimes a constitutive one (as at 1039a8); on the
slipperiness of ‘from’, see Metaphysics V. 24, 1023a26–b11.
The closing remarks of conversation 1 (1038b30–4) suggest a negative outcome,

for A. Since the dialectic concludes here, and A makes no further contribution, he
seems to have conceded to B. Then the final stages sum it all up:
is ousia; and that no common predicate indicates a this something but rather a such.
Otherwise, many difficulties follow, including the third man. (1038b34–1039a3)
To the nature of these concluding remarks I shall return.
4. The First Conversation: The Background

The progress of this conversation is striking. It is written, for sure, in terse prose, and
the exact delineations of the interchange between the two parties are disputable.
What is indisputable is that both sides to the exchange have a complex view of the
way things are with the universal. Notably, the position of A is detailed and
structured, as well as being subject to revision and reconsideration. It begins, as
I suggested, with an inference (from a claim about causation to a claim about the
universal’s being ousia); and develops by means of an implicit account of how the
universal is related to the particular, namely by the presence (enuparchein, at
1038b18, 29) of one thing (the universal) in another (the particular), where the two
relata are not merely distinct but separate (chôris, 1038b32, chôrista, 1038b29). From
his first concern for causes and principles comes A’s interest in the title of the
universal to being ousia; but that title is ultimately rejected.
But in this rejection A and B have some quite different principles at stake. A wants
to provide a theory of causation. This ambition dictates both that the universal have
some priority and that it be separate from the things of which it is the cause. Thus
even though causation is not explicitly the focus of B’s later objections, the implica-
tions of supposing that the universal is a cause are. Causation lies at the heart of A’s
position. But then he is trapped in the dilemma that either we have a causal relation
that seems to proliferate ousiai31 or we have a causal relation wherein the cause is
somehow posterior to the effect. B presses the ontological profligacy of A’s view, and
seems to conclude that this account of the causal role of the universal is untenable
because it doubles the entities in its ontology. But his parsimony is not merely
negative; it underpins the claim that the ousia of something is peculiar to it
(1038b10) and hence the shape of the argument at 1038b8–15, which allows a
relation between something and its ousia, only if it is one of numerical identity.
Somewhere in the background to this conversation lurks Plato. Most obviously,
of course, the objection with which the passage closes comes from the Parmeni-
des (‘many difficulties follow, including the third man’, 1039a2–3, compare Parmenides
31
This is the effect of the sub-argument of 1038b29–30, especially if the text includes P Æ P fi Æ at 29.
132a–b, 132d–133a).32 In the present context, moreover, the eponymous character of

the Parmenides has more to offer B.
B focuses attention on three aspects of his opponent’s view: the relation between a
universal and the things that fall under it (hence, from 1038b8 onwards, how should
we say that the universal is of something, or the something from the universal?); the
scope of any theory of universals (hence, at 1038b18 ff., will limiting the scope of
his theory of universals allow A to continue to treat them as principles?); and
the ontology of universals (are they, for example, separable? 1038b27 ff.). In the
Parmenides, Parmenides’ objections to Socrates’ theory of forms turn on just these
three issues. The puzzle of the relation between forms and particulars runs right
through this stretch of the dialogue (from 131a ff.: are the forms divided to cover the
particulars? Or wholly in each of them? Or what?). The issue of the ontology of forms
is introduced by Socrates’ first claim that forms are separate from the things that
share in them: Parmenides 130b. And the opening discussion between Socrates and
Parmenides (130a–d) asks just how far this theory need extend (130c7): must there
be forms beyond sameness and difference, or the just and the good? Must there be a
form of man (130c2)? What is the scope of the theory? What is more, just as in Z. 13
B implies that parsimony is a principle for ontology, so equally Parmenides’ closing
arguments (notably the greatest difficulty, 133b ff.) will not work as objections to
Socrates’ claims without a tacit reductive assumption. In general, then, the arguments
of the Parmenides press not only on whether the theory of forms is coherent, but also
on whether it is excessively expensive of entities.33 The arguments of the Parmenides,
I suggest, provide a detailed antecedent to B’s approach to A on universals: either
directly, or mediated by the topoi of the Academy.34
A different feature of the Parmenides’ arguments might give us pause here: the
theory they tackle bears considerable affinity with the theory of forms of the
Phaedo.35 A’s view recalls, in several ways, Socrates’ theory to account for ‘the what
it is’ (75d).
First, A argues from a claim about what causes ‘most of all’ to a claim about the
principles of things, and then to a conclusion about ousia. This collocation of
interests—causes, principles, ousia—is characteristic of the later sections of the
Phaedo, the passage known as Socrates’ autobiography. Thus, for example, at 96a,
the search for the aitia of each thing—including the hot and the cold, man and
32
On the Aristotelian name of the Third Man, see Fine (1993), 203 ff.
33 34
See McCabe (1994a), 78 ff. See Fine (1993).
35
The correlation is closer than between the Parmenides and any other dialogue’s account of forms
(with the exception of the Timaeus whose discussion is concerned with the natural world—men, horses,
animals, as well as with likeness and unlikeness, and goodness). So we find especially relational or more
generally metaphysical forms like unity, similarity, equality, or largeness (Parm. 129a–b, 130b, 131; and
Phaedo 74b, 75c, 101a ff.); evaluative forms such as beauty and goodness (Parm. 130b and Phaedo 76d–e,
100b–d); and substantial forms, notably, if controversially in both contexts, fire (Parm. 130c and Phaedo
105c). For the teleology apparently raised by the question of ridiculous things like mud and hair at Parm.
130c–d, compare Phaedo 96 ff.
animal36—is connected with understanding what each thing is; and this echoes an
earlier interest in ‘the ousia itself of whose being we give an account,37 when we ask
and answer questions’ (78d1–2). Further, at 99a–d, Socrates declares an interest in
classifying causes and rank-ordering them in terms of their epistemic importance: so
he would be interested in understanding how we may discover principles ‘most of
all’.38
Second, in the context of this discussion of causation, some of Socrates’ interest in
the Phaedo lies in the way in which causes are regular and repeatable: so neither
private nor peculiar to the effect. This emerges both from the puzzles about causal
explanation (e.g. how can addition and division both explain one thing’s becoming
two? (96e)) and from the discussion of teleology, where the explanation of things will
be common to all that it explains (98b). This commonality, too, seems to lie at the
base of the universal’s claim to being a principle, on A’s view (1038b11). When it
comes to teleological explanation it is hardly surprising to find a connection between
causation and principles, just because one dominant account of teleological explan-
ation rests on the idea of a principle (an archê) which is directive.39
Third, the last argument of the Phaedo offers two accounts of causation. The first,
in response to the puzzles about proper causal explanation (the head and one/two,
96d–97a, 100c–101c), is the simple-minded answer, according to which ‘the beauti-
fuls are beautiful by virtue of the beautiful’ (100d). For this account, the exact relation
between the beautiful and the beautifuls is left unspecified. Later, however, the
‘cleverer answer’ (105c) rests on a distinction between the properties of things that
are ‘in us’ and the forms that are not; here the thought is that there may be causal
relations explained by our properties, where those properties in turn are explained by
a form. The language of being ‘in us’ (102d7, 103b5, etc.) brings in a contrast between
the possessors of properties and the properties that are possessed: a contrast on
which the final argument for the immortality of the soul depends. But this is the very
distinction that causes A trouble at 1038b25, and into which he falls by virtue of his
own talk about things being in things, in his attempt at a defence of his view at
1038b17.40
36
Socrates’ predecessors were interested in the explanation of animals and man (Phaedo 96b), even
horses (96e). After such a context reference to the third man (1039a2–3) may not after all be rebarbative.
37
Give an account = ºª ØÆØ; at Phaedo 99e ff. the giving of an account is connected both to
causation or explanation and to the dialectical procedure of question and answer.
38
Socrates’ autobiography offers a thumbnail sketch of earlier attempts to explain (to give the Øa ) of
these phenomena—Aristotle himself describes this as the search for ÆN ÆØ and IæåÆ (e.g. at Metaphysics
982a5 ff.).
39
At Phaedo 97c, 98c, Socrates uses the language of ordering ŒE, ØÆŒE rather than ruling,
¼æåØ; but at 100b5 he insists that the forms are starting points or principles in causation, ¼æåÆØ
I KŒ ø. At de anima 405a15 Aristotle points to F as an Iæå for Anaxagoras—a figure of note, of
course, in Socrates’ autobiography. Of course the Timaeus deals with similar issues, in similar terms; that is
itself a matter of Plato’s own intertextuality.
40
Kı æåØ, 1038b18, 29, is, I suggest, an Aristotelianizing of e K E at Phaedo 102d ff.
Fourth, the Phaedo is expressly undecided about what the relation between forms
and particulars actually is (100d). This—if A is an exponent of the Phaedo—allows
him licence to speculate about whether it is a relation of presence in, or something
more loosely described, hence the tentativeness of 1038b17. This issue, I shall suggest
below, is taken up in the second conversation.
Fifth, the Phaedo rests some of its argument about the immortality of souls on the
claim that souls are akin to forms.41 If this is to show anything about the nature of the
soul after death, it needs to insist on some kind of separation: of forms from
particulars, as for souls from bodies (see Phaedo 67c). By contrast, the language of
separation is rare elsewhere in Plato to describe the relation between forms and
particulars, with the exception of the Parmenides.42 In Z. 13 A is under attack for
finding himself committed to the view that the properties of things—if that is what
universals are—are separable and ousiai; and this leaves him in difficulties about how
they could be causes or principles at all (1038b29; and see 1038b32).
Sixth, the entire discussion of causation from Phaedo 96a ff. is set in the context of,
first, the critique of the views of others and, second, of the defence of one’s own
position in dialectical exchange (especially 101d–e). In this, of course, it is echoed by
the Parmenides, which shows us the young Socrates in discussion about his theory of
forms. The ostentatious biography of the Parmenides, one might think, has the effect
of recalling the autobiographical features of the discussion in the Phaedo, and of
recalling in particular the argumentative context. The two dialogues are well
matched; and it would hardly be surprising that at the very least the Academic
tradition continued to see them as companion pieces.
There are not only verbal echoes of the Phaedo in A’s position, but also argumen-
tative ones.43 Consider the development of the account of causation in Socrates’
autobiography.
This begins with a principle: ‘the beautifuls are beautiful by virtue of the beautiful’.
This simple-minded answer exemplifies how causal explanation should be con-
strained (neither should opposites explain the same thing, nor should the same
thing explain opposites) and how it should be amplified (there is such a thing as
‘the beautiful’ which is not one of the things it explains). But it falls short, we might
think, of explanation, falls short of giving us something more than this austere
formula, and tells us nothing either about teleology or about more constructive
ways of accounting for why what happens, happens. The second, cleverer answer
which Socrates elaborates next (102b ff.) might be thought to respond to that
shortfall, in offering an account of explanation that, while it is not universally
41
This appears formally at 78b ff., and again as an assumption in the last argument, 102d ff., as I have
argued (1994), 47–8.
42
The Politicus (e.g. 262e, 280b) and the Sophist (e.g. 253d) describe not the relation between form and
particular, but the relation between forms, or between classes, as separation.
43
The simple-minded answer is proposed in a context where there is a difference of opinion going on:
Phaedo 101c ff.
applicable, is at least informative in individual cases: thus fever may be explained by

the presence of heat, or someone’s being alive by the presence of soul.44 This
suggestion that causation is involved in what is in things cannot be used for every
case, as the simple-minded answer can (the beautifuls are always beautiful by virtue
of the beautiful), but it can nevertheless tell us something significant for those cases
where it applies. It is exactly this move that A makes in responding to the first attack
on the theory he has put forward. For at 1038b16 ff. he faces the challenge that if the
explanation gives the ousia of something, we shall have just one thing, not two,
and the illumination that the universal seems to provide will fail. Consequently,
he modifies his view to cover cases where the collapse of the explanation into the
explanandum will be blocked: for example, in the case of the relation between the
genus (e.g. animal) and the species (e.g. man). In such cases, A suggests, we shall still
have a role for the universal; and it can still provide some kind of explanation; but it
does not risk identification with the thing it is supposed to explain. Just such a
modification is offered by the clever answer of the Phaedo, which improves on
simple-mindedness by citing something that is not the eponym of the thing to be
explained. And it does so, just like A, by couching the explanation in terms of
properties (as B comes to construe it) in things. A’s dialectical position, that is to
say, is closely, and peculiarly, connected to the position put forward by Socrates in
the last argument of the Phaedo.
A minor pay-off from this suggestion might be this. There is, as I have pointed out,
a thoroughly tricky sequence of thought at 1038b16–23, in the course of which
A proffers at least the revision of his original claim (1038b17–18), but whose
connection to the argument that follows is tricky. But if we find the Phaedo in the
background here, we may understand just why A might follow the modification of his
view (‘But perhaps, while it is not possible that the universal is ousia in the way in
which the essence is, it can be present in this, e.g. animal can be present in man and
horse?’) with a claim about the account (‘Then it is evident that there is an account of
it’). After all, just this sort of account is the object of the inquiry in the Phaedo, as we
have seen. But in that case, we may imagine the sequence of thought to follow
through from B’s previous objection. For at 1038b15–16 he complains that the
universal figures in the wrong place in our understanding of things: when we talk
about the ousia we talk about something that is not said of a subject, but the
universal, by contrast, is said of subjects. The focus of the objection is that this will
make the universal posterior in the order of explanation, so that it must fail to be a
principle. A’s riposte is that the universal can still (as in the clever answer) provide
explanation because it is in things; and in that case, there will be an account of it, so
that it is itself explanatory. A picks up, that is to say, the objection at 1038b15–16 as
an objection to how the universal is related to a logos—where for A the crucial role of
44
I have benefited here from discussion with Owen Gower; on the autobiography, see Gower (2008).
a logos is in explanation and causation. His riposte serves (via the clever answer) to
enable there to be a proper explanatory logos of the universal, and restores its claim to
be a principle. A’s restriction of his original claim thus turns on just how explanation
and causation works. The entire discussion between A and B echoes the last argu-
ments of the Phaedo—in reflecting on the conditions for good explanation at the
same time as they offer theories that are designed to do some explaining.
The background of the first conversation, then, is peculiarly Platonic. A is the heir
to the Phaedo, and all of his putative remarks may be derived thence, even if they are
recast in Aristotelian terminology. B, on the other hand, is the heir to the Parmenides,
where variants of the main burden of his arguments can be found.
5. The Second Conversation

The argument seems to restart at 1039a3, ‘again it is clear also as follows’, where what
is clear seems to be the general conclusion just reached ‘that no universal attribute is
ousia; and that no common predicate indicates a this something but rather a such’
(1038b35–1039a2). Now, however, the puzzlement moves from the relation ‘ousia of
something’ to the relation treated hitherto as its converse ‘something from (ek)
ousiai’. When we talk about ousiai being ‘from’ (ek) something, do we mean that
they are actually composed of other ousiai? A general denial of this was part of B’s
reductive view (hence what seemed to be dilemma offered at 1038b23–30); and it was
apparently conceded by A in the face of B’s final salvo. So why labour the point now?
Well, the argument is about to change direction: away from the causal terminology
of the previous argument towards a discussion of the relations of composition. This
may be prompted by earlier talk of the universal being ‘in’ things (1038b18); or of
things being ‘from’ the universal (1038b24). But it takes as its starting point some of
what has been agreed between A and B by the close of the previous conversation.
In particular what follows focuses on the claim that ‘none of the things in the
account . . . exists separately from them nor in anything else; I mean, for example,
that there is no animal besides the particular animals’ (1038b31–3). The argument
that follows proceeds via the denial that the universal is the sort of ousia that exists
separately from its particulars to consider what we might still say about its being in
things.
This—derived from the joint conclusion of A and B—becomes, I shall say, the
position of C. It is in part the outcome of the previous discussion; but it is presented
as making a quite specific claim: that when we say that ousiai are not from other
ousiai, what we mean is that ousiai are not composed of other actual ousiai. C begins
by justifying the claim that no actual ousia is from other actual ousiai, by considering
number:
For it is impossible that ousia can be from ousiai present in it actually; for things that are thus
actually two are never actually one, though if they are potentially two, they can be one (e.g. the
double line consists of two halves, but potentially; for the actualization of the halves separates
them), so that if the ousia is one, it will not be from ousiai present in it. (1039a4–8)
C’s position thus is already determinate in ways distinct from the claims of either
A or B, for his view exploits notions of actuality and potentiality that are new to the
discussion of the universal. What is two in actuality cannot (at the same time, etc.) be
one in actuality, but can only be so potentially. If one is separated into two, that will
actualize the two (and render the one only potential). So if an ousia is one, it will not
be from actual ousiai.
This argument has echoes in the puzzle about one and two in the Phaedo which
provokes the simple-minded answer, and its view of forms as separate. So C has
something in common with A. He shares something with B, too: a reductionist drive,
which is here deployed in terms of a restricted account of metaphysical separation,
brought about by the actualization of two halves of a whole. But the Phaedo’s talk of
causation and explanation has receded in favour of talk about composition: C would
not be satisfied by a version of the simple-minded answer (‘the ones are one by virtue
of the one’). Any causal talk that might remain from the previous exchange is
residual, and relies only on whatever causal impetus we might allow to the actuality,
which does the separation into two (1039a7).
An atomist like Democritus might say this, too—and he might be right, at least
according to C.
For he says one thing cannot come from two nor two from one; for he makes the atomic
magnitudes ousiai. So it is evident, likewise, that this will be so for number, if number is a
synthesis of units, as is said by some; for two is either not one, or there is no unit in it actually.
(1039a9–13)
Here the move in the discussion from causes to composition becomes quite clear.
Democritus’ account of the relation between one and two rests on the thought that
there are elemental entities (atoms) that cannot change, even though they can be held
to make up complex wholes. Atoms are not composed of atoms, even if molecules
are. Democritus is committed to denying that any of his basic entities is such as to be
divided or made greater: and this is a question of their metaphysical nature, not their
causal relations—the problem is not the lack of a nuclear fission machine.
There are others who might say something similar. Consider the exponent of
Socrates’ dream (Theaetetus 201e ff.)45 who would say that the primary entities are
elemental, of which everything else (including numbers, Theaetetus 204b11 ff.) is
composed. The dreamer offers this in the context of an account of knowledge, and an
attempt to explain knowledge by reference to basic elements:
45
There is a literature about who this is, with which I shall not concern myself here, since the
philosophical question, about how the position is developed in the dialectic of the argument, is separate
from the historical question, however important the latter may be.
Soc: Now, my friend, a little while ago, if you remember, we were inclined to accept a certain
proposition which we thought put the matter very well—I mean the statement that no account
can be given of the primaries of which other things are constituted, because each of them is in
itself incomposite (asuntheton); and that it would be incorrect to apply even the term ‘being’ to
it when we spoke of it, or the term ‘this’, because these terms signify different and alien things;
and this is the reason why a primary is an unaccountable (alogon) and unknowable thing.
(Theaetetus 205c)46
C’s antecedents, that is to say, may include this section of the Theaetetus. That
suggestion is supported, moreover, by what happens next in the Theaetetus. For
what Socrates offers here is itself thoroughly dialectical. What seemed a virtue of
this view (the dream theory) at the beginning of Socrates’ discussion (that each of
these primary entities has only either a name or a private account, an oikeios logos,
which are the elements of composition of a definition or some account that could
constitute knowledge) turns out to be a vice. After all—as Socrates goes on to argue—
if the elements of things are unknowable, so are the complexes; if the complexes are
knowable, so must be their constituent parts:47
If anyone maintains that the complex is by nature knowable and the element unknowable, we
shall regard this as tomfoolery, whether it is intended to be or not. (Theaetetus 206b)
The dream theory, if it is supposed to explain knowledge, fails.

Against C, equally, this might read as a fatal objection, because C’s thesis is about
accountability as well as composition. After all, in denying A’s claim to the universal’s
standing as a principle, the view that survives the debate between A and B must be
able to explain as much as A can. And that this tells against C is manifest from the
puzzle that is next brought to bear:
If no ousia can be from universals, because a universal indicates a such not a this,48 and if no
composite (suntheton) ousia can be from actual ousiai, every ousia would be incomposite
(asuntheton), so that there would not even be an account (logos) of any ousia. But it is thought
by all and has been previously stated that definition is either only or primarily of ousia; yet now
it (definition) is not even of that. There will not, then, be definition of anything. Or, rather, in
one way there can be, and in another there cannot. (1039a14–19)
C’s opponent (D) has been reading the Theaetetus: the language of composition and
of the oikeios logos49 is forcibly brought to mind here and exposed. Against the theory
of composition offered by Socrates’ dream—the theory into which C has been led by
his compromise on the previous conversation—D offers the very objection that
46
Trs. M. C. Levett in Burnyeat (1990).
47
This is to ride roughshod over a very difficult and contentious passage; I apologize for doing so, and
refer my reader to the magisterial Burnyeat (1990).
48
This remark recalls the origins of C’s thesis in the rejection of A’s modifications at 1038b25.
49
ºª, 1039a18; although it is not said to be NŒE, that thought would carry over from the thesis of
the first conversation that the ousia of something must be peculiar to it.
Socrates lines up against the dreamer. D is, of course, relying on earlier claims in
Z that one of the conditions of ousia should be that it provides the basis for
knowledge or understanding (1028b1–3); but his objection derives from the argu-
ment of the dream, rather than from ammunition provided by the Phaedo.
I conclude that the parties to conversation 2 have the debate of the Theaetetus in
their background. Undoubtedly, there are other antecedents to both positions (Dem-
ocritus, explicitly). However, in forcibly recalling the language of the dream, Aristotle
recalls also both sides to the dream debate. C and D are both represented in the
Theaetetus, and represented there as in conversation with each other. And this
conversation echoes the impasse of the Platonic antecedent: will there be no defin-
ition of anything? Or can we take comfort from the Aristotelian promissory note: ‘in
one way there can be, and in another there cannot’?
6. The Dialectic of Z. 13
The two conversations of Z. 13 are, I conclude, distinct, involving two different
parties in each case, and a reorientation of the discussion from causation to com-
position. By the conclusion of the chapter no resolution is reached, even if a
compromise is promised. When the chapter closes on ‘and what is said will be
more evident from what follows’ (1039a22–3), what is ‘what is said’? Will what
follows offer some way out of the impasse? The dispute about the universal’s claim to
be ousia falls foul of a general tension of this work: between, on the one hand, the
ontological priority of ousia and, on the other, its role as the basis for definition and
understanding.50 In this context, this tension plays out in a series of different
positions about the nature and role of the universal itself; but none of these positions,
so far, is decisive. Instead, the chapter closes with the promise of further and better
clarity; I suggest that this is not forthcoming until the entire sequence of discussions
about the universal is complete—if then.
Something else, however, is going on here as well. The chapter, as I have argued,
revolves around a series of different positions, the ammunition for which comes from
Plato. Indeed, in the first conversation, both sides to the discussion rely heavily on
material to be found in two different dialogues. The references are dense and,
I submit, unmistakable—both in the language and in the structures of the argument.
Equally, in the second conversation, the two sides to the dispute occupy positions
that are first opposed to each other by Plato himself, in Socrates’ dream. The allusions
are hardly surprising. After all, for both Aristotle and his readership the Platonic
dialogues are deep background; they would readily recognize the complex
intertextuality.
50
This has been at issue since Z. 1, e.g. 1028a33–6.
But there may be more to it than that. I have suggested that the two conversations
of Z. 13 are conceptually distinct (the first occupies itself with questions about
causation and its ontological commitment; the second with the explanation of
composition). They are, moreover, carefully demarcated: the allusions to the Theae-
tetus do not begin until the Phaedo and the Parmenides are effectively done. What is
more, in the changeover between one section of the chapter and the other, as I have
suggested, the personnel change: the two protagonists of the first conversation take
up a different view of the universal from that offered by A and criticized by B at the
beginning of the chapter; but that view too (the view of C) is vulnerable to attack
(by D). A is distinct, then, from C (A insists on the separateness of the universals,
where C is interested in composition); without a great deal more metaphysics their
views could not be made consistent, and certainly neither view implies the other.
B and D, likewise, are distinct: B’s concern is with ontological parsimony, while D is
interested in the accountability demanded by Socrates in his attack on the dream.
Someone might, of course, be interested both in parsimony and in accountability; but
neither flows immediately from the other. The careful demarcations of the chapter,
that is to say, have the role of delineating these four distinct positions. The demar-
cations are underlined by the care with which the intertextuality with the Platonic
dialogues is managed.
So in each conversation a pair of positions is dialectically opposed, and neither
position turns out decisive. On the contrary—the discussions force the reader into
the position of needing an Aristotelian compromise: so in the second conversation, in
one way there can be definition, and in another there cannot. We should avoid,
therefore, supposing that any of these Platonic parties to the debate represent
Aristotle himself. C, for example, offers a view that seems at the outset to be the
outcome of the routing of the Platonizing A; but that view is itself vulnerable to
attack. D, on the other hand—if he concedes the outcome of conversation 1, both
demands and is unable to provide definition. Likewise, although we might easily
suppose that B represents Aristotle, if C shares both his reductive impulses and A’s
interest in principles, he falls foul of D’s closing complaint. To repeat, these two
conversations, Platonically inspired as they are, remain dialectical: they pose the
puzzles, rather than solving them.
Perhaps, then, we should rethink some of the stage directions. Recall the closing
stages of conversation 1:
is ousia; and that no common predicate indicates a this something but rather a such.
Otherwise, many difficulties follow, including the third man. (1038b30–1039a3)
This passage recapitulates the (Platonic) terms of the discussion and appears to
represent something like an agreement: but between whom? It is regularly assumed
that ‘those looking’ are the composite audience of the discussion—Aristotle and his
readers.51 However, the second conversation suggests that whatever agreement there
may be at this stage between A and B, if that issues in the position of C, it is not
simply ‘clear’ in some decisive or factive sense. For C’s position is subsequently
brought under attack in the next conversation. So the clarity at the end of the first
conversation must after all be relative ‘to those who look at the matter’; and those
who look must be the parties to the conversation, rather than the reader. That this is
‘clear’ does not provide us with a definitive conclusion: what we are given, on the
contrary, is a series of puzzles derived from the interplay between the parties to the
conversations.
Indeed, the expression ‘looking at’ (theôrein) is, I suggest, heavily loaded. After all,
theôrein is an important term of art both for Plato and for Aristotle.52 Before we fall
into the trap, moreover, of supposing that this sort of looking is merely staring in
amazement at the ultimate objects, we should notice its deployment in this argu-
mentative context. First, it describes the process of the argument—this is a discursive
kind of looking, ‘from these things’, ek toutôn (1038b34). Second, it describes the
agreement between the two parties: for both A and B must be included in the subject
of the looking. This sort of looking is a cooperative, joint activity. And third, it
describes their stance: this is the reflective looking that comes about through
dialectic—the detached view of those arriving at a conclusion after considering the
to and fro of argument between different positions. By this stage in the whole
complex argument about the universal, therefore, the position of A has been aban-
doned, and he has joined forces with his original opponent. His opponent, equally,
has made progress during the argument: from the blank denial that the universal is
ousia with which he began he has progressed to a more complex position, itself
developed out of what A has said in his own defence. So should we find here
definitive conclusions for the argument about universals as a whole? No: the dialectic
is not over yet; what we are shown here is how it is to be conducted.
At 1039a3, Aristotle seems to add a rider to the previous discussion: ‘This is
evident also from the following . . .’. Here we might think that Aristotle is proceeding
in what is commonly thought of as his usual way—offering one argument and then
another,53 properly to hammer home the point he wishes to make. That is not quite,
however, what happens in the next sequence.
First of all, to whom is it evident? Maybe ‘this is evident’ simply picks up the earlier
remark (‘it is clear to those looking’, 1038b34) so that what is evident is indexed, via
the progress of the preceding conversation, to the participants in the previous
51
Thus Ross’ ‘If, then, we view the matter . . .’ (my emphasis) (1953) (see n. 10). There is here no explicit
first-person plural, however.
52
The expression has connotations, for both Plato and Aristotle, of detached inspection (see
Nightingale (2004)); as I have argued elsewhere, for neither Plato nor Aristotle is it necessarily a solitary
activity (see Chapters 6 and 16). Notice its appearance at Metaphysics 995a34.
53
Marked by the repeated Ø at 1039a3, more heavily loaded than at 1038b15; compare Ø b ŒÆd
1038b23, discussed by Gill (2001).
conversation. Or it may be intended as a broader claim: ‘it is evident’ simpliciter—

that is, to Aristotle and his reader. But in the latter case, it is hard to see just how the
next conversation is supposed to fit; for the next conversation itself renders puzzling
the conclusions of the first. The claim of anything to be evident or clear, at this stage
in the discussion, may be unwarranted; so perhaps it had better be indexed to the
different parties to the debates.
So should we perhaps think of something’s being evident not as factive, but as
relative, evident to someone or from some point of view? An immediate example may
perhaps be forthcoming from within Z. 13: if—as is often suggested—the remarks
that follow the revision of A’s thesis at 1038b16–18 are to be attributed to A, then
what is evident is evident to A (1038b19). If B interrupts immediately, he is non-
committal about whatever is clear; if he does not start speaking again until, for
example, 1038b23, he is arguing that whatever was said to be clear either is or results
in impossibility and absurdity.54 I conclude that clarity need not be factive. Aristotle’s
own position may be undeclared by the end of Z. 13.
The overall dialectical structure carries through to the rest of the chapters about
the universal, Z. 14–16. There are detailed allusions to specific Platonic dialogues;
and these Platonic allusions are markedly specific to particular stretches of conver-
sation. In each case, moreover, the disputes are offered to the reader as still unre-
solved puzzles, with the ebb and flow of the argument pressing now for, now against,
the claims of the universal to be ousia.55 Most of the time, therefore, it is not just very
difficult, but mistaken, to attribute some particular remark to the author of the
Metaphysics. On the contrary, most of the attributions should be to the positions
whose Platonic background in some particular dialogue is so richly described. In
what follows I offer a view of the dialectic of the next conversation, and then a brief
conspectus of the following four.
7. Collection and Division: Conversation 3

