Turner - 2011 - Ontological Nihilism
Turner - 2011 - Ontological Nihilism
Turner - 2011 - Ontological Nihilism
Ontological Nihilism
Jason Turner
DOI:10.1093/acprof:oso/9780199603039.003.0001
This essay explores Ontological Nihilism, and for just this reason. After
discussing what the thesis would amount to if it were to have any plausibility
whatsoever (§1), I present (§2) and clarify (§3) a crucial challenge for it. I show
what is wrong with two less plausible attempts to meet this challenge (§4), and
then argue that the proposal thought by many to be much more promising (§5)
succumbs to the same problems as the less promising attempts (§§6 and 7). It
Page 1 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
turns out that, in order to make Ontological Nihilism viable without smuggling
an ontology in through the back door, we need a bloated ideology, indefinitely
many brute, necessary connections, and a deep‐seated holism about the
structure of reality. Once we understand why these are costs of the view, we see
just what we gain by thinking of the world as built up out of things.
1. ONTOLOGICAL NIHILISM
Ontology, Quine tells us, asks what there is; and while this onto‐ logical question
can be answered in a word—‘everything’—there is still room for disagreement
about cases (1948, 1). When we encounter this case‐by‐case disagreement, we
occasionally come across views that can best be described as kinds of
Ontological (p.4) Nihilism. Compositional nihilists, for instance, hold that there
are no composite objects: nothing has parts. So‐called nominalists (of the good,
old‐fashioned ‘nothing is abstract’ type) could just as well be called abstractional
nihilists: they claim that there are no abstract objects. Perforational nihilists are
those who, like the Lewis' (1970) Argle, say that there are no holes. And so on.
When an Ontological Nihilist says that there isn't anything at all, we might
naturally think he endorses the following claims:
The Nihilist need not be quite as crazy as all that, though—he can hold that
there isn't anything at all without endorsing either of (1) or (2). He thinks the
blob‐less blob of reality has a rich structure which our ordinary beliefs (usually)
track. But he insists that this structure does not involve any things, any entities—
any ontology.
Page 2 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Let me explain. At the simplest level, we describe the world by combining two
different types of expressions.2 We take noun (p.5) phrases—paradigmatically,
proper names such as ‘Bertrand’ and ‘Gottlob’, and quantifier phrases such as
‘every philosopher’ or ‘some logicist’—and combine them with predicates—such
as ‘thought about language’ or ‘didn't notice the inconsistency in Basic Law V’.
The noun phrases latch on to some things, and predicate phrases describe and
differentiate these things.
Such descriptions of the world implicitly suppose that it has a certain sort of
structure—an ontological structure. Ontological structure is the sort of structure
we could adequately represent with a pegboard and rubber bands. The pegs
represent things, and the rubber bands represent ways these things are and are
interrelated.
To say ‘Bertrand thought about language’, for instance, is to hang the thought
about language rubber band on the peg labeled ‘Bertrand’. And to say ‘Some
logicist admired every philosopher who didn't notice the inconsistency in Basic
Law V’ is to say that, somewhere on the pegboard, there is a peg which (a) has a
logicist rubber band hanging on it, and (b) has an admires rubber band
stretching from it to each of the pegs with the didn't notice the inconsistency in
Basic Law V band on.3
A similar challenge faces e.g. the more conservative perforational nihilist, who
claims that there are no holes. At first blush, the perfo‐ rational nihilist's claim
seems incredible, implying that we are radically deceived about the nature of the
Page 3 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
world. Suppose, for instance, that you just crossed a bridge like that in figure 1.
If someone asks you why you crossed on the left, you will likely say
and point out that you were not keen on dropping through to the river below.
Perforational nihilists insist that there are no holes; since (3) seems to entail that
there are, they should reject (3). So it seems they must say that you were
radically mistaken about the nature of the bridge—and that crossing on the
right‐hand side would have been fine.
Of course, perforational nihilists want to say neither thing. They grant that some
bridges are indeed shaped like the one in the diagram, and that you should not
walk on the right‐hand‐sides of such bridges. And they will say that there is
something right about your utterance of (3): even though there are no holes,
there is some important fact, relevant to bridge‐crossing, that you were getting
at with (3) and which explains your reluctance to cross on the right. They object
only to the idea that this important fact involves a special kind of entity called a
‘hole’. On their view, crossing on the right‐hand side of the bridge is not bad in
virtue of its relation
There are two reasons perforational nihilists should go further. First: doubters
may worry that if there were no entities deserving to be called ‘holes’, the world
just wouldn't have enough structure to guarantee that (3) is a good thing to say
in the envisaged circumstances. Perforational nihilists can assuage these doubts
Page 4 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Second and more important: if we stop with a negative thesis, we do only half
the job of metaphysical inquiry. Metaphysics asks what the fundamental
structure of the world is and how this structure accounts for the richness and
variety of experience. Simply to tell us what the world isn't like is not yet to tell
us what the world is like. A complete metaphysical picture will tell us what the
world is like, and if it is indeed not a blatant error to appeal to (3) when
explaining how we cross bridges like the one in the diagram, a complete
metaphysical picture will tell us why.
The perforational nihilist answers our question in the simplest way by telling us
what useful but hole‐free fact (3) is getting at. For instance, the perforational
nihilist might think that, although there are no holes, certain physical objects
have a special shape property, that of being perforated. Furthermore, he claims,
And he will say that (4) is the metaphysically perspicuous truth we have been
getting at with (3) all along.
If (3) is the only useful hole‐involving sentence we ever say, this will be enough.
But it is not; we communicate many other important facts by talking about holes.
So we need more than just this particular, one‐off explanation—we need an
account of how hole‐ talk communicates important facts generally.
(H) Thereisaholein____,
the perforational nihilist will tell us the important fact we are communicating is
(P) ____isperforated.6
Page 5 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(p.9) The perforational nihilist thus tells us what he thinks this hole‐free world
is like—he thinks it is filled with perforated things—and how apparent talk of
holes really gets at these perforation facts.7
But let's not bind the nihilist to the commitments of any particular philosophy of
language. We demand merely that the perforational nihilist tell us, for any claim
involving holes that he takes to be getting at some important fact, exactly what
important, hole‐free fact he thinks it is getting at—regardless of whether it is
finitely stateable, easily recognizable by anyone capable of talking about holes,
etc.8
‘There is a hole in the bridge’ says something very different about the bridge
than ‘There is a hole in the argument’ does about the argument, so we shouldn't
demand that the perforational nihilist paraphrase these in the same way. But,
insofar as the nihilist thinks that various claims about holes are getting at similar
facts, they ought to paraphrase them in similar ways. And insofar as we think
that various claims about holes are getting at similar facts, we ought to look
askance at strategies which paraphrase them differently. Nobody thought the
facts gotten at by ‘There is a hole in the bridge’ (p.10) and ‘There is a hole in
the argument’ were similar in the first place. But we do think a bridge's having a
hole is somehow similar to a door's having a hole; if a perforational nihilist
paraphrases ‘There is a hole in the bridge’ and ‘There is a hole in the door’ in
radically different ways, he thereby denies that these claims are getting at
similar facts after all. And, the more convinced we are of these facts’ similarity,
the more the nihilist must do to convince us of our error.9
Page 6 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
and so on also get at important facts.10 But since the Ontologi‐ cal Nihilist denies
that there is anything at all—and hence denies that there are buildings, people,
shirts, or marshmallows—he must think the sentences in (5) are somehow
misleading. He needs to tell us what this building‐, people‐, shirt‐, and
marshmallow‐free world is like, and why its being this way makes the sentences
in (5) worth saying. So he also needs a paraphrase scheme: a way to trade in
ontological, pegboard‐and‐rubber‐band presupposing claims for ‘ontologically
innocent’ facts—facts which do not entail that there is anything.
