Beliefs Do Not Report Beliefs (More on the Specification Assumption)

Last night I read Kent Bach's very interesting article "Do Belief Reports Report Beliefs?". Kent rejects the specification assumption (SA). Here is Kent's formulation of it:

Specification Assumption: (SA) Belief reports specify belief contents, i.e., to be true a belief report [of the form 'S believes that p'] must specify a proposition [that p] the person believes.

Kent's reason for rejecting (SA) is different from Delia Graff's. Kent is motivated by Frege/Kripke puzzles, like:

(1) Lois Lane disbelieves that Clark Kent can fly
(2) Lois Lane believes that Superman can fly
(3) John disbelieves Paderewski had musical talent.
(4) John believes Paderewski had musical talent.

The puzzle is that it seems that these can all be true, even though 'Clark Kent' and 'Superman' co-refer, and the two occurrences of 'Paderewski' co-refer. John might, for instance, believe that Paderewski, the pianist, had musical talent, but disbelieve that Paderewski, the statesman, had musical talent.

Kent rejects (SA). 'S believes that p' expresses a dyadic relation between S and the proposition that p. But p may not be the exact content of any of S's belief states. 'That p' describes the content; it doesn't specify it. So, the exact content of S's belief may be considerably more specific than the content of 'that p'. Whether 'that p' adequately describes the content of S's belief will depend on the context of the reporter.

I find the view very interesting. I think it might even be right. But there is an apparent puzzle. Consider the following exchange:

Jason: John believes Paderewski had musical talent.
Delia: You are wrong, Jason. John doesn't believe that Paderewski had musical talent.

Jason and Delia seem to disagree. But as a matter of fact, John believes that Paderewski, the pianist, had musical talent, but he doesn't believe that Paderewski, the statesman, had musical talent. So, given Kent's view, we may suppose that they are both saying something true! But now we also know that the following is true:

A proposition p and its negation, not-p, contradict iff p and not-p are evaluated at the same circumstance of evaluation.

Delia and Jason are located at the same circumstance of evaluation (if there are times in the circumstance, let's assume that they are speaking at the same time). So, if Delia denies what Jason affirms, and they are both saying something true, then they do not disagree.

Or if they do disagree, then either Delia is not denying what Jason affirms, or they are not both saying something true.

That's a puzzle.

So is Kent's view in trouble?

I don't think so.

Kent apparently holds that belief reports can express different relations in different contexts of use. Here is a quote:

"[t]he relation in question is not the belief relation. If it were, then Peter would bear the belief relation both to the proposition that Paderewski had musical talent and to the proposition that Paderewski did not have musical talent, in which case he would believe contradictory propositions. The descriptivist theory [Kent's view] specifically denies this. Besides, because the relation in question would vary from one context to another, it could not be the belief relation"

So, it seems to me that what Kent says about 'belief' is quite similar to what some epistemic contextualists say about 'know'. Some epistemic contextualists (Keith DeRose, I think) say that 'know' expresses different knowledge relations depending on the epistemic standards that are salient in the context. I take Kent to be saying that 'belief' expresses different relations depending on the standards for describing belief that are salient in the context.

Delia and Jason, to the extent that their claims express propositions, do not really disagree. They express different propositions, but, and this is the cool part: 'S believes that p' still ascribes a dyadic relation!

I think this view is really neat.