Attention Shifting from Emotional Faces in High-Risk Infants and Relations with Later Social-Communicative Behavior

Background: Studies with individuals with ASD and their first-degree relatives have found difficulties in face processing as well as more general atypicalities in visual attention.  Prospective studies of infants at high-risk for autism (HRA; defined as having an older sibling with ASD) allow researchers to examine the origins of these difficulties to help understand how these domains could be linked to later developmental difficulties.

Objectives: The present study examined attention disengagement from emotional faces in HRA and low-risk control (LRC) infants.  Attention shifting was then analyzed alongside prospective social-communicative behavior with the aim of using early visual attention to predict later development within the broader autism phenotype.

Methods: Attention disengagement in HRA and LRC was tested in 6- (HRA: 30; LRC: 34), 9- (HRA: 34; LRC: 25), and 12-month-olds (HRA: 29; LRC: 28). Infants receiving a subsequent clinical judgment of ASD were not included in the present sample. Each trial presented one of three emotional faces (happy, fearful, neutral) in the center of the screen for 1s, then the face remained present while a distracter appeared on the left or right. A Tobii T60 eye-tracker recorded gaze, and for each emotion, latency to shift from face to distracter and percentage of no-shift trials were calculated.  Parents completed the Communication and Symbolic Behavior Scales (CSBS) at 18 months. 

Results: For latency to shift from faces, infants showed a main effect of emotion (p = .012), with slower shifting from fearful as compared with happy and neutral faces (ps < .015). A significant age*group interaction was also found (p = .025), with LRC showing decreasing shift latency with age, while HRA show the opposite pattern (Figure 1).  Analysis of no-shift percentage revealed a main effect of emotion (p < .001), with a greater percentage of no-shifts found for fearful faces compared with happy and neutral (ps < .05), and happy greater than neutral (p = .027). A significant age*group interaction (p = .033) revealed that HRA show increasing no-shift percentage with age, while LRC show no age-related differences. Correlations were run at each age between eye-tracking measures and CSBS 18-month social and total scores.  At 6 months, HRA showed a positive association between disengagement latency from fearful faces and CSBS total scores (p = .044; Figure 2).  At 9 months, HRA showed a positive association between no-shift percentage to fearful faces and CSBS social scores (p = .05).  No other correlations were significant.

Conclusions: Similar to past work, we found slower shifting and increased no-shifts for fearful faces, and this emotion-specific pattern was comparable for LRC and HRA. HRA infants with more normative responses to fearful faces also showed better social-communicative outcomes. Across emotions, our results show diverging trajectories of increasing and decreasing disengagement efficiency during the first year of life in low-risk and high-risk infants, respectively. These findings are representative of the broader autism phenotype, as this sample excluded infants with an ASD diagnosis. Future work will examine whether infants later diagnosed with ASD differ on these emotion and attentional measures.