A Methodological Framework for Capturing Cognitive-Affective States in Collaborative Learning

2025-07-01

AI Summary

This study addresses the challenge of non-intrusively identifying individuals’ real-time cognitive-affective states in collaborative learning settings. We propose a video-assisted self-report framework grounded in retrospective cue-based recall (RCR), integrating a predefined cognitive-affective state taxonomy with prompt-guided annotation to minimize disruption to natural collaboration. By comparing label frequency and temporal distributions across reporting modalities, we uncover dynamic patterns in emotion and attention evolution during group interaction. Our approach significantly enhances ecological validity and scalability of state annotation, offering a high-fidelity, low-interference paradigm for educational data mining. It enables fine-grained modeling and timely intervention in adaptive learning systems.

Technology Category

• Cognitive Modeling & Cognitive Systems: Affective Computing
• Computer Vision: Video Understanding & Activity Analysis
• Machine Learning: Multimodal Learning

Application Category

• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Identification of affective and attentional states of individuals within groups is difficult to obtain without disrupting the natural flow of collaboration. Recent work from our group used a retrospect cued recall paradigm where participants spoke about their cognitive-affective states while they viewed videos of their groups. We then collected additional participants where their reports were constrained to a subset of pre-identified cognitive-affective states. In this latter case, participants either self reported or reported in response to probes. Here, we present an initial analysis of the frequency and temporal distribution of participant reports, and how the distributions of labels changed across the two collections. Our approach has implications for the educational data mining community in tracking cognitive-affective states in collaborative learning more effectively and in developing improved adaptive learning systems that can detect and respond to cognitive-affective states.

Problem

• Identifying cognitive-affective states in collaborative learning groups
• Comparing self-reported vs probe-constrained state reporting methods
• Improving adaptive learning systems by tracking cognitive-affective states

Innovation

• Retrospect cued recall paradigm for states
• Constrained reports on pre-identified states
• Analysis of label frequency and distribution