Abstract

Conflicting desire ascriptions—of the form ⌜S wants p⌝ and ⌜S wants ¬p⌝—falsify every extant semantics that takes the orthodox, and theoretically powerful, belief-based approach to desire ascriptions: these semantics often wrongly predict that conflicting desire ascriptions can't both be true. In response, I propose a semantics of some-things-considered desire, which I model with questions, lending further evidence to the view that questions are central in our cognitive lives.