That the discussants of the previous chapter are realigned at the beginning of Z. 14 is
suggested by the way in which they are described from the outset:
It is clear from these very things what is the consequence also for those who both say that the
ideas are separate ousiai and make form from the genus and the differentiae. (1039a24–6)
The considerations previously adduced will tell also against this next group. The
conversation that ensues takes place, let us say, between E and F.
54
Compare also a later passage where something clear may be derived from the hypothesis of forms,
which not everyone accepts: 1039a29.
55
This, indeed, is one way of understanding what is promised at 1039a21–3.
In the first section of the chapter (1039a24–33), F invites an elaboration of E’s

position. E holds the conjunction56 of two views: first that the ideas are separate
ousiai; and second that the form is from (ek) the genus and the differentiae.
F reformulates this conjunction as:
If there are forms, and if animal is in man and horse,57 then animal is either the same in
number in each case, or different. (1039a26–8)
Suppose E agrees58 that when animal, for example, occurs in the account of both man
and of horse this is the same genus in each case, so it is one in account
(1039a28–30).59 Further, it is part of the commitment to forms that all of the items
in the account will be separate ousiai: from the species, man, to the genus, animal,
and the differentia, biped (1039a30–3).
But then E is faced with the dilemma anticipated at 1039a28. Either animal60 is one
and the same in number in man and in horse (just as you are one and the same as
yourself, 1039a33–4); or animal is different in number in each thing (1039b7). If
E chooses the first horn, he must explain how something can be one61 and the same,
and still in separate things, without committing himself to saying that some one thing
is also separate from itself (1039b1–2). Moreover, if he thinks about this in terms of
the genus and the differentiae, then:
If, on the one hand, animal shares in biped and in manyped,62 then it will have opposite
properties at the same time, while remaining one and the same individual, which is impossible.
If, on the other hand, it does not do so, what would someone mean by saying that animal is
biped or is footed? (1039b2–5)
Even if E tries various different accounts of the relation between the genus and the
species—is it sharing, so that the genus shares in the differentiae? Is it composition or
mixture, so that the genus is ‘perhaps composed of the differentiae, and in contact
with them, or mixed’ (1039b5–6)?—F briskly avers that all of these are impossible
or absurd (1039b6). This pushes E onto the second horn of the dilemma, as he
concedes that the genus is different in each thing (1039b7). But then, says F, the
things whose ousia is animal will be indefinite (determined, that is to say, by an
indefinite number of genera), and the form animal will thus itself be many. Neither
outcome is tolerable.
56
The claim that Plato’s Parmenides (or a school version of its arguments) lies in the background here
(see e.g. Ross (1953) ad loc.) is belied by the complexity of E’s view, emphasized at 1039a25: the Parmenides
does not treat of forms in terms of genus and differentiae.
57
F’s move (from ‘from’ to ‘in’) owes a good deal to conversation 1.
58
Bº at 1039a29 must be internal to E’s view, since F denies that there are forms.
59
Contrast the view of C, or of Democritus, for whom the elements are one in number, i.e. units or
monads (see 1039a12–13).
60
Both text and interpretation are vexed here; see ad loc. Heinaman (see n. 4), Burnyeat et al. (1979),
Frede/Patzig Metaphysik Z (1988).
61
Following Ross (1953) ad loc.
62
This is the only occurrence of the expression º ı in the Metaphysics; it is treated here as the
contradictory of ‘two-footed’.
The chapter then closes with a sequence of further arguments against the same
position, urging both the risk of the proliferation of ideas (1039b11–14) and the
incoherence of the claim that things are ‘from them’ (1039b14–16), at whatever level
of reality (1039b16–17). This gives F the conclusion that ‘there are not forms as some
say there are’ (1039b18–19). The conclusion itself is thus markedly dialectical, fitting
the way the argument of the whole chapter develops. For it is derived only from the
claims advanced in this context by E, and denies only the theory under those claims
(‘as some say there are’).
These ‘some’, we might say, are those who proceed in the manner of the Eleatic
Stranger in the Sophist and the Politicus.63 The Eleatic Stranger is the exponent of
elaborate schemes of ‘collection and division’ which account for the way things are in
hierarchical structures that bear some resemblance to the taxonomy of species by
genus and differentiae. And the language of Z. 14 recalls the terminology of the
Eleatic Stranger’s divisions: two-footed, many-footed, winged, walking on land, etc.
in a gamut of sometimes haphazard differences.64 So he may be one who makes the
form from the genus and the differentiae. But he does so in the context of a
discussion of ontology: should we say that there is just one thing, or many, just
material entities or ideal ones, too? (Sophist 242–9). In this context, he talks about
ideas (ideai 253d): so he may well be committed to the ideas being in some sense or
another ‘separate’ entities at the same time as they are the items in a collection and
division. The Eleatic Stranger, that is to say, may occupy E’s opening position.
Collection and division is, in the Eleatic Stranger’s view, the way to explain
knowledge and expertise. For it is by understanding these structures of ideas that
we may become knowledgeable:
Surely the person who is able to do that distinguishes sufficiently a single form (idea) spread
out over many distinct and separate (chôris) items; and many forms different from each other
but embraced from without by one single one; and a single form, again, collected together into
one through many wholes, and many separated altogether apart. This is to know how to
distinguish by kinds (kata genos) how things can commune with each other and how they
cannot. (Sophist 253d)65
Aristotle’s argument turns on the very thing that the Stranger is puzzled about here—
the interrelations between forms, and how exactly we might say that they are ‘spread
out’ and still each one. For the Eleatic Stranger, that is to say, the question is about the
relations between forms: and the puzzle is about how forms may be separate from
63
The ES’ method has some affinities, of course, with the method of which Socrates declares himself
enamoured, Phaedrus 266b, Philebus 23d.
64
1039b2–5: Compare Plt. 266b, 266d, e, 276c, e. but º ı at Tim. 92a3; compare Plt. 262, 266e,
Phdr. 265e on just how the divisions should be made, as well as the oddities of the Sophist divisions, e.g.
219e. The insistence that we must make the divisions at the right place is, one might suppose, the correlate
of Aristotle’s supposition that the definitions of things fit natural kinds.
65
My translation, McCabe (2000), 208 ff.
each other and how, in so being, they may retain both their identity and their
explanatory function.66 That the interrelations at the level of forms are the primary
focus of attention in conversation 3 of the Metaphysics is made clear in F’s final
move—that the problems he proposes will be reiterated at the level of perceptible
things (1039a16–17). This is in clear contrast to the previous chapter, where the
puzzles were about the relations between forms and particulars. Likewise, many of
the concerns of the Sophist and its companion dialogues are in clear contrast to the
problems about forms and particulars pressed in the Parmenides. The contrast
between conversations 1 and 2 on the one hand, and conversation 3 on the other,
lies also in their Platonic backgrounds.
What is more, the central puzzle of the Sophist turns on just how the entities we
propose could be interrelated.
Shall we refuse to attach being to motion and rest, nor anything to anything else at all, but posit
them in our accounts as unmixable and say that it is impossible for them to share in each other
in this way? Or shall we collect everything together into the same thing as being capable of
communing with each other? Or some, and some not? (Sophist 251d)
The Eleatic Stranger argues that it is impossible to say that none of the kinds (genôn)
commune, or that they are all mixed together. Instead, if we are to be able to talk at
all67 then there must be some rational (and differential) relations among forms: some
forms must mix, and some must not (252e).
Both verbal and argumentative echoes with conversation 3 are, I suggest, striking.
Argumentatively—both texts are dialectically structured. The Eleatic Stranger argues
against the two extreme possibilities (no communion of kinds; total communion of
kinds) to some kind of compromise. F adopts a similar strategy, pressing the dilemma
centred on the question whether the form of the genus is one in number, or not.
Verbally—consider, for example, the different ways in which the relations between
forms, or genê, or ideai68 are described in both texts: mixture, attachment, sharing,
separation.69 Or consider the way in which the context of the Sophist, a discussion of
ontology, is picked up by the insistent questions about the numerical identity of
forms in Z. 14. This theme is continued through the discussion of the ‘great kinds’
(Sophist 255e ff.). The thought in both texts is this: if forms are to account for our
ontology, that account must be monitored lest it generate either too many entities or
66
That is the effect of the long section discussing the great kinds, 255e ff.
67
This is the running theme at Soph. 243 ff.; compare F’s interest in how we repeat ourselves in speech,
1039a29 with the Eleatic Stranger’s claim that if the forms do not combine at all, we shall be unable to
speak, 252c.
68
YÅ: Sophist 253d and 1039a27; ª Å: Sophist 253b and 1039a26, b11; N ÆØ: Sophist 253d and
1039a25.
69
Mixture: e.g. Sophist 252e and 1039b6; attachment: Sophist 252c and 1039b6; sharing: ÆºÆØ
at Sophist 251d7, åØ at Sophist 256a1, and at 1039b2; separation: Sophist 252c and at 1039b1.
entities that impose a contradiction on our explanations from the start.70 Z. 14,
I suggest, alludes to the Sophist and perhaps also to the Politicus.
What do I mean that allusion to involve? I suggest, in the first place, that it is both
deliberate and obvious: that E’s composite view is developed and constrained just as
the general claims about the koinônia genôn are in the central pages of the Sophist,
and that Z. 14 is written to call attention to this. But this, in turn, brings out the
dialectical features of the exchange: so that we should be wary of attributing to
Aristotle himself the positions that are apparently taken up by F. For the role of
F in this passage is to scrutinize what E is offering—not to offer theses of his own, but
counterarguments to the view of E before him. These counterarguments may be fatal,
so that their negation may be positively affirmed. But the discussion is still very much
limited to the dialectical context.
That context, moreover, is limited to this chapter. First, the position taken by
E and developed over the chapter is neither implied by, nor does it imply, the position
of A in Z. 13. A’s concern is to explain the relation between particulars and their
causes, while E’s concern is at a different level of ontology, about the relation between
one form and another. It is conceivable that someone might make a claim about
causes, even the sort of claim that goes with the clever answer of the Phaedo, without
getting tangled up in further issues about the natural relations between causes (they
might suppose, for example, that this sort of causal account is such that it is not
replicated at the level of the causes themselves). Equally, it is conceivable that such a
claim about causes might not be glossed in terms of genus and differentiae; one might
well suppose, for example, that the relation between equal itself and the equal sticks
or stones is designed to tackle the equality of the sticks and the stones, rather than
attempting to articulate nature at its joints. Conversely, the koinônia genôn might not
tell us anything about causation, at least as the Phaedo may be thought to understand
it—someone interested in this kind of explanation might not be making any com-
mitment at all to whether it explains causation.
Second, the position taken by E is distinct from that taken by either C or D. If E is
interested in the relations between genus and differentiae, he need not construe that
relation as one of composition—indeed, he is clearly undecided on how that relation
works (1039b6). Even if it is a composition relation (as it might be developed from
1039a6), it is not a relation that takes accounts to arise from the combination of
incomposites into something composite. E, that is to say, need not subscribe to any
claim about the foundations of reality made by C or D, or to anything said by either
C or D about how the accounts of reality correspond to its structure; or conversely.
Equally, the counterarguments of F are not to be identified with either the position of
C (who proffers an extreme account of self-identity, namely incompositeness) or of
D (who takes composition to provide accountability). Instead, I suggest, the counter-
arguments of F are ad hominem: arranged specifically against the view of E.
70
Compare 1039b2–3 with the Eleatic Stranger’s insistence on the principles of proper division.
So each of the three conversations of Z. 13–14 involves opposed views about the
relations between things that might be said universally and other things (in some
cases, other things said universally; in other cases, particulars). And in each case
we can understand the dispute as lying, roughly, between a reductive view (B, C, and
F) and an idealist one (A, D, and E). But we should not therefore be tempted to
conflate the three reductive views into a single continuous position, or to suppose
that the idealist views coalesce.
If that is right, then we might draw two different conclusions about the parties to
these debates. First, although the idealist positions can be found in different Platonic
dialogues—and they are, if I am right, expressly alluded to by Aristotle’s choice of
language—they are not the same position; we do not have here a single Platonic view,
but rather a varied collection of views, which can be connected only at the most
general level, as idealist positions. Second, and therefore, we should be equally wary
of adding up the reductive positions into a single view, let alone into a single view
belonging to Aristotle in propria persona. After all, some of each of the reductive
views (B, C, and some part of F) are to be found in the Platonic dialogues, in two cases
(C and F) in the same Platonic dialogues as provide the source of the idealist
counterparts. Aristotle may well agree with some of the reductive strategies we see
outlined here; but he need not suppose that the argument is complete yet, and may
not suppose that merely to prefer the reductive outcome to the idealist one is the way
to resolve what seemed to be puzzling about the universal. The dialectical context,
that is to say, continues to dominate the debate.
8. The Remaining Conversations

The subsequent chapters, I maintain, follow the same methodology: a series of
distinct conversations, each located in some specific Platonic dialogue, where the
idealist view is ranged against a set of reductive considerations. For reasons of space,
I cannot delineate the structure of the arguments in detail: so I restrict myself here to
pointing to the allusions.
Conversation 4 turns on the objects of knowledge, and it provides dense allusions
to the conversation about knowledge, belief, and ignorance in Republic V (compare
1039b32–1040a3 with Republic 477e–478e). As a consequence, the opening of the
conversation (1039b20–7)—which seems to be spoken by Aristotle in propria per-
sona—may simply be the beginning of an encounter between a view from the
Republic and its opponent, both views dependent on agreements reached earlier in
Z. Once again, neither interlocutor need be committed to the views of their prede-
cessors in Z. 13 or Z. 14.
Conversation 5 (1040a8–b4) turns on questions about names and common
terms—a discussion heavily resonant of the Cratylus and parts of the Theaetetus
(201–2). Now the Platonism is explicit, for the argument focuses on whether an idea
can be defined. Here once more, then, the parties to the discussion come into view
(1040a8–9) and remain noticeable throughout (1040a28–30) until the concluding

passage (1040b2–4).
Once again this conversation takes place between a distinct idealist position and its
opponent; and while they have connections and affinities with their earlier counter-
parts, they neither imply nor are implied by them.
Conversation 6 (1040b5–27) concerns the natural world. It offers a dispute
between someone who proposes a strictly materialist account of the nature of
the natural world (elements made into composites by cooking, 1040b9) and someone
else who supposes that explanation in the physical world must refer ultimately to
some kind of principle of unity, for which his opponent’s material causation will not
do. What happens here, I suggest, is that Aristotle turns his attention to the Timaeus
(reinforced by some of the remarks about natural philosophy that occur in Socrates’
autobiography in the Phaedo) and its account of the way the world is constituted. So
conversation 6 is between someone (like Timaeus himself) who takes the compos-
ition of the world to be from actual metaphysical properties (the components of
world-soul: being, sameness, difference, Timaeus 35), in dispute with someone (like
Aristotle himself, perhaps, in biologist mode) who supposes that we must explain the
parts of animals in terms of their potentialities. Once again the focus of attention here
is localized, in exactly the same way as earlier conversations have been, and focused
on a particular Platonic dialogue.
In the final conversation (1040b27–1041a3) we seem to hear the Socrates of the
Parmenides speaking: that forms are separate and one over many. His opponent
offers arguments about just how one over many can be explained that sound much
like the objections of Parmenides in the same dialogue. Parmenides, after all,
famously has reservations about the reductive strategy he espouses (135a–b). Might
that explain the puzzling close of conversation 7?
So even if we hadn’t seen the stars nonetheless, I think, there would be eternal ousiai beyond
the ones we knew; so that now also if we don’t know what [ousiai] there are, nonetheless, there
are perhaps necessarily some. (1040b34–1041a3)
We may hear Aristotle here speaking for himself in the general reductive pressure on
the Platonist; but does he develop a positive and detailed view of his own? By the
close, are we to suppose that he has reached a decisive conclusion?
So that none of the things said universally is ousia nor is there an ousia of ousiai, is evident.
(1041a3–5)
Or does the dialectical context persist even here?
9. Conclusion
To this last question I refrain from offering an answer. But I hope to have shown that
the dialectical context is more pervasive in these chapters than is usually allowed. In
particular, they contain several clearly demarcated conversations between opponents

(sometimes sharing considerable common ground) of which at least one is repre-
sented by the views found in some particular Platonic dialogue. The anti-Platonist
arguments are often negative, and certainly specific to the view of whichever dialogue
is in question. This specificity dominates the dialectical exchanges. In each case,
I have argued, the Platonic views as they are represented here neither imply nor are
implied by the other Platonic views represented in the other conversations. And in
each case the counterarguments are specific to those views; it is not at all obvious that
the counterarguments, considered collectively, add up to a properly demarcated
position, or to a view that we might describe as Aristotle’s position. Instead, what
we have are detailed arguments against the opponents, and illumination along the
way about the nature and demands of ousia, and about the constraints that operate
on our conceptions of the universal.
If this is right, we might say two quite different things about how these chapters
come out this way. The first would be that these puzzles are common fare in the
Academy; and that Aristotle is here recording the debates with which he was familiar,
debates that have their origins (origins that are still distinct from each other) in
various Platonic dialogues. What appear to be verbal allusions to individual Platonic
passages here would then be the residue of their Platonic background; and they
would tell us something about how the arguments of the Academy were received by
Aristotle.
Or we might think—as I incline to do—that these allusions are in fact rather more
resonant (for example, the sequence of thought about legesthai and logos at
1038b15–19, or the puzzles about knowledge, belief, and ignorance, especially at
1040a1–2), and are designed to recall the individual dialogues where those views are
represented. If that is plausible, it tells us a good deal about how Aristotle approached
reading Plato; and how he expected his audience to respond to his reading. In the first
place, the reading is dense, careful, and detailed—and sometimes even literary (as
I suggested for the dense composition of 1039b27 ff.). In the second place, the
separation of the conversations in these chapters attests the view on the part of
their author of the separate, discrete, philosophical credentials of different Platonic
dialogues.
On either account, this Plato is not a single monolithic set of doctrines, but rather a
set of views, loosely or generically similar, but distinct, dialogue by dialogue (across a
broad spectrum of the dialogues: Phaedo, Republic, Sophist, Politicus, Theaetetus,
Parmenides, Cratylus, Timaeus). If the dialectical structure of these arguments is
meant to be taken this way, then it speaks to an assumption on Aristotle’s part that
his audience will pick up the Platonic backgrounds to his arguments, and will be able
to see the care with which they are separated and dealt with individually. This may be
a matter of arguments within the Academy; but those arguments are manifested
here in a peculiarly literary way: for one of the marks, as I have suggested, of the
allusiveness of the conversations lies in their peculiar affinities to the language and
context of particular dialogues. This, I submit, shows us an Aristotle who had been
well trained by his master in how to read, and how to write about what he read.71
71
I am very grateful indeed to Verity Harte, Ursula Coope, Bob Sharples, and Peter Adamson for
comments on various earlier drafts of this chapter, and to the members of the King’s College seminar on
Metaphysics Z and the ICS seminar on Aristotle Reading Plato for discussion. I should like to thank the
Leverhulme Trust for the Major Research Fellowship during the tenure of which I wrote this chapter.
16
With Mirrors or Without?
Self-Perception in Eudemian
Ethics vii. 12
1. Reading Aristotle
Aristotle—so I was brought up to think—does not bear comparison with Plato when
it comes to the style of his philosophical writing. For—so this comparison goes—
Plato writes proper Greek, carefully (even sneakily) composed, while Aristotle’s
works are in note-form, sparsely literary and condensed.1 But my youthful assump-
tions did Aristotle an injustice: not only in respect of his style but also in respect of its
service to the dialectical engagements we find in his work.2 For now, I am interested in
EE 7.12 (possibly a last chapter, rather than a first3); it is an object lesson, I shall say, for
my younger self. Attending to its shape and structure, I shall hope to show in what
follows, makes a difference to how we understand the chapter’s philosophical point.
This chapter is both carefully wrought and thoroughly dialectical. Carefully
wrought: Aristotle makes an important point through puns. The issue is whether
the self-sufficient person will have, or should have, friends. At 1244b12–14 one of the
partners to the debate is claiming both that he should have fewer and fewer friends, as
his self-sufficiency increases, and that he would care nothing for this diminution: the
claim has, that is to say, both an objective and a subjective component. But the point
about ‘caring nothing’, about the subjective aspect, is made with the verb oligôrein—
‘think little of ’—itself cognate with the objective expressions for ‘few’.4 This, I
suggest, is a deliberate piece of word-play, designed to bring out the objective/subjective
1
On this, see Barnes (1995), 7 ff.
2
Many first chapters are notably ornate (e.g. rhet 1.1; de anima 2.1); the ornaments are in service to the
complex dialectical to-and-fro to be found in these chapters. See McCabe (1994b).
3
The text of the EE, including the division of the books, is famously difficult, even after we may have
decided the question of its relation to the EN. I have learned a great deal from Whiting’s detailed and rich
account. I have not the space here for a detailed discussion of the differences between us, but have recorded
some of the focal points in the notes; references to Whiting are to (2012) unless otherwise indicated.
4
The comparative for oligos may be supplied by elassôn: see LSJ s.v. VI. 1; compare the Platonic
association of oligôrein with peri elachistou poieisthai in the context of Socrates’ version of Achilles’ choice:
Apol. 28d, 30a.
WITH MIRRORS OR WITHOUT ? 341
contrast that turns out to be central to the chapter as a whole.5 That suggestion is
supported by a second play, on the question whether the self-sufficient person will be
lacking in friends (met’ endeias), and whether he will miss them (deisthai at
1244b20–1).6 Either the lecture notes were of the crass kind—to include the
jokes—or this chapter is much more carefully composed than my earlier self would
have been disposed to admit.
Second, the dialectical structure. At the opening of EE VII. 2 Aristotle offers an
account of the method to follow:
We must, then, take the account which will best explain the opinions held on these matters and
will resolve the puzzles and contradictions. And this will happen if the opposed opinions are
seen to be held reasonably; such an account will be most in agreement with the phenomena;
and it will come about that, if what is said is true in one sense but untrue in another, both the
opposed opinions will remain. (EE 1235b13–18)7
7.12 follows this method in a marked manner,8 and its structure—complex though it
is—is best understood in those terms. For it offers, I shall argue, a double assault on
its puzzle about friendship—does the self-sufficient man need friends?—by means of
a series of opposed endoxa—where those include both the views of others and the
social or psychological phenomena that need to be explained. And the outcome
provides both an explanation of the endoxa (it shows why they are reasonable in
the first place), a rationale for their opposition (it shows why they are both reasonable
at once), and a resolution of the difficulties they jointly pose (they are each true in one
way, but not in another). In so doing, as is Aristotle’s wont, he provides a much
deeper account of what is going on in friendship—especially, as I shall say, in what is
involved in doing things together, in ‘togetherness’.9 He does so—and this is what
will interest me here—in the context of a distinctive discussion of self-perception and
friendship; on that discussion turn the modifications of the original endoxa, to show
how they might indeed be reasonably held true, at once. The chapter, then, is
dialectical through and through (all the more so, I shall suggest en passant, in
Aristotle’s allusion to a Platonic dialogue, the Symposium).
5
Kosman (2004) suggests that the chapter is designed to draw our attention to the subject’s conscious-
ness, and thence to consciousness shared with friends. Below, }10, I suggest a different account, in which
reflectiveness is more significant than Kosman allows.
6
See Whiting (2006) and (2012) on the text of 1244b9.
7
Translations of the EE are my own; other texts are translated as indicated.
8
Hence the repetition of eulogôs (a term of art at 1235b16?) at 1244b33, 1245a39, 1245b13.
9
If the chapter is, as I claim, carefully wrought, then the use of the sun-prefixes and their explication
throughout the central passage of argument is significant: e.g. the initial puzzle denying the importance of
suzên, 1244b7; the comparison between living together 1244b7 and living alone, 1244b17 (retaining monos
here, see n. 41); the insistent sun-prefixes, 1244b25–6; suzên glossed in terms of something in common
compared to something together, 1245a11–15; and other ‘together’ expressions e.g. 1245a32, 1245a37,
1245b3, 4, 5, 8, 10; 1245b22, 24.
342 WITH MIRRORS OR WITHOUT ?
2. Three Texts about Self-Perception and the Friend

