Among Quine's main concerns in his “On What There Is”, there was that of solving a problem of expressibility for ontological denials. His proposed solution to such a problem was, in a purely Carnapian vein, a shift of attention to the semantic features of ontological claims - what Quine called the strategy of “semantic ascent”. Quine's relevant assumption is that talk about language is much less controversial than first-order talk of worldly items. My contention is that the semantic ascent strategy fails as a somehow “neutral” means to clarify ontological disputes and that it is better understood as a means to resolve or even dissolve such disputes.
1 Introduction
“There are more things in Heaven and Earth than are dreamt of in your philosophy”, says Hamlet. But imagine for a moment you are Horatio and suppose you don’t agree: you do not think that there is (or might be) something besides the entities countenanced in your philosophy. How would you express that belief? After all, how can you talk about and refer to what Hamlet believes in but you do not? Alternatively, suppose that upon reflection – maybe because Hamlet’s theory is the best available theory – you are somehow led to believe what Hamlet says, but you still feel uneasy with such a belief. Then, you would believe that there are somehow things that you do not believe to exist. But is there any intelligible way of expressing such a belief?
In what follows, I shall try to show how Quine’s strategy in answering the first question of expressibility (“How can one coherently refer to things in whose existence he or she does not believe in?”) – the strategy of “semantic ascent”– can in effect be exploited not only to answer positively to the second question, but is somehow exploited in arguing for those very ontological views that are Quine’s main polemical targets. To make things as clear as possible, I shall distinguish between four different views of ontological disagreement, and spell out the sense in which the last three of them share in the end the same underlying dialectical strategy – i.e. a shift of the subject of the disagreement from ontology to language. My contention is that such a strategy of semantic ascent is better understood as a means to resolve or even dissolve ontological disputes, rather than as a somehow “neutral” means to clarify this kind of disagreements.
Quine identifies at least two compelling reasons “for operating on a semantical plane” in an ontological debate1:
i) “One reason is to escape from the predicament […] of not being able to admit that there are things which McX countenances and I do not. So long as I adhere to my ontology, as opposed to McX’s, I cannot allow my bound variables to refer to entities which belong to McX’s ontology and not mine. I can, however, consistently describe our disagreement by characterizing the statements which McX affirms.” ii) “Another reason for withdrawing to a semantical plane is to find common ground on which to argue. Disagreement in ontology involves basic disagreement in conceptual schemes; yet McX and I, despite these basic disagreements, find that our conceptual schemes converge sufficiently in their intermediate and upper ramifications to enable us to communicate successfully on such topics as politics, weather, and, in particular language. In so far as our basic controversy over ontology can be translated upward into a semantical controversy about words and what to do with them, the collapse of controversy into question-begging may be delayed.”
The first reason Quine identifies is that in expressing an ontological disagreement
one might express self-defeating claims such as “There are things that you
countenance but I do not”. Of course, as it stands, such a claim is far from being obviously
self-defeating. A given statement is self-defeating to the extent that it defeats the very
aim it is designed to serve. In this sense, “Thank God for atheism” is unmistakably
self-defeating. But in order to construe the statement “There are things that you
countenance but I do not” as a self-defeating one, one has to assign a particular
“ontological responsibility” to phrases such as “there are” or “something”, namely to what in
logical jargon are called the idioms of existential quantification. Quine’s notion of
“ontological commitment”, in effect, brings with it such an idea of ontological
responsibility as regards our idioms of quantification. Things can be made even clearer
if we put the matter in terms of belief, for some kind of belief is what “ontological
commitment” seems to come down to2. Then in asserting a statement such as “There
1 Quine (1948, p. 16).
2
The second, related reason to operate on a semantic level that Quine identifies is somewhat Carnapian. Namely, it is the idea that talk about language constitutes a fairly “common ground on which to argue”. On Quine’s picture, in effect, in disagreeing about ontology we may, more or less inadvertently, slip into equivocation and question begging, precisely because of the aforementioned problem of expressibility. Quine’s relevant assumption here is that talk about language is much less controversial than first-order talk of worldly items. This isn’t in general true, I believe. However, as we shall see, the problem is that the semantic ascent strategy is bound to fail as a “neutral” means to clarify ontological disputes, for it is largely used as a means to resolve or dissolve such disputes. But let us now review Quine’s reasons in more detail, and let us take into account some of their more significant consequences.
