Infant Gaze to Faces Across Interactive Contexts

Background:  Infant gaze is an important social-communicative behavior and is particularly relevant among infants at risk for autism spectrum disorder (ASD), who have been shown to exhibit atypicalities in gaze-related behavior in the first year of life. A common scenario in research settings, in which conditions need to be standardized across participants, is to measure infant social-communication behavior with an unfamiliar examiner, including during diagnostic assessments such as the Autism Diagnostic Observation Schedule. It is important to understand how infant behavior in structured contexts with novel social partners relates to behavior during interaction in a more familiar context (i.e., play) with a caregiver to assess whether skills in such structured settings generalize to other contexts.

Objectives:  To investigate the relationship between gaze behavior during two different interactive contexts: structured testing with an unfamiliar examiner and unstructured play with a parent, in infants without and with a sibling with ASD.

Methods:  Participants were classified into one of three groups based on diagnostic assessment at 36 months: low-risk non-ASD (n=53), high-risk non-ASD (n=66), or ASD group (n=17; 16 high-risk, 1 low-risk). At 6, 9, and 12 months, infants participated in two interactive contexts: 1) an unfamiliar examiner administered the Mullen Scales of Early Learning, a standardized developmental test, and 2) a play interaction with a parent, during which infants (in a high chair with a tray) and their parents (seated across from their infants) were provided a set of age-appropriate toys and instructed to play as they would at home. Infant social-communicative behavior was coded during the first 6 minutes of the examiner-administered Mullen Visual Reception subtest and during the 3-minute parent-child play interaction at each visit. The frequency of gaze to an adult’s face was coded following procedures in Ozonoff et al. (2010). Frequency counts were then divided by the total duration coded to create rates of gaze per minute in each context.

Results:  Regression analyses were conducted to examine the associations between infant gaze to face behavior in the two interactive contexts at each age, controlling for potential effects of familial risk or diagnostic classification. At 6 months, infant gaze to faces during Mullen testing approached significance in predicting infant gaze to face during parent-child play, β=0.25, standard error=0.14, p=.08. At 9 months, infant gaze to faces during Mullen testing significantly predicted infant gaze to face during parent-child play, β=0.34, standard error=0.10, p<.001. At 12 months, infant gaze to faces during Mullen testing significantly predicted infant gaze to face during parent-child play, β=0.46, standard error=0.14, p<.01.

Conclusions:  Infant gaze behavior, a potential early marker of ASD, was significantly associated between two interactive contexts by 9 months of age, controlling for outcome group. Infants with higher levels of gaze to the face of an examiner during structured testing were likely to have higher levels of gaze to the face of a parent during unstructured play. This supports the validity of observing gaze behavior within laboratory settings, which appears to provide a representative measurement of an infant’s skills in more naturalistic settings.