Wolfgang Schwarz

Consider this:

Thought experiments of the kind “what would X refer to if...” do not give us any clue as to what the true semantics of the term is. We do not have stable and interesting intuitions on reference, because “reference” is a technical term.

In the case on “aluminium” it is important to decide if this is a term in a scientific theory of some kind or not. If the term is used in scientific theories, the theory will determine what it means that the term is in fact empty.

“Aluminium” in ordinary natural language might be a completely different story. Reference and meaning are not the scientific ones (the same applies of course to “water”), so procedures should and would differ. What these procedures and the meaning of the term is, is far more difficult to determine. What can not be the case is a kind of descriptivism that makes it altogether very unlikely that there isn’t any aluminium at all – “phlogiston”-type examples show that this can not be the true.

With reference of names in counterfactual situation, Kripke seems to rely strongly on semantic intuitions: “what we would say, what scholars would say” (s. the example with “Columbus” in N&N, “Moses”, “Jonah”). Intuitions are not as clear cut as he takes them to be, compare for instance the (methodologically speaking weak but interesting) work by Stich & Machery, Semantics, Cross Cultural Style that investigates into differences between cultures on “semantic intuitions”.

You are of course right, “no reference”-cases as discussed e.g. in Perry, 2002 are one of the main problems for causal theories.

M.

Well, if you use "reference" as a technical term in semantics (or ornithology, for that matter), then intuitions about reference may not be relevant to what you call "reference". But I was talking about our ordinary conception of reference. Or rather, I meant to be talking about how we are disposed to describe various counteractual situations; whether, for example, we would say that Moses didn't exist if we found out that such-and-such. These dispositions I take to be semantically important, even though I wouldn't speak of "reference" in semantics at all. (I am happy with intensions alone, and it would be silly to claim that our terms "refer to" their intension.)

I agree that our intuitions are often not as clear cut as Kripke assumes. If scientists or people in other cultures have different intuitions of the relevant kind about some terms, I'd say they don't mean (exactly) the same by these terms as we do. Certainly other people *could* mean something slightly different with a term. To what extent this actually happens I don't know. I don't think the research by Stich et al proves much in this respect.

I should have mentioned that I wasn't arguing against 'theories of direct reference', and I didn't want to point out that they have problems with empty terms. I only wanted to point out that philosophers (we descriptivists in particular) should pay attention to the conditions under which we would say that there is no x. This will show us e.g. that we should not put too much weight on causal conditions.

(Something seems to be wrong with the special characters in your comment, I'll look into that when I come home.)

the "special characters" were ment to be Anf?hrungszeichen, I copied the text from Word into the comment box.

M.

Ah thanks, I was going to ask you that. Not sure how I should handle entities. For now I've just unescaped them in the comment.

Simply claiming that empirical results do not prove much is no rebuttal--in no way.

The facts remain: with a probe that was directly drawn from Kripk's N&N, we obtained some statistically significant differences between groups. We had predicted these differences. (Notice that we are not claiming that this is the last empirical word on the issue--mnore work is needed to see whether our results generalize. But this is evidence nonetheles-sufficient for one of the top journals in psychology...).

Moreover, we have shown that with the 2 types of probes under consideration, there is considerable intra-cultural variability. Thus, I am curious to know who is "we" referred to by the expression "our" intuition?

The move "Different meanings/concepts" is common in reply to our work (and other cross-cultural work). It strikes me as more problematic than is usually thought. Suppose that most citizens of Hong-Kong have a descriptivist theory of the reference of mentalist terms like "beliefs", while most Americans have a referentialist. Suppose that neuroscience shows that our folk theory of mind is massivbely erroneous. The claim "There are no beliefs" would then be true for Hong-Kongers, but false for Americans. Thus, there are no beliefs for Hong-Kongers, while there are beliefs for Americans. Now, the question is: ok, but do beliefs exist?

You may reply: it depends on what you mean by "belief" when you ask the question.

But you do not really want to do this. For this shows that most metaphysical/ontological debates are merely terminological--mere linguistic disagreement in the absence of any disagreement about the facts.

I didn't say that empirical results in general do not prove much. I'm just not very convinced by the setup of the particular tests you carried out.

If people in Hong-Kong are disposed to use "belief" for whatever satisfies a certain role in some speculative theory of mind, whereas Americans are disposed to use "belief" for whatever stands in a certain causal relation to their word "belief" -- which I think is the situation you describe --, then it seems clear to me that the term has a very different meaning in the two communities. E.g., for people in Hong Kong it would be analytically true that if there are any beliefs, they satisfy the speculative theory of mind, whereas for Americans this would be a substantial hypothesis.

