Abstract
Cognitive workload, the mental effort required to process task demands, is a key determinant of performance in high-stakes environments. While neurophysiological measures such as electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) have been studied, the combination of these modalities in multiple signal domains is yet to be explored. This study investigates the efficacy of a multimodal approach by using an open-access Stroop task-based dataset. We develop separate pipelines for EEG, de-oxygenated haemoglobin (Hb), and oxygenated haemoglobin (HbO), along with a multimodal analysis. Employing various classifiers, we note that: (1) frequency and wavelet attributes consistently outperform time-domain measures, and (2) multimodal fusion yields a peak accuracy of .993. Overall, these results highlight the efficacy of EEG–fNIRS fusion in building robust cognitive workload classifiers. Moreover, we identify the most informative feature sets and cortical channels, revealing consistent contributions from prefrontal regions that may serve as reliable biomarkers of cognitive workload.
Recommended Citation
Manawadu, Isuri Wijechandra; Gahalawat, Monika; Rojas, Raul Fernandez; Chetty, Girija; and Subramanian, Ram, "Tri-Domain EEG–fNIRS Fusion Approach for Stroop-Based
Cognitive Workload Classification" (2025). ACIS 2025 Proceedings. 243.
https://aisel.aisnet.org/acis2025/243