Representing metarepresentations: Is there a theory of mind module?

摘要

The mental representation of minds, or Theory of Mind (ToM), has been theorized to be produced by a cognitive module specialized for representing mental representations (domain-specific cognition) or, alternatively, to overlap with a broader ability to represent representations (domain-general cognition). Some key evidence for domain-specific cognition is that autistic children fail a test of ToM, the False Belief test (FBT), but pass a test of non-mental representations, the False Photograph test (FPhT). Individuals with a ToM deficit but no other deficit in representing representations would be evidence against the theory that ToM is accomplished using the same processes as representing non-mental representations. However, I argue that because the FBT might require representing multiple levels of representation (what individuals see as well as what they subsequently believe), the FPhT is an insufficient comparison because it only requires representing one level of representation (what photographs show). If there are multiple levels of representation in the FBT, then the pattern of FBT failure with FPhT success could indicate either failure on a more difficult representational test or failure on a mental domain test (because the FBT would be both of these). In order to test whether individuals who fail the FBT but pass the FPhT have general difficulty representing multiple levels of representation (metarepresentations), I compared performance among children and adults on two new tests of non-mental metarepresentations, the real-time and paper-and-pencil Meta Photograph tests (MPhT), both of which involve photographs of photographs. Whereas individuals who passed the FBT performed better on the MPhT than on a control photograph test (the new Strange Blindness test, SBT), individuals who failed the FBT but passed the FPhT performed worse on the MPhT than on the SBT. Thus, the pattern of FBT failure with FPhT success does not necessarily indicate a domain-specific deficit or preclude a domain-general deficit (because the MPhT, failed by these individuals, has no mental content but does require representing [non-mental] metarepresentations). Overall, I argue that domain-general theories of ToM are at least as well supported as domain-specific theories, and that ToM is closely related to representations of non-mental representations and metarepresentations.