Measuring Mental Workload with EEG+fNIRS

Aghajani, Haleh; Garbey, Marc; Omurtag, Ahmet · 2017 · Crossref

DOI: 10.3389/fnhum.2017.00359

archive: archived pipeline: cataloged

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the capability of a hybrid functional neuroimaging technique combining electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) to quantify human mental workload (MWL). The research is motivated by the need for accurate, non-invasive physiological measures to monitor cognitive load in human-machine interfaces, where traditional self-report or behavioral methods are often delayed or intrusive. The authors hypothesized that concurrent EEG and fNIRS would provide superior discrimination of MWL levels compared to either modality alone, leveraging complementary neural and hemodynamic signals. The experiment involved 17 healthy subjects performing a letter n-back task, a standard working memory paradigm, with difficulty parametrically varied by changing the n-value from 0 to 3. Data were acquired using 19 whole-head EEG channels and 19 fNIRS channels positioned on the forehead to cover the prefrontal cortex. Signal preprocessing included band-pass filtering and artifact rejection, leading to the exclusion of three subjects due to poor signal quality or low task accuracy. Features were extracted from EEG (frequency band power, phase locking value, phase-amplitude coupling), fNIRS (hemoglobin amplitude, slope, and statistical moments), and a new category of hybrid features based on zero-lagged correlations between EEG frequency bands and fNIRS hemoglobin levels, reflecting neurovascular coupling. These features were fed into a linear support vector machine (SVM) classifier using 10-fold cross-validation, with principal component analysis applied to reduce dimensionality. The study systematically assessed the impact of feature subsets and window sizes (5 to 25 seconds) on classification performance. Results indicated that the hybrid EEG+fNIRS system achieved significantly higher classification accuracy than either EEG or fNIRS alone in discriminating between different levels of mental workload. The inclusion of hybrid neurovascular features contributed to this improved performance. The study also evaluated the effect of window size, finding that shorter windows (5–10 seconds) were sufficient for robust classification, supporting the feasibility of real-time monitoring. Specific metrics such as sensitivity, specificity, and predictive values confirmed the robustness of the hybrid approach. The findings suggest that EEG+fNIRS is a promising tool for developing passive brain-computer interfaces and other applications requiring continuous monitoring of user cognitive states. By combining neural activity with cerebral blood flow information, the hybrid system offers a more comprehensive biomarker for MWL than uni-modal approaches. This work provides a validated framework for integrating multi-modal brain signals to enhance the accuracy of cognitive state detection, with implications for improving human-computer interaction safety and efficiency in complex operational environments.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• neuro workload indices
• workload measurement

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data