Promoting social attention in 3-year-olds with ASD through gaze-contingent eye tracking.

Young children with autism spectrum disorder (ASD) look less toward faces compared to their non-ASD peers, limiting access to social learning. Currently, no technologies directly target these core social attention difficulties. This study examines the feasibility of automated gaze modification training for improving attention to faces in 3-year-olds with ASD. Using free-viewing data from typically developing (TD) controls (n = 41), we implemented gaze-contingent adaptive cueing to redirect children with ASD toward normative looking patterns during viewing of videos of an actress. Children with ASD were randomly assigned to either (a) an adaptive Cue condition (Cue, n = 16) or (b) a No-Cue condition (No-Cue, n = 19). Performance was examined at baseline, during training, and post-training, and contrasted with TD controls (n = 23). Proportion of time looking at the screen (%Screen) and at actresses' faces (%Face) was analyzed. At Pre-Training, Cue and No-Cue groups did not differ in %Face (P > 0.1). At Post-Training, the Cue group had higher %Face than the No-Cue group (P = 0.015). In the No-Cue group %Face decreased Pre- to Post-Training; no decline was observed in the Cue group. These results suggest gaze-contingent training effectively mitigated decreases of attention toward the face of onscreen social characters in ASD. Additionally, larger training effects were observed in children with lower nonverbal ability, suggesting a gaze-contingent approach may be particularly relevant for children with greater cognitive impairment. This work represents development toward new social attention therapeutic systems that could augment current behavioral interventions. Autism Res 2020, 13: 61-73. © 2019 International Society for Autism Research, Wiley Periodicals, Inc. LAY SUMMARY: In this study, we leverage a new technology that combines eye tracking and automatic computer programs to help very young children with ASD look at social information in a more prototypical way. In a randomized controlled trial, we show that the use of this technology prevents the diminishing attention toward social information normally seen in children with ASD over the course of a single experimental session. This work represents development toward new social attention therapeutic systems that could augment current behavioral interventions.