In a recent article published in this journal, Lea (1991) argues against psychological theory which proposes access to mental states through empirical investigation. His argument depends upon demonstrating certain logical flaws in the arguments of proponents of this kind of empiricism, thereby ruling their conclusions out of court. In response to Lea, I will show that empirical investigation of mental states is possible only when undertaken as a reflexive theoretical practice. Unlike Lea's argument which remains within the Cartesian/Kantian paradigm of a priori structures and categories, an argument based on reflexive theoretical practice takes account of the grounded conditions under which knowledge becomes validated in social settings, involving close attention to the status of language in context of use. I will show that certain elisions in Lea's argument, particularly in his invocation of Wittgenstein's work on language use, eliminates for him the possibility of dealing with language in this manner. Furthermore, I will also show in the course of undertaking the above tasks, that Lea's mode of augment is itself perfectly consistent with the form of apriori theorizing he wishes to reject.