Abstract

I hold that a belief report characterizes the subject's belief not only by its truth conditions, but also by the token mental representations involved in it (based on conversational hints). To what extent does a belief report specify the mental representations required to make it true? I advance two surprising theses: (i) many reports specify representations by actually referring to them, and (ii) it is not clear that any ordinary reports simply leave open what sorts of representations are required for their truth. I defend these claims from the arguments of Marga Reimer and other critics.