Saturday, September 30, 2006

Another false View

I talked about a view in my last post that is certainly false. Here is another view that I have been thinking about, which is also certainly false. Let's suppose, as per the true, that counterpart theory is false and that individuals are in multiple possible worlds. This seems to cause problems for Lewis' account of truth in fiction because his account best fits with counterpart theory and descriptivisim about names.

Here is a view though. Suppose that 'Alyosha' is in fact empty. But there are worlds in which people speak something very much like english except that when they utter the sentences of The Brothers Karamozov they express truths and the symbol 'Alyosha' is a name that refers. However, there are lots of such worlds. There are worlds where 'Alyosha' is used as a name for one guy and worlds where it is used as a name for some other guy. Given transworld identity, it looks like we cannot appeal to some of these worlds when giving an account of truth in fiction without being arbitrary. We might appeal to them all and supervaluate in some sense. But another thing that we can do is pretend that any the individual, x, such that the symbol 'Alyosha' is used as a name for x in one world and any individual, y, such that the symbol 'Alyosha' is used as a name for y in another world are identical.

I know this theory is not worked out. But, I am wondering about necessary connections that might be noted between things that people commonly take to be true in a fiction and what various accounts say would be true in a fiction.

Names in Fiction

Let's assume that 'Alyosha' is an empty name that appears in "The Brothers Karamozov". One might think that if 'Alyosha' is in fact empty, then it is necessarily empty. Here are two reasons to think so.

First, one might be convinced by Williamson style arguments for the conclusion that necessarily everything necessarily exists. If that is the case, then if 'Alyosha' possibly refers and if names are regid designators, then 'Alyosha' in fact refers. But of course names are rigid designators (in some sense). So, if 'Alyosha' possibly refers, then 'Alyosha' in fact refers. Equivalently, if 'Alyosha' does not in fact refer, then 'Alyosha' does not possibly refer. Equivelanly, if 'Alyosha' is in fact empty, then it is necessarily empty.

I want to set aside Williamson style arguments and assume that some things contingently exist. I hope everyone can play along with this.

A second reason to believe in fact empty names are necessarily empty is the following: if 'Alyosha' is in fact empty but it is possible that 'Alyosha' is not empty, then the actual world 'Alyosha' is a different name than the 'Alyosha' of the other possible world. This is a widespread belief in semantics. But I want to consider an alternative view.

Consider the following case. One that I do not buy, but I know some of you will like it. Suppose that sperm S and egg E are not in fact united. If that is the case, then given the necessity of origens and the denial of Williamsonian conclusions, the thing that would have resulted from their uniting does not in fact exist. However, we might introduce a name, 'Fred', by saying "I hereby name whatever would have resulted from the union of S with E 'Fred'" Here it looks like Fred is in fact empty yet it possibly refers.

If we accepted this case, then we would have to deny that for any symbol type N, if N is in fact an empty name but it is possible that N is not an empty name, then the N as used in the actual world is a different name than the N of the other possible world. Thus we have undermined our second reason for believing that 'Alyosha' is necessarily empty.

However, here is a response that someone might have if he wants defend the view that 'Alyosha' is necessarily empty. He might admit that there are strange situations like that involving the name 'Fred' where we have a name that is in fact empty yet possibly non-empty. However, he might say that the situation with 'Alyosha' is not the same. We did not introduce the name 'Alyosha' with a description that happened to pick out a particular merely possible object. Moreover, we can only get names like 'Fred' by invoking such a description. So, if 'Alyosha' is in fact empty, then it is necessarily empty.

This much, some of you might recongnize, is roughly the content of the exchange that David and I had last Wednesday. But now I want to add something for the side of the wierdos. Suppose we said that when an emtpy name is introduced and there is no discription that uniquely picks out a possible referent, then the name is super vague or multiply ambiguous. I am thinking that we might endorse a kind of meaning pluralism according to which sentences involving the name 'Alyosha' express multiple propositions. One proposition for every merely possible individual who satisfies the conditions laid out in the story. In this case, we might say that 'Alyosha' is in fact many empty names and there are many stories expressed by 'The Brothers Karamozov'. Moreover, we could then adopt something like Lewis' strategy for truth in fiction without endorsing counterpart theory.

Okay, I know that this is a false view. But I thought that I would put it out there anyway.

Wednesday, September 27, 2006

Gunk in Discrete Space

So suppose that space is discrete at w. Does this rule out the possibility that there is gunk in w? It would surely be weird to suppose there were--for then it seems there would be objects that exactly occupy a region though no proper parts of those objects exactly occupy any region. Though this is weird, it does not seem contradictory or incoherent. So is gunk compatible with discrete space?

Saturday, September 23, 2006

A Modest Proposal

I wish to confess attraction to a version of Feldman's analysis of knowledge in Epistemology. According to the analysis,

S knows p iff:
(i) S believes p,
(ii) p is true,
(iii) S is adequately justified in believing p, and
(iv) S's justification for p does not essentially depend on any falsehood.

