Karen Margrethe Nielsen

Aristotle's Theory of Decision

The concept "prohairesis" enters philosophical discourse with Aristotle. Whereas Socrates thought the
agent's state of character depends exclusively on his epistemic state -- whether he has knowledge of good
and bad1 -- Aristotle identifies virtue as a state of character concerned with decision (hexis prohairetik,
1106b36; 1139a23). A virtuous person decides on virtuous action for its own sake2 (1105a32, 1144a19) -he chooses the right action for the right reason,3 and sticks with his decision even if doing the right thing
comes at the price of pain, as in courageous acts,4 or requires that he forfeit excessive or base pleasures, as
in temperate acts.5
The state of the agent's prohairesis -- whether he makes the right decision for the right reason, and whether
he stands by his decision when the time to act arrives -- is therefore a better indicator of the agent's
character than his act per se. What matters to our assessment of a person's character is not simply what he
does (though that matters, too), but why he does it, and whether his doing the right thing was a fluke, or the
expression of a firm and settled disposition. Just as it is possible to reach the right conclusion by valid
inference from false premises, it is possible to do the right thing for the wrong reasons. And just as it is
possible to hold true beliefs for no reason at all, it is possible to do the right thing without prior deliberation
and decision. That is why Aristotle insists that "[f]or actions in accord with the virtues (kata tas aretas), it
does not suffice that they themselves have the right qualities. Rather, the agent must be in the right state
when he does them. First, he must know [that he is doing virtuous actions]; second, he must decide on them
(prohairoumenos), and decide on them for themselves (kai prohairoumenos di'auta); and third he must also
do them from a firm and unchanging state (bebais kai ametakints echn)" (EN II, 4, 1105a28-33)6.
Aristotle defines decision as a special type of desire, a "deliberate desire to do an action that is up to us
(bouleutik orexis tn ephhmin)" (1113a11). Like his predecessors, Aristotle is convinced that the

Socrates' reservations about the identification of virtue with knowledge in the Meno concerns the lack of
teachers of virtue, for he ventures that if virtue were knowledge, there would be teachers of it. But he has
found none. This objection does not show that Socrates rejects the definition of virtue as knowledge -- the
evidence suggests that he rejects the inference. All that matters for determining whether an agent has virtue
is his epistemic state: Laches, 199c-e; Protagoras; Meno 88c-89d.
2
In choosing virtuous action for its own sake, he does not choose it for the sake of being virtuous -- as
if the perfection of his character were the final justification for any virtuous act. He will choose the act
because it is the right thing to do, and not for any further non-moral end. As Aristotle, like Plato and
Socrates, is convinced that virtue is necessary for happiness, there is no question of sacrificing one's
eudaimonia to do the right thing.
3
"Whoever stands firm against the right things (ha dei) and fears the right things, for the right end, in
the right way, at the right time, and is correspondingly confident, is the brave person; for the brave person's
actions and feelings accord with what something is worth, and follows what reason prescribes" (EN III, 10,
1115b17-20).
4
"The brave person will find death and wounds painful, and suffer them unwillingly, but he will endure
them because that is fine or because failure is shameful. Indeed, the truer it is that he has every virtue and
the happier he is, the more pain he will feel at the prospect of death. For this sort of person, more than
anyone, finds it worthwhile to be alive, and knows he is being deprived of the greatest goods, and this is
painful. (...) It is not true, then, in the case of every virtue that its active exercise is pleasant; it is pleasant
only insofar as we attain the end" (EN III, 10, 1117b 7-14; 15-16).
5
"The temperate person's appetitive part must agree with reason; for both [his appetitive part and his
reason] aim at the fine, and the temperate person's appetites are for the right things, in the right ways, at the
right times, which is just what reason also prescribes" (EN III, 12, 1119b15-18).
6
Throughout, I quote from T. H. Irwin (translation and notes), Aristotle: Nicomachean Ethics, 2nd
Edition (Indianapolis: Hackett, 1999). I rely on Bywater's OCT Aristotelis Ethica Nicomachea (Oxford:
Clarendon Press, 1894) for the Greek.

virtuous person will be motivated to act "in accordance with virtue", i.e. virtuously;7 in fact, this is what it
takes to be virtuous.8 Depending on how he defines virtue, each philosopher will think of different states as
basic in his moral psychology. In Socrates' case, knowledge of good and bad is thought to guarantee
virtuous action; in Aristotle's case, a settled disposition to make the right decisions, is equally thought to
guarantee virtuous action. Aristotle identifies this state as prudence, phronsis. Only a prudent man will
satisfy all three of Aristotle's criteria for virtue of character. He will do the right thing, knowing that what
he does is virtuous, and will decide on the virtuous actions for their own sake, and will do them from a firm
and unchangeable character. "It is by deciding (t prohaireisthai) on good and bad that we are men of a
certain character (poioi tines esmen)" (1112a2), Aristotle says. Making the right decisions is necessary and
sufficient for virtue, in the sense that nothing more is required no further act or character trait than the one
that produced the right decision.9 The virtuous man not only makes the right decision, he also lacks a
motive to act contrary to his decision, and hence acts in accordance with his decision whenever nothing
external prevents him. Insofar as the agents conception of what virtue requires is reflected in his decision,
and decision motivates him to act ceteris paribus, decision is the pivotal term in Aristotles ethics and
moral psychology.
Aristotle's choice of "prohairesis" as the concept that ties together his ethics and moral psychology is
utterly unprecedented in Greek philosophy. While Aristotle certainly did not coin the word (as Cicero
coined the noun "qualitas" from "qualis" in an attempt to translate the Greek "poion"), he is the first to put
the verbal noun to systematic philosophical use. The forensic speeches of Aristotle's contemporary
Demosthenes are ripe with examples of "prohairesis", and Aristotle may have picked it up from the legal
context, where the defendant's character and intentions will be relevant for the jury's assessment.10 There
7

Again, Aristotle's strict criteria for acting virtuously (adverbial construction) must be kept in mind.
Acting virtuously is not the same as performing a virtuous act.
8
I take the presence of a firm and unchangeable state to require the absence of desires that could
potentially topple the agents decision. Insofar as the merely continent agent has such desires, he will not
act virtuously, even if he satisfies the other criteria.
9
Because the absence of wayward desires doesnt amount to the presence of anything over and above
the right prohairesis, Aristotle may omit reference to them when he singles out the conditions for virtue: x
may be a sufficient condition of y in C even if the presence of z in C could have prevented y from obtaining
in C even though x obtained.
10
In Demosthenes' speeches, defendants, accusers and witnesses use prohairesis to refer to plans and
purposive actions as well as to their choice of profession or way of life, their "prohairesis tou biou".
"Prohairesis" can refer both to the act of choosing x over y, the thing chosen, as well as the purpose behind
an action, that for the sake of which something is chosen. This conforms to Aristotles use of prohairesis to
pick out both what the agent decides to do and that for the sake of which he does it, a double usage that
reflects his conviction that the virtuous agent will decide on virtuous activity for its own sake, i.e. as a
constituent part of the happy life. Any act chosen as a constituent part of happiness (as opposed to as an
instrumental means) will be decided on for its own sake. That an agent has made a certain "choice of life"
reveals that he considers this life best. Demosthenes' speech Against Aristocrates (translated with an
introduction by A. T. Murray (Cambridge: Harvard University Press, 1939)) illustrates this point.
Charidemus is said to be the same as Philiscus, with respect to his "choice of life": they are both tyrants (n
d'homoios ekeinos tout prohairesei tou biou (141)). Earlier in the same speech (127), the Athenians are
warned against awarding citizenship to the volatile Charidemus, because "There is neither any stability nor
devotion on the side of those who live for the purpose of getting more than their fair share (para toutois tois
epi t tou pleonektein prohairesei zsin)". The conception of the good life that motivates Charidemus is
mercenary. The agent's "choice of life", then, refers to the end the agent values above all, and that he has
organized his life around. The allegiance need not be political; it can also be to pleasure and debauchery.
The speech Against Olympiodorus, now considered spurious, provides a vivid example of a pleasureseeker. (Pseudo-Demosthenes, Against Olympiodorus, translation and introduction by A. T. Murray
(Cambridge: Harvard University Press, 1939)). The craven Callistratus is trying to persuade the court to
disinherit his brother-in-law Olympiodorus on the grounds that Olympiodorus lifestyle is depraved. In a
letter to the court, the family describes the defendant in unflattering terms: The defendant Olympiodorus,
then, is a person of this sort. He is not only dishonest, but in the opinion of all his relatives and friends is
proven by the manner of life he has adopted (t prohairesis tou biou) to be mentally deranged

appears to be a linguistic watershed some time in the mid-fourth century, when the composite verb appears
with increasing frequency both inside and outside philosophical discourse.11 Plato only uses the simple verb
hairesis, though not systematically. The noun prohairesis appears only once in the entire Platonic corpus,
at Parmenides 143c.12
Voluntary Action and Decision in Nicomachean Ethics, III
Aristotle presents his "official" analysis of prohairesis in book III, where it appears in the context of an
analysis of the voluntary (to hekousion) and the involuntary (to akousion). In the Nicomachean Ethics,
Aristotle identifies voluntary action with action that is caused by the agent's desires -- whether they are of
the appetitive, spirited, or rational kind that Plato identified in Republic IV. Furthermore, the agent must
know the particulars of his action.

(melancholan dokn): to use the language of the lawgiver Solon, he is beside himself (paraphronn) as no
man ever was, for he is under the influence of a woman who is a prostitute (porn). And Solon established
a law that all acts shall be null and void which are done by anyone under the influence of a woman,
especially of her stamp. (The disgruntled in-laws follow up this impeccable syllogism -- based on Solon's
general principles about the volatile influence of women -- by pointing out that it would actually be in the
advantage of Olympiodorus himself if the court ruled in favor of the plaintiff: no money, no debauchery).
The date of the speech, the year 343 or 342 B. C., supports the idea that that Against Olympiodorus is not
by Demosthenes. As Norman DeWitt notes in his introduction, "Scholars are virtually unanimous in
refusing to believe that Demosthenes, at the height of his career, would have stooped to handle so unsavory
a matter for an unimportant personage". But if the speech is not by Demosthenes, we have additional
evidence that the phrase "prohairesis tou biou" was in general use in mid fourth-century Athens. The
author, whoever he was, puts the phrase into the mouths of Callistratus' unimportant and petty-minded
relatives. The phrase prohairesis tou biou also occurs in the so-called Erotic Letter, another spurious
work that both modern and ancient critics have been reluctant to attribute to Demosthenes. (PseudoDemosthenes, The Erotic Letter, translated with an introduction by Norman W. DeWitt and Norman J.
DeWitt (Cambridge: Harvard University Press, 1949)). In the Erotic Letter, whose author may have been
familiar with the Phaedrus, the lover (erasts) promises to advise his love-interest (eroumenos) on
questions of education and his "choice of life" -- prohairesis tou biou (2). This, the letter-writer confesses,
is partially the purpose (prohairesis) of the letter. But he also wishes to praise the boy, though not to the
point of shaming him as lovers often do. The letter-writer goes on to provide pedantic tips to the boy about
his choice (prohairesis) of sports. It seems that the verb prohaireisthai could be used in a multitude of ways
even within the same page without drawing attention to itself -- which seems to prove that prohaireisthai
was a staple verb in Greek in the mid-fourth century.
11
For a survey of instances, consult R.-A. Gauthier and J.Y. Jolif, L'Ethique Nicomaque:
Introduction, Traduction et Commentaire, vol. II, 1, 2 ed. (Louvain: Publications Universitaires, 1970), p.
189. Eugen Kullmann traces the history of the term and its precedents in philosophical, literary and
rhetorical discourse up to the mid-fourth century in his dissertation Beitrge zum Aristotelischen Begriff der
'Prohairesis' (Basel: Im Selbstverlag des Verfassers, 1943).
12
R.-A. Gauthier and J.Y. Jolif, L'Ethique Nicomaque: Introduction, Traduction et Commentaire,
vol. II, 1, 2 ed. (Louvain: Publications Universitaires, 1970), p. 189. Gauthier and Jolif fail to note that
Plato employs the verb prohaireisthai elsewhere, though in an unsystematic and unselfconscious way (e.g.
Theaetetus 147d5, Sophist 251e1, Phaedrus 245b4, Theages 128a8). There is one notable exception. In the
spurious Definitiones, attributed to Plato, we find a definition of virtue that is uncannily Aristotelian:
"kalokagathia hexis prohairetik tn beltistn" (412e8). In his introduction to the Definitiones (in J. M.
Cooper (ed), Plato: Complete Works (Indianapolis: Hackett 1997)) D. S. Hutchinson suggests that the
Definitiones, a dictionary of philosophical terms, was compiled in the fourth century Academy. In
Democritus, Fragment 96, we find the first and only instance of the composite verb prohaireisthai in the
testimony of the philosophers somewhat misleadingly dubbed the "Presocratics": "A generous man is not
one who looks for a return but one who has chosen to confer a benefit (ho eu dran prohremenos)". But I
dont count this as compelling evidence for a systematic use of the word prior to Aristotle.

The discussion of decision follows hard on the heels of a pithy, but for our purposes important, analysis of
"mixed" actions. An action is mixed if no one would choose it for its own sake, since it is bad in and of
itself, but a particular agent still chooses to perform it because he finds that it is the best alternative
available to him under the circumstances. No person in his right mind would voluntarily jettison his cargo
for its own sake, Aristotle submits, but in order to save himself and his shipmates in a storm, any sane man
would throw his cargo overboard. Similarly, no upright man would choose to debase himself just for the
sake of it, thinking that it is better, but when a tyrant threatens to kill his family unless he submits, any sane
man would comply. In both of these cases, the agent calculates the costs and the benefits of the per se bad
action, and finds that it is better, all things considered, to perform it.13 He voluntarily chooses x for the sake
of some end, y, an end that he takes to justify the high price exacted by the action. On the basis of his
analysis of mixed actions, Aristotle concludes that "decisions distinguish characters from one another better
than actions do" (EN 1111b5-6). After all, good men may face tough choices. For, he writes, "it can be hard
to judge (diakrinai) what goods should be chosen at the price of what evils, and what evils should be
endured as the price of what goods. And it is even harder to abide by one's judgments, since the results we
expect are usually painful, and the actions we are compelled to perform are usually shameful" (EN
1110a29-33).14 Thus, until we know what an agent decided to do for the sake of what end, and what
alternatives seemed available to him at the time he made his decision, we cannot determine whether the
agent is vicious, crazy, or simply incapable of good deliberation, or an honest and upright person in
unfortunate circumstances. We cannot know, moreover, whether he did what he did because he judged that
it was best, all things considered, or whether he broke down under stress, and acted contrary to his
judgment. Actions by themselves are poor guides to character, since one and the same action may be bad in
the abstract, but best under the circumstances. This is presumably one of the reasons why Aristotle defines
virtue as a "state that decides" (hexis prohairetik, EN 1106b36; 1139a22). The agent's character is
revealed by what he decides to do for what ends, and whether he abides by the decision.15
Aristotle defines decision as a "deliberate desire to do an act that is up to us" (EN 1113a11). This definition
is obviously in need of elaboration. What, for starters, is a "deliberate desire" (orexis bouleutik)? And
when is an action "up to us" (eph'hmin)? Decision, Aristotle says, is a species of the voluntary (to
hekousion). If an action is performed as the result of decision, it is voluntary, but the reverse does not hold
(EN 1111b6-7; Rhetoric 1368b10-12). While Aristotle famously holds that animals and children act
voluntarily, they do not thereby make decisions, since they lack deliberative imagination.16 Nor do women
13

