Virtual agents with personality: Adaptation of learner-agent personality in a virtual learning environment

Abstract

Virtual agents are artificial intelligent artifacts that mimic natural conversations with users. The media equation posits that human-agent interaction mirrors the social cues prevalent in human-to-human relationship. As such, personality adaptation-convergence (similarity-attraction) and divergence (complementary-attraction) in human-agent interaction have been explored, particularly in the context of e-commerce, robot and scenario-based simulation. The present work extends prior studies by investigating personality adaptation in learner-agent interaction with respect to a virtual learning system. Pursuant to this goal, an introverted and an extroverted pedagogical agent were developed, as operationalized through vocal and animation parameters, and were then assessed by forty introverted and forty extraverted learners in a 2 (agent personality type) X 2 (learner personality type) experiment. The results of this study provided clear evidences in support of the complementary-attraction principle; emotional and motivational aspects of learning were significantly enhanced when the pedagogical agent exhibited personality (extraversion type) that complemented the learner's own. Theoretical and practical implications are discussed in this paper.