Spontaneous visual perspective-taking in human–robot interaction as a methodological artifact of spatial stimulus arrangement

Abstract While prior research has demonstrated spontaneous visual perspective-taking (VPT) towards robots using static images or videos of existing robots, little is known about whether VPT occurs with robotic avatars or in immersive environments like virtual reality (VR). The purpose of this research was to extend the existing research to robotic avatars and VR as a medium for interaction. Study 1 confirmed that participants spontaneously adopt the perspective of both robotic and human avatars in static images, with no significant difference in VPT between the two agents. Study 2 compared VPT in video and VR conditions, revealing unexpectedly higher VPT in the video condition, despite VR’s greater sense of copresence. We suspected that the spatial context of stimulus presentation played a critical role. Therefore, Study 3 further explored this by manipulating the placement of stimuli in the robot’s versus the shared environment, revealing that VPT was higher when stimuli were presented exclusively in the robot’s environment. This finding suggests that earlier VPT research may have a critical methodological limitation, as it did not account for spatial context in real human–robot interaction scenarios, which could limit its applicability to real-world robot interactions.