(p.11) We can simplify the Nihilist's task by pretending the target language to
be paraphrased is that of first‐order logic without names (but with identity). This
language is generally thought sufficient for talking about ontological structure:
its existential quantifier, ‘∃’, means there is, and it can form all sorts of sentences
that talk about what there is, what there isn't, and how things are interrelated.
We also ask the Nihilist to paraphrase only part of our ontology‐ involving talk.
In particular, we make him paraphrase only claims from well‐established
scientific theory (or, at least, first‐order consequences of well‐established
scientific theory).
We thus make the Nihilist's task both easier and harder. Easier, because we take
from his shoulders the burden of deciding which sentences deserve paraphrase.
A Nihilist ought not paraphrase everything we say: some of what we say just isn't
getting at any important fact. (Nihilists need not paraphrase e.g. ‘Phlostigon is
emitted during combustion’.) But Nihilism is plausible only if it can recover at
least the claims of our (incredibly fruitful) best science— surely if any claims
ever get at important facts, these do.
Harder, because focusing on these sentences gives us the right to demand that
the Nihilist paraphrase systematically. Even if
are metaphysically misleading, they clearly get at very similar facts. But if
sentences get at similar facts, then they should be paraphrased in similar ways.
With this in mind, we demand the following of our Ontological Nihilist: give us a
systematic recipe for taking any sentence of a first‐order language (with
predicates assumed to be predicates of our best science) and cooking up the
ontologically innocent claim it was supposed to be getting at all along.
Page 7 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
The core idea is that we somehow manage to linguistically convey that the world
has a certain ontological structure—that there are some things of a certain kind
K, pegs with the K‐rubber band hanging from them. When someone performs the
right sort of linguistic activity, we say that the individual is ontologically
committed to K s.
The ‘right sort’ of linguistic activity is sincere assertion of the right sentences,
properly understood.12 But the sentence has to be the right one—I cannot
commit myself to unicorns just with any old sentence. I have to use a sentence
that says that there are unicorns. So I am ontologically committed to unicorns if
and only if I understand and sincerely assert a sentence that says that there are
unicorns; and in general I am ontologically committed to Ks if and only if I
understand and sincerely assert a sentence that says that there are Ks.13
What does it mean to say that ‘∃’ is a ‘variable binder’? In first‐ order languages,
we can take a sentence open in a variable ‘x’ and prefix it with ‘∃x’ or ‘∀x’ to get
a new sentence. If the original sentence was open in other variables, the new
sentence is open in those variables, too. Otherwise, the sentence is closed and
Page 8 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
can be evaluated for truth. Variable binding is this turning of open expressions
into closed (or less open) ones.
Quantifiers ‘quantify proper’ by saying how many pegs are arranged the way its
postfixed formula says. ‘∃x… ’, for instance, says that least one peg is … ; ‘∀x… ’
says that every peg is . … If we have more sophisticated quantifiers than these,
we can also say, for instance, that infinitely many pegs are …, or that most pegs
that are … are also _ _ _.
Here's the idea. Introductory logic texts often tell us that we can read ‘∃x(… x
… )’ as a sort of quasi‐English (p.14)
But we could do the same work with separate expressions: one which means
‘there is something’, one which means ‘everything’, and a third which means ‘is
an x such that… x…’. The third expression would be a variable‐binder; the first
two, quantifiers proper.
That the variable‐binder is innocent: suppose that we had a language with ‘A’
and only one sentence‐making operator, ‘B’, which means ‘It is possible for there
to be someone who believes that something…’. No ontologically committal
sentence could be formed in that language. We could only use it to talk about
what possible believers could or couldn't believe. But we can talk about that all
Page 9 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
day without ever saying anything about what there is. The variable binder
doesn't suffice for ontological guilt.
That the quantifiers proper are to blame: ‘∃p’ means there is something. We
commit ourselves ontologically when we say that there is something which is
some way or another, and ‘∃p’ is the expression we use to say this.
(p.15)
A pair of pegs will satisfy this
predicate exactly when one of
them has the ’F’ rubber band
hanging from it, the other has Figure 2. A rubber band for (8)
the ‘G’ rubber band hanging
from it, and the ‘R’ rubber band
is stretched between them. We might think of (8) as picking out a single rubber‐
band structure—one involving three rubber‐bands glued together, as in figure 2.
But ‘λ’ does not fasten that rubber band to any pegs. If you want to say that an F
thing Rs a G thing—if you want to stretch this complex rubber band between a
pair of pegs—you've got to find a way to plunk down some pegs to stretch it
between.
Page 10 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
where the only difference between (9) and (10) lies in the meanings of
‘there are’ and ‘there schmare’ (and cognates, like ‘every’ and
‘schmevery’), respectively.
When we press him on the meaning of ‘there schmare’, this Nihilist refuses to
say anything informative. He merely insists over and over again that it is
ontologically innocent and can be uniformly replaced for ‘there are’ to turn
falsehoods into truths.
Bewildered, you ask further: ’Is eulb a color?’ He says, ‘Yes’. He tells you that
eulb is a cool color, the color of the sky, lying on the spectrum between red and
green. You ask what color complements eulb, and he replies, ‘yellow’. He even
insists, ‘Contrary to what most people think, purple is not a combination of red
and blue. It's a combination of red and eulb.’ He denies any sentence that you
assert using the word ‘blue’, but happily assents to the sentence that results
from it by a systematic replacement of ‘blue’ for ‘eulb’.
You will soon think that when Eustace says ‘eulb’, he means blue—the color you
have known and loved all along, the color of the sky and of bluebirds, the color
you have always called ‘blue’. (p.17) And so, even though he won't use the
word ‘blue’ for it, you will suspect that, insofar as the two of you have any real
disagreement at all, it is only disagreement about which word to use for certain
shades. You certainly aren't disagreeing about anything's color.
Suppose we think of you and Eustace as speaking subtly different languages: the
‘blue’‐language and the ‘eulb’‐language, respectively. Then your understandable
Page 11 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Eustace and I seem to mean the same thing by all of our terms other than
‘blue’ and ‘eulb’, and he uses ‘elub’ in exactly the same way that I use
‘blue’. But, since our words get to mean what they mean thanks to the way
we use them, ‘blue’ in my mouth and ‘eulb’ in his should have the same
meaning. Since ‘blue’ in my mouth means blue, ‘eulb’ in his mouth must
mean that, too.
There is a general lesson here. Suppose L 1 and L 2 are languages that are
exactly alike except that, where L 1 has an expression α, L 2 has a different
expression, β. If ϕ is a sentence in L 1 that uses a, we write it as ϕ α, and ϕ β will
be the expression of L 2 that is just like ϕ α except that β is replaced everywhere
for α. The line of thought just sketched relies on the principle:
(∗) If every term (other than α and β) is interpreted the same way in L 1 as
it is in L 2, and if the speakers of L 1 utter ϕ α in all and only the
circumstances in which speakers of L 2 utter ϕ β, then α and β have the
same interpretation also.
