Neural oscillatory dynamics reveal altered top-down and integrative mechanisms during face processing in autistic children and unaffected siblings of autistic children.

Face processing is fundamental to social communication and has been a major focus of autism research. While event-related potential (ERPs) studies of face processing have produced mixed results, little work has examined neuro-oscillatory dynamics, which may better capture the integrity of underlying networks. To address this gap, EEG was recorded from children aged 8-13 across three groups: autistic (n = 50), non-autistic (n = 38) and siblings of autistic children (n = 26), during a visual oddball task. In a blocked design, participants viewed faces and objects, presented upright and inverted (non-targets), to assess the face inversion effect (the FIE; a larger or delayed N170 to inverted than upright faces), and responded to infrequent shadow versions (targets). Analyses using permutation statistics and linear mixed models focused on non-target stimuli, quantifying face-related ERPs (P1, N170) and oscillatory activity associated with sensory and attentional processing (theta, alpha, gamma). Across groups, faces elicited earlier P1 and larger N170 amplitudes than objects, and showed a FIE. Furthermore, the rightward lateralization of the FIE was reduced for autistic participants. Analyses in the frequency domain revealed greater induced theta for inverted versus upright stimuli and for faces versus objects, revealing face specific effects, and stronger theta for inverted faces for the autistic and sibling groups, suggesting greater cognitive effort in processing these social stimuli. Gamma-band inter-trial phase coherence exhibited face selectivity only in the non-autistic group, pointing to differences in early network synchronization in autistic children relative to their non-autistic peers, whereas alpha event-related desynchronization did not vary by group or category. Altogether, these findings support altered neural synchronization/efficiency for autistic participants and siblings of autistic children, that is specific to face stimuli and seen despite largely typical sensory driven encoding. These data suggest that neural oscillatory assays are more sensitive to face processing differences in autism than broadband ERPs and that these oscillatory assays may be endophenotypic.