Is Face Recognition Really Impaired in Autism Spectrum Disorders? the Role of Gaze Direction

Background: Face recognition is often, although not always, reported to be impaired in high functioning individuals with an autism spectrum disorder (ASD). One potential explanation for this deficit is motivated by findings that face recognition might be a type of expertise. That is, individuals with an ASD may not develop expertise in face perception because of less time spent looking at faces (e.g., Gretolli, Gauthier, & Schultz, 2002). There is some evidence that participants with autism tend to avoid looking at others, and especially looking at the region of the eyes (e.g., Kanner, 1943; Langdell, 1978; Osterling & Dawson, 1994). Thus, it is possible that lack of social interest underlies face recognition deficits in autism. A related, but slightly different hypothesis, is that the aversion to eye contact in ASD that is often reported clinically may play a role in reported face recognition deficits.  A test of this hypothesis stems from social cognition studies that have shown that typically developing children and adults better remembered faces in which the eyes were directly gazing at them compared to faces in which the eyes were averted (e.g., Hood, Macrae, Cole-Davies & Melanie Dias, 2003). 
Objectives: We tested the hypothesis that, in contrast to findings in typically developing individuals, face recognition in ASD would not be facilitated by direct gaze. Instead, we expected to find the opposite pattern in the ASD participants: better performance for averted gaze faces relative to direct gaze faces.
Methods: Eighteen high functioning children and adolescents with an ASD and 19 age and IQ matched typically developing controls were shown a series of images of faces. The gaze of the face was either direct or averted.  We then tested memory for these faces in an old/new recognition task.  All faces in the old/new recognition task were shown with their eyes closed. 
Results: A group by condition (direct gaze, averted gaze) ANOVA revealed a group by condition interaction, F(1,35)=5.2, p<.05, but no main effect of group or condition. The general pattern of results suggested that the typically developing group better remembered the direct gaze faces than the averted gaze ones, whereas, in contrast, the ASD participants showed the reverse effect. Interestingly, there was no difference between groups for recognition of faces with averted gaze (controls = 59%, ASD = 63%), whereas the ASD group recognized fewer direct gaze faces (54%) than the control group (67%; t =-2.21, p < .05).
Conclusions: These results imply that the direction of gaze plays a critical role in face recognition performance in those with an ASD. There may be an important link between gaze aversion in ASD and face recognition abilities. Given that previous studies have primarily used face stimuli with direct gaze, it is important to consider conclusions about “impaired” face processing in light of the current study.