Gaze Following in Infants at Risk for Autism: The Role of Eye Information, Head Turns and Salient Objects

Background: The ability to follow other people’s gaze is a key component of joint attention (JA), and Autism Spectrum Disorders (ASD) are associated with poor JA skills. Surprisingly, a recent eye-tracking study (Bedford et. al., 2012) found that infants with later emerging ASD were as accurate in gaze following as control children. In contrast, Elsabbagh et. al. (2012) found different ERP responses to dynamic gaze shifts in infants who were later diagnosed with an ASD as compared to typically developing infants. Thus, despite processing eye-related information differently, infants with later autism seem to be able to follow gaze. No study of infants at risk has investigated the tendency to follow eye direction alone, without an accompanying head turn. 

Objectives:  The aim of the present study was to investigate gaze following in 10-month olds at risk for autism (due to having an older sibling with ASD), with and without head turns. 

Methods:  The study included a group of children at high risk (HR) for ASD (n=24) as well as a group of children at low risk (LR, n=11). Using live eye tracking technology, we recorded the infants’ gaze patterns when they observed a model who gazed at one of two objects in front of her. The primary measure was a difference score (DS) with the number of incongruent gaze shifts subtracted from the number of congruent gaze shifts made by the child, aggregated over several trials. The design included three conditions; one in which the model turned his/her head towards the toys (eyes and head condition), one in which the model shifted his/her gaze towards the toys without turning the head (eyes only condition) and one where the model turned his/her head towards empty spaces with no objects present (empty condition). 

Results: The analysis revealed a group x condition interaction (p < 0.05). The HR group showed significantly higher gaze following accuracy in the eyes and head condition than in the eyes only and empty conditions (p:s < 0.05, bonferroni corrected). The gaze following accuracy in the LR group did not differ between conditions.  The effect could not be explained by a general group difference in number of gaze shifts. Both groups were able to follow gaze accurately (DS > 0) in the eyes and head and empty conditions, but only the LR group followed gaze accurately in the eyes only condition.

Conclusions:  While performance was modulated by condition in the HR group, no such modulation was found in the controls. This could suggest that infants in the HR group rely more on features such as head turns towards target objects, and fail to follow gaze when such cues are absent (i.e. in the eyes only condition). Lack of modulation in the LR group could indicate that these infants are more sensitive to the communicative meaning of the gaze cue, and less reliant on salient features such as head turns and objects.