Autism Spectrum Disorder (ASD) children often experience substantial difficulty adapting to unfamiliar environments and complex sensory input. To support skill execution and transfer across varying contexts, we developed PanoVR, a panoramic virtual reality (VR) system that gradually modulates multi-sensory stimulation. In a four-week longitudinal study, 11 children with ASD completed training sessions progressing from familiar to semi-familiar and unfamiliar scenes. Behavioral results showed continuous improvement in task performance, while behavioral patterns remain stable, indicating that increased environmental complexity did not introduce additional variability. Functional near-infrared spectroscopy (fNIRS) analyses revealed significant reductions in global effective connectivity (EC) and decreased between-participant variability, suggesting a shift toward more efficient, lower-load coordination across brain regions. These findings indicate that gradual multi-sensory modulation can enhance behavioral performance and neural efficiency, enabling strategies learned in familiar settings to transfer to more challenging contexts. The results provide practical implications for designing adaptive VR-based interventions for ASD.
ACM CHI Conference on Human Factors in Computing Systems