Abstract
Atypical face perception in "high-functioning" individuals with Autism Spectrum Disorder (ASD) is either interpreted as a consequence of socio-communicative behaviours (Schultz, 2005), or resulting from abnormally local/detailed perceptual strategies (Behrmann et al., 2006). This is exemplified by a decreased ability to discriminate facial identities when access to local facial cues is minimized (e.g. Morin et al., 2015). We aimed to assess if global or local perceptual strategies were used by individuals with ASD with below average IQ during a face identity discrimination task. Nineteen individuals with ASD (17 male, 2 female) and below average IQ (WASIFSIQ=81) were presented with synthetic face stimulus pairs (Wilson et al, 2002) that were both front-facing (same-view condition), allowing for local identity judgements (e.g., comparing noses), or front-side facing (view-change condition) with one face oriented front and the other side facing (20 deg.). The view-change condition decreased access to local information, resulting in greater reliance on a global analysis to complete the task. Participants were asked to answer 'same' or 'different' during face pair presentation; eye fixation data were collected during task completion. Percentage of time spent looking at each region of interest was calculated. Although there was no main effect of view condition (p=0.76), a main effect of region of interest was found (p< 0.001). The eyes and the hairline were viewed significantly more than the nose, mouth and hairline for both front-front (local; p< 0.001), and front-side (global; p< 0.001) conditions. Hence, these results indicate a pattern of viewing that isolates specific areas of the face (i.e., local processing) as opposed to rapid use of several regions to create a perceptual 'whole' (i.e., global processing) in individuals with ASD. By isolating local and global processing in synthetic faces, this study demonstrates that individuals with ASD with below average IQ prefer local processing strategies.
Meeting abstract presented at VSS 2018