Abnormalities in Early Visual Processing of Faces in Autism

Background:   Abnormal face processing is an often-reported deficit in autism spectrum disorders (ASD). However, it is unknown whether this is due to abnormalities in face processing itself, or in visual processes preceding face perception. A suggested atypical visual process in ASD is visual segmentation, i.e. the integration of details into a global figure and segregation of this figure from its background. It is unknown at what moment in time, i.e. during initial visual processing, visual segmentation, or object processing, abnormalities in visual processing of faces in ASD are firstly present.

Objectives: The current study investigated three processes involved in face perception: initial visual processing, visual segmentation and object processing. Using electroencephalography (EEG), it was studied which of these visual processing stages are abnormal in persons with ASD.  

Methods: Adults with ASD and controls viewed three types of stimuli, while brain activity was recorded using EEG: orientation-defined textured faces, houses, and homogeneous stimuli. Stimuli were presented for a short duration, which is previously shown to evoke multiple ERP peaks, associated with different visual processes. These include an initial negative peak for all stimuli at occipital and occipito-temporal sites, associated with initial visual processing. In addition, there is a visual segmentation peak (i.e. a difference between faces and houses versus homogeneous stimuli) at occipital electrodes, and a positive and negative peak at occipito-temporal electrodes associated with object processing of faces and houses. Differences between groups on latency and amplitude of these peaks were of interest in the current study.  

Results: Preliminary results indicated abnormalities in initial visual processing at occipital and occipito-temporal sites: longer latencies were present for ASD than control subjects. A trend towards this longer latency was also present in visual segmentation but interestingly not at later peaks at occipito-temporal sites. Differences between groups in peak-amplitude were also indicated at occipito-temporal sites in initial processing as well as the later negative peak, associated with object processing. No peak amplitude differences between groups were present for initial processing or visual segmentation at occipital electrodes.  

Conclusions: These results indicate delayed and diminished initial visual processing in ASD, as well as delayed visual segmentation. However, this latency difference was not present at later time-points at occipito-temporal sites, associated with object processing. Object processing did however seem to be abnormal regarding peak amplitude in ASD. Together, these results show that already at an early stage visual processing is abnormal in ASD.