Mental Workload Classification during simulated flight operations based on cardiac and neural dynamics recorded using the MUSE 2 low-cost system
DOI: 10.54941/ahfe1003017
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 investigates the feasibility of using the Muse 2, a low-cost and portable system combining electroencephalography (EEG) and photoplethysmography (PPG), to classify mental workload during simulated flight operations. The research addresses the limitations of subjective workload assessments and bulky, expensive research-grade instrumentation, aiming to validate whether consumer-grade sensors can provide objective, real-time physiological data in ecologically valid, high-movement environments. Five male pilot students participated in the experiment, performing two traffic pattern tasks in a flight simulator: a low-load condition involving monitoring an instructor’s flight and a high-load condition involving active piloting. Concurrently, participants completed a passive auditory oddball task to elicit event-related potentials (ERPs). Data were collected using the Muse 2 headset, which records EEG from five electrodes and heart rate via PPG. The researchers extracted three types of features for classification: EEG frequency-domain features (delta, theta, alpha, beta, and gamma bands), ERP features derived from the P300 response to auditory targets, and heart rate metrics from the PPG signal. Classification models were trained using Linear Discriminant Analysis for frequency and heart rate data, and a Riemannian geometry-based pipeline with Xdawn spatial filtering for ERP data. Group-level analyses confirmed successful manipulation of mental workload, with the high-load condition yielding higher subjective ratings, more auditory counting errors, higher heart rates, and lower frontal alpha power compared to the low-load condition. Crucially, the P300 amplitude was significantly lower in the high-load condition, consistent with the resource allocation theory of attention. In terms of classification accuracy, the EEG frequency-based classifier achieved the highest mean accuracy at 93.2%. However, post-hoc analysis revealed that this performance was primarily driven by beta and gamma band features, which are heavily contaminated by electromyographic (muscle) artifacts due to the pilots' motor activity during flying. Classifiers using only delta and theta bands reached 61.0% accuracy, while the alpha-only classifier performed at chance level (50.9%). The ERP-based classifier achieved a robust 77.8% accuracy, and the heart rate-based classifier reached 75.8%, though with higher variance. The study concludes that while low-cost systems like the Muse 2 offer promising prospects for mobile neuroergonomics, their limited number of electrodes restricts advanced noise reduction techniques, leading to signal contamination. The high accuracy of frequency-based classifiers was largely an artifact of muscle activity rather than neural dynamics. Consequently, the authors identify ERP-based classification as the most reliable trade-off between accuracy and responsiveness for estimating mental workload in this context, despite the requirement for auxiliary auditory stimuli. The findings highlight the challenges of using dry, low-channel EEG systems in dynamic operational settings and underscore the need for careful interpretation of high-frequency EEG features in mobile applications.
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 | — | — | 124 | 2026-08-10 |
| 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 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 123 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 11 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-10 |
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.
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