Atypical neural response to human faces in autism spectrum disorders (ASD) has been hypothesized to develop from reduced social drive and consequent inattention to faces. Face perception is indexed by specific markers of hemodynamic (fusiform gyrus) and electrophysiological (N170) activity. Face-sensitive brain regions are demonstrated to activate when faces are merely implied by context, suggesting that intrinsic facial features are not necessary to evoke face-specific neural responses and that face-related activity is modulated by top-down processes, such as sensitivity to contextual cues. Given the limited temporal resolution of hemodynamic neuroimaging, it is not known whether the influence of context is exerted at early stages of face processing or at later stages, driven by feedback from higher cortical regions. Furthermore, individuals with ASD are noted to exhibit both behavioral and neural anomalies in face processing, but it is not known if and when these anomalies can be modulated by context.
Here we harness the temporal resolution of event-related potentials (ERPs) to (a) determine the stage of processing at which context exerts its influence on face processing and (b) examine neural mechanisms underlying the influence of context in children with ASD and typically developing (TD) children. We predicted that TD children, but not those with ASD, would show contextual modulation of face-related activity at the earliest stages of face perception, as indexed by shorter latency ERP components (N170).
Children with ASD (N=24; mean age=12) and matched TD children (N=24) participated in the study. EEG was recorded with a 256 electrode Geodesic Sensor Net while participants viewed: images of faces degraded such that internal features were no longer visible (DF1), followed by intact faces alone (faces) and further contextual cues disambiguating the degraded face (e.g., bodies with degraded faces), followed by a second set of degraded faces (DF2). Peak amplitude and latency were extracted for the N170 from electrodes over occipitotemporal scalp. To examine the effect of context on face perception, analyses contrasted N170 responses to faces to and degraded faces prior to and subsequent to establishing social context (DF1 and DF2, respectively).
A main effect of condition (p<.001) indicated that both groups displayed enhanced N170 amplitude to DF2 relative to DF1 (p<.01). N170 amplitude to faces was also greater than that elicited by DF1 (p<.01), but faces did not significantly differ from DF2. Children with ASD had significantly slower latency to DF1 and DF2 compared to TD (ps<.05).
This study demonstrates that social context modulates face-related brain activity at early stages of face processing. Ambiguous stimuli, devoid of intrinsic facial features, elicited comparable response to faces only after exposure to intact faces. This modulation by social context was also observed in children with ASD. Consistent with prior research, although children with ASD showed an intact response in terms of signal strength, they displayed decreased processing efficiency, reflected in longer latencies to face-related ERP components. These findings suggest the potential value of treatments involving awareness of social context for children with ASD.