There are in fact three parallel chapters in Aristotle’s ethical works that talk about
the self-sufficient man and his friends, and do so in the context of some kind of
claim about self-perception and self-knowledge: Magna Moralia 2.15, especially
1213a8–27; Nicomachean Ethics 9.9, especially 1170b5–14; and Eudemian Ethics
7.12, especially 1245a29–b4. About these three texts we might think that they all
say pretty much the same thing, and so might conclude that obviously the EN chapter
(which picks up on subtle stuff from the psychological treatise, de anima10) is the
most interesting and sophisticated; or we might think that they do not say the same
thing, but anyway the EN chapter is the most interesting and sophisticated (just
because it picks up on the subtle stuff from de anima). Either conclusion may be too
hasty.
2a With mirrors . . .
I begin with the Magna Moralia:
Let us leave on one side the question of what god contemplates; for we are making a study, not
about god’s self-sufficiency, but about man’s—whether the self-sufficient man will need
friendship or not. Now suppose someone looks at his friend and sees what he is and of what
sort11 [and says to himself] ‘this man is just such as I’—if we imagine a friend of the closest
sort—whence the saying ‘This is another Heracles, my friend is another I.’12
Now knowing oneself is, as some of the wise have said, a very difficult thing; and it is very
pleasant, too (‘to know oneself is pleasant’13). But we are unable14 to contemplate ourselves by
our own resources (that we are unable [sometimes cannot] to do so is made evident from the
things we blame others for while escaping our own notice doing the very same thing. This often
happens through favour or emotion: these things darken good judgement for many of us).
10
On this, see Kosman (1975), (2005); Osborne (1983), (1998); Caston (1998), (2002), (2004), (2005);
Sisko (2004); Johansen (2005); and Chapter 14.
11
There seems to be a lacuna here.
12
The text here is a mess: but however it should read, it seems that here egô = ‘self ’, as also at 1213a11,
1213a24. This passage suggests that the saying had two clauses, the first about Heracles, the second about ‘I’
or the self. In the companion passages in the EE and the EN, then, we might expect some expression for
‘self ’ in the second clause, such as the reflexive autos (pace Whiting (2012)). The saying seems to refer to
some joint enterprise on which Heracles is engaged with a friend, without whom he is just one against
many (cf. Plato, Phaedo 89c); the allusion is perhaps to his being helped in the slaying of the Hydra by
Iolaus (see Plato, Euthydemus 297); see n. 67.
13
Armstrong (1935) construes this as a proverb, to get rid of the repetition.
14
Does ‘we are unable to contemplate’ mean that we cannot ever contemplate ourselves by our own
resources? Or that we are (sometimes, even often) unable to contemplate? The corollary—that we decry in
others what we escape our own notice doing—implies, I think, that these are somehow faults in us, and so a
weakness, not a total incapacity of our natures. So ‘we are unable to contemplate’ means ‘we are sometimes
unable to contemplate’ and ‘knowledge is impossible’ means that this kind of self-knowledge cannot be a
permanent state of mind; compare the point about theôria at EN X. Further issues of modality recur in
Aristotle’s discussions of perception and intellection; I return to this issue below, and see Johansen (2005)
and Chapter 14.
So just as when we want to see our own faces we look into the mirror and see, likewise when
we want to know ourselves, looking into our friend, we know. For the friend is, as we say,
another I. Then if it is pleasant to know oneself, and this knowledge is impossible without
another who is a friend, then the self-sufficient man will need a friend in order to know himself.
(MM 1213a8–27)15
This MM text represents a view that has acquired some notoriety in later philosoph-
ical thought.16 It supposes that the friend is the mirror of the self and that we get
self-knowledge by looking at ourselves in them. Self-knowledge, then, is treated as
self-perception: and this treatment will be the focus of my attention in what follows.
Its notoriety, perhaps, derives from two quite different aspects—the first is the sheer
instrumentality of the way it thinks about friendship (so the first is an ethical matter);
the second is the model of the mind in which seeing ourselves has looking in a mirror
as a suitable analogue (so this is both an epistemological and a metaphysical matter).
I shall begin with the second, and its epistemological dimension: why might it be
problematic to think that my friend is like a mirror, so that I may perceive myself
(and thence come to know myself ) in him? I shall return, at the end, to ethics, and to
metaphysics, too.
There are (at least) two problems from the start in thinking about self-perception.
i) When I (seek to) perceive myself,17 how do I parse this ‘myself ’? Is the self I see
the object of my seeing (so I see myself in just the same way as I see my goldfish: that
the object of my perception is in fact identical to me is somehow an opportune
accident). Or is the self I see seen as the subject of the seeing (so that I recognize that
this is seeing done by me, when I see myself )? Likewise, if I seek to perceive myself in
my friend, is my aim to see myself as an object of perception (that patch of pink, for
example, or that patch of salt-and-pepper brown, or that face) or to see myself as the
subject of perception, as seeing (the seer of that patch of pink, etc.)?
The mirrors may illuminate: if I see myself in a mirror, the self that I see seems to
be the object of my seeing; mirrors tell me less about myself as the perceiving subject.
Suppose, for example, that we sought to understand conscious awareness (not, as
I shall argue, the only construal of my seeing myself seeing); it seems to me we would
be unlikely to stand before a mirror to find it out.
ii) When I perceive myself, however I do that, what is the content of my percep-
tion? Perhaps I see a raw sight: that patch of pink, or this patch of salt-and-pepper
brown. Or perhaps the content is more cooked: I see that this woman is wearing
15
Translation by St. G. Stock in Barnes (1984). I have made minor modifications especially to retain the
prominent visual vocabulary of the original.
16
E.g. Ryle (1949), 186.
17
The same question, mutatis mutandis, for knowing: the case of seeing may be thought the
most problematic in respect of our own assumptions about the private and privileged nature of perception:
see }10.
spectacles; or I see that this woman wearing spectacles is me.18 Perhaps it is even
twice-cooked (or ordered): I see that I am seeing that this woman wearing spectacles
is me.19
Again, the mirrors illuminate the point: if I see myself in the mirror, I may see my
appearance raw on the surface of the mirror (and subsequently work out—by some
other means than perception, perhaps—that this fright is me). Or I may cook it a bit:
I see that the fright in the mirror is me. Can I cook it any more, though, with the
mirror: can I see, in the mirror, that this is me seeing myself in the mirror?
Now suppose that I see myself in my friend as my mirror, à la Magna Moralia.
The model comes from Plato (Alcibiades 133): when I see myself in my friend, I see
the little image of me in her eye.20 What I see, then, is not myself seeing, but the
appearance of me, myself as the object of my seeing; and the image of myself is
the cause of my seeing myself. Then how cooked is my seeing? Rawish, if my seeing
just is receiving the image of the little person in the other’s eye. More cooked, if my
seeing contains more cognitive content (this image is tiny, and I am large: but I see it
as myself ). But can it be cooked twice, any more in the image in my friend’s eye than
in the mirror? Can I see myself as the subject of my gaze?
And anyway might this not—to return now to the ethical issue—be an objection-
able way of coming to know myself? If I see myself mirrored in my friend, my friend
is herself conceived as a mirror, as nothing more than the surface in which I am
myself reflected. So my sense of my friend’s identity, like my sense of her claim on
me, is etiolated: limited to the instrument of my self-perception (and self-knowledge)
that she is here taken to be. And this, we might object (in a Kantian turn, perhaps), is
no way to treat others. Nor—even if we reject the thought that we should treat others
as ends in themselves—is it anything like the right attitude we should have to those
special others, our friends: if they are our mirrors, we use and exploit them—and that
is no way to be their friend.
2b . . . or without: EN 9.9
But neither EE nor EN mention mirrors; nor should they, for the account that both
give of self-perception is much more complex than what we find in the MM. In
particular, where the mirrors of the MM seem limited to seeing myself as an object,
both EN and EE take us to perceive ourselves as perceiving, and so as the subjects of
perception.21 The question of the instrumentality of self-perception I shall leave till
the end.
18
Is my seeing myself de re (the woman I see is in fact me) or de dicto (I see that the woman I see is me)?
The de dicto case, at least, is cooked.
19
I mention here—to ward it off—the suggestion that this twice-cooked seeing is somehow merely
metaphorical, not proper seeing at all—that would depend on what non-metaphorical seeing was con-
ceived to be in the first place. See Chapters 9 and 14.
20
The Greek for ‘pupil’ is the same as the Greek for ‘doll’, ‘girl’ (korê); Alc. 133a, in alluding to this,
brings out the objective features of this account.
21
See Kosman (2004).
Consider, first, EN 9.9.22 It begins, as the MM does, with the problem of the self-
sufficient man: ‘it is disputed, in respect of the happy man, whether he needs friends,
or not.’ So there is a dilemma. On the one hand, ‘they say that there is no need for
friends for the blessed and self-sufficient’ (1169b3–5). And yet, on the other, ‘it is
strange to make the blessed man a solitary; for no-one would choose to have all good
things by himself; for man is a political animal, and naturally fitted to living together’
(1169b16–19). This dilemma is tackled, in the sequel, ‘more in accordance with
nature’ (1170a13) by focusing on the nature of living, determined by thinking or
perceiving. Thence (I quote here at length):
But if 23 living24 is itself good and pleasant (and it seems to be, also from the fact that everyone
desires it, and decent and blessed people most of all, since for them life is most desirable, and
their vital activity is most blessed), and if the one who sees perceives25 that he sees, the one who
hears perceives that he hears, the one who walks perceives that he walks26 and similarly in the
other cases there is something that perceives that we are in activity, so that if we perceive, it
perceives that we perceive, and if we think, it perceives that we think; and if perceiving that we
perceive or think is perceiving that we exist27 (for as we said, existence is perceiving or
thinking); and if perceiving that one lives is pleasant in itself (for living is something naturally
good and perceiving what is good as being there in oneself is pleasant); and if living is desirable,
and especially so for the good, because for them existing is good and pleasant (for perceiving
together28 what is in itself the good, in themselves, gives them pleasure); and if as the good
person is to himself, so he is to his friend (since the friend is another self ),29 then just as for
each his own existence is desirable, so his friend’s is too, or to a similar degree.
22
On this chapter, see especially Cooper (1977a), (1977b), (1990); Irwin (1988), 391 ff.; Price (1989),
103 ff.; Whiting (2006).
23
The sentence structure is bulky: a series of protaseis leads to a conclusion that the friend is as
choiceworthy as oneself.
24
I take the translation of EN from Rowe in Broadie and Rowe (2002), except that I translate zên and
cognates here and elsewhere as ‘living’ (rather than ‘being alive’), to mark out its normativity, and
sunaisthanesthai and cognates as ‘perceiving together’ (rather than as ‘concurrently perceiving’).
25
The modality of this claim matters (see n. 14): does Aristotle mean that the person who sees always
also perceives that he sees? So, indeed, it is taken by Caston (e.g. 2002) and also by Kosman (1975), (2004).
But there are doubts about this in the parallel passages from the de anima—see Johansen (2005) and
Chapter 14.
26
This may come as a surprise after the two preceding examples, of perceptions. And it is not at all
obvious that we always perceive that we walk: I can perambulate absent-mindedly or, thinking fiercely
about something quite different, end up somewhere else completely unawares.
27
The parenthesis makes clear that Aristotle is not here dealing with, for example, diabolical doubt, but
rather attending to our perception of our own active flourishing.
28
The expression here and at 1170a10 is sunaisthanomenoi, ‘perceiving together’. What are the relata of
‘together’? Ross/Urmson have ‘he needs, therefore, to be conscious of the existence of his friend as well’.
I shall ask whether we should allow talk of ‘consciousness’ to overshadow the significance of the prefix sun-
‘together’, in both the EN and the EE (sunaisthanesthai and cognates do not occur in the MM parallel). See
e.g. Kahn (1979b), 22 ff.
29
The expression here is heteros autos (1170b6), making play with the reflexive in the previous line;
likewise at 1166a31 the context is about reflexives, and the expression for ‘another self ’ is allos autos.
Compare and contrast the MM version (see n. 12).
But, as we saw, the good man’s existence is desirable because of his perceiving himself, that
self being good; and such perceiving is pleasant in itself. In that case he needs to be perceiving
together with the friend—that he exists too—and this will come about in their living together,
conversing, and sharing their talk and thoughts; for this is what would seem to be meant by
‘living together’ where human beings are concerned, not feeding in the same location as with
grazing animals. (EN 1170a25–b14)
Aristotle’s argument seems to run like this: suppose that the good man lives by fully
actualizing his capacities, and that perceiving that he is fully actualizing his capacities
is part of the pleasure of his life. Then suppose that the good man’s friend is ‘another
self ’; then the good man will get pleasure from perceiving that his friend is fully
actualizing his capacities too. But the friend’s full actualization depends on30 his
perceiving that he is doing so; so they should do this perceiving ‘together’.
This argument is—to say the least—compressed: in three respects in particular.
First, it does not explain just what the self-perception is that so contributes to the best
life. Is it the contemplation of my full actualization (so that such contemplation adds
to the goodness and to the pleasure, 1170b4–5, of that life)? Is it that full actualization
of my cognitive faculties must be conscious? Or is there any other way in which my
perceiving myself actively perceiving might be a component in the best life for me?
Consequent on this unclarity, second, the chapter does not explain just what work
the conception of the friend as ‘another self ’ is doing here.31 Is the thought that
because my friend is another me, then in perceiving that he sees, I perceive myself ?
This would fit, we might at first think, with the MM account, because it seems to treat
the vision of myself that I get from my friend as the object of my perceiving. Or is the
idea that in perceiving him, I perceive what my perceiving is like (since it is just
analogous to mine)? My perception of my friend would thus reveal myself to myself
as the subject of perceiving. But why would I need a friend for that? Does the thought
that he is another self generate any richer view of his own perceiving than as an
instrument to my own, or any better account of why perceiving together should
matter to my life?
Third, the value of all this self-perception is not immediately transparent. Is it
that self-perception (and perception with my friend) is an adjunct to my best life
(a pleasure that we may enjoy along the way)? Or is the value we get from it an
30
The chapter seems to shift from the thought that living is pleasant and good (1170a26) and thence that
perceiving that one is living is (also) pleasant and good (1170b1) to the thought that existence is worth having
because we perceive it (1170b9) and thence that we must perceive together with the friend (1170b10–11). This
shift (which seems to turn on the role of the friend as ‘another self ’ at 1170b6–7) might perhaps be explained
by its having the (different) argument of EE 7.12 in the background.
31
In EN 9.4, Aristotle talks about the good man’s relations with himself and with his friends. Reference
to himself occurs within the scope of his thoughts (e.g. 1166a13), his wishes (e.g. 1166a23–4), and even,
perhaps, his feelings of pleasure and pain (1166a27). Reflexivity, thence, is part of the content of his
cognitive states, and not just his consciousness of them (pace Kosman (2004)). In the parallel between
himself and his friends, likewise, the reference to his friend occurs within the scope of his own descriptions
of his friendship (1166a29–31), which suggests that he sees his friend as another self (1166a31–2).
essential feature of the best life: ‘the good man’s existence is desirable because of his
perceiving himself, that self being good; and such perceiving is pleasant in itself ’
(1170b8–10)? And then what does my friend contribute? Aristotle’s conclusion is a
strong one: ‘he needs to be perceiving together with the friend’. But is that need a part
of our natural functioning in, or a central pleasure of, the best life, or is it just a means
to the end of living well; and if the latter, is it a means in default of some better?
Perhaps I should take care to perceive my friend, who is another self, for this is nearly
as good as perceiving myself: and perceiving myself is hard, so that I need my friend
to help me do so. The MM makes this assumption explicit; not so, here. Or perhaps
the pleasure I get from perceiving my friend is additional to the pleasure I already get
from perceiving myself anyway. In that case, how are we to explain the strong
modality of the conclusion? Perhaps, instead, my perceiving my friend is somehow
intrinsic to the best life for me: but how?
The EN may take the relation between my friend and me to be one of analogy (‘just
as for each his own existence is desirable, so his friend’s is too, or to a similar degree’
1170b7–8), not causally linked (there is no mirror). On that account, either we
should imagine the two friends realizing their own actualization in tandem (in
which case the ‘other self ’ idea is doing no serious work; and there is no account
given of why I should attend especially to my friend at all32); or the friend is somehow
a part of my own self-realization. This may alter the focus somewhat: I perceive my
friend seeing; I perceive that she is another me; so perceiving her seeing leads me to
perceive myself seeing—so I perceive myself (and her) not as the object, but as the
subject of the seeing. And that, after all, is what we should expect here: for the burden
of Aristotle’s argument weighs on the role of proper functioning in the best life (e.g.
1169b29–70a4, 1170a25–b5). He exploits two claims for which he has already argued:
the first that we should understand the good in terms of active functioning; the
second that we can see this in terms of the full functioning of our cognitive faculties—
in particular, in not only perceiving, but also perceiving that we perceive; not only
knowing, but knowing that we know.33 So perceiving our friend, who is another self,
is just (or nearly) like perceiving ourselves (and so gives us pleasure, etc.). But what
we perceive are our quasi-selves seeing; this is the good that we perceive in them.
What we perceive in them, then, is their (or our) perceiving; so both they and we are
perceived as the subject of seeing: when we perceive that we perceive, we perceive
ourselves as perceiving.34
32
And yet ‘in that case he needs to be perceiving together with the friend—that he exists too’,
1170b10–11, is presented as an inference.
33
The surprising ‘perceiving that we walk’ fits well the connection of this passage to the de anima’s
discussion of the natural working of the different faculties of soul.
34
Whiting argues that this is not enough to explain why such perception should be good or pleasant:
what if I perceive myself doing, or perceiving, something bad? We need, she suggests, some constraint on
the objects of perception for the argument to generate a conclusion about pleasure. This passage, however,
sidesteps that objection by focusing on the intrinsic goodness of proper functioning (hence the ‘more in
accordance with nature’ point) so that what I perceive, when I perceive myself perceiving, is already a good
Now self-perception, and perception in our friends, is twice-cooked: for an

important thing is to see that we see. Rightly so, for among the antecedents of this
passage of the EN are, surely, not only the de anima discussion of self-perception, but
also one of its antecedents, the Charmides.35 In both Charmides and de anima the
paradigm of self-perception is perceiving oneself as perceiving, at least in the latter
text because we thus perceive ourselves as properly functioning. If this last feature is
central, then what we perceive, when we perceive that we perceive, is ourselves as
perceivers. The de anima makes it clear that this sort of perception is complex and
twice-cooked: for it includes both my perception of myself, perceiving, and the
content of what I perceive (see 425b13 ff.).
What, then, is it thus to perceive myself perceiving? Is Aristotle talking about self-
conscious awareness, or even just consciousness?36 Do we need a friend for that? If
not, then what is he talking about? And whatever that is, is it any way to treat a
friend?
2c The context of the ‘other self ’: the puzzles of EE 7.12
The argument of the EE, I shall suggest, is different again; and so differs, in both its
account of the nature of self-perception and its account of its value, from both the
MM and the EN, at least as the latter is most readily construed. The chapter begins
with the challenge to the phenomena of friendship: that the self-sufficient man, like
god, won’t need anything at all, so friends for him will be redundant, even actively
discouraged. In what follows, the challenge is met both directly and indirectly, with
the result that its power is both recognized (as the programme of 7.2 requires) and
diverted by a discussion of proper functioning. This chapter has affinities with EN
9.9, therefore; but it seems markedly different in effect. This is the result of this
chapter’s focus, as I shall argue, on man’s failure to be self-sufficient. As a conse-
quence it ends up, first, with a surprisingly different take on natural teleology; and
second, with a different account of the ethical structure of friendship.37 This may
encourage us to believe that the two chapters are quite different; or else to conclude
that EN 9.9 should be amplified by what we learn from EE 7.12.
EE 7.12 has a careful dialectical structure. It starts with a puzzle, about the
supremely self-sufficient person and friendship (1244b1–21); then connects the
thing, irrespective, at least as far as this passage goes, of the objects of the first-order perception. I shall suggest
below that the same is true of the parallel EE passage.
35
With the Theaetetus in the background, too; see Chapter 14.
36
As Kosman (1975), (2004) and Whiting (2012) both suppose, although Whiting has a more inclusive
account of what might be involved in ‘self-awareness’. In what follows I shall take ‘consciousness’ to focus
especially on the ‘what it is like for me’ to perceive (see Nagel (1979)); this is distinct from the kind of
reflexive awareness described by apperception; see Johansen (2005). It is distinct again, I shall claim, from
the reflectiveness upon which we engage with our friends, on Aristotle’s account.
37
I leave to one side whether this tells us anything about the relative dating of EE and EN; I confess to
thinking that the MM is weakly beside the point, and derived from different concerns in Plato than either
EN or EE.
answer to this puzzle with self-perception and how best to live (1244b21–1245a10).
Next the puzzle is reformulated in different terms (1245a11–26) and finally resolved
by thinking about the friend as ‘another self ’, and the connection between friendship
and self-perception (1245a26–b9); this eliminates the difficulties and explains the
phenomena (1245b9–46a25).38
The chapter is shaped, thus, around the two versions of the puzzle about virtue and
friendship: but they are versions that are distinct.39 The first version (1244b1–24)
opposes two views about friendship and the supremely virtuous man (1244b4–5). On
one view, he has neither need nor desire for friends—for he is like god (1244b1–15).
On the other view, the virtuous man40 does have friendships, but only of the
‘virtue’ sort;41 and because he is a virtuous man, he gets it right with whom to live
(1244b15–21).
The second version of the puzzle begins in the same way as the first, with the self-
sufficient man:
To choose to live together would seem to those looking at it to be in some way simple-minded
(first in the case of the things common also to the other animals, such as eating together or
drinking together42—for what difference does it make whether these come about for people
who are close to each other, or apart, if you were to take away the power of speech? But then the
sharing in casual speech is another such;43 and at the same time it is not possible for friends
who are self-sufficient either to teach or to learn; for if he learns he is not as he should be; and if
he teaches, his friend is not; but likeness is friendship). (EE 1245a11–18)
The puzzle is formulated, again, as a dilemma; and the first limb replicates the first
limb of the first version at 1244b2 ff. For both focus on the ‘godlike man’—the person
who is ‘in every respect self-sufficient’ (1244b3), even to the extent of having nothing
38
I follow the text of Walzer/Mingay’s OCT except where otherwise noted.
39
Whiting (ad loc.) acknowledges that there is a new move at 1245a11, but takes the run of the
argument to be made throughout in the context of the parallel with god, so that the chapter ends with the
comparison with god. On the account I offer there are two versions of the puzzle: one version invokes
the comparison with god, the other an account of natural function, leading up to the ‘other Heracles’. The
two puzzles are connected, but they could be—and here are—treated separately.
40
The person who needs nothing (1244b17); the first part of the chapter focuses on someone fully
virtuous, not on someone who is nearly so.
41
The text is vexed. Walzer/Mingay give: all’ ou di’aretên philos monon. Susemihl gives alla di’aretên
philos monos. Rackham reads all’ho di’aretên monos, and translates ‘the only real friend is one loved on
account of goodness’. Solomon reads ho di’aretên monos, and translates ‘the friend through excellence is
the only friend’. If we allow that the chapter is carefully written, this phrase comes in the second limb of the
dilemma, introduced by alla mên kai and an appeal to the phenomena at 1244b15. The present phrase,
then, is the counter to 1244b7, houtos gar autôi hikanos suneinai: in which case the careful word order
should contrast hikanos suneinai with monos (the MS reading). Perhaps, then, the sentence would read best
as all’ou di’aretên philos monos: with the last word used in its predicative sense, ‘alone’. Hence ‘the virtue-
friend is not alone’: that is, virtue-friends live together. This rather dramatic formula flags, I think, the
central theme of the chapter—what it is to live together, or alone.
42 43
Cf. EN 1170b12–13. I.e. casual speech is as unimportant as feeding together.
left to learn (1245a16). Here, however, the direct comparison with god is missing;
and the second limb seems to have a different tone.
But indeed it [living together] seems44 so [sc. to be a good thing] and we all take greater
pleasure in good things when we share them with friends, insofar as is possible for each and of
as much of the good as he is capable; but among these one man is capable of sharing in bodily
pleasure, another in artistic contemplation, another in philosophy.45 And (for this) it is
necessary to be together with the friend (hence the saying ‘far friends are a burden’) so that
they should not be away from each other while this is happening. This is why love seems to be
like friendship—for the lover wants to live together,46 although not as he really should, but for
perception.47(EE 1245a18–26)
For whereas the second limb of the first puzzle simply alluded to the mutual
pleasure of virtuous friends (1244b17–19), the second limb of the second puzzle is
cagier about the self-sufficiency of anyone.48 Instead, Aristotle presses the thoughts
that mutual pleasure occurs ‘insofar as is possible for each and with as much of the
good as he is capable of ’ (1245a20–1); that these friendships work in the context of
different sorts of pleasure (bodily, artistic, philosophical, as well as the physical
pleasure of erotic friendships 1245a21–6); and that for all these pleasures being
together is necessary (1245a23–4).
So there are two versions of the puzzle about the self-sufficient man, differing in
their second limbs. In the first, the virtuous man is merely good at getting hold of the
right kind of friends; they seem to be a pleasant but not necessary adjunct of his life.
The second version, however, rejects the absolute conceptions of the first, supposing,
instead, that we are talking about being as good as possible.49 This generates a
discussion of what is involved in self-fulfilment—doing things together—which
seems to suggest that we need our friends there with us, even for philosophy. In
this context it has a more everyday air, so that it urgently needs resolving. The pair of
puzzles is bracketed by warnings about how the comparison with god may mislead us
(1244b21–3, 1245b12–19); and they are, we might think, counterpoised50 so that we
see that quite a lot has changed between them. What is it that has changed?
44
This, like the second limb of the first version of the puzzle, presents the phenomena.
45
Taking the point of the previous paragraph, these sorts of sharing need to be important, not trivial
like feeding at the same trough or sharing in casual speech.
46
Whiting here follows the MSS in reading eu zên, ‘living well’, and rejects Casaubon’s emendation to
suzên. But this seems to me to miss the point of the introduction of erôs, which is here taken both to
confirm the living together of friends, and somehow to spoil it (since love, unlike friendship, is too reliant
on sensation). Without suzên, I think, the connectives (‘this is why’ and ‘for’) lose their point.
47
aesthêsis: ‘sensation’? Is the point here that the lover gets sensuous enjoyment, not the real benefits of
love? Then aesthêsis for ‘sensation’ is a pun of some kind, perhaps, contrasting this point (lovers fall for a
pretty face) with the chapter’s overall interest in perception, aesthêsis.
48
This is anticipated at 1244b11, where Aristotle adds a caveat about whether it is possible for anyone to
be self-sufficient.
49
Whiting (ad loc.) argues that this requires us to take what I call the second version of the puzzle as a
point secondary to the absolute concerns of the first.
50
E.g. the logos/ergon pair at 1245a28 picks up the logos/pragma pair at 1244b31–2.
3. The First Puzzle: Living Together

Returning to the first puzzle, we may see that it focuses—perhaps surprisingly—on
what it is like to live with someone: either the self-sufficient man is sufficient to live
with himself,51 or the virtuous man is good at choosing his friends to live with him.52
This question of living together, I suggest, is central throughout the chapter.53
Tackling the puzzle, Aristotle warns against the analogy with god (1244b23), and
demands that we get a grip, instead,54 on ‘what it is to live in actuality (kat’energeian)
and as an end’. If to live is to perceive and to know,55 he infers, so to live together is to
perceive together56 and to know together (1244b24–6). This looks as though it is
somehow or other obvious: but what on earth does it mean?
Aristotle seems to begin57 his explanation with self-perception and self-knowledge:58
‘For oneself to know and for oneself to perceive are the most choiceworthy for each
person; and that is why the desire for living is natural to everyone’ (1244b26–8). So
self-knowledge and self-perception are somehow the basis for a natural desire to live a
51
This is emphatic at 1244b7: and compare the repetition of suzên at 1244b20. On Whiting’s account
this is one of a series of references to the Philebus: that there are such references seems right (and does not
preclude the Symposium allusion I find here), but in my view the focus of attention is not so much on
pleasure, as on what is involved in ‘together’.
52
The text of 1244b20–1 is problematic; but the sense seems clear.
53
See nn. 9, 41. Note the repetition of the verbs with sun- prefixes in the sections of the chapter that
offer the resolution: 1245a37, 1245b3, 4, 5, 8, 10, 22, 24. Reading monos with the MSS at 1244b17, as
parallel to monôtês at EN 1169b16, 1170a5, and ou rather than ho, to give the proleptic: ‘but nor will [the]
virtue-friend be solitary. For . . .’, the puzzle offers a contrast between doing things together and the
virtuous man living on his own.
54
So there is a contrast, at 1244b22–3, between what escapes our notice when we think about god and
what is clear, when we think about actuality and ends. This contrast, on my view, is the source of the second
version of the aporia.
55
This is explained in the next sequence of argument; then the question about living together is revisited
at 1245a10.
56
sunaisthanesthai: here the sequence of the argument demands that this mean ‘perceiving together’;
see nn. 24, 28.
57
The text, once again, is corrupt; see the detailed discussion by Whiting, ad loc. At 1244b26–7 should
we read auto to aisthanesthai kai auto to gnôrizein (as Rackham has it: ‘perception and knowledge
themselves are the thing most desirable . . .’); to hautou aisthanesthai kai to hauton gnôrizein (Bonitz,
followed by Susemihl and Walzer/Mingay, which supports Solomon’s ‘self-perception and self-knowledge
is most desirable to every one’); to auton aisthanesthai kai to auton gnôrizein (Kosman and Whiting ‘what
is desirable for each person is that he himself perceives and that he himself knows’)? The pronouns,
however we read the disputed text, are prominent; and at least some of the time are reflexive (arguably at
lines 26, 27, and 30; surely at 32 and 33; and again, whatever the argument, at 1245a4–5 and 10). The issue
is repeatedly whether these reflexives refer to the subject of the perceiving (I myself perceive, as Kosman
supposes), to its object (I perceive myself ), or both (I perceive myself perceiving something or other, where
the subject is part of the complex content of the perception, as I shall suggest below). At the very least
the argument that follows shows that we must be talking about the subject, since it turns on who is
doing the living in question (see Kosman (2004)). In the context of the chapter’s earlier play with the
contrast between the subjective and the objective, I suggest that we should see this run of pronouns as
deliberate elaboration and focus; the complex nature of Aristotle’s conclusion, if I have it right, explains
why the text has become fraught.
58
If we read the text printed by Susemihl and Walzer/Mingay, the explanation for this follows, 1244b28
ff., introduced by gar; and the conclusion at 1244b33–4 is expressed as persuasive, eulogôs.
life (1244b29–1245a11),59 both because this marks the knowledge or the perception as
somehow mine (not impersonal nor alienable: 1244b33) and because this makes me
somehow share in the excellent character of determinacy. For to wish to perceive
oneself is to wish oneself determinate in this desirable way: by participation (kata
metalêpsin) in the capacities of perceiving and knowing:
For perceiving one becomes perceptible60 in the same way and in the same respect as he first
perceives, and likewise knowing one becomes knowable.61 And for this reason one also wants
to live always,62 because one wants to know always, and this is that he himself should be the
knowable thing. (EE 1245a7–11)
This argument has a strong teleological cast: to become perceivable in this way—that
is, by perceiving—is something we naturally aim towards, part of our choice of life.63
Becoming perceivable like this, that is, is not something that happens just by virtue of
our exercising our capacity to perceive, for if it did, there would be no significance in
my aiming at being perceived over and above my aiming to perceive. And yet we
become perceptible as soon as we perceive. Why? And why is that so important? To
this question I shall return.
But still, what does this have to do with friendship? By now the trajectory of the
chapter has become somewhat unclear: Aristotle had seemed to be engaged on an
account of why we should live together (1244b25–6), but now he focuses again on
our self-sufficiency. After all, if we aim at self-knowledge and self-perception, and we
can become the objects of perception by perceiving, then what need do we have of
anyone else to achieve our ends? Thus the puzzle of the self-sufficient man and his
friends arises again, for a new reason, one developed from the nature of man himself,
rather than from the comparison to god.64 So Aristotle moves on to reprise the
59
The argument from the table of opposites is presented as about the nature of things (1245a1, 3), about
regularity (1245a1, 9–10), and about capacities or powers (1245a6). This makes clear that we are talking of
a natural desire to have a life, not merely to stay alive. If the chapter turns on an account of character and its
structure, then the table of opposites fits easily, since it is about how various aspects of character go
together: see e.g. 1245a20–6. Whiting (2012) connects the table of opposites to Metaphysics 12.9, and
argues that it is here that the chapter switches from considering the subjects of perception and knowledge
to the objects.
60
aisthêtos can mean ‘perceptible’ or ‘perceived’. In what follows I take it to mark a possibility, compare
de an. e.g. 425b26: if what is aisthêtos has an actualization, then what is actualized must be just possible. It is
hard to construe ‘perceptible’ here, or ‘knowable’ at 1245a10, without some of the connotations of the
object of perception; but if perception is cooked, its objects will include itself.
61
In the parallel chapter in the EN, the higher-order faculty seems to be perception throughout,
1170a29 ff.
62
Or, ‘always wants to live’: aei may be janus-faced here (compare Heraclitus DK22B1).
63
The emphasis on pleasure throughout, and the allusions to the Philebus on which Whiting focuses,
marks this teleology: functioning well has its own accompanying pleasures. I maintain, however, that the
significance of pleasure is only as a symptom of proper function in this chapter, which aims at under-
standing joint function, rather than its attendant pleasure. This, of course, fits the dialectical structure,
since the pleasure of friendship counts among the phenomena.
64
Hence in the lead-up to what I call the second version of the puzzle, at 1245a11, the language of phusis
is prominent, 1245a1, 3, anticipated, I take it, in the turn to energeia and the end at 1244b23; this is why
I take there to be a single puzzle in this chapter, but in two versions.
account of the self-sufficient man in the first limb of the second version of the puzzle:
(1245a11–18) if what I should be aiming at is self-perception and so on, then is living
together a matter of indifference, or even folly?65 Or is there some other account to be
given of the phenomena that we all take greater pleasure in good things when we
engage in them with friends, some other explanation of living together (1245a18–26),
as the second limb demands?
4. Perception and the Friend

