Navigating Communication Patterns and Personalities in User Preference During Human-Agent Interaction

Abstract: This study delves into the dynamics of human-agent interaction (HAI), specifically focusing on how communication patterns and personalities (both users’ and virtual ones) influence user preferences of Embodied Virtual Agents (EVAs). Under the Big Five personality framework, we designed two personality-wise opposite EVAs - a shy SA and a confident CA. We then examine such effect of communication pattern and personality matching on user preferences. The experiment involved an online survey with 87 participants, consisting of questionnaires, video evaluations, and interactive tasks. Results demonstrated significant differentiation in the agents' perceived personalities and revealed user preference for SA, although the reasons remain to be elucidated. The preliminary findings on the simultaneously effective factors of similarity-attraction principle and complementary-attraction principle are unprecedented and will be further investigated in our future work.