Analyzing Learners' Visual Behavior Patterns Related to Information Processing and Cognitive States in Problem-Solving Learning

Abstract

This study explores the possibility of evidence-based educational prescriptions by analyzing learners' information processing and cognitive states in problem-solving learning. Learners' scanpath patterns were regarded as information processing processes, and groups were differentiated according to the similarity of learners' scanpaths. I collected and analyzed multimodal data including visual behaviors, logs, and stimulated recall responses from 60 university students and found differences in cognitive states between groups. Moreover, the predictive model confirmed that pupil dilation size and selective attention, which are indicators of cognitive states, statistically significantly predicted problem-solving outcomes. This study's results confirm that visual behavior can be a useful indicator for detecting real-time changes in information processing and cognitive states during learners' problem-solving processes. This finding is significant in that it provides pedagogical implications for designing individualized instruction customized to each learner from the perspective of multimodal learning analytics.