DISCRIMINATION OF MENTAL WORKLOAD LEVELS IN HUMAN SUBJECTS WITH FUNCTIONAL NEAR-INFRARED SPECTROSCOPY

SASSAROLI, ANGELO; ZHENG, FENG; HIRSHFIELD, LEANNE M.; GIROUARD, AUDREY; SOLOVEY, ERIN TREACY; JACOB, ROBERT J. K.; FANTINI, SERGIO · 2008 · Crossref

DOI: 10.1142/s1793545808000224

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 functional near-infrared spectroscopy (fNIRS) to discriminate between different levels of mental workload in human subjects. The research is motivated by the potential for fNIRS to serve as a non-invasive, portable monitor for brain activity, specifically targeting the prefrontal cortex where higher-order cognitive functions occur. The authors aim to determine if hemodynamic changes associated with working memory tasks can be reliably classified to distinguish varying cognitive loads, which could enable real-time assessment of mental workload for applications such as dynamic human-computer interfaces. The experimental protocol involved five healthy subjects performing working memory tasks while fNIRS data was collected from their foreheads. The tasks required subjects to count colored sections on the visible sides of a rotating cube, with the number of colors (0, 2, 3, or 4) defining the workload level. Workload