Abstract

The paper concerns the semantics of belief reports. The following definition is proposed: if rational agent x does not accept any sentence, which is true under x’s interpretation of expressions iff not-p, and x does not reject any sentence, which is true under x’s interpretation iff p, then (x believes that p iff x accepts  which is true under x’s interpretation iff p). The term ‘x’s interpretation of an expression e’ is to be understood: something which x actually refers to by using e (rather than a object she intends to refer). This is to say that to ascribe a belief to rational agent one should use words which have the same reference as words used by the agent. We believe it could solve Krpike’s puzzles about belief on the price that sometimes names used by the agent refer to others entities than when we use them.