Cross-Task Consistency of Electroencephalography-Based Mental Workload Indicators: Comparisons Between Power Spectral Density and Task-Irrelevant Auditory Event-Related Potentials

Ke, Yufeng; Jiang, Tao; Liu, Shuang; Cao, Yong; Jiao, Xuejun; Jiang, Jin; Ming, Dong · 2021 · Crossref

DOI: 10.3389/fnins.2021.703139

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 cross-task consistency in estimating mental workload (MWL) using electroencephalography (EEG). While EEG-based MWL estimators are promising for adaptive human-machine systems, features often fail to generalize across different task types due to task-specific neurophysiological responses. The authors hypothesized that task-irrelevant auditory event-related potentials (tir-aERPs) might offer more consistent MWL indicators than power spectral density (PSD) of ongoing EEG, as tir-aERPs reflect residual cognitive resources and are less dependent on task-specific strategies. The experiment involved 17 healthy subjects performing two distinct tasks: a verbal N-back task and a multi-attribute task battery (MATB), each with easy and hard difficulty levels. EEG was recorded while participants performed these tasks and were simultaneously presented with novel, task-irrelevant complex auditory probes. The researchers extracted tir-aERP components (N1, early P3a, late P3a, and reorienting negativity) and relative PSDs in theta, alpha, and low beta bands from the intervals between probes. Statistical analyses included bootstrapping-based ANOVAs to examine effects of MWL and task type, and support vector machines (SVM) were used to evaluate cross-task classification performance. Results indicated that tir-aERP amplitudes for N1, early P3a, late P3a, and reorienting negativity significantly decreased as MWL increased in both the N-back and MATB tasks. Crucially, task type did not significantly influence the amplitudes or topographical layouts of these MWL-sensitive tir-aERP features. In contrast, while relative PSDs in theta, alpha, and low beta bands were also sensitive to MWL, their patterns and topographies were significantly affected by task type. Consequently, cross-task classification models based on tir-aERP features significantly outperformed those based on PSD features. These findings suggest that tir-aERPs are more robust and consistent indicators of MWL across diverse task types compared to traditional PSD features. The study provides new insights into the neuropsychological essence of MWL, highlighting that while spectral power is task-dependent, the attenuation of responses to task-irrelevant stimuli reflects a more general limitation of cognitive resources. This implies that tir-aERPs could be a superior feature set for developing generalizable, real-time MWL monitoring systems in complex, safety-critical 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