Abstract
Background: In the Don't Look paradigm, subjects view images while avoiding specific features, e.g. eyes or mouth in faces, e.g. social or non-social stimuli onscreen. Previous studies (Shimojo et al., VSS '12 & '13) found reduced orienting to social stimuli and reduced gaze to faces and eyes in high-functioning autism (HFA) compared to neurotypical (NT) controls. We extend these studies to investigate gaze differences using video stimuli. Methods: HFA (N=5) and NT (N=6) adults viewed videos of happy and disgusted facial expressions while gaze was monitored. Three conditions specified different instructions: Avoid Eyes (AE), Avoid Mouth (AM), and Free View. Each condition included 4 s and 20 s (5 repeats of 4 s) dynamic changes of expression. Groups were compared on 1) fixation frequencies to the head, face, eyes, and mouth, and 2) slopes of fixation frequencies over time. Results: HFA and NT subjects used different strategies to achieve equal success in avoidance conditions. NTs shifted to unavoided facial features, while HFA subjects split into two subgroups: one (HFA1; N = 2) that shifted within the face at a significantly slower rate (p<0.02 in AE, p<0.006 in AM), and one that shifted outside the face (HFA2; N = 3), looking 30% less at the head and face in AE (p<0.01), and 47% less at the face in AM (p<0.02). For the whole sample, face fixations were negatively correlated with Autism Quotient scores (p<0.01). In the HFA2 group, significant decreases in head (p<0.001) and face (p<0.05) fixations occurred over both stimulus durations. Discussion: Different solutions to the Don't Look problem segregate HFA subgroups from NT controls. One HFA subgroup lacks the tendency to focus on and maintain interest in socially relevant facial features, while the other attends more slowly to these features.
Meeting abstract presented at VSS 2014