Abstract
Abnormal visual attention to faces and difficulty interpreting eye gaze cues are core deficits of autism and have widespread consequences for social communication. Specifically, autistic people have difficulty computing the trajectory of eye gaze, understanding the referential nature of gaze, and assigning social relevance to gazed-at objects. We developed a “serious” computer game intervention (SAGA: Social Games for Autistic Adolescents) intended to improve looking time to faces and understanding about using eye gaze cues via learning within a narrative storyline and adaptive difficulty progression. Participants interact with animated human characters and discover that eye gaze cues are useful for guiding their own goal-directed behavior to solve problems in the game, simulating how these cues are discovered and used in the real world. We evaluated the effectiveness of SAGA in a hybrid Phase 1 and 2 randomized clinical trial with intervention (N=20) and waitlist control (N=20) groups of autistic adolescents. The intervention group was instructed to play the game at home 90-min/week for 2 months. SAGA was feasible, acceptable (high participant compliance), and generally safe (no adverse events). Participants completed a standardized laboratory gaze perception task pre- and post- intervention with accuracy and eye-tracking metrics as outcome measures. Participants viewed an image of a person in a complex scene looking at a target object (TO); they had to identify the TO from a list of 4 plausible alternatives from the scene. At pre-test, there was no association between looking time to faces and accuracy in either group. At post-test, however, the intervention group showed a positive association between looking time to faces and TO accuracy (no association in controls). Mean looking time to faces did not increase from pre- to post-intervention timepoints. Therefore, we conclude that SAGA improved social visual attention to faces for the purpose of interpreting eye gaze cues.