Differences in Overt but Not Covert Gaze-Following in Young Infants at Risk for Autism Spectrum Disorders

Background: Children and adults with autism spectrum disorders (ASD) show abnormalities in real-world gaze monitoring, but often perform typically in laboratory-based paradigms examining gaze cueing. Two types of gaze cueing—both present from infancy in typical development—often appear intact in ASD: overt (i.e., looking in the direction of gaze) and covert (i.e., faster detection of gaze-cued targets). Typical cueing responses in children with ASD could either suggest that gaze cueing is typical in autism or that performance is achieved via learned alternative mechanisms. Examining gaze cueing in infants at risk for ASD could disambiguate these possibilities and may reveal an early marker of autism.

Objectives: This study used a Posner-style gaze cueing paradigm to examine how infants at low risk (LR) and high risk (HR) for ASD overtly and covertly respond to gaze, and investigated how this responsivity predicted future social and cognitive functioning.

Methods: Thirty-seven infants (9 HR) aged three to five months viewed trials in which a face appeared on-screen and shifted gaze to the left or right.  One second after this shift, the face disappeared and a target appeared either on the side congruent, or incongruent, with the gaze.  Infant eye movements were videotaped and coded by blinded researchers. Covert response to gaze was measured using the latency of the infant’s look to the target. Overt responses were defined as looks in the direction of the on-screen gaze shift at least 100ms after the shift, but before the target appeared.

Results: There was no difference between groups in the time to congruent versus incongruent targets (e.g., covert gaze cueing), although for neither group was the effect significant. The LR group, however, was significantly more likely to overtly follow gaze before the target appeared (p<.001). There was no group difference in error rates for overt cueing (e.g., shifting left after rightward gaze).  For LR infants, errors in overt responding predicted worse communicative (r=-.64, p<.05) and social (r=-.69, p<.01) skills at 12 months on the Ages and Stages Questionnaire. These correlations were in the opposite (positive) direction, although not significant, for the HR group.

Conclusions: Past research has suggested that gaze cueing can rely on physical (e.g., pupil movement) or social (e.g., understanding intention) routes, and that physical factors may explain intact gaze cueing in ASD. Consistent with these findings, covert gaze cueing is not impaired in HR infants.  At-risk infants, however, do show an atypical overt gaze-following response; compared to LR infants, HR infants are less likely to correctly follow gaze shifts before a target appears. This suggests a potential dissociation between the mechanisms of overt and covert cueing, and future studies should explicitly test this possibility. Further, though the small sample precludes definitive conclusions, the processes underlying gaze behavior may differ between LR and HR infants, as a behavior that predicts worse social outcomes for LR infants (looking indiscriminately away from the face) does not appear to serve the same function in HR infants.  Future research will examine how these responses relate to diagnostic outcomes.