Improving sensitivity to eye gaze cues in autism using serious game technology: study protocol for a phase I randomised controlled trial

Abstract

Introduction: Autism spectrum disorder (ASD) is characterised by impairments in social communication. Core symptoms are deficits in social looking behaviours, including limited visual attention to faces and sensitivity to eye gaze cues. We designed an intervention game using serious game mechanics for adolescents with ASD. It is designed to train individuals with ASD to discover that the eyes, and shifts in gaze specifically, provide information about the external world. We predict that the game will increase understanding of gaze cues and attention to faces.

Methods and analysis: The Social Games for Adolescents with Autism (SAGA) trial is a preliminary, randomised controlled trial comparing the intervention game with a waitlist control condition. 34 adolescents (10-18 years) with ASD with a Full-Scale IQ between 70 and 130 and a minimum second grade reading level, and their parents, will be randomly assigned (equally to intervention or the control condition) following baseline assessments. Intervention participants will be instructed to play the computer game at home on a computer for ~30 min, three times a week. All families are tested in the lab at baseline and approximately 2 months following randomisation in all measures. Primary outcomes are assessed with eye tracking to measure sensitivity to eye gaze cues and social visual attention to faces; secondary outcomes are assessed with questionnaires to measure social skills and autism-like behaviours. The analyses will focus on evaluating the feasibility, safety and preliminary effectiveness of the intervention.

Ethics and dissemination: SAGA is approved by the Institutional Review Board at Pennsylvania State University (00005097). Findings will be disseminated via scientific conferences and peer-reviewed journals and to participants via newsletter. The intervention game will be available to families in the control condition after the full data are collected and if analyses indicate that it is effective.

Trial registration number: NCT02968225.

Keywords: mental health.

Conflict of interest statement

Competing interests: EMW was at the Department of Psychology at Pennsylvania State University when she contributed to this work. She is currently working for Daybreak Games in San Diego, California.

© Author(s) (or their employer(s)) 2018. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Figures

Figure 1

CONSORT diagram for SAGA protocol. ASD, autism spectrum disorder; CONSORT, Consolidated Standards of Reporting Trials; SAGA, Social Games for Adolescents with Autism.

Figure 2

Screenshots from multiple training conditions in the intervention game. (A). The avatar is instructing the participant to select 1 of 5 possible drawer locations using pointing, shoulder direction, head direction, and gaze cues in a room scene. (B). The avatar is directing the participant to select 1 of 6 possible tools to put on the peg board using pointing, shoulder direction, head direction, and gaze cues in a tool shed scene. (C). The two avatars are engaged in an episode of joint attention on the bowl and are inviting the participant to select it from 5 possible objects using pointing, shoulder direction, head direction, and gaze cues in a library scene.

Figure 3

Schematic illustration of the intervention game structure. The game is designed to train learning about three functional uses of eye gaze cues including the use of gaze to reference locations and objects in the world via a single informant and in episodes of joint attention between multiple informants (A). The game is organised around three sequential phases. The tasks in phase 1 are structured to help participants learn that eye gaze is an important cue to solving problems in the game. The tasks in phase 2 help participants learn to estimate precise gaze trajectories by making target gazed-at objects closer together and to ignore salient objects that are not the target gazed-at object. Episodes of joint attention are also introduced in phase 2 in which participants have to determine the target object that two avatars are looking at together. This is difficult because the timing of the non-verbal cues to identify the object is not perfectly synchronous between the two avatars. In phase 3, the tasks are structured around helping participants learn the difference between a goal-directed gaze cue (eg, looking at a target object to solve a puzzle) and a non-goal-directed gaze cue (eg, looking around at all the objects before deciding which one to select). To complete a phase of the game, participants must finish all levels within a phase. Each phase has multiple levels (B). Levels are defined by the number of non-verbal cues avatars use to guide participants to solve puzzles in the game. Easy levels have multiple cues. Level progression increasingly focuses learning to use eye gaze cues exclusively by stripping away other cues. Within each level, there are six stages (C). Each stage represents the number of potential objects or locations that the participant has to discriminate between based on the cue from the avatar. In the easiest stage, the participant chooses between two objects or locations that the avatar is pointing, directing shoulders, head, and gaze to (as in level 1), whereas in stage 6, the participant chooses between six possible objects or locations that the avatar could be referring to with the non-verbal cue(s). Within each stage, participants have five trials. They must perform with 80% accuracy to advance to the next stage, and they must finish all stages within a level before they can progress to the next level within a phase. When they do not reach 80% accuracy within a stage, they are returned to the previous stage to reify the learning where they were recently successful. Sometimes that means they are returned to later stages of previous levels.