February 06, 2004

Se Se Se

[UPDATE: Rewritten from top to bottom.]
I quasi-promised to have something to say about Jason Stanley's remarks on de se desire reports. Briefly: I haven't read the literature in question, but I think Stanley's discussion can be spun to support my views, which are:

(i) There's no such thing as a de re attitude. Everything we think about, we think about some way. If someone thinks about an object in a certain way w, the best way to express that attitude is with a sentence in something like the language of thought, including a term for w (ii) It often makes sense to ascribe attitudes de re. An ascription, "A attitudes that x phis," de re for x, will be true iff A attitudes some sentence "y phis" such that y and x co-refer (as used by A and the ascriber, respectively)--possibly with a contextual restriction on the range of allowable x.

[These views may not be explicit, or even implicit, in the linked post.]

Now, Jason points out that

(1) Brian wants to win the race

always expresses a de se desire, and that

(2) Brian wants himself to win the race

is almost always de se, but that we can contrive a case in which it's not; Brian is running a race and looking at the race on the Jumbotron, and he thinks "I want the red-headed runner to win," not realizing that he is the red-headed runner on the Jumbotron. (An example like this is due to Mark, in Brian's comments.)

This makes sense in terms of (i)-(ii) if we add:

(iii) De se attitudes are just attitudes that contain "I." No difference between saying de dicto "I think that I am cold" and that I have the de se thought that I am cold.

For, in order to make the non-de se reading of (2) come out, we need to find some alternate description under which Brian thinks of himself. His thought simply can't reach out and grab himself in a non-first-personal way; he has to think "I want that guy to win," or "I want the red-head to win," or "I want the guy who looks like [picture] to win," or even "I want the guy that Jason bet on to win."

Each of these can be described in Brian's language of thought--I just did it--and that description seems more fundamental than the description given by "Brian wants himself to win." "Brian wants himself to win" is a de re ascription, grounded on the co-reference of "himself" with the term B that actually shows up in Brian's language of thought (if you will). (2) is acceptable as an ascription of Brian's desire if we're in a position where the salva reference substitution of "himself" for B is allowed. (I'm being agnostic as to whether an impermissible substitution would falsify (2)--thus making (2)'s truth value context-dependent--or whether it would make (2) true but misleading.)

But (2) is guaranteed true if Brian's original desire was "I want to win." And if (1) always reports the desire "I want to win," this explains why the most natural reading of (2) is as equivalent to (1).

I think my view is equivalent to the view that Jason mentions on which (1) expresses a relation between Brian and a property. On my view, (1) has to express a desire of Brian's "I want to..."; the ... is the property to which Brian is being related.

Posted by Matt Weiner at February 6, 2004 01:04 PM