In the above case, of course, the ‘blue’‐language was L 1, the ‘eulb’‐ language L
2, ‘blue’ was α and ‘eulb’ was β. Since you and your interlocutor meant the same
thing by your other expressions, (∗) licenses the conclusion that ‘blue’ and ‘eulb’
mean the same thing in your respective mouths.
If we don't understand (∗) in this way, it will prove too much. Imagine two
communities that differ linguistically only in that one uses ‘green’ and one uses
‘grue’. The green speakers are just like us, except they have never read
Goodman 1979/1983 and never entertained the predicates ‘grue’ and ‘bleen’.
The ‘grue’‐speakers are just like the ‘green’‐ones, except (a), they have never
entertained a predicate that works the way ‘green’ does, and (b) although this
community calls things ‘grue’ exactly when the ‘green’‐speaking community calls
them ‘green’, they have different linguistic intentions. The ‘grue’‐speakers fully
intend, when they encounter green‐ looking things for the first time after the set
future date, to not call them ‘grue’ anymore. And they fully intend to call blue‐
looking things encountered for the first time after this date ‘grue’.16
Page 12 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Could the Quiet Nihilist defuse the appeal to (∗) by his mere insistence that
‘schm∃’ doesn't mean the same thing as ‘∃’ does? I doubt it. Suppose Eustace
insisted vehemently that ‘eulb’ did not mean the same as ‘blue’ in our mouths. It
is then as though he stipulates the following:
(S1) ‘Eulb’ applies to exactly those things ordinary people would call
‘blue’ under ordinary conditions.
Page 13 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(S2) ‘Eulb’ is not interpreted the same way as (is not intensionally
equivalent to) ‘blue’.
It is not at all clear that these constraints are jointly satisfiable. If there is a
property B that applies to exactly those things that ordinary people would call
‘blue’ under ordinary circumstances, an ideal interpreter will be pressured to
interpret ‘blue’ as meaning B. When Eustace comes by and makes stipulation
(S1), the ideal interpreter will have no choice but to interpret ‘eulb’ as B. But
then she will have no way to satisfy (S2) without re‐interpreting ‘blue’ as
something other than B. No ideal interpreter would give Eustace that sort of
control over the interpretation of everyone else's ‘blue’—any reasonable
principle of charity will have her make Eustace, rather than the rest of us, speak
falsely. So, insofar as she ensures (S1)'s satisfaction, she will have good reason
to leave (S2) unsatisfied.
What goes for Eustace goes for the Quiet Nihilist: having insisted that sentences
such as (10) are true in exactly those situations where we assert (9), he cannot
also insist that ‘schm∃’ has a different meaning than ‘∃’. Insofar as we grant his
first insistence, we have good reason to think that ‘schm∃’ means (is
intensionally equivalent to) ‘there is’ after all.
The crucial difference between these charity arguments and my (∗) argument
above is simply that the charity arguments take place at the level of sentences
whereas mine takes place at the level of words. There are thus ways to resist the
charity arguments that do not likewise affect the (∗) argument. To take one well‐
discussed example, there might, as Lewis (1983, 45–55, and 1984) argued, be a
so‐called naturalness constraint on interpretation: try, inter alia, to give each
word as natural and un‐gerrymandered a meaning as possible. This constraint
will of course be balanced against other interpretative constraints like charity.
But it is crucially a constraint about the interpretation of words rather than the
interpretation of sentences. It may very well be that every interpretation of
Page 14 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
parties A and B that makes all of B's sentences ϕ synonymous with A's t(ϕ) does
so by assigning overly gerrymandered meanings to the individual words of A's or
B's language. And so the naturalness constraint may, as a result, require some of
B's ϕs to have different meanings from A's t(ϕ)s after all.19
But the (∗) argument relies on an interpretative principle about the meanings of
words: when A and B use the words α and β in the same way, against a
background of other, shared words with matching interpretations, then α and β
must be interpreted the same way, too. No naturalness constraint or other word‐
level constraint can conflict with (∗): the shared words have the same
interpretations and hence the same semantic properties in both (p.22)
languages, and if we give α and β, the same interpretation, they will have the
same semantic properties in both languages, too. If β, gets a highly natural
interpretation, for instance, then nothing about naturalness can keep α from
getting the same interpretation. And if β has a gerrymandered interpretation, the
fact that this interpretation snuck in against the semantic background of B's
language means that there can be no naturalness‐inspired bar to giving α the
same interpretation against the same semantic background.20
The considerations underwriting the (∗) argument are much more fine‐grained
than those supposed to underwrite standard charity arguments, and the two
kinds of argument ought not be confused. Even those who suspect interpretative
charity arguments in general ought to find (∗) plausible and thus reject Quiet
Nihilism.
will be paraphrased into another atomic sentence, say ‘Q’. But these
atomic sentences don't invoke any pegboard structure. They just say that
thus‐and‐so is the case, where thus‐ and‐so is some ontology‐free state of
reality.
When we press the Propositional Nihilist to tell us more about what these
sentences mean, he also refuses to say anything helpful.
Page 15 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(p.23) I doubt that Propositional Nihilism is untenable in the way that Quiet
Nihilism is. But it has several defects that we should not pass over.
Since these sentences are not truth‐functional compounds, they must each be
paraphrased as some atomic proposition. And each of these is logically distinct,
so if the Nihilist translates two of these as the same sentence, he will collapse
distinctions that he shouldn't. So, insofar as he wants his paraphrases to
preserve our ability to make these sorts of distinctions, he will need to
paraphrase each of these by a different atomic proposition:
Suppose the Propositional Nihilist paraphrases these as ‘A’ and ‘B’, respectively.
Since (13) entails (14), ‘A’ ought to entail ‘B’.
Page 16 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
is not. That ‘A’ entails ‘B’ but not ‘C’ is, according to the Nihilist, a brute fact: it
admits of no more basic explanation. And, although everybody has some brute
facts somewhere or another, the Propo‐ sitional Nihilist has more than his share:
presumably there will be indefinitely many valid (and indefinitely many invalid)
inferences between atomic propositions, and the validity (or invalidity) of each
one will be a further brute fact.21
4.2.3. Holism
Finally, the view is holistic: it cannot make sense of reality's global structure
being somehow ‘built up’ out of its various local structures.
We ontologically minded folk think, more or less, that there are only a limited
number of ways things can be and can be related, and the way reality is in toto is
fixed by how each thing is and relates to its fellows. For instance, when I say (p.
25)
and
Page 17 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
A Propositional Nihilist might instead complain that he has not really incurred
all these costs. I can see no way for him to keep holism and ideological bloat off
the bill. But he can at least try to quit the charges of inferential unsystematicity.
Let's look at two ways he might do this.
Page 18 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(p.27) But here he waves a red herring. §4.2.2's complaint wasn't that the
inferences between the atomic sentences were syntactically indiscernible—not
worn on their syntactic faces, as it were. The complaint was that the inferences
have to be semantically brute: there is no explanation whatsoever, syntactic or
otherwise, for their validity.
Let's look at the appeal to natural language. Hawthorne and Cortens (1995, 151)
point out that while the inference
is clearly not. And these two inferences are syntactically indistinguishable; the
validity of (19) and invalidity of (20) are not worn on their syntactic faces.22
But (as Hawthorne and Cortens note on the same page) this does not make the
inferences brute: there is a simple semantic explanation for the difference in
(19)'s and (20)'s validity. ‘Happily’ is an adverb which, when attached to a verb
that picks out an action V, creates another verb which is still a kind of V‐ing. But
‘allegedly’ is an adverb which, when attached to a verb that picks out an action
V, does not create a new verb that picks out a kind of V‐ing. (19)'s validity is
neither syntactically discernible nor brute.
