Refining Mathematical Task Difficulty for Accurate Mental Workload Estimation

Abstract

Accurate classification of mathematical difficulty is essential for quantifying mental workload (MWL) in cognitive research. This study introduces a novel method that defines difficulty based on the number of interim values stored during each stage of verbal arithmetic. Twenty-six participants completed six levels of math tasks, with task performance and eye-tracking data (fixations and saccades) collected to evaluate the framework. Performance results showed significantly higher accuracy in Levels 1 and 2 and meaningful differences between Levels 3, 4, and 6, validating the method’s ability to capture task difficulty. Eye-tracking data identified distinct MWL patterns but produced different groupings, suggesting that physiological metrics and performance capture complementary aspects of MWL. Fixation and saccade measures effectively distinguished lower from higher tiers, particularly separating Levels 1–2 from 3–6. Together, these findings support the robustness of the classification framework and underscore the value of using multiple modalities to evaluate MWL in mental arithmetic tasks.