I here depart from M. Stocker's analysis of "mixed acts" in "Dirty Hands and Conflicts of Values and
of Desires in Aristotle", reprinted in his Plural and Conflicting Values (Oxford: Clarendon Press, 1990). I
defend my view in "Dirtying Aristotle's Hands? Aristotles Analysis of 'Mixed Acts' in the Nicomachean
Ethics III, 1".
14
Note that an action can be performed under compulsion (ex ananks) without being involuntary
according to Aristotle: when the tyrant threatens a man into debasing himself in order to save his family
from death, the man deliberates, and chooses whatever action seems best to him under the circumstances
(1110a19-26; 29-31). And if the action is the result of a decision, it is eo ipso voluntary. Aristotle
repeatedly refer to the things compelled men do as "praxeis" (1110a 4;5;6;7;1110a12; 13;14;15;19), and I
take this to imply that although he uses the simple verb "haireisthai" (1110a 12; 19; 30; 1110b4; 7) rather
than the compound verb "prohaireisthai" throughout the mixed acts passage (III, 1), this does not entail that
the actions of the compelled man are merely instrumental (poisis). For the praiseworthy compelled man
makes a correct judgment (1110a29) about what he ought to do (1110a25). And as I argue below, this
presupposes that his choice is ultimately grounded in a conception of happiness: the correct one, as it
happens.
15
Insofar as the weak agent reaches the right decision, but does not abide by it, actions are not
irrelevant for distinguishing between characters. But the nature of the agent's decision is a better guide to
his character than his actions. It is furthermore the only positive factor that need be present for virtue the
only other factors are factors that must be absent.
16
Terence Irwin has argued that on Aristotles considered view, it is not the case that an agent is
responsible for doing x if and only if he did x voluntarily. This simple theory conflicts with Aristotles
unwillingness to challenge the common-sense view that animals and young children are not responsible for
their actions even thought they act voluntarily. Rather, Aristotle should be thought to endorse the
complex theory that an agent is responsible for doing x if and only if (a) there is some deliberative

or natural slaves have the ability to make authoritative decisions. Slaves lack the mental resources to live a
life in accordance with decision (zn kata prohairesin, Politics 1280a31-34). Women tend to abandon their
decisions once made. This means that women and natural slaves can never lead a happy life, by Aristotle's
standards, for the happy life is a life well lived, and a life well lived is a life that expresses virtue, and
virtue, in turn, is a "state that decides" (hexis prohairetik). Like children and animals, however, women do
act voluntarily.17
In the EN, Aristotle defines the voluntary as "what has its origin (arch) in the agent himself when he
knows the particulars that the action consists in" (EN 1111a20-21). The origin of an action is in the agent if
he desires the action, and this desire is the (direct) cause of what he did. The involuntary (to akousion), by
contrast, is what is caused by force (bia) or by ignorance (agnoia).18 An agent does something by force if
the origin of what he is doing is external, and he contributes nothing to the action. If someone grabs hold of
your hand and slaps someone else with it, the slapping is involuntary by Aristotle's lights (EN V, 1135a27;
EE II, 1224b13-14)19. There is one important proviso to this rule: you should not be pleased by what
happened. If you are pleased by the slapping, the action falls into Aristotle's curious third category -- it is
"non-voluntary" (ouk hekousion), neither voluntary nor involuntary. An action is involuntary if it is against
your desire; it is non-voluntary if the source of the motion is external, but you are not acting contrary to
your desire. That is why the agent's reactive attitude reveals his or her character. Ignorance is a cause of
involuntary action when the agent lacks true beliefs about (1) who he is, (2) what he is doing, (3) to whom,
(4) with what, (5) for what result, or (6) how he is doing it (gently or hard). When Hamlet strikes his sword
through the tapestry, he kills a man voluntarily, but he does not kill Polonius voluntarily, since he performs
the action believing that he is killing an intruder, but not believing that he is killing Polonius. However,
Hamlet's lack of remorse shows that the act of killing Polonius was not involuntary, but rather non-

argument which, if it were presented to him, would be effective about his doing x, and (b) he does x
voluntarily. This makes the capacity for deliberation and decision concerning the desire that is responsible
for ones doing x a necessary condition for responsibility, although the actual exercise of that capacity is
not. T. H. Irwin, Reason and Responsibility in Aristotle, in A. O. Rorty (ed), Essays on Aristotles Ethics
(Berkeley: University of California Press, 1980), pp 117-156.
17
In this paper, I self-consciously bracket Aristotle's gender-parochialism, and refer to the agent
interchangeably as "he" or "she", in the absence of an all-purpose pronoun to pick out a person as a person.
Aristotle's conception of decision is not inherently masculinist, although his application of the theory can
be criticized on that count. In Politics III, 9, Aristotle argues that states exist for the sake of the good life,
and not for the sake of life only, by pointing out that if life in itself were the end of the state, even slaves
and brute animals could form a state. "But", he argues, "they cannot, for they have no share in happiness or
in a life based on decision (zn kata prohairesin)" (1280a31-34). In order to live in accordance with
decision, we need to direct our actions towards a conception of happiness. Natural slaves lack the ability to
deliberate about what kind of life is best, and hence they cannot make decisions about praxis (although they
can, presumably, make decisions about instrumental action, poisis). This deficiency is the generic
distinguishing mark of natural slaves. It explains why some "from the hour of their birth" are marked out
for subjection, not for rule (Politics, 1254a23). Women occupy a borderline position with regard to virtue
and decision. At Politics I, 1252a33-b4, Aristotle denies that nature has put women on a par with (natural)
slaves. As he insists, "there are many kinds of subjects" (1254a24). The failure to develop a properly
functioning ruling part of soul means that women lack the ability to make authoritative decisions about
praxis. Masters are masters because they are better psychologically equipped; they have a fully capable
ruling part of soul. This is the part that decides (1113a6-7). The ruling element in the female soul, by
contrast, is easily dethroned: a certain softness (malakia) is endemic in women. In fact, this softness
"distinguishes the female sex from the male" (EN 1150b15-16). Aristotle defines malakia as the tendency
to be overcome by pain.
18
Compare Aristotle's corresponding definition of the voluntary in Rhetoric 1368b19.
19
In the EE, book II, section 8, Aristotle specifies that an agent does something by force only if the
thing he does is contrary to his impulse (horm) -- whether that impulse is a rational or a non-rational
desire. I take it that Aristotle is here using horm as a catch all for desire, as he uses orexis as a catch-all
elsewhere.

voluntary, and hence he is subject to blame for killing Polonius, not just for killing any old intruder.20 If I
strike out against you, thinking that I am wielding a plastic Halloween sword, but it is the real thing, I do
not impale you voluntarily, though I did strike out against you voluntarily. Again, if I am pleased that I
impaled you, though I didn't intend to, my impaling you is non-voluntary, but not in-voluntary.
Whether an action is voluntary, involuntary, or non-voluntary, then, depends both on the agent's cognitive
condition as well as his desires. If the agent knows who he is, what he is doing, to whom, with what, for
what result, and how he is doing it, and he is not forced to do what he is doing, the origin (arch) of the
action is in him, and he acts voluntarily. In the Nicomachean Ethics, Aristotle seems to maintain that if an
action was performed voluntarily (hekn), then it was also "up to" the agent to perform it. But an action can
be "up to" a specific agent even if he chooses not to perform it: anything that it is in our power to do as
specific agents is "up to us", including unrealized possibilities. By defining decision as a "deliberate desire
to do an act that is up to us" (EN 1113a11), Aristotle is underlining that we don't decide to do what we
think is impossible or necessary.21
This brings us back to the first part of Aristotle's definition of decision as a deliberate desire. We
deliberate in order to find actions that are up to us to perform which promote our ends -- ends that we desire
because we judge them to be such that we ought to pursue them for their own sake. Decisions, then, are the
result of a wish for an end (boulsis) and deliberation about how to attain that end. If x is a means to y, we
don't decide to do x unless we think y is worthy of pursuit in its own right. That's why deliberation that
takes an appetitive desire (epithumia) or a spirited desire (thumos) as its starting point does not result in
decision (prohairesis) in Aristotle's technical use of that term (1111b10-19). I can't decide to take exercise
as a means to health unless I desire health for the sake of something that I judge to be unqualifiedly good
(EN 1139a31; 1140b4-7).22 Aristotle preserves the connection between acting on a decision and acting
from forethought that we find in the 4th-century BC orators, most notably Demosthenes, in whose speeches
acting from decision, "ek prohaireses", is used interchangeably with acting from forethought, "ek
pronoias". Premeditated murder, then, is the same as murder "ek prohaireses" for Demosthenes23 But
20

For a full discussion of the Hamlet-case, see A. Kenny, Aristotle's Theory of the Will (New Haven,
CT: Yale University Press, 1979) p. 53.
21
In De Fato XIV, Alexander of Aphrodisias appears to depart from Aristotle's position on the
voluntary and what is up to us (eph'hmin). For Alexander, anything that comes about from an assent that is
not forced is voluntary, while what comes about from an assent that is "in accordance with reason and
judgment" is "up to us" (183.27-30). An action is "up to us" only if it depends causally on reason and
judgment (i.e., on deliberation and assent). Alexander thus identifies the class of actions that are up to us
with the class of actions that we have decided to perform, rather than with the class of actions that we could
decide to perform. Consequently, he denies that unrealized possibilities are "up to us". This view does not
sit well with Aristotle's contention that we deliberate about what is up to us (1112a32) -- for deliberation is
certainly about unrealized possibilities. Alexander is led to conclude that while everything that is up to us is
voluntary, not everything that is voluntary is up to us. He commits himself to the same unorthodox view in
Quaestio 3.13, "Some points concerning what is up to us". R.W. Sharples, Alexander of Aphrodisias: On
Fate (London: Duckworth, 1983); R. W. Sharples, Alexander of Aphrodisias: Quaestiones 2.16-3.15
(Ithaca, NY: Cornell University Press, 1994). Im not here taking a stand on the vexed question about
whether Aristotles account of voluntary action is libertarian. I dont think anything I say above, including
the observation that Aristotle refuses to permit that we can decide to do something that is necessary or
impossible, forces us to accept that he is a libertarian.
22
Decision is the proximate cause of praxis (1139a31), and the aim (telos) of praxis, as opposed to
poisis, is acting well in itself (eupraxia) (1140b7). The prudent man knows what eupraxia consists in,
since he knows "what is good and bad for a human being" (1140b4-6). That is, if I choose x for the sake of
y, and think that y is desirable for its own sake, then I technically speaking decide on x. Note that the
transitivity is not limited to actions that immediately lead to an end desired for its own sake because it is
good in itself. If I desire x for the sake of y, and y for the sake of z, judging that z is an end that ought to be
pursued in its own right, I can still decide to do x (as well as y) as long as the justification is of the required
kind.
23
As Gauthier and Jolif note (p. 190), Demosthenes employs the term in a sense which approximates
Aristotle's in the Second Philippic, 16; Against Meidias, 44, 66; and Against Leochares, 57. An action is the

while Aristotle takes premeditation to be a necessary condition for decision -- the agent must have
deliberated about her action -- he also requires that the agent have judged that it is best, all things
considered, that she act in the way that she has decided (EN 1140b4-7; 1140b16-20; 1112b16-20). If the
agent does not think she ought to commit murder, then her murdering her enemy is not done "ek
prohaireses", even if it is done in cold blood. No one decides on the spur of the moment, but deciding
requires more than technical deliberation: it requires weighing the pros and cons in order to determine both
how one ought to act relative to ones ends, and what ends are worth pursuing.
In a fit of etymologizing, Aristotle explains that
apparently, a decision is voluntary, but not everything voluntary is decided. Then perhaps
what is decided is what has been previously deliberated. For decision involves reason and
thought, and even the name itself would seem to indicate that [what is decided
prohaireton] is chosen [haireton] before [pro] other things (EN 1112a14-17).
The prefix "pro-" has been understood in three senses: temporal, preferential, and teleological (denoting
what is decided for the sake of some end). Aristotle may have all three in mind in his pithy etymological
explanation: What we decide to do is what we have deliberated about in advance, in the light of some goal
for the sake of which we decide to undertake the action, but concluding the process of deliberation also
requires that we decide between the available alternatives -- minimally, whether we should act or not -- and
decide whether the end is worth the cost of the means (1112b16-20). Deliberation is only meaningful when
the agent suspects that there may be more than one alternative open to her: at the very least, she can either
perform the action or abstain (1113b8-11). In deciding to perform an action, we decide to do one thing
rather than another, and this has led some commentators to translate decision as "preferential choice". 24
However, the preposition "pro" may equally denote the things we do for the sake of certain ends, to be
specific, for the sake of the most complete end, happiness (teleological). Not just any choice counts as a
decision -- only actions that are chosen for the sake of happiness are decided upon. Other choices, however
deliberate, fall short of being decisions.
This brings us to what has been one of the most contentious issues is the discussions of Aristotelian
practical reason. Interpreters have differed sharply over the scope they ascribe to Aristotelian deliberation.
Do we deliberate about means only, or does Aristotle's theory leave room for deliberation about ends?
Aristotle's discussion of mixed actions seems to leave room for both kinds of deliberation: for he observes
result of a decision if it is premeditated. Aristotle explicitly uses the expressions "ek pronoias" and "ek
prohaireses" as synonyms in EN V, 1135b25-26. Alexander of Aphrodisias follows suit in Quaestio 3.13,
108.14-18 and equates the sense "from deliberation", "from decision", and "from forethought". See also W.
T. Loomis, "The Nature of Premeditation in Athenian Homicide Law", The Journal of Hellenic Studies,
vol. 92 (1972), pp. 86-95.
24
See e.g. W. F. R. Hardie, Aristotle's Ethical Theory, 2nd ed., chapter IX, "Choice and the Origination
of Action" (Oxford Clarendon Press, 1980), p. 162. D. Charles adopts Hardie's terminology in Aristotle's
Philosophy of Action (London: Duckworth, 1984). A. Kenny, Aristotles Theory of the Will (New Haven,
CT: Yale University Press, 1979) renders "prohairesis" as "purposive choice" or simply "choice". Kenny
apologetically defends his translation in a note: "'Purposive choice' seems to me the least misleading
translation of prohairesis. Its clumsiness reflects the fact that no natural English concept corresponds to
Aristotle's. I shall sometimes abbreviate to 'purpose' or 'choice'" (Kenny, p. 69). S. Broadie, Ethics with
Aristotle (Oxford: Oxford University Press, 1991), p. 78, settles for "rational choice", while R. Sorabji
suggests "deliberate choice" in Necessity, Cause and Blame. Perspectives on Aristotles Theory (Ithaca,
NY: Cornell University Press, 1980). The translations "preferential choice" or "purposive choice" strike me
as unfortunate, insofar as both foreclose the question of what the force of "pro" really is by narrowing the
range of available interpretations down to one. "Rational choice" is better, but resonates too much with the
contemporary "rational choice theory", which Aristotle's theory certainly is not. "Deliberate choice" is
better, especially since Aristotle defines prohairesis as "orexis bouleutik" (1113a11), but the translation
moots many of Aristotle's observations in the sections on decision and deliberation (EN III, 2 and 3).
Irwin's restrictive use of "decision" to render prohairesis avoids these difficulties.