A similar point holds for higher‐order languages. Even though they have no
complete axiomatization, they do make room for semantic explanations of
validity. The explanations come from the model theory for those languages,
which makes them semantic, not syntactic, explanations.
Propositional Nihilism is not like either of these cases. It lacks not just a
syntactic account of the inferences’ validity, but it lacks a semantic one, too. Its
atomic propositions, recall, do not have semantic values that encode any more
detailed structure. They are propositional blobs—they can be true or false, but
that's all we can (p.28) say about them. After he has told us that there is a true
atomic claim ‘A’, and that it is what we were getting at all along with (13), the
Propositional Nihilist has nothing left to say. In particular, he has no story about
what ‘A’ means that would let him explain why it entails ‘B’ but not ‘C.
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Call this the tag‐language. It has a fully specified syntax. It remains to give it a
semantics. The semantics we give it is quite simple: every content tag is
interpreted so as to encode one of the Propositional Nihilist's atomic facts. And
inference tags, and all of their parts, are semantically empty23
Propositional Nihilists can easily create a syntactic inference system that will
mirror the inferential structure of the target language: paraphrase any sentence
ϕ of the target language as ⌜P ϕ⌝ for some content tag P. Then say that
. In this case, since we always
Page 20 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
If we think that ϕ entails ψ, we think that whatever important fact we get at with
ϕ cannot be true if the important fact we get at with ψ is false. Anyone who
paraphrases ϕ so that it does not entail the paraphrase of ψ effectively says that
we are wrong about this (p.30) relationship: the important fact we get at with ϕ
could be true even if the important fact we get at with ψ is false.
Let's spell this out more carefully. Ontological nihilists of any stripe want to
‘save the appearances’—to explain why it is sometimes useful to talk as though
there are certain kinds of things even though there aren't. But part of saving the
appearances is ‘saving the practices’—explaining why certain natural transitions
involving these kinds of things are so useful.
Consider again the bridge in figure 1. We noted that someone who doesn't
believe in holes needs to tell us why you can point to
even in the absence of holes, to explain why you crossed on the left instead of
the right.
In one sense, to explain an action is to explain why someone did it. Explanations
of this sort usually cite some beliefs and desires: roughly, I can explain my A‐ing
by pointing out that I desired that C be the case and I believed that if I A‐ed, C.25
And I can explain why I A‐ed instead of B‐ed by pointing out that I desired that C,
believed that if I A‐ed, C, and believed that if I B‐ed, not‐C. Call this sort of
explanation a descriptive explanation—it describes why somebody acted in a
certain way.
Page 21 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
and your desire not to plummet into the river to give such an explanation.
Your (21)‐ and (22)‐like beliefs were true, and your behavior well‐ motivated,
because your bridge was perforated; Hal's counterpart beliefs were false, and
behavior ill‐motivated, because his bridge was not.
But this move works only if (4) (plus some background assumptions) entails (22).
If you could walk over perforated parts of bridges unharmed, the truth of (4)
wouldn't be relevant to how you crossed. (p.32) So the paraphrase scheme
‘explains the practices’ only if it preserves the validity of certain inferences.
Page 22 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
It is raining,
It is snowing,
It is cold,
and so on. Notice that, despite the ‘it’ in each sentence, none of these say that
any thing is raining, snowing, or cold. These sentences simply ‘place’ certain
meteorological features—simply say that rain, snow, or coldness are manifest
without saying that any particular object is manifesting them. Unlike most
English sentences, these are not talking about the arrangements of rubber
bands on pegs. (p.33) If they are doing anything even in the neighborhood of
that, they are throwing rubber bands on peg‐free areas of the board.26
we could say
(24) It is catting.
Just as ‘it is raining’ says that rain is going on without saying that there is any
thing which is raining, (24) says that catting is going on without saying that any
particular thing is a cat.
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(25) ∃xA(x),
as
(26) It is A‐ing.
We can now paraphrase very simple sentences. How do we deal with more
complex ones? We need to tread carefully around them. Consider, for instance:
(28) says that some things are positively charged and some things are negatively
charged; this is the sort of sentence the Nihilist should paraphrase into
something he takes to be true. But our best science rules out (27) (or so I am
told), and so the Nihilist ought to paraphrase it as something he takes to be
false.
But the temptation should be resisted, for these sentences say the same thing.
The semantics of feature‐placing sentences treat the ‘it’ as empty and the
predicate ⌜A‐ing⌝ as expressing a proposition. The ‘it’ is needed simply to fill a
syntactic requirement, and isn't doing any semantic work. (Some languages do
not have this syntactic requirement, and their corresponding feature‐placing
sentences are simply verbs. The Spanish counterpart of ‘it is raining’, for
instance, is the conjugated verb ‘llueve’.) But if ‘is positive‐charging’ and ‘is
negative‐charging’ express propositions all by themselves, then any ‘and’
between them simply conjoins those propositions, regardless of where the ‘it’
shows up.28 (29) and (30) are equivalent, so they can't respectively paraphrase
both (27) and (28).
Page 24 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Let's make this proposal more precise. Suppose we begin with a stock of simple
predicates A, B,. …Then we help ourselves to some predicate functors,
expressions that combine with predicates to make new predicates. For instance,
we help ourselves to a predicate conjunction functor ‘&’, which combines with
any two predicates to create a third. If P and Q are any predicates, then ⌜P & Q⌝
is their conjunction. Likewise, we help ourselves to a predicate negation functor
‘~’: if P is a predicate, ⌜~ P⌝ is its negation.
Actually, at this point we might as well drop the syntactic pretense that the
feature‐placing language's expressions ⌜is A‐ing⌝ are predicates. From the
perspective of the semantics, these things are sentence‐like—they are truth‐
evaluable all on their own, and only demand a (semantically empty) ‘it’ to satisfy
Page 25 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
a quaint demand of English syntax. We can make the semantics and the syntax
march in step by letting the predicate‐functor combine with predicates to make
sentences. Suppose we write this functor ‘Δ’: then for any predicate A, whether
simple or complex, ⌜Δ(A)⌝ is the Nihilist's sentence meaning ⌜It is A‐ing⌝.
If our feature‐placing language has the simple predicates of the target language
and the three predicate functors ‘&’, ‘~’, and ‘Δ’, we can paraphrase the target
language into it simply and smoothly.
The paraphrase strategy relies on two facts. First, every sentence in a first‐order
language with only one‐placed predicates is equivalent to a truth‐functional
compound of sentences of the form
(33) ⌜∃x(…x…)⌝,
(34) ⌜Δ(P)⌝,
where P is the functor reduction of ‘…x …’. This gives us feature‐ placing
replacements for each sentence of the target language without any of the costs
of Quiet or Propositional Nihilism. The feature‐placing option, it seems, gives us
Nihilism on the cheap.
etc.
Page 26 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
The paraphrase scheme already in place is nice; can we extend it to deal with
relational predicates? We will need to say that, just as we can ‘place’ the feature
associated with a one‐placed predicate A by prefixing it with a ‘Δ’, we can also
somehow ‘place’ the relational feature associated with a many‐placed predicate
R by prefixing it with a ‘Δ’, too. Just as ⌜Δ(A)⌝ says that it is A‐ing, ⌜Δ(R)⌝ will, in
some sense or another, say that it is R‐ing.
