Specific Events That Impact the Topography of Salience When Individuals with and without ASD View Naturalistic Social Scenes

Background: Social interactions are dynamic events that unfold in real time and in complex environments. When faced with open-ended, fast-paced scenes, knowing both what is important to process and when something is important to process is critical for adaptive social action. Research from our laboratory has shown that typically-developing (TD) individuals are remarkably attuned to what is most relevant to process: during open-ended viewing of complex scenes, visual scanning of TD viewers converges on a common location during more than 80% of viewing time. By contrast, individuals with ASD encounter great difficulty when faced with open-ended social scenes, showing less attunement to socially relevant information as scene complexity increases. The current study expands on these results by examining how specific content within a social scene may guide or fail to guide visual scanning in a group of TD viewers and viewers with ASD.

Objectives: To examine how specific elements within naturalistic social scenes (e.g. vocalizations and facial expressions of varying affect) elicit or fail to elicit convergent visual scanning in TD adolescents and adolescents with ASD.

Methods: Adolescents with ASD (mean age = 16.67 (3.92) years; n = 21) and TD controls (mean age = 16.86 (4.5); n = 17) were matched on age and verbal function. Eye-tracking data were collected during viewing of video scenes depicting realistic social interactions. We used kernel density estimation to quantify the level of convergence of visual scanning at each moment in time for each group. The onset and offset of specific events (positive, negative, and neutral facial expressions and vocalizations) were coded on a frame-by-frame basis. Level of convergence in visual scanning corresponding to each coded event was examined for both groups.

Results: Preliminary analyses reveal higher visual convergence in visual scanning among TD viewers compared with ASD viewers across all categories of scene content. For TD viewers, results showed a significant effect of variation in affective content on levels of scanning convergence. For ASD viewers however, there was no significant effect of affective content on levels of convergence in visual scanning.

Conclusions: Variation in affective content is an important factor that affects the attention of TD adolescents when viewing scenes of dynamic social interactions. By contrast, the level of convergence in visual scanning among viewers with ASD is not significantly modulated by changes in affective content. Reduced modulation in visual scanning due to changes in affective content in ASD may be accompanied by increased attention to socially irrelevant distractors. Future research will utilize similar methods to identify unique aspects of scene content that drives convergence in visual scanning in ASD viewers.