Feature Binding of Social Versus Non-Social Stimuli In Children with ASD

Background: Observed differences in electrophysiological markers of face perception in ASD have been interpreted according to several theoretical frameworks.  Social theories of autism, such as the social motivation hypothesis, posit that atypicalities in face processing reflect core dysfunction in social brain mechanisms or consequent developmental deviance.  In contrast, neuropsychological theories of autism, such as the temporal binding deficit hypothesis, describe face processing anomalies as reflective of general differences in cognitive mechanisms, such as a bias towards local versus global processing. Nevertheless, both frameworks describe disrupted configural processing of faces.  In EEG studies, configural processing is typically indexed by high-frequency gamma oscillations, presumed to mark integration of individual features of a familiar visual percept, or feature binding. Though gamma band activity (GBA) has been demonstrated to be affected in ASD during face perception, little is known about integrative visual processing of non-social stimuli in ASD.

Objectives: We compared feature binding, as indexed by GBA, in social versus non-social stimuli to investigate the specificity of perceptual coherence anomalies in ASD and to explore their consistency with cognitive-perceptual versus social theories of ASD.  The former predicts atypical perceptual binding to both classes of stimuli, a reflection of problems with the feature-binding process itself; the latter suggests that anomalies would be evident only during perceptual binding of social stimuli, reflecting specific vulnerability with social information.

Methods: High-density event-related potentials (ERPs; 128 channel Hydrocel Geodesic Sensor Net) were recorded from 19 children with autism and 19 typically-developing peers matched for age (~12), sex, IQ (~110), and handedness. Participants viewed familiar social stimuli (faces, compared with inverted faces), and non-social stimuli (letters, compared with pseudoletters). Gamma band oscillations were extracted at frontal and occipital electrode groups spanning a 500 ms window concurrent with stimulus onset. 

Results: Significantly different patterns of GBA were observed between groups when viewing social stimuli; during perception of non-social information comparable brain activity was observed.  Comparing face and inverted faces, TD children showed a gamma burst elicited by faces that was attenuated by inversion.  In contrast, children with ASD displayed reduced GBA to both faces and inverted faces.  When viewing non-social stimuli children with ASD and typical counterparts displayed similar patterns of GBA, selective to letters compared to pseudoletters.

Conclusions: Results suggest that deficits in feature binding reflected by reduced GBA are specifically social in nature.  When comparing social information, children with ASD exhibited a predicted reduction in feature binding.  In contrast, when viewing non-social information children with ASD showed normative patterns of brain activity, reflecting intact feature binding.  This pattern of results emphasizes the import of the visual content being processed when evaluating visual-perceptual mechanisms in ASD.