The “Face Deficit” in Visual Attention: Parsing Heterogeneity in ASD

Background: As early as 6 months of age, children later diagnosed with ASD prioritize social information differently than typically developing children. Early atypical attention to people, in particular, is hypothesized to lead to downstream effects on social cognition that cause significant impairment in everyday social interaction. Visual attention to faces has recently been measured using infrared eye-tracking technology, and while the preponderance of evidence suggests reduced attention to faces in the ASD population versus typically developing controls (TDC), many studies have been hindered by small sample sizes that do not allow for adequate power to explore covariates that might shed light on the mechanisms responsible for these differences. The present study fills this gap by exploring visual attention to faces and objects in a sample of 372 carefully phenotyped children with ASD and TDC.

Objectives: Parse heterogeneity in visual attention to faces in a large sample of children with ASD and TDC, with the goal of explaining variance associated with reduced attention in the ASD group.

Methods: Children with ASD (N=252) and typically developing controls (N=120) watched 3 16-second sets of 4 video clips, 1 in each quadrant of a 32-inch screen, while their gaze data was collected at a rate of 60 Hz using a Tobii X120 system. Each set of videos included 2 neutral faces and 2 objects in action (e.g., car racing down a highway). Total fixation duration to faces relative to faces+objects was calculated for each participant, and this proportion was correlated with phenotypic variables hypothesized to explain individual variance.

Results: Consistent with the extant literature, children with ASD looked proportionately less at faces than TDCs, F(1,370)=5.55, p=.02. In the ASD group alone (but not the TDC group), more looking to faces was associated with higher scores on the Benton Test of Face Recognition (r=.16, p=.02), higher scores on the ADHD subscale of the Child and Adolescent Symptom Inventory (CASI; r=.14, p=.03), and lower overall scores on the Vineland Adaptive Behavior Scales (VABS; r=-.12, p=.05). Looking to faces was not associated with age, IQ, scores on the Social Responsiveness Scale, the Repetitive Behavior Scale, CASI social phobia or anxiety subscores, the VABS socialization or communication subscores, or the Brief shift score. In-depth analyses are ongoing, and will include sets of conceptually related scales (e.g., Benton and Let’s Face It!) from which latent variables can be extracted and used to predict variance in attention to faces.

Conclusions: Atypical attention to social stimuli is one of the most robust findings in autism research, and was replicated in the present study comparing large samples of children with ASD and typically developing controls. Within the ASD group, a number of variables were found to correlate with individual differences in gaze to faces, including face processing skill and a measure of adaptive behavior. Planned analyses will model the unique and shared contributions of each of these individual variables to differences reflected at the group level.