Let us begin by taking into consideration Quine’s first reason to withdraw to the semantic plane in ontological debates. Such a reason, as I have already said, has to do with the very possibility of expressing disagreement. Quine’s worry is due to the apparent conflict between what looks like a plausible way to understand the truthmaking underpinnings of existential claims and what appears to be a rather uncontroversial way to characterize ontological debate. We might reconstruct Quine’s reasoning as follows. Suppose, for the sake of explanation, that among the things that Hamlet believes inhabit Heaven there is Pegasus; and suppose that Horatio doesn’t believe that there is any such thing. A prima facie plausible way to describe how they might express their ontological beliefs in such a disagreement – call it the “Naïve View of Ontological Disagreement” – seems to be the following3: [Naïve View of Ontological Disagreement] “Pegasus exists” is such that (i) it can be used by Hamlet to express that he believes in the existence of Pegasus (ii) its negation can be used by Horatio to express that he does not believe in the existence of Pegasus properties – plus the general intellectual requirement that if one becomes aware that one’s belief that p commits one to the further belief that q, then one should either believe that q or cease to believe that p. (p.9, italics mine) 3 This way of reconstructing Quine’s reasoning is partly founded on some illuminating remarks I found in B. Hellie’s unpublished paper On What Exists. The Naïve View accounts for the intuition that the disagreement between Horatio and Hamlet is about Pegasus. More precisely, by saying “Pegasus exists” Hamlet would be talking about Pegasus and by saying “Pegasus does not exist” Horatio would be talking about Pegasus as well. Pegasus indeed is the object of their debate. This is what happens in any ordinary disagreement, or at least so it seems4. Moreover, the Naïve View seems also to account for the intuition that Hamlet’s and Horatio’s choice of words is the most straightforward in order to express their own views about the matter in question. In other words, there seems to be no apparent reasons for them to choose a different way to express their opinions. According to Quine, however, a plausible case can be made to the effect that such a characterization is not as uncontroversial as it might seem at a first glance.
In particular, “Pegasus does not exist” cannot be used by Horatio to express that he does not believe in the existence of Pegasus. In effect, in order for this kind of assertions to be meaningful, it seems that the name or denoting phrase that features in them must succeed in picking out something. But if “Pegasus” succeeds in picking out something, then what it succeeds in picking out has, after all, some measure of existence, and “Pegasus doesn’t exist” will be then false. This is a genuine puzzle – and one with a quite long historical pedigree – so, according to Quine, something in such a picture must be given up. Otherwise, Horatio would find himself in a somewhat puzzling rhetorical disadvantage for he simply cannot disagree with Hamlet.
Following the path traced by
vided that such a denial is regimented as ¬∃x(x pegasizes).
On the Quinean View, “Pegasus does not exist” can be used by Horatio to express his disagreement with Hamlet, because his denial, as it is now formulated, does not lead him to a contradictory conclusion anymore. From the first order formula “¬∃x(x pegasizes)” we cannot derive that there is something that does not exist, but simply that none of the existing things has the properties that Pegasus is thought to have.
But why don’t think just that Horatio might in the end be able to refer to and talk about things in whose existence he does not believe? In order to give an affirmative answer to such a question, one must seriously entertain the possibility that an item might be the value of a variable in Hamlet’s domain and yet not exist, i.e., not be the value of a variable in Horatio’s own domain. This looks like a quite plausible characterization of their disagreement, but as we just saw, on Quine’s account such a characterization leads straight to contradiction. The only other option to avoid such a contradiction amounts to postulating some sort of ambiguity in our notion of existence, and this is something that Quine simply finds utterly unintelligible. Quine firmly refuses the idea to the effect that “exist” might have an equivocal meaning, and indeed views a disagreement about this as a merely verbal dispute. In fact, Quine disambiguates between the (alleged) alternative meanings by choosing for himself the word “is”5, so that his sense of existence should be identified with nothing but the “weakest” of the senses of existence endorsed by the (not so) imaginary philosopher Wyman6, namely what, by Wyman’s own lights, would be the most unrestricted sense – the sense in which whatever can be thought of or referred to, by any means, is said to exist.
