When humans encounter other humans in public spaces, they effortlessly avoid collisions and implicitly negotiate space usage. We investigate how mutual gaze understanding contributes to such incidental and brief interactions. To balance realism and experimental control, we conduct most experiments in virtual reality (VR). In a first step, we precisely measure eye, head and body movements in humans shifting their gaze to peripheral targets while walking. We transfer the measured data to virtual humans (avatars), whose realism we evaluate in a separate experiment and then use them for all subsequent VR experiments. In a first series of VR experiments, we consider encounters where interactions between participant and avatar are not necessary. Specifically, the participant encounters a series of avatars in a corridor of a virtual building and we study how the avatars’ gaze influences the participants’ gaze and their actions. In variants of the experiment, we ask participants to conduct simple, everyday tasks and assess how this affects gaze interactions with the avatars. To validate the VR setting, we repeat part of the experiment in the real world where participants encounter confederates while negotiating a pre-defined path through a building. Importantly, the VR used closely resembles the building in which the real-world experiment is performed. In a second series of VR experiments, we study situations in which a space conflict is imminent. In the basic setting, participant and avatar approach an opaque sliding door from opposite sides. Once the door opens, one of the actors has to give priority to the other, so that both can proceed. We study how mutual gaze is used to negotiate these situations, and how this is modulated by spatial constraints on either actor and environmental context (e.g., which actor enters or exists a confined space like an elevator). Together, our experiments will provide insight as to how gaze expression and mutual gaze understanding aid humans to move smoothly and safely though public spaces. While the present project focuses on dyadic (two-person) interactions, the approach is readily extendable to multi-user scenarios, such as moving through a crowded train station.

Deutsche Forschungsgemeinschaft (DFG) - Project number 563097773