that some actions are so base that no amount of threats should ever lead us to prefer them, no matter how
worthy our end is per se. He adduces Alcmaeon's matricide in Euripides' play as an example of such a
categorically unjustified action (EN 1110a29). Aristotle's censure of Alcmaeon implies that an agent must
always inquire into what ends are worth pursuing at what cost, and this is only possible if he is able to
consider the worth of the end relative to the cost of the means. It seems, then, that Aristotle permits
deliberation about what ends we should pursue.
This result appears to be sharply at odds with the impression left by Aristotle's discussion of the objects of
deliberation and decision. In his summary of the "official" discussion of deliberation in the Nicomachean
Ethics, book III, Aristotle states that
we have found, then, that we wish for the end (boultou men tou telos), and deliberate
and decide about things that promote it (bouleutn de kai prohairetn tn pros to telos);
hence the actions concerned with things that promote the end are in accord with decision
and are voluntary (EN 1113b3-5).
The proliferation of such claims in EN III has convinced some interpreters -- for instance Aurel Kolnai -that as far as Aristotle is concerned, we never deliberate about what ends to pursue or which desires to
satisfy. Deliberation merely seeks means to the satisfaction of pre-existing desires. Aristotle, in other
words, is an instrumentalist about practical reason.25 Aristotle's distinction between wish (boulsis) and
decision (prohairesis) appears to confirm this impression: "we wish rather for the end (to telos), but decide
on what promotes the end (ta pros to telos)" (1111b26). For instance, we wish to be healthy, but decide on
things that produce health, and we wish to be happy, but we do not decide to be happy. Isnt this a clear
endorsement of instrumentalism? The instrumentalist holds that we always deliberate on the basis of some
pre-existing desire for an end, and that this desire is essentially non-rational: it is not the result of previous
reasoning about what ends one ought to pursue26. Most scholars now agree that the instrumentalist
interpretation of Aristotle is mistaken. As David Wiggins has pointed out, the "ends" Aristotle have in mind
are not ends in general, but ends that are constitutive of specific crafts, such as health is for medicine, or
persuasion is for oratory, or an end which all human beings, qua human beings, share, namely happiness.
Just as no physician qua physician deliberates about whether to heal, no human being deliberates about
whether he should promote his own happiness, Aristotle says. This does not, however, rule out deliberation
that aims to specify the content of happiness. Therefore, although the highest end is fixed for all human
agents, and not subject to deliberation, this end is entirely formal. In order to serve as a guide to living, the
content of the highest end must be specified through a process of deliberation. That is, we must determine
what happiness consists in.27
Wiggins' ingenious interpretation supports the view that deliberation can be about ends. This view has
come to dominate the literature on Aristotle's theory of practical reason. Beginning with L. H. G.
Greenwood's introductory essay to his 1909 study Aristotle: Nicomachean Ethics vi, a host of philosophers
have argued that appearances to the contrary, Aristotle is not an instrumentalist about practical reason. As
is well known, Greenwood provides a non-instrumental interpretation of what Aristotle means when he
claims that we "wish for the end", but "deliberate and decide about the things that promote the end" (ta pros
to telos) (1113b3-4).28 The expression "ta pros to telos" may pick out both instrumental means to an

25

A. Kolnai, "Deliberation is of Ends", reprinted in E. Millgram (ed.), Varieties of Practical Reasoning

(Cambridge, Mass: MIT Press, 2001), pp 259-278. The paper was originally published in Proceedings of
the Aristotelian Society, suppl. vol. 36 (1962).
26
This way of bringing out the idea behind the instrumentalist interpretation of Aristotle is T. H. Irwin's:
Aristotle's First Principles, chapter 15, 178, "The Scope of Deliberation", pp 335-6.
27
D. Wiggins, 'Deliberation and Practical Reason', reprinted in E. Millgram (ed.), Varieties of Practical
Reasoning (Cambridge, Mass: MIT Press, 2001), pp 279-299. The first version of the paper was printed in
Proceedings of the Aristotelian Society 76 (1975-1976).
28
L. H. G. Greenwood (trans. and ed.), Aristotle: Nicomachean Ethics vi (Cambridge: Cambridge
University Press, 1909), introductory essay. His argument is developed by, among others, J. M. Cooper,
Reason and Human Good in Aristotle (Cambridge, Mass: Harvard University Press, 1975), pp 22-23; and

independent end and "constituent means", as Greenwood calls them; what the ends of "health" or "living
well" consist in, as opposed to what causally brings them about. Thus, having a well-functioning heart is a
constituent of health, not an independent means to health, just as sugar is a constituent means of the
pudding, rather than an instrument in the production of pudding. Deliberation can thus be a way to discover
the instruments to bring a given end about, but, more importantly, it can be a way to discover which
elements constitute the end. The only end that is fixed for all human agents is eudaimonia.29 The most
important task of practical rationality is thus to specify what happiness consists in, and to determine which
ends we should pursue given our specification of happiness. It gives content to an end that would otherwise
be entirely formal. In deciding to perform an action, we judge, explicitly or implicitly, that it is the best way
to promote the kind of life that we judge to be best.
Aristotle's definition of prudence, phronsis, confirms this interpretation. For phronsis is deliberative
excellence, and deliberative excellence consists in the ability to determine what kinds of activities are good
or bad for a human being:
Prudence (phronsis) is a state grasping the truth (hexin aleth),30 involving reason,
concerned with action (meta logou praktik) about things that are good or bad for a
human being (1140b4-6).
The rejection of instrumentalism has one very important implication for Aristotle's distinction between
moral and intellectual virtue. A first-time reader of book VI of the Ethics could be excused for going away
with the impression that Aristotelian virtue (aret) is excellence in desiring the right ends, while prudence
(phronsis) is excellence in finding the right means to those ends -- i.e. mean-ends deliberative excellence.
This neat, but ultimately mistaken, division of labor ignores Aristotle's claim that full virtue requires
deliberative excellence. Before we can know what ends to pursue, and before we can decide which desires
to satisfy, we must specify which ends are worth pursuing. The fact that a desire is mine is not sufficient
grounds for seeking to satisfy it. If I don't judge that the desire ought to be satisfied, I don't have a reason to
satisfy it. This means that with the exception of the desire for happiness as a formal designation of the
highest end, all rational desires (boulseis), and not just decisions (prohaireseis), arise through deliberation.
Thus, all rational desires (with the exception of the desire for happiness) are the products of a judgment
(krisis) about what ends we should pursue, and not just desires we happen to have:
What we deliberate about is the same as what we decide to do, except that by the time we
decide to do it, it is definite; for what we decide to do is what we have judged [to be
right] as a result of deliberation (ek ts bouls krithen). For each of us stops inquiring
how to act as soon as he traces the principle to himself, and within himself to the guiding
part (to hgoumenon), for this is the part that decides (EN 1113a2-6).

T. H. Irwin, "Aristotle on Reason, Desire, and Virtue", Journal of Philosophy (1975), pp 567-578, and
Aristotles First Principles (Oxford: Clarendon Press, 1988), esp. chapter 15.
29
The reason is that everyone agrees that it is a complete and self-sufficient end; it is that for the sake
of which we choose everything we choose, and someone who is happy lacks nothing. Aristotle seemingly
wavers between endorsing psychological and ethical eudaimonism in the texts transmitted to us. In EE
1214b6-8 he claims that anyone who is capable of living in accordance with his own decision (kata tn
hautou prohairesin) does set up (thesthai) a target for living finely, a target which he "looks towards" when
he performs all his actions (pasas tas praxeis). This is a statement of psychological fact. But the
manuscripts are not unanimous: a variant of the same passage reads "dei thesthai" -- which implies ethical
eudaimonism. I have not found any conclusive evidence to suggest that Aristotle is a psychological
eudaimonist in the Nicomachean Ethics.
30
"Hexin aleth" literally: "a true state". Aristotle's emphasis on prudence as a state grasping
"truth" should alert us to a difficulty in attempts to portray the conclusions of the prudent man's
deliberation as imperatives rather than judgments. Aristotle is apparently not afraid to think that
making a true judgment suffices to motivate the agent.

In making a decision, I fasten on the action that I judge to be the best, all things considered, for promoting
my end. But unless this end is entirely formal (i.e., unless it is happiness in the abstract) it is not itself
beyond deliberation. When the price is too high, I may decide, on reflection, to abandon it, as Alcmaeon
should have abandoned his end when he realized that he could only attain it by matricide.
By classifying decision as a desire, Aristotle holds that when I judge that I ought to pursue a certain course
of action, I will be motivated to pursue that course of action. Perhaps one might object that deciding to
pursue a certain course of action does not by itself amount to desiring that course of action. I can decide to
go to the dentist without desiring to go to the dentist. Faced with such objections, Aristotle would counter
that any judgment about what is best invariably produces a desire for the course of action we judge to be
best: "for when we have judged [that it is right] as a result of deliberation, our desire to do it expresses our
wish" (EN 1113a11-13)31. Granted, I may not be jumping with joy at the prospect of having my teeth
pulled out, but if I judge that it is for the better, I will have an impulse towards that action. In what follows,
I will argue that Aristotle holds an internalist conception of decision. Judging that "I should ! in
circumstances C" is sufficient to motivate me to ! in circumstances C, provided that I don't suffer from
practical irrationality. This inner relationship between judging that one ought to !, having a desire to ! (a
decision) and !ing, unless some internal or external obstacle hinders the desire from being effective,
explains why Aristotle only cites the agents prohairetic state, and no other positive factor, as a measure of
his virtue.
Aristotle's Explanation of Akrasia
The upshot of my thumbnail analysis of Aristotle's theory of decision is that Aristotle thinks that practical
judgments have the power to motivate the agent. I decide to pursue a course of action by judging that it
would be best for me to pursue it (EN 1110a29-33; 1113a2-6) -- I do not need an independent desire to ! in
C in order to be motivated to ! in C if this is what I think I should do. Judging that this is what I should do
suffices to motivate me to pursue that course of action, as the judgment necessarily causes the prohairetic
desire for the action.32
31

This rendition of 1113a13 is somewhat controversial. There is a dispute about what the correct text at
1113a13 really is. Oxford Classical Texts reads "kata tn bouleusin" (in accordance with deliberation) in
lieu of "kata tn boulsin" (in accordance with wish). There is evidence both internal and external to
Aristotles corpus that "kata tn boulsin" expresses Aristotle's view. First, the claim Aristotle advances if
we read "kata tn boulsin" rather than "kata tn bouleusin" fits hand in glove with observations Aristotle
makes both in the EE (1226b17-19), and in in De Anima (433a24-25). In EN VI, 9, Aristotle denies that a
person who is good at finding means to satisfy his non-rational desires possesses euboulia, good
deliberation. The incontinent agent, who acts on appetite contrary to right reason and decision, can use
rational calculation (logismos) to reach his end (1142b17-22). But this use of rational calculation does not
issue in a decision. His calculation about how to attain the object of appetite must therefore differ from the
deliberation that issues in a decision. And the difference must consist in the kind of desire that motivates it.
Second, the reading "kata tn boulsin" is corroborated by the earliest extant commentary on the Ethics, by
the 2nd century AD commentator Aspasius Aspasius writes: Boulsis appears to be very close to
prohairesis, since first of all, it is in the rational part of the soul, where what most controls prohairesis is
(to kuritaton ts prohaireses), and second, because it is part of prohairesis. [W]henever (hotan) intellect
(nous), after having deliberated, has approved and chosen (sunaises kai heltai), wish (boulsis), being a
desire, has an impulse together with it (sunexhorm aut) [Or: springs forward together with it] (In Eth.
Nicom. 68, 27-30). Aspasius adds that linguistic usage supports Aristotle's technical analysis.
Unfortunately, the text of the commentary breaks off at this point -- it is not clear from the surviving scraps
what Aspasius' intended supporting example is. What is important for my purposes is that a manuscript
reading "kata tn boulsin" is at least as old as Aspasiuss commentary. One extant manuscript of the EN,
"Mb" ("cod. Marcianus 213") supports Aspasius rather than the OCT by reading "kata tn boulsin".
Aspasius may have been commenting on a manuscript of which codex Marcianus 213 is a descendant. If
this is so, then "Mb" has a very old pedigree. If he did not, then there are at least two manuscript traditions
for "kata tn boulsin", and not just one.
32
My interpretation of Aristotelian decision rests on a distinction between the act of deciding
(prohaireisthai) -- making a judgment based on deliberation about what is best -- and the product of that

10

This position is open to a rather obvious objection: decisions don't always result in action. There's
procrastination, there's weakness of will, in fact all kinds of psychological obstacles. How, then, can
decision motivate per se? Aristotle is no stranger to practical irrationality. He tackles the phenomenon head
on in the Nicomachean Ethics book VII. The possibility of practical irrationality seems to face Aristotle
with two equally unpalatable alternatives. He can either (a) take the occurrence of practical irrationality at
face value, as a phenomenon to be explained rather than eliminated from his psychological theory. The cost
seems to be that he will be forced to deny that practical judgments motivate by necessity -- reason can be
"dragged around by passion like a slave", as Socrates puts it. Alternately, he can (b) retain his belief in the
motivating power of reason, and deny the possibility of practical irrationality. But then he risks fighting an
uphill battle against rock-solid evidence (introspective and inferential) that people can and do act contrary
to their deepest held convictions about what they ought to do. Aristotle famously wants to defend the
phenomena, and so it seems that he should land firmly and painfully on the first horn of the dilemma.
In this section, I will argue that the choice facing Aristotle is not quite as stark as this description entails.
The failure of the agent's judgment about what is best, all things considered, to guide her actions in each
act, a decision (prohairesis). I have maintained that decisions motivate in themselves, and that judging that
one ought to ! in C is sufficient to make a decision. Does this really entail that judging that one ought to !
in C is sufficient to motivate the agent to ! in C? One might object that the all-things-considered judgment
is not per se motivational, since what motivates is the desire that is the product of the judgment. But insofar
as this desire appears as a necessary consequence of the judgment -- nothing could possibly prevent the
agent from having the desire once she has made the judgment -- we do not go wrong if we think that the
judgment motivates necessarily, if not per se. The relationship between the judgment and the desire is both
causal and necessary. Making the judgment necessarily produces the desire: nothing could prevent it,
although the relationship between the desire and the action is non-necessary: the desire could be defeated. I
take the past tense at 1113a4 (krithen) and a12 (krinantes) to entail that the decision (which as we know is a
type of desire) is a product of the judgment. My interpretation of the relationship between judgment and
decision has affinities to the one defended by David Charles in Aristotles Philosophy of Action (London:
Duckworth, 1984). But Charles has since changed his view, holding that a prohairesis is a state that can be
understood both "a form of cognition (albeit as a distinctively desiderative form) or as a form of desire,
which involves cognition as to what is best to do)" (D. Charles, "Aristotle's Desire", in V. Hirvonen, T.
Holopainen, and M. Tuominen (eds.), Mind and Modality: Studies in the History of Philosophy in Honor of
Simo Knuuttila (Leiden: Brill, 2006), pp 19-40; p. 34). Although I was initially sympathetic to this reading,
Im skeptical due to Aristotles explicit denial in EN III, 1112a1-13, that decision is some kind of belief
(doxa). His reasons would make Hume happy, as Aristotle underlines that beliefs and decisions are distinct
existences. Aristotle writes: "Beliefs are divided into true and false, not into good and bad, but decisions are
divided into good and bad more than into true and false". Aristotle's implicit conclusion is that it would be
wrong to say that decisions and beliefs form identical genera. Still, he recognizes that this argument does
not prove that decision is not a species of belief: "But", he continues, "Neither is decision the same as any
kind of belief" (all' oude tini) (1112a1). He lists several reasons why this is so: (1) Decisions to do good or
bad actions, not our beliefs, form the characters we have. Here, the argument seems to be that it is possible
for a person to have the right beliefs without making the right decisions. Having general beliefs about good
and bad won't alter your dispositions if you never act on them. (2) We do not believe to take and avoid, we
decide to take or avoid. (3) We praise people for deciding on the right thing, while we praise them for
believing rightly. Aristotle then adds a clarification that is important for our purposes: "We may grant that
decision follows or implies belief (ei de proginetai doxa ts prohaireseos parakolouthei) But that is
irrelevant (ouden diapherei), since it is not what we're asking; our question is whether decision is the same
as some sort of belief (ei t'auton esti dox tini)" (EN 1112a12-13). I take this passage, with its insistence
that decision follows or implies belief, to support my interpretation. For if decision is identical to a
cognitive state, then belief is the best candidate. Decision (the product of an act of deciding, which in turn
is identical to the act of judging that something ought to be done) is not identical to any belief. It is a desire
that is the causal consequence of the agent's endorsement of certain evaluative beliefs concerning what he
thinks he ought to do under specific circumstances. Im indebted to David Charles for discussing the textual
evidence in support of either interpretation with me. I cannot in this paper give my full reasons for rejecting
Charless new view, but hope to do so in a later paper.

