Tuesday, December 19, 2006

MM Sider (and Bennett): Whether 'exist' admits of precisifications

I would like to object to Sider’s claim (with which Dan appears to agree) that “questions of fundamental ontology … are not susceptible to the non-fact-of-the-matter argument since there are no multiple candidates for ‘exists’ to mean” (206); the standpoint from which I make the objection has some connections with the Carnapian view of existence, but they also differ significantly. I would object on a similar basis to Sider’s argument for 4-D in “Against Vague Existence,” Philosophical Studies 114, 142-144.

Sider assumes that there is a non-arbitrary meaning for the logical expression ‘there is’ or its formal counterpart ‘∃’ – non-arbitrary unlike the alleged meaning he envisages (204) on which it has to do with what Nelson Goodman says – such that it ranges “unrestrictedly over absolute everything, except perhaps non-“concrete” things” (204). He also thinks that non-arbitrary meanings such as this for the more basic logical constants are determined in part by eligibility, given the “existence of logical joints in reality” (205). I am happy to grant all of this (although later I will come back to discuss to what extent one should grant the latter assumption).

We have the practice of, and the freedom for, ascribing new “temporary” meanings to expressions with specific literal meanings, under certain constraints. This is what we do when we convey implicatures, and (perhaps just a particular case of the former) when we produce new metaphors. We apply this freedom to expressions whose literal meaning is fundamentally determined by eligibility, such as natural kind terms. Thus, perhaps ‘to swallow’ has a basic meaning such that it refers to natural events; but nothing stands in our way to use it “temporarily” for what ATMs sometimes do with our credit cards. I see no reason why we cannot do the same with more “abstract” or “formal” expressions, like the apparatus of reference – the quantifiers, the referential expressions which may occupy the “positions” in logical form “occupied” by the variables they bind, the identity sign. To provide a philosophical account of how this is implemented, we would need at the very least an account of how those “temporary” meanings are created, and one of what exactly is the literal meaning of the referential apparatus (even if Sider is right about the role of eligibility considerations, there might well be more to it, such as its cognitive and inferential role). But we do not need such accounts to accept the possibility of extending the practice to such a case.

A plausible case in which this applies to the apparatus of reference, granting Sider’s assumptions, is in my view that of explicit reference to, quantification over, and identification of, fictional characters; given a sufficiently elaborated philosophical account of the kind envisaged in the previous paragraph, we could develop along these lines a figurative view of fictional characters, close, I think, to views put forward by Stephen Yablo for abstract entities.

Now, a complication for Sider’s claim results from the fact that “temporary” meanings such as the metaphorical meaning for ‘to swallow’ turn easily into standard, conventional meanings. It is no easy matter to contend that afterwards they are still not “literal”, even if the meanings are related (so that the resulting ambiguity is not like the one in ‘bank’). Obviously, the figurativist about fictional characters I have envisaged would acknowledge that the “extended use” of the referential apparatus he posits is a fully standardized one.

Let us now apply this to Sider’s discussion. Here are three different candidates for ‘exists’/’there is’ to mean, granting his assumptions: (i) the nihilist is right about an “ontologically fundamental” core given solely by the eligibility considerations Sider mentions, and a figurative account along the sketched lines applies to the use of the referential apparatus that both the chaste endurantist and the friend of 4-D invoke. (ii) it is the chaste endurantist who is right about the core. (iii) it is the friend of 4-D who is right. Now, one could think that the availability of these candidates does not contradict Sider’s claims, such as those quoted at the beginning, because it will still be the case that “eligibility” considerations as a matter of fact select one of the candidates, no matter what it is and independently of whether or not we can come to know what their verdict is. But I think this would be to quick; it is here that I need to go back to this assumption I said before I was granting to Sider.

Eligibility considerations, as pointed out by Sider, have to do with sensible replies to Putnam’s “model-theoretic” argument, and similar arguments with an anti-realist basis for the indeterminacy of reference, translation or meaning. They come to the fact that causal-explanatory relations between language use and the objective extra-linguistic world by themselves contribute to determine meaning, independently of whether or not they are part of our linguistically stated beliefs. But they should be applied with care, and holistically. To go back to the analogy I have been repeatedly using, perhaps they can be invoked to argue that expressions such as ‘to put’, ‘to walk’, ‘to swallow’, ‘on’, ‘in’, etc., have, among their equally standard/conventional meanings, an “ontologically fundamental” core (the “more physical” meanings) distinct from more “figurative” ones; although it is not clear how exactly they do.

