Logical, analogical, and psychological reasoning in autism: A test of the Cosmides theory

An important regulatory process in the development of behavior is cognition. However, cognition as a blanket term is far too broad to be useful. Rather, specific types of cognition need to be examined separately. One proposal is that one type of human reasoning evolved in a social context, to solve social problems. Here, we report two experiments that use autism to test a prediction from that theory: that social intelligence should be independent of nonsocial intelligence. Autism was chosen because deficits in social reasoning ("theory of mind") are well known. The question we tested was whether their theory of mind deficit was dissociable from abstract and relational reasoning ability. In particular, we expected that the abnormalities in the behavioral development of children with autism would be regulated by abnormalities in theory of mind reasoning rather than other forms of reasoning. Children with autism and matched controls were given tests of abstracting reasoning, which did not involve mental state understanding. Results showed that children with autism performed comparably to the control groups, both on a test of transitive inferential reasoning and on a test of analogical reasoning. These results lend support to the specificity of the theory of mind hypothesis for autism and to Cosmides' theory of the evolution of social intelligence. They also show that cognition as a regulatory process in development needs to be examined in highly