11

and every situation does not entail that reason is impotent -- a slave that is always or customarily dragged
around by the passions. Reason is not a natural slave of the passions, to use Aristotle's otherwise
discredited term, even though reason is sometimes subjugated by wayward passions. It is perfectly possible
to accept that the agent's judgment that she should ! suffices to motivate her to ! without denying the
reality of practical irrationality. Aristotle recognizes this possibility.
There are many types of practical irrationality, however, and not all of them conform to the phenomenon
Aristotle discusses. Contemporary philosophers typically define akrasia, or "weakness of the will" 33 in
morally neutral terms: an agent acts akratically if he acts contrary to his best judgment, even if his best
judgment is atrociously bad. Aristotle, in contrast, defines akrasia in morally invested terms: an agent acts
akratically only if he correctly identifies the right action under the circumstances, and thinks that this is
what he ought to do. The weak agent (a) has the right supposition (hupolpsis),34 (b) makes the right
decision about how to act,35 but (c) is led to abandon the right decision due to (d) appetite36 for (e) base
pleasures37 that (f) most people can resist38. Furthermore, (g) the agent must act voluntarily 39, and
consequently cannot prefer the inferior alternative because of ignorance (di'agnoian) of any of the
particulars of the action (although he can act in ignorance (agnon) of some relevant particulars of the
action). For Aristotle, "weakness of the will" is a character flaw because it makes us do what we shouldnt
do, not because he presumes that it is unqualifiedly good to act in accordance with one's evaluative
judgments. If those judgments are false, one would be better off losing nerve and abandoning the original
decision.40 The inquiry into weakness of will therefore belongs in a treatise on ethics, not in a treatise on
psychology proper.
33

"Incontinentia" is the Latin translation of "akrasia". It is often objected that "weakness of will" is an
anachronistic translation of "akrasia", as the Greeks lacked a concept of the will. Aristotle does not
recognize a separate mental faculty called the "will", but rather attributes the functions of the will to reason.
If we presuppose that the will is a faculty separate from reason, weakness of will would indeed be
anachronistic. But it seems that this presupposition is itself anachronistic, as T. H. Irwin has pointed out
("Who Discovered the Will?", Philosophical Perspectives, vol. 6: Ethics (1992), pp. 453-473). For
Aristotle may be thought to have an intellectualist concept of will. That the Greeks concept(s) of will
differ(s) from post-Augustinian ones does not show that they lacked a concept of the will, only that
Aristotle, Augustine, Thomas Aquinas, Maxim the Confessor, and others disagree about what the right
concept of will really is. We dont want to define the concept so narrowly that we lose sight of this debate.
34
"The incontinent person knows (eids) that his actions are base, but does them because of his feelings
(dia pathos)" (1145b12-13; note that "dia pathos" could be translated because of his affliction, referring
to his weak character, not to his feelings). "We might be puzzled about what sort of correct supposition
(ps hupolambann orths) someone has when he acts incontinently" (1145b21). Aristotle does not think it
matters whether the person has the right supposition or whether he has knowledge, as those who believe
can have as strong of a conviction (pistis) as those who know.
35
The incontinent makes a decision that he abandons: 1148a13-17; 1151a5-7; He abandons the right
decision: 1151a29-33.
36
Akrasia is caused by epithumia:1147a33-34. Akrasia strictly speaking is not caused by spirit, thumos:
1149a20-26, people who abandon their right decisions because of spirit are only called incontinent by
transference.
37
"The person who is prone to be overcome by pleasures is incontinent (akrats); the one who
overcomes is continent (enkrats); the one overcome by pains is soft (malakos); and the one who
overcomes them is resistant (karterikos)" (1150a12-15).
38
The excessive appetites cannot be "bestial" in kind or degree, or be caused by deformities or diseases
(1149b27-1150a1). For those who suffer from such afflictions "are outside [human] nature", since they lack
decision and calculation (ou gar echei prohairesin oude logismon). Just like beasts, they cannot act
temperately or intemperately, nor can they abandon their decision, for they have none. Note that this
criterion is not the same as the demand that the agent act voluntarily, for beasts act voluntarily according to
Aristotle: they act on their desires.
39
The akratic acts willingly: 1152a15.
40
So, a person suffering from anorexia who judges that all things considered, she should never eat more
than an apple a day, would benefit if she were to cave in and eat a proper meal. Although he recognizes that
there are such people (1151b24-25), Aristotle curiously denies that there are many people who suffer from

12

Aristotle wants to determine exactly what the epistemic state of the akrats is. To this end, he organizes his
inquiry around Socrates' denial that anyone ever chooses an inferior alternative while possessing
knowledge (epistm) of what is best and his "debunking" explanation of akrasia as nothing other than
ignorance. We are therefore in position to determine where Aristotle stands by establishing where he stands
relative to Socrates' position in the Protagoras.
In his preliminary exposition of the puzzles about incontinence, Aristotle describes Socrates' position as
follows:
First of all, some say [the incontinent] cannot have knowledge [at the time he acts]. For it
would be terrible, Socrates used to think, for knowledge to be in someone, but mastered by
something else, and dragged around like a slave. For Socrates used to oppose the account [of
incontinence] in general, in the belief that there is no incontinence; for no one, in Socrates'
view, supposes while he acts that his action conflicts with what is best; our action conflicts
with what is best only because we are ignorant [of the conflict] (di'agnoian) (VII, 2, 1145b2227).
Aristotle promptly observes: "this argument, then, contradicts things that appear manifestly" (amphisbtei
tois phainomenois enargs) (1145b27-28). Read in isolation from the remainder of Aristotle's argument,
this sentence may be thought to be ambiguous. Does the adverb "enargs" ("manifestly") go with
"amphisbtei" ("contradicts") or with "phainomenois" ("things that appear")? In the first case, Aristotle
would be saying that "Socrates' argument manifestly contradicts the things that appear", thereby expressing
his disagreement with Socrates' diagnosis of akrasia as ignorance. In the second case, he would be saying
that "Socrates' argument contradicts appearances that are manifest", thereby suspending judgment on the
truth or falsity of Socrates' diagnosis.41 Some have been led by the first interpretation to see in Aristotle's
argument a wholesale rejection of Socrates' intellectualism in EN VII.42
Aristotle's subsequent discussion reveals that he is convinced that some form of ignorance underlies the
akratic's behavior. But it is not the kind that makes the action involuntary, nor is it ignorance of good and
bad, for that would make the agent vicious rather than weak. It must be ignorance of a different kind. That
some form of ignorance difference from these two underlie the akratics behavior is confirmed at several
points in the subsequent discussion, where Aristotle says that "[i]f ignorance causes the incontinent person
to be affected as he is, we must look for the type of ignorance that it turns out to be; for it is evident, at any
rate, that before he is affected the person who acts incontinently does not think [he should do the action he

feeling appetites for necessary pleasures less than most people. He thinks people are more inclined to go to
excess.
41
I am indebted to C. C. W. Taylor for pointing out the grammatical ambiguity underlying the different
interpretations.
42
E.g. M. Nussbaum, The Fragility of Goodness (Cambridge: Cambridge University Pres, 1986),
chapter 8, "Saving Aristotle's Appearances". Nussbaum presupposes the first interpretation of Aristotle's
remark: "Aristotle first reports some of our most common beliefs and sayings about akrasia, concluding his
summary with the words, 'These, then, are the things we say (ta legomena)' (1145b20). Next he presents the
Socratic view that nobody does wrong willingly: we choose the lesser good only as a result of ignorance.
Of this theory he says brusquely, 'This story is obviously at variance with the phainomena'. He then sets
himself to finding an account of akratic behavior that will remain faithful to the 'appearances' in a way that
the rejected Socratic account does not" (p. 240). For a criticism of Nussbaum that takes the second
interpretation as authoritative in the light of the ensuing discussion, see J. M. Cooper's review of
Nussbaum, "Aristotle on the Authority of 'Appearances'", Philosophical Review, 97 (1988), pp 543-564,
reprinted in his in Reason and Emotion (Princeton, NJ: Princeton University Press, 1999) pp 281-291, p.
286: "It is evident that, according to his method, Aristotle attaches some weight to the Socratic theory, at
least to something fundamental in it, as well as attaching weight to the appearance with which it conflicts.
Otherwise he could not think, as he does, that there is anything here to puzzle over".

13

eventually does] (1145b28-31)43. And having concluded his "scientific" (phusiks) explanation of akrasia
in VII, 3, Aristotle asks: "How is the ignorance resolved, so that the incontinent person recovers his
knowledge?" (1147b6-7). Although it will transpire that ignorance is not the ultimate cause of the agent's
irrationality -- his wayward appetite is -- it is still the proximate cause, a momentary blindness brought
about by his wayward appetite. The agent acts in ignorance (agnon), although not because of ignorance
(di'agnoian). In this way, Aristotle attempts to preserve the insights in Socrates' description of akrasia as a
form of ignorance, while specifying an explanation of this ignorance that Socrates presumably did not
share, as it presupposes the existence of desires that bypass the agent's better judgment.44 Aristotle's
explanation is laid out in EN VII, 3. 45
In EN VII, 3 Aristotle picks up on a distinction that Socrates introduced in the simile of the aviary in the
Theaetetus (196d-199c). Having a capacity to know is here likened to having a bird in ones cage, while
exercising the capacity is likened to holding the bird in ones hand. Aristotle notes that "we speak of
knowing in two ways", having knowledge (echn) and using it (chrmenos). An agent uses his knowledge
at t1 only if he attends to that knowledge at t1 (thern). He has the knowledge at t1 if he could attend to it
at
t1,
in
conformity
with
Aristotles
distinction between first and second actuality from De Anima (412a23; 417a28) and Metaphysics (1048a34;
1072b24; 1087a20). It would be astounding (thaumaston), Aristotle argues, if someone were to both have
knowledge and attend to it at t1, but still do wrong, while it is not extraordinary at all if someone has
knowledge at t1, but fails to attend to it at t1, and does wrong as a consequence (1147a8-10). This should
alert us to the thrust of the explanation of akrasia that Aristotle is about to present: akrasia is the result of
some kind of failure to use a piece of knowledge that one has in ones cage. Its crucial to determine what
kind of failure to use Aristotle has in mind. Aristotle compares those suffering from strong appetites to men
who are intoxicated, mad or even asleep. Which trait do these men have in common? How are they blocked
from attending to knowledge they have? Aristotles intoxicated man is oinomenos: he drinks moderately,

43

The text Irwin translates is possibly corrupt. C. C. W. Taylor has suggested to me that the phrase
"hoti ouk oietai ge" is a corruption of "hoti gar ouk agnoiei". The sentence then reads "... it is evident that
before he is affected the person who acts incontinently is not ignorant", which makes more sense both
grammatically and philosophically. Aristotle nowhere in his explanation suggests that the person who acts
contrary to his decision and supposition changes his mind about how he ought to act. It's ignorance about a
particular that causes akrasia -- though not ignorance of a particular the awareness of which is needed for
the action the agent actually performs to be voluntary.
44
Whether he could have been brought around to Aristotles point of view is another matter. Insofar as
Aristotles explanation of akrasia presupposes the existence of desires that are not the product of judgments
about what is best, or what one ought to do, Socrates would have to swallow a medium-size camel to accept
Aristotles analysis.
45
Aristotle identifies different kinds of incontinence. The "impetuous" incontinent, who suffers from
propeteia, impulsiveness, never pauses to deliberate, but rushes into action on impulse (1150b19-29). An
impetuous person is like a dog who barks at all visitors without pausing to ascertain whether they are
friends or foes -- his appetite is so fast or so intense that he does not wait for reason, but follows
appearance. The "weak" incontinent, on the other hand, who suffers from lack of strength (astheneia), does
pause to deliberate, and reaches a decision (prohairesis) about what he should do (1151a5-7). But because
he is misled by his appetite (epithumia) he acts contrary to his decision. The "weak" incontinent, then, fails
to act in accordance with his own judgment about what he should do in a particular situation, and this is a
failure of practical rationality, because his actions will not be guided by his conception of the happy life.
Insofar as Aristotle counts even impetuousness as a kind of akrasia, he must think that the impetuous agent
acts contrary to deliberation he has completed at some point at a greater temporal remove from the situation
in which he succumbs to passion than the weak incontinent (see Irwin, comment ad 1150b19-29). If this is
so, the impetuous person has deliberated, and has adopted a conception of the happy life, but he will still
need to complete his thinking about how to act here and now. But this presents a problem for Aristotle's
description of decision as the efficient cause of action (hothen h kinsis) (1139a31). For nothing is
supposed to intervene between an efficient cause and its effect (Physics, 243a33-34). See D. Charles,
Aristotle's Philosophy of Action, p. 139.

14

and is thus tipsy, not punch-drunk like the methusos.46 Nor is the madman punch-drunk: madmen are aware
of most things that happen around them, and sometimes excessively aware of their own principles. Sleepers
are often conscious of their own dreams. So the common trait cannot be that these men are knocked out. If
they were knocked out, their actions would not be voluntary. Rather, Aristotle emphasizes that these men
merely say the words without attending to their knowledge. Saying the words is not necessarily a sign
that the agent understands what he is saying, for people who suffer from strong feelings, or who are mad,
intoxicated, or asleep, even recite demonstrations and verses of Empedocles, but for all that, they are still
not exercising their knowledge. Ask for an account, and you may get either more or less than what you
asked for. The same goes for learners who merely string together the words (1147a21). We should not
assume that these men fail to exercise their capacity for knowledge because they dont speak Greek or
dont understand the meaning of individual Greek sentences. The failure is rather their inability to give a
correct account of the subject matter of their words. In the Ion, Plato charged rhapsodes with lacking
understanding, although they presumably understand the dictionary meanings of the words and the
sentences that they speak. The rhapsodes fail to know their subject because they dont see the right
connections between the lines and words they proclaim. In EN VII, 3, Aristotle charges the akrats with a
temporary affliction of the same kind: And so we must suppose that those who are acting incontinently
also say the words in the way that actors do (hste kathaper tous hupokrinomenous) (1147a22-24). Actors
may very well assent to the lines they recite. But for all that they won't be able to explain the subject matter.
As far as Aristotle is concerned (EN VI, 2, 1139b32; 1139a19-32, with explicit reference to the Analytics;
also 1140b35), knowledge (epistm) is a demonstrative state (hexis apodeiktik). A man has knowledge if
he (1) has the appropriate sort of confidence (pistis) and (2) knows the principles (archai), and (3) knows
them better than the conclusion. It is a demonstrative state, because one must be able to prove the
conclusion from the principles. This entails that one only knows (has epistm of) a specific proposition if
one knows its relation to other propositions.47 Exercising one's knowledge means paying attention to only
the right inferential connections. Unlike Platos rhapsode, the akrats does have the ability to exercise his
knowledge while he is not under the influence of passion. His ignorance does not consist in a failure to
state individual premises in his practical deliberation, but in a momentary failure to make only the right
inferential connections between them. His deliberation derails after it has reached the intended station.
Aristotle now ties his distinction between having and using to the distinction between knowledge of the
major premise in the practical syllogism and knowledge of particulars (the minor premise(s) of the practical
syllogism (EN 1146b35-1147a10). His explanation should show how the agents reaches the right decision,
but still acts against it, and must thus explain how the agent reaches the right conclusion about what he
should do (after all, the decision is produced by the agents assent to the conclusion I should ! in
circumstances C), but abandons it because of passion (pathos). The explanation Aristotle seeks should
show how those affected by strong feelings are like the tipsy, mad or sleeping person, in that he does not
use the knowledge that he has. The ignorance cannot be ignorance of individual premises, for if the agent is
ignorant of the major premise, he will be momentarily vicious, and if he is ignorant of the minor
premise(s), he will be acting involuntarily.
To the delight and frustration of his readers, Aristotle depicts the plight of the weak incontinent by casting
his final piece of deliberation 48 in the form of two competing "practical syllogisms". The "good" syllogism
46

I'm indebted to David Charles for drawing attention to the distinction between different states of
inebriation in Greek. In my experience, when "oinomenos" is translated "drunk", readers used to a Northern
or Eastern drinking culture will think of a man who is "punch-drunk" (i.e. drunk by Scandinavian, Slavic or
Russian standards), while those who approach the translation from the perspective of a more genteel
drinking culture will read "drunk" as "tipsy".
47
After I completed an earlier draft of this paper, I became aware that David Charles emphasizes this
aspect of knowledge in his unpublished manuscript on NE/EE VII.3. I have deepened my argument in line
with Charless.
48
Readers will immediately note that I dont accept J. M. Coopers account of the scope of deliberation in
Aristotle, although I won't defend my view here. See J. M. Cooper, Reason and Human Good in Aristotle
(Indianapolis: Hackett, 1975), chapter 1, for a defense of the view that the practical syllogism is not part of
the deliberative process. A. Kenny offers a rejoinder to Coopers view in Aristotles Theory of the Will
(New Haven: Yale University Press, 1979), p. 113.