Aristotle starts his unravelling of the puzzle (‘we must consider the truth from this
point’, 1245a27) after the second version is complete, with a fresh formulation of the
relation between friends (1245a29–35), before returning to the question of percep-
tion (1245a35–b9). He concludes the chapter by explaining, seriatim, all the opinions
that went to make up the puzzles, and showing how the puzzles no longer do any
harm (1245b9–1246a25).
Friends, first:
For the friend wishes to be,66 as the proverb says, another Heracles,67 another self. But he is
torn apart68 and it is hard for them to come to be in one.69 But although by nature the friend is
65
The hieratic tone of 1244b26–1245a10 encourages the comparison with god, while the second version
of the puzzle becomes more earth-bound; see 1245a13–14.
66
Is this intentional (the friend really has this in mind to do) or essential (this is what it is for someone
to be a friend)? In what follows I shall suggest that Aristotle trades on the ambivalence of expressions like
this.
67
See also MM 1213a12 (and n. 12) which suggests that the proverb includes both clauses, ‘another
Heracles, another self ’; this tells against Whiting’s suggestion that we should read allos houtos. Aristotle’s
addition of Heracles to the ‘other self ’ tag adds several things. First (see Irwin (1988), 395 ff.) it specifies an
individual, so that the ‘other self ’ has in fact a determinate identity (which is not mine). Second, the
scenario is one of a joint collaboration, as when Heracles was helped in slaying the Hydra by his nephew
Iolaus (Euthydemus 297c1 ff.). Seeking ‘another Heracles’ imagines a situation where two are joined in a
single purpose, and where both helper and helped are of Herculean stature—hence ‘another’: there should
be two of them engaged on the task at once (we may be misled here by a common English idiom of
comparison; ‘Tony Blair was another Margaret Thatcher’ does not imply that the two prime ministers
actually collaborated). Third, as Kosman makes clear ((2004), 147, citing Plutarch Lives 29.3), the
collaboration is a result, or a feature, of the relationship between the two Heracles; in some versions, the
relation is a familial one, but that too would be a case of philia. That, in turn, still needs explanation.
68
This ‘tearing apart’ has already figured at 1240b 28–30: ‘The one who is absolutely good seeks also to
be a friend to himself, because he has two things in him which want to be friends by nature, and which are
impossible to tear apart.’ (I am grateful to Julia Annas for pressing consideration of this passage on me).
Notice that in the earlier chapter what cannot be torn apart are the two parts of a single soul; the same
image seems to be in play here, where what is torn apart is the very thing that finds coming together as
one difficult. In both cases, the model is best understood as a whole with parts: in 7.6, a whole whose parts
are inseparable (for here we are talking about self-love); in 7.12, parts that are difficult to keep as a whole
(here we are talking about love between friends). I shall return to this below in accounting for the sun-
terminology so important in this chapter.
69
The text seems to be a mess; and the suggestions for fixing it not much better. Walzer/Mingay and
Rackham read chalepon panta eph’henos genesthai, and Rackham translates ‘it is difficult for all to be
realized in the case of one person’ (Solomon offers something similar). Susemihl has chalepon ta eph’henos
genesthai. If, as I incline to think, this passage both alludes backwards to the discussions of moral
what is most akin, one is like another in body, another in soul, and one in respect of one part of
body or soul, another of another. But nonetheless the friend wishes to be70 as it were a
separated71 self. (EE 1245a29–35)
Two elements of Aristotle’s terminology here should give us pause: that the other self
is a separated self; and that this separation explains and modifies our natural
inclination to friendship.72
Therefore to perceive the friend must be somehow to perceive oneself, and somehow to know
oneself. (EE 1245a35–37, my italics)
But what exactly is involved in that ‘somehow’ (here I come to the object of my
exercise)? How is this self-perception done? And how does it treat my friend? It
explains, apparently, all sorts of pleasure:
. . . even to enjoy vulgar pleasures together and to live together with a friend is pleasant
(because73 his74 perception always comes about at the same time) and all the more so with
the more divine pleasures—the reason for this is that it is always more pleasant to contemplate
oneself in the better good. (EE 1245a37–b1)
psychology in 7.6, especially 1240a14–21 and 1240b15–21 (see previous note), and to the discussions of the
nature of love in the Symposium (see section 5), then the point here must be supposed to be the converse of
what is said in 7.6 in the same language. There the thought was that the good man has a well-integrated
soul—unlike the way in which the bad man is somehow many (1240b16); here the thought is that friends
are actually torn asunder, and that it is difficult for them to come together (as the single-souled good man
would do). Actually to become a single-souled man is impossible, of course; but a loose formula here would
express the aspiration to become one (compare Plato, Protagoras 358b where the aspiration is a painless
life) against the facts of being separate or ‘pulled apart’.
70
Again, the ambivalence of ‘wishes to be’ is noticeable (see n. 66): is the claim that the friend is by
nature a separated self, or that the friend actually wants to be a separated self? If there is a Symposium
background here, of course, that would not be quite right, since friends don’t want to be separated, but to
coalesce. On my account, Aristotle has a less fanciful account of the togetherness of friendship than
Aristophanes.
71
Separated/divided; or separable/divisible: diairetos; compare the account of one individual’s soul in
7.6, and see n. 60 on the ambivalence of similar expressions; here I take the problem to be that the friend is
indeed separated, but if there is an allusion to Aristophanes in the Symposium here, the ambivalence will be
explained, since it is exactly the problem of the whole-natured creatures that they are separable.
72
Aristotle is at pains to deny that the pleasure in friendship is explanatory of how we should live;
instead, against the opponent in the second version of the puzzle, he insists that the living together happens
to generate the pleasure, rather than the pleasure explaining the living together; hence the hôste of result at
1245a37. The dialectical structure is prominent: eulogôs at 1245a38; and again, to deal with the first limb, at
1245b13, and again at 1245b37, and 1246a13.
73
The explanatory gar at 1245a38 picks up the result clause at 1245a37.
74
The MS have ekeinou, so of the friend; Walzer/Mingay take Robinson’s suggestion of hautou = of
oneself (reading back from the supposed conclusion at 1245a1). But either way the point is unclear: does
Aristotle mean that the perception is of my friend (or myself ), or that it is done by my friend (or myself )?
Whiting takes the latter way, and retains the MS reading; and this seems right to me, especially in light of
what is to come: so at this stage, what Aristotle is interested in is the subject of perception. The sequel then
insists that when we enjoy pleasures together we contemplate ourselves as perceiving; and this is to see
oneself ‘in the better good’. The switch between the subjective and the objective is repeatedly the source of
unclarity: when I perceive my friend perceiving, he, perceiving, is the object of my perception; but this, as
I shall suggest below, is an essential feature of Aristotle’s proposal about friendship and self-perception.
This explains the phenomena, namely the pleasures we enjoy, by amplifying how
perception—of oneself or of one’s friend—is supposed to be a good thing. For, if
perception is itself a good thing to do, then seeing one’s friend as perceiving is seeing
him in a good state. Then, if that goes along with some kind of seeing of oneself
(perceiving) then one sees oneself in a good state, and the better, the better. Self-
perception (or contemplation) is a good thing, part of our end, because when we do
that we see ourselves doing well (functioning well); and this gives us pleasure.
Likewise, then, when we live with a friend, another self, we perceive them doing
well, and this is the source of the pleasure of friendship. But is this anything other
than exploiting them?
Why should one contemplate oneself (perceiving)?
. . . This (being in the better good)75 is sometimes an affection, sometimes an action, sometimes
something else. If it is that76 one lives well, and thus also that the friend lives well, and that they
are active together in the living together, their joint enterprise is most of all among the things
included in the end. That is why we should77 contemplate78 together and feast together—but
not in the pleasures of eating and necessary pleasures (for those associations seem to be, not
communions, but sensuous enjoyment)79—but the end which each person is able to attain, in
this end he wishes80 to live with another. If not, people choose most of all to do well and to
suffer81 well at the hands of their friends. (EE 1245b2–9)
75
What? The contemplating oneself or the being in the better good? (Rackham translates ‘and this is
sometimes a passive, sometimes an active experience, sometimes something else’ (my italics).) The former
would allow the business of contemplation to be something we can do better or worse; and can improve at
doing. The latter seems more appropriate given what happens next (so Whiting), fits in with the underlying
theme that perceiving itself—which one may contemplate—is a good, and might allow us to explain the
infinitives at 1245b4 ff.
76
This might be indirect speech, suggesting that the object of contemplation is oneself living well and
one’s friend living well too. Or it could mean that the good state (‘this’ from the previous sentence) is living
well, etc. So the ‘things included in the end’ at 1245b4 are the central point, and the contemplating together
and feasting together turn out to be what the joint enterprise is. Whiting’s version takes the run of
infinitives at 1245b4–6 to depend on ‘it is always more pleasant’; this seems to me to load the chapter in
favour of a discussion of pleasure where the focus in fact is on good functioning and the end.
77
Walzer/Mingay take Fritzsche’s addition of dei; Whiting translates ‘it is more pleasant to contemplate
together . . .’.
78
More below on what is involved in contemplation, but the present passage does not require us to
understand it as exclusively intellectual; it seems to include the possibility of contemplating oneself in
perceptual mode.
79
apolauseis: picks up, and revises, what we had in the first version of the puzzle, at 1244b18.
80
Again the expression ‘wishes’ may describe a natural state, or an intentional one; see nn. 60, 71.
81
This suggests that the best joint enterprise, contemplating together in a proper communion, is in fact
the ‘something else’ of b2, because the conclusion picks up the second-best versions in what we do in the
case of second-best outcomes: do and suffer. What would that mean? I think Aristotle’s point—if indeed it
is about contemplation throughout this passage—is that there are ways in which we can contemplate
ourselves, and be contemplated (be the objects of our friends’ theoretical gaze, to emphasize the parallel
with vision), which are worth pursuing, but are not the best possible way of contemplating together. On
this account of the passage, the best contemplating includes oneself as its subject, while in the lesser
versions either we are the object of someone else’s contemplation or perception (we suffer it) or they are the
object of ours (we do it).
Either affection or action or something else: what something else is that?82 If our
living well is the end, and so is the friend’s doing so, then living well together is
actualizing together. Failing the fulfilment of this end (1245b8), we choose, in fact, to
do things and to suffer things at the hands of our friends.
Aristotle offers, therefore, two different models of our activity with our friends.
The second, and secondary, model is the reciprocal treatment—the ‘doing to them’
and ‘being done to by them’—we engage in when we fail to engage in the first, in true
joint activity. That joint activity, the primary model, is neither ‘doing to’ nor
‘suffering at the hands of ’, but actualizing together: and that involves, at least,
contemplating ourselves as active: so perceiving ourselves and our friends as
perceivers.
But even ‘actualizing together’ is itself a concession, something we need to do
because in fact our natural capacities are diminished in comparison to the self-
sufficiency of god. For notice the way in which this section of the chapter follows
on from the second limb of the second version of the puzzle. In neat counterpoint to
the way in which the first puzzle was followed by a detailed account of what it would
be to be entirely self-sufficient, the treatment of the second version of the puzzle
presses hard on the thought that our friendships are determined by our capacities
(1245a20–1), by the difficult way things are by nature (1245a31–4), and, notwith-
standing the difficulties,83 by our natural inclinations and ends (1245a34–5). By now,
that is to say, Aristotle is focusing his attention on the way in which our natural ends
are compromised, because we are not gods (even Heracles is a demi-god, and
sometimes needs help). The resulting explanation of our relations to our friends,
therefore, derives from how we think about self-sufficiency and our failure to reach it.
Since it takes a friend to complete us (because the best natural characteristics are
scattered about, because the other self is separated), we actualize together as part of
living in the best way possible. And this actualizing together, it seems, is somehow an
activity of contemplation or perception: an activity that includes perceiving ourselves
as perceiving.
Consider, thus, the resumptive section of the chapter, before Aristotle closes by
providing accounts of the phenomena:
So it is evident both that we should live together and that everyone especially wishes to do so,84
and that the happiest and best person is most of all someone who is such [sc. as to live together
and to wish to do so]. But that it was not evident according to the argument is also a reasonable
conclusion to reach, since the argument said something true. For the putting together of the
82
Compare 1244b18–19, where the first version of the puzzle does not consider this ‘something else’:
work has been done on how we understand friendship and self-perception between the first puzzle and
here.
83
The qualification at 1245a34 emphasizes that these are cases where we are not dealing with the ideal.
84
Again, this is ambiguous between the objective and the subjective.
two things85 provides a solution,86 although the case of god was truly described.87 For that god
is not such as to need a friend, we think should apply also to the comparandum. But by this
argument, the excellent man won’t even think; for that is not how god is best, but in being
better than to think of anything except88 himself. The reason for this is that our doing well is by
virtue of another, whereas for him doing well is of himself. (EE 1245b9–19)
The opening of this paragraph insists on the imperative conclusion: we should live
together. It requires, that is to say, stronger grounds than merely that we happen to
derive pleasure from these associations. What is more, this conclusion is based on
facts about human nature: this is humanity’s natural tendency, as well as its inclin-
ation, and thence a matter of character: the best person is such as to live together with
another. Aristotle’s point, therefore, is not that as a matter of fact the virtuous person
can accommodate friends in his life; but rather that the commonality of friends is an
intrinsic part of the best person’s nature. The argument has shifted in focus, there-
fore, from the first version of the puzzle to Aristotle’s response to the second; and that
is now his point. The trouble was caused, he suggests, by the comparison with god; by
the assumption of the first version of the puzzle that man is godlike, rather than
human. But for humans, by contrast, ‘doing well is by virtue of another’.
So during the working out of the puzzles in the dialectical sequence of the chapter,
Aristotle has modified his account of what it is to be a virtuous person. In the first
puzzle, he treated the virtuous person as the human equivalent of god, as someone
solitary and self-sufficient, whose ends are best met by his remaining alone. But the
intervening passages rethought this conception, and allowed us to see that the person
under consideration is someone who is not really like god at all. Instead this is
someone who just operates as best he can, within the limitations of his humanity (this
just is what his humanity is). He has, as I have suggested, ends from which he must
necessarily fall short, even though he still does the best he can. This contrast between
the two figures on Aristotle’s stage deals with the inconcinnities of the two versions of
the puzzle: in the first, man was just like god, in the second he crucially falls short.
This contrast explains why god may be alone, but we are not: instead, in the best case,
we actualize together with our friends. Actualizing alone is god’s nature, but it is
beyond us, whose nature it is—according to this argument—to actualize in commu-
nion. This actualizing in communion is a matter not of one party using the other to
their own ends, or even of mutual exploitation to the ends of both, but an account of
what human nature is, frail as it is: such as to actualize in communion.
85
This refers back to 1244b34, as Whiting observes, ad loc., and the two things there seem to be the
choosability of living well, and the good for the agent of such a life.
86
I follow Whiting’s preference for the MS here. Reading hê lusis ouk estin, with Walzer/Mingay, the
passage needs to say something about where the solution fails; it is not clear just how that would work at
this stage in the discussion.
87
Here the sequence of thought fits well with my suggestion, above, that there are two versions of the
aporia in the chapter.
88
Or, on a different text, ‘anything alien rather than’.
This kind of actualizing must include, it seems, perceiving ourselves as perceiving.

But unlike god, we are not mere self-perceivers, or merely engaged in reciprocal self-
perception with our friends. How exactly, then, does this joint actuality answer to the
chapter’s earlier insistence on self-knowledge and self-perception? And how does it
explain (or is it explained by) the claim that the friend is another self?
5. Failing to be God
It is a striking feature of this chapter that Aristotle’s language at this point, as he talks
about the friend as ‘another self ’ in the context of our failure to be god, recalls89
another account of the natural condition of love—that given by Aristophanes (with
several nods to Empedocles90) in his speech in the Symposium (189–93).91 The
friend, the other self, ‘is torn apart92 and it is hard for them to come to be in
one’93 (1245a31). In Aristophanes’ story, the whole-natured creatures, who were
self-sufficient and arrogant, aspired to be gods. But Zeus was furious at their hubris:
and to teach them a lesson, he cut them in half. As a consequence, they were
separated from their proper halves—and spent their lives trying to be reunited with
them—in vain. Aristophanes’ account rested on a view about natural identity; he
supposed that our present natural condition, that is to say, is of a kind of failed or
deficient identity, something we may be always seeking to improve or complete; and
he suggests that this is to be understood in terms of what it is not—self-sufficiency.
For man—on Aristophanes’ account—spectacularly fails to be like god; and this is
what explains his nature, his desires, and his attachments. It explains his aspirations,
too: becoming self-sufficient, becoming whole, is the way to be blessed and happy,
impossible though that may turn out to be.
89
Aristophanes’ speech, like Aristotle’s arguments here, is about love and friendship and happiness and
being blessed; see 193d5. Aristotle is a careful reader of Plato, as I have argued in Chapters 14 and 15, giving
clear allusions to Platonic texts and arguments, but not attributing them directly. This passage, I suggest, is
another such; and even if the allusion is looser than direct (as several people have suggested to me), the
parallel is still, I think, instructive, especially in giving us a model for what joint perception would be, in the
context of questions about identity. Compare also EN 1166a34–5.
90
E.g. DK31B17.
91
See Stern-Gillet (1995), 123.
92
See especially Symp. 190–2. Although Plato does not use the expression diaspastai, Aristophanes’
speech is littered with the two prefixes dia- and sun- (e.g. at 191e9), emphasizing the separation and
coming together that, on the account I give, is the focus of Aristotle’s attention here. Aristotle uses diaspaô
infrequently, usually of forcible tearing apart: e.g. De caelo 313b20; Meteor. 372b19; Rhet. 1386a10; Pol.
1303b13; and see below on the earlier passage in EE 7.6. The vivid language and its connotations of force
bring to mind the forcible splitting of the whole-natured creatures in Aristophanes’ account of love. Both
Aristotle and Aristophanes are giving an account of human nature in these terms; see Symposium 189d5–7,
191d1–3, 193c5.
93
With the Symposium in mind, where the whole-natures are chopped up, we might think that this
expression is striking: compare ‘making one from two’ at Symp. 191d2; and the Empedoclean version, e.g.
DK31B17.
In this passage of the EE, I suggest, Aristotle is exploiting something like Aris-
tophanes’ account: hence ‘by nature the friend is what is the most akin’94 (1245a32);
‘nonetheless the friend wishes to be as it were a separated self ’ (1245a34–5); and
hence the sudden remark about love, erôs, at 1245a24. In both accounts, the friend or
the lover is understood in terms of the self—in terms, that is to say, of our failure to
be a single self, a failure to have a complete identity. That failure, further, is
understood in terms of natural deficiency, and of a failure to be god, a failure to be
self-sufficient and complete. For to be complete, we need our friend, our other,
separated self; only then will we be able to actualize—when we actualize together.
Aristotle, like Aristophanes, undermines the possibility of an individual’s reaching
self-sufficiency. Instead, the best we can do is to get together with another self; and
our aiming at this explains how we behave. Aristotle’s discussion of friendship here
reflects our real feelings of inadequacy and our vain searching to compensate for it.
It is this conception of identity and its failures, therefore, that underlies this
account of what it is to be a friend and this account of the lives we share: an account
that is built on a series of thoughts about nature and teleology, about the essence of
man and his aspirations. It is, therefore, no coincidence that Aristotle’s description of
the friend is put in terms that embrace both expressions of desire and intention (the
friend ‘wants to be . . .’) and claims of essence (a friend ‘really is’ a separated self ): the
combination of the two shows up how Aristotle appeals to a teleological account that
rests not on the perfectibility of man, but on his imperfection.
The discussion of the idea of a divided self is anticipated in an earlier chapter of the
EE, 7.6, which tackles the question of self-love.95 Should we say that self-love is the
paradigm of friendship, or an aberrant or metaphorical version of it? After all, ‘being
loved (being befriended) and loving (befriending) are in two separate things’
(1240a14–15).96 As a consequence, Aristotle concludes, the model for self-love
must be like the model for continence and incontinence97 where we must suppose
that the soul is somehow divided: insofar as the soul can be divided, there can be self-
love; insofar as not, then not (1240a20–1).
This account of the divided self, presented as it is in the context (in 7.12) of our
failure to be god—presented, indeed, in the course of a chapter whose dialectical
strategy is to show just how we should concede that we fail to be gods—echoes the
Aristophanic account of love (of philia, friendship, as well as erôs, 1245a24), even
94
Compare the use of oikeios in Symp.: e.g. at 192c1, 193d2.
95
I am grateful to Julia Annas for insisting on this point, against my stronger claim that Aristotle is
deliberately invoking Aristophanes’ speech.
96
Solomon has ‘loving and being loved require two separate individuals’; Rackham ‘being loved and
loving involve two separate factors’. The expression for separate, diêirêmenois, at 1240a14 is cognate with
diairetos at 1245a35. Compare Physics 254b31, for the same expression in respect of the movement of living
organisms, and 258a5 ff. for the amplification of the point in terms of (something like parts and) wholes.
See Furley’s classic (1978), and his reservation that Aristotle is here not specific about this being a division
into parts, 166: the same reservation applies at EE 1240a14.
97
Compare e.g. EN 1102b13 ff., EE 1224b21 ff.
down to the way in which Aristophanes’ story invokes the comparison with god in
the hubristic challenge of the whole-natured creatures. Indeed, it seems plausible to
suppose that some such account, if not this very account, is being alluded to by
Aristotle here: a supposition that is reinforced by the careful composition of the
chapter, down to the level of its literary complexity.98 Even if the connection is looser
than a direct allusion, the model that Aristophanes provides for us can offer a useful
commentary on just what is going on in EE 7.12, when Aristotle tells us about the
‘other self ’.
6. Self-Perception in the EE: What Mirrors?

Aristotle’s allusion in 7.6 to the problem of explaining incontinence reflects a broader
problem of understanding any kind of reflexive conception: self-love, self-perception,
self-motion. For we should not say that something moves and is moved by itself in
the same respects at the same time; instead, if something seems to be a self-mover,
then part of it moves, and part is moved (even if the part that moves is then moved,
accidentally). In cases such as these, that is to say, where something seems to be agent
and patient at once, he suggests dividing, and supposing that the agent is one part, the
patient another.99 The self that loves, on this account, is in some sense divided from
the self that is loved; but considered together, they count as the same self, the proper
possessor of self-love. Likewise, we might think, self-perception is of one part of the
self by another; each part belonging, nonetheless, to the same self, which thus
perceives itself. The language throughout 7.12 of both division and togetherness
reinforces the thought that this is the burden of ‘perceiving together’, of sunaisthêsis.
If this is what we should say about self-perception, how is it accounted for by my
relationship with my friend, where he is another self? What is the relation between
our perceiving together, my perceiving myself, and my perceiving myself as perceiv-
ing (all of which are involved in my end)? Aristotle takes himself to be entitled, in the
closing sections of the chapter, to the claim that ‘if it is possible to live with and share
the perceptions of many at the same time, it is most desirable that these [sc. friends]
should be as many as possible; but since this is most difficult, the activity of joint
perception must exist among fewer’ (1245b20–4). And he reaches this point by
means of the claim that friendship as he has described it will generate some kind
of self-perception (and so be a good thing). But what exactly does that involve?
Think, first, about two things it is not:
• It is not an objective account: the issue is not that I see my reflection in my
friend’s eyes. There are no mirrors here; this account of self-perception is far
98
Compare Whiting (2012) on allusions to the Philebus here.
99
If the language is not always of parts (see Furley (1978), n. 99), it nonetheless allows for a conclusion
that can be read, as at EE 7.12, as the separation of two countable individuals, which can come together
somehow into a joint enterprise. On this issue see the classic discussion at Plato Republic 435e ff.
from the MM. For throughout EE 7.12 Aristotle insists that what is important is
my perceiving; and in that case, perceiving that I perceive should not be meant as
my merely becoming the object of my own perception (self-perception, on this
account, cannot be raw, as first-order perception is often construed).
• It is not, either, the sort of subjective account that suggests that self-perception
(or joint perception) is the natural and immediate accompaniment of first-order
perceiving. So the issue is not that I am self-conscious when I see.100 For self-
perception to be this kind of immediate consciousness is at odds with the
teleological cast of the whole account, which insists that this sort of becoming
perceptible is something we should aim at (that is, it is both the object of my
deliberate striving and of my natural tendencies) and something that, without a
friend, we may well miss. (We had better not, then, read ‘perceptible’ at 1245a8
as some kind of regularly activated natural potentiality.)
Instead, therefore, we should think of this as a genuine possibility: as soon as
I perceive I can be perceived in the fashion in question, but not necessarily am I so
perceived. Now, if we are to be perceived as perceivers, as the subjects of the
perception, then it follows immediately that when we perceive we become perceptible
as such: and this explains the detail of ‘perceptible in the same way and in the same
respect [sc. as the perceiving]’ (1245a7–8). So becoming perceptible and then
becoming perceived (if we succeed in our aims) is correlated to our perceiving: and
likewise for knowing: the self that becomes perceptible is the self that perceives, as
such: what happens, somehow or other, is that we perceive ourselves as perceiving—
twice-cooked.101
How then does what I do with my friend convert into self-perception (or self-
contemplation)? Well, if I have the trajectory of 1245a38–b9 right, it does not do so
in simple terms, where the friend is a mirror, the object of my perception (so that, at
one remove, I too am the object of my perception, just because he is another self ); or
yet where he is the subject (where he perceives me, so that I am the subject of my
perceiving at one remove—this is a harder case to imagine102). Instead Aristotle
emphasizes the fact that it is an actualization that is realized by the composite of
myself-and-my-friend. I aim to perceive myself as perceiving; but if my nature is
incomplete, I cannot do this on my own (or not, at least, if this kind of self-perception
is normative, difficult, not mere self-awareness). Instead, I need my friend to com-
plete the thing: it is, somehow, a joint enterprise, on which I and my friend, my
Heracles, are engaged together.
100
Nor is this the account of self-perception that many find in the de anima; but see Chapter 14.
101
So by the time we get to the close of this section of the chapter, at 1245a10, Aristotle has filled out the
programme announced at 1244b23: that the question will be answered if we attend to living and the
actuality.
102
Is this sort of reciprocal action the second-best model, the action and passion described at 1245b1–9?
7. Team Spirit
How does this differ from the account of self-perception with my friend offered in EN
9.9? Both accounts suggest, first, that it is an activity (so I perceive myself as
perceiving) and second, an activity that is hard to achieve (so not mere conscious-
ness).103 But the EN suggests, we might think, that the virtuous man does this with
his friend by analogy with himself (1170b5–6); and suggests, therefore, that the
fulfilment of this natural capacity is well within our grasp. I may find it hard to
perceive myself perceiving; but by seeing my friend in the same condition, I may
come to see it for myself. There is something of the mirror here, no doubt: because
the argument rests on the thought that this aspect of fulfilment, the fulfilment of self-
perception, comes about just because I see my analogue in another.104 The perceiving
of myself that I may eventually come to do, that is to say, comes about because of the
parallel case I see in my friend. He is not my mirror, but he is my Doppelgänger; by
looking at his activity I see my own.
In EN 9.9, then, it seems that ‘myself perceiving’ is the content of my perception,
just as ‘myself walking’ or ‘myself existing’. The second-order perception, that is to
say, has as its content at least the first-order activity. But is that all there is to it? Is the
point of self-perception that it is merely reflective, merely able to give me some
reference to myself in the perceiving I do with my friend? Is the content of this sort of
perception exhausted by myself as perceiving—at least on the EE account?
The EE, by contrast with the EN, presents a far bleaker world, where we are
incapacitated by nature to perceive ourselves, since we ourselves are incomplete—
we are not, as Aristophanes had pointed out, gods (to make this point is the effect of
the complicated dialectical structure of the chapter). Instead, to fulfil our natural
capacity of perceiving ourselves, we must do so in concert with a friend (this is the
‘activity of seeing together’, 1245b24, see 1245b3–4). That is the point of Aristotle’s
insistence that this actualization is primary to the secondary business of reciprocal
action and passion, at 1245b2; and the qualification that this joint activity is somehow
perceiving myself, somehow knowing myself (1245a36–7). This has a far more radical
appearance than the EN: the activity is fulfilled by being done together. So there is no
mirroring at all; instead this is a genuinely joint activity. As a consequence, my friend
and I are genuine collaborators: we do—whatever we do—together, and neither uses
the other as a mirror, or a likeness, either.
But what then is involved in this kind of joint activity? I might be able to talk about
rowing a boat together, and imagine that for it to work we each have to do our bit, for
the team. I might say the same, even, for a joint activity of knowing or learning;105
103
If the EN refers across to the de anima, then we had better not think that the latter is just about
consciousness, either.
104
I have some reservations about this, however; perhaps instead EN 9.9 needs to be reread in the light
of EE 7.12.
105
Notice the significance of this in the account of the less than perfect man, 1245a17–8.
that, surely, is what we are doing now—since I certainly have no privilege except by
virtue of having written what you are reading. But could we say this for perceiving?
How could we think that you and I might perceive together? Has Aristotle just been
misled by Aristophanes—whose whole creatures could, indeed, share a perceptual
faculty? We, who are irrevocably parted, surely cannot.
In (nearly) closing, I offer some thoughts to combat your incredulity here. Our
immediate assumption that we cannot perceive things together—except, perhaps, in
some etiolated or metaphorical way—relies, perhaps, on perception’s being, at its
base, private and privileged (that is why it is often thought to be raw). But if
perception is a bit more cooked than raw, it may not be inner, or private, in quite
the same way (contrast the perception of this patch of puce with the perception that
this music is loud). What, then, if it is twice-cooked? If self-perception merely stands
in for the business of my awareness of my perceptual states, we might still deny that it
could be joint. But suppose self-perception is more reflective, more something we do
as a difficult actualization of a cognitive faculty: what then?
Suppose, that is, that self-perception includes my perception of myself as perceiv-
ing, but is not exhausted by it. Instead, perhaps, its content would include my first-
order perceptions, too; and would bring them into the purview of my reflection. Why
could not this reflective sort of perception, which would include its own second-order
features, be something I could do with someone else?
If we are perceiving something together, and reflecting perceptually at the same
time on what we are perceiving, the process of perceiving may be rich in content, and
productive in terms of the development of our natural capacities.106 For example,
I may see that grey wagtail over there just because we have practised bird-recognition
on our ornithological expeditions; and my doing so is itself a part of our joint
reflective perception. I may get better at playing tennis by practising doubles with
you, and acquiring an improved sense of where the ball is relative to my racquet and
yours by seeing it repeatedly coming right at us over the net, and by perceiving that
I am seeing it that way. I may enjoy music alone by reflective perception of what
I hear, and have heard; all the more so when we listen together, and I think of us both
as perceiving the same cadence, as our appreciation of music, how we hear it,
develops over years of listening together.107 In ethical cases, too, my sense of moral
perception may be enhanced by our seeing the situation together; and by doing so
over time as our friendship matures. Why should we not be able to think of a rich
perceptual life together, just as we might have a rich shared intellectual life? And
when we do, if that life is reflective enough, it will focus our attention both on what
106
It may include, e.g., good judgement in choosing our friends (1244b19–20); or teaching and learning
together (which the self-sufficient person does not need, 1245a16ff.). These examples demand the kind of
rich reflection involved in the development of our other ethical capacities; on this see Burnyeat (1980),
McDowell (1998), Goldie (2004). For a slightly different approach, see Sherman (1993).
107
Notice the joint enterprise of music and philosophy at 1245a22.
we see and hear, and on who we are who do so, and who our companions are who
share or refract our point of view. What is more, it is this very reflectiveness that
explains our progress—in ornithology, in tennis, in musical appreciation, and in
doing the right thing. Is it the mischief of sceptical arguments108 that prevent us from
seeing Aristotle’s point in the EE—that we can have a genuinely shared life of the eye
and the ear?109
8. ‘Another Self, Another Heracles’

