
Mixed reality (MR) integrates virtual content with the physical world, enabling users to place virtual objects in real environments and interact with or observe them. As MR technologies advance, such experiences are becoming increasingly common. However, it remains unclear how the visual and interactive representation of virtual objects influences users, and few studies have examined users’ behavioral responses to virtual objects. We investigated whether representation factors (interactivity, transparency, and size) affect users’ sense of presence and their behaviors toward the object (e.g., avoidance or displacement). Here, interactivity refers to whether users can touch the virtual object. In two experiments (desk-scale and room-scale) conducted, participants performed a reaching task toward a real target located behind a virtual object whose representation factors were manipulated. Presence and behavior were assessed using subjective ratings and objective measures from tracking data and video observations. Perceived presence varied with interactivity, transparency, and size, whereas avoidance and displacement behaviors showed no reliable differences across conditions. Nonetheless, the results suggest that behavioral responses may emerge when interaction demands are stronger or the scale of interaction is larger. Overall, representation affected perceived presence but did not reliably change avoidance or displacement behavior in this task.

Improving drivers’ risk prediction ability can reduce the accident risk significantly. The existing accident awareness training systems show poor performance due to the lack of immersive sense. In this research, an immersive educational system is proposed for risk prediction training based on VR technology. The system provides a highly realistic driving experience to driver through 360 degrees video using VR goggle. In the nearly actual driving scene, users are expected to point out every potential dangerous scenario in different cases. Afterwards, the system evaluates users’ performances and gives the corresponding explanations to help users improve safety awareness. The results show that the system is more effective than previous systems on improving drivers’ risk prediction capability.