Let Me Tell You What You Believe: The Semantics of Belief Reports

Michael Nelson

The following theses seem independently plausible.

(1) The sentence S believes that a is an F , where S designates some agent, a is a syntactically simple singular term, and F is a simple predicate, is true (relative to some context C and with respect to some world w) just in case the designation of S (relative to C) stands in the belief relation at w to the proposition expressed by the sentence a is an F (relative to C).
(2) The proposition expressed by a is an F relative to C is the singular proposition represented as < , being an F-er>.
(3) Standing in the belief relation to a proposition that p is a matter of grasping and internally affirming the proposition that p.

Call the conjunction of these theses Naive Russellianism. Although independently plausible, these theses together entail that if m and n are syntactically simple, co-referring (relative to C) singular terms, then S believes that m is an F is true (relative to C) iff S believes that n is an F is true (relative to C). But many think that this is obviously false. My aim in this paper is to defend Naive Russellianism, and in particular to attempt to show that this consequence is not unacceptable.

One main line of argument against Naive Russellianism runs as follows. Surely the Joker believes that Bruce Wayne is rich and well-bred; but he hardly thinks that Batman is either rich or well-bred. Just ask, "What do you think of Batman," and he will retort, "Well, given that he lives in a cave, he quite clearly is neither well-bred nor terribly wealthy." Ask him, "What about Bruce Wayne?" He will say, "He is both wealthy and well-bred, as everyone knows." So clearly Naive Russellianism is false.

The above argument relies upon the following principles.

Disquotation: If S is fully competent with respect to a is an F and sincerely and reflectively assents to it (in some context of type C), then S believes the proposition expressed (relative to C) by a is an F .

Rationality: If S is fully rational, then there is no piece of content such that S stands in the belief relation to it and its negation at the same time.

Converse Disquotation: If S is fully competent with respect to a is an F and sincerely and reflectively dissents from it (in some context of type C), then S does not believe the proposition expressed (relative to C) by a is an F .

I argue that a Naive Russellian can legitimate the rejection of both Rationality and Converse disquotation. Largely following Nathan Salmon, I attempt to construct a view of the belief relation which delivers this consequence. I then attempt to answer objections that arise from the denial of these two principles.

The particular problems discussed include: the role propositional attitude ascriptions play in explaining and legitimating actions, inferences and belief revisions; how information about the way the agent (allegedly) believes the proposition in question comes to be conveyed by certain utterances of belief sentences; issues of context-sensitive versus context-invariant accounts of certain linguistic phenomena.