Divergent Perceptuomotor Recalibration in Virtual Reality and Video-Passthrough Mixed Reality on the Same Head-Mounted Display
TL;DR
Compared visually guided manual pointing in VR and MR-VPT using the same headset suggests that perceptuomotor recalibration is shaped not only by display-level constraints but also by the action-relevant visual information available within the interaction environment.
Abstract
Virtual reality (VR) and video-passthrough mixed reality (MR-VPT) can be delivered on the same headset, but it is unclear whether these two interaction modalities produce comparable perceptuomotor behavior. Although delivery through the same headset controls many display-level characteristics, VR and MR-VPT differ in both their visual-feedback pipelines and the action-relevant visual information available to guide movement, such as whether the surrounding environment and the user's body are synthetically rendered or preserved through passthrough. This study compared visually guided manual pointing in VR and MR-VPT using the same headset. Forty adults were assigned to either a VR or MR-VPT group and completed a pointing task in the physical, unmediated reality (UR) before and after performing the same task in their assigned XR modality. The analyses revealed numerous important insights. First, groups had comparable baseline performance in UR prior to XR exposure. Second, the two modalities produced opposite initial biases during XR exposure: the VR group undershot the target, whereas the MR-VPT group overshot. Third, although both groups reduced error with continued exposure to XR, there were differences in the patterns of adaptation: the VR group adapted more slowly and exhibited stronger distance-dependent undershooting than the MR-VPT group. Finally, after XR exposure, both groups showed significant undershooting aftereffects with the VR group retaining a persistent residual undershoot, suggesting incomplete de-adaptation. The results suggest that perceptuomotor recalibration is shaped not only by display-level constraints but also by the action-relevant visual information available within the interaction environment. These findings highlight the need to validate perceptuomotor interaction techniques separately in VR and MR-VPT, even when they are implemented on the same hardware.