Thursday, January 20, 2005


One of the perils of a mixed philosophy/politics blog is that if I write about one thing too long, I probably lose all my readers who do the other thing. But today we talked about Kripke's Wittgenstein on Rules and Private Language in the Gibbard "Normativity of Meaning" seminar and I want to talk about that with my ~3 remaining philosophical readers.

I don't think the skeptical problem of the first half of the book -- What makes it true that some agent's use of the '+' symbol represents addition rather than quaddition, where you add if the numbers are below 57 and otherwise the answer is 5? -- is a particularly big problem. My two points of contention with Kripke are these:
(1) He neglects the possibility of giving a synthetic reduction of the normativity of meaning.
(2) His objections to an idealized dispositional theory aren't good.

(1) is pretty simple. On page 37, he talks about how the dispositionalist will give a descriptive account of the addition relation. He says that "this is not the proper account of the relation, which is normative, not descriptive." But as naturalistic moral realists using Kripke and Putnam's own theory of reference are aware, normative and non-normative concepts can have the same reduction base. Just as water and H2O (assuming the Putnam intuition which I don't have, but whatever) are concepts that pick out the same property, normative and non-normative concepts can pick out the same property. If you see the dispositionalist as trying to run a synthetic reduction of this sort, the normative/descriptive contrast doesn't have any great significance.

If (2) is right, a dispositional theory can even work as a conceptual analysis, and not merely as a synthetic reduction. I think that one can solve the skeptical problem with an idealized dispositional theory -- you consider a counterfactual situation where people have perfect calculative power, flawless memory, and long enough lifespans to hear the longest numbers for sums. If they would say that 68+57=125 in this situation, they're doing addition, and if they say 5 it's quaddition. Kripke claims that an idealized version of the dispositional theory fails because any idealization sufficient to get addition as a result is so gerrymandered as to be question-begging. The account must have been constructed with the idea of getting addition and not quaddition to be the answer. I think this is far too quick. If we can isolate the set of psychological processes having to do with meaning, we can fix everything else so that those psychological processes will be the ones expressed in behavior. Whether one is an adder or a quadder will depend on what one would do in this counterfactual situation. The question hasn't been begged here -- if the psychological processes having to do with meaning are the right way, the subject will be quadding.


Justin said...

I haven't read Kripke's book in awhile (not since my 1st year of grad school), but I found it pretty gripping when I did. I'm probably ultimately sympathetic to the dispositionalist response, but...

Regarding (1): to establish a "synthetic reduction", it's crucial that you be able to explain away certain intuitions as to what's possible on the grounds they're misreported. E.g., suppose you have the intuition that water could have been XYZ. I say no way -- you don't really have this intuition. What you have is the intuition that the watery stuff around here could've been XYZ and that we could've called this stuff "water". This IS a real possibility, but you misdescribe it when you say water could've been XYZ. Water couldn't have been XYZ; it's necessarily H20.

How does the analogy go with the dispositionalist reduction of meaning? (I was originally going to say what I think it is, but I have to go so I'll just leave it as a question and return to the matter in a second post.)

Neil Sinhababu said...

I don't think I can answer that question in the abstract. If you give me a specific case where the idealized dispositional theory seems counterintuitive, I can make moves to explain away intuitions in that case. Kripke, as far as I recall, gives no such cases.

What's so intuitive about the idealized dispositional theory? Well, in cases where the agent doesn't live long enough to hear the entire sum, or where there's some failure in memory, or some failure of calculation, this doesn't affect our intuition that the agent still means the same thing. So we can idealize away from mortality, faulty memory, and calculative failure. This isn't to rule out the possibility of a counterexample, but to shift the burden of proof a little, so that the dispositionalist's opponent now is charged with producing a counterexample to show where our intuitions go against the theory.

Justin said...


Sorry I took so long to write again. Writing what I was originally going to here would take up too much space, so let me just say this about (1): it's not obvious that you get Kripke style synthetic reductions for normative properties, or for various other properties. Suppose I think that '+' means quus, and to justify this I say the following.

Look, if we punched you in the brain just this hard and asked you what '+' means, you would say it means quadition rather than addition. This is a certain disposition of yours. Unlike the disposition you discuss which involves perfect intelligence, THIS disposition is the one that matters to meaning. I know it's not a priori evident that since you have this disposition, when someone asks you what 58+67=, you should say 5. But I'm postulating a synthetic, a posteriori reduction of the relevant normative property involved to the disposition I mentioned.

This would be completely unsatisfactory. Analogously, it may be that the type of synthetic reduction you describe is possible, but before this becomes a real objection here some work needs to be shown regarding how it would go. I'm inclined to think that what's really needed is a kind of a priori reduction of the normativity of meaning to the sort of idealized disposition you describe -- if such a reduction can be carried out at all, I suspect it will need to be done a priori.

Maybe that's Ok b/c maybe it can be done a priori.

I'll say something about 2 tomorrow.

Neil Sinhababu said...

I think I understand the problem now. Let's see if this is it:

Purely normative concepts don't have descriptive content. That's what it means for them to be purely normative concepts. But to do a synthetic reduction of a concept, it needs to have some descriptive content so you can locate in the world the things that fall under it. Then you might be able to determine that all the things that fall under it, and only those things, also fall under some lower-level concept. This is how you get synthetic reduction. Since you can't do this with purely normative concepts, they aren't apt for synthetic reduction.

There is another way to locate a normative concept in the world -- consider lots of situations, have moral intuitions about them, and use this to determine where the normative properties are. I guess the trouble here is that this doesn't seem to give you a synthetic reduction. If you find out, by considering counterfactuals, that in all possible worlds where something is morally good it maximizes pleasure, you've got a conceptual analysis going and not a synthetic reduction.

This would explain why people like the Cornell realists who want synthetic reductions think that there are laws in which moral concepts irreducibly figure. These laws keep normative concepts from being purely normative, and allow us to find some descriptive content to place them in the world.

Anonymous said...

I've probably just misunderstood what you've said, but surely just because normative and non-normative concepts can pick out the same things, that doesn't mean that they do? 'It is what I ought to do' is a normative concept, and 'Eating babies causes pain' is a non-normative concept, but you'd be pretty surprised if someone tried to tell you they point to the same thing.

Anonymous said...

PS ...I suppose what I'm trying to say is that there's no evidence for why the descriptive conditions for meaning should be the same as the normative conditions.