Second-Order Joint Attention: Children with Autism Spectrum Disorder Showed Abnormal Eye Movements in Response to Others’ Gaze Following

Background: Considerable empirical evidence shows that children with Autism Spectrum Disorder (ASD) have an intact ability to follow others’ gaze direction, which seems to contradict clinical observations that children with ASD have difficulty with joint attention. In real-life, joint attention has an interactive and reciprocal nature and may involve a multi-order process—we don’t merely respond passively to others’ gazes, but also interpret others’ responses to the our own gaze direction and modify our attention accordingly.

Objectives: We propose a multiple-layer concept of joint attention that defines first-order joint attention as attentional responses to others’ gazes and second-order joint attention as gaze responses to others’ first-order joint attention. In the present study, we focused on the second-order joint attention in children with ASD compared with typically developing (TD) children. Specifically, we aimed to explore how children with and without ASD respond to others’ gaze-following when they have initiated joint attention, such as gaze duration on the object followed or not followed by others.

Methods: Twenty-four children with ASD and 24 age- and IQ-matched TD children completed a computer-based gaze-contingent eye-tracking task. In the task, children had to choose one of two objects to look at (first-looked-at object, FLO) and a computerized face would follow children’s gazes towards the FLO (congruent condition), unfollows the gazes towards the other object (incongruent condition), or close the eyes (closed-eyes condition). Novel data-driven temporal-course analyses were used to explore how children with ASD and TD children allocated their visual attention according to others' gaze responses. Furthermore, novel data-driven correlation analysis was used to explore whether looking at eyes would influence subsequent looking time on objects the computerized face gazed at.

Results: TD children, but not children with ASD, were sensitive to others’ gaze responses, as evidenced by their higher proportional FLO-looking time in the congruent condition than in the incongruent and closed-eyes conditions, and also higher proportional NFLO-looking time in the incongruent condition than in the congruent and closed-eyes conditions. Such sensitivity occurred around 1700 ms after the computerized face shifted its gaze (Figure 1A & 1B). Also, children with ASD looked at eyes less than TD children during 875-1700 ms after the computerized face shifted its gaze (Figure 1C). Furthermore, we found that TD children, but not children with ASD, who looked more at eyes during around 100-1300 ms tended to look more at the object the computerized face gazed towards during around 1500-4000 ms in the incongruent condition (Figure 2).

Conclusions: Our study bridges a significant gap in the literature by examining the second-order joint attention in both children with and without ASD. We found that TD children, but not children with ASD, showed sensitivity to others’ gaze responses to their own gaze. The absence of the sensitivity in children with ASD was possibly due to their reduced eye-looking time caused by their diminished social motivation relative to TD children. Our study provides insights into our understanding of abnormal social cognition in the context of ecologically valid social interactions in children with ASD.