15

of reason depicts the reasoning leading up to the judgment that produces his decision (prohairesis) 49, while
the "bad" syllogism depicts the derailment which occurs when appetite intervenes and "hijacks" the minor
premise of the agent's final deliberations and puts it to its own sinister use (EN 1147a31-35).50 A standard
reconstruction, which has been dubbed the "Cornell model" 51 is this:
Nothing sweet ought to be tasted
This is sweet

Everything sweet is pleasant

This is sweet

Don't taste this ("Avoid this")

Appetite enters:
The agent incontinently eats

("Charlie Bucket") 52

("Augustus Gloop")

49

Because Aristotle thinks that virtue is defined with reference to the agents decision I cannot accept J.
M. Coopers suggestion (Reason and Human Good in Aristotle (Indianapolis: Hackett, 1975)) that for
Aristotle, a decision is a general policy captured in the major premise of the syllogism (p. 46). Whether the
agent is virtuous or not does not merely depend on whether he adopts the right general policy, it depends on
whether he can determine when the policy applies. A decision to perform a particular action may of course
be defeated by appetite or anger, but this does not help Cooper, since this objection would apply to his
interpretation as well as mine. Moreover, a definition should cite actual causes (and not mention possible
defeaters, as there could be an awful lot). And being able to determine when a policy applies is a positive,
necessary condition for virtue. D. Charles adduces another reason for denying that the decision (what
Charles calls "preferential choice" is captured by the major premise: "Preferential choice is described as a
proximate cause of praxis (1139a31) and of action more generally (701a4-5, cf. a34-36). Since the
proximate efficient cause is simultaneous with the effect -- in the sense that nothing intervenes between
them (243a33-34), there can be no further stages of deliberation which intervenes between the preferential
choice and the praxis. Hence preferential choice cannot be for a general policy antecedent to action which
requires technical deliberation to implement" (Aristotle's Philosophy of Action, pp. 139-140). By the same
token, no practical syllogism can intervene between the agent's deliberation and decision and his action, the
view that Cooper defends (Reason and Human Good in Aristotle, p. 46).
50
I take "teleutaia protasis" (1147b9) to refer to the last premise of the syllogism (see Cooper's defense of
this claim in Reason and Human Good in Aristotle, note 62, p. 49). But as will momentarily become clear, I
understand the weak akrats' failure to "use" the "teleutaia protasis" to consist in his failure to relate the
minor premise only to the good major premise, and thus his failure to use it only as a minor premise in the
syllogism leading up to the good conclusion. It is not a failure to be conscious of it, or even conscious of it
in relation to other premises. The weak akrats may therefore be conscious of the "good conclusion" even
as he acts, thought this falls short of exercising one's knowledge (epistm) of it. This result is inimical to
the view of those who, with Hardie (Aristotle's Ethical Theory, pp. 287-289), take "teleutaia protasis" to
refer to the conclusion of the "good syllogism", for they cannot explain the agent's failure to exercise his
knowledge as a failure to connect the conclusion to any other proposition. The failure to exercise or attend
to the conclusion must consist in lack of consciousness of it, or something or the sort. On the view I defend,
the akrats does not exercise epistm of the good conclusion (as defined in the Analytics and in EN
1139b32 and 1140b35), as he does not know it as part of a body of demonstrative knowledge. If I base C on
P1 and P2, but I do not know P1 better than C, as I misapply P1elsewhere, I don't know C strictly speaking,
although I can be conscious of C even as I act.
51
The name was suggested to me by C. C. W. Taylor. Having spent a semester as visiting professor at
Cornell, Taylor noted that it enjoyed widespread acceptance there. I will discuss Taylors alternative
reconstruction below. Taylor attributes different minor premises to the two syllogisms.
52
This syllogism is not spelled out by Aristotle, but the context makes it plausible that this is the form
he intends -- Aristotle has just mentioned the inference "Everything sweet ought to be tasted ("pantos
glukeos geuesthai dei"); this, some one particular thing, is sweet; so taste this". I'll assume that he intends
the first premise of the virtuous syllogism to have the same form (infinitive + dei-construction, or, what
amounts to the same from a psychological perspective, a gerundive). For an alternative construction, see
below.

16

For convenience, and following established Medieval practice, I have taken the liberty of baptizing both
syllogisms. The names reflect the content of the syllogism, not their logical form. The left hand "syllogism
of reason" is named after the modest and virtuous protagonist of Roald Dahl's classic children's book
Charlie and the Chocolate Factory, Charlie Bucket.53 Aristotle's sweet-tooth syllogism, on the other hand,
is named after a less admirable character in the same book, the gluttonous Augustus Gloop. As you may
recall, Augustus "was deaf to everything except the call of his enormous stomach", and nearly drowns
himself in a sea of liquid chocolate.54
Unlike the major premise of "Charlie Bucket", the major premise of "Augustus Gloop" lacks the force of a
gerundive. There is no term indicating that the agent thinks that everything sweet should be tasted, ought to
be tasted, or is to be tasted. It simply states that "everything sweet is pleasant".55 The reason Aristotle
formulates the major premise of the Augustus syllogism in this way, is presumably that appetite does not
by itself reach conclusions about what one ought to do. Appetite does not have practical principles. Only
reason makes all-things-considered judgments based on deliberation about what the best kind of life would
be; these are judgments about what the agent ought to do in particular situations to promote his happiness. 56
It should also be noted that it is entirely rational to hold both the first premise of "Augustus" and of
"Charlie" to be true at the same time. The practical problem occurs if the major premise of "Augustus" is
allowed to determine the agent's actions, in opposition to what the agent thinks he ought to do. Augustus'
failing, then, is not simply judging that everything sweet is pleasant (Charlie, our virtuous agent, might
concede as much), his failing is being tempted to taste this particular sweet thing because of the "call of his
enormous stomach". Even if he first rehearses the "good syllogism", he soon attaches the minor premise of
perception "this is sweet" to the belief that "everything sweet is pleasant", and since he has an enormous
appetite for everything sweet, the conclusion of adding these beliefs up determines his actions. He should
listen less to the hungry roar in his stomach, and more to the voice of reason.57
How, exactly, does the conflict between reason and appetite play out? Aristotle describes the motivational
conflict of the "weak" incontinent as follows:
53

Roald Dahl, Charlie and the Chocolate Factory (New York: Alfred A. Knopf, 1996). Charlie's family
has nothing but cabbage soup for supper, they feast on double helpings of cabbage soup on Sundays.
Despite severe malnourishment, Charlie does not fall for the temptation to sample Willy Wonka's products
during the tour of the factory. Although Charlie does not think that sweets are bad per se, he still exercises
restraint under particular circumstances. We can imagine that he rehearses the syllogism under those
circumstances.
54
Chapter 17: "Augustus Gloop Goes up the Pipe", p. 72. The attentive Dahl-reader may object that
Augustus is a glutton on principle, not out of weakness or even impetuousness. He has adopted the
maximum consumption of sweets as his highest practical principle. But I'll give Augustus the benefit of the
doubt, and assume that he actually wants to be a good and self-controlled boy. He does not judge that all
things considered, he ought to gobble up whatever chocolate he can get his sticky, stocky fingers on.
55
Unlike the example which Aristotle presents in the preceding paragraph (1147a29), which includes
the principle "Everything sweet must be tasted" (pantos glukeos geuesthai dei). This principle has the force
of an all-things-considered judgment. It is the principle of a chocolate sybarite.
56
Aristotle's distinction in De Motu Animalium 701a23-25 between the premise of the "good" (dia tou
agathou) and the premise of the "possible" (dia tou dunatou) applies to all types of practical syllogism, and
maps the same distinction as the "universal" - "particular" distinction which Aristotle relies on in the EN.
By a "premise of the good" he does not just refer to the major premise which is an all-things-considered
judgment. (i.e. a belief abut what is good on the whole). As he points out (700b28-29), "that for the sake of
which" we act is always presented to us in the guise of the good, and since pleasure is an apparent good, a
major premise that mentions pleasure as a feature of some object, e.g. "all sweets are pleasant", therefore
counts as a premise of the good.
57
Aristotle proposes a mental exercise: just like first-strike ticklers brace themselves before a
counterattack, we should brace ourselves against weakness: "For some people are like those who do not get
tickled themselves if they tickle someone else first; if they see and notice something in advance, and rouse
themselves and their rational calculation, they are not overcome by feelings, no matter whether something
is pleasant or painful" (EN 1150b22-25). There are mental prophylactics; ways to focus one's attention
away from aspects of a situation that tend to trigger unwanted behavior.

17

Suppose, then, that someone has the universal belief hindering him from tasting; he has
the second belief, that everything sweet is pleasant and this is sweet, and this belief is
active (energei); but it turns out that appetite is present in him. The belief, then, [that is
formed from the previous two beliefs] tells him to avoid this, but appetite (epithumia)
leads him on, since it is capable of moving each of the [bodily] parts (EN 1147a31-35).
Again, appetite is not represented as a universal all-things-considered judgment of the kind that makes up
the major premise of the syllogism of reason. It enters from "outside" the agent's deliberations about what
is best. The incontinent agent would not endorse the principle "everything sweet ought to be tasted" as a
part of his explicit deliberation. It is furthermore clear from Aristotle's description that the incontinent man
has reached a decision not to eat this particular sweet, the conflict is between the dictate of reason ("don't
taste") and the impulse of appetite. Even so, Aristotle thinks both that there is some kind of cognitive
deficiency underlying the incontinent agents failure to stand by his decision and that "in a way, reason and
belief make him act incontinently" (1147b1-2). Why does he think that reason and belief "in a way" (ps)
make him act incontinently? Because the second belief "this is sweet" awakens his appetite, and this
particular belief is contrary to the correct reason ("Don't taste this") not in itself (kathhautn), but
coincidentally (kata sumbebkos). That is, while there is no real contradiction between holding both "this is
sweet" and "Dont taste this" to be true at the same time, the incontinent attaches the "minor" premise to the
wrong major premise, namely "Everything sweet is pleasant". Because "This is sweet" is permitted to
become active relative to this major premise, it is the cause of an appetite that runs contrary to right
reasonthe conclusion of the good syllogism. Aristotle, then, conceives of the real conflict as a conflict
between two desires, the appetitive desire and the decision, and since decision is the product of a judgment
of reason based on deliberation about what is best, it is also a conflict between appetite and reason -precisely the kind of conflict that Hume denies is possible. The conflict is a conflict between desires of
different kinds. In EN I, 13, 1102b21, Aristotle writes that the incontinent has "impulses (hormai) in
contrary directions". The conflict is not simply a conflict between an evaluative judgment and an appetite.
For the evaluative judgment produces a desire the decision. And the decision, although it is an impulse in
the right direction, is defeated.
How, then, does this explanation relate to Aristotles insistence that there is an epistemic error involved in
cases of akrasia? Having laid out the explanation above, Aristotle immediately asks: "How is the ignorance
resolved, so that the incontinent person recovers his knowledge?" (1147b6-7). His answer is cryptic, but
ultimately intelligible: "Since the last premise (teleutaia protasis) is a belief about something perceptible,
and controls action, this is what the incontinent does not have when he is being affected. Or [rather], the
way he has it is not knowledge of it, but, as we saw, [merely] saying the words, as the drunk says the words
of Empedocles" (1147b9-12). If the reconstruction above is correct, and the minor premise is identical for
both the "good" and the "bad" syllogism, the agent merely says the words because he does not connect
the minor premise only to the right major premise, but lets it be dragged off where it does not belong. He
has one thought too many. And having one thought too many is a cognitive deficiency: it reveals that the
agent is not attending to his knowledge. As Socrates taught Aristotle, exercising ones capacity for
knowledge requires the ability to give an account, and that requires the ability to see how ones convictions
hang together. Having and attending to ones practical knowledge requires making only the right judgment,
and not mixing in considerations where they do not belong. A mathematician who makes a valid deduction
from true premises and then goes on to misapply some of these premises elsewhere does not really know
his premises.
Aristotle takes strong affections (path) to be the ultimate cause of the akratic's failure: For spirited
reactions, sexual appetites, and some conditions of this sort clearly [both disturb knowledge and] disturb
the body as well, and even cause fits of madness in some people (1147a15-17). To move from Dahls
innocent childrens story to a slightly less innocent story: When Humbert Humbert seizes up the recently
orphaned Lolita, he sees a thirteen-year-old girl in need of a protector. Heres a thirteen year old orphan
girl, Humbert thinks to himself: I should protect her (an enthymeme). But Lolitas vulnerability, the
reason he should protect her, also awakens his dormant sexual desire for nymphettes. We may say that
Humbert, due to his desire for young nymphettes has one thought too many in the course of reaching the
right conclusion, I should protect her. By thinking about the right thing to do, Humbert ends up with a

18

desire a stronger desire, it turns out for the wrong course of action. His train arrives at the right station,
but then moves beyond it due to his sexual desire. Drawing on a distinction from EN III, 1, Aristotle can
argue that ignorance of good and bad is not the ultimate cause of the wrongdoing, nor is ignorance of e.g.
the minor premise that "this is sweet" the ultimate cause. After all, the agents attention to the minor
premise is what awakens his desire. Rather, the presence of strong affections makes the agent act in
ignorance: "Action caused by ignorance would seem to be different from action done in ignorance. For if
the agent is drunk or angry, his actions seems to be caused by drunkenness or anger, not by ignorance,
though it is done in ignorance, not in knowledge" (1110b24-27)58.
Aristotle concludes his phusiks explanation of akrasia by noting that the result Socrates was looking for
would seem to come about, as what is dragged around because the incontinent is affected is not knowledge
strictly speaking (epistm), but perception.59 Because he fails to connect the minor premise to only the
right major premise, the incontinent agent's knowledge of good and bad is not really dragged around by
passion. For knowledge of the minor premise ("This is sweet") does not amount to knowledge in the full
extent (kuris epistme), but rather perceptual knowledge (ts aisthtiks) (EN 1147b13-17).60 This
58

