Abstract
The hypothesis that autistic people have impaired metarepresentational ability (and, as one result, lack a 'theory of mind') is described in the context of other psychological work concerning the development and function of the human mind. Evidence for the hypothesis is outlined and evaluated. The claim that impaired metarepresentational ability underlies the social, communicative and imaginative deficits of autism (see paper by Uta Frith in this issue) is discussed. It is concluded that impaired metarepresentational ability is unlikely to be the primary, or fundamental, behavioural difficulty in autism, but that the hypothesis provides an important new way of looking at and understanding autistic behaviour.