The Impact of Educational Agents on the Effectiveness of Human-Machine Collaborative Learning: A Meta-Analysis Based on Learning Motivation and Academic Performance

Educational agents, serving as pivotal mediators in human-machine collaborative learning, have been extensively deployed in teaching practice through multimodal interaction. However, their empirical efficacy remains inconclusive. This study establishes a theoretical framework for how educational agents influence cognition, emotion, and motivation, followed by quantitative analysis of 32 experiments and quasi-experiments. Results indicate that educational agents exert a significant positive effect on learning outcomes (SMD = 0.901), with this effect moderated by factors including subject domain, intervention modality, and student educational stage. It is recommended that practical applications prioritize subject-specific adaptation and refined agent design, driving agents' evolution into comprehensive learning partners that integrate cognitive, affective, and motivational support. This approach will enhance the overall efficacy of human-machine collaborative learning.