Dynamic Visual Search Strategies During Natural Viewing in 12-24 Month-Olds with Autism

Background: The natural environment of toddlers contains a wide array of competing attentional demands: from other people in motion—talking, gesturing, and interacting—to the physical features of the surrounding world (clothing, toys, furniture).  In response, people in general, and toddlers in particular, have only limited attentional resources to deploy. By analyzing how individuals direct those limited resources in pursuit of a goal, we may gain insights into both the goals that guide behavior in particular contexts, and how those goals may differ in young children with autism. In a natural viewing task using dynamic scenes of social interaction, this experimental framework was operationalized by measuring the first fixations following a movie scene cut.  Related work with adolescents and school-aged children with autism spectrum disorders (ASD) demonstrated that these individuals were more likely to direct their fixations to body and object areas than to the face, while their typically-developing (TD) peers were significantly more likely to shift their first fixations to the eyes of onscreen actors. In younger children, results indicated that TD toddlers were more likely than children with ASD to shift their attention to the eyes after a scene cut.

Objectives: To understand differences between toddlers with and without ASD in the rapid deployment of initial fixations when presented with new visual information, and to understand how these initial fixations relate to subsequent fixations as well as to sustained looking preferences that persist throughout an entire viewing period.

Methods: Children with ASD and age- and non-verbal IQ-matched TD controls, between the ages of 12-24 months, watched dynamic social scenes of young children playing.  Scene cuts provided instances where new visual information required a viewer to shift attention from an old location (in the previous frame) to a new target location (in the current frame). Eye-tracking technology was used to collect visual scanning and fixation data.  Dependent measures included reaction times to shift gaze following a movie scene cut; location of first, second, and third fixations within the scene following a cut; overall fixation time spent looking at different regions; and rate of convergence on a new location.

Results: Preliminary results suggests that while reaction times to shift visual attention following a change in visual information are similar between-groups, TD children are more likely to direct their first fixation towards the eyes of on-screen actors than their peers with ASD.  Results also show that initial fixation patterns in children with ASD are very highly correlated with overall fixation patterns, whereas visual search strategies vary between initial fixations and subsequent fixations in TD children, suggesting that the gaze strategies of TD children are more strongly modulated by scene context.

Conclusions: This study explores how the analysis of initial fixations may serve as a proxy for ‘social intuition’ (the first reactions that guide behavior in novel situations) and how ongoing deployment of visual resources varies as novel visual information is presented and as content and contextual cues change over time.