Visual Scanning of Familiar and Unfamiliar Faces in 12-Month-Olds Later Diagnosed with ASD

Background: Eye-tracking studies have provided evidence of atypical scanning of faces in individuals with ASD as well as their first-degree relatives.  Prospective work examining infants with an older sibling with ASD provide an opportunity to look for markers of atypical face processing, some of which might relate to the broader autism endophenotype, and importantly, others that might be predictive of later ASD outcome

Objectives: The present work examined measures of visual scanning during the first year of life in infants as it might relate to later ASD outcome.  Many infant sibling studies have thus far focused on differences between high-risk and low-risk infants, but with children reaching an appropriate age for ASD classification, we can now begin studying the relationship between early measures and later ASD outcome.

Methods: As part of a longitudinal study of infant siblings of children with ASD and typically-developing children, 12-month-olds were presented with side-by-side neutral images of their mother and a stranger.  A Tobii eye-tracker monitored eye gaze during presentation.  The present analyses focused on looking behavior during the first 10s of presentation and examined looking time to each picture as well as looking to the eye, mouth, and face regions.  Proportion of looking to each of these three regions was also analyzed.

Eighteen children who had eye-tracking data at 12-months and an ASD assessment on a follow-up visit at 24- or 36-months were classified into three groups for the present analyses: (1) children classified as having ASD (‘outcome’); (2) children at high risk for ASD but who have been classified as typically developing (‘HRA-‘); (3) children at low-risk for ASD who have been confirmed to be typically developing (‘LRC-‘).  Each child in the outcome group was matched to a child in both the HRA- and LRC- groups based on initial cognitive ability.

Results: Preliminary results include 6 infants per group and are presented descriptively due to the limited sample size.  Data collection is ongoing and an additional 10-20 children will soon be old enough to be included in the present analyses.  The proportion of time spent on the face was similar for LRC- (M=0.98, SD=0.04) and HRA- (M=0.95, SD=0.06), and smaller for the outcome group (M=0.87, SD=0.10).  Proportion of time on the mouth region was similar for outcome (M=0.11, SD=0.14) and HRA- (M=0.11, SD=0.15) and smaller for LRC- (M=0.03, SD=0.06).  For proportion of time on eyes, HRA- resembled outcome when scanning mom’s face (M=0.36, SD=0.39 and M=0.32, SD=0.39, respectively; LRC-: M=0.74, SD=0.14), but when scanning stranger’s face, HRA- more closely resembled LRC- (M=0.64, SD=0.38 and M=0.63, SD=0.40, respectively; outcome: M=0.51, SD=0.43).  Total time on eyes was similar for LRC- (M=3637ms, SD=1775) and HRA- (M=3672ms, SD=2953) and reduced for outcome (M=2814ms, SD=2961).

Conclusions: Preliminary analyses show infant face scanning patterns that are similar for the broader autism endophenotype and other patterns that appear to differentiate outcome and non-outcome children.  As samples grow, this work will be important for identifying early markers of ASD.