A Preliminary Head-Mounted Eye-Tracking Study of Individuals with ASD Touring a Museum of Art

Background: Eye tracking is a powerful tool for studying attention patterns, revealing differences in attention to people’s mouths, eyes, and activities among individuals with typical development (TD) and autism spectrum disorders (ASD). Many current eye tracking studies have used table-mounted devices to examine pre-recorded video stimuli. However,  commercial head-mounted eye trackers can be expensive, limiting accessibility. Gaze tracking in real world situations may offer insight into the difficulties individuals with ASD experience in authentic social interactions.

Objectives: 1) To construct an untethered head-mounted eye-tracking device; 2) To contrast patterns of attention in individuals with ASD and TD towards people in paintings and social situations.

Methods : We used inexpensive electronic components to construct untethered head-mounted eye trackers. A tour guide escorted pairs of participants (n = 9, ASD; n = 13, TD) through an art museum to view four paintings featuring a variety of social content. For each painting, participants completed three tasks: (1) silently observing the painting, (2) listening while the guide spoke, and (3) answering open-ended and rating questions asked by the guide (QA) . Behavioral coding was completed on all participants. Eye-tracking data has been analyzed for (n = 5, ASD; n = 3, TD). To track gaze, we detected the pupil centroid and estimated a map from its position to calibration targets in scene videos.

Results: Linear mixed model analyses revealed a task x group interaction for within-painting regions of interest (p < .01). During the QA task, participants with ASD looked more at the painting background than did TD individuals (p < .05). This effect was absent in the silent observation task, and marginally significant (p = .080) during the listening task. While looking at the guide or the other participant, participants with ASD looked less at people’s eyes than did TD participants (p < .05). There also was a task x group interaction for duration spent looking at real world versus the paintings (p < .05), with ASD and TD participants showing opposite patterns of attentional allocation. Behavioral coding showed a higher number of shifts of attention to the guide (p < .05) and to the painting (p < .05) by participants with ASD than by TD participants. Participants with ASD gestured more than TD participants (p < .05).

Conclusions: High-functioning adults with ASD show similar gaze patterns towards paintings as do TD adults during silent viewing. These patterns become more atypical as the social demands imposed by the guide increase. During real-world social interactions, individuals with ASD showed decreased attention towards the eyes of others. Individuals with ASD tended to monitor the real world more than typical individuals, and engaged in more discrete shifts of attention during social interaction, suggesting either executive function or social information processing deficits. More prototypical patterns of attention towards paintings in the absence of communicative processing demands may be due to an equalizing effect of art on attention, or the static nature of the painting compared with the dynamic, social nature of interaction with the guide.