Perception of basic emotions from facial expressions of dynamic virtual avatars

Abstract

Virtual Reality experiences featuring realistic Virtual Humans with convincing facial expressions are a useful tool to improve social skill in humans. For this reason several investigations have been carried out on the recognition of virtual avatar emotions, based on dynamic and static facial cues originated by basic emotions developed by Ekman. Dynamism and realism of facial expressions are both important aspects of the process of face-to-face interaction in everyday life. In this paper we present the results of a research aiming at investigating the impact of the combination of dynamic facial expressions corresponding to particular emotions with a high level of realism of virtual faces. A study where we have measured the level of intensity in the correspondence between facial expressions of virtual avatars and emotional stimuli perceived by an observer was carried out on two groups of participants with different expertise in Virtual Reality. Results show a high level of intensity in this correspondence in both groups through the evaluation of two variables: time response and the score assigned to each emotion. We suggest that the use of dynamic virtual avatars offers advantages for studying emotion recognition in a face in that they recreate a realistic stimuli in emotion research.