However, non-fictitious advocates of such a view would just insist that what Quine is really doing here is equivocating. For what he takes to be the most unrestricted sense, is just the restricted sense by their lights, namely the sense in which quantification expresses only actual existence. In other words, such philosophers would argue, the extension of the notion of actual existence constitutes only a proper subset of the universe of discourse in their semantic framework. And this may be stated also as a difference between being and existence.
Such an idea has been famously central to the works of Alexius
6 “Wyman, by the way, is one of those philosophers who have united in ruining the good old word “exist” […] in an ill-conceived effort to appear agreeable, genially grants us the nonexistence of Pegasus and then, contrary to what we meant by nonexistence of Pegasus, insists that Pegasus is. Existence is one thing, he says, and subsistence is another. The only way I know of coping with this obfuscation of issues is to give Wyman the word “exist”. I’ll try not to use it again; I still have “is”. (1948, p. 3) [Inflationist View of Ontological Disagreement] “Pegasus exists” is such that (i) it can be used by Hamlet to say that he believes in the actual existence of Pegasus (ii) its negation can be used by Horatio to say that he does not believe in the actual existence of Pegasus
The problem of expressibility is dissolved as long as what follows from Horatio’s denial is just that Pegasus is a possible object whose actual existence is denied. Notice, however, that it is still according to a Quinean standard of ontological commitment that the inflationist is admitting all sorts of non-actual objects. That is, it is still according to Quine’s view on ontological commitment (and responsibility) that such a view might be qualified as “inflationary”. The Inflationist seems committed to the idea that, whether possible or actual, anything in her domain of discourse enjoys some kind of existence. This observation is worth of attention, since advocates of the “Deflationist View” – in a certain sense the mirror-view of the inflationist one, given their common appeal to “quantificational ambiguity” – want to escape, as we shall see, any charge of inflationism precisely by denying the Quinean standard of ontological commitment.
Let us take stock now. And let us review the dialectical strategies that underlie the
Quinean and Inflationist Views of ontological disagreement in their attempt to
resolve (or dissolve) the alleged problem of expressibility. As it turns out, they are all
of the same kind, namely they are strategies that typically operate on a semantic
level. The apparent reason for operating on such a level is that of avoiding
equivocation and/or contradictory talk in ontological disputes; but the genuine reason for
“going semantic” seems that of arguing for determinate ontological positions. In
other words, Quine’s concern in “On What There Is”, was – to borrow Morton
Despite the appearances, on the other hand, the Inflationist alternative makes a somewhat analogous shift from disagreement about what there is to disagreement about the meaning of certain phrases. The inflationist takes the superficial grammatical form at face value. “Pegasus” does refer to something, namely to a possible object. Hence, in taking for granted the prima facie semantics of existential claims the inflationist is – as a matter of fact – taking stance on semantic matters. Moreover, she escapes charges of contradictory talk by means of a further move: i.e. by postulating an ambiguity in what she means with expressions such as “there is” and “exists”.
Even if the Quinean and the Inflationist approaches look as completely different accounts of the problem of expressibility, they both exploit the very same strategy, i.e. non-neutral forms of semantic ascent. As a result, on Quine’s view, for instance, we can disagree about whether there is something with all the Pegasus-properties, but we cannot say that “Pegasus” refers to a possible object. On the inflationist view, on the other hand, we can disagree about whether Pegasus is an actual object, but still we cannot deny being to Pegasus.