But in just what sense? What happens to a many‐placed predicate when it gets
prefixed with ‘Δ’? The Nihilist has only two viable options: either say that the
prefixing creates a new predicate, or say instead that it creates a sentence. On
the first option, if R is an n‐placed predicate, ⌜Δ(R)⌝ is an n − 1‐placed predicate.
Then ‘A(repels)’, for instance, would be a one‐placed predicate—the Nihilist's
paraphrase of our complex predicate ‘repels something’. On the other option,
attaching ‘Δ’ to a predicate always creates a (p.38) sentence, no matter how
many places the predicate had to begin with. On this proposal, ‘Δ (repels)’ is the
Nihilist's paraphrase of our sentence ‘Something repels something’.
6. PREDICATE FUNCTORESE
6.1. The combinatorial functors
According to the first proposal, attaching ‘Δ’ to, say, the predicate ‘orbits’ makes
a new predicate, ‘Δ(orbits)’. Since ‘orbits’ has two places, this new complex
predicate has just one. And, although it is hard to say in any straightforward way
what this predicate means, the idea is straightforward: ‘Δ(orbits)’ is the Nihilist's
feature‐ placing paraphrase of our one‐placed predicate ‘orbits something’.
Then, to make a sentence out of this predicate, we can attach another ‘Δ’ to it:
‘something orbits something’ is paraphrased as ‘ΔΔ(orbits)’.
This proposal suggests a natural paraphrase strategy. Every first‐ order sentence
is equivalent to one in prenex normal form: one which begins with a block of
quantifiers followed by a quantifier‐free open sentence. But any block of
quantifiers can be converted to a block of existential quantifiers sprinkled with
negations; say that a sentence that begins with existential quantifiers and
negations which are then followed by a quantifier‐free open sentence is in
prenex existential form. Now, if we can find some n‐placed predicate equivalent
to any quantifier‐free sentence open in n variables, we have a straightforward
way to paraphrase any first‐order sentence ϕ : first, convert ϕ to prenex
existential form
Page 27 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Δ…Δ…Δ(P)
This will work so long as we can always come up with a complex predicate for
each quantifier‐free open sentence. The Nihilist already has many resources
needed for this job. For instance, he can turn any sentence open in only one
variable into a complex predicate using just the functors‘~’ and ‘&’. And, via a
natural extension of‘~’ and ‘&’ to multi‐placed predicates, he can trade in some
other sentences, too. We extend ‘&’ so that, if A is an n‐placed predicate and B an
m‐placed predicate, ⌜A& B⌝ is an i‐placed predicate, where i is the greater of n
and m, so that ⌜(A& B)x 1,…, x i⌝ is equivalent to ⌜Ax 1,…, x n & Bx 1,…, x m⌝.31
Then, for instance, he can turn the open sentence
and paraphrase
as
(36) ΔΔ(orbits)
~ Δ ~ Δ(orbits)
Δ ~ Δ(~orbits)
~ ΔΔ(‐orbits).
(p.40) But each of these are already tagged as respective paraphrases for:
Page 28 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Since (35) is not equivalent to any of the sentences in (37), it should not be
paraphrased as anything in (36).
(perhaps with negations sprinkled through the block of quantifiers) and trying to
find a complex predicate P so that
But it's crucial that the variables in this equivalent one‐predicate open sentence
occur in the same order that they do in the original. If x 1 is the first variable
bound in the block of quantifiers, it needs to be the first of P's arguments, if x 2 is
bound second, it needs to be the second of P's arguments, and so on.
In (35), ‘x’ is bound first and ‘y’ is bound second. So we need to find a predicate
P where ⌜P(x,y)⌝ is equivalent to the open sentence ‘y orbits x’. No truth‐
functional compound of ‘orbits’ will do the trick. We need something else.
If we had the predicate ‘is orbited by’, our troubles would be over: ‘y orbits x’ is
clearly equivalent to ‘x is orbited by y’ (or, in other notation, ‘orbited by(x,y)’).
Then we could paraphrase (35) as
(p.41) Where will we find this predicate? We might add it to our stock of
primitives. But we have seen that we do better, avoiding ideological bloat and
inferential brutality, if we find a way to build it up from ‘orbits’. And indeed we
can, by introducing another predicate functor: the inversion functor, ‘INV’.
Where R is any two‐ placed predicate, ⌜INV(R)⌝ is a predicate that means ⌜is R‐ed
Page 29 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
by⌝. That is, for any two‐placed predicate R, the open sentence ⌜R(x, y)⌝ is
equivalent to ⌜INV(R)(y, x)⌝. With INV in hand, (35) gets paraphrased as
(39) ~Δ ~Δ(INV(orbits)).
Two particular ways of swapping positions turn out to be very useful. One way is
to simply invert the last two of a predicate's positions; call this minor inversion.
Or we could move a predicate's last position to the front and bump the rest down
a notch; call this major inversion. If ‘INV’ is a minor inversion functor, then (for
quaternary R, say) ⌜INC(R)(w, x, z, y)⌝ is equivalent to ⌜R(w, x, y, z)⌝. And if ‘𝕀𝕟𝕧’
is a major inversion functor, ⌜𝕀𝕟𝕧(R)(z, w, x, y)⌝ is equivalent to ⌜R(w, x, y, z)⌝. It
turns out that these two functors, wisely deployed, can generate any
rearrangement of predicates’ positions we might like.
One final issue needs resolving before we will be ready to paraphrase everything
in the target language. Consider the open sentence
(p.42) To deal with (40), we give ourselves a padding functor, which adds a
‘dummy’ place to a predicate. That is, for any predicate P and variable y, ⌜P(x 1,
…, x n)⌝ is equivalent to ⌜PAD(P)(y, x 1, …, x n)⌝. (The new variable, y, is a dummy
in that it does no work—as we ontologically minded folk would say, whether or
not some objects satisfy ⌜PAD(P)⌝ has nothing to do with what object gets
assigned to y.)
How does this help with (40)? First note that ‘PAD(attracts)(x, y, z)’ is equivalent
to ‘y attracts z’, because the object assigned to the place added by ‘PAD’—in this
case, ‘x’—doesn't make a difference to the predicate's satisfaction‐conditions. So
(40) is eqivalent to
because we replace (40)'s second conjunct with something equivalent. But when
we conjoin a two‐placed predicate and a three‐ placed one, we get a new three‐
placed predicate that is satisfied by some things only if the first two of them
Page 30 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
satisfy the two‐placed predicate and all three satisfy the three‐placed one. So
(41) will be equivalent to
making ‘(attracts & PAD(attracts))’ just the predicate we're looking for.
Call the Nihilist who paraphrases the target language into functorese the
Functorese Nihilist. The Functorese Nihilist avoids the (p.43) costs of
Propositional Nihilism. He avoids ideological bloat by limiting his new primitive
expressions to six. And he avoids inferential unsystematicity by appealing to
logical relations between complex predicates—relations encoded in functorese's
own attendant logic. This logic mirrors the predicate calculus: if ϕ entails ψ in
the predicate calculus, then the functorese paraphrase of ϕ entails the func‐
torese paraphrase of ψ in predicate functor logic.