This, I suggest, is what Aristotle has in mind in EE VII. 12 by the togetherness that
comes from friendship: and it is this that he takes to explain the phenomena (that we
both should, and desire to, live together, 1245b9–11), the puzzle (that if god neither
needs nor desires a friend, the godlike person should be the same, 1245b12–15), and
the importance of friendship to our development of the best life.
In all three of the passages I have discussed, my friend is ‘another self ’. The
locution is both striking and unclear, and thus brings out the fact that the reflexives
are the focus of Aristotle’s attention. But ‘another self ’ means, I suggest, quite
different things in the three contexts I have discussed.
• In the MM the friend is another self because he is my mirror: looking at him,
I see myself back again. There is, we might think, only one self here: me. My
friend is another self only in the sense that I am reflected in him. This makes
clear just how instrumental is the MM account: there is only one self to be the
focus of attention, the ‘otherness’ of my friend is because he isn’t quite me.
I exploit him to see myself (and if he fails to mirror me, presumably, he should
be thrown away).
• In the EN the picture is a different one, because my friend is imagined to be, like
me, the subject of his functioning senses: it is as such a subject that in him I see
myself seeing. The fulfilment I get from him may then be from his full func-
tioning; so he is another self, just like me. In this sense, even if my enjoyment of
him is egocentric (and this is reasonably disputed), it is not exploitative.
• In the EE the picture seems different again; and the ‘other self ’ expression is, if
I am right about it, odder. For here the joint functioning of friendship is the way
to fulfil my function; so that—as Aristophanes would have construed it—the self
is a composite entity, made up of the two of us, engaged on the joint enterprise
of self-perception and self-knowledge. I do not exploit him; nor is my
108
Kosman (2004) takes this point, but does not, I think, suppose that the development of skills or
values is involved in these reflexive activities.
109
It is a connected mistake to suppose that all such rich perception is in fact a combination of raw feels
with something else that is cognitive but non-perceptual. This mistake is often attributed to Plato, but we
should be more wary, I think; see Chapter 9.
functioning along with him egocentric from my own point of view; instead, we
function together for our joint benefit.
Does this last just beg the question against the challenge of self-sufficiency? I think
not: for it was set up by the discussions about the ‘other self ’ at 1245a30–4. There the
modification of our self-sufficiency was argued in terms of the respect in which we
are akin to others. It is this kinship that explains our natural purposes together; and
the failure of our natures to stand alone makes kinship central. The point would be
lost without the long account, throughout the chapter, of the nature and significance
of self-perception to our functioning; of the reasons why we should not think of
ourselves as functionally self-sufficient; and of the way in which intercourse with a
friend can thence be understood not in terms of the pleasure it brings (which is, if
anything, merely concomitant) but in terms of the way that togetherness contributes
to our reaching our natural ends, actively engaged in our best possible functions. In
the EE, then, our friends are our companions, not our instruments. Kant might be
pleased.110
110
Antecedents of this chapter were given at the Keeling Conference and at the B Club in Cambridge.
I am most grateful to the audiences on both occasions for their constructive comments, and especially to
my Keeling commentator, Julia Annas, and to my colleagues Peter Adamson and David Galloway. Jennifer
Whiting was extremely generous in discussion of this difficult chapter of the EE, both before and during the
conference, for which I am very grateful. I should like to record, also, a more personal debt, to Jennifer and
to Bob Heinaman for their kindness and support during the week of the conference. My gratitude and
appreciation are also owed to the Leverhulme Trust for the Major Research Fellowship during the tenure of
which I wrote this chapter. Thanks also to Fiona Leigh both for her skill as an editor and for her patient
care in bringing this volume to fruition. I would like, finally, to record a lasting debt to Peter Goldie, for his
friendship and for wonderful conversation over many years about moral perception, both sadly curtailed
far too early in October 2011.
Bibliography
Annas, J. (1980) ‘Truth and Knowledge’, in M. Schofield, M. F. Burnyeat, J. Barnes, eds., Doubt
and Dogmatism (Oxford: Clarendon) 84–104.
Annas, J. (1981) An Introduction to Plato’s Republic (Oxford: Oxford University Press).
Annas, J. (1982) ‘Plato’s Myths of Judgement’, Phronesis 27: 119–43.
Annas, J. (1993) The Morality of Happiness (Oxford: Oxford University Press).
Annas, J. (1994) ‘Virtue as the Use of other Goods’, in T. Irwin and M. Nussbaum, eds., Virtue,
Love and Form (Edmonton: Academic) 53–66.
Annas, J. (1999) Platonic Ethics, Old and New (Ithaca: Cornell University Press).
Annas, J. and Rowe, C. (2002) eds. New Perspectives on Plato: Modern and Ancient (Cam-
bridge, MA: Harvard University Press).
Anscombe, G. E. M. (1958) ‘Modern Moral Philosophy’, Philosophy 33: 1–19.
Armstrong, C. J. (1935) Aristotle: Eudemian Ethics (Harvard: Loeb Classical Texts).
Badham, C. (1865) Platonis Euthydemus et Laches (London: Williams and Norgate).
Bailey, D. (2006) ‘Knowledge in Plato’s Charmides’ (unpublished).
Barker, A. (1981) ‘Aristotle on Perception and Ratios’, Phronesis 26: 248–66.
Barnes, J. (1979) The Presocratic Philosophers (London: Routledge).
Barnes, J. (1981) ‘Aristotle and the Methods of Ethics’, Revue Internationale de Philosophy 34:
490–511.
Barnes, J. (1984) ed. The Complete Works of Aristotle (Oxford: Oxford University Press).
Barnes, J. (1995) ed. The Cambridge Companion to Aristotle (Cambridge: Cambridge Univer-
sity Press).
Barnes, J., Schofield, M., Sorabji, R. (1979) eds. Articles on Aristotle IV: Psychology and
Aesthetics (London: Duckworth).
Bollack, J. and Wismann, H. (1972) Héraclite ou la separation (Paris: Les éditions de minuit).
Bostock, D. (1986) Plato’s Phaedo (Oxford: Oxford University Press).
Brandwood, L. (1976) A Word Index to Plato (Leeds: Maney).
Brentano, F. (1973) Psychology from an Empirical Standpoint, trs. A. C. Rancurello,
D. B. Terrell, L. McAlister (London: Routledge).
Broadie, S. (1991) Ethics with Aristotle (New York: Oxford University Press).
Broadie, S. (2005) ‘On the Idea of the Summum Bonum’, in C. Gill, ed., Virtue Norms and
Objectivity (Oxford: Oxford University Press) 41–58.
Broadie, S. and Rowe, C. (2002) Aristotle: Nicomachean Ethics: Translation, Introduction and
Commentary (Oxford: Oxford University Press).
Brunschwig, J. (1973) ‘Sur quelques emplois d’opsis’, in Zetesis, Festschrift for E. J. De Strycker
(Antwerp: Utrecht) 24–39.
Brunschwig, J. (1986) ‘The Cradle Argument in Epicureanism and Stoicism’, in M. Schofield
and G. Striker, eds., The Norms of Nature (Cambridge: Cambridge University Press) 113–44.
368 BIBLIOGRAPHY OF WORKS CITED
Brunschwig, J. (1999) ‘Un détail négligé dans la caverne de Plato’, in L. Bové, ed., La recta ratio:
criticiste et spinoziste: mélanges en l’honneur de Bernard Rousset (Paris: Presses Paris
Sorbonne) 65–76.
Brunschwig, J. (2003) ‘Revisiting Plato’s Cave’, in J. J. Cleary and G. M. Gurtler, eds.,
Proceedings of the Boston Area Colloquium in Ancient Philosophy (Leiden: Brill) 19: 145–77.
Burnyeat, M. F. (1976a) ‘Protagoras and Self-refutation in Plato’s Theaetetus’, Philosophical
Review 85: 172–95.
Burnyeat, M. F. (1976b) ‘Protagoras and Self-refutation in Later Greek Philosophy’, Philo-
sophical Review 85: 44–69.
Burnyeat, M. F. (1977) ‘Examples in Epistemology: Socrates, Theaetetus and G. E. Moore’,
Philosophy 52: 381–96.
Burnyeat, M. F. (1980) ‘Aristotle on learning to be good’, in A. Rorty, ed., Essays on Aristotle’s
Ethics (Berkeley: University of California Press) 69–92.
Burnyeat, M. F. (1981) ‘Aristotle on Understanding Knowledge’, in E. Berti, ed., Aristotle on
Science (Padua: Antenore) 97–139.
Burnyeat, M. F. (1982a) ‘Idealism and Greek Philosophy: What Descartes Saw and Berkeley
Missed’, Philosophical Review 91: 3–40.
Burnyeat, M. F. (1982b) ‘Reading Heraclitus’, New York Review of Books, May 13, 1982.
Burnyeat, M. F. (1990) The Theaetetus of Plato with translation by M. J. Levett (Indianapolis:
Hackett Publishing).
Burnyeat, M. F. (1992) ‘Is an Aristotelian Philosophy of Mind still credible? A Draft’ in
M. C. Nussbaum and A. Rorty, eds., Essays on Aristotle’s Ethics (Oxford: Oxford University
Press) 15–26.
Burnyeat, M. F. (1997) Culture and Society in Plato’s Republic (The Tanner Lectures, Harvard
University).
Burnyeat, M. F. (2000) ‘Plato on Why Mathematics is Good for the Soul’, in T. J. Smiley, ed.,
Mathematics and Necessity: Essays in the History of Philosophy (Oxford: Oxford University
Press) 1–82.
Burnyeat, M. F. (2002) ‘De Anima II. 5’, Phronesis 47: 28–90.
Burnyeat, M. F. (2002) ‘Plato on How to Speak What is Not: Euthydemus 283a–288a’, in
M. Canto-Sperber and P. Pellegrin, eds., Le style de la pensée: receuil de textes en homage à
Jacques Brunschwig (Paris: Les belles lettres) 40–66.
Burnyeat, M. F. (2003) ‘Apology 30B2-4: Socrates, Money and the Grammar of Gignesthai’,
Journal of Hellenic Studies 123: 1–25.
Burnyeat, M. F. et al. (1979) Notes on Zeta (Oxford: Sub-faculty of Philosophy).
Caizzi, F. D. (1996) Platone Eutidemo (Milan: Mondadori Bruno Scolastica).
Canto, M. (1987) L’intrigue philosophique: essai sur l’Euthydème de Platon (Paris: Les belles
lettres).
Caston, V. (1998) ‘Aristotle and the Problem of Intentionality’, Philosophy and Phenomeno-
logical Research 58: 249–98.
Caston, V. (2002) ‘Aristotle on Consciousness’, Mind 111: 751–815.
Caston, V. (2004) ‘More on Aristotle on Consciousness: Reply to Sisko’, Mind 113: 523–33.
Caston, V. (2005) ‘The Spirit and the Letter: Aristotle on Perception’ in R. Salles, ed.,
Metaphysics, Soul and Ethics in Ancient Thought (Oxford: Oxford University Press) 245–320.
BIBLIOGRAPHY OF WORKS CITED 369
Cooper, J. M. (1970) ‘Plato on Sense-perception and Knowledge (Theaetetus 184–6)’, Phronesis

15: 123–65.
Cooper, J. M. (1977a) ‘Aristotle on the Forms of Friendship’, Review of Metaphysics 130:
619–48 and in Cooper (1999).
Cooper, J. M. (1977b) ‘Friendship and the Good in Aristotle’, Philosophical Review 86: 290–315
and in Cooper (1999).
Cooper, J. M. (1990) ‘Political Animals and Civic Friendship’, in G. Patzig, ed., Aristoteles
Politik (Göttingen: Vandenhoeck and Ruprecht) and in Cooper (1999).
Cooper, J. M. (1996) ‘Eudaimonism, the Appeal to Nature, and “Moral Duty” in Stoicism’, in
S. Engstrom and J. Whiting, eds., Aristotle, Kant and the Stoics: Rethinking happiness and
duty (Cambridge: Cambridge University Press) 261–84.
Cooper, J. M. (1999) Reason and Emotion (Princeton: Princeton University Press).
Cooper, J. M. (1997) ed. Plato: Complete Works (Indianapolis: Hackett).
Cornford, F. M. (1939) Plato and Parmenides (London: Routledge).
Cornford, F. M. (1941) The Republic of Plato (Oxford: Oxford University Press).
Cornford, F. M. (1965) ‘Mathematics and Dialectic in the Republic, VI-VII’ in R. E. Allen, ed.,
Studies in Plato’s Metaphysics (London: Routledge) 61–95.
Crisp, R. (1996) How Should One Live? (Oxford: Clarendon).
Cross, R. C. and Woozley, A. D. (1964) Plato’s Republic: A Philosophical Commentary (London:
Routledge).
Denyer, N. C. (1991) Language, Thought and Falsehood in Ancient Greek Philosophy (London:
Routledge).
Denyer, N. C. (2001) Plato: Alcibiades (Cambridge: Cambridge University Press).
Denniston, J. (1978) Greek Particles (Oxford: Oxford University Press).
Derrida, J. (1972) ‘La pharmacie de Platon’, in La Dissémination (Paris: Éditions du Seuil)
77–214.
Détienne, M. and Vernant, J.-P. (1978) Cunning Intelligence in Greek Culture and Society
(Sussex: Harvester).
Diels, H. and Kranz, W. (1968) eds. Die Fragmente der Vorsokratiker 6th edition (Zurich:
Wiedmann).
Dixsaut, M. (1998) ‘What is it Plato calls “Thinking”?’, Boston Area Colloquium in Ancient
Ferrari, G. R. F. (1990) Listening to the Cicadas (Cambridge: Cambridge University Press).
Fine, G. (1992) ‘Inquiry in the Meno’, in Richard Kraut, ed., The Cambridge Companion to
Plato (Cambridge: Cambridge University Press) 200–26.
Fine, G. (1993) On Ideas: Aristotle’s Criticism of Plato’s Theory of Forms (Oxford: Oxford
University Press).
Fine, G. (1996) ‘Conflicting Appearances: Theaetetus 153d–154b’ in C. Gill and M. M. McCabe,
eds., Form and Argument in Late Plato (Oxford: Clarendon) 105–34.
Fine, G. (2003) ‘Knowledge and Belief in Republic V–VII’, in Fine, Plato on knowledge and
forms (Oxford: Oxford University Press) 85–116.
Fine, G. (2007) ‘Enquiry and Discovery: A Discussion of Dominic Scott, Plato’s Meno’, Oxford
Studies in Ancient Philosophy 32: 331–67.
Fine, G. (2010) ‘Aristotle and the aporêma of the Meno’, in Verity Harte, Mary Margaret
McCabe, Robert W. Sharples, and Anne Sheppard, eds., Aristotle and the Stoics Reading
Plato (London: Bulletin of the Institute of Classical Studies Supplement) 107: 45–72.
Fontenrose, J. (1978) The Delphic Oracle, its Responses and Operations (Berkeley: University of
California Press).
Foot, P. (1978) Virtues and Vices (Berkeley: University of California Press).
Forster, E. S. (1933) ‘Further Emendations in the Fragments of Theophrastus’, Classical
Quarterly 27: 140–1.
Frankel, H. (1970) ‘Studies in Parmenides’, in R. E. Allen and D. Furley, eds., Studies in
Presocratic philosophy vol. 2 (London: Routledge) 1–47.
Frankel, H. (1974) ‘A Thought Pattern in Heraclitus’ in A. P. D. Mourelatos, ed., The
Presocratics (Princeton: Princeton University Press) 214–28.
Frede, D. (1989) ‘The Soul’s Silent Dialogue: A Non-aporetic Reading of the Theaetetus’,
Proceedings of the Cambridge Philological Society 215: 20–49.
Frede, D. (1999) ‘Plato on What the Body’s Eye Tells the Mind’s Eye’, Proceedings of the
Aristotelian Society 99: 191–210.
Frede, D. (2012) ‘The Endoxon Mystique: What Endoxa are and What They are Not’, Oxford
Studies in Ancient Philosophy 43, 185–219.
Frede, M. (1987) ‘Observations on Perception in Plato’s Later Dialogues’ in Frede, Essays in
Ancient Philosophy (Oxford: Oxford University Press) 3–9.
Frede, M. (1999) ‘On the Stoic Conception of the Good’, in K. Ierodiakonou, ed., Topics in
Stoic Philosophy (Oxford: Oxford University Press) 71–94.
Frede, M. and Patzig, G. (1988) Aristoteles ‘Metaphysik Z’, Text, Übersetzung und Kommentar
2 vols (Munich: Beck).
Frege, G. (1997) ‘Thought’, trs. Geach and Stoothof, in M. Beaney, ed., The Frege Reader
(Oxford: Oxford University Press) 325–45.
Furley, D. J. (1978) ‘Self-movers’, in G. E. R. Lloyd and G. E. L. Owen, eds., Aristotle on Mind
and the Senses (Cambridge: Cambridge University Press) 165–80.
Furth, M. (1974) ‘Elements of Eleatic Ontology’, in A. P. D. Mourelatos, ed., The Presocratics
(Princeton: Princeton University Press) 241–70.
Geach, P. (1960) ‘Plato’s Euthyphro: An Analysis and Commentary’, The Monist 50: 369–82.
Gerson, L. (2003) Knowing Persons (Oxford: Clarendon).
Gerson, L. (2009) Ancient Epistemology (Cambridge: Cambridge University Press).
Gettier, E. (1966) ‘Is Justified True Belief Knowledge?’, Analysis 23: 121–3.
Gifford, E. H. (1905) The Euthydemus of Plato (Oxford: Oxford University Press).
Gill, C. (1996) Personality in Greek Epic, Tragedy and Philosophy: The Self in Dialogue (Oxford:
Clarendon).
Gill, C. (2002) ‘Protreptic and Dialectic in Plato’s Euthydemus’, in T. M. Robinson and
L. Brisson, eds., Plato: Euthydemus, Lysis, Charmides (Sankt Augustin: Academia Verlag).
Gill, M. L. (2001) ‘Aristotle’s Attack on Universals’, Oxford Studies in Ancient Philosophy 20:
235–60.
Goldie, P. (2004) ‘Emotion, Reason and Virtue’, in P. Cruse and D. Evans, eds., Emotion,
Evolution and Rationality (Oxford: Oxford University Press) 249–68.
Gomez-Lobo, A. (1977) ‘Plato’s Description of Dialectic in the Sophist’, Phronesis 22: 29–47.
Goody, J. (1987) The Interface Between the Written and the Oral (Cambridge: Cambridge
University Press).
Gower, O. (2008) ‘Why is there an Autobiography in the Phaedo?’, Ancient Philosophy 28:
329–46.
Halliwell, S. (2006) ‘An Aristotelian Perspective on Plato’s Dialogues’, in F. G. Hermann, ed.,
New Essays on Plato (Swansea: Classical Press of Wales) 189–211.
Hamlyn, D. W. (1968) Aristotle’s de anima Books 2 and 3 (Oxford: Clarendon Press).
Harte, V. (2002) Plato on Parts and Wholes: The Metaphysics of Structure (Oxford: Oxford
University Press).
Harte, V. (2006) ‘Beware of Imitations: Image Recognition in Plato’, in F. Hermann, ed., New
Essays on Plato (Swansea: Classical Press of Wales) 21–42.
Harte, V. (2007) ‘Language in the Cave’, in D. Scott, ed., Maieusis (Oxford: Oxford University
Press) 195–215.
Harte, V. (2010) ‘Republic X and the Role of the Audience in Art’, Oxford Studies in Ancient
Hartman, E. (1977) Substance, Body and Soul (Princeton: Princeton University Press).
Havelock, E. (1963) A Preface to Plato (Oxford: Oxford University Press).
Hawtrey, R. S. W. (1981) Commentary on Plato’s Euthydemus (Philadelphia: Diane
Publishing).
Heinaman, R. (2003) ‘Plato: Metaphysics and Epistemology’, in C. C. W. Taylor, ed., From the
Beginning to Plato (London: Routledge) 356–93.
Heinaman, R., Metaphysics Z [Online], Available: <http://www.archelogos.com> [8 October
2014].
Hett, W. S. (1964) Aristotle on the Soul, Parva Naturalia (Cambridge, MA: Loeb Classical
Library).
Hicks, R. D. (1907) Aristotle, de anima (Cambridge: Cambridge University Press).
Housman, A. E. (1901) ‘Fragment of a Greek Tragedy’, The Cornhill Magazine.
Hume, D. (1978) A Treatise on Human Nature, 2nd edn., ed. L. A. Selby-Bigge (Oxford:
Oxford University Press).
Hursthouse, R. (1999) Virtue Ethics (Oxford: Oxford University Press).
Ioppolo, A. M. (1980) Aristone di Chio e lo stoicismo antico (Naples: Bibliopolis).
Irwin, T. H. (1988) Aristotle’s First Principles (Oxford: Clarendon).
Irwin, T. H. (1995) Plato’s Ethics (Oxford: Oxford University Press).
Jameson, G. (1958) ‘Well-rounded Truth and Circular Argument in Parmenides’, Phronesis 3:
15–30.
Johansen, T. (2005) ‘In Defense of Inner Sense’, Proceedings of the Boston Area Colloquium in
Ancient Philosophy 21: 235–76.
Kahn, C. H. (1969) ‘The Thesis of Parmenides’, Review of Metaphysics 22: 700–24.
Kahn, C. H. (1973) The Verb ‘To Be’ in Ancient Greek (Dordrecht: Reidel).
Kahn, C. H. (1979a) The Art and Thought of Heraclitus (Cambridge: Cambridge University
Press).
Kahn, C. H. (1979b) ‘Sensation and Consciousness in Aristotle’s Psychology’, in J. Barnes,
M. Schofield, R. Sorabji, eds., Articles on Aristotle: 4: Psychology and Aesthetics (London:
Duckworth) 1–31.
Kahn, C. H. (1996) Plato and the Socratic Dialogue (Cambridge: Cambridge University Press).
Kant, I. (1964) Groundwork of the Metaphysic of Morals, trs. H. J. Paton (New York: Harper).
Kirk, G. S. (1954) Heraclitus: The Cosmic Fragments (Cambridge: Cambridge University
Press).
Kirk, G. S. (1974) ‘Natural change in Heraclitus’, in A. P. D. Mourelatos, ed., The Presocratics
(Princeton: Princeton University Press).
Kirk, G. S. and Raven, J. (1966) The Presocratic Philosophers (Cambridge: Cambridge Univer-
sity Press).
Kneale, W. and Kneale, M. (1962) The Development of Logic (Oxford: Clarendon Press).
Kornblith, H. (2001) ed. Epistemology: Internalism and Externalism (Cambridge, MA: MIT
Press).
Korsgaard, C. (1983) ‘Two Distinctions in Goodness’, Philosophical Review 92: 169–95.
Korsgaard, C. (1996) The Sources of Normativity (Cambridge: Cambridge University Press).
Kosman, L. A. (1975) ‘Perceiving that We Perceive: On the Soul III, 2’, Philosophical Review 84:
499–519.
Kosman, L. A. (2004) ‘Aristotle on the Desirability of Friends’, Ancient Philosophy 24: 135–54.
Kosman, L. A. (2005) ‘Commentary on Johansen’, Proceedings of the Boston Area Colloquium
in Ancient Philosophy 21: 277–83.
Kraut, R. (1973) ‘Egoism, Love and Political Office in Plato’, Philosophical Review 82: 330–44.
Kuhner, R. and Gerth, B. (1992) Ausführliche Grammatik der griechischen Sprache (Hannover:
Hahnsche Buchhandlung).
Locke, J. (1975) An Essay Concerning Human Understanding, ed. P. H. Nidditch (Oxford:
Long, A. A. (1970) ‘The Principles of Parmenides’ Cosmogony’, in R. E. Allen and D. Furley,
eds., Studies in Presocratic philosophy vol. 2 (London: Routledge) 82–101.
Long, A. A. (1996) ‘Socrates and Hellenistic philosophy’, in A. A. Long, Stoic Studies (Cam-
bridge: Cambridge University Press).
Long, A. A. (2005) ‘Platonic Souls as Persons’, in R. Salles, ed., Metaphysics, Soul and Ethics in
Ancient Thought: Themes from the work of Richard Sorabji (Oxford: Oxford University
Press) ch. 8.
Long, A. A. and Sedley, D. N. (1987) The Hellenistic Philosophers (Cambridge: Cambridge
University Press).
Long, A. G. (2004) Character and Dialectic: The Philosophical Origins of the Platonic Dialogue
(unpub. PhD dissertation: University of Cambridge).
Lycan, W. G. (1996) Consciousness and Experience (Cambridge, MA: MIT Press).
Lyons, J. J. (1963) Structural Semantics (Cambridge: Cambridge University Press).
McCabe, M. M. [as M. M. Mackenzie] (1978) ‘The Tears of Chryses: Retaliation in the Iliad’,
Philosophy and Literature 2: 3–22.
McCabe, M. M. (1981) Plato on Punishment (Berkeley: University of California Press).
McCabe, M. M. (1982) ‘Paradox in Plato’s Phaedrus’, Proceedings of the Cambridge Philological
Society 208: 64–76.
McCabe, M. M. (1986) ‘Putting the Cratylus in its Place’, Classical Quarterly 36: 124–50.
McCabe, M. M. (1988a) ‘The Virtues of Socratic Ignorance’, Classical Quarterly 38: 331–50.
McCabe, M. M. (1988b) ‘Impasse and Explanation: From the Lysis to the Phaedo’, Archiv für
Geschichte der Philosophie 70: 15–45.
McCabe, M. M. (1994a) Plato’s Individuals (Princeton: Princeton University Press).
McCabe, M. M. (1994b) ‘Arguments in Context: Aristotle’s Rhetoric and the Platonic

challenge’, in A. Nehamas and D. Furley, eds., Aristotle’s Rhetoric: Philosophical Essays
(Princeton: Princeton University Press) 129–65.
McCabe, M. M. (1998) ‘Silencing the Sophists: The Drama of Plato’s Euthydemus’, Proceedings
of the Boston Colloquium in Ancient Philosophy 14: 139–68.
McCabe, M. M. (2000) Plato and his Predecessors (Cambridge: Cambridge University Press).
McCabe, M. M. (2001) ‘Developing the Good Itself by Itself: Critical Strategies in the Euthyde-
mus’, Plato: online journal of the International Plato Society 2.
McCabe, M. M. (2004) ‘Irony in the Soul: Should Plato’s Socrates be Sincere?’ in M. Trapp, ed.,
Socrates (Aldershot: Ashgate) 17–32.
McCabe, M. M. (2005) ‘Extend or Identify: Two Stoic Accounts of Altruism’, in R. Salles, ed.,
Metaphysics, Soul and Ethics: Festschrift for Richard Sorabji (Oxford: Oxford University
Press).
McCabe, M. M. (2010) ‘Banana Skins and Custard Pies: Plato on Comedy and Self-Knowledge’,
in J. M. Dillon and L. Brisson, eds., Plato’s Philebus: Selected Papers From the Eighth
Symposium Platonicum (Sankt Augustin: Academia Verlag) 194–204.
McCabe, M. M. (2013) ‘The Stoic Sage in the Original Position’, in V. Harte and M. Lane, eds.,
Politeia (Cambridge: Cambridge University Press) 251–74.
McCabe, M. M. (2014) ‘Thinking Outside the Frame: Plato, Quinn and Artaud’, in D. Freeman
and D. Matravers, eds., Figuring Out Figurative Art (London: Routledge).
McCabe, M. M. (2015) ‘First Chop your logos’, in L. Castagnoli and V. di Lascio, eds., Ancient
Logic (forthcoming).
McDowell, J. (1998) Mind, Value and Reality (Cambridge, MA: Harvard).
MacIntyre, A. (1981) After Virtue (London: Duckworth).
Marcovich, M. (1967) Heraclitus (Merida: The Los Andes Press).
Mates, B. (1979) ‘Identity and Predication in Plato’, Phronesis 24: 211–29.
Matson, W. I. (1980) ‘Parmenides Unbound’, Philosophical Inquiry 2: 345–60.
Matthews, G. (1999) Socratic Perplexity (Oxford: Oxford University Press).
Meinwald, C. (1991) Plato’s Parmenides (New York: Oxford University Press).
Meinwald, C. (1992) ‘Goodbye to the Third Man’, in R. Kraut, ed., The Cambridge Companion
to Plato (Cambridge: Cambridge University Press) 365–96.
Moore, G. E. (1903) Principia Ethica (Cambridge: Cambridge University Press).
Moravcsik, J. (1992) Plato and Platonism (Oxford: Oxford University Press).
Mourelatos, A. P. D. (1970) The Route of Parmenides (New Haven: Yale University Press).
Mourelatos, A. P. D. (1973) ‘Heraclitus, Parmenides and the Naı̈ve Metaphysics of Things’, in
E. Lee, A. Mourelatos, R. Rorty, eds., Exegesis and Argument (Assen: Van Gorcum) 16–48.
Nagel, T. (1976) ‘Moral Luck’, Proceedings of the Aristotelian Society Supplement 1: 138–51.
Nagel, T. (1979) ‘What is it Like to be a Bat?’ in Mortal Questions (Cambridge: Cambridge
University Press) 165–80.
Nails, D. (2002) The People of Plato (Indianapolis: Hackett).
Narcy, M. (1984) Le philosophe et son double: un commentaire de l’Euthydème de Platon (Paris:
Vrin).
Nehamas, A. (1985) ‘Meno’s Paradox and Socrates as a Teacher’, Oxford Studies in Ancient
Nehamas, A. (2004) ‘Epistêmê and Logos in Plato’s Later Thought’, in A. Nehamas, Virtues of
Authenticity (Princeton: Princeton University Press) 224–48.
Nightingale, A. (2004) Spectacles of Truth in Classical Greek Philosophy (Cambridge: Cam-
bridge University Press).
Nussbaum, M. C. (1972) ‘Psuchê in Heraclitus’, Phronesis 17: 1–16, 153–70.
Nussbaum, M. C. (1979) ‘Eleatic Conventionalism and Philolaus on the Conditions of
Thought’, Harvard Studies in Classical Philology 82: 63–108.
Nussbaum, M. C. (1982) ‘Saving Aristotle’s Appearances’, in M. Schofield and M. Nussbaum,
eds., Language and Logos (Cambridge: Cambridge University Press) 267–94.
Nussbaum, M. C. (1986) The Fragility of Goodness (Cambridge: Cambridge University Press).
Nussbaum, M. C. and Rorty, A. (1992) eds. Essays on Aristotle’s De Anima (Oxford: Oxford
University Press).
O’Shaughnessy, B. (2003) ‘Proprioception and the Body Image’, in B. O’Shaughnessy,
Consciousness and the World (Oxford: Oxford University Press) 628–55.
Ong, W. J. (1982) Orality and Literacy: The Technologizing of the Word (London: Routledge).
Osborne, C. (now Rowett) (1983) ‘Aristotle, De anima 3.2: How Do We Perceive that We See
and Hear?’, Classical Quarterly 33: 401–11.
Osborne, C. (1987a) ‘The Repudiation of Representation in Plato’s Republic and its Reper-
cussions’, Proceedings of the Cambridge Philological Society 33: 53–73.
Osborne, C. (1987b) Rethinking Early Greek Philosophy (Ithaca: Cornell University Press).
Osborne, C. (1996) ‘Space, Time, Shape and Direction: Creative Discourse in the Timaeus’, in
C. Gill and M. M. McCabe, eds., Form and Argument in Late Plato (Oxford: Clarendon)
179–212.
Osborne, C. (1998) ‘Perceiving White and Sweet (Again): Aristotle De anima 3.7,
431a20–431b1’, Classical Quarterly 48: 433–46.
Owen, G. E. L. (1974) ‘Plato and Parmenides on the Timeless Present’, in A. P. D. Mourelatos,
ed., The Presocratics (Princeton: Princeton University Press) 271–92.
Owen, G. E. L. (1975) ‘Eleatic Questions’, in R. E. Allen and D. Furley, eds., Studies in
Presocratic philosophy vol. 2 (London: Routledge) 48–81.
Owen, G. E. L. (1986a) ‘Tithenai ta phainomena’, in G. E. L. Owen, Logic, Science and Dialectic
(London: Duckworth) 239–51.
Owen, G. E. L. (1986b) ‘Notes on Ryle’s Plato’, in G. E. L. Owen, Logic, Science and Dialectic
(London: Duckworth) 85–103.
Owen, G. E. L. (1986c) ‘The Place of the Timaeus in Plato’s Dialogues’, in G. E. L. Owen, Logic,
Science and Dialectic (London: Duckworth) 65–84.
Papineau, D. (2002) Thinking about Consciousness (Oxford: Oxford University Press).
Parker, R. (2000) ‘Greek States and Greek Oracles’, in R. Buxton, ed., Oxford Readings in Greek
Religion (Oxford: Oxford University Press) 76–108.
Penner, T. (2005) ‘Socratic Ethics: Ultra-Realism, Determinism, and Ethical Truth’, in C. Gill,
ed., Virtue, Norms and Objectivity (Oxford University Press) 157–87.
Penner, T. and Rowe, C. (2004) Plato’s Lysis (Cambridge: Cambridge University Press).
Politis, V. (2004a) Aristotle and the Metaphysics (London: Routledge).
Politis, V. (2004b) ‘Aporia and Searching in the Early Plato’, in V. Karasmanis, ed., Socrates,
2000 Years Since his Death (Delphi: European Cultural Centre of Delphi) 303–17.
Price, A. (1989) Love and Friendship in Plato and Aristotle (Oxford: Oxford University Press).
Pryor, J. (1999) ‘Immunity to Error Through Misidentification’, Philosophical Topics 26:

