Social virtual reality elicits natural interaction behavior with self-similar and generic avatars

Abstract

Social Virtual Reality (VR) allows to interact in shared virtual environments while embodying computerized avatars which display behavior in real-time. The technique mimics real social interactions in its preservation of the spatial relatedness of social gaze and other facets of non-verbal behavior, but the extent to which people behave naturally in such artificial situations remains largely unknown. Here we show in 128 participants who interacted in dyads that the coordination of gaze and speaking behavior closely follows patterns known from face-to-face interactions: eye gaze to a partner's eye region was relatively enhanced while listening compared to while speaking and at the end of a speaking turn compared to the beginning of a turn. Gaze, speaking and smiling behavior were sensibly adapted to differing conversation topics (small talk, personal talk, talk about conflicting opinions). In contrast to written communication on the internet, anonymization – here realized using generic as opposed to self-similar avatars – was not associated with behavioral disinhibition or any differences in subjective experience, possibly due to a closeness-generating effect of direct eye contact despite the concealment of one's own and the interaction partner's identity. Our results indicate that social VR elicits natural interaction behavior and may be used to implement anonymized face-to-face interactions without the negative side-effects often associated with anonymization.