Some (e.g., Jonathan Schaffer (2009, 368‐70) and David Chalmers (2009, 118);
see also Burgess and Rosen 1997, 185‐8), perhaps led by reasons such as these,
treat functorese as the Nihilistic language of choice. But I think the Nihilist's
hopes are misplaced if they are placed in functorese, for—despite its other
laudable features—I doubt that functorese has the primary qualification for the
Nihilist's paraphrasing job: that of being ontologically innocent. Even though it
avoids the ills that beset Propositional Nihilism, it falls straight into the ills of
Quiet Nihilism.
Page 31 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(p.44) Step One: Begin with F, and introduce a new language F λ as follows: it
has all the same primitive predicates and sentential connectives as F, and it
retains the feature‐placing functor ‘Δ’. These expressions are to be interpreted
in the same way as they are in F. But F λ does not have the other four predicate
functors; instead, it has variables and the abstraction operator ‘λ’ from §3.2.
Step Three: We appeal to (∗) from §4.1.1. If L 1 and L 2 are languages that differ
only in that L 1 has a term α where L 2 has a term β, this principle says:
(∗) If every term (other than α and β) is interpreted the same way in L 1 as
it is in L 2, and if the speakers of L 1 utter ϕ α in all and only the
circumstances in which speakers of L 2 utter ϕ β, then α and β have the
same interpretation also.
Now consider the target language, T, that the Functorese Nihilist wants to
paraphrase. It has all the same predicates as F δ: F uses for simple predicates
the predicates of T, and F δ inherits its simple predicates from F. Furthermore,
these predicates are to be interpreted in the same way in T and F δ, for the same
reasons. Also, T and F δ share the same truth‐functional connectives, which are
also to be interpreted in the same way. The only expressions that T and F δ differ
about are ‘δ’ and ‘∃’, and the Nihilist will say that ϕ δ is true in exactly the cases
where we say that ϕ ∃ is true. So, by (∗), ‘δ’ in F δ is interpreted the same way as
‘∃’ is in T.
We finish the argument with the following observations. We know that ‘∃x’ in T is
interpreted as ‘there is something that is an x such that…’. So the appeal to (∗)
in Step Three tells us that ‘δx’ in F δ must also be interpreted as ‘there is
something that is an x such that …’. But, by the construction of Step Two, we
know that ‘δx’ is interpreted in F δ as ‘Δλx’ from F λ. And we also know that ‘λx’
in F λ is interpreted as ‘is an x such that…’. So ‘Δ’ in F λ must be (p.45)
interpreted as ‘there is something that…’. But by the construction of Step One,
‘Δ’ in F has the same interpretation as ‘Δ’ in F λ; thus, ‘Δ’ in F is interpreted as
‘there is something that…’. Hence, F is not ontologically innocent after all; its
supposedly innocent expression ‘Δ’ is a quantifier proper in disguise.
Page 32 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
6.3.2. An objection
It is tempting to think that the (∗) Argument proves too much and so can't be
right. The main idea runs something like this:
We all agreed back in §5 that when we attach ‘Δ’ to a one‐ placed predicate
A we got an expression that meant ⌜It is A‐ing⌝. And we all agreed that ⌜It is
A‐ing⌝ did not mean, and did not entail, ⌜∃xAx⌝. If the (∗) Argument were
right, it would show that we were mistaken to even agree to this much—it
would show that even the feature‐placing language using only one‐placed
predicates was really quantificational all along. But surely we could use
sentences like ‘It is catting’ and ‘It is treeing’ without thereby saying that
there is a cat or that there is a tree!
But (∗) doesn't entail this impossibility. Suppose we came across a tribe of
‘feature‐placers’ that spoke just such a language—that had the same one‐placed
predicates as T, predicate‐functors ‘&’ and ‘~’, and an expression ‘ΔFP’ that
attaches to predicates to create sentences.
By mimicking the steps gone through above, we can transform their language
into a similar one that has the same predicates, no predicate functors, and a
variable‐binding operator ‘δ FP’ that means ‘ΔFP λ’. Then we compare this new
language to the fragment T 1 of (p.46) our first‐order target language that uses
only one‐placed predicates to see whether or not the two satisfy the antecedent
of (∗).
In order for both languages to satisfy this antecedent, the tribe must use ‘δ FP’ in
just the same way we use ‘∃’. But recall from §4.1.2 that ‘use the same way’ must
be understood dispositionally: it's not enough that they in fact use ‘δ FP’
whenever we use ‘∃’. For any counterfactual situation C, they must be disposed
to apply ‘δ FP’ in C exactly when we are disposed to apply ‘∃’ in C.
Page 33 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
But the tribe might not be thus disposed. They may, for instance, be disposed to
extend their feature‐placing language to many‐ placed predicates in the way to
be suggested below, in §7. Or they may have some other dispositions entirely, or
have no such dispositions at all. And if they are disposed to extend their
language in a way that doesn't turn it into functorese, then (∗) gives us no
reason to think that ‘ΔFP’ in their mouths means ‘there is’.
Since it is in principle possible for there to be tribes like the one imagined with
these sorts of dispositions, it is in principle possible for there to be an
ontologically innocent feature‐placing language like the one described in §5. The
argument does not prove too much after all.
Isn't it just clear that Predicate Functorese is the natural extension of the
innocent feature‐placing language from §5, and that its ‘Δ’ is the
ontologically innocent multi‐placed extension of ‘It is … ‐ing’?
In reply: no, it isn't clear at all. First, it is not clear that the functorese extension
is the natural way to extend ‘It is … ‐ing’; perhaps the extension to be discussed
below is more natural. But even if it were the natural extension, this need not
make it ontologically (p.47) innocent. A number of philosophers have thought
that ontologi‐ cally guilty expressions naturally emerge when we extend feature‐
placing languages to deal with troublesome cases. We start out saying things
like ‘it is catting over here’ and ‘it is dogging over there’, but then run into
various kinds of troubles expressing everything we want to express. For
instance, we get into trouble deciding whether placed features ‘go together’ or
not (Evans 1975, Quine 1992), or how features placed yesterday relate to
features placed today (Evans 1975, Strawson 1954, 1963), etc. So we extend our
feature‐placing language by adding some pegs to place these features on; we
then know whether or not features go together, either right now or over time,
based on whether or not they're on the same peg.
There is particular reason to think that something like this happens when we
extend the one‐placed version of the feature‐placing language to full Functorese.
We can think of ‘placing features’ as throwing rubber bands onto a peg‐free
board. We say ‘It is elec‐ troning’, and throw the ‘electron’ rubber band on the
board; ‘It is protoning’, and throw the proton rubber band on the board; etc.
The other predicate functors let us make complex rubber bands out of simpler
ones. But we have real conceptual difficulty understanding the Functorese
Nihilist's preferred extension of ‘Δ’ to relational predicates. What have we done
when we say ‘Δ(orbits)’? We have somehow thrown part of the ‘orbits’ rubber
band down on the board while keeping the other part up. But what are we going
to do with the part that we’ve kept off the board? Suppose we prefix the new
complex predicate with ‘~Δ~’. Intuitively, this tells us that, for any other place
Page 34 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
where we could throw a rubber band down, we must make sure the other half of
this (kind of) rubber band gets thrown there also. But now it no longer matters
simply that thus‐and‐so a rubber band has been thrown on the board—it also
matters where it's been thrown, and where it could be thrown, too. In other
words, certain locations of the board now matter. Certain locations on the board
have to count as possible parking places for parts of polyadic rubber bands, and
it becomes significant when parts of two different rubber bands land on just one
of these special locations.
