The Impact of Virtual Agent Locomotion and Postures on Social Perception and Brain Activity in MR

When people interact with virtual characters, the character's non-verbal cues can change the user's perception of and engagement with it. In this work, we investigate the impact of virtual agent locomotion and emotional posture on user engagement, as measured by subjective questionnaires and EEG signals. We found the EEG-based engagement index was significantly lower in the postural agent condition, which was opposite to the questionnaire-based engagement result. From these results, we differentiate social engagement from task engagement and discuss their influence on overall perceived engagement.