Agent Persona and Motivational Discourse: A Two-Step Process to Enhance Self-Efficacy and Learning in Virtual Environments

The expanding integration of Virtual Pedagogical Agents (VPAs) in online learning environments necessitates a rigorous understanding of the design factors that optimize their effectiveness. Grounded in Bandura’s Social Learning Theory, this research conducts a two-part analysis of how an agent's persona and discourse influence learner outcomes. Study 1 (N = 55) demonstrated that participants reported significantly greater affiliation with a peer agent than an expert (p = .001). This heightened identification was positively correlated with increased self-efficacy (p = .021). However, aligning with persistent challenges in VPA literature, this affective enhancement did not translate into improved learning performance, revealing a gap between affective and cognitive outcomes. Study 2 (N = 24) investigated whether agent discourse could bridge this gap using the peer model with either motivational or neutral scripts. Motivational discourse resulted in significantly superior improvements in both self-efficacy (p < .001) and learning performance (p = .036). Mediation analysis indicated that self-efficacy did not mediate this relationship (p = .169), suggesting parallel processing pathways. These findings propose a critical two-step design framework: a peer persona is essential for fostering identification, while motivational discourse is required to convert affective gains into tangible learning improvements for online systems.

Keywords: Agent Persona, Virtual Pedagogical Agent, Self-Efficacy, Identification, Motivational Discourse, Online Learning.

Volume 3, ISBN 978-989-758-833-4, ISSN 2184-5026, pages 2440-2449.  DOI: 10.5220/0014498100004021 
Lien éditeur : https://www.scitepress.org/ProceedingsDetails.aspx?ID=V9ESoTp7g/4=&t=1