In this context, Aristotle speaks of the drunken man (ho methun), not the tipsy one. But I think his
reasoning applies to the tipsy man as well, if tipsiness is the pathos that makes him inattentive. The angry
man (ho orgizomenos) need not be furious; he could just be annoyed. It's frequently noted that Aristotle
denies that ignorance caused by drunkenness is a valid plea for mercy: he seconds the policy according to
which wrongdoing committed under the influence justifies double penalties: "Indeed, legislators also
impose corrective treatments for the ignorance itself, if the agent seems to be responsible for the ignorance.
A drunk, for instance, pays a double penalty; for the principle is in him, since he controls whether he gets
drunk, and his getting drunk causes his ignorance. They also impose corrective treatment on someone who
[does a vicious action] in ignorance of law that he is required to know and that is not hard [to know]. And
they impose it in other cases likewise for any other ignorance that seems to be caused by agents
inattention; they assume it is up to him not to be ignorant, since he controls whether he pays attention. But
presumably he is the sort of person who is inattentive. Still, he is himself responsible for becoming this sort
of person, because he has lived carelessly" (EN III, 5, 1113b30-1114a5). The incontinent, just as the tipsy,
person would seem to be inattentive.
59
His emphasis that the failure is perceptual explains his appeal to sleepers earlier on in the chapter. For in
De Insomniis, 1 (454a1-6) we learn that being awake is "the exercise of sense-perception", and that sleep is
the contrary of this.
60
An alternative reconstruction of the agent's deliberations from the time he starts deliberating, through
the time he makes his decision, to the time he abandons his decision and acts incontinently, is the
following, the core of which has been suggested by C. C. W. Taylor. Aristotle nowhere spells out the exact
form of the premises of the syllogism of reason, and so we may think that those premises do not mention
sweetness at all:
(1) Nothing unhealthy ought to be tasted
(2) This is unhealthy
(3) So, this should not be tasted

(4) Everything sweet is pleasant

(5) This is sweet
Appetite enters:
The agent incontinently eats

The agent no longer pays attention to premise (2) and presumably not to the conclusion (3) when he
proceeds to think that everything sweet is pleasant and this is sweet. At the time of action, (2) and (3) are
no longer active, while (4) and (5) are. After he has devoured the sweet, the agents inattention disappears,
he again pays attention to (2), reiterates the "good syllogism", and consequently regrets what he has done.
Unlike the Cornell reconstruction, Taylor's model does not assume that the two syllogisms have any
premises in common. This reconstruction has one supposed advantage over the Cornell model: the agent's
ignorance of the minor premise may be complete neglect of it, not just the failure to exercise knowledge by
using a premise where it does not belong. A drawback is that "health" is never mentioned as a consideration
in the context of Aristotle's argument. Furthermore, the Cornell model permits us to explain in a
straightforward way why the agent's deliberation goes astray: perceiving that "This is sweet" as part of the
virtuous deliberation can in itself awaken the incontinent agent's latent appetites. Taylor's model, with
different minor premises, does not.

19

acknowledgement importantly does not imply that Aristotle's explanation of akrasia conforms to Socrates'
debunking account, since Socrates' position is that the agent's ignorance about good or bad -- captured by
the major premise in the Aristotelian syllogism -- is the cause of his wrongdoing. Aristotle permits that an
agent can know what is better and worse, and still fail to act on this knowledge due to inattention. What he
does not permit is that an agent may have all and only the right inferential connections present-in-mind,
have made the right decision, and still fail to carry it out because he is overcome by contrary appetites.
Appetites cause incontinent episodes by skewing the agents appreciation of the particulars of her
circumstances.
Now, if it is possible to act incontinently, due to the breakdown of inferential connections due to passion,
we may want to puzzle over the following description of the relationship between concluding a practical
syllogism and acting. Immediately before depicting the two syllogisms, Aristotle makes the following
observation:
[O]ne belief is universal (katholou doxa); the other is about particulars (peri tn
kath'hekasta), and because they are particulars, perception (aisthsis) controls them. And
in the cases where these two beliefs result in one belief (mia gentai), it is necessary, in
one case, for the soul to affirm (phanai) what has been concluded, but in the case of
beliefs about production, to act at once on what has been concluded. If for instance
everything sweet must be tasted, and this, some one particular thing, is sweet, it is
necessary for someone who is able and unhindered (ton dunamenon kai m klumenon)
also to act on this at the same time (hama touto kai prattein) (EN 1147a25-31).61
This description seems to imply that making an inference about what one should do necessitates the
corresponding action. There is an analogy between the necessary relationship between the premises and
conclusions of a theoretical syllogism and the relationship between reasons for acting and action.
On the basis of a similar section in De Motu Animalium 701a and following, some critics have even been
led to attribute to Aristotle the fantastic theory that the conclusion of a piece of practical reasoning is an
action. But on the most obvious reading, this is a mistake. The passage I just quoted fortunately does not
support this interpretation. Aristotle clearly distinguishes between drawing the conclusion and acting on it.
The thing the agent must "also act on" is, I take it, the content of the conclusion of his deliberation ("to
sumperanthen"). In this case, "This should be tasted". This indicates that in the Ethics, at least, Aristotle
does not identify the conclusion of a practical syllogism with the action that the agent has chosen to
perform. That which has been concluded is rather the content of a doxa, or a belief, and the belief that "I
should !" is not identical to the action I will perform if I act in accordance with my convictions, namely a
token !-type action. A further piece of evidence that in the Nicomachean Ethics, the conclusion of
deliberation is not identical to the action is Aristotle's reference to "two beliefs resulting in one" ("mia").
The feminine numeral "mia" has "doxa" as its antecedent: what has been concluded when I complete a
practical syllogism is not the action I decide to perform, but that I should perform this action. And this is a
belief.62 Even without addressing the intricacies of the Greek, it is a puzzle of the identity-thesis ("the
conclusion is the action") how the agent comes to "syllogize" or "put together" the two beliefs in the first
place, if not by drawing the conclusion that the premises jointly entail. After all, it is necessary that the
agent see what follows from the premises. Without seeing what the premises entail, the agent would simply
not have a motive to perform the action that the premises jointly give him a reason to perform63.
61

I take "hama" to mean that action is "immediate", in the sense that nothing else is needed. This
coheres with Aristotle's description of decision as the efficient cause of action (1139a31).
62
See J. M. Cooper, Reason and Human Good in Aristotle (Cambridge, Mass.: Harvard University
Press, 1975) p. 48 n. 61; M. Nussbaum, "Practical Syllogisms and Practical Science", in Aristotle's De
Motu Animalium (Princeton, NJ: Princeton University Press, 1978), p. 204, n. 50. D. Charles (unpublished
manuscript on NE/EE VII, 3) has argued that rather than "doxa", "mia" has "protasis" as antecedent, and
uses this to conclude that "protasis" may refer to the conclusion of the syllogism. "Teleutaia protasis" thus
means "final proposition".
63
Philip Clark is, in his own words, "one of few philosophers willing to follow Aristotle's lead in
thinking that the conclusion of a piece of practical reasoning is an action" ("The Action as Conclusion",

20

Many commentators have been embarrassed by Aristotle's insistence that the premises of a practical
syllogism "necessitate" the action which the conclusion picks out. The embarrassment is exacerbated by the
fact that Aristotle adds a conspicuous qualification to the claim that it is "necessary" to act on "what has
been concluded". If you are capable, and not prevented from acting, you will act. If judging that one ought
to stay away from x is sufficient to motivate the agent to stay away from x, how can the judgment also be
overridden? What are we to make of his claim that it is necessary to act, at once, on what has been
concluded, for anyone who is able and unhindered? The first reaction may be to ask, well, is it necessary or
is it not? In the following section, I attempt to give credence to Aristotle's "necessity-with-qualifications"
theory of the motive power of decision.
Internalist Interlude: Practical Judgment and Motivation
In order to see how Aristotle can navigate these choppy waters, it will be helpful to distinguish between
two conceptions of motivational internalism, captured by a strong and a weak thesis. The problem this
distinction will help us sort out is this: Given that Aristotle accepts the possibility of practical irrationality,
how can he be an internalist about the motivating power of the all things considered-judgments? I will
defend the weak thesis as an adequate formulation of internalism, and argue that Aristotle can think of
decisions as intrinsically motivational although they don't always lead to action.
The terms "internalism" and "externalism", in circulation for half a century, are currently in a state of
semantic flux.64 Rather than attempt to capture some uncontroversial "core" meaning of the terms -- an
impossible task, I suspect -- I will simply stipulate the weak and strong sense of "motivational internalism".
My stipulations owe their inspiration to Michael Smith's definitions of strong and weak internalism in The
Moral Problem.65 According to Smith, proponents of strong internalism hold that it is a conceptual truth
that if an agent judges that it is right for her to ! in circumstances C, then she is motivated to ! in
circumstances C.66 But the expression "having a motive to ! in circumstances C" has fuzzy boundaries.
After all, I can be somewhat motivated to ! in C, while still being more motivated not to ! in C. If judging
that it is right for me to ! in circumstances C just makes me somewhat inclined to ! in circumstances C, but
never sufficiently motivated to cause me to ! in C, it seems that the internalist will have to give up a very
appealing type of reasons-explanation: citing the agent's judgment won't suffice as an explanation of why
she !ed in C when that is what she did. Part of the appeal of internalism about practical judgment is, I take
it, that it allows us, in circumstances to be specified shortly, to explain the agent's actions by citing her
judgments about what she should do.
On my preferred interpretation, then, the strong thesis states that:
(1) If an agent judges that she should ! in circumstances C, she will ! in circumstances C, unless she is
prevented by external obstacles.
Canadian Journal of Philosophy vol. 31, 4 (2001), pp. 481-506. Whatever the philosophical merits of this
theory are thought to be, the theory does not obviously have an Aristotelian pedigree, for the reasons I have
canvassed above.
64
It is, I take it, an uncontroversial fact that Aristotle is an externalist about reasons for action. An
agent may have good reasons to ! in circumstances C without actually desiring to ! in circumstances C.
For she may lack the true belief that !-ing in C is in her own best interest. The function-argument
underwrites this interpretation. The question about what is the highest good for any agent can be answered
objectively, Aristotle submits, and so, therefore, can questions about what any agent ought to do. I'll leave
this intriguing aspect of Aristotle's ethics to the side.
65
M. Smith, "The Externalist Challenge", chapter 3 in his The Moral Problem (Oxford: Blackwell
Publishers, 1994), p. 61.
66
The idea behind the strong thesis is that there is a conceptual connection between judging that
something is right and being motivated to pursue it. I deny that the connection is conceptual. Thus, the fact
that the agent is sufficiently motivated to ! in circumstances C does not follow logically from the fact that
she has judged that she ought to. The connection is causal.

21

In other words, according to the suitably fleshed out "strong" formulation of the idea behind motivational
internalism, only non-psychological obstacles may prevent the agent from !-ing in C if she judges that she
ought to -- shackles, flat tires, and other "acts of God". Absent such intervening causes, the agent will ! in
C, since that is what she is motivated to do in virtue of her judgment. Unless she changes her mind, she will
act. Adherents to the strong conception thus hold that the agent's practical judgment at t1 in each and every
case disposes the agent to act as she thinks she should at t1. If she doesn't, no psychological account will
ever explain why she didn't.
This, I submit, is not the internal relation between judgment and motivation which Aristotle accepts. He
accepts a weaker thesis. The weaker thesis does not place such stringent demands on the connection
between practical judgment and motivation to act. While preserving the internal connection between the
two, the weaker conception nevertheless allows that psychological states -- i.e. desire-states that are not the
product of practical all-things considered judgments -- may intervene and sway the agent's will contrary to
her judgment. The weaker thesis thus requires its adherents to hold that the possibility of practical
irrationality -- including akrasia -- does not rule out a full-blooded internalist position. We may present the
weaker thesis as a conditional with a disjunctive consequent67:
(2) If an agent judges that she should ! in circumstances C, then, unless she is prevented by external
obstacles, either she will ! in circumstances C, or she is practically irrational.
If the agent fails to act in accordance with her judgment, and there are no external obstacles, we must be
able to explain her behavior in terms of practical irrationality -- which is not to say, I hasten to add, that her
action is beyond our comprehension in principle, in which case the weaker thesis would be a cop-out. The
second, but not the first expression of the idea behind internalism allows that the relation between practical
judgment and motivation is a defeasible one. Internal obstacles, such as irrational desires, cravings, fears
and proclivities may prevent the agent from acting on her considered judgment.
Why should the externalist permit the internalist to conceive of the relationship between practical judgment
and motivation as a defeasible one? Isn't the possibility of practical irrationality the major piece of evidence
in the case against internalism? Many have thought so.68 To see why were justified in thinking that the
weaker thesis is a legitimate formulation of the internalist position, it helps to consider an argument from
physics. Consider the following arguments about the dispositions of matches, which Robert Brandom
presents in his Articulating Reasons69:
1. If I strike this dry, well-made match, then it will light (p-->q)
2. If p and the match is in a very strong electromagnetic field, then it will not light (p & r --> - q)
3. If p and r and the match is in a Faraday cage, then it will light (p & r & s --> q)
4. If p and r and s and the room is evacuated of oxygen, then it will not light (p & r & s & t---> - q)
In this case, as Brandom points out, the relationship between p and q is defeasible -- the addition of further
premises leads the conclusion to oscillate from q to not-q. The inference in 1 is "non-monotonic" -although q follows from p, q does not follow from p and r. The addition of another premise unsettles the
inference. Perhaps the denial of all possible unsettling conditions is tacitly presupposed in 1. and 3. But
trying to cash this thought out appears to reduce the expression of the relationship between p and q to
67

Again, following Smith, The Moral Problem.

