An Ethogram of Behaviors Guiding Dynamic Visual Scanning in 12-24 Month-Olds with Autism Spectrum Disorder

Background: Previous research has demonstrated that two-year-olds with autism spectrum disorder (ASD) look more to mouths, bodies and objects when viewing video scenes of caregivers, while typically-developing (TD) toddlers look more to their eyes. Related research has shown that two-year-olds with ASD preferentially attend to physical contingencies, regardless of their social context.  These analyses of visual fixation to discrete regions of interest suggest that toddlers with ASD attend more to physically salient aspects of a scene while TD toddlers preferentially attend to elements that have social adaptive value. These differential patterns of visual fixation align with findings within a separate but similarly motivated body of research investigating reward processing among individuals with ASD. Investigation of neural processing of reward among individuals with ASD suggests that individuals with ASD may process reward—particularly social reward—differently than typically-developing individuals, and may learn less than typically-developing individuals from socially mediated rewards. While summary analyses of fixation provide useful indices of differences between individuals with and without ASD, natural social interactions incorporate many dynamic and complex cues that are not limited to discrete regions of interest. The current study, rather than summarizing fixations in relation to regions of interest, aims to characterize functionally meaningful social actions that drive the attention of TD children but fail to capture the attention of children with ASD.

Objectives: This study aims to characterize functionally distinct elements within naturalistic stimuli that contribute to differences in dynamic visual scanning of children with and without ASD.

Methods: Eye-tracking data were collected as 12-24 month-old TD children (N = 28) and children with ASD (N = 48) viewed naturalistic videos of peer interactions. Children were matched on chronological age and non-verbal function.  We quantified dynamic visual scanning using kernel density analysis, and the resulting measures of fixation density, at each moment in time, allowed us to model the allocation of visual resources for both groups throughout all video scenes. Using these measures, we tested for between-group differences in dynamic visual scanning. In parallel, we categorized functionally distinct categories of social behavior occurring in the videos by means of an ethographic analysis.  We then analyzed the catalog of behaviors during frames of the videos when the visual scanning of children with ASD and TD children differed significantly.  

Results: Preliminary analyses suggest that socially functional elements of naturalistic video scenes, including gaze cues and intense facial expressions, serve as salient social signals to TD 12-24 month-olds, and drive their visual scanning more strongly than that of 12-24 month-olds with ASD. In contrast, preliminary results suggest that physical motions and vocalizations drive the scanning behavior of both children with ASD and TD children in a more similar fashion.  

Conclusions: Socially relevant functional behaviors appear to drive attention more strongly among TD children than among children with ASD. The social and communicative outcomes of children with ASD may thus result from a divergent foundation of knowledge gained from atypical allocation of attention from an early age.