Using Head-Mounted Eye Tracking to Examine Sticky Attention in Naturalistic Child-Parent Social Interaction

Background: Children with autism spectrum disorder (ASD) often display atypical profiles of attention in constrained laboratory tasks (Chita-Tegmark, 2016), but whether these differences manifest in more naturalistic situations remains largely unexplored. One such attentional difference concerns the ability to rapidly disengage attention from one stimulus to another. This “sticky attention” phenomenon has predominantly been studied using a highly constrained and screen-based experimental paradigm, known as the gap-overlap task. Although previous research has demonstrated an impaired ability to disengage in children with ASD and this ability has been associated with ASD severity (Elsabbagh et al., 2013), very little is known about “stickiness” in more naturalistic and real-world contexts – the contexts in which children interact with the environment and their social partners.

Objectives: To determine whether young children with ASD (24-48 months old) exhibit evidence for “sticky attention” -- as operationalized across a number of different quantitative metrics spanning various timescales (from seconds to minutes) -- by using head-mounted eye tracking to measure and quantify moment-by-moment visual attention during free-flowing parent-child toy play, as compared to age-matched typically developing (TD) children.

Methods: Child-caregiver dyads played with 24 novel toys in a toy room space while wearing head-mounted eye trackers (for this study, only child eye tracking was analyzed). Figure 1 shows the experimental set-up. Following eye-tracker set-up, each dyad was given the toys to freely play with, and encouraged to play as they would at home.

Results: Data was acquired from 9 child-caregiver dyads in the ASD group and 16 dyads in the TD group. Sticky attention was operationalized in four ways and spanning different timescales, shown in Figure 2: (a) duration of individual toy looks, (b) proportion of time spent on each object, (c), number of objects looked at throughout the session, and (d) number of returns to objects previously looked at throughout the session.

T-tests revealed that groups did not differ on average look duration or across quartiles of look durations (all ps > .10). There were no differences between the ASD group and the TD group on proportion of time spent on the most looked at object (p = 0.85). Groups did not differ on total number of objects looked at during the session (p = 0.30) or on total number of returns to previously looked at objects per session (p = 0.11). Examination of time courses or finer-grained investigations of these behaviors also revealed highly similar patterns of looking and interacting with toys (Figure 2).

Conclusions: Our initial results suggest that “sticky attention”, measured during naturalistic toy play and operationalized in different ways and across different timescales, is not present in 24- to 48-month old children with ASD. Although many researchers have observed abnormalities of attentional disengagement in ASD (but see Fischer et al., 2016) and have speculated on its real-world importance, it remains an open question how such behavioral differences measured using highly constrained laboratory tasks may manifest in real-world behavior. Future analyses will explore the role of the caregiver on the child’s visual looking patterns.