Same Day, Different Gaze: Task Effects on Eye Gaze to Social Stimuli

Background:

Atypical eye gaze is a canonical symptom of Autism Spectrum Disorder (ASD; Kanner, 1943), but attempts to use eye-tracking to analyze aberrant gaze patterns have produced mixed results (Papagiannopoulou et al., 2014). This variability may be due to different task demands, which are known to affect eye-tracking patterns in TD individuals (e.g., Triesch et al., 2003; Radun et al., 2014).

Many passive-listening studies have found that individuals with ASD look less at the eyes of unfamiliar actors than their neurotypical (TD) peers (Klin et al., 2002; Sterling et al., 2008; White, et al., 2015). However, emotion identification tasks often don’t show these group effects (Bal, et al., 2010; Sawyer et al., 2012; Spezio et al., 2007). This variability in results may reflect differences between task-driven versus passive gaze in individuals with ASD.

Objectives:

To analyze the effect of task type (passive viewing versus emotion-identification) on eye gaze patterns to social stimuli in children and adolescents with and without ASD.

Methods:

Participants were children and adolescents ages 10-17, 19 with ASD (3 females) and 25 TD controls (8 females), who were not significantly different in language ability as measured by the CELF-5 (ASD M = 113, TD M = 114), IQ as measured by the K-BIT (ASD M = 117, TD M = 107), and gender.

Participants watched videos of human actors while eye gaze was recorded in two separate tasks. During one task, participants passively viewed adolescents talking about their lives. In the other task participants were asked to identify the emotion on a speaker’s face.

We calculated emotion-identification accuracy in the second task as well as percent dwell times to the actors’ face, eyes, and mouth in both tasks.

Results:

There was a main effect of diagnosis for gaze time to the face, with ASD participants looking less across both studies. We also found a significant interaction of task and diagnosis on gaze to the eye region. Post-hoc tests revealed that participants with ASD looked less to the eye region than TD participants during passive listening. In the emotion identification task, there were no differences in eye gaze or in accuracy rates across groups.

Conclusions:

Participants with ASD looked less at the eyes than their TD peers when passively watching a video, but both groups showed similar gaze patterns during emotion identification from faces. Reduced social motivation among individuals with ASD could explain reduced gaze to the eye region in the passive viewing task. It is possible that this lack of social interest is then overridden by a desire to successfully complete a task (McClelland, 1987), thereby eliminating the group difference in gaze pattern during the emotion identification task. In summary, our analysis shows that gaze patterns of adolescents with ASD to faces can change with the addition of a concrete task demand. Therefore, variability of results in previous eye-tracking studies may reflect effects of the tasks, rather than the natural propensity of individuals with ASD to gaze at faces.