Bernard Williams is a notable exception, see his "Internal and external reasons", reprinted in his Moral
Luck (Cambridge: Cambridge University Press, 1981), p. 107. While defending internalism, Jonathan
Dancy grants that the existence of akrasia is a major stumbling block for internalists (Jonathan Dancy,
Moral Reasons (Oxford: Blackwell Publishers, 1993). But Dancy argues, convincingly to my mind, that the
internalist has the resources to accommodate the existence of akrasia and other forms of practical
irrationality -- such as the depressive's failure to act as he thinks he ought.
69
R. B. Brandom, Articulating Reasons: An Introduction to Inferentialism (Cambridge, Mass.: Harvard
University Press, 2000), p. 88.
68

22

vacuity -- q follows from p except in the cases where conditions that prevent q from following from p
obtain. Adding a ceteris paribus-clause to make the inference monotonic will yield a Pyrrhic victory,
Brandom argues. He therefore enjoins us not to treat a ceteris paribus-clause as a panacea, "a deus ex
machina that magically removes the non-monotonicity" 70: the ceteris paribus-clause rather signals that the
inference is non-monotonic. Similarly, it wont do for the internalist to claim that agents are always
motivated to act in the way they judge they ought to act, all things considered, unless for some reason they
arent. That won't save the strong formulation of internalism, since it reduces the formulation to a truism "either the agent is motivated to act in accordance with her judgment or she isn't". The list of defeaters must
be finite.
Inferences about an agent's particular practical judgments and her motivation are non-monotonic. My claim
is that conceding this point does not threaten internalism. We can see why this is so by considering the
following inferences about the motivational efficacy of my practical judgment in different psychological
environments:
1. If an agent judges that she should ! in C then she will be motivated to ! in C (p -->q)
2. If she judges that she should ! in C, but she is tempted by the sensuous gratification of not !-ing, then
she will not be motivated to ! in C (p & r --> - q)
3. If she judges that she should ! in C, and she is tempted by the sensuous gratification of not !-ing, but she
has developed a psychological skill which permits her to deflect such temptations when they arise, for
instance some cognitive technique, then she will be motivated to ! in C ( p & r & s --> q)
And so on.
Let us assume that the agent's judgment about what she ought to do in my example corresponds to the
match in Brandoms example. The agent's wayward desires are like the strong electromagnetic field, while
the focusing technique functions as a psychological Faraday cage. What the agent will be motivated to do
depends on her practical judgment. But the presence of intervening psychological states may override this
motivation. To put the point in slightly different terms, the general propensity of all-things-considered
judgments to provide a motive for action may be "masked". Just as we can predict the course of two billiard
balls that collide on a table on the assumption that there is no third ball in play, we can predict how an
agent will act on the basis of her practical judgment on the assumption that she is not practically irrational.
At this point, I think we are entitled to at least a comparative thesis: the claim that practical judgments
motivate is no more undermined by the existence of akrasia than the claim that dry, well-made matches
light when struck is undermined by the existence of electromagnetic fields. Both claims concern defeasible
relations. The existence of motivational states internal to the agent which prevent the agent from acting on
her practical judgment -- sudden anger, wayward passions, psychological compulsions or other proclivities
does not by itself imply that there is no internal connection between practical judgment and motivation.
Perhaps one might object that the general truth that practical judgments motivate, ceteris paribus, does not
permit us to make any inferences about the motivational consequences of a particular agent's practical
judgments. If I know that the agent is weak-willed -- of the Augustus Gloop-type -- and I know that he is
touring Willy Wonka's chocolate factory, I may want to doubt whether his judgment about what he ought to
do, all things considered, will prevent him from tasting. But the reason we hesitate before inferring that
Augustus will abstain is not that practical judgments don't motivate, but our knowledge of Augustus'
appetite. We suspect that his judgment will be prevented from manifesting its effect. We make inferences
to the next case with a proviso, just as we do in the matchstick-case: when I judge that "if I strike this dry,
well-made match it will light", I assume that the match is not in a very strong electromagnetic field. That, I
hasten to add, does not make the presupposition part of the premises of the inference -- for the list of
potential defeaters would be infinitely long, and hence I could never reach my conclusion since the number
of absent conditions cited in the premises would be endless. There is a lesson to be learned here about the
putative "causal contribution" of absences: it is nil. As D. M. Armstrong writes, "omissions and so forth are
not part of the real driving force in nature. Every causal situation develops as it does as a result of the

70

Brandom, Articulating Reasons, pp. 88-89.

23

presence of positive factors alone".71 There is thus an explanatory asymmetry between the absence of
preventing causes and the presence of preventing causes. The absence of preventing causes plays no role in
explanations, while the presence of preventing causes does. The only time one might want to mention the
absence of preventing causes in an explanation are the cases where there is reason to suspect that there may
be some in effect. But the function of citing absent preventive causes is to allay a fear that we may have, or
that the person we're presenting the explanation to may have. The function of mentioning the absence of
preventing causes is not mentioning a contributing cause -- a cause which somehow causally helped bring
about the effect.72
Aristotle, then, is in his full right when he holds that the judgment "I should ! in circumstances C" suffices
to motivate the agent to act accordingly unless she is hindered by external forces or practical irrationality. If
Augustus, the weak-willed boy, fails to act as he thinks he should, it does not show that practical judgments
are impotent. There is no need to go looking for additional desires to unveil the "real" motive force behind
our decisions, or the real motive force which makes decisions effective. Decisions are effective in the
absence of intervening causes. That's how they "necessitate" action. I will therefore venture a wild
conjecture: Aristotle's talk of necessitation is really disguised talk of causes. His claim is that decisions are
sufficient causes of action.
Brad Inwood on Aristotelian Decision and Stoic Assent
I am not about to launch a full-scale defense of internalism in this paper. That would be a tremendously
challenging task. My objective up to this point has been to forestall an objection that is likely to arise in
discussions of Aristotle's theory of action, in casu, his theory of deliberate decision. Aristotle takes the
existence of irrational behaviour at face value -- as a phenomenon to be explained rather than eliminated
from our psychological theories. How, then, can he think that practical judgments motivate? My aim has
been to give credence to the thought that it is an entirely consistent position to be an internalist about
practical judgment, but nevertheless grant the possibility of practical irrationality, in particular, the
possibility of akrasia.

71

D. M. Armstrong, "The Open Door: Counterfactual versus Singularist Theories of Causation" in H.

Sankey (ed.) Causation and Laws of Nature (Dordrecht: Kluwer Academic Press, 1999), pp. 175-185; 177.
Thus, I side with those who think that ordinary ways of speaking sometimes distort the true causes at work
in nature. Claims like "His failure to turn off the faucet caused the tub to run over" and "Lack of antibiotics
caused the patient's death" don't single out the true cause of death or flooding. The true cause of death was
not the lack of antibiotics, but the bacterial infection -- an infection which would have been stopped if the
patient had been treated with antibiotics. Similarly, the true cause of the tub's running over is the water
which kept pouring from the open tap -- not the agent's failure to turn the water off. It is all too easy to infer
that causation by absence is a real phenomenon if one fails to distinguish between possible preventers and
real causes. Preventers work by blocking an effect which would otherwise have been manifest. Preventers
can, moreover, be blocked by what Armstrong elegantly calls "meta-preventers". In Brandom's example
and in my motivational example, the Faraday cage and the focusing technique both function as metapreventers: they block the effect of a preventer. My position on causation by absence importantly does not
entail that omissions are morally innocent. We can be held responsible and reasonably be praised or blamed
for what we do as well as for what we don't do; for what we cause as well as for what we fail to cause. The
excuse "I didn't do nothing" sometimes points to a moral shortcoming rather than an excusing
circumstance.
72
Again, someone may object that "I helped bring it about that the house burned down" by failing to call
the fire department when I noticed the fire. Granted, I can certainly be blamed for failing to prevent the fire
from developing. But my inaction did not contribute causally to the house's burning down. My moral
failure is omitting to prevent the fire from developing. The house burned down because of the presence of a
flame, oxygen, flammable material etc.

24

Aristotles analysis of akrasia has led some commentators to deny that Aristotle conceived of decisions as
both intrinsically motivational and as the result of judgments about what one ought to do. 73 Their inference,
in a nutshell, runs as follows: "If it is possible to act contrary to ones prohairesis, then either (a) it takes
something more than a decision to be motivated to act (the decision is not intrinsically motivational) or (b)
a decision cannot be the result of a practical judgment". One critic, Brad Inwood, appears to fall down on
alternative (b), and proposes a theory of Aristotelian decision according to which deciding is issuing a selfdirected command.74 Inwood is presumably influenced by R. M. Hare when he wants to attribute a view of
decision as a self-directed command to Aristotle.75 Hare maintains that whoever sincerely assents to a
judgment that he ought to ! in C thereby sincerely assents to a self-directed command "Let me ! in C!",
and he holds that it is tautologous that "we cannot sincerely assent to a command addressed to ourselves
and at the same time not perform it, if now is the occasion for performing it, and it is in our (physical and
psychological) power to do so".76 I believe Alfred Mele states the obvious when he complains that "not
only is the latter claim not tautologuous, it is false".77 My aim in the present section will be to explain first,
why Inwood's imperatives-interpretation of Aristotle's conception of decision does not solve any problems
that cannot be solved on the judgment-model I have presupposed, and second, to show that what Aristotle
has to say about decision gives us reason to infer that deciding just is judging that one ought to ! in
circumstances C, all things considered.
In his landmark study Ethics and Human Action in early Stoicism (1985) Inwood rejects the view that I
have been defending: that Aristotle conceives of decision as the product of the agent's assent to the
proposition "I should ! in C". Inwood is sympathetic to Aristotle's theory of decision, but only after he has
given it a thorough, and, I will argue, distorting overhaul.
Inwood is convinced that Aristotle's "imperatival" model remedies certain blatant shortcomings in a Stoic
intellectualist theory of decision. The Stoics, as we know, think of all desires as the product of a judgment
of reason: by assenting to the presentation of something as to-be-chosen, reason has an impulse (horm)
towards that course of action. Unlike Aristotle, who explains incontinent action as the result of non-rational
desires, the Stoics follow Socrates in thinking that we always act in accordance with our beliefs about what
is best. What goes by the name of "passion" is nothing but erroneous judgments about what is worth
pursuing or avoiding. As Plutarch puts the Stoics' view: "[P]assion is vicious and uncontrolled reason
73

I will discuss B. Inwood, Ethics and Human Action in Early Stoicism (Oxford: Clarendon Press,
1985), pp. 61-62.
74
For the command-theory of decision, confer D. J. Allan, "Aristotle's Account of the Origin of Moral
Principles", in R. Sorabji, M. Schofield, and J. Barnes (eds.), Articles on Aristotle: Ethics & Politics
(London: Duckworth, 1977), pp 72-78: "Reason, then, can issue a command to the appetitive faculty; the
words epitattei and keleuei are repeatedly used in this connexion. If this command is obeyed (but of course,
there may be contrary forces at work), an actual desire for the good will ensue, and an end will have been
established" (p.75). It's important to note that Allan, unlike Inwood, denies that command necessarily lead
to action. Allan's command-theory is not introduced as in an attempt to remedy the putative shortcomings
of a judgment model in explaining how reason motivates. That makes it philosophically more plausible that
Inwood's theory.
75
Inwood (p. 16) argues that "Aristotle uses imperatival language in describing how action issues from a
combination of desire and belief", and cites as evidence De Motu Animalium 701a32, where appetite says
"poteon moi", and EN 1147a29, where "the motive force of desire is expressed with similar
language":"pantos glukeos geuesthai dei". Furthermore, reason issues orders (keleuei, epitattei), Inwood
notes. But note that 1147a29 is a dei + infinitive construction; in other words, what Aristotle says (and this
isn't controversial) is: "everything sweet should be tasted". Appetite similarly says "I must drink". Inwood
cites EN 1143a4-11 as another Nicomachean passage that supports the "imperatival" model. This is the
passage where Aristotle discusses the distinction between sunesis and phronsis. I return to it below. All in
all, a more reasonable conclusion seems to be that Aristotle takes reason's orders to be of the form "I should
do x", or, in the case of appetite, "I must do x" and that he takes such sentences to express true or false
propositions (in the case of reason) or quasi-propositions (in the case of animal appetite).
76
R. M. Hare, Freedom and Reason (Oxford: Oxford University Press, 1963), p. 79.
77
A. R. Mele, Irrationality: An Essay on Akrasia, Self-Deception and Self-Control (Oxford: Oxford
University Press, 1987), p. 20-21.

25

which acquires vehemence and strength from bad and erroneous judgment".78 The weak agent does not
really do what people think he does, namely act contrary to a belief which he holds, and is aware of
holding, at the time of action. Instead, he changes his mind when the time to act approaches. This
vacillation between beliefs about what is best occurs because he lacks knowledge of good and bad. While
virtue is "consistent, firm and unchangeable reason",79 vice is the vacillating reason. The incontinent agent
suffers from weak assent.
As psychological monists, the Stoics maintain that "the natural instrument of appetite and regret, or anger
and fear, is the same part of soul (...)":
For appetite and anger and fear and all such things are corrupt opinions and judgments,
which do not arise about just one part of the soul but are the whole commanding-faculty's
inclinations, yieldings, assents and impulses, and, quite generally, activities which change
rapidly, just like children's fights, whose fury and intensity are volatile and transient
owing to their weakness (On Moral Virtue, 446F-447A).80
This debunking account of motivational conflict presupposes that an agent necessarily acts in accordance
with his evaluative judgments about what is best. What to the untrained eye appears to be a tug-of-war
between different motivational considerations -- passion's "!ing is sweet" and reason's "you should not !" - is really vacillation between different convictions about what should be done. This entails that for the
Stoics, as for Socrates, weak akrasia is the result of ignorance about what is better and worse -- it is not
possible to know that x is better than y, and still do y as a result of non-rational desires.
How does Inwood conceive of the antecedents of action on the Stoic and the Aristotelian models? While
our evidence suggests that the Stoics think that judgments motivate, Aristotle, according to Inwood,
explicitly denies that practical judgments ever suffice to motivate the agent. Inwood's reason for drawing
this contrast is that Aristotle accepts the possibility of akratic action. Inwood finds the Stoic idea that assent
to the proposition that something should be done motivates the agent so implausible that he is motivated by
charity to improve upon the Stoic theory. The only way to salvage the Stoic position, according to Inwood,
is to attribute to them the view that each presentation of something as to-be-chosen is accompanied by two,
and not just one, linguistic entities (lekta). One is a proposition, the other a self-directed imperative. We
assent is to the proposition, but act on the imperative.81 There are no sources attesting to such a theory, but
Inwood is nevertheless convinced that charity recommends this interpretation. Inwood's interpretive move,
then, is to make the Stoic theory more plausible by supplementing the views found in our sources with what
Inwood takes to be Aristotle's "imperatival" conception of motivation. We may attempt to clarify Inwood's
move as follows: Let "I" be an internalist conception of the relationship between decision and motivation.
Let "J" be the theory that decisions are the result of judgments about what one ought to do, and let "M" be
the theory that decisions are the result of self-directed imperatives. Inwood thinks that Aristotle holds I and
M, and that he rejects J. Inwood furthermore thinks that the Stoics hold I, and, on the face of it, J. He wants
78

Plutarch, On Moral Virtue, 441D, quoted from A.A. Long and D. Sedley's translation The Hellenistic
Philosophers, vol 1 (Cambridge: Cambridge University Press, 1987), p. 378, 61B.
79
Plutarch, On Moral Virtue, 441D.
80
That the inspiration for Stoic intellectualism and psychological monism is Socrates' argument in the
Protagoras is, I think, uncontroversial. That is partly why it is illuminating to use Stoic views to interpret
Socrates' own argument. Plutarch's "children's fight" simile is arguably a bit misleading. It intimates that
there are several "homunculi" parties to the motivational conflict ("two children"), when in fact there is
only one, a party, the hegemonikon, that cannot stick to his convictions. If we think of each quarrelling
child as a weak conviction, not as a homunculus, the simile makes more sense. Both contestants are
offspring of the same parent: reason, but they never occur simultaneously. I quote Plutarch from A. A.
Long and D. Sedley, The Hellenistic Philosophers, p. 412; 65G.
81
As Inwood acknowledges, this is not a view that our sources attest to. But he consoles himself that
"nothing stands in the way of the hypothesis that more than one lekton or even that lekta of different kinds
accompany a single presentation". This absence of competing evidence is used as a license to attribute the
theory to the Stoics "hypothetically" (Ethics and Human Action in Early Stoicism, p. 61).