272–304.
Rabinowicz, W. and Ronnow-Rasmussen, T. (2000) ‘A Distinction in Value: Intrinsic and for
Its Own Sake’, Proceedings of the Aristotelian Society 100: 33–52.
Ramelli, I. and Konstan, D. (2009) Hierocles the Stoic: Elements of Ethics, Fragments and
Excerpts (Atlanta: Society of Biblical Literature).
Rawls, J. (1972) A Theory of Justice (Oxford: Oxford University Press).
Reale, G. and Scolnicov, S. (2002) (eds.) New Images of Plato: Dialogues on the Idea of the Good
(St. Augustin: Academia Verlag).
Reinhardt, K. (1974) ‘The Relation between the Two Parts of Parmenides’ Poem’ in
A. P. D. Mourelatos, ed., The Presocratics (Princeton: Princeton University Press).
Robinson, R. (1953) Plato’s earlier dialectic (Oxford: Oxford University Press).
Ross, W. D. (1953) Aristotle’s Metaphysics 2 vols. Second ed. (Oxford: Oxford University
Press).
Ross, W. D. (1956) Aristotelis de anima (Oxford: Oxford Classical Texts).
Ross, W. D. (1961) Aristotle: de anima (Oxford: Clarendon Press).
Rowe, C. J. (1985) Plato: Phaedrus (Warminster: Aris and Phillips).
Rowe, C. J. (1995) Plato: Statesman (Warminster: Aris and Phillips).
Rowe, C. J. (2010) Plato and the Art of Philosophical Writing (Cambridge: Cambridge
University Press).
Ruben, D. H. (1991) Explaining Explanation (London: Routledge).
Rutherford, R. (1995) The Art of Plato. (London: Duckworth).
Ryle, G. (1949) The Concept of Mind (London: Hutchinson).
Ryle, G. (1965) ‘Plato’s Parmenides’, in R. E. Allen, ed., Studies in Plato’s Metaphysics (London:
Routledge) 97–147.
Sainsbury, R. M. (1988) Paradoxes (Cambridge: Cambridge University Press).
Sainsbury, R. M. (2011) ‘English Speakers Should use “I” to Refer to Themselves’, in
A. Hatzimoysis, ed., Self-knowledge (Oxford: Oxford University Press) 246–60.
Savile, A. (1996) ‘The Heart of History’, Proceedings of the Aristotelian Society 96: 197–213.
Scheffler, S. (1982) The Rejection of Consequentialism (Oxford: Oxford University Press).
Scheffler, S. (1988) ed. Consequentialism and its Critics (Oxford: Clarendon Press).
Schmid, W. T. (1998) Plato’s Charmides and the Socratic Ideal of Rationality (Albany: SUNY
Press).
Schneewind, J. B. (1996) ‘Kant and Stoic Ethics’, in S. Engstrom and J. Whiting, eds., Aristotle,
Kant and the Stoics: Rethinking Happiness and Duty (Cambridge: Cambridge University
Press) 285–302.
Schofield, M. (1977) ‘The Antinomies of Plato’s Parmenides’, Classical Quarterly 27: 139–58.
Schofield, M. (1984) ‘Ariston of Chios and the Unity of Virtue’, Ancient Philosophy 4: 83–96.
Schofield, M. (1996) ‘Likeness and Likenesses in the Parmenides’, in C. Gill and M. M. McCabe,
eds., Form and Argument in Late Plato (Oxford: Oxford University Press) 49–78.
Scott, D. (1995) Recollection and Experience (Cambridge: Cambridge University Press).
Scott, D. (2006) Plato’s Meno (Cambridge: Cambridge University Press).
Sedley, D. N. (1990) ‘Teleology and Myth in the Phaedo’, Proceedings of the Boston Area
Colloquium in Ancient Philosophy 5: 359–83.
Sedley, D. N. (1996) ‘Three Platonist Interpretations of the Theaetetus’, in C. Gill and

M. M. McCabe, eds., Form and Argument in Late Plato (Oxford: Oxford University Press)
79–103.
Sedley, D. N. (1998) ‘Platonic Causes’, Phronesis 43: 114–32.
Sedley, D. N. (1999) ‘The Ideal of Godlikeness’, in G. Fine, ed., Plato 2: Ethics, Politics, Religion
and the Soul (Oxford: Oxford University Press) 309–28.
Sedley, D. N. (2003) Plato’s Cratylus (Cambridge: Cambridge University Press).
Sedley, D. N. (2004) The Midwife of Platonism: Text and Subtext in Plato’s Theaetetus (Oxford:
Sharples, R. W. (2003) ‘Theophrastus, On Dizziness’, in W. W. Fortenbaugh, R. W. Sharples,
and M. Sollenberger, eds., Theophrastus on Sweat, Dizziness and Fatigue (Leiden: Brill)
169–249.
Sherman, N. (1993) ‘The Virtues of Common Pursuit’, Philosophy and Phenomenological
Research 53: 277–99.
Shoemaker, S. (1968) ‘Self-reference and Self-awareness’, Journal of Philosophy 65: 555–67.
Shorey, P. (1907) ‘Emendation of Plato Charmides 168b’, Classical Philology 2: 340.
Silverman, A. (2002) The Dialectic of Essence (Princeton: Princeton University Press).
Sisko, J. E. (2004) ‘Reflexive Awareness Does Belong to the Main Function of Perception: Reply
to Caston’, Mind 113: 513–22.
Sorabji, R. R. K. (1979) ‘Body and Soul in Aristotle’, in Barnes, Schofield, and Sorabji, eds.,
Articles on Aristotle Vol. 4: Psychology and Aesthetics (London: Duckworth) 42–64.
Sorabji, R. R. K. (1992) ‘Intentionality and Physiological Processes: Aristotle’s Theory of Sense-
Perception’, in Nussbaum and Rorty, eds., Essays on Aristotle’s De Anima (Oxford: Clarendon)
195–226.
Sorabji, R. R. K. (1993) Animal Minds and Human Morals (London: Duckworth).
Sorabji, R. R. K. (1999) ‘Soul and Self in Ancient Philosophy’, in M. J. C. Crabbe, ed., From Soul
to Self (London: Routledge) 8–32.
Sprague, R. K. (1962) Plato’s Use of Fallacy (London: Routledge).
Stenzel, J. (1940) Plato’s Method of Dialectic, trs. D. J. Allen (Oxford: Oxford University Press).
Stern-Gillet, S. (1995) Aristotle’s Philosophy of Friendship (New York: SUNY Press).
Stokes, M. C. (1971) One and Many in Presocratic Philosophy (Cambridge, MA: Harvard
University Press).
Strawson, P. (1959) Individuals (London: Routledge).
Striker, G. (1996a) ‘Following Nature: A Study in Stoic Ethics’, in G. Striker, Essays on Hellenistic
Epistemology and Ethics (Cambridge: Cambridge University Press) 221–80.
Striker, G. (1996b) ‘Plato’s Socrates and the Stoics’, in G. Striker, Essays on Hellenistic
Epistemology and Ethics (Cambridge: Cambridge University Press) 316–24.
Szlezak, T. (1999) Reading Plato, trs. G. Zanker (London: Routledge).
Taylor, C. (1989) Sources of the Self: The Making of the Modern Identity (Cambridge: Cambridge
University Press).
Tye, M. (1995) Ten Problems of Consciousness (Cambridge, MA: MIT Press).
Van der Ben, N. (1985) The Charmides of Plato: Problems and Interpretations (Amsterdam:
B. R. Grüner).
Verdenius, W. J. (1966) ‘Der Logosbegriff bei Heraklit und Parmenides’, Phronesis 11: 81–98.
Vernant, J.-P. (1981) ‘The Union with Metis and the Sovereignty of Heaven’, in R. Gordon, ed.,
Myth, Religion and Society (Cambridge: Cambridge University Press) 1–15.
Vlastos, G. (1946) ‘Parmenides’ Theory of Knowledge’, Transactions of the American Philo-
logical Association 77: 66–77.
Vlastos, G. (1955) ‘On Heraclitus’, American Journal of Philology 76: 337–68.
Vlastos, G. (1983) ‘The Socratic Elenchus’, Oxford Studies in Ancient Philosophy 1: 27–58,
71–4.
Vlastos, G. (1991) Socrates, Ironist and Moral Philosopher (Cambridge: Cambridge University
Press).
Vogt, K. (2012) Belief and Truth: A Skeptic Reading of Plato (Oxford: Oxford University Press).
White, N. P. (1992) ‘Plato’s Metaphysical Epistemology’, in R. Kraut, ed., The Cambridge
Companion to Plato (Cambridge: Cambridge University Press) 277–310.
Whiting, J. (2006) ‘The Nicomachean Account of Philia’, in R. Kraut, ed., The Blackwell Guide
to Aristotle’s Nicomachean Ethics (Oxford: Oxford University Press) 276–304.
Whiting, J. (2012) ‘The Pleasure of Thinking Together: Prolegomenon to a Complete Reading
of EE VII.12’, in F. Leigh, ed., The ‘Eudemian Ethics’ on the Voluntary, Friendship, and Luck:
The Sixth S.V. Keeling Colloquium in Ancient Philosophy (Leiden: Brill) ch. 5.
Wiggins, D. (1971) ‘Sentence Meaning, Negation and Plato’s Problem of Not-being’, in
G. Vlastos, ed., Plato I (New York: Doubleday) 268–303.
Wiggins, D. (1982) ‘Heraclitus’ Conceptions of Flux, Fire and Material Persistence’, in
M. Schofield and M. Nussbaum, eds., Language and Logos (Cambridge: Cambridge Univer-
sity Press) 1–32.
Wilberding, J. (2004) ‘Prisoners and Puppeteers in the Cave’, Oxford Studies in Ancient
Williams, B. A. O. (1976) ‘Moral Luck’, Proceedings of the Aristotelian Society Supplement 1:
115–35.
Williams, B. A. O. (1978) Descartes: The Project of Pure Enquiry (London: Penguin).
Williams, B. A. O. (1985) Ethics and the Limits of Philosophy (London: Fontana).
Williams, B. A. O. (2003) ‘Plato’s Construction of Intrinsic Goodness’, in R. W. Sharples, ed.,
Perspectives on Greek Philosophy: S.V. Keeling Lectures in Ancient Philosophy (Aldershot:
Ashgate) 1–18.
Williamson, T. (2000) Knowledge and Its Limits (Oxford: Oxford University Press).
Woodbury, L. (1958) ‘Parmenides on Names’, Harvard Studies in Classical Philology 63:
145–60.
Woods, M. (1967) ‘Problems in Metaphysics Z Chapter 13’, in J. M. E. Moravcsik, ed., Aristotle
(Garden City: Anchor) 215–38.
Zagzebski, L. (1996) Virtues of the Mind (Cambridge: Cambridge University Press).
Index Locorum
AESCHYLUS 425b20–2 291, 296

Agamemnon 211 81n47 425b20 287n28, 290n44
Supplices 628 81n51 425b21–2 299
[ALEXANDER] 425b21 284n10, 291n48, 300
Problemata 4.42 66 425b22–5 291
425b22–4 296
ANAXIMANDER DK12B1 59, 78n35 425b22 283n4, 290, 300n74
ANTIPHON 1.7 79n41 425b23–5 300
425b23 302n82
ARISTOPHANES 425b26 ff. 289n38
Equites 1354 67 425b26–426a26 300
Thesmophoriazusae 930 67 425b26 352n60
Clouds 365 ff. 89 426a2 290
ARISTOTLE 426a6–16 300n75
Analytica posteriora 426a13 290n44
1.1–2 25n91 426a15 ff. 296n65
1.2 13n47 426a20 ff. 286n16
2.19 99b22–27, 191n8 426a23–5 283n6
Analytica priora 87 426a26 296n65
24b24 308 426a27 ff. 289n38
26b26 86 426a27–b7 300
de anima 182n21, 283–309, 348, 362n103 426b5 301n77
405a15 322n39 426b8–14 300n75
2.1 340n2 426b12–21 300
2.5 284n8, 306 426b14–29 300n75
417a2 ff. 291n46 426b20–1 300
417a3–6 183n34 426b28 ff. 300n76
417a3–5 288n33 426b29–427a6 300n75
417a9 ff. 283n6 427a9–14 300
422a20–3 298 3.3 301n77, 305
422a21 298n71 427a17 ff. 286n17
2.12 301n77, 306 427a19–21 308
3.1 301n77 427a 21 ff. 286n16
425a6 301n77 427b3 304n87
3.2 178n14, 307 428a14 ff. 305n89
425b12–25 283–4 428b3–4 305n89
425b12–20 174n5 428b3 57n45
425b12 283n2, 287, 291n48 3.4 306
425b13 ff. 348 3.5 306, 307n90
425b13–17 305 3.7 296n64
425b13–15 287 de caelo 313b20 358n92
425b13–14 294 de generatione et corruptione 332b32 ff. 69
425b13 283n2, 287, 289n38, 290n44, de partibus animalium 1.1 96
294n58 de sensu et sensibilibus
425b15–17 287 437a26–9 286n24
425b15 287n29 449a5–20 296n65
425b16 287n29 de somno et vigilia 455a12 ff. 284n11,
425b17–20 174n5, 294 288n34, 296n66, 297n68
425b18 289n38 de sophisticis elenchis 166b 128
425b19 283n1, 290n43, 294n58 Ethica Eudemia 340–65
425b19–20 299 1224b21 ff. 359n97
380 INDEX LOCORUM
ARISTOTLE (cont.) 1245a11–15 341n9

1235b13–18 341 1245a11 349n39, 352n64
1235b16 341n8 1245a13–14 353n65
1240a14–21 353n69 1245a16 ff. 363n106
1240a14–15 359 1245a16 350
1240a14 359n96 1245a17–18 362n105
1240a20–1 359 1245a18–26 350, 353
1240b15–21 353n69 1245a20–1 350, 356
1240b16 353n69 1245a22 363n107
1240b28–30 353n68 1245a23–4 350
7.12 25, 340–365 1245a24 359
1244b1–24 349 1245a26–b9 349
1244b1–21 348 1245a27 353
1244b1–15 349 1245a28 350n50
1244b2 ff. 349 1245a29–35 353, 354
1244b3 349 1245a30–4 365
1244b4–5 349 1245a31–4 356
1244b7 341n9, 349n41, 351n51 1245a31 358
1244b9 341n6 1245a32 341n9, 359
1244b11 350n48 1245a34–5 356, 359
1244b12–14 340 1245a34 356n83
1244b15–21 349 1245a35–b9 353, 361
1244b15 349n41 1245a35–7 354
1244b17–19 350 1245a35 359n96
1244b17 341n9, 349n40, 351n53 1245a36–7 362
1244b18–19 356n82 1245a37–b1 354
1244b18 355n79 1245a37 341n9, 351n53, 354n72, 354n73
1244b19–20 363n106 1245a38 354n72, 354n73
1244b20–1 341, 351n52 1245a39 341n8
1244b20 351n51 1245b1–9 361n102
1244b21–45a10 349 1245b2–9 355
1244b21–3 350 1245b2 362
1244b22–3 351n54 1245b3–4 362
1244b23 351, 352n64, 361n101 1245b3 341n9, 351n53
1244b24–34 303 1245b4 ff. 355n75
1244b24–6 351 1245b4–6 355n76
1244b25–6 341n9, 352 1245b4 341n9, 351n53, 355n76
1244b26–45a10 353n65 1245b5 341n9, 351n53
1244b26–8 351 1245b8 341n9, 351n53, 356
1244b26–7 351n57 1245b9–46a25 349, 353
1244b28 351n58 1245b9–19 356–7
1244b29–45a11 352 1245b9–11 364
1244b30 351n57 1245b10 341n9, 351n53
1244b31–2 350 1245b12–19 350
1244b32 351n57 1245b12–15 364
1244b33 341n8, 351n57, 351n58, 352 1245b13 341n8, 354n72
1244b34 357n85 1245b20–4 360
1245a1 352n59, 352n64, 354n74 1245b22 341n9, 351n53
1245a3 352n59, 352n64 1245b24 351n53, 362
1245a4–5 351n57 1245b25 341n9
1245a6 352n59 1245b37 354n72
1245a7–11 352 1246a13 354n72
1245a7–8 361 Ethica Nicomachea 340–65
1245a9–10 352n59 Book 1 255n114
1245a10 351n55, 351n57, 352n60 1.6 26n96
1245a11–26 349 1098b4 308
1245a11–18 349, 353 1099a23 308
INDEX LOCORUM 381
1100a33 308 1004b25 74n7

1102b13 359n97 1005b11 ff. 44
2.1 25 1005b19 ff. 39, 44
1113a4 ff. 308 1005b19–22 133
1140b8 ff. 308n91 1005b20–2 176n9
1141b9 ff. 308 1005b22 14n50
1143a6 ff. 308 1006a11–12 134
1143b5 308 1006a12 45
7.1 284n10 1008a2 ff. 44
1045b2–6 262n2 1009a6 ff. 304n87
1166a13 346n31 1010a12 35n2
1166a23–4 346n31 1017a3 ff. 283n6
1166a27 346n31 1023a26–b11 319n30
1166a29–31 346n31 VII (Z) 26
1166a31–2 346n31 1028a33–6 329n50
1166a31 345n29 1028b1–3 328
1166a34–5 358n89 1029a5 69
9.9, 284n11 305, 340–65 Z13–16 310–39
1169b3–5 345 1038b1–2 310
1169b16–19 345 1038b4–6 311
1169b16 351n53 1038b6–39a3 311
1169b29–70a4 347 1038b6 311
1170a5 351n53 1038b7–8 314
1170a10 345n28 1038b7 314
1170a13 345 1038b8–15 318, 319, 320, 321
1170a25–b14 345–6 1038b8–9 314, 315
1170a25–b5 347 1038b8 314
1170a26 346n30 1038b9 ff. 314
1170a29 ff. 352n61 1038b9–15 314
1170a29–b1 303 1038b9–12 315
1179b1 346n30 1038b10 314, 320
1170b4–5 346 1038b11 322
1170b5–6 362 1038b12–15 315–6
1170b6–7 346n30 1038b14–15 318
1170b6 345n29 1038b15–19 338
1170b7–8 347 1038b15–16 314, 316, 317, 317n18, 324
1170b8–10 347 1038b15 317n19, 330n52
1170b9 346n30 1038b16 ff. 324
1170b12–13 349n42 1038b16–23 315, 324
1179b10–11 346n30, 347n32 1038b16–18 316, 331
10.7 342n14 1038b16 316n14, 317n20
Magna Moralia 342–4, 361, 364 1038b17 317n19, 322, 323
1213a8–27 342–3 1038b17–18 324
1213a11 342n12 1038b18 ff. 321
1213a12 353n67 1038b18–23 317
1213a24 342n12 1038b18 317, 317n20, 320, 322n40, 324
Metaphysica 310–39 1038b19–20 317n20
982a5 ff. 322n38 1038b19 317n20, 331
983b2 81n49 1038b20–3 319
983b6 ff. 46 1038b23–30 324
995a33–b3 307, 313 1038b23–9 314, 318, 319
995a34 330n51 1038b23–7 319
986b31 77n25 1038b23 317n16, 317n20, 330n52, 331
987a30 ff. 48 1038b24 324
998a24–b4 90–1 1038b25 322, 327n48
III.1 (B.1) 295n63 1038b27 ff. 321
IV (ˆ) 30n107 1038b29–30 314, 318, 319, 320n31
382 INDEX LOCORUM
ARISTOTLE (cont.) 1039b32–40a3 336

1038b29 318n26, 320, 322n40, 323 1040a1–2 312, 338
1038b30–1039a3 329 1040a1 313n3
1038b30–4 314, 319, 320 1040a5–7 312
1038b30 319 1040a8–b4 312, 336
1038b31–3 324 1040a8–9 312, 337
1038b32 320, 323 1040a28–30 337
1038b34–1039a3 320 1040b2–4 337
1038b34–5 311, 314 1040b2 312
1038b34 313n3, 330 1040b3 313n3
1038b35–1039a3 311, 324 1040b5–27 312, 337
1039a2–3 320, 322n36 1040b5 312, 313n3
1039a3–23 311 1040b9 337
1039a3 311, 313n3, 324, 330, 330n52 1040b18 313n3
1039a4–8 325–6 1040b25–7 312
1039a6 335 1040b26 313n3
1039a7 326 1040b27–41a5 313, 337
1039a8 319n30 1040b27–8 313
1039a9–13 324 1040b34–1041a3 337
1039a11 313n3 1041a3–5 310, 313, 337
1039a12–13 332 1041a5 313n3
1039a14–19 327 XII.9 307n90, 352n59
1039a16–17 334 XIII (M). 4, 5 26n96
1039a18 327n49 Meteorologica 372b19 358n92
1039a21–3 331n55 Politica
1039a22–3 311, 328 1.2 308
1039a22 313n3 1253a38 308
1039a24–b19 312 1303b13 358n92
1039a24–33 332 Physica 70
1039a24–6 312, 331 193a3–9 30n107
1039a24 313n3 202a20 ff. 47n28
1039a25 332n56, 334n68 209a23 69
1039a26–8 332 244b10–245a11 286n24
1039a26 334n68 254b31 359n96
1039a27 334n68 258a5 ff. 359n96
1039a28–30 332 VII.5 291n46
1039a28 332 Rhetorica
1039a29 331n54, 332n58, 334n67 1.1–2 308
1039a30–3 332 1.1 340n2
1039a33–4 332 1377b20–7 307
1039b1–2 332 1386a10 358n92
1039b1 334n68 Topica
1039b2–5 332, 333n64 1.1 284n10, 294n58
1039b2–3 335n70 100a30 ff. 313
1039b2 334n68 101a34–6 313
1039b5–6 332 1.10 294n58
1039b6 332, 334n68, 335 1.11 294n58
1039b7 332 1.15 283n6
1039b11–14 333
[ARISTOTLE]
1039b14–16 333
Problemata 2.42 66
1039b16–17 333
1039b18–19 333 CICERO
1039b17–19 312 Academica II.130 250n96, 251
1039b20–40a7 312 de finibus
1039b20–7 336 III.50 231, 251n98, 253
1039b20 312 III.51 250, 251
1039b27 ff. 338 III.52 256
INDEX LOCORUM 383
IV.43 250n96 DK22B50 20n71, 41, 42n22, 46, 48, 56n43,

IV. 69 251 58, 63, 64
IV.79 250 DK22B51 42, 51, 54, 61, 64
DIOGENES LAERTIUS DK22B53 48, 54, 57, 60
II DK22B54 58
30 249 DK22B55 57
31 249 DK22B56 49n30
32 249 DK22B57 40, 47, 49–51, 54, 59, 64
VII DK22B58 45
1 249 DK22B59 38, 40, 46, 70
37 250 DK22B60 38, 40, 46, 47, 48, 58, 70, 131
102–3 251 DK22B61 38, 41, 45, 47, 48, 131
102 230 DK22B62 38, 52, 53, 57, 62, 72
104–5 251n102 DK22B63 53
105 255 DK22B64 60n53
160 250, 251 DK22B66 60n53
174 36 DK22B67 38, 45, 54, 56, 57, 58
IX DK22B72 53, 54, 64
22 81n46 DK22B73 53
EMPEDOCLES DK31B17 358n90, 358n93 DK22B75 53
EURIPIDES DK22B77 53
Heracles 76 43n23 DK22B78 56
Medea 663–88 8n26 DK22B79 56
DK22B80 40n15, 44, 54, 57, 59, 60, 63, 72
HERACLITUS DK22B82–3 56
DK22B1 20n71, 40n15, 42, 48, 51, 53n40, DK22B84a 60, 72
54, 56n43, 58, 60, 63, 64, 352n62 DK22B86 56
DK22B2 40n15, 42, 51, 53n38, 54, 56n43, DK22B88 38, 41, 45, 48, 52, 53, 54, 59,
58, 63, 64 60, 64, 72
DK22B3 57 DK22B89 51, 59n50, 60, 64
DK22B7 56n44, 57 DK22B90 59, 60
DK22B8 54 DK22B91 1n1, 35–8, 46, 60, 70
DK22B9 38, 40, 41, 45 DK22B99 35n2, 49, 51, 54
DK22B10 42n22, 44, 46, 48, 52, 54, 56n43, DK22B101 52, 53, 189n36
64, 70, 72 DK22B101a 57
DK22B12 1n1, 35–8, 45, 48, 54, 70 DK22B102 42, 56
DK22B13 45 DK22B103 38, 40, 46, 131
DK22B15 38, 45, 57 DK22B106 49
DK22B17 52, 54 DK22B108 43, 58
DK22B18 54, 58 DK22B111 40, 41
DK22B21 57 DK22B113 53, 54
DK22B22 58 DK22B114 56, 58, 64
DK22B23 42, 45 DK22B115 51, 53
DK22B26 53, 57, 60 DK22B116 53
DK22B27 52n37 DK22B117 59n48
DK22B28a 52n37 DK22B118 59n48
DK22B30 46, 59, 60, 64 DK22B123 58
DK22B31 20n71, 39, 46, 53n40, 59, 60 DK22B124 59
DK22B32 56nn43, 44, 56n44, 58 DK22B125 38, 40, 42n22, 46, 47, 60, 65–72
DK22B34 53 DK22B126 45, 59, 72
DK22B36 51, 59, 60, 64
DK22B40 49 HERODOTUS
DK22B41 43, 54 1.140 43n24
DK22B45 2n4, 51, 53, 58, 189n36 1.172 43n24
DK22B48 45, 56n44 3.20 43n24
DK22B49a 1n1, 35–8, 47, 48, 54, 58, 3.62 69
63, 70, 72 4.28 43n24
384 INDEX LOCORUM
HERODOTUS (cont.) 22–5 75

5.62 43n24 26–33 75
7.140–2 4 28 81n52
34–41 75, 79
HESIOD 39–41 76n19
Theogony 494 78n31 39 81n52
HIEROCLES 41 77n29
Elements of Ethics 45 ff. 209 42–9 75
50–2 75
HOMER 51 81n52
Iliad 52 81n46
1.225 129 53 78
11.638 ff. 66 54 76n19, 81n52
19.180 78n35 55 79
23.361 81n51 60 76
24.407 81n51 61 76
Odyssey 10.234 66 DK28B9–19 73
HOMERIC HYMN TO DEMETER 207 ff. 66 DK28B9 77n28
MELISSUS DK30B5 69 DK28B16 79
DK28B19 76
PARMENIDES
DK28A23 77n25 PHOTIUS Bibliotheca 526b40 ff. 66
DK28A24 77n25 PINDAR
DK28A33 77n25 Nemeans
DK28A34 77n25 7.104–6 270n54
DK28A37 77n28 8.21 79n41
DK28B1 Olympians 2.92 81n51
4 78n31 PLATO
29–30 77n30 Alcibiades
29 78n37, 81n52 132c ff. 185n28
30–2 76 133 344
30 81n52 133a 183n34
32 43n23 Apology
DK28B2–7 73 20–3 103n9
DK28B2 74 21a ff. 11n36
4 81n52 21b ff. 9n30
5 75 21c1 11n37
DK28B5 78 21c5 11n38
DK28B6 22b4 11n38
1–2 75 22d3 11n39
4–9 76 22e 11n38, 15–16
5–9 76n19 28d 340n4
5 78 30a 340n4
7 79 30b 266n42
9 78 32c 173n1
DK28B7 Charmides 8, 173–189, 290n43, 292–6,
3 78 309, 348
5 79 155c–d 184
DK28B8 74 156e 189
1–49 73 157d 183
3–6 73 157e ff. 183
3–4 75 158c–d 183
5–21 75 158e–159a 183
14 78n37 159a 13n45, 293n55
14–15 79 160d–e 183
15–16 79 160e1 186
17 81n52 164b4 173
INDEX LOCORUM 385
164c–e 10n31 Epistle 7 6n16

164c–d 173 Euthydemus 8n22, 14n50, 16n58, 17n59,
165b–c 9n30 18n63, 28, 29, 125–37, 190–206,
165c 292n51 228–57, 258–79
165c7 173 271–2 201
166c–e 9n30 272c ff. 201n34
166c–d 179 272c–d 264
166c2–3 173 272c 241, 249n92, 264
166c3 292n51 272e 264
166d 293 273c ff. 129
166d1 189 273d 264
166e5–6 173 275 ff. 203
167–9 186, 292 275a–b 241
167 ff. 210, 212n17 275a 264
167a ff. 149n20 275d–278e 200
167a 180 275d 203
167b–169a 292n50 275e 129, 132
167b11–c2 174 276a–277c 132
167c–168a 174 277d ff. 201n37
167c–d 180–1 278b6–7 264
167c 293n55 278e ff. 251n101
167c4 181 278e–280a 238
167c8–d2 293 279a–c 233n31
167c8 294n57 279a 264
167d 292 279a3 236n43
168a3–4 175 279a7 238n55
168a6–8 175 279b4 ff. 238n55
168b–d 174 279b4–c2 264
168b2–3 175 279b6 249n93
168b10–c2 176n9 279c 263
168c4–7 176n10 279d6 264
168d–169a 174 280a 242n65
168d–e 181, 292 280a6–b3 265
168d1–2 176 280a6 236
168d3 177 280a7–8 265n34
168d9–e1 174n5 280b 238
168d9–11 294 280b1 238
168d10 177, 294n58 280b2–3 236
168e 293n55 280b2 236
168e1–3 181n18 280d 263
168e4 186 280e 263
168e5–6 178 280e6 240
168e9–a1 178 281 23, 124n67
168e9–10 174 281a3 236
169d–171c 179n16 281b–e 233–4, 266n39, 269n52
169c 16 281b–d 233n31
169e 188 281b–c 238
170a ff. 177n12 281b 238, 263
173a ff. 184n25 281b1 236
Cratylus 336 281b3 236
386d 39 281b5 263n27
390c 108n28 281b6 236
398d 108n28 281b7–c3 244
402a 35n2 281b7 236, 240n63
411 48 281b8 266n38
436 111n34 281c–e 134
439 48 281c1 265n34
386 INDEX LOCORUM
PLATO (cont.) 298b–c 129

281d–e 266, 269n52, 270n53 298c 130
281d 270 298c4–5 131
281d1 240 298d–e 127
281d6–8 254 301a ff. 98
281d6 236 303b 134
281d7 236 303e 135
281e 263 304c ff. 10n34
281e4–5 243 Euthyphro 9n28
281e5 236 5a–b 8n24
282a2 269n51 5d 8n24
282c 265n35 6c–d 13n44
282d 263 10a ff. 255n117
283–8 204 10a–11b 261n18
283e–288a 133, 200 10d ff. 8n24
285c–d 129 10e–11a 142n10
285d–286b 273 15a–e 8n24
287–8 269n52 15e 16
287 92 Gorgias 8n20, 93, 97
287b 9n29, 241, 273 466–79 8n25
287d 274n62 466–70 267n47
288d–293a 204n43, 237 466e–470e 261n20
288b–c 273 467c–468e 259n8
288d 269n52 469–81 260n10
288d6–7 269n52 488b ff. 12n41
289d 269 505c ff. 16
289e 269n52 Hippias Major 287c–289d 262n25
290 28n103, 108n28 Laches 8n20
290b 274 Lysis 260n13
290e ff. 10n32 217a–218c 142n10
290e–291a 272–3 220 263n27
291a–b 9n29 Menexenus 12n43
291a 237n52 Meno 18n63, 28, 190–206, 228–57, 260n13
291b–c 238 71b 192n14
291b 135, 269 72a–b 13n44
291b6 238 75d 101n3, 108n28
292b1–2 270n53 79e 13n48
292d 202n38 80a8–b4 199
293a 134 80d 104
293b ff. 38n7 80d5–e5 192
293b–297b 200 80d5–8 191n9
293b–d 201 80d6–7 193n16
293b 200, 201 80d6 199n32
293d2–5 201 80e1 195n22
293d6 201 80e2 205
293d8 202n39 80e3–5 191n12
293e–294e 201 81b ff. 28
294d9 201n37 81c–d 203
295a5–7 202 81c6 106n22
296a–d 202 81d 203
296b 132 81d6 205
296c–d 28, 132, 202 82 ff. 11n40
296d 206 82b–86c 197
297 342n12 84a 197
297c1 ff. 353n67 84c 197
297e–298a 128–9 85c 197
298a 130 85d9–10 201n35
INDEX LOCORUM 387
86b–c 203 135e 147, 148

