Evidence for infants' understanding of false beliefs should not be dismissed.

but also in face-processing, gaze monitoring and joint attention [3,8]. Failures in low-level inputs to ToM computations could account for their deficits on these tests. Without co-morbid intellectual disability, individuals with autism seem to have intact capacities for metarepresentation and recursion, as indexed by false-photograph tests and mathematical ability [4,9]. All known cases of patients with ToM deficits arising from brain lesions involve deficits in either low-level social input systems or higher-level domain-general abilities. Orbitofrontal patients with deficits on ToM tasks have lower-level social deficits in face-processing and tracking intentions [5]. As Apperly et al. detail, medial frontal and temporoparietal junction (TPJ) patients have either executive function deficits, general metarepresentational deficits, or no ToM deficits [1]. When Baron-Cohen, Leslie and Frith published their original paper 'Does the autistic child have a theory of mind?', they argued that ToM is 'one of the manifestations of a basic metarepresentational capacity' ([10], p. 37; emphasis added). We think it is time to recapture the insights of their original proposal, and abandon the quest for the neural substrate of the fabled ToM module. Apperly et al.'s analysis of TPJ patients' performance shows that it might be more promising to focus on the domain-general and uniquely human ability of metarepresentation. Testing joint attention, imitation and play as infancy precursors to language and theory of mind. A mathematician, a physicist, and a computer scientist with Asperger Syndrome: performance on folk psychology and folk physics tests. In their response to Leslie [1], Ruffman and Perner (R&P) reiterate their position that there is no need to explain Onishi and Baillargeon's (O&B) recent findings [2] with 15-month-olds in terms of attributing false beliefs (FB). Here we put forward three reasons why their points do not explain the infants' performance. (1) We are not surprised that Leslie [3] did not respond to Perner and Ruffman's 'neurological' argument [4], according to which 'cells in the brain code for configurations of persons relating to objects'. To support their argument, they cited: (i) a neural-network model [5], which hypothesized rather than demonstrated the forming of associations in the prefrontal cortex between two rather than three stimulus features; and (ii) a neurophysiological study [6] showing that cells in the rat's hippocampal region are activated differently for novel and familiar arrangements of pictures, without demonstrating that those cells coded for episodes rather than familiarity of arrangements per se. Although these studies suggest that brains could …