Assessment of mental workload across cognitive tasks using a passive brain-computer interface based on mean negative theta-band amplitudes

Gallegos Ayala, Guillermo I.; Haslacher, David; Krol, Laurens R.; Soekadar, Surjo R.; Zander, Thorsten O. · 2023 · Crossref

DOI: 10.3389/fnrgo.2023.1233722

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the challenge of generalizing mental workload assessment across different cognitive tasks using passive brain-computer interfaces (pBCIs). While EEG-based BCIs can detect workload in real-time, classifiers trained on one task often fail to generalize to others due to task-dependent neural signal variations. The authors propose a novel algorithm focusing on frontal theta oscillations to improve cross-task classification performance, aiming to create a more robust system for neuroadaptive human-computer interaction. The researchers utilized a published dataset from Zhang et al. (2018) involving 15 participants performing six cognitive tasks: a calibration task (T0) with no-workload and workload conditions, and five experimental tasks (T1–T5) with low and high workload conditions. These tasks included N-back, backward span, arithmetic addition, word recovery, and mental rotation. EEG data was recorded using a 64-channel system. The methodology involved preprocessing steps including bandpass filtering (4–7 Hz theta band), artifact removal via iterative trimming of extreme voltage values, and a unique scaling procedure. This scaling mapped signal amplitudes to arbitrary universal factors to standardize data across subjects and sessions. Feature extraction focused on the mean negative theta-band amplitudes from six prefrontal channels identified as exhibiting the largest changes in theta negativity. Classification was performed using support vector machines in a subject-dependent task transfer setup, where models were trained on calibration data (T0) and tested on unseen experimental tasks (T1–T5). The results demonstrated that the proposed algorithm achieved binary classification accuracies of 92.00% and 92.35% when distinguishing between low or high workload conditions versus the initial no-workload condition. This performance significantly outperformed previous approaches using Filter Bank Common Spatial Patterns. However, the algorithm did not perform above chance levels when directly comparing high versus low workload conditions. Additionally, while applying independent component analysis (ICA) prior to preprocessing yielded stable above-chance results across all tasks, it did not surpass the performance of the previous state-of-the-art methods. The study concludes that while the proposed algorithm cannot fully replace existing general-purpose classification methods, it offers superior performance for specific workload comparisons, particularly in detecting the presence of workload against a baseline. The findings highlight the potential of focusing on mean negative theta-band amplitudes with specialized preprocessing to enhance cross-task generalization in pBCIs. This approach provides a feasible method for real-time mental workload monitoring in diverse scenarios, contributing to the development of more adaptive and reliable neurotechnology systems.

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 qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

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

• neuro workload indices
• workload measurement