Once we've gone this far, we've all but introduced pegs. The point of using pegs
to represent objects in a model of reality is (p.48) that they mark out certain
locations on the board as special, as potential landing sites for parts of rubber
bands. (We use pegs to mark these locations for practical reasons: they keep the
rubber bands from sliding around.) Since pegs represent objects, this is to say
that objects are special landing‐sites for features. Once it matters where one
part of a rubber band has been stuck, we've smuggled in ontology. So, even if
the functorese ‘Δ’ is the natural extension of the one‐placed feature‐placing
language, there is good reason to think it is an extension that introduces
ontology— and so good reason to think that the (∗) Argument was right all
along.
We could let Δ turn many‐placed predicates into sentences. Just as ‘It is raining’
says that rain is going on, and ‘Δ(proton)’ says that protoning is going on, we
can understand ‘Δ (orbits)’ as saying that orbiting is going on and ‘Δ(repels)’ as
saying that repelling is going on.
Saying that orbiting is going on will be the Nihilist's way of paraphrasing our
claim that something orbits something else. Thus for any n‐placed predicate R,
⌜Δ(R)⌝ will be the Nihilist paraphrase of ⌜∃x 1… ∃x n(R(x 1, …, x n))⌝.
Page 35 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
return to these (p.49) faithful friends.) Then we can find complex predicates
equivalent to the embedded open sentences, and paraphrase (43)–(44) as
respectively.36
(47) ∃x~∃y…
And the current proposal has nothing to say about sentences of this sort.
The first of these says that something neither attracts nor repels anything else;
the second says that something attracts at least one thing but repels nothing.
The Nihilist ought to be able to distinguish cases in which it is good to say one of
these but not the other. But he cannot, using just ‘Δ’ and ‘∑’, give these two
different paraphrases.
Page 36 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
resist any explanation, for the expressions ‘Δ’, ‘∑’, and so on are taken as
semantic black boxes— the Nihilist has nothing to say about them except that,
when attached to predicates of a certain sort, they produce sentences fit for
certain sorts of paraphrases. But with no further story as to what these
expressions mean, he cannot explain the inferences they participate in.
Holism: In §4.2.3, we noted that we ontologically minded folk can think of the
more global fact expressed by (44) as being somehow ‘built up’ out of the fact
that an x attracts a y, the fact that a y repels a z, and the fact that the y being
attracted is the same y as the one doing the repelling.
But a Nihilist who paraphrases (44) as (46) thinks of this fact as essentially
‘placing’ a complex feature in reality—of deploying, in a peg‐free way, a complex
rubber band of the shape in figure 7. And although we make this complex
feature by gluing together the ‘attracts’ and ‘repels’ rubber bands, we cannot
think of the deployment of this complex rubber‐band structure as being
somehow ‘built up’ out of the deployment of the ‘attracts’ and ‘repels’ rubber
bands. The mere fact that these two rubber bands have been thrown on the
board isn't enough to guarantee that they overlap in the required way. A
deployment of an ‘attracts’ rubber band corresponds to an x attracting a y, and
the deployment of a ‘repels’
This gives us some reason to reject Ontological Nihilism. But this is hardly
headline news. After all, only a few metaphysicians would ever have suspected
Ontological Nihilism of truth in the first place.
Page 37 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
When we see the evils we must embrace in order to make Onto‐ logical Nihilism
work, though, we gain a better appreciation for our ontology. It is through
ontology that we can think of reality's global structure as built up out of more
local structures, for it turns out that a set of ‘pegs’, of things, is crucial for this
sort of bottom‐up picture. It is by identifying things across different local
structures that we can build up more global structures. By picking out which
things in this local structure are identical to which things in that one, we have a
way to link those two structures together to come up with a more global one.
And it is by thinking of the world ontologically that we can understand the
validity of certain inferences: they are valid because the pegboard structure
described by one claim fits or (p.52) doesn't fit with the structure described by
the other. Thinking of the world in an ontological way provides us with the
resources to offer powerful systematic explanations of a wide variety of
pervasive facts. That, perhaps, is part of why ontology has been important to
philosophy all along.
University of Leeds
REFERENCES
Bibliography references:
Boolos, George, Burgess John P. and Jeffrey Richard C. (2002) Computability and
Logic, 4th edn. (Cambridge: Cambridge University Press).
Burgess, John P. and Gideon Rosen (1997) A Subject With No Object (Oxford:
Oxford University Press).
Page 38 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Goodman, Nelson (1979/1983) Fact, Fiction, and Forecast, 4th edn. (Cambridge,
Mass.: Harvard University Press).
Hirsch, Eli (2002) “Quantifier Variance and Realism”, Philosophical Issues 12,
Realism and Relativism: 51–73.
Lewis, David (1983) “New Work for a Theory of Universals”, The Australasian
Journal of Philosophy 61: 343–77. Reprinted in Lewis (1999: 8–55).
Lewis, David and Lewis, Stephanie (1970) “Holes”, The Australasian Journal of
Philosophy 48: 206–12. Reprinted in Lewis (1983b: 3–9).
Page 39 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Quine, Willard van Orman (1948) “On What There Is”, The Review of
Metaphysics 2: 21–38. Reprinted in Quine (1980: 1–19) .
——(1976) The Ways of Paradox and Other Essays, 2nd edn. (Cambridge, Mass.:
Harvard University Press).
Schaffer, Jonathan (2009) “On What Grounds What”, in (Chalmers et al. 2009:
347–83).
Seppänen, Aimo (2002) “On Analysing the Pronoun IT”, English Studies 83/5:
442–62.
Stanley, Jason (2002) “Context and Logical Form”, Linguistics and Philosophy
23/4: 391–434.
Page 40 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
van Inwagen, Peter (1990) Material Beings (Ithaca, NY: Cornell University
Press).
Notes:
(*) Thanks to Karen Bennett, Andrew Cortens, Cian Dorr, John Hawthorne, Hud
Hudson, Ricardo Mena, Christopher Pynes, Starr Turner, Ted Sider, Robbie
Williams and Dean Zimmerman for helpful comments and discussion.
(1) Ontological Nihilist, that is. For stylistic reasons, I’ll often drop the
‘Ontological’, letting the capital ‘N’ do the disambiguating work.
(2) At least, we do when using the languages with which I have any familiarity;
perhaps some languages do not at bottom operate this way. If so, it would be
interesting to see what kind of metaphysics native speakers of these languages
produce.
Page 41 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(6) We will need an even more general scheme to deal with e.g. ‘There are six
holes in this cracker’; see Lewis and Lewis (1970) for a fuller treatment and
critical discussion.
(7) A paraphrase scheme might not be the only way a perforational nihilist can
give us this account. Perhaps instead he could explain why hole‐talk is a useful
fiction using a non‐paraphrase methodology similar to that found in Field (1980).
But I stick to paraphrase schemes in the text because I cannot see how to even
begin a Fieldian treatment of Ontological Nihilism.
(8) Likewise, we need not insist that ‘paraphrases’ avoid semantic ascent;
anyone who thinks there are no ultimately egocentric facts—no facts that must
be stated using terms like ‘I’ or ‘you’, for instance—may fairly take Kaplan's
(1989) semantics for indexicals as providing a ‘paraphrase’, in our sense, of
tokens of sentences of the form ⌜I am F ⌝ even though, as Kaplan argues, there is
no way to provide the account as a translation from sentences to sentences all in
the ‘material mode’.
