Most recent semantic theories about attitude ascriptions(AAs) are contextualist in that they hold that AAs are context sensitive: AAs are associated with different truth conditions in different contexts.
One quite articulated version of contextualism about AAs is Mark Richard's theory, which very roughly says that an AA of the form 't V that S' (S a declarative sentence, V an attitude verb, t a term) is true with respect to a context c and a possible world w iff in w the extension of t w.r.t. (c, w) accepts a (sentence-like) representation that 'translates' the representation determined by S in c, where the translation satisfies the restrictions on translation operative in c. Since a context's restrictions on translation are determined by the intentions of speakers and hearers of the context it follows that according to Richard AAs are context sensitive.
Now, this semantics doesn‘t seem to accommodate our disagreement intuitions: Suppose Susan discusses the question whether Hammurabi believed that Hesperus = Phosphorus (Hammurabi was a legendary Babylonian astronomer, who supposedly didn't assent to (the Babylonian translation of) 'Hesperus = Phosphorus' but who did assent to 'Hesperus = Hesperus'). Suppose further that the conversation's participants are interested in Ham's de dicto beliefs, so that what is at issue is (roughly speaking) whether Ham believed something that is expressed by our uses of 'Hesperus = Phosphorus'. Suppose finally that Susan is a Russellian about AAs and so at some stage of the conversation sincerly, after reflection etc. utters
- Hammurabi believed that Hesperus = Phosphorus.
Let Nat be involved in a different conversation of exactly the same kind as Susan's. But Nat, let's assume, is a die-hard Fregean about AAs and so at some stage of the conversation utters
- It is not the case that Hammurabi believed that Hesperus = Phosphorus.
Intuitively, Susan and Nat disagree; after all, there appears to be some sentence S such that Susan believes what S expresses, and Nat believes what S’s negation expresses. So, they believe things that, all dialetheist joking aside, cannot be true together. However, according to Richard’s semantics that need not be the case, because in Susan’s context ‘Hammurabi believed that Hesperus = Phosphorus’ could express something different from what the same sentence expresses in Nat’s context.
To circumvent this difficulty it is tempting to adopt a version of relativism about AAs: Let an AA have the same Richardian truth-conditions in every context; but instead of conceiving of truth-conditions simply in terms of functions from possible worlds to truth-values, let them be functions from pairs of worlds and restrictions on translations to truth-values.
Natural fleshings out of this idea solve the above problem from disagreement. But this idea amounts to alethic relativism: The truth of what an AA expresses in a context depends on more than just possible worlds. Now, many authors, most recently Cappelen and Hawthorne, claim that relativism about AAs is bound to fail. What do you, gentle readers, think: Are there sound arguments against relativism about AAs?