The analysis of negative existentials – and the resulting contrast between a view such as Quine’s and a view such as the Inflationist one – is indeed perfectly illustrative of how the implicit or explicit shift to a semantic level in ontological debate has an argumentative function, as long as such a shift is intended to convey “introductionist”7 and eliminativist positions (or “realist” and “antirealist” positions, as you might say).
According to the picture I’m trying to suggest, in an ontological disagreement, say, about the existence of “ordinary objects”, it won’t then be enough for a believer in their existence to merely state her view by asserting, “There is a round table in the office”. This will not suffice to commit her to the existence of tables, for her claim, as Carnap would have it, is somehow “incomplete”8. She will be committed to the existence of such entities only after having qualified her claim to the effect that it has to be interpreted at face value, so that it has the appropriate truth-conditions according to which such entities must exist for the statement in question to be true9. To put it in a more familiar terminology, before one is allowed to read off any ontological commitment from a given claim, one must somehow establish how to interpret this very claim. The main reason for this resides in the fact that eliminativist challenges – i.e. the expressions of disagreement about the existence of a given sort of entity – enter the “playfield” precisely in relation to the interpretation of the antieliminativist’s claims. This is why Quine questions the naïve semantics of negative existentials: the problem of expressibility is nothing but a problem with the alleged ontological consequences of a given talk.
In other words, familiar eliminativist moves seems to make sense only on the “semantical plane” or at the formal level, to use Carnap’s terminology. Precisely because eliminativists argue to the effect that, in light of a correct interpretation of the language concerned, there is nothing which is referred to or described in material talk, say, of Fs. Typically, then, disbelievers in the existence of some kind of objects will provide a story about how claims apparently about them should be interpreted in ways that do not carry any commitment to such things. The dispute, once it has reached such a stage – i.e. the semantical plane or formal level – seems somehow bound to be carried on by its participants at this very same stage. Thus, believers in the existence of Fs would answer – indeed they look as they are somehow forced to answer – to the eliminativist challenge by arguing for the correctness of their (material) talk about Fs. Once the semantic question is raised, it seems, there appears to be no easy way out for believers in the existence of certain kind of entities who are willing to refuse to play the game10, so to speak. In meaning what they say, 7 The terminology is borrowed from Carrara and Varzi (2001). 8 Carnap (1935, p. 77) 9 Of course, in the case of inflationist positions there may be the further semantic qualification concerning the sense of “existence” in question. 10 Compare Lewis (1996, p. 559): “No matter how far-fetched a certain possibility may be, no matter how properly we might have ignored it in some other context, if in this context we are not in fact ignoring it but attending to it, then for us now it is a relevant alternative. It is in the contextually determined domain.” That is, any possibility that one actually attends to cannot be properly ignored (this is what Lewis calls the “Rule of Attention”). such believers are, as a matter of fact, undertaking the commitment to stick with what is implied by a literal understanding of their assertions11.
The semantic ascent strategy’s underlying assumption is that we can make sense of a disagreement only within a shared (by the disputants) semantic framework. In order to avoid equivocation and contradictory talk in a given disagreement, we need to clarify what parties to such a dispute mean with their words. Generally speaking, the “clearist” intentions seem entirely justified, for there could be no genuine disagreement without reciprocal understanding. However, in an ontological disagreement things become somewhat trickier, given the use of “non-neutral” semantic ascent strategies, such as those explicitly advocated by Quine and implicitly presupposed by the Inflationists. For the proposed clarifications simply boil down to providing interpretations of the claims in question that supports specific ontological views (somehow begging the question in relation to other ontological positions). But then how can parties to an ontological disagreement be in the end able to avoid equivocation and question begging?