Now, as I said I am prepared to grant that the objective world comes equipped with “logical” or “formal” features, contributing as part of the global eligibility considerations to determining the correct semantics for a given language or representational system. But given the unclarity concerning their nature, and even more about how exactly they intervene in the global eligibility considerations, I do not find at all out of the question that neither facts about linguistic use, not facts about eligibility properly understood in this global way, give any verdict on the three meanings envisaged before for ‘exists’. In any case, I think the model I have sketched provides a conceptually coherent possibility that suffices by itself to refute Sider’s claims quoted at the beginning, rejecting a “dismissive” attitude (to use Bennett’s term) towards questions of fundamental ontology, in Sider’s own terms.

This discussion has, I think, consequences for the Bennett’s paper. I guess the form of dismissivism I envisage here is, in her taxonomy, a form of semanticism; although the view that there is no fact of the matter whether, say, the 4-D position is just a legitimate figurative extension of the referential apparatus, or is rather rendered correct by pure eligibility considerations applies to the natural kind existence, sounds also close to her “anti-realism”. I share with others that posted before on this blog the concerns about her objections to the analyticity of existentially committing claims. My main concern is whether there is room for a distinction between real believers and hermeneutic nihilists.


Dan López de Sa said...

Very interesting post!

My main worry is, I think, somehow similar to one Sider himself expresses. For your three views (i)-(iii) being (determinately) different requires, I take it, that it is true that there is something that exists according to some but not the others.

Let’s focus on (ii) and (iii), and let ‘exists-(ii),’ ‘exists-(iii)’ and cognates stand for the candidates. Now candidates (ii) and (iii) being (determinately) different requires, I take it, either that it is (determinately) true (a) that there is something-(ii) that is not a thing-(iii) or that it is (determinately) true (b) that there is something-(iii) that is not a thing-(ii). Hence what is required to be true by (ii) and (iii) being different in turn crucially involves (unrestricted) quantification: (a) that there is something-(ii) that is not a thing-(iii), (b) that there is something-(iii) that is not a thing-(ii). (a) or (b) being (determinately) true thus require that they be true according to all admissible precisifications of the quantifier, i.e. they be true according to both (ii) and (iii). But this is not so. (a) is not (determinately) true, as ‘there is-(iii) something-(ii) that is not a thing-(iii)’ is false; and (b) is not (determinately) true either, as ‘there is-(ii) something-(iii) that is not a thing-(ii)’ is false. (I take it that Sider would probably agree with this, I offer my own elaboration here.)

I am afraid I do not quite see how the “fictional” strategy would block this kind of argument. As Williamson 2003 says: “If there are fictional Xs, then they are things too, whether or not they are Xs, and whether or not they are real Xs (if that is a different question). But that does not commit me to the claim that if, according to a story, there is a golden mountain, then there is a golden mountain, any more than it commits me to the claim that if, according to a story, the streets of London were paved with gold, then the streets of London were paved with gold.” (p. 421) In any of the cases, it seems to me, the previous kind of argument would go through. (A version of this worry might perhaps then also apply to the “fictionalist” proposal of Rosen & Dorr 2002.)

Sorry if I have missed and/or misconstructed something important!

manolo g-c said...

The argument appears to depend on a peculiar policy for subcripting, allowing subscription in two places of what is supposed to be just one expression, the existential quantifier 'there is something': after 'is' and after 'something'. Can you run it with (a) something-ii is not a thing-iii vs (b) something-iii is not a thing-ii?

In any case, the argument by Sider to which I was objecting appeals not just to the unrestricted character of the existential quantifier, but also to the claim that "(unrestricted) existence is a natural kind" (see his 2003, 143, where it is the third premise in the argument). I cannot see how this line of reply takes that feature of the argument into consideration, and so I cannot see how it is similar to Sider's.

