Abstract
Whilst people with autism process many types of visual information at a level commensurate with their age, studies have shown that they have difficulty interpreting facial expressions. One of the reasons could be that autistics suffer from weak central coherence, ie a failure to integrate parts of information into globally coherent wholes [Frith, 1989 Autism: Explaining the Enigma (Oxford: Blackwell); Frith and Happé, 1994 Cognition 50 115 - 132]. To test this hypothesis we presented autistic and (age and IQ matched) normal children with pairs of facial images. Their task was to decide whether the faces showed a similar expression, or whether either the eyes or the mouths were similar (ignoring the rest of the face). Stimuli in the latter condition were digitally manipulated, eg the eyes stemming from a happy face were composited within a face displaying sadness. Although autistics were expected to show relatively greater difficulty comparing whole facial expressions, we proposed that they should be better than normal children when judging the similarity of individual expressive features. Results are discussed in terms of the central coherence theory and current theories of visual information processing in autism.
