Perceptual Matching of Facial Expressions and Identities in Children with Autism Spectrum Disorders

Background:   Children with autism have a perception which fundamentally differs from ours. According to the Weak Central Coherence theory, children with autism experience difficulties with central coherence and, as a consequence, they have difficulties in holistic processing (e.g., of faces) and they have a processing bias for details. According to the Enhanced Perceptual Functioning hypothesis, children with autism have superior qualities in detecting and discriminating stimuli. Moreover, children with autism have difficulties with dynamic stimuli and in filtering out irrelevant details.  

Objectives:   Using dynamic face stimuli, we wanted to investigate possible strengths and weaknesses in the perception of children with autism. In particular, we examined whether they were better or worse in matching tasks that required focusing on the identity or the emotion of faces with dynamic facial expressions.  

Methods:   Two groups of 24 boys, matched for age (M = 9 y, range between 7 and 11 y) and IQ (M = 104, range between 90 and 120), were tested. One group had received the clinical diagnosis of autism based on DSM-IV criteria and ADI-R scores, without attention or concentration problems and using no medication. The control group was representative for the general population, in the same range of intelligence. The stimuli consisted of faces of four different female adults expressing four different emotions: happiness, anger, surprise, disgust. Each trial consisted of a series of 25 consecutive images showing the expression dynamically (25 fps), followed by a blank interval (1 s) and a static test frame. Participants had to indicate whether the static test frame was same or different as the series preceding it. In one block of trials, the matching criterion was identity, in another block it was emotion. In both identity and emotion matching, the irrelevant aspect (emotion and identity, resp.) could be congruent or incongruent.  

Results:   The autism group tended to perform worse and slower than the control group but the overall group differences were small. Identity matching was easier than emotion matching. More interestingly, there was a large interaction effect of task and group: The autism group performed worse and slower than the control group on emotion matching, while they performed equally well and fast on identity matching. Congruent trials were performed better and faster than incongruent trials in both groups. Both groups were faster at congruent than incongruent match trials. Moreover, the autism group, but not the control group, appeared faster on the incongruent different trials than on the congruent different trials. In other words, an irrelevant difference was beneficial for the autism group only.  

Conclusions:   We did not find large difficulties in processing the dynamic face stimuli. Task difficulty and group differences depended on the specific task requirements. Children with autism performed equally well as control children on identity matching but they were less good and slower at matching facial expressions regardless of identity. Children with autism were faster when the task-relevant differences between stimuli were accompanied by task-irrelevant differences, whereas control children filtered these out more efficiently.