Dan López de Sa said...

The policy for subscripting involved in my argument is, I think, the standard one—assuming that precisifications are (or can be seen for present purposes as) new expressions which precisify the vague ones. Suppose that ‘is bald-1’ and ‘is bald-2’ are the two relevant precisifications for ‘is bald,’ differing just in that someone with 162531 hairs is bald-1 but not bald-2. Hence ‘Everyone who is bald-1 is bald’ is not true, given that, although ‘Every one who is bald-1 is bald-1’ is true, ‘Everyone who is bald-2 is bald-1’ is false. Mutatis mutandis for the case of ‘everything’—with the crucial difference, of course, that the relevant vague expression needs to be used in stating what is required for the candidates to be (determinately) different in the case of ‘everything’ but not in the case of ‘is bald’!

As to your second point, you’re right: my elaboration of the Lewisian argument does not seem to make any (direct) claim about the naturalness of everything, whereas the second of the two arguments in Sider 2003 (§3) does mention this. However, I took your post to be objecting to the idea that ‘everything,’ in virtue of expressing unrestricted quantification, does not admit of precisifications—which is shared both by my elaboration and the first of the two arguments in Sider 2003 (§2).

How would people feel about discussing next here Sider’s even more recent contribution to the debate, forthcoming in the meta-metaphysical volume?

Happy Holydays!

Dan López de Sa said...

Sorry, that was a typo. It should read: "Hence ‘Everyone who is bald-1 is bald’ is not true, given that, although ‘Every one who is bald-1 is bald-1’ is true, ‘Everyone who is bald-1 is bald-2’ is false."

manolo g-c said...

I am still unclear about the subcripting policy that Dan follows in the case of quantifiers.

In any case, the main assumption behind the reply, that "the difference between the alternative candidates is a determinate difference" begs the question at stake, for it requires that there determinately exists something which belongs to one of the precisification for 'exist' but not the other - which is precisely what is at stake. I cannot see why one should grant this assumption.

It is as if one required that, for two precisifications for the English predicate 'bald' to be really different, a determinately bald person should fall under the extension of 'bald' according to one but not the other. Of course, in the case of 'bald', given that the vagueness of 'there is' is not at stake, we can say that two different precisifications do satisfy the requirement that there determinately exist something that falls under the extension of 'bald' according to one but not the other precisification; but what we cannot make is the relevantly analogous assumption for the case that concern us, mentioned before.

Dan López de Sa said...

Hmm... Charges of “begging the question” are often tricky, or so they seem to me.

True, the main views about the nature of vagueness have it that the candidates differ determinately. This in turn “requires,” in the sense of suffices for, something which is incompatible with one of the views on what was at stake.

Hence, one would have thought, we have a nice argument against such a view, which assumes independently plausible (although admittedly controversial) views on the nature of vagueness. I can only see this to "beg the question" in a sense in which, for not "begging" it, an argument should leave the question open :-)!

manolo g-c said...

The first sentence in the second paragraph, with its 'true' and 'the main views', is unargued for rhetoric; if I am right, one of the main contenders, vagueness as semantic indecision extended to unrestricted quantifiers does not "have it" that, in general, the candidates differ determinately.

It is "unargued for" because it does not address the consideration in the final paragraph in the previous post, which was intended to object to the assumption behind that now rhetorically emphasized claim that one's metalanguage includes precise notions corresponding to the object-language vague ones whose vagueness one is characterizing.

Dan López de Sa said...

I have to confess I am a bit lost at this point.

I take that it is common ground that “the main views” “have it” that candidates differ determinately for most vague expressions other than unrestricted quantifiers. And I took it that this was enough as to ground my previous more general claim you refer to. (Of course, this leaves open that there might be views that make the vagueness of unrestricted quantifiers somehow different in kind from that of most other vague expressions. See the discussion in Sider 2003 §2, I also mention this possibility at pp. 403-4.)

Besides, I aimed to have addressed the consideration you also mention. As I understood it, the consideration makes explicit how it is that “a view” that has it that (in general) candidates differ determinately “requires,” in the sense of suffices for, something which is incompatible with one of the views on what was at stake. And this was why you think that there was a question begged, no?