An answer to such a question might be that of taking the ontological
consequences of a given talk less seriously. The view I shall define as the “Deflationist
View” is meant to provide such an ontologically innocent semantic framework. Such
a view is intended as a reasonable way to accept both the Naïve View and its strict
consequences – without accepting however their ontological costs. For the sake of
explanation, I shall focus on a promising version of this view that has been recently
advocated by Thomas
Hofweber does not deny that quantifiers often function as Quine says they do,
namely as carrying ontological commitment to objects. For instance, Hofweber
maintains, a sentence like “Something ate my cheese” can be true only if there is
some “object or entity that ate the speaker’s cheese.” In this sense, there seems to be
“no doubt that there is a close connection between the truth of some quantified
statements and ontology” (2005, p. 268). In particular, Hofweber holds that we
should distinguish between two different ways in which quantifiers actually function
in natural language. This, again, amounts to postulating some form of ambiguity in
the meaning of “existence”. The first way our quantifiers may function is the one that
is spelled out by the objectual interpretation, i.e. the interpretation under which, for
instance, the sentence “something ate my cheese” “will only be true if reality
contains an object or entity that ate the speaker’s cheese” (2005, p. 268). Such an
interpretation, Hofweber points out, “imposes some conditions on the domain of entities
that our discourse is about”. (2005, p. 271), and this is why he calls it the “domain
conditions” reading. In other words, on such a reading, when we quantify we
somehow try to reach out, so to speak, the sphere of our language, so that the success or
failure of such an attempt is strictly dependent on the actual composition of such an
11
One of the uses we have for quantifiers is to communicate information that is in certain respects lacking with sentences that have quantifiers in them, and apparently have them essentially (given the information we have). (2005, p. 271)
The example that Hofweber asks us to consider in support of such a claim is the case in which one might utter a quantified statement like the following:
There is someone that Fred admires very much, and that person is also admired by many detectives despite the fact that one has forgotten that the detective in question is Sherlock Holmes. As it happens, there is no such person “out there”. So, on the objectual interpretation (the “domain condition” reading), the sentence simply turns out false. But Hofweber points out that it still has a true reading – the reading in which the quantifier is regarded as a mere “placeholder” for the forgotten part of the information12. On the inferential role reading of quantifiers the concern is somehow wholly internal to the sphere of language spoken by the relevant speaker, for what counts are the inferential relations between statements.
So far, so good. But how all this is meant to apply in view of a Deflationist account of ontological debate, and in particular in relation to the problem of expressibility? Hofweber’s account seems to make room for the possibility of sticking to the Naïve View, while dissolving the problem of expressibility that concerned Quine so much. For the inference from the negative existential claim “Pegasus does not exist” to “There is something that does not exist” would be both trivial and ontologically unloaded, precisely because the reading of the quantifier in such a case should be the “inferential role” reading13. Then the Deflationist View would roughly amount to a qualified version of the Naïve View: [Deflationist View of Ontological Disagreement] “Pegasus exists” is such that (i) it can be used by Hamlet to say that he believes in the existence of Pegasus (ii) its 12 In fact, he argues that “if we had not forgotten who it was, we could have used a name or some other term for this person instead of the quantifier. But given that we forgot we have to use a quantifier to communicate less than the complete information. Furthermore, it doesn’t matter here whether or not the person admired is real and exists. In either case we can get useful information out of what we know about Fred. Whether the person admired is real is not of importance here. The quantifier has to be a placeholder no matter what the original term was, whether or not it referred to some entity, failed to refer, or had some completely different function” (2005, p. 272). 13 “In their internal use the inference from ‘F (t)’ to ‘F (Something)’ is always and trivially valid, no matter what ‘t’ is. Quantifiers in their internal reading are not directly related to ontological issues, i.e., the truth of ‘F (something)’ or ‘Something is F’ with the quantifier in its internal use, does not by itself settle the ontological question about Fs.
It pushes the issue back to the quantifier free instances.” (2005, p. 274). negation can be used by Horatio to say that he does not believe in the existence of Pegasus (provided that the quantifier featuring in the quantified statement “There is something that does not exists” that follows from the ontological denial is construed in its internal – ontologically innocent – reading).
According to Hofweber, both Quine, in envisaging a problem of expressibility, and Inflationists, in inflating the domain of entities with non-existent objects, make just “the mistake of collapsing this [internal] use of quantifiers into the domain conditions use” (2000, p. 23). However, one might argue against such a deflationist account from both the Quinean and the Inflationist perspectives.
