산수문제 해결 과정에서의 동공 반응과 수행력: 시행별 정답률과 평균 정답률의 상호작용에 대한 다층모형 분석

초록

The purpose of this study is to explore the potential of pupil response as a performance indicator beyond cognitive load by analyzing repeatedly measured pupil responses using a multilevel model (MLM). To this end, pupil time-series data measured during arithmetic problem-solving were utilized, and the area of pupil response was calculated as the area under the curve (AUC). Task complexity and trial-level accuracy were set as within-person (Level 1) variables, while average accuracy rate—the mean of accuracy across all trials—was set as the between-person (Level 2) variable. The study involved 45 participants, providing a total of 3,240 data points from 72 trials per individual for analysis. The results of the multilevel analysis showed that the model specifying random slopes for task complexity and trial-level accuracy was selected as the final model. Furthermore, pupil AUC significantly increased as task complexity and average accuracy rate increased, whereas a higher trial-level accuracy was associated with a decrease in pupil AUC. Notably, the cross-level interaction between trial-level accuracy and average accuracy rate was significant, confirming that the impact of trial-level accuracy on pupil response varies depending on an individual's overall performance level. These findings suggest that pupil response can serve as a physiological indicator reflecting differences in individual performance levels and decision-making processes beyond simple cognitive load. This study demonstrates the potential for using pupil response to precisely identify a learner's performance state and provides basic data for developing physiology-based feedback indicators to supplement instructor-learner interactions in online educational environments.

키워드