A Comparative Analysis of Cognitive Load Surrogates Using the Eye-Tracker in Code Comprehension

Cognitive load assessment is essential for understanding the mental effort involved in many complex tasks, such as code comprehension. While electroencephalography (EEG) is a reliable method for measuring cognitive load, its intrusive nature and practical constraints limit its use in real-world scenarios. This study investigates the potential of alternative lowintrusive physiological signals, such as those from eye-Tracking, to estimate cognitive load during code comprehension tasks under complex real conditions. Using data from a controlled experiment, we extracted eye-Tracking features and compared them with standard EEG-based cognitive load measures, analyzing their correlations and feature rankings. By segmenting the code into regions and analyzing eye-Tracking features at the region level, we identify cognitive load patterns that are specific to particular programming constructs. Notably, metrics such as revisit counts and reading time show strong alignment with cognitive load. These results suggest that eye-Tracking technology offers a promising tool for cognitive load estimation, providing a non-intrusive, practical, and cost-effective solution, especially for real-Time monitoring in software development environments.