李佳欣1 刘彩霞2*
（1.江苏师范大学 智慧教育学院，江苏 徐州 221116；2.江苏省教育信息化工程技术研究中心，江苏 徐州 221116）
【摘 要】直接洞察学习者的内在状态，是突破传统交互数据局限、实现视频学习深度个性化的关键。生理信号的采集与分析技术为此提供了实时、客观的感知路径。然而目前研究对如何系统性地利用这些数据驱动学习者的自适应学习仍缺乏清晰的路径。为此，本研究通过对2015年至2025年的43篇实证文献进行系统性文献综述，构建并应用“状态—干预—情境—效果”整合性的分析框架，系统地梳理该领域的研究进展。研究发现：①现有研究高度集中于对认知状态的测量，对情感、行为状态的客观识别严重不足；②干预策略以“静态预设”为主，仅有2项研究探索了实时自适应干预；③干预效果呈现出显著的情境依赖性，受学习者先验知识与任务知识类型的影响；④现有研究的技术应用模式多利用其进行事后“解释”，而在实时“预测”与“干预”的实证研究环节存在断层。本研究系统地揭示了该领域从“解释性研究”迈向“自适应系统”所面临的核心矛盾与关键技术断层。未来研究应通过多模态融合的方式构建学习者的全息画像，致力于发展学习状态时序预测模型，从而为跨越“解释—干预”断层、搭建真正的自适应视频学习系统奠定基础。
【关键词】视频学习；生理技术；多模态融合；自适应学习
How Do Physiological Technology Drive Adaptive Video Learning? A Systematic Literature Review of Empirical Studies
LI Jiaxin1 and LIU Caixia2*
(1. School of Smart Education, Jiangsu Normal University, Xuzhou 221116, China; 2. Educational Informatization Engineering Technology Research Center, Xuzhou 221116, China)
Abstract: Directly gaining insights into learners’ internal states is key to overcoming the limitations of traditional interaction data and achieving deep personalization in video-based learning. Technologies for the collection and analysis of physiological signals provide a real-time, objective pathway for such perception. However, a clear path for systematically leveraging this data to drive adaptive learning remains lacking. To address this, this article conducted a systematic review of 43 empirical studies published between 2015 and 2025. An integrated analytical framework of“state–intervention–context–effect”was constructed and applied to systematically synthesize research progress in this field. The findings reveal that: (1) Research is highly concentrated on the measurement of cognitive states, while objective identification of affective and behavioral states is severely insufficient. (2)Intervention strategies are predominantly“statically preset”, with only two studies exploring real-time adaptive interventions. (3) Intervention effects exhibit significant contextual dependency, influenced by learners’prior knowledge and the type of task knowledge. (4) The current application mode of technologies in existing research primarily utilizes them for post-hoc“explanation”, revealing an empirical gap in the chain of empirical research concerning real-time“prediction” and“intervention”. This article systematically uncovers the core contradictions and key technological gaps that the field faces in transitioning from “explanatory research”to“adaptive systems”. Future research could construct holistic learner profiles through multimodal fusion and focus on developing temporal prediction models for learning states, thereby laying the foundation for bridging the“explanation–intervention”gap and realizing truly adaptive video learning systems.
Keywords: video-based learning; physiological technology; multimodal data; adaptive learning
下载： 
生理技术如何驱动自适应视频学习？.pdf
