In his book The Nature of Mathematical Knowledge and in a series of articles, Philip Kitcher attacks the traditional conception of a priori mathematical knowledge. The reliabilism he develops as an alternative situates all our knowledge within a psychological framework. However, in Frege‘s Epistemology he claims that Frege‘s conception of a priori knowledge is compatible with a psychological account. Kitcher attributes to Frege a traditional concept of proof, according to which mathematical and logical proofs are psychological activities. I shall argue that Kitcher‘s interpretation conflicts with Frege‘s anti-psychologistic injunction against confusing reasons with causes. Moreover, the psychological explanation obscures one of the most interesting features of a priori knowledge.