87–8 263n27 136e 139n4
87e 249n93 137 ff. 12n41
87d 232n25 137a 149
87e5 231n20 137b 11n40
87e6–88d3 231–2 137b2 147
88a3 236 137c–142a 147
88b4 234 137c–d 150
88b7–8 236 139b ff. 98
88c 236 139b7 157
88c2 236 139c 157
88c6 242 139c1 156
88c7 236 139c6 156
88d4–89e3 235n40 139d–e 157
88e 236 139e 156, 158
88e1 236 141d4 163
89a1–2 242 142b–155e 147
89a1 243n73, 255n118 142b–143a 150–1
89a2 234 143b7 148
97a ff. 198 144b1 164
98a 192n14 144b2 164
Parmenides 3, 4, 8n22, 84n1, 91, 138–169, 200, 144b5 164
291n47, 320–5, 329, 332n56, 337, 338 146d 158, 159
127d 139n2 146e 159
127e 140 155c 150
128 81n46 155e–157b 147
128b ff. 140 157b–159a 147
128d 140n5 157b–159b 164
129a ff. 140, 141, 246n78 158a 164
129a–b 321n35 158b4 164
129a 143n14 158c2 148
129c 141 159b–160b 147
129e 141 159b 165
130–5 271n57 160b–163b 147, 161
130 22n82 160b–d 160–1
130a–d 143, 321 160b7 161n31
130b 321, 321n35 160c6 161n31
130c–d 321n35 160e 161
130c 321n35 163b–164b 147, 161
130c2 321 164b–165d 147
130c7 321 164b–e 162
130d 143 164d 162
130e–133b 143 165a–b 162
130e 89 165a 148
131 321n35 165b–c 162
131a ff. 321 165e–166c 147
131a–e 144 165e–166b 162
132a–b 144, 321 166c 10n35, 139
132b–d 144 Phaedo 7n19, 93, 95, 97, 321–6, 329, 335,
132d–133a 321 337, 338
133b ff. 321 57a–59c 139n3
133b–134e 143 67c 323
133c ff. 22n83 72 ff. 196
134a3 175n8 72e ff. 10n33
134e9–135c2 145 73a–e 20n70
135a 144 74 ff. 94, 186n30
135c–136a 146 74 231n20
388 INDEX LOCORUM
PLATO (cont.) 275d–e 19

74b ff. 231n16 276e 108n28
74b 321n35 Philebus 8n20, 101n4, 168, 305n89,
74c 141n7 351n51, 352n63
75b10 201n37 14c 13n48
75c 321n35 17a 108n28
75d 321 19c ff. 12n42
75e 201n37 23d 333n63
76d–e 321n35 33–4 20n70
78b ff. 323n41 36c 12n42
78b–79e 142n12 38c ff. 104, 301n79
78d 244n76 38c 301
78d1 322 38c5–7 116
79c 342n12 38e 15n55
96 ff. 321n35 38e3 117
96a ff. 66, 86, 323 Politicus 101n4, 333–6, 338
96b 322n36 262 333n64
96d–97a 322 262e 323n42
96e 322, 322n36 266b 333n64
97c 322n39 266d 333n64
98b–c 89 266e 333n64
98b 322 272b ff. 149n20
98c 322n39 280b 323n42
99a–d 322 285d 108n28
99c 89 287a 108n28
100a 94 Protagoras 260, 276
100b–d 321n35 329a 19
100b5 322n39 334 13n49
100c–101c 322 334c–d 8n21, 20
100d 246n79, 322, 323 351–8 259
100e 262n25 351 ff. 241
101a ff. 321n35 352–7 237
101c ff. 323n43 358b 353n69
101d–e 323 Republic 8n21, 93, 95, 100–27, 199,
102a–b 9n27, 139n3 208–27, 229n7, 260, 261n17, 272,
102b ff. 244n75, 323 277, 338
102c–103c 142n10 327a ff. 224, 237
102d ff. 322n39, 323n41 376e ff. 83
102d7 322 Book 4 260n10
103b5 322 424a 211n15
105c 321n35, 322 430 ff. 210
107d 85 430–1 220
108c 85 432e 24n87
118a 139n3 435e ff. 360
Phaedrus 19n68, 88, 101n4 438c7 175n8
228a–b 18n62 439e 15n54
229b ff. 84–5 454a 101
229e–230a 85 454a1–9 101
242c ff. 85 474 ff. 18n64
246a ff. 18n63, 85 476–80 21n74, 174n4, 276n67
265e 333n64 476e 74n10
266b 333n63 477c–e 174n4
266c 108n28 477e–478e 336
274a8 17n60 479d 19n67
274b9 17n60 Book 6 93
274e–275a 17–18 490a–b 105
275a7 20n17 495a 24n87
INDEX LOCORUM 389
505 ff. 143 515c7 224

505a 117, 120 515d–e 224
505a2–3 101 515d 103n9
506 ff. 88 515d5–7 102
506d–507a 84 515d5 119
506e 102 515e1 103n8
507d–508b 121 516a 220
507e ff. 120 516a5 106n21
508 118 516a9 118n56
508b 120 516b 105, 108, 118, 301n78
508d 120 516b9 119
509 118 516c4-e2 211
509a–c 85 516c5 218
509b 220 516c9 214n26
509b9–10 107, 120 516b6 224
510–11 18n64 516e3–517a7 211
510a 211n14 517 118
510c5–d3 119 517a 23n84, 102
511a1 106n21 517b4 118n56
511b–c 24n87 517b8 119
511b 22n77, 102, 105, 108 517c 105, 108
511b4–c2 119 517c1 106n21, 119
511b4 102, 106n21 517d4 224
511b7 106n21 517d5 118n56
511c2 119 518c–e 116
511c5 102 518c 24n87, 107n23
511c8 118n56 518c10 118n56
511d 208 519c10 106n21
511d7 208n2 519d2 106n21
511e 211n14 520c4–5 106n21
Book 7 16, 100 523 ff. 94, 183n34
514a1–2 211 523–5 15n54, 186n31, 212n17, 301n79
514a2–3 211n12 523a–525b 110–15
514a2 24n87, 211n16 523a2–3 110
514a1–515c3 211 523a5–8 115
514a5 211 523a5 110n33, 115
514b 220 523a7 115, 116n52
514b2 212n16 523a8 115
514b5 212n16 523a10 115
514b8–9 212n16 523b ff. 103
514c4–516c3 211 523b 115
515a4 215 523b2 113
515a5–b1 214 523b3–4 110
515a5–8 212, 214 523c2–3 112n41
515a5 212n18, 214n23 523c10 ff. 111
515a6 212n16, 213n22 523d5–6 112n41
515a9–b1 218 523d5 111
515b2 212, 212n18 523e2 111n35
515b4–5 212 523e7 115
515b4 214n26 524a–c 92
515b5 212n16 524a3–4 111, 112n41
515b7–9 212, 213 524a3 110n33, 113
515b8 213n22, 214n26 524a7 115
515c1–2 212, 213 524a8 112n41
515c1 213n22 524a9–10 111n37
515c5 ff. 224 524a10 112n41
515c5 211 524b4–5 104, 113
390 INDEX LOCORUM
PLATO (cont.) 532d 102

524b5 113 533a 102
524b10–c1 114 533b 116n52
524c4 112n41, 113 533c 102
524c7 114 533c4 119
524d 114 534b 119
524d8–525a3 117 534b3-d1 124
524e1 106n21 534b3–c5 100
524e5–6 104 534d8–535a1 100
524e5 103, 115 534d9 119
524e6 119 535a 103
524e10 115 535a1 119
525a1 117, 118n56 537c–d 118
525a2 117 537c 103, 119, 211n15
525a3 117 537c7 108, 224
525b 104 537d–e 24n87
525b1 ff. 117 538–9 101n5, 103
525b3 117 595c ff. 19n65
525c1 106n21 596b 219
525c2 118n56 596b10–c1 219
525d–e 116n52 596c3 219
525d 103 596c4–7 220
525d5–6 104 596c7–9 220
525d6 103n8 596d1 220
526a 119 596d7-e4 220
526a2 104 596e ff. 222n39
526b1–2 103n8 596e–597e 219
525c6 118n56 597b2 222
526e1 106n21 597c–d 222n39
526e4 106n21 597e 83–4, 222n39
527a–b 123n63 597e3 222n39
527b 103 598a1 222
527d 103, 107n23 598a7–10 222
527d7–e3 102 598b8–c4 222
527e3 106n21 598c ff. 223
528a 102, 103, 123 599a ff. 84
528a5 119 Sophist 29, 74n12, 91, 101n4, 149, 161, 200,
528d 103 333–6, 338
528d7 119 230b ff. 21n76
529a–b 103 237–49 168
529a 116n52 242–9 333
529b3 118n56 242 20n71
529d1 103n8 243 ff. 334n67
529d5 106n21 243–5 276n67
529e 103n11 243c ff. 82n53
529e5 106n21 244 ff. 98
531–7 101 244b–245d 140n6
531d 108 244b–d 8n23
531d2 119 246e 113n44
531e–532a 103 251d 334
531e4 119 251d7 334n69
532a5–b4 100–101 252c 334n67, 334n69
532b1 106n21 252e 334, 334n69
532b2 119 253b 334n68
532b4 102 253d–e 108n28
532c6 118n56 253d 323n42, 333, 334n68
532d–533a 103n11 254 ff. 168
INDEX LOCORUM 391
255 ff. 98 204b11 ff. 326

255e ff. 334n66 205c 327
256a1 334n69 206b 327
263 15n55 210 ff. 168
263e 104, 301n79 201c–202a 98
Symposium 4, 8n22, 84n1, 139–40, 341, 209a 113n44
351n51, 354nn70, 71, 358–60 Timaeus 85, 93, 95–9, 143, 149,
187a 42n22 322n39, 338
189–93 358 27d 95
190–2 358n92 29c–d 85, 95
191d1–3 358n92 35a ff. 168
191d2 358n93 35 337
191e9 358n92 49a ff. 168
193c5 358n92 49d 48
193d5 358n89 52a–c 97–8
201d ff. 11n40 92a3 333n64
210–12 21n74
210e 105 PLOTINUS Enneads V.5.1 62–5, 21n75
Theaetetus 8n22, 8n23, 17n59, 29, 149, 200, PLUTARCH
212n17, 301–3, 308–9, 326–8, 329, 338, de communibus notitiis
348n35 1071F-1072A 250n95
152 ff. 48, 168 Lives 29.3 353n67
153d ff. 22n80 SEXTUS EMPIRICUS
154 ff. 174n6 Against the Mathematicians
154–5 106n22 7.6 82n55
157 48 11.64–7 251
169–86 22n80 11.64 250
172a–177b 306 Outlines of Pyrrhonism 82n54
176b 302
SIMPLICIUS de caelo 557.20 81n46
176e4–5 306
177a1 306 SOPHOCLES Electra
181e ff. 39 644 81n46
184 ff. 88 1067 78n36
184–6 112n40
184b ff. 22n78, 90 STOBAEUS
184b10 301n77 2.82.20–83.4 253n107
185b8 301n77 2.84.2 253n106
185b10 301n77 2.84.18–85.11 254
186b 301 2.85.11 256
188 ff. 113n43 THEOPHRASTUS
189–90 104 de igne 4, 6, 69
189e–190a 15, 116 de sensu 52, 69
189e7 116 de vertigine 9–10, 65–72
190a 301 Metaphysics
190a2–3 301 5b14 69
190a3 117 6a12 67
191c ff. 301n81 7a15 69
191d4–7 301n81 8a21 67
197c ff. 301n81
201–2 336 THUCYDIDES 2.41–2 81n51
201e ff. 326 ZENO OF ELEA DK29B4 43n23
General Index
Academy, the, 26, 249, 295, 321, 338 belief 6, 14–15, 18n64, 22, 30, 84, 94, 95,
accounting, accountability, 9, 13, 20, 23, 41–3, 110n32, 116, 117, 118n55, 119, 174, 175,
49, 52, 53n40, 56, 58, 87, 100, 102, 105, 130, 183, 184, 186, 187, 190, 191, 197, 198, 199,
134, 177, 190, 196, 202n38, 204, 207, 206, 226, 208, 219, 262, 278, 301, 336, 338
230–7, 310, 317–24, 325, 327, 329, beliefs, belief-set 91, 92, 104, 134, 135, 167
334, 335, bewilderment, see aporia (impasse,
Adamson, P. 189n38, 309n93, 339n71, 365n110 bewilderment)
allegory 83–99 Blackburn, S. 189n38
allusion 3–4, 229, 247, 328, 329, 331, 335, 336, blindness 23n84, 224
338, 341, 342n12, 351n51, 352n63, 354n71, Bollack, J., and Wismann, H. 49nn30, 31,
358n89, 360 51n34, 57n45
Annas, J. 64n57, 107n25, 219n34, 229n6, Bostock, D. 5n13
231n15, 237n48, 238n54, 240n63, 249n90, Brandwood, L. 101n3
258n3, 260n12, 261n17, 264n28, 353n67, Brentano, F. 24, 297
359n95, 365n110 Brittain, C. 227n46
and Rowe, C. 259n5 Broadie, S. 124n68, 264n28, 268n49, 279n71
Anscombe, G. E. M. 238n53, 261n21 and Rowe, C. 345n24
answer 90, 102, 103, 104, 111n37, 114n46, 132, Brown, L. 101n3, 124n69
213n20, 322 Brownrigg, M. 309n93
simple-minded, safe 87, 88, 89, 93, 94, 98, Brunschwig, J. 185n28, 209n3, 217n32
262n25, 322, 323, 324, 326 Burnyeat, M. 5n13, 15n56, 19n68, 24n90,
cleverer 322, 323, 324, 325, 335 26n99, 30n106, 56n42, 81n48, 82n56,
see also question and answer 90n5, 104n12, 106n22, 107n24, 108n26,
antilogic 101, 102 110n29, 112n40, 118n57, 121n61, 122n62,
aporia (impasse, bewilderment) 7, 10, 17, 102, 123n65, 124nn67, 68, 173n2, 178n14,
103, 138, 139, 148, 160, 167, 168, 199, 183n24, 189n38, 217n32, 218n33, 227n43,
270n54, 273, 295, 296, 297, 313, 328, 228n4, 232n22, 237n49, 262n26, 264n28,
351n54, 357n87 266n42, 283n6, 284n9, 286nn22, 23,
argument 3, 4–10, 30, 35–40, 44, 48, 50, 54–5, 288n33, 299nn72, 73, 327nn46, 47,
61, 63–4, 83–99, 101, 102, 103n8, 125, 126, 363n106
127, 129, 131, 132, 134, 136, 140, 144, 147, et al. 316nn11, 14, 319n28, 332n60
148, 149, 150, 151, 163, 164, 165, 166, 167,
168, 191, 203, 205, 228, 242–3, 247, 249, Caizzi, F. D. 233n28
257, 274, 310–11, 313, 323, 338 Campbell, L. 22n81
conditions of 9 Canto, M. 239n58
framed, see frames, framing of dialogues Caston, V. 24n89, 283nn1, 3, 4, 284n11,
lazy 203, 205 285n15, 286nn22, 24, 287nn25, 26,
relations 28–9 289n40, 290nn42, 43, 292nn49, 50,
sequential 20 297n69, 306, 342n10, 345n25
Armstrong, C. J. 342n13 causation 22n77, 86, 106, 107, 109, 113, 114,
Atkins, M. 207n45 119–24, 178n15, 216, 218, 220, 251, 252,
Aufderheide, J. 20n70, 31n109 254, 265, 267n48, 268, 271, 272n58, 289,
291, 293, 302, 344, 347
Badham, C. 233n29 cave 101, 107, 110, 118, 123, 208–27, 311,
Bailey, D. 173n2 314–25, 326, 328, 329, 335, 337
Balogh, Z. 279n71 Chappell, T. 124n69
Barker, A. 289n38, 300n75 ‘Chinese whispers’ 4, 84n1, 138, 139, 149, 166
Barnes, J. 5n14, 38n6, 42n20, 73n1, 80n45, 125, chronology of the dialogues 27–9, 104n13, 200,
262n23, 284n9, 340n1, 343n15 228–30
Baumann, P. 309n93 changing one’s mind 28, 229nn7, 8, 276n67
394 GENERAL INDEX
chronology of the dialogues (cont.) Descartes, R. 73

