Frontal EEG electrodes predict cognitive workload 20% better than full scalp

A new study accepted to EMBC 2026 systematically evaluates which scalp regions of EEG (electroencephalography) contribute most consistently to cognitive workload prediction. Led by Jacob Wong and colleagues, the researchers built a model-agnostic framework that trains models using features from anatomically defined scalp regions only. They analyzed four publicly available EEG workload datasets covering diverse tasks (e.g., memory, arithmetic), recording hardware, and electrode montages. Under both mixed-subject and subject-independent evaluation protocols, region importance was quantified via a performance-based rank aggregation method.

Across all datasets, frontal electrode groups delivered relative rank improvements of 15-20% compared to full-scalp baselines while using far fewer channels. Fronto-central regions provided the most stable predictive utility across experimental conditions, whereas posterior and occipital regions showed inconsistent contributions. These findings suggest that workload-relevant EEG signals are most reliably captured from frontal and fronto-central sites, enabling the design of simpler, cheaper, and more generalizable EEG-based cognitive monitoring systems for human-centered and safety-critical applications like aviation, driving, and industrial control.