Can Virtual Reality be Used for Behavioral Analysis in Autism Spectrum Disorder? a Simulation for Interpersonal Distance Preference Assessment

Background: Virtual reality technology (VRT) is frequently addressed as a possible approach for rehabilitation in individuals with autism spectrum disorder (ASD) by promoting improvement in social interaction abilities and executive function. However, few studies use fully immersive virtual reality setups and evidence on the ability of ASD subjects to interact in such environments in the same manner as they do in the real world is scarce.

Objectives: In this study we aim to evaluate the potential of VRT to assess behavioral measures in the study of interpersonal distance preference in ASD, interacting with an adult for conversation.

Methods: Five male teenagers with high-functioning ASD (9-12years, 10.8±1.64) and four neurotypical male teenagers (8-15years, 11.25±2.99) performed a stop-distance paradigm in two environmental settings: a real environment (open room) and its virtual replication. Participants were instructed to stop (or request it from the other person or avatar) when the interpersonal distance was comfortable for having a conversation. Two experimenters conducted the study: one male and one female of ages 25 and 30, respectively, without any previous contact with the participants. Once the participant established the distance, a waist-to-waist measure was taken with a digital laser measure. In the virtual replication, two virtual humans (VH) with the experimenters’ apparent age were used. The experiment varied randomly in walking mode (approaching: starting away and walking towards; receding: starting close and walking away), the walking person (experimenter or participant), experimenter’s gaze orientation (direct or averted) and gender of the experimenter. The same conditions were imposed in the virtual setup, where the user wore a head-mounted stereoscopic display (Oculus® Rift) and its position in the room was tracked via an infrared emitter (WorldViz® PPT), placing the participant in the virtual room matching his real position. When the participant established the distance to the VH, it was registered based on the virtual positions of him and the VH. Values were transformed into the eye contact distance. Statistical analysis were conducted with α=0.05.

Results: Distances in the control group, for both real and virtual setups, matched a normal distribution (p=0.2 for real and p=0.10 for virtual setups), whereas ASD group did not (p<0.01 for both settings). Positive significant Spearman’s correlations were found between real and virtual measured distances, for both ASD (ρ(79) = 0.58, p<0.01) and controls (ρ(63) = 0.39, p<0.01). In the ASD group, the real and virtual setups resulted in statistically equal distance evaluations (Wilcoxon signed ranks test: ṡ+=38.58; ṡ-=43.38; Z=-1.113; p=0.27; N=80). When comparing with the control group (Wilcoxon-Mann-Whitney test), the mean interpersonal distance of ASD group differed statistically in both real (U=2007; W=5247; p=0.026) and virtual setups (U=1812; W=5052; p<0.01).

Conclusions: Results indicate that the virtual setup achieved comparable results with the real setup. Moreover, differences from the control group found in both setups show that the computerized simulation was able to identify the deficits observed in the real environment. Such results enforce the idea that fully immersive technology might aid in the study and intervention in ASD, expanding the possibilities of rehabilitation techniques when incorporating this type of systems.