Eye avoidance in young children with autism spectrum disorder is modulated by emotional facial expressions.

Abstract

Individuals with autism spectrum disorder (ASD) exhibit a reduced duration of eye contact compared with typically developing (TD) individuals. This reduced eye contact has been theorized to be a strategy to relieve discomfort elicited by direct eye contact (Tanaka & Sung, 2016). Looking at threatening facial expressions may elicit more discomfort and consequently more eye avoidance in ASD individuals than looking at nonthreatening expressions. We explored whether eye avoidance in children with ASD is modulated by the social threat level of emotional expressions. In this study, 2- to 5-year-old children with and without ASD viewed faces with happy, angry, sad, and neutral expressions, while their eye movements were recorded. We observed the following: (a) when confronted with angry faces, the children with ASD fixated less on the eyes than did TD children, persistently across time; (b) the group differences in the overall eye-looking time were rarely found for happy, neutral, and sad faces; (c) the ASD group showed eye avoidance for neutral faces between 1,000 ms and 2,900 ms after the stimulus onset. Additionally, both groups spent more time looking at the angry faces than the faces showing other emotions. Considering that the children with ASD spent less time looking at the eyes of the angry faces than other emotional faces, the results suggest a combination of vigilance to threatening faces and an avoidance of the eyes in children with ASD. Our study not only extends the gaze aversion hypothesis but also has implications for the treatment and screening of ASD.

What Can Eye Movements Tell Us about Subtle Cognitive Processing Differences in Autism?

“…A further subtle but significant finding showed that when an accidental event happened to a puppet in the interlocutor’s hand, ASD participants took significantly longer to begin to monitor the face of the interlocutor compared to both TD and SLI participants (see Figure 6). These findings are indicative of an ASD-specific difference in the allocation of social attention, and this is consistent with previous studies [98,111,128,131]. Delays in looking to the interlocutor’s face indicate that participants with ASD, unlike TD counterparts, did not immediately use the interlocutors face as a social reference marker to indicate detection of the ‘accident’.…”

“…One explanation for this atypicality could be that neutral faces are perceived as more ambiguous in social information for the ASD group, and hence, shifting fixation away from the eyes of neutral faces may be adopted by autistic individuals as a compensatory strategy to alleviate any potential threatening feeling associated with neutral faces. However, a very recent study from Wang, Lu, Zhang, Fang, Zou and Yi [128] found evidence of eye avoidance exclusively for angry faces in ASD children compared to TD controls. There have also been other inconsistent reports for eye avoidance.…”

The influence of emotional face distractors on attentional orienting in Chinese children with autism spectrum disorder

“…As an extension to this, the current study itself directly reveals a visual disengagement difficulty for central angry faces at the endogenous attention level in children with ASD. Furthermore, considering that angry faces convey obvious threatening information, this delay could reflect hypervigilance for threats when they are presented centrally in this group [ 45 ], and this hypervigilance could result in less flexible attentional disengagement from this type of stimuli in ASD children.…”

Out of mind, out of sight? Investigating abnormal face scanning in autism spectrum disorder using gaze‐contingent paradigm

“…To examine how the eye‐looking time changed over time, we proposed a novel data‐driven method based on a moving‐average approach (Dankner, Shalev, Carrasco, & Yuvalgreenberg, ; Wang, Hu, et al, ; Wang, Lu, et al, ) with a cluster‐based permutation test to control the family‐wise error rate (Maris & Oostenveld, ). Specifically, we segmented each set of trial data (600 sample data in total) into epochs of 250 ms (30 sample data), with 29 sample data overlap, resulting in 571 epochs for each trial.…”