This essay sketches the potential implications of Wittgensteinian thought for conceptualizations of socalled fictive mental states, e.g. mental calculating, imagination, pretend play, as they are currently discussed in developmental psychology and philosophy of mind. In developmental psychology the young child's pretend play and make-belief are seen as a manifestation of the command of an underlying individualistic “theory of mind”. When saying “This banana is a telephone” the child's mind entertains simultaneously two mental representations, a primary or veridical representation about the (...) real properties of banana's and a pretend representation. It is the task of psychology to explain how this “double knowledge” does not result in conceptual chaos. Various sorts of internal mechanisms are postulated. In this essay it is argued that the threat of chaos is misconceived, and that the solutions are irrelevant. Following Wittgenstein's sparse remarks about the secondary sense of words, I argue that pretend language does not refer to underlying individual mental representations but to the child's creative transference of words used in one, primary, domain of application to another, secondary domain. The command and use of the secondary domain logically presupposes the command and use of the primary domain. Since the latter domain is necessarily public and social, fictive mental states cannot be dealt with purely individualistically as current mentalism assumes. (shrink)