developmentalist 27, 200, 228, 229nn6, 7, 8 detachment, reflective, see reflection,
late (critical) period 27, 200 philosophical
middle period 27 Détienne, M., and Vernant, J.-P., 78n31
Socratic period 27, 229 developmentalism, see chronology of the
unitarian 229n10 dialogues
Cleary, J. 124n69 dialectic 7, 9, 24n87, 26, 28n103, 60–4, 74n7,
cognition 175, 176, 179, 180, 197, 204, 211n14, 77n24, 78n31, 79, 80, 81, 90, 91, 92, 93, 94,
212–13, 221, 224, 226, 227, 251, 261, 306 95, 99, 100–24, 134, 135–7, 140, 145, 146,
comedy 1n2, 8 148, 149, 153, 165, 166, 167, 184, 187, 203,
compresence of opposites 16n58, 142, 143, 211n15, 224, 226, 230, 273, 275, 288n33,
231n16, 233n33, 243, 244, 251n98, 294n58, 295, 296, 307, 310, 311, 313, 314,
252n103 315, 318, 320, 322n37, 323, 324, 326n45,
consciousness 24–6, 182, 188, 210, 225, 284–7, 327, 328–31, 333, 334, 338, 340, 341, 348,
288n35, 291, 294, 297n69, 304, 305, 306, 352n63, 354n72, 357, 359, 362
341n5, 345n28, 346n31, 348, 361, 362 dialectical context 2, 3, 81, 311, 335, 336, 337
phenomenal 188n33, 285, 286, 288, 289, 290, dialectical device 73, 77, 82
294, 295, 300, 301, 304, 305, 307, 309 dialectical partners 26
self-consciousness 10, 112, 188n33 dialecticians 272
subjective 24, 210 rules of 73–6, 81
‘what it is like’ 24 dialogue 74, 93, 125, 126, 168, 223, 228–30,
consequentialism 232nn22, 24, 234, 235, 295, 309
236n41, 237–47, 255, 256, 257, 258–79 form 8–12, 21, 24, 27
consistency, inconsistency 91–5, 97, 99, 104, internal 15–17, 20, 91–2, 102n6, 104, 115,
134, 135, 273, 274 116–18, 183n24, 301–2
contradiction 8n22, 9, 10n35, 16, 29, 37, 38, Platonic dialogues passim
44–8, 49n30, 50, 54, 61, 63, 76n19, 79n43, Diels, H. 40n13, 56n44
80, 82, 92, 94, 103, 111n37, 112, 132, 133, disagreement 3, 8, 136
134, 141, 151, 166, 200, 204, 244n75, 272, logic of 2
273, 335, 341; see also Law of discovery 190–6, 197, 198, 199, 200, 203–6, 207
Non-Contradiction discrimination, see judgement
contradictories 37, 39, 40, 45, 47, 50, 55, 58, 87, distance, reflective, see reflection, philosophical
131, 141, 147, 148 divided line 103, 107, 110, 119
conversation, philosophical 1–31, 95, 100–24, Dixsaut, M. 104n12, 116n51
132, 134, 135–7, 272, 274, 310–39 drama 1n2, 92, 168
conditions for 103, 212–15 dramatis personae 140–6
internal, see dialogue, internal of the dialogues 12, 13, 125, 126, 132, 173n1,
Coope, U. 309n93, 339n71 179, 183
Cooper, J. 106n22, 112n40, 169n34, 231n13, doxa, as contradictory of paradox 36, 37, 45, 47,
233n32, 345n22 50, 57n45, 64
Cornford, F. M. 105, 139n2, 147n19, 160n29 Doyle, J. 124n69
cosmology, 39, 56n44, 57n45, 58–60, 76, 77, 85,
94, 95, 96, 203 education 208, 209, 211, 213, 223–5, 227, 273
Coxhead, M., 22n81 philosophical, 21, 110n29, 208, 209
cradle arguments, 209–11, 213, 216 see also learning
craft, craftsmen 11, 16, 174n2, 218, 219, 220, emotion 174, 176, 179, 196, 261, 342
221, 222–3, 265, 270, 271, 27 ‘episodism’ 19n69, 204
divine, 93, 97, 219, 221 epistemology 9, 11, 22, 30, 50, 59, 82, 90, 95,
see also skill 103, 104, 105n19, 111n36, 121n60, 123,
creation, creator 93, 96, 222; see also demiurge 190–207, 219, 278, 307, 343
Cross, R. C. and Woozley, A. D. 105n18 externalist, see externalism
internalist, see internalism
death 93, 230 eschatology 94, 97
of Socrates 7n19 esoteric doctrines 6n16
demiurge 96, 97, 98 experience 10n33, 22, 23, 24, 25, 42, 52, 53, 78,
demonstration 25n91, 90, 312 81, 106, 108, 111n34, 184, 196n24, 213n22,
Denyer, N. 46n27, 73, 169n34, 185n28, 189n38 216, 278, 355n75
GENERAL INDEX 395
explanation 13, 14, 17, 20, 23, 50, 61–4, 79, 83, another self, another I, 342–8, 353–8, 360–5
85–90, 91, 94, 96, 98, 99, 120, 121, 122, 125, as a mirror 342–4, 347, 360–4
138, 143, 144, 146, 192n14, 194, 215, 228, virtue-friendship 25–6, 349n41, 351n53
244, 246nn78, 79, 252, 256, 261, 262, Furley, D. 210n7, 359n96, 360n99
265n31, 268, 270, 271, 275, 276, 277, 278,
284, 285, 291n48, 298, 302, 322–6, 329, Gallagher, P. 279n71,
335, 337, 341 Galloway, D. 279n71, 365n110
(not) self-explanation 17, 94 Geach, P. 262n26
externalism 191–9, 204, 205, 206, 224n40, 225 Gerson, L. 21n73, 208n1
Gettier, E. 22n81
failure 3n7, 11, 13, 28, 135, 136, 199, 223, 236, Gifford, E. H. 233nn28, 29
238, 245, 259, 261, 276, 348, 356, 358, Gill, C. 124n69, 169n34, 279n71
359, 365 Gill, M. L. 316n14, 317n20, 330n53
fallacy 127, 128, 136, 157, 262n26 god, gods 42, 43n23, 61, 89, 95, 96, 97, 99, 219,
falsehood 29, 30, 92, 200, 203, 204, 301n77, 308 220, 222, 342, 348–53, 356–64
fiction 127, 136, 184n25, 226 goddess, in Parmenides’ poem 73–82
Fine, G. 4n8, 14n51, 26n95, 105n16, 107n25, ‘god’s eye view’ 56–9
111n37, 118n55, 159n27, 190, 191, 193n15, godlike person 349, 357, 364
197n25, 198n28, 207n45, 208n1, 219n34, on not being god 356, 358–60, 362
295n62, 321nn32, 34 self-sufficiency of god 356, 359
first-order (assertions, claims, arguments, Goldie, P. 363n106, 365n110
responses, perception etc.) 2, 4, 6, 9, 12, Gomez-Lobo, A. 101n4
15, 20, 23, 37, 45, 48–51, 55, 58, 61–4, 116, good 21, 24, 88, 92, 93, 96–7, 100–24, 138, 179,
177n12, 178, 188, 198, 199, 204, 207, 189, 200, 202, 230–57, 258–79, 295, 301,
283n3, 285, 287, 288, 289, 290, 296–301, 303, 307, 308, 321, 345–63; see also value
302, 305, 305, 307, 308, 347n34, 361, 362 conditional 134, 248
flux 39, 46, 47, 48, 59, 61, 62, 95 consequential 234, 235, 236n41, 237–47, 255,
Foot, P. 238n53 256, 257, 258–79
forgetting, forgetfulness 8n21, 13n49, 18, 20, derivative 239, 240, 243, 244, 246, 247, 248,
196n24 252, 253, 254, 255, 256n119
Frankel, H. 41n16, 78n32 form of the good, see form, forms (ideas)
Frede, D. 5n12, 36n5, 64n57, 72n3, 104n12, exclusive 234n34, 243–51
107nn23, 25, 115n48, 124n69, 169n34 extrinsic 235, 246, 247, 250, 253, 254, 255,
Frede, M. 106n22, 112n40, 249n90 267n43
and Patzig, G. 315n10, 316n15, 317nn17, 21, in itself 227, 235, 238, 243, 244, 246, 252,
318n24, 319n27, 332n60 253n107, 269, 271, 273, 274, 275, 277, 345
Frege, G. 125, 126, 228n2 indifferent, see indifferents, Stoic theory
Fontenrose, J. 4n9 instrumental 232n21, 233n28, 234, 235, 236,
form, forms (ideas) 21–23, 26, 87, 88, 94, 98, 99, 238, 239, 241, 242, 243, 246, 255, 259, 263,
101n4, 138–46, 147, 148, 149, 151, 162, 267n45, 270, 271, 274, 343, 344, 364
209, 219, 231n16, 246n78, 312–13, 321, intrinsic 124n68, 126, 231n14, 232n22,
322, 323, 326, 331–5 235n38, 246, 247, 250, 253, 255, 257,
Aristotelian 310 267n45, 268, 277, 347
contact with 21, 22 knowledge of the good 23, 122–4
direct relation to soul? 21 learning to be 25, 227n43
‘greatest difficulty’ 22n83 putative (apparent) 230n12, 233n28, 234,
knowledge of 22, 209 238, 240, 241–5, 248, 249, 251, 254, 255,
of the beautiful 22 264, 266, 269n51, 274, 275
of the good 84–5, 88, 93, 104–10, 116, 117, seeing the good 100–24
118, 119, 120, 122, 124, 220, 224, 276, 277 sovereign 110, 123–4, 260, 268
theory of 27, 107, 138–46, 149, 155, transformative 23n85, 122–4, 267n43, 274
167,168, 321 unconditioned, unconditional 248, 251,
Forster, E. S. 65, 66n2, 68 265, 267,
frames, framing of dialogues 8n22, 9, 10, 12, 17, Goody, J. 5n10
18, 20, 24n87, 181, 182, 183, 187, 188, Gower, O. 124n69, 324n44
211n12, 212n16, 217n31, 262, 270, 275, 293 Greenberg, A. 22n81
friends, friendship 4, 25–6, 220n36, 340–65 Guthrie, W. K. C. 39n12, 59n46
396 GENERAL INDEX
Hamlyn, D. 283n1, 284n9, 287n30, 288n35 161–3, 164–9, 312, 318, 319, 332, 353n67,
Harte, V. 31n109, 124n69, 189n38, 207n45, 354n71, 359, 360n99
211n14, 212n19, 213n20, 214n24, 216n28, austere 142–5, 149, 151–60, 164–9
219nn34, 35, 227n46, 309n93, 339n71 generous 142–5, 149, 151–60, 164–9
Hartman, E. 47n28 inquiry 7, 13n48, 14, 90, 103, 104, 107, 108, 122,
happiness (eudaimonia) 23, 93, 123, 218, 232, 179, 187, 189, 190–207, 262, 269, 278, 295
233n28, 234, 235–40, 247, 248, 255, paradox of 54, 58, 190–207
258–61, 263, 265, 267nn43, 47, 268–70, see also paradox, Meno’s
272, 275–8, 345, 358n89 inspiration, divine 11
Havelock, E. 5n10, 229n9 intellectualism 189, 255, 258–60, 276
Hawtrey, R. 233nn28, 29, 32, 236n45, intentionality 185n28, 196, 204, 306
237nn49, 50 intentional objects 174n3, 194, 195, 296
hearing 20n71, 175, 177, 178; see also listening intentionalist model of teleology 96
Heinaman, R. 314n4, 316n13, 317n18, 332n60, interlocutors 2, 4, 8, 10–14, 17, 19n67, 20, 102,
365n110 201n36, 274
Halliwell, S. 294n58 internalism, internalist conditions 191, 194–9,
Hejawi, H. 309n93 206, 210, 224, 273, 274, 277
Hett, W. S. 284n11 on knowledge 11, 225
Hicks, R. D. 283n1 interrogative 12–17, 191, 192, 193n15, 200
higher-order, second-order (knowledge, interruption 10n32, 262, 270n53, 272–3
thought, reflection, stance, levels of intertextuality 4, 5, 7n18, 26, 28, 322n39,
thought) 2, 3, 9n28, 11, 12, 13, 16, 17, 22, 328, 329
25, 28, 45, 48–51, 55, 57, 58, 61–4, 116, 117, telegraphed 28
119, 120, 121, 166, 175, 177–82, 186–9, Ioppolo, A. M. 251n100
199, 200, 204, 205, 206, 207, 243, 288–91, irony 1, 237n52, 241, 269n52
293, 294, 296, 299, 302, 303–8, 352n61, Irwin, T. 111n37, 123n64, 233n28, 236n41,
362, 363 258n3, 259n6, 264n28, 267n46, 345n22,
history of philosophy 29–31, 258, 286, 292 353n67
fractures in 30
scientizing account 30 Jaeger, W. 310n2, 314n6, 316n15, 317n17
Hossack, K. 169n34 Jameson, G. 81n52
Housman, A. E. 8n26 Johansen, T. 124n69, 283nn1, 3, 4, 284n11,
Hughes, G. 316n11, 319n28 286n24, 287nn25, 26, 31, 289nn40, 41,
Hume, D. 226 290n43, 291n47, 292n50, 297n67, 303n86,
Hursthouse, R. 238n53, 262n22 306, 342nn10, 14, 345n25, 348n36
joke 1, 2, 83, 136, 341
identity 1, 2, 31, 47, 48, 62, 63, 80, 88n3, 91, 130, judgement 15n56, 22n78, 90, 93, 97, 99, 103,
153–64, 166, 213n22, 232nn21, 23, 236n45, 104n14, 110n32, 113, 116, 117, 183n24,
242, 243n73, 274, 277, 320, 334, 335, 344, 262n23, 297–308, 342, 363n106
353n67, 358, 359
ignorance 9n30, 11, 16, 50, 51, 52, 53, 85, 174, Kahn, C. 27n101, 28n103, 35n3, 38n8, 40n13,
184, 203, 232–43, 249, 254, 255, 263, 42n21, 43nn23, 24, 49nn31, 32, 51n34,
266–7, 269, 276, 336, 338 52nn36, 37, 53nn38, 39, 40, 36n44, 60n54,
illusion 73, 76n19, 94, 111, 115, 124 74n8, 75n12, 78n37, 80n44, 229nn6, 10,
imitation 19n65, 83, 217n31, 219, 222, 225 237n48, 259n5, 284nn9, 11, 286n18,
impasse, see aporia (impasse, bewilderment) 345n28
incomplete Kant, I. 231n13, 248, 255n114, 344, 365
argumentatively 95, 99 Kapantais, D. 214n25
cognitively incomplete 94, 113 Kirk, G. 35nn2, 3, 38nn8, 9, 39n12, 40nn13, 15,
nature 361, 362 41n19, 42n21, 43nn23, 24, 49nn30, 31, 33,
indeterminacy 39, 44, 45, 46, 48, 61, 229, 51n34, 59n46, 67, 68, 69
236n43, 257 Kirk, G., and Raven, J. 76n23, 78nn31, 32
of translation 30 knowledge 10–12, 14, 15, 18, 21–4, 28, 30, 31,
indifferents, Stoic theory 230, 231, 247–57 49, 50, 51, 52, 53, 55, 57, 61, 93, 95, 98,
‘preferred’ and ‘dispreferred’ 230, 250–6 104–10, 116, 118n55, 120, 132, 144, 150,
individuals 46, 47, 48, 61, 62, 63, 64, 111n37, 153, 160, 161, 164, 173–89, 190–207,
114n46, 140, 142, 143, 145, 149, 151–60, 208–27, 229, 232–7, 242, 244, 246, 247,
GENERAL INDEX 397
249, 265n36, 269–73, 275, 276, 278, Leigh, F. 31n109, 227n46, 365n110
292n51, 293, 304, 306, 318n24, 326, 327, Lesher, J. 64n57, 72n3
328, 333, 336, 338, 342n14, 343, Levett, M. L. 327n46
351n57, 352 life 23, 110, 123, 124, 202, 217, 218, 224, 225,
‘all-or-nothing’ 192 227, 230, 231, 239, 240n63, 241, 244, 245,
and perception 21–4, 90, 104–22, 188, 251, 253, 254, 255, 256, 268, 271, 272,
208–27, 292n51, 293, 294, 295, 296 274–6, 278, 303, 305, 306, 308, 345–7, 350,
and virtue 25, 122–4, 206 352, 354, 357, 363–4
as understanding 19n68, 122–4, 173n2, ‘how best to live?’ 6, 239n57, 268, 278, 306,
202n39, 225, 293 307, 308, 349
disavowal of 103n9 normativityof, value of 124, 345n24
explanatory role of 13 listening 20, 363; see also hearing
externalist account of, see externalism literacy 5, 17–20
foreknowledge principle 190n5 Lloyd, G. 169n34
higher-order 175–9, 182n22, 188, 204, 293, logic, principles of, conditions for 90, 91, 92,
296; see also higher-order, second-order 103, 114, 118, 125, 126, 128, 130, 132, 133,
(knowledge, thought, reflection, stance, 134, 136, 137, 167, 176n9
levels of thought) Long, A. 124n69
holistic 187n32, 189, 192, 202n40, 206, 225 Long, A. A. 75n17, 76nn20, 21, 22, 23, 77nn25,
knowing that one knows 11, 190–207 27, 231nn14, 15, 243n72, 249n90, 253n107,
internalist account of, see internalism, 254n113, 264n28
internalist conditions Long, A. A. and Sedley, D. N. 230n11, 248n84,
knowledge of knowledge 173, 174, 175–9, 253n105, 108, 254n110,
189, 205n44, 292, 293 luck 202, 258, 262–6, 272, 274, 276
objects o, 208–9, 224, 225 Lycan, W. G. 285n14, 286n20
of the self 173, 187–9 Lyons, J. 174n2
reflective 180–9, 204, 205, 295, 296
self-intimating 107 McDowell, J. 363n106
self-knowledge 173–189, 202n38, 208, 209, MacIntyre, A. 248n89
211, 217, 221, 225, 226, 227, 292, 306, Mackie, H. 189n38
340–65 Marcovich. M. 35n2, 40n13, 41nn18, 19,
simple 208 43nn23, 24, 49n31, 66n2, 67, 68
state of mind 21, 206, 208 Mates, B. 154n25
see also understanding; wisdom Matson, W. 75n17, 76n19
Knox, B. 64n57 Matthews, G. 102n6
Korsgaard, C. 124n68, 235n38, 240n61, Meinwald, C. 147n19, 150n21
246n78, 248n87, 255nn114, 116, 267n45 Mellor, H. 189n38
Kosman, L. A. 24n89, 182n21, 283nn1, 3, memory 5, 17–18, 20–1, 26, 197; see also
284n11, 285n12, 287nn25, 30, 288n35, recollection
289nn40, 41, 295n61, 297n68, 306, 341n5, metaphysics 155, 163, 167, 177, 178, 226, 235,
342n10, 344n21, 345n25, 346n31, 348n36, 242, 254, 257, 258, 259n8, 260, 263, 267,
351n57, 353n67, 364n108 276, 277, 278, 307, 321n35, 326, 329,
Kraut, R. 261n17 337, 343
methodology (philosophical method) 9, 77, 79,
Lacewing, M. 279n71 100, 104, 108, 109, 219, 262, 277, 284,
Lacey, A. 309n93 333n63, 336, 341
law of excluded middle (LEM) 44 mind 87, 89–93, 105n16, 107, 110, 114, 115,
law of non-contradiction (LNC) 30n107, 44–8, 116, 118, 120, 122, 191, 192, 193, 194,
61, 64, 131, 133, 150, 201 197, 198, 206, 214, 219, 226, 269, 276,
gross 132, 133–4, 136, 150 278, 293, 301n81, 306,
refined 133–4, 136 342n14, 343
Lear, J. 82n56, 124n69 Mind, Anaxagorean 88, 89
learning 18n62, 21, 24n90, 25, 91, 101, 106n22, ‘missing persons’ 3n7, 15n54
117, 120, 123, 175, 195, 197, 198, 200, 201, monism 40, 48, 63, 73–82, 91, 140n5, 141, 142,
202n41, 203–7, 227n43, 235n38, 249, 313, 146, 149, 165, 167
362, 363n106; see also education Moore, G. E. 235n38, 246n78
Lee, K. 43n23, 72n3 Moravcsik, J. 21n73
398 GENERAL INDEX
Mourelatos, A. 49n29, 51n35, 73n1, 76nn20, paradox 1–4, 9, 10–11, 35–64, 65–72, 76, 79, 80,
22, 77nn26, 27, 78n32, 81n50 82, 131, 136, 147, 153, 156–60, 163, 164,
myth 6n15, 18n63, 75, 83–99 165, 188, 217, 223
dilemma 73–82
Nagel, T. 24n89, 182n20, 265n31, 285n13, liar 2n6
348n36 Meno’s, paradox of inquiry 11, 14, 105, 121,
Nails, D. 173n1 190–207
Narcy, M. 237n49 reflexivity 216–17, 225, 292
narrative 4, 10, 262 resolution of 3, 38, 39, 40, 44–6, 47, 50, 54,
reported 8 55, 61
natural inherence 142n10 river 1–2, 20, 35–8, 39, 40, 45, 46, 47, 49, 50,
nature, natures 96, 97, 101, 142, 143, 144, 146, 51, 54, 55, 61, 62, 63
149, 153, 156, 157, 158, 163, 164, 168, 209, writing 18, 20
211, 213, 222n39, 249n90, 250n94, 255, Zeno’s 147–8
261, 277, 303, 307, 308, 335, 337, 342n14, Parker, R. 4n9
345, 347n34, 352, 353, 354n70, 356–7, 359, parsimony 143, 144, 145n17, 166, 168, 243, 320,
361–2, 365 321, 329
normativity of 25 particular, particulars 16n58, 17, 19n67, 23, 94,
necessity 89, 93, 96, 97, 98 111n37, 140–6, 148, 149, 151, 155, 157,
Nehamas, A. 107n24, 121n61, 173n2, 190n6 162, 166, 183, 242, 244, 278, 318–23, 325,
Nightingale, A. 102n7, 105, 123n66, 330n52 334, 335, 336
nihilism 74, 82, 91, 167, 168 parts 140–4, 150–60, 162–4, 209, 210, 217, 327,
normativity 25, 101, 103, 104n14, 117n53, 122, 337, 353n68, 359n96, 360n99
210, 211, 225–7, 304, 306, 308, 345n24 Penner, T. 124n69, 261nn19, 20, 279n71
Nussbaum, M. 42n20, 51n35, 53n39, 75n17, Penner, T., and Rowe, C. 123n64, 258n1,
260n9, 262nn22, 23, 24, 265n31 258n3, 259nn7, 8, 260nn13, 16, 275n65
perception 16, 21–6, 57, 80, 82, 90, 95, 96, 97,
O’Brien, D. 137n3 106–24, 161, 173–89, 209, 212nn17, 18,
O’Shaughnessy, B. 209n5 213, 251, 283–309, 340–65
objectivity 258–62, 268n50, 275–9, 340, brute 22, 23, 178n15, 180, 181, 182,
344n20, 351n57, 354n74, 356n84, 360 183n24, 185–7
observer 3, 20, 118 capacity 24, 25, 180–9, 283n4, 284n11, 302,
Ockham’s razor 89, 94 303, 352, 362
omniscience 132 civilized 180, 181, 182, 186–7, 187, 188
one 138–42, 144, 147–69 cognitive 22n78, 24, 25, 180–9, 210,
Ong, W. 5n10 212nn17, 19, 215, 226, 293, 299, 300, 302,
ontology 59, 61, 74, 77, 97, 98, 99, 140–5, 303, 305, 307, 309, 344, 346n31, 364n109
148–50, 151, 160, 165–7, 169, 174n3, common objects 301, 308
315, 318, 319, 320, 321, 328, 329, 333, common sense 284n11, 297n68
334, 335 ‘cooked’ 24, 300n74, 343–4, 352n60, 363
opposites 40–4, 45, 48, 49, 51, 61, 64, 79, 88, faculty 283n4, 284n11, 287n26, 289n38, 290,
135, 140–3, 151, 153, 159, 160, 163, 164, 291, 297, 298, 300, 306, 352n61, 363
176, 240, 243, 248, 254, 323, 352n59; see first-order 283n3, 285, 287–90, 296–301,
also compresence of opposites 302, 304, 305, 307, 308, 347n34, 361, 362
opposition of unity (in Heraclitus) 40–4, 48, higher-order, second-order 180–9, 283,
50, 54, 55, 57, 60, 284n10, 287n31, 288–91, 293, 294, 299,
unity of, 40–4, 47, 48, 50, 51, 52, 54, 55, 56, 302–8, 352n61, 362, 363
57, 58, 60, 61, 64 joint 25–6, 356, 358, 360–5
oracle 11, 16, 58 knowledge as 21–4, 224
oral culture, orality 5, 26, 28 moral 25, 224, 363, 365n110
Osborne, C., see Rowett, C. (formerly Osborne) non-cognitive 22n78, 180–9
Owen, G. E. L., 27, 59n47, 82n56, 147n19, normativity of 25, 225–7, 304, 306,
156n26, 262n23, 313n3 345n24, 361
objective 209–10, 344n20, 351n57, 354n74,
painting 19, 219, 222 356n84, 360
Papineau, D. 210n9, 286n20 perceiving that we perceive 283–309
GENERAL INDEX 399
perceptible object 73, 175, 210, 222, 224, plurality 141, 142, 144, 163, 164, 165, 168,
284, 287, 288–91, 292n49, 293, 299, 297n68
300, 301, 305 pluralization 142, 150, 151, 156–60, 289
piecemeal 22, 224, 285 poets, poetry 83–4, 86
‘raw’ 22, 24, 106, 109, 110, 112, 115, 116, 119, point of view, perspective 16, 126, 140, 186n30,
120, 121n61, 122, 183n24, 212n17, 226, 209, 213, 216, 218, 219, 222, 223, 224, 225,
293, 300n74, 301, 302, 303, 308, 309, 343, 226, 227, 273, 285, 286, 310, 313, 319, 331,
344, 361, 363, 364n105 364, 365
self-perception 173, 175, 176, 178, 180, 208, object 25, 120
210, 211, 215–17, 221, 222, 223, 226–7, subject, subjective 24, 25, 209n5, 210, 227
287–91, 292, 294–8, 302, 304, 340–65 Politis, V. 102n6, 103n8, 114n47, 120n59,
special objects 175, 177, 178, 297, 299, 302, 124n69, 189n38
304, 308 posset 65–72
subject 286, 287nn29, 31, 288, 291, 293, 299, pretenders to knowledge 11, 12, 16, 184n25
300, 301, 305, 341n5, 343, 344, 346, 347, Price, A. 279n71, 345n22
351n57, 354n74, 355n81, 361, 364 properties 142, 144, 147, 150–60, 161, 163, 164,
subjective perception 209–10, 285, 286, 288, 165, 168, 176, 246n78, 252, 302, 322–4,
293n54, 304, 306, 340, 351n57, 354n74, 332, 337
356n84, 361 proposition 14, 19, 22, 55, 86, 112, 131, 135
‘twice-cooked’ 24, 344, 348, 361, 363 propositional content 22, 109, 110, 114, 115
unmediated 21, 22n77, 23, 105n18, 106, 107, non-propositional thought 23
109, 115, 118, 120, 121, 212n17, 226 proprioception 209, 210, 222, 225
perfection, perfectibility 25–6, 244, 359, protreptic 134, 272, 274
362n105 provocation 1, 2, 3, 17, 114, 119, 136, 221n37
perspective, see point of view, perspective Pryor, J. 216n29
persons 74n7, 75, 80, 88n3, 135, 136, 145, 146, psychology 9, 86, 87, 90, 92, 104n14, 109, 121,
150, 167, 265, 271, 274–7, 291n48, 303, 174, 175, 177–82, 184, 185, 186, 229,
306, 344 242n67, 248, 249n90, 258n2, 259n8, 260,
first-person 225, 272n59, 284n10, 291n48, 261, 276, 341
304–6, 330n51 conditions of dialectic 103, 107, 110, 122,
identity of persons 274–7 123, 124, 193, 194, 196
impersonal 218, 225, 226, 284n10, 291, of conversation 3, 14, 15–16
292n52, 305, 352 of interrogative 14, 16–17
second-personal reference 73 moral 353n69
third-person 283n2 of perception 118, 120, 187n32, 173–89
phenomena, phenomenal world 22–3, 76, 77, of puzzlement, paradox, etc. 10, 45, 64,
80, 81, 82, 87, 88, 98, 107, 111n37, 142n11, 114, 167
149, 197, 198, 200, 204, 284n10, 286, pun 1
291n48, 294, 302, 305, 322n38, 341, 348, puzzle, puzzlement, perplexity 2, 4, 5, 9n27, 10,
349, 350n44, 352n63, 353, 355, 356, 364 14, 16, 17, 20, 29, 30, 37, 58, 60, 61, 64, 73,
moral 260–2, 268, 275–78 79, 86, 87, 90, 92, 94, 103, 112–19, 121, 122,
philosopher, the 21–4, 83, 88, 89, 92, 103, 104, 141, 143, 144, 145, 152, 163, 165, 167, 168,
105, 107, 109–10, 117, 118, 119, 120, 176n10, 178, 189, 191, 192, 193–6, 197,
122n62, 209, 211, 217–18, 224–7, 274, 276, 198, 200, 207, 208, 212, 221, 225, 226,
302, 306 240n62, 261, 270, 272, 278, 291–301, 313,
philosophers rule 22, 94, 108, 123, 226, 272 321, 322, 324–34, 338, 341, 348–57, 364
as solitary 20–1, 23
philosophical principles 143, 152, 156, 166, 169 question and answer 12–14, 91, 92, 100, 102,
philosophy 2, 4–7, 9, 12, 20, 83, 86, 88, 109, 132, 103, 108n28, 116, 119, 127, 132, 201–2,
135, 140, 146–50, 350, 363n107 204, 206, 322
history of 29–31, 229–30, 258
institutionalization of 30 Rabinowicz, W., and Ronnow-Rasmussen,
and literature 4–7, 86 T. 235n38
paradigms of 6 Rackham, H. 349n41, 351n57, 353n69, 355n75,
positions in 3 359n96
pluralism 140n5, 142, 278n70 Ramelli, I., and Konstan, D. 209n4
400 GENERAL INDEX
Rawls, J. 231n13 Rudebusch, G. 124n69

reading Aristotle 340–1 Rutherford, R. 5n13
reading Plato 26, 228–30, 231, 232n22, 235n38, Ryle, G. 147n19, 159–60, 188n33
237n47, 242–3, 247, 249n93, 253, 257,
259n5, 275, 295, 296, 308–9, 327, 338 Sainsbury, M. 2n6, 216
Reale, G., and Scolnicov, S. 261n17 salience 13
reason 96, 97, 98, 100, 111n35, 113, 138, 163, Savile, A. 30n104
166, 167, 186, 1994n20, 236, 238, 254, 233, Scaltsas, T. 124n69
271, 273, 274, 301 scepticism 24, 30, 39, 61, 82, 94, 103n9, 141,
reasoning 10, 76, 79, 80, 81, 82, 88, 90, 92, 93, 182n22, 237, 241, 278, 364
94, 95, 96, 118, 167, 192n14, 194, 273, Scheffler, S. 235n37, 239n59, 259n4
274, 301 Schneewind, J. 231n14, 248n85
reasons 86, 89, 90, 119, 120, 135, 136, 252, 273 Schofield, M. 82n56, 144n15, 147n19, 163n32,
recollection 10n33, 18n63, 20n70, 28, 106n22, 169n34, 187n32, 189n38, 249n91,
190, 191, 192n13, 195, 196–9, 200, 201n37, 253n109, 257n124
202n41, 203, 205, 206, 207; see also science 23, 30, 77n25
memory Scott, D. 14nn51, 53, 104n15, 110n31, 119n58,
reductionism 89, 96, 97, 99, 143n13, 321, 325, 189n38, 190, 191, 196n23, 198n29, 202n41,
326, 329, 336, 337 207n45
Reeve, M. 72n3 secret dialogue of soul, see dialogue, internal
reflection, philosophical 1, 2, 4, 9, 10, 12, 23, 25, second-best journey 21
29, 36, 37, 55, 61, 103, 109, 110, 116, 117, second-order, see higher-order, second-order
118, 122, 168, 180, 182, 185, 186, 187–9, (knowledge, thought, reflection, stance,
194, 195, 196, 198–9, 200, 201, 202n41, levels of thought)
204, 205, 206, 211, 212, 225, 226, 227n43, Sedley, D. 6n17, 271n57, 274n61
229n6, 230, 243, 246, 257, 285, 286, 293, seeing 21, 23, 24, 25, 105, 106n21, 109, 110–15,
295, 296, 302–9, 341n5, 348n36, 362–4 121, 127, 130, 135, 173–89, 210, 211n12, 212,
higher-order 2, 116, 117, 121, 188 214, 215, 216, 217, 221, 223, 224, 225, 226,
reflective detachment 2, 4, 16, 116, 117, 301–5 283, 286, 287, 288, 290n44, 291, 294, 297,
reflective distance 2, 4, 9, 16, 139, 166 298, 299, 300, 308, 343–4, 347, 355, 362–4
reflexivity 174–5, 176nn9, 10, 177–9, 180, 188, better 21, 210, 218, 224, 226, 227, 308,
189, 202n38, 210, 216, 217, 223, 225, 226, 355, 363
227, 284n11, 287n29, 291, 292, 293, 294, joint 26
296, 297n69, 305, 346n31, 348n36, 351n57, mental 24, 115–21
360, 364 understanding 86 see also sight; vision
Reinhardt, K. 77n27 self 340, 342n12, 343–65
relations 98, 127, 128, 130, 135, 136, 139, 142, -knowledge, see knowledge
143, 144, 145, 150, 152, 155–60, 161, 162, -reference 2, 215–18
163, 164, 168, 174, 175–89, 201, 204, 205, -refutation 75, 76, 82n54, 201
213, 216, 221, 222, 244n75, 253, 319, 322, -sufficiency 340–3, 345, 348–53, 356–9,
325, 326, 333–6 363–5
The Relation Argument 174, 175–89 -verification 73, 75
relativism 22n80, 268 sense 25n92
responsibility 93, 261, 265 -organ 25n92, 284, 287n31, 292n49, 296,
Revenge of the Pink Panther 127 302n83, 303
Riordan, P. 279n71 see also perception
Robinson, R. 105n18, 108 Sharples, R. 65n1, 66n2, 339n71
Ross, W. D. 81n49, 283nn1, 4, 7, 315n10, Sheffield, F. 189n38, 309n93
316n15, 317nn16, 17, 21, 22, 330n51, Sherman, N. 363n106
332nn56, 61 Shoemaker, S. 216n29
and Urmson, J. O. 345n28 sight 283, 284, 286, 287, 288, 289n39, 290–4,
Rowe, C. 17n61, 27n101, 120n59, 123n64, 296–301, 302, 304, 305, 307, 343
124n69, 279n71 Silverman, A. 101n4
Rowett, C. (formerly Osborne) 84, 95n8, Sisko, J. 283n1, 292n50
283nn1, 5, 284nn9, 11, 286n19, 287nn25, skill 23, 108n28, 126, 184n25, 201, 234, 259,
27, 30, 31, 296n64, 342n10 263, 265, 268, 270, 276, 308, 364n108; see
Ruben, D. H. 244n77 also craft, craftsmen
GENERAL INDEX 401
Smith, N. 189n38 Third Man arguments 141n7, 144, 291n47, 311,

solipsism 75 320, 321n32, 322n36, 329
Solomon, J. 349n41, 351n57, 353n69, 359n96 Thomas, J. 279n71
sophists, sophistry 9n29, 14, 16n58, 19n69, 91, thought 101, 110–15, 116, 117, 118, 119, 121,
92, 125–37, 191, 200–7, 249, 262, 264, 144, 145, 148, 163, 178–80, 186, 188, 189,
269n52, 272, 273, 275, 276, 298 275, 290, 303, 306, 307
‘amazing sophist’ 208, 218–22, 225 thinking about thinking 11
‘noble sophist’ 21n76 Tor, S. 20n71, 31n109
sôphrosunê 173–89 touch 22, 24, 102, 104, 105, 106, 119; see
Sorabji, R. 178n14, 272n59, 286n22, 292n50, perception
299n73, 309n93 tragedy 1n2, 8
soul 84, 85, 90–93, 95, 102–24, 184, 185, 186, truth 11, 13, 18, 19, 22, 30, 36, 37, 38, 39, 41, 45,
187, 189, 199, 202, 231, 232, 234, 235, 236, 51, 52, 56, 58, 61, 81–2, 102, 103n11, 104,
237, 242, 260, 275, 301–4, 306, 308, 309, 105, 106, 119, 120, 121, 122, 125, 146, 154,
322, 323, 324, 347n33, 353n67, 354, 359; 200, 204, 212, 213, 215, 221, 222, 273, 274,
see also internal dialogue 278, 304, 306, 353
world-soul 98, 337 truth-function 77
Sprague, R. K. 228n3, 233n32, 238n56 truth-value 19
Stenzel, J. 101n4 Tye, M. 285n14, 286n20
stereoscopy 224–7
Stern-Gillet, S. 358n91 understanding 10, 19, 30, 48, 51, 52, 53, 64, 109,
Stokes, M. 59nn49, 52, 60n53 110, 117, 120, 121, 122–4, 146, 153, 185,
Strawson, P. 62 190, 196, 197, 202n39, 205, 209, 216, 219,
Striker, G. 231n15, 238n54, 249n90, 250n95 225, 226, 227, 293, 296, 308, 322, 324, 328;
structure 88, 96, 165, 277 see knowledge; wisdom
causal 17, 107, 120, 121, 218, 268, 293n54 (or knowledge?) 19n68, 107n24, 120, 121,
of conversation 8, 12, 116 122–4, 173n2
ethical 124, 270, 348 unhypothesized beginning 105
of explanation 14, 19, 88, 119, 121–3, 204, unity 138–69
268, 278 universals 4, 156, 157, 164, 203, 310–39
of knowledge 13 universalizability 143, 166, 168
of life 275 ‘unwritten doctrines’ 6n16
logical 19, 31, 90, 118, 127
of mind 219 value 234, 235–57, 258–79, 308, 346, 348,
ontology 143, 144, 165, 166, 315 364n108
of perception 114, 116, 212n19 evaluative dualism, 248, 249, 251n102, 256,
psychological 15, 118 257, 260, 277, 278
of reality 138, 142, 160, 296n65 source of, 235n40, 239–43, 248, 250, 252,
teleological 143 254–7, 260, 265–75, 276, 277
of value 257, 262, 271n56, 274 see also good
of world 93 Vasilakis, D. 21n75
sun 84, 85, 88, 93, 94, 95, 97, 110, 118, 120, Vazquez 20n71
220, 224 Vernant, J.-P. 39n10
synoptic view, synoptic dialectic 103, 104, 109, Vogt, K. 24n90
119, 120, 121, 122, 224, 226 virtue 12n43, 13n45, 20, 25, 26, 31, 109, 123–4,
systematicity 14, 92, 101n4, 121, 209 173–89, 195, 199, 202n39, 203, 206, 209,
of explanation 87, 90, 91, 138 211, 227, 229, 230, 232–7, 238nn53, 55,
of knowledge or understanding 87–9, 108, 242, 248, 249, 250, 251, 255, 258–79, 292,
121, 209 293, 294, 295, 303, 306, 308, 309, 349,
Szlezak, T. 6n16 351, 357
cognitive or epistemic 20n72, 24, 25, 123–4,
Taylor, C. 272n59 173n2
teleology 88, 89, 95–9, 143, 255, 303–4, 321n35, development 25, 274
322, 323, 348, 352, 359, 361 -friends, see friends, friendship
Textor, M. 24n88, 309n93 habit 25
Theuth 17–18 moral vs intellectual? 25, 123–4, 227
402 GENERAL INDEX
virtue (cont.) Whittle, A. 279n71

person 349–51, 357, 362 Wiggins, D. 39n12, 59n52, 62, 74n11,
practice 25 309n93
reflection 25 Wilberding, J. 221n38
vision 22, 24, 25n92, 119, 122, 297, 346, 355n81 Williams, B. 30n105, 73nn5, 6, 231n14, 232n22,
moral, 224n40, 239n57, 260n12, 261n21, 265n31, 267n45,
see also seeing 277n65
Vlastos, G. 27, 35nn2, 4, 51n34, 53n38, 59n52, Williamson, T. 22n81, 205n44
79n42, 229n5, 237n48, 243nn69 72, 259n5, wisdom 11, 16–18, 23, 123–4, 126, 134, 227,
260nn12, 15, 264n28, 268n49 258–79, 293, 294, 306, 308n91
Withey, M. 22n81
Walzer, R. R. and Mingay, J. M. 349nn38, 41, Wittgenstein, L. 35n1, 75
351nn57, 58, 353n69, 354n74, 355n77, Woodbury, L. 76n19
357n86 Woods, M. 316n11
Wasmuth, E. 22n81 Woolf, R. 31n109, 207n45
White, N., 111n27 writing, written word 6, 7, 17–20
Whiting, J. 26n98, 349n30, 350nn46, 49,
351nn51, 57, 352nn59, 63, 353n67, Yunis, H. 189n38
354n74, 355nn75, 76, 77, 357nn85, 86,
360n98, 365n110 Zagzebski, L. 20n72, 22n81, 24n90

Mary Margaret McCabe - Platonic Conversations-Oxford University Press, USA (2015)

Uploaded by

Copyright:

Available Formats

Mary Margaret McCabe - Platonic Conversations-Oxford University Press, USA (2015)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mary Margaret McCabe - Platonic Conversations-Oxford University Press, USA (2015)

Uploaded by

Copyright:

Available Formats

Platonic Conversations

I. On Dialectic and Method

II. On Knowledge and Virtue in Plato

III. On Aristotle’s Conversations with Plato

Bibliography of Works Cited 367

1. Paradox and Conversation

what is the provocation? What happens—to the hearer or to the speaker—when

2. Three Kinds of Conversation

3. Philosophy and Literature

4. Reading the Platonic Dialogues

The dialogues are constructed in various different ways to present discussions

5. Reﬂection, Thought, and Knowledge

show us these conversations set in a reﬂective context, where other interlocutors

6. Questions and Answers

intentional content: so when the question is answered, it is answered as an answer to

the observation that no particular provides us with an explanation, or gives us the

8. The Trouble with Writing

But he receives an unexpected reply:

9. Memory and Listening

10. Perception and Knowledge

11. Consciousness and Perception

Both philosophers regularly take reﬂection to be essential to improving our

allows the near-perfection of joint seeing—each understands the other’s subjective

12. Aristotle Reading Plato

14. Philosophy and its History

1. The River Argument

reasons, both may be thought to be true. Instead of common-sense, unarticulated

2. Other Paradoxical Arguments in Heraclitus

and qualiﬁed statements like 12:

directly—whether Heraclitus wishes to assert the contradictories simpliciter (‘sea

3. The Unity of Opposites and the Opposition of Unity

overlook—and perhaps we should not look beyond—the explicit association of logos

It is irresistible to conclude, from the conjunction of 50, 1, and 2 (invited by their

emphasizes not so much unity as opposition (‘everything through everything’). Or in:

4. Contradiction and Resolution

Aristotle defends this view in the chapters that follow.

5. One Under Many: Individuation

6. Up a Level: Trouble in General Theoretical Remarks

As it is generally construed, 57 (with 106) attacks Hesiod’s reputation for knowledge

In fragments 1 and 2, Heraclitus deploys the terms ‘common’ and ‘private’ to

So because the unity of opposites is essentially connected to its converse, the

7. The God’s-Eye View

At ﬁrst, these fragments ﬁt into a well-determined picture of Presocratic cosmology;

9. Dialectic and Metaphysics

Wiggins’ account of the problems of individuation is certainly inﬂuenced by the

sameness is an expression of the unity of opposites (88), or an aspect of the common

The famous ‘posset’ fragment of Heraclitus is cited ﬁrst by Theophrastus, de vertigine

It could, however, be a modiﬁcation by Theophrastus of some Heraclitean original.

however, on the phenomenon that generates such a contradiction allows us to see

1. The Rules of Dialectic

2. Alêtheia and Doxa

3. Dualism and Antithesis

Pythagorean side-effect, is fundamental to the structure of Parmenides’ entire

argument, it is also a consequence that he foresaw and wished to exploit.46 He subtly

5. Truth and the Alêtheia

1. Myth and Allegory

explanation is systematic. Aristotle, when he talks about demonstration and under-

Formally, a dialectical argument represents two lines of (deductive) argument, where

3. Myth and Allegory Again