The interaction of learner expertise and instructional role of a pedagogical agent on learner perception of agent, perceived cognitive load and task performance

Pedagogical agents are computerized animated characters that interact with the learner as a conversational partner in computer-based instructional environments. Computer-based instruction (CBI) utilizing pedagogical agents is becoming a popular tactic for instructional delivery. Because of this new focus on the use of agents to deliver CBI, research examining the most effective design of these environments is called for. Along with design considerations of the instructional messages delivered, designers must be cognizant of the impact of affective perceptions of agents on learning.;The purpose of this study was to examine the interaction of learner expertise and pedagogical agent instructional roles on perceptions of the agent, perceived cognitive load and task performance. In this study, agents interacted with the student as either an expert or peer tutor. Instructional messages were tailored to the role of the agent. Perceptions of agents were tested in the context of social learning theory (Bandura, 1977) and the social agency effect (Moreno, Mayer, Spires, & Lester, 2001). Effects of instructional messages were tested in the context of cognitive load theory (Sweller, 1988) and the expertise reversal effect (Kalyuga, Ayres, Chandler, & Sweller, 2003).;Participants were 75 undergraduate students in a classroom technology integration course in an urban university setting in the Mid-South. Participants were randomly assigned to one of the two pedagogical agent tutoring roles, where they worked on an instructional planning task using the Integrating Technology through Inquiry model (Morrison & Lowther, 2001). Data were analyzed using 2 (agent role) x 2 (learner expertise) factorial ANOVAs for each dependent variable and regression analysis.;Findings showed that learners at both high and low expertise levels rated the peer agent higher in perceptions of pedagogical efficacy and attitude (attitude differences were not statistically significant). This supports research describing the social agency effect (Moreno et al., 2001). It also appears learner expertise does not have an impact in terms of the instructional message but does influence which variables contribute to task performance. Findings regarding effects of messages on cognitive load and the expertise reversal effect were inconclusive. Suggestions for designers of pedagogical agent environments are discussed.