Participation in Pedagogical Agent Design: Effects on Training Outcomes

Abstract

Pedagogical agents have the potential to increase engagement and learning for trainees completing e-learning courses. However, little research to this date has been conducted to determine the conditions that make these agents most effective. The current study examined the role of agent design control in improving learner reactions and learning. A sample of 164 e-learners completed a Microsoft Excel training course; half of the learners were given the opportunity to design a pedagogical agent that suited their preferences, while the others were assigned an agent with predetermined features. Those who helped design their agent were randomly assigned to one of four conditions which allowed them to participate in the configuration of their agent’s (a) appearance, (b) personality, (c) feedback style, or (d) all of the above. Findings demonstrated that participation influenced scores on a post-training declarative knowledge test, although this effect depended on the type and amount of participation permitted. Specifically, feedback participation decreased declarative knowledge, while participation in multiple agent characteristics increased declarative knowledge. Contrary to expectations, participation regarding agent feedback also decreased utility reactions and self-efficacy. No type of participation influenced affective reactions or engagement. Although some forms of participation yielded trends in the expected direction, effects were not statistically significant. On the whole, this study expands the learner control literature, identifying a new form of learner control that has beneficial effects on knowledge acquisition.