An EEG-based mental workload estimator trained on working memory task can work well under simulated multi-attribute task
archive: archived pipeline: cataloged
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 EEG-based mental workload (MW) estimators across different cognitive tasks. While passive Brain-Computer Interfaces (pBCIs) that monitor MW via EEG are effective for enhancing human-machine interaction, existing classifiers typically fail when applied to tasks not included in their training data (cross-task generalization). Previous cross-task studies reported performance at or below chance levels, attributed to task-specific EEG patterns, mismatched workload levels, and temporal effects. This paper investigates whether a regression model trained on working memory (WM) tasks can accurately estimate MW in a complex, simulated multi-attribute task (MATB), and whether cross-task feature selection can mitigate these generalization failures. The experimental design involved 17 healthy participants who performed verbal and spatial n-back tasks (WM) and the Multi-Attribute Task Battery (MATB). EEG data were recorded using 30 channels and processed to extract power spectral density features across seven frequency bands. The researchers employed a Support Vector Machine Regressor (SVMR) rather than a classifier to accommodate potential workload mismatches between tasks. A key methodological contribution was the implementation of a cross-task performance-based Recursive Feature Elimination (RFE) algorithm. This algorithm selected a subset of "salient features" (SF) by optimizing the correlation coefficient (COR) between predicted and actual workload in a cross-task validation set, thereby isolating MW-related EEG patterns from task-specific noise. Results demonstrated that within-task regression performance was high, with mean CORs exceeding 0.718. In contrast, cross-task performance using all features was poor, with CORs not significantly different from zero. However, applying the cross-task RFE-selected feature subset significantly improved cross-task estimation. Specifically, the correlation coefficient for the NM condition (trained on n-back, tested on MATB) reached 0.740 ± 0.147 for the feature selection data and 0.598 ± 0.161 for the validation data, a significant improvement over the chance-level performance observed with all features. Topographic mapping revealed that the most contributory features were distributed across frontal, parietal, and occipital lobes, suggesting that MW estimation relies on widespread neural activation patterns rather than isolated task-specific regions. The significance of these findings lies in the demonstration that a viable approach exists for generalizing MW estimation across cognitively distinct tasks. The study confirms that while task-specific EEG patterns exist, there are common MW-related features that can be identified and isolated through performance-based feature selection. This suggests that MW is a global neuroergonomic state that shares underlying neural mechanisms across different cognitive demands. These results provide a promising pathway for developing robust, adaptive pBCI systems that can monitor operator workload in real-world, multi-task environments without requiring task-specific retraining.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | canonical_url | — | — | 1 | 2026-08-09 |
| extract | success | cached | — | — | 4 | 2026-08-23 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-08-09 |
| enrich | success | semantic_scholar | — | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.8-27b-gittensor | summ-v5 | 3 | 2026-08-23 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-09 |
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.
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
- Theoretical Contribution: theory or model