角色化多智能体支持议题式学习的模式构建与效果检验OA

In an era of value pluralism,cultivating future citizens with sound personality and a sense of responsibility has be-come an important mission of education.Issue-based learning adopts authentic dilemmatic,conflicting and public issues as carriers.It promotes disciplinary knowledge learning through value clarification and rational argumentation and guides students to form sound val-ue judgments amid clashes of diverse values.Its learning process is characterized by"issue orientation-problem driving-rational de-liberation-social participation".However,conventional technological support suffers from limitations such as over-prescribed proce-dures,delayed feedback on argumentation,and insufficient coordination of multiple technical resources.To address these limitations,this study introduces large language model-driven multi-agent technology,constructs the"Yixiang Tiankai"multi-agent system,and defines four types of agents:Issue Tracer,Argumentation Modeler,Perspective Challenger,and Deliberation Navigator.Drawing on this system,this study designs an issue-based learning model supported by multi-agents,advancing classrooms toward a new form of mul-ti-agent collaboration among teachers,students and machines.A quasi-experimental study was carried out within primary school Ethics and the Rule of Law courses.The findings reveal that the core value of multi-agent support lies in improving the quality of hu-man-AI interaction.Specifically,role differentiation and functional specialization improve the matching between cognitive support and students' cognitive needs.The heterogeneous perspectives of multiple agents generate productive cognitive tension and stimulate higher-order cognitive activities including perspective taking,critical rebuttal,and deliberative decision-making.At the group argu-mentation level,multi-agent support optimizes the structural relations among claims,evidence,reasoning and rebuttals and thereby improves the quality of argumentative dialogue.Regarding learning performance,its benefits are more salient in tasks requiring high cognitive demands such as argumentation scripting and solution design.Its effects on critical thinking dispositions differ across di-mensions,with the most prominent promotion observed in reflective skepticism.