Scholarlink

<p>A widespread issue in video-based learning is low learner engagement. Based on the Interactive, Constructive, Active, Passive (ICAP)framework, most learners tend to stay in a&#xa0;low-investment passive mode. Because the cognitive state of passive learning is crucial, identifying students' learning modes for targeted interventions is of great concern.&#xa0;Using attentional features, particularly evaluating joint attention via inter-subject correlation (ISC), offers a highly promising approach for identifying these modes. By manipulating learning expectations, this study explored attentional characteristics by synchronously collecting eye-tracking data from a teacher during recording and 94 students during video viewing, assessing teacher-student joint attention via ISC. The results revealed that learners in the constructive and active groups exhibited significantly lower ISC than those in the passive group, indicating a shift from a passive, synchronous reception mode to an autonomous, top-down cognitive engagement. Moreover, constructive learning was characterized by a lower frequency of text-image transitions and a higher average pupil diameter, these physiological metrics may reflect the depth of internal processing and cognitive effort. Mediation analysis revealed that ISC mediated the effect of learning modes on performance. Based on the aforementioned eye-tracking features and learning experiences, we constructed machine learning classifiers, finding that the XGBoost model achieved the highest accuracy, with&#xa0;ISC and subjects' learning experience being the most influential predictors in the model. This study demonstrates how to quantify different levels of engagement through physiological metrics, enriching empirical research on the ICAP framework and offering practical implications for personalized video-based learning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Attentional characteristics and learning mode recognition of different learning patterns in video-based learning

  • 期刊论文

  • 发表日期：

    2026年7月9日

查看全文
  • Yang Peng,
  • Han Bai,
  • Riman Ga,
  • Qing Lai,
  • Zheng Liang,
  • Qingbai Zhao

摘要

A widespread issue in video-based learning is low learner engagement. Based on the Interactive, Constructive, Active, Passive (ICAP)framework, most learners tend to stay in a low-investment passive mode. Because the cognitive state of passive learning is crucial, identifying students' learning modes for targeted interventions is of great concern. Using attentional features, particularly evaluating joint attention via inter-subject correlation (ISC), offers a highly promising approach for identifying these modes. By manipulating learning expectations, this study explored attentional characteristics by synchronously collecting eye-tracking data from a teacher during recording and 94 students during video viewing, assessing teacher-student joint attention via ISC. The results revealed that learners in the constructive and active groups exhibited significantly lower ISC than those in the passive group, indicating a shift from a passive, synchronous reception mode to an autonomous, top-down cognitive engagement. Moreover, constructive learning was characterized by a lower frequency of text-image transitions and a higher average pupil diameter, these physiological metrics may reflect the depth of internal processing and cognitive effort. Mediation analysis revealed that ISC mediated the effect of learning modes on performance. Based on the aforementioned eye-tracking features and learning experiences, we constructed machine learning classifiers, finding that the XGBoost model achieved the highest accuracy, with ISC and subjects' learning experience being the most influential predictors in the model. This study demonstrates how to quantify different levels of engagement through physiological metrics, enriching empirical research on the ICAP framework and offering practical implications for personalized video-based learning.