Patterns of Visual Exploration of Complex Social Scenes Are Increasingly Deviant over Time in Preschoolers with ASD: A Longitudinal Eye-Tracking Study

Poster Presentation
Friday, May 11, 2018: 5:30 PM-7:00 PM
Hall Grote Zaal (de Doelen ICC Rotterdam)
N. Kojovic1, M. Franchini2, Z. Daniela3, H. F. Sperdin4 and M. Schaer5, (1)Developmental Imaging and Psychopathology Lab, UNIGE, Geneva, Switzerland, (2)Dalhousie University / IWK Health Centre, Halifax, NS, Canada, (3)Medical Image processing Laboratory, University of Geneva, Geneva, Switzerland, (4)Developmental Imaging and Psychopathology (DIP) Lab, Office Médico-Pédagogique, Department of Psychiatry, University of Geneva School of Medicine, Geneva, Switzerland, (5)Developmental Imaging and Psychopathology Lab, University of Geneva, Geneva, Switzerland
Background: Numerous studies using eye tracking have highlighted atypical visual exploration patterns in individuals with ASD, e.g. less focus on eyes and faces compared to typically developing (TD) individuals. Little is known about the mechanisms driving the emergence of these atypical patterns and their change with age.

Objectives: We developed a method that defines age-appropriate dynamic “norms” of visual exploration of complex social scenes. This method allows definition of characteristic gaze behavior for both groups, and quantification of its change with age (i.e. change in group focus precision), thus discriminating normal and atypical developmental processes. It allows to, without any a priori, measure for each child with ASD the deviance in his gaze pattern from the “norm”, based on a gaze of TD children.

Methods: A 3-minute cartoon was displayed on a Tobii eye-tracker device for 16 ASD males (aged 3.5±0.3) and 17 TD males (aged 3.3±0.3) and same procedure was repeated a year later. A measure of dispersion of gaze data was calculated for each frame of the video, in both groups. For each frame of the video we created “normative” gaze pattern distribution by employing kernel density distribution estimation on the raw gaze data of TD individuals. We then calculated the Proximity index-PI for each patient, and averaged these values for the duration of the video. Higher values indicate the visual exploration of the individual is being more similar to the one of TD subjects. The PI was further correlated with clinical behavioral data and a measure of the complexity of social plot (number of interacting characters).

Results: While TD gaze behavior showed increasing focus with age (less dispersion, p < 0.0000 h2= -0.47), ASD children showed a more dispersed gaze pattern a year later (p < 0.0001, h2= -0.04). We found positive correlation between the change in PI and change in measures of cognition (R2 = 0.57, p = 0.008). In a group of males with ASD, higher PI characterized scene sequences with less social content (one character alone), while in sequences with two or three characters interacting this index was lower (p = 0.004).

Conclusions: Our results show that this data-driven method can be used to characterize change in visual exploration in typical and atypical development. In one year TD preschoolers showed increased attunement in visual exploration of complex social scenes. The opposite was observed in our group of males with ASD. This finding highlights the importance of intense early intervention to help preventing the derailment