Title: Evaluating the Impact of Regional EEG Signals on Cognitive Workload Forecasting

Abstract: Precise and broadly applicable measurement of cognitive workload via electroencephalography (EEG) is essential for systems focused on human factors and safety. Despite the widespread adoption of EEG for assessing workload, the degree to which specific scalp regions contribute consistently to these assessments across different tasks, datasets, and individuals has not been fully established. To address this, we introduce a framework for evaluating EEG-based workload prediction at the regional level, where models are developed and tested using features derived solely from electrodes mapped to specific anatomical areas. Our study conducts a comprehensive analysis across four public EEG workload datasets, which vary in task requirements, hardware specifications, and electrode configurations. We measure the significance of each brain region through a performance-driven, model-agnostic method, applying both mixed-subject and subject-independent evaluation schemes. To guarantee reliability across varying experimental setups, outcomes are synthesized using a rank-based aggregation technique. The results demonstrate that, across all datasets and under subject-independent testing, electrode clusters in the frontal area achieve a relative rank position improvement of 15-20% compared to the full-scalp baseline, all while utilizing significantly fewer sensors. Frontal-central zones display the most consistent predictive power, whereas posterior and occipital areas show less stable contributions across different conditions. These insights suggest that the most reliable EEG markers for workload are concentrated in frontal and fronto-central regions, offering a foundation for creating streamlined and robust EEG-based workload monitoring solutions.

Source: arXiv Generated at: 2026-06-03 00:00:00 UTC