Objectives: In the present study, we developed a Virtual Reality (VR) based system that systematically manipulated social parameters during interactions and examined affective and physiological variation in response to this manipulation.
Methods: Thirteen pairs of age-matched children with ASD and typically developing (TD) children (age 13-18 years), participated in this study. VR-based social interactions were designed to project virtual human characters (avatars) who displayed different eye-gaze patterns and stood at different distances while telling personal stories to the participants. We measured physiological responses (cardiovascular (ECG) and electrodermal (EDA) signals) via wearable biofeedback sensors and collected a clinical observer's report on the levels of affective state (e.g., anxiety) for each participant who completed two 1.5-hour sessions. The social parameters of interest, eye-gaze and social distance, were examined in a 4x2 design. Four types of eye-gaze (direct, averted, normal-while-speaking, and flip-of-normal) dictated the percentage of time an avatar looked at the participant. Two types of social distance, termed invasive (1.5ft away) and decorum (4.5ft away), characterized the avatar-to-participant distance. Subjective reports and physiological data collected during the task were analyzed to examine the affective and physiological variation in response to social parameter manipulation.
Results: The ASD group's physiological signals showed significant changes to trials rated as eliciting "low anxiety" (LA) versus "high anxiety" (HA). The TD children also showed significant physiological reactions to experimental stimuli for LA and HA trials in similar and different ways than their ASD counterparts. Both ASD and TD groups had significant increases in EDA (e.g., galvanic skin response), with a stronger variation for ASD between LA and HA for Invasive social distance and varying eye gaze. The experiment condition of Averted eye-gaze with distance varying elicited significant increases in ECG (e.g., Pre-Ejection-Period) for ASD but not the TD group.
Conclusions: The results show that the VR system elicits variations in affective ratings and physiological responses to changes in social experimental stimuli for ASD and TD children. This research allows comparative analysis between ASD and TD groups and may enhance our ability to understand the specific vulnerabilities in social communication of children with ASD. Work incorporating adaptive physiological monitoring into an expanded set of VR-based social interaction utilizing machine-learning mechanisms aimed at exploring specific social communication and sensory vulnerabilities of children with ASD is underway.