The effects of personality congruence and conversational stance on human–agent interaction in virtual reality

As Embodied Conversational Agents (ECAs) become central to fostering social interactions in immersive environments, personalization has emerged as a key factor in their effectiveness. Since individuals differ in how they perceive, interpret, and respond to social cues, aligning agent personality with that of users is an important design consideration. However, how user-agent personality alignment influences social perception such as rapport and persuasion remains unexplored. This study examines how personality alignment between users and ECAs influences interaction outcomes across two debate-based conversational contexts: opinion agreement or disagreement. We designed a multimodal ECA that expressed personality through verbal and nonverbal behaviors, tailored to each participant’s Big Five profile across three conditions: Similar, Neutral, and Inverse. Results revealed that personality similarity improved perceived relational quality, especially in agreement contexts. We further found that these effects were differentially shaped by individual personality traits, indicating that responses to personality-adaptive ECAs vary systematically across users. These findings provide design insights for personality-adaptive ECAs, highlighting the importance of personalized interaction strategies that account for both conversational context and user personality traits.