Animated pedagogical agents: Does their degree of embodiment impact learning from static or animated worked examples?

This study examined the impact of varying an animated pedagogical agent's level of embodiment in a learning environment consisting of static or animated multimodal worked examples illustrating how to solve multi-step proportional word problems. In the fully embodied condition, the agent was programmed to coordinate spoken instructional explanations with non-verbal forms of communication (locomotion, gesture, and gaze) to support learning. In the minimally embodied condition, the same agent provided the spoken instructions but remained static on the screen. In the voice-only condition, the spoken instructions were provided without an agent. Of the 174 college-age participants, those provided with the fully embodied agent produced more conceptually accurate answers to near and far transfer items than their counterparts in the voice-only condition. Moreover, participants that studied animated worked examples, where the problem's subgoals were presented sequentially over time, outperformed their peer, provided with static examples with simultaneously-presented subgoals on measures of transfer.