26

to show that they in fact hold M rather than J. This is how he attempts to salvage the Stoic's internalism
from what he takes to be its prima facie implausibility -- the implausible claim that assenting to
propositions can cause action.
In her otherwise favorable review of Inwoods book, Gisela Striker admits that she is unconvinced both by
the purported problem and by Inwood's "solution" to it. 82 As Striker notes, why should issuing a selfdirected imperative have any more force than assenting to a proposition? As I hope to show, Striker is right
both to be nonplussed by the purported problem and by Inwood's remedy. I will maintain, pace Inwood,
first, that Aristotle does think that assent to the proposition that something is to be done is sufficient to
motivate the agent, ceteris paribus. In other words, Aristotle holds J and I, just like the Stoics do on the
most obvious reading of the extant sources.83 Second, I will argue that there is no reason to think that
directing imperatives at oneself is a more successful way to be motivated than assenting to propositions
about what one ought to do. An imperative does not have more of a kick-in-the-butt effect than assent to
the proposition that something is to be done.
Inwood's analysis starts out with a familiar "child-trapped-in-burning-house" scenario. The agent has a
presentation of the trapped child, accompanied by the associated proposition "it befits me to save the
child".84 The agent in question assents to this proposition. This, according to Inwood, is sufficient for the
agent to "know" that he should save the child, but it is not sufficient to make the agent move, according to
Inwood :
[A] Does he move? That is to say, is there an impulse? No. He does not move yet. [B]
For although the proposition is a true statement about what he should do, it does not
follow that he will act. Knowing what to do provides no guarantee by itself that one will
act. Aristotle has recognized this, both in his discussion of akrasia [C] and in another
passage from the Nicomachean Ethics which is of more interest here [NE VI,10, 1143a410]. 85
Everything Inwood says in the middle part (the explanatory [B]-part) of the quoted passage is true. But it
does not warrant the negative answer he gives in the first part (the [A]-part). For there could very well be
an impulse, provided that the agent is not weak-willed. Inwood's answer to the "Does he move"-question
rests on an externalist intuition about the relationship between judgment and motivation that he neither
questions nor defends. He is convinced that assent to the proposition "I should ! in circumstances C" has
merely cognitive consequences. It gives rise to "knowledge" (presumably in a "loose" sense of knowledge),
it certainly does not give rise to action. What is missing from the account, Inwood maintains, is a desire.86
Now, as I argued earlier on, Aristotle is not compelled by his acceptance of the reality of akrasia to deny
82

G. Striker, "Review of Inwood: Ethics and Human Action in Early Stoicism", Ancient Philosophy vol.
19, no. 1 (1989), 91-100.
83
True, the theories seem to differ in their emphasis of deliberation. It is questionable whether the Stoics
had a theory of deliberation (Inwood thinks we lack any evidence that they did): "There is no secure
references to deliberation in our sources for the Stoic theory" (p. 44).
84
Inwood, Ethics and Human Action in Early Stoicism, p. 61.
85
Inwood, op. cit., p. 62.
86
John Burnet is (mis)guided by a similar intuition: In cases of weakness, the moral syllogism "may even
be completed; but in the absence of orexis to which it can present itself, nothing happens. For dianoia by
itself moves nothing" (commentary to 1147a31-5, with reference to 1139a35) (The Ethics of Aristotle
(London: Methuen, 1900)). But Burnet, like so many scholars of a similar persuasion, cuts Aristotle off
mid-sentence. For Aristotle's full statement reads: "thought by itself (dianoia aut) moves nothing; what
moves us is goal-directed thought concerned with action (all h heneka tou kai praktik)" (1139a35-37). In
other words, Aristotle's view is that it is not thought qua thought, but thought qua practical thought that
moves the agent. So there is one kind of thought that moves: practical thought. If we allow that thought
may produce desire Burnet's puzzle disappears, for we won't have to assume that desire is already present
to explain why the agent acts. If we allow that desires are defeasible, as I have urged, we may even explain
why he does not act.

27

the practical efficacy of assent to a proposition that something is to-be-chosen (hs haireton). It is perfectly
possible to be an internalist about the motivating force of assent to such propositions, and accept that our
best judgment about what it is right for us to do can be overridden: compare the matchstick-case. The
argument from akrasia is a non-starter; and it certainly cannot be used to show that assent to the
proposition 'it befits me to save the child' is never efficacious, as Inwood seems to think.
Inwood presents a potentially much more damaging argument when he turns his attention to Aristotle's
discussion of the subordinate virtue of "sunesis", or comprehension, in EN VI, 1143a4-11. This is the
passage he refers to in the [C]-part of the quotation above. Sunesis is one of three "subordinate" intellectual
virtues which Aristotle compares to phronsis (prudence); the other two are euboulia (excellence in
deliberation) and gnom (consideration). An agent who possesses phronsis will always act in accordance
with his (excellent) decisions, Aristotle claims, for phronsis is praktik, it is practical (1141b21). That is
why prudent men never succumb to temptation, while men who merely have theoretical wisdom may. The
prudent man knows what the right thing to do is, and he always acts accordingly. Now, in the passage in
book VI, Aristotle has distinguished full-fledged prudence from mere comprehension. What, exactly, is
sunesis? Aristotle does not devote more than a page to the question. Here, in Irwin's translation, is an
expanded version of the passage which Inwood quotes in support of his externalist interpretation of
Aristotelian practical judgments:
Comprehension (sunesis) is neither about what always is and is unchanging nor about just
anything that comes to be. It is about what we might be puzzled about and might
deliberate about. That is why it is about the same things as prudence (phronsis), but not
the same as prudence. For prudence is prescriptive (epitaktik; Inwood, translates
"imperatival"), since its end is what action we must or must not do (ti gar dei prattein
m, to telos auts estin), whereas comprehension only judges (it is 'kritik monon')
(1143a4-10).
Inwood glosses the paragraph as a crystal-clear endorsement of externalism about the motive force of
practical judgment. If sunesis merely judges that something ought to be done or not done, and hence lacks
"prescriptive" or "imperatival" force, while prudence does not just judge, but also prescribes and indeed
ensures that the agent will act in accordance with the result of his deliberation, it seems that having reached
a judgment about what ought to be done will not suffice to motivate the agent. Unfortunately, Inwood does
not dwell on his argument at this point. Instead, he immediately wants to attribute a similar distinction to
the Stoics: "There is no reason why the Stoics should not have seen the same distinction and realized its
importance" 87.
The problem, as far as I can tell, is that Inwoods interpretation misconstrues the role of sunesis in
Aristotle. For a few lines further on, Aristotle states that "comprehension consist in the application of belief
to judge someone else's remarks (allou legontos) on a question that concerns prudence, and moreover it
must judge them finely, since judging well is the same as judging finely" (1143a13-16). This implies that
sunesis, comprehension, is not the ability to make fine judgments about what it "befits" oneself to do, to
paraphrase Inwood. It is the ability to judge finely about other's judgments about what they ought to do.
When I listen to my friend's explanation of her deliberation in a difficult situation, I possess sunesis if I can
determine whether she gave weight to the right reasons, if I can judge whether she deliberated well or badly
under the circumstances relative to her ends. Similarly, I possess sunesis if I am a fine judge of the
deliberations of the characters in a play or a person on trial. The person with sunesis is a good judge of
other people's judgments. R. B. Louden has brought this out well in his paper "What is Moral Authority?
Euboulia, sunesis, and gnom vs. phronsis".88 He writes: "The job of sunesis is to issue in a correct
judgment of someone else's choice and action, a skill that good judges and juries (among others) must
possess. Unlike phronsis, sunesis itself does not issue in choice (prohairesis) and action. Rather, the end
result of sunesis is simply a judgment concerning someone else's choice and action that has already
87

Inwood, Ethics and Human Action in Early Stoicism, p. 62.

R. B. Louden, "What is Moral Authority? Euboulia, sunesis, and gnom vs. phronsis", Ancient
Philosophy 17 (1997), 103-118.
88

28

occurred".89 The sunetos can answer the question: "Did x make the right decision?". He cannot, in virtue of
having comprehension, answer the question: "What should be done?". Having the virtue of sunesis
therefore does not amount to being a prudent person (phronimos), since a man may be a fine judge of
others, but a deficient manager of his own affairs.90 This explains why sunesis is not prescriptive, or
imperatival: "epitaktik".
We should therefore be skeptical about Inwoods appeal to Aristotle's remarks on the difference between
comprehension and prudence. As a consequence of his interpretation of the sunesis-passage, Inwood
attributes an externalist view of practical judgment to Aristotle. This paves the way for his attempt to
assimilate the Stoic theory of rational assent to a hormetic impression to a theory that Aristotle in fact does
not defend. Aristotle does not deny that making a judgment about what one ought to do is sufficient to
motivate action, for the sunetos qua sunetos does not make judgments about what he himself should do
under the circumstances. If he commands anyone, it isn't himself: it goes without saying that Fred's
judgment that Harry made the right decision will not motivate Harry, nor will it motivate Fred, to do what
he thinks Harry correctly decided to do.91 Aristotle's discussion of the prescriptive force of prudence and
the merely critical function of comprehension consequently does not commit him to externalism about
moral judgment.
But even if Aristotle had conceived of prudence as imperatival, it would not help solve the problem that
Inwood thinks needs solving. Introducing an imperative does nothing to ward away the akrasia-bogeyman.
If it is a real problem that assent to propositions lack intrinsic motivational force, then both Aristotle and
the Stoics have a problem. And even if we concede (as I think we should not) that assent to a proposition
about what one should do lacks motivational force, exactly the same argument could be made concerning
self-directed imperatives. Charity permits the historian to ascribe a view to a philosopher because it is the
most cogent philosophical view that can be reconstructed on the basis of the source-material. But
externalism about the relationship between practical judgment and motivation enjoys no such elevated
status. Inwood's argument for reducing impulse to a self-directed command seems to attribute a petitio
principii to the Stoics: They assume that self-directed commands will be obeyed: "[T]he description of
impulse as the reason of man commanding him to act suggests how the Stoics bridged the gap between
practical knowledge and action. They did so, I suggest, by construing impulse, which is the immediate
89

Louden, "What is Moral Authority? Euboulia, sunesis, and gnom vs. phronsis", p. 112.
The Magna Moralia 1197b15-18 implies that sunesis is an inseparable part of phronsis, thus the
virtues imply each other. Louden argues that "the author of the Magna Moralia" (who, according to
Louden, is not the author of either Ethics) distorts Aristotle's position. The Ethics keep the subordinate
virtues apart from phronsis.
91
T. H. Irwin does not accept that Aristotle's reference to the sunetos as a good judge of "what another
says" confines this minor virtue to the ability to make expert judgments about someone else's affairs.
Indeed, it would be strange if I could not apply the same principles to myself, given that we are both
human, and hence should aim at the same highest good; the activity of the soul in accordance with virtue.
At least at a high level of generality, the questions that concern prudence would be the same for all rational
beings. In the note ad loc to his translation of the Nicomachean Ethics 1143a8, Irwin proposes an
alternative interpretation of Aristotle's terse remarks. The man with comprehension is skilled at judging the
consequences of an action, e.g. "If you apologize to him, he will be less resentful", while the prudent man
knows what he ought to do, e.g. "Since you must remove his resentment, you must apologize to him". This,
I add, makes comprehension (sunesis) the moral equivalent to cleverness (deinots) in production. Just like
the clever man does not inquire whether his end is good, but is merely adept at finding effective means to
given ends, the man with comprehension does not in virtue of his comprehension understand how he should
act. He has a virtue that is "kritik monon". "If you want y, you had better do x". But of course, such
judgments (even if they are correct) won't suffice to motivate the agent. Irwin does not offer any specific
reasons for his interpretation. It therefore seems to me that Louden's interpretation is superior, insofar as it
is better supported by the text: Aristotle actually does note that the man with comprehension is a good
judge of "what another says". How we decide between Louden's and Irwin's proposals turns on whether
"being a good judge of what another says" is an accidental property of having comprehension, or whether it
is an essential property of having comprehension. Note that whether we prefer Louden's or Irwin's
interpretation, Inwood is wrong.
90

29

cause of action, in grammatical terms as a command to oneself which one obeys".92 Well, it would be very
convenient to be able to define oneself out of a psychological problem like this: "Problem: How does a
piece of practical knowledge motivate? Solution: By being necessarily correlated with a self-directed
imperative. What is a self-directed imperative? Answer: An imperative that will be obeyed by the person
who issues the imperative". Inwood acknowledges that the last clause in the passage I quoted "may be
startling". He then proposes three philosophical reasons to support the contention that "assent to a hormetic
proposition is extensionally equivalent to obedience to a self-given command"93, each of which is either
implausible or, it seems, outright question-begging. Like Hare, Inwood presupposes that any self-directed
imperative necessarily leads to action, while assent to propositions with the main verb in the gerundive
does not. But I dont see any reason to accept this.
The protagonist of Italo Svevos novel Zeno's Conscience, a chain-smoker of last cigarettes, quite rightly
observes that an excess of self-directed commands may perpetuate an addiction rather than end it. The act
of self-conquering has an addictive potential in and of itself: "Striking a beautiful attitude, one says: 'Never
again'. But what becomes of that attitude if the promise is then kept? It's possible to strike the attitude only
when you are obliged to renew the vow".94 The attempts to quit give birth to Zeno's "quitting-disease".
Zeno can issue imperatives against his nasty proclivity to smoke only if he continues to smoke. Therefore,
his last-cigarette disease is impervious to both resolutions and imperatives -- and even to electroshock
therapy, it transpires. Any self-directed imperative may be disobeyed. But that does not show that selfdirective imperatives are causally impotent as such, only that they can be defeated. By what? By other
resolutions, beliefs, desires, imperatives, take your pick. In this respect, self-directed imperatives are
entirely on a par with all-things-considered judgments. Inwoods "imperatives"-remedy is vulnerable to
Aristotles saying: "When water chokes us, what must we drink to wash it down?" (EN 1146a35-6).95 If I
can disobey the command, I can certainly disobey the command to obey the command -- there's no magic at
the meta-level. But the fact that water may choke us does not mean that it is impossible to wash anything
down with it.96 The possibility of akrasia is not a tie-breaker between the "imperatives"-model and the
"judgment"-model.
In this paper, I have attempted to defend a model for understanding Aristotle's theory of decision as a
theory which holds that decisions are the product of all-things-considered judgments, and which holds that
decisions motivate in and of themselves. This means that decisions are both intrinsically motivating states,
and states that can be evaluated as right or wrong relative to an objective standard: their contribution to
promoting the happy life. As Aristotle puts it, "Truth is the function of whatever thinks; but the function of
what thinks about action is truth agreeing with correct desire" (EN 1139a29-31). This interpretation seems
to pull Aristotle's theory of decision in a more intellectualist direction than what most interpreters have
recognized. That should not alarm us. Aristotle can accept the possibility of weak-willed action while
holding that practical judgments motivate.
92

Inwood, Ethics and Human Action in Early Stoicism, p. 62.

Inwood, op. cit., p. 63.
94
Italo Svevo, Zeno's Conscience , trans. W. Weaver (New York: Alfred A. Knopf, 2001), pp. 13-14. The
first chapter, "Smoke", offers a splendid case-study of incontinence, superior in psychological acuity to
most of the scenarios analytic philosophers tend to concoct to illustrate some point or other about
motivation.
95
Aristotle presents the saying as a proverb, so it did not originate with him.
96
"If some impulses were commands we obeyed and some were commands we did not obey, we would
have to explain why some were blocked from having effect. We should need a new power in the soul to
hinder some activations of the powers of presentation, reason, assent, and impulse. All the orthodox Stoics
are agreed that there is no such additional and obstructive power in the mind of man" (Inwood, op. cit., p.
63). Inwood assumes that a psychological monist is at a loss to explain how assent, or, as he conjectures, a
self-directed command, could fail to issue in action. This, however, shortchanges the intellectualist. The
Stoics do indeed offer an explanation: weak assent. The agent changes his mind when the time to act
arrives. The Stoics, then, deny that there is such a thing as "clear-eyed" or "strict" akrasia: acting contrary
to a conviction consciously or subconsciously entertained at the time of action. The "new power in the
soul" needed to explain akrasia is lack of knowledge, which leads to vacillation, not a competing mental
faculty.
93

30

31