Cognitive load patterns affect temporal dynamics of self-regulated learning behaviors, metacognitive judgments, and learning achievements

This study investigated how students' cognitive load patterns related to their self-regulated learning (SRL) and performance. One hundred and eleven medical students were required to accomplish a diagnostic task in a computer-based learning environment. The metacognitive aspects of SRL were reported by students at different points during the task, and cognitive load was rated after students finished the task. We extracted SRL behaviors from system logfiles, and students' performance was measured as diagnostic accuracy, which indicated whether they achieved an accurate diagnosis. A two-step cluster analysis was conducted to identify students' cognitive load patterns based on their perceived intrinsic, extraneous, and germane load. We found three cognitive load patterns: knowledge builders with a low-level intrinsic and extraneous load but high-level germane load, task-loaded learners with a high-level intrinsic and extraneous load but low-level germane load, and overloaded learners with high-level intrinsic, extraneous, and germane load. Moreover, we performed epistemic network analyses to compare the SRL behavior networks of the three clusters. The results showed that knowledge builders exhibited stronger connections associated with monitoring behaviors, while task-loaded and overloaded learners constructed more connections associated with help-seeking behaviors. The transitions between evaluation and solution elaboration behaviors were more frequent for knowledge builders than for the other clusters. Furthermore, knowledge builders reported higher scores in two types of metacognitive judgments: ease-of-learning and judgment-of-learning. Knowledge builders also achieved significantly higher diagnostic accuracy than task-loaded and overloaded students. These findings contributed to a more comprehensive understanding of how cognitive load interacts with SRL.