I say it is a version of Feldman's analysis, because (iv) is explained as follows:

(J) S's justification for p does not essentially depend on any falsehood iff the body of evidence that supports p for S is such that it's not the case that deleting any false proposition from that body of evidence results in S's not being justified in believing p. (It is "of the essense" of S's justification for p that it depends on falsehood if the right-hand side of (J) obtains.)

Feldman does not accept (J). I believe he thinks that in order for the account to be plausible, it requires not only (J) but some sort of "no defeater" clause. I deny that there are knowledge defeaters, however. So I am interested in the viability of an account of knowledge that includes only (i-iv), where (iv) is analyzed in terms of (J).

This account yields some perhaps surprising verdicts.

In Gettier cases, the target proposition is not known since deleting a falsehood from the knower's body of evidence destroys justification.

In a Feldman-Gettier case, where all propositions explicitly reasoned through are true, there is still a falsehood in S's grounds such that its deletion from S's evidence result in a lack of justification. So in Feldman-Gettier Nogot/Havit, S infers directly from S's evidence regarding Nogot that someone in S's office who has Ford papers, drives a Ford, etc., and from this S infers that someone in the office owns a Ford. But here, the original belief is arguably "Gettiered"; deletion of all falsehoods from S's evidence base for the proposition that someone in the office has Ford papers, etc., results in loss of justification for that proposition.

Russell's stopped clock case is treated differently. S knows what time it is if it is time t, the clock says it is t, it seems to S as if this is so, and S has some sort of belief/evidence for the general reliability of clocks. That there is some accident involved is irrelevant; accident is involved in any case of knowledge. (Duncan Pritchard has a mountain of paper on epistemic luck. I think this issue would be worth checking out but I am not familiar with the details of Pritchard's proposals.)

Perhaps more interestingly, the view can be employed to explain how testimony can be a generative source of knowledge, rather than serving merely to transmit it. For some reason or other, A may not know p, but A may testify that p to B, and B may come to know p on the basis of A's testimony.

Perhaps even more interestingly, this suggests a problem for epistemicism about vagueness. Suppose a philosophical neophyte, suddenly made interested in the problem of vagueness by Andy's tupperware dish full of rubble and a bit of classical logic, travels the world consulting vagueness experts on the sharp cutoff for heapness. Each expert, after carefully reflecting on the question, tells the neophyte that the cutoff is 4; any number of grains less than that a heap does not make. It's plausible to suppose that the experts do not know their own answers (insert whatever motivation you like for their asserting it here). But suppose they're right. Now the neophyte has a justified true belief where his evidence is the wealth of expert testimony. And by the analysis, the neophyte knows the cutoff for heapness. The problem for epistemicism is clear; epistemicists hold that the cutoffs are unknowable in cases of vagueness. (I'm blurring use-mention stuff here, but the problem is hopefully clear enough.)

I think this view of knowledge comports well with the standard view and is not liable to any significant problems that I am aware of. So would someone please make me aware of some significant problems?

Thursday, September 14, 2006

Kant's Cleavage

(Lots of fools apparently assess the cascade of craptasticity in "Two Dogmas" and conclude that there is no analytic-synthetic distinction.

That conclusion is unwarranted. But there are lots of features that analytic sentences were traditionally supposed to have and I doubt, for reasons independent of Quine, that they have any of them. So while I think there are analytic sentences, of course, I wonder if there is any philosophical usefulness to the distinction if analytic sentences don't in fact have the features they were traditionally supposed to have. That's my question for you guys.

So here are some of the features:
1. Analytic sentences are necessarily true.
False. With respect to an appropriate context, a usage of 'that student is a student' is intuitively analytic, though it is fobviously contingent (and a posteriori).

2. Analytic sentences encode propositions that are knowable a priori.
False. See (1).

3. Analytic sentences are true in virtue of meaning.
Wtf. 'That table is brown' is true because its meaning, the proposition that that table is brown, is true--that table is brown. Similarly, 'all bachelors are unmarried males' is true because its meaning, the proposition that all bachelors are unmarried males, is true--all bachelors are that way. There is no coherent distinction I know of between truth in virtue of meaning on the one hand and truth in virtue of meaning plus the way the world is on the other. So wtf.

4. Analytic sentences are such that understanding the terms in them and the way they're put together is sufficient for being in a position to know that they're true.
False. Given the panoply of perverse philosophical views on the market, it's not hard to counterexample this. Perverse logicians understand logical truths but are not in a position to know them given the perversity of their views. And so on.

What's left? Again, I don't deny that there's a distinction between analytic and synthetic sentences, but the sorts of metaphysical and epistemological claims made about them seem like crapezoids. Is there some other reason to be interested in this notion?