The first objection amounts to pointing out that where Hofweber sees a difference
in meaning (agreeing on this with the Inflationists, at least on the surface), a Quinean
might be in its own right in seeing none. For such a philosopher what Hofweber calls
a “collapsing” of uses would be, in fact, just a “coinciding” of uses. In the Quinean
picture, the inferential role is strictly associated to the domain conditions and there’s
simply no way one can alleviate the ontological burden of his or her quantifications.
The second objection might be done along the same lines, but this time from the
Inflationist’s standpoint: the latter, as
However, a further problem has to do with such a semantic picture and its
consequences as to the very possibility of ontological disputes. For even if Hofweber’s
account still allows for an “ontological” reading of quantifiers (one might even talk
of such use as the “literal reading”) – thus apparently allowing ipso facto for
meaningful ontological disagreements – there would be always room for questioning
whether one’s use of quantifiers qualifies in the end as ontologically loaded or
ontologically unloaded. One might argue that precisely as a result of the pervasivity of
semantic underspecifcation that Hofweber himself has compellingly illustrated, there
might not be a principled way to tell the ontologically loaded uses from the
ontologically unloaded uses. Philosophers such as Steve
The problem with the Deflationist solution is that the very search for “ontologically neutral” ways of expressing our views seems to lead straight to deflating the disagreement itself, for no parties to the dispute could then safely draw inferences as to the ontological beliefs of the other disputants just from what they say. So how can the believers in the possibility of ontological disagreement react to such a picture?
The choice is somehow mandatory. They will react by providing a different
semantic story, a story that makes room for the possibility of limiting the seemingly
pervasive effects of semantic underspecification, and in which, there are indeed clear
cases of strict and literal talk. As it might be clear, the pattern here is exactly the
same as the one that we have identified in the case of the ontological disagreement
between believers and disbelievers in the existence of some kind of entities. The only
difference is that the level of the disagreement has changed: now the dispute has
turned into a metaontological disagreement between believers and disbelievers in the
possibility of ontological disagreement itself. While, on the one hand, such believers
will advocate a broad semantic picture that we may define as a “literalist” picture of
ontological discourse; disbelievers will appeal to an opposite – “non-literalist” –
account14. One can regard, then, the shift to the “semantical plane” not only as a
dialectical move within first-order ontological debates, but also as a dialectical move
within the second-order controversy about the nature of ontological debates.
Interestingly, one might even distinguish between restricted and unrestricted forms of
literalism and nonliteralism. For instance, Quine’s view in relation to denoting phrases
would qualify as a “restricted” form of nonliteralism. In spite of this, however,
Quine’s view about ontological commitment seems founded on a form of literalism.
According to Quine, in effect, the literal interpretation of quantifiers should be
considered as the default interpretation when it comes to serious ontological reflection –
at least as long as one cannot replace the ontologically committing talk with an
ontologically innocent paraphrase. Only in this latter case, in fact, the quantification along
with its supposed commitment might count, and excused, – somehow
“retrospectively” – as a mere façon de parler15.
14 The literalist picture may exploit notions such as Lewis’ (1983) “eligibility of the
referent”. See Sider (2007) for its application in metaontology. According to Sider, just as
the world is structured in natural kinds that function as “referential magnets” for the
meaning of our natural kind terms, thus the world “comes equipped with logical joints”
that function as referential magnets for the meaning of our quantifiers. The nonliteralist
picture may make appeal, on the other hand, to fictional talk and “games of
makebelieve”.
It seems, then, that ontological debates can be settled only insofar as the semantic dispute between literalists and nonliteralists has been settled16. I think there is something true about this, but nothing we should worry about. For this does not mean that ontological issues entirely turn on semantic issues. What exists does certainly not depend on the language we speak, but what we can say and how we express disagreement about what exists does indeed depend on the language we speak. Determining our ontological beliefs is, to be sure, a highly complex theoretical endeavor that crucially requires a whole host of considerations about what parts of our discourse convey such beliefs and how they manage to do so, and the merits of such considerations, of course, will be, in turn and somehow inevitably, debatable.