The present study was concerned with children's ability to identify a prior expectation of their own as false belief. In Expt 1 a sharp improvement was found between 3 years and 5 years not only in the ability to identify own beliefs but also in the explanation of what caused that belief and in the ability to infer another person's belief from exposure to misleading information. Experiment 2 ruled out that 3‐year‐olds' failure with belief identification was due to a memory problem or to a misunderstanding of the test question. Experiment 3 excluded the further possibility that 3‐year‐olds simply failed belief identification because of embarrassment about having said something false. The younger children's difficulty with the identification of own beliefs contradicts Wimmer, Hogrefe&Sodian's (1988) proposal that young children's difficulty in understanding another person's false belief is solely due to a failure to understand informational origins of beliefs. The present finding also speaks against the Cartesian assumption that the mind is transparent to itself, an assumption which underlies the perspective taking tr