As embodied agents assume increasingly social roles in VR, visual realism and embodiment alone are insufficient for users to perceive emotional attunement, support, and humanness. Prior work shows that verbal attunement and non‑verbal mimicry each improve social evaluations of agents, yet behavioral mimicry has been studied mainly outside real‑time conversational contexts, leaving a gap in understanding how users respond when an agent simultaneously generates context‑responsive dialogue and adapts its non‑verbal behavior. To address this, we built an embodied AI counselor that integrates conversational AI with real‑time facial‑expression and posture mimicry, capable of delivering either verbally attuned or neutral responses. In a 2 × 2 within‑subjects study with 20 participants, we manipulated verbal attunement (attuned vs. neutral) and behavioral mimicry (present vs. absent). Results indicated that verbal attunement was the most reliable driver of perceived empathy. Behavioral mimicry showed a marginal positive relationship with perceived humanness, and higher mimicry exposure yielded exploratory positive associations with empathy, positivity, and humanness, especially among female participants. Overall, multimodal synchrony is not a simple additive design strategy; careful consideration of how verbal and non‑verbal behaviors are combined is needed when designing empathic conversational agents in VR.
Review