Visual Processing Strategies Used for Face Perception in School-Aged Autistic Children

Background: Human face perception is considered essential for developing typical social interaction skills that are crucial for day-to-day functioning. Consequently, atypical processing of face information has been linked to the social differences that characterize autism. One prominent hypothesis suggests that the unique, detail-oriented visual processing style used by autistics may negatively affect their ability to identify faces when a global analysis is optimal, such as, when faces are presented from different viewpoints. Various studies have assessed the nature of face perception in autism primarily using face images presented only in a front-view orientation. Yet, the effective identification of faces in real-life settings often requires an integration of information across viewpoints (e.g. front and profile). Studies assessing abilities of face identity discrimination in typically developing children and adults have found significant age related differences in abilities, specifically when different head orientations (views) are considered (Mondloch et al., 2003, Habak, Wilkinson & Wilson, 2008).  Assessing face identification abilities across ages and viewpoint is therefore important in understanding how and when differences in such abilities emerge between autism and typical development.

Objectives: To assess the face identity discrimination abilities of school-aged autistic and non-autistic participants in a view specific manner, where access to local face attributes is available (same view) or minimized (different views). 

Methods: Ten autistic and ten typically-developing school aged children matched for full-scale IQ and age (8 to 12 years) performed a face identity discrimination task using synthetic, computer-generated face images (Wilson et al., 2002). These face images consist of simplified (hair and skin texture removed), ecologically-validated stimuli, extracted from traditional face photographs in both frontal (“front”) and 20° side (“side”) viewpoints. A target face was presented for 1000 milliseconds and then followed by two choice faces, one of which had the same identity as the target. Performance was measured using face identity discrimination thresholds (amount of facial geometry change needed to discriminate between faces) for conditions where the target and choice faces were presented in the same view (front-front view) and in different views (front-side view).

Results: As was found in a similar study examining facial identity discrimination across viewpoints in adolescents and adults with and without autism (Morin et al., IMFAR 2010), mean identity discrimination thresholds for the autistic group were higher for the viewpoint change condition (front-side view) when compared to the typically-developing group. 

Conclusions:  A decrease in performance for the viewpoint change condition, as indicated by higher mean identity discrimination thresholds, suggests that facial identity discrimination in school-aged autistic children may be more difficult when (i) access to local cues, such as individual facial features, is minimized, and/or (ii) increased dependence on a global, integrative analysis is introduced to the face task. These findings will be presented along with those from non-social tasks comparing local and global spatial perception, in order to assess whether minimized access to local information specifically affects socially-contingent face perception, or generalizes across complex types of visuo-spatial information in autism.