(I would prefer not to speak of 'descriptivist' versus 'referentialist' theories here. For one, the causal theory of reference is best seen as a special case of modern-day descriptivism. More importantly, the interesting question I believe is just under what conditions people are disposed to apply which terms to which objects. To test this, one would have to present subjects with lots of different scenarios and ask how they would describe them on the assumption that they actually obtain. I am certain that *nobody* uses "belief" in the way a purely causal theory would predict -- e.g. applying it to rabbits on the assumption that our term "belief" was once introduced as a name for rabbits --, and nobody uses it in the way predicted by a simple description theory in which only purely qualitative, non-indexical conditions are allowed in the descriptions. The actual application conditions are somewhere in between, and probably quite indeterminate and context-dependent.)

The difference between factual and verbal disagreement in metaphysics is subtle, and tricky: do philosophers who claim that only material objects have parts mean the same by "part" as philosophers who believe that events and sets and stories and tasks also have parts? Do Lewis and Stalnaker mean the same by "possible world", but disagree about their nature? I don't think a clear distinction can be drawn here.

Not to sound overtly naive, but if this is so as you described: What then *is* belief reconstructed through observation of the language use of persons with some "indeterminate and context-dependent" application conditions?
Isn' t that really vexing?

As for the point with Machery, as I faintly recall but didn't bother to look up, the study is on proper names only (where the object of reference is - at least for philosophers - simply given, which is not the case with general terms). Turning to "belief" is simply changing the subject. I meant to make a point mentioning this data on semantic intuitions.

M.

1/ The study is only on PNs indeed. Nonetheless, it is reasonable to expect intuitions about natural kind terms to behave similarly. Indeed, among philosophers, the 2 types of words are supposed to trigger the same type of intuitions (Putnam's cat thought-exp is formally similar to the Godel case). And it is often believed that their theories of reference are similar.
So this is a reasonable hypothesis. We hope to get some data about this--but real experimental work is very very very slow.

2/ are referentialist theories a type of descriptivism? It is possible to formulate a descriptivist theory (causal descriptivism) that accounts for all the intuitions that are supposed to support the causal-historical account of reference (Kroon for instance on NKTs).
From this, it does not follow that the causal theory of reference as formulated by Kripke, Putnam and Devitt is a type of descriptivism. This is clearly not the case. According to Kripke's, Putnam's and Devitt's approach, the reference relation is not supposed to depend on the satisfaction of ANY description. You may find this obviously wrong. But this is the idea.

3/ I do not think that you get the dialectic right. The issue is: how do you establish a theory of reference for a type of words T? Answer: by looking at people's intuitions about the reference of words of type T. The correct theory of reference is the one that captures people's intuitions about the reference of T words in possible and actual situations.
But then, what if people have different intuitions about reference? Say, if 2 cultural groups (Chinese and American, say) tend to have different intuitions about the same cases.
You then face a dilemma: you can claim that words of a given type (T) refers in one way in one culture, in another way in another culture; or you can claim that theories of reference for T do not depend on intuitions.
The second one is a dead end, and the first one has plenty of annoying consequences. (The different meaning reply suggests that you want to endorse the first horn.) For instance, you cannot disagree with Chinese speakers of English. Moreover, there is a substantial of intra-group variation and there may be some variation within the same individual across times. And certainly you want to be able to disagree with yourself.
Thus, the issue is not about what "belief" (or any particular word) means, but about what are the consequences of a cross-cultural diversity in intuitions for the project of developing a theory of reference. If our results stand and generalize from PNs to NKTs (this is a big IF-we are aware), then what are the consequences for the understanding how PNs and NKTs refer? The argument bears on all PNs and on all NKTs.

4/ I want to emphasize the fact that we do NOT believe that this is the last empirical word on this question. We agree that more work is needed.
But we find it quite amazing when philosophers who didn't do any empirical work and who would not have predicted our results, tell us that this is not the right type of data and that they are quite sure that with the right probes we would have found the *good* results--ie the results that fit with their preconceptions. We may be wrong, but we are the only ones to have any serious empirical data to lay on the table.
And, I insist, we used Kripke's own thought-experiments. No double standards please: if this was good enough when Kripke wrote the text, that should be good enough for us.