(9) There are subtleties. Perhaps the perforational nihilist uses one recipe to
paraphrase ‘There is a hole in the bridge’ as ϕ and uses a very different recipe to
paraphrase ‘There is a hole in the door’ as ψ. Nonetheless, if ϕ and ψ are
themselves clearly very similar facts, then the differences in the recipes used to
get to them from the original hole‐sentences do not mean the nihilist is denying
any intuitive similarity.
(10) Of course, as a Nihilist, he denies that there are any facts at all. But we
should think of his ‘fact’‐talk as merely a useful turn of phrase for explaining his
view to us doubters. He will talk about facts only while trying to get us into the
spirit of his view; once we are fully converted to Nihilism, he promises to show
us how to understand what he was saying without any ‘fact’‐talk at all. Similarly
for his talk about ‘the world’, ‘structure’, ‘sentences’, and so on.
(11) Although see Cartwright (1954), Rayo (2007), Richard (1998), and van
Inwagen (1998, Thesis V) for discussion.
(12) If Joe mistakenly thinks that ‘unicorn’ means zebra, he might not
ontologically commit himself to unicorns when he says ‘There are unicorns’.
Thanks here to Ted Sider.
(13) This, more or less, is how Mark Richard (1998) seems to understand the
notion; and Peter van Inwagen (1998, Thesis V) is perhaps best interpreted this
way, too. Agustín Rayo (2007, 2008) suggests instead that I'm ontologically
committed to Ks if I understand and sincerely assert a sentence with truth‐
conditions which demand that there are Ks. Little hangs on this distinction in
what follows.
Page 42 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(14) There are, in fact, quite a few languages that go by the name of ‘lambda‐
abstraction’. We are here concerned with the first‐order fragment of typed
lambda‐abstraction languages with categorical quantifiers (Gamut 1991, 102–15;
see Hindley and Seldin 1986, 266–86 for the untyped counterpart). Since the
languages have categorical quantifiers, only ‘λ’ can bind variables; since they
are typed and first‐order, ‘λ’‐abstraction can only form first‐order predicates.
(16) Since they are otherwise just like their ‘green’‐speaking counterparts, they
intend to say things like ‘this sapphire is both blue and grue’ after the future
date. But they have not yet introduced a term to use for green things observed
after this date.
(17) One caveat: we should not be concerned with the speakers’ dispositions to
utter sentences containing both disputed words. For instance, we shouldn't
demand that (∗)'s antecedent not be satisfied in the above ‘blue’/‘eulb’ case
simply because the ‘eulb’‐speaker is disposed to assert ‘eulb things are not blue’
and you, at least after serious reflection, are not disposed to assert ‘eulb things
are not blue’. The question is whether, setting aside the way the speakers think
these terms interact, we should interpret them the same way; (∗) is supposed to
give us a guide for determining whether speakers' assertions of this sort are
plausible, and as such it should not be overly sensitive to these assertions
themselves. Cf. §4.1.3.
(19) This might very well suggest that the ‘grue’‐speakers from §4.1.2 really
meant ‘green’ by ‘grue’. But this is perhaps to make the constraint too strong: an
ideal interpreter's injunction to give words an ungerrymandered interpretation
ought not outweigh a community's explicit intention to use an expression in a
gerrymandered way.
(20) The situation is even better: if the rest of the respective languages’
semantic backgrounds are the same, a lot of other interpretative issues between
both parties must have been settled in favor of similar meanings for their
languages. Given that this much has been fixed, and given that the only
Page 43 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
remaining expression is used the same way by all parties, it becomes very hard
to think of any interpretative constraint that could pressure us to assign α a
meaning intensionally inequivalent to β.
(22) Hawthorne and Cortens's original invalid example, ‘He ran halfway up the
hill; therefore, he ran up the hill’ seems to have a different syntactic form than
their valid example, ‘He ran quickly up the hill; therefore, he ran up the hill’. In
the first case, ‘quickly’ modifies ‘ran’; in the second, ‘halfway’ modifies ‘up the
hill’.
(23) They are thus like the semantically empty ‘it’ of weather reports; see §5
below.
(24) Hawthorne and Cortens (1995, 151–2); cf. van Inwagen (1990, 128).
(26) This may not be quite right. The semantics of ‘is raining’ may make it a
predicate of places. Even though the ‘it’ must be semantically empty (see
Seppänen, 2002, 445–53), ‘is raining’ may nonetheless include a location ‘slot’ at
the semantic level, filled in by context in a bare assertion of ‘It is raining’ but
explicitly filled in construc tions such as ‘It is raining in Austin’ or bound as in
constructions such as ‘Wherever Joe went, it rained’ (cf. eg. Stanley, 2002, 416–
18). Out of charity towards the Nihilist, though, we ignore these complications
here; cf. the next note.
Page 44 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
(28) Compare, for instance, ‘It is raining and it is cold’ with ‘It is raining and
cold’. This transparency of ‘it’ to truth‐functional operators is one reason
semanticists think it semantically empty; cf. Seppänen (2002, 448).
(29) Hawthorne and Cortens (1995, 148‐9) suggest using adverbs to solve the
problem, rendering a sentence such as ‘There is a red cat’ as ‘It is catting redly’.
While my suggestion here is similar in spirit, by not distinguishing between
feature‐placing verbs and adverbs, it is somewhat more streamlined: for
instance, it can paraphrase (27) without deciding (as Hawthorne and Cortens's
proposal would have to) which of ‘is positively charged’ and ‘is negatively
charged’ to turn into a verb and which into an adverb.
(30) This follows from the fact that a sentence of the form (33) is equivalent to
one using only one variable; cf. Boolos et al. (2002, 274‐5).
(31) To say that one open sentence P is equivalent to another, Q, is to say that P
can everywhere be replaced for Q salva veritate (at least in languages without
opaque contexts). ‘Equivalence’, in this sense, is as dependent upon where
variables are placed as it is upon where predicates are placed.
(32) And perhaps unary ones, if we take ⌜INV(P)⌝ to be equivalent to P for unary
P.
(33) And an identity predicate, shared (I assume) by the target and paraphrase
languages.
(34) More precisely, (open or closed) sentences of the form ⌜δxϕ⌝ are to be
interpreted as ⌜Δλxϕ⌝.
(35) At least, we ought to agree insofar as we are not troubled by, or are setting
aside, worries that we can only make sense of the feature‐placing languages on
the model of weather sentences, which in turn must be thought of as covert
predicates of places as discussed in n. 26.
(36) A different option involves complicating the ‘Δ’‐functor, giving it extra ‘slots’
for more predicates and paraphrasing (43) as Δ(electron, proton ǀ orbits). The
idea here is that the predicates on the left side of the‘ ǀ’ indicate unary features
to be placed, and those on the right side indicate many‐placed features to be
placed ‘in between’ the unary features, as it were. But it is not clear how to
extend this to more complex cases; see Sider and Hawthorne (2003) for a
version of this proposal and a discussion of some of the difficulties involved.
(37) If throwing two rubber bands on the board so that they look like figure 7 did
the trick, it would be by making ‘places’ on the board important and thereby
smuggling ontology back into the picture, as discussed at the end of §6.3.2.
Page 45 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020
Ontological Nihilism
Page 46 of 46
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press,
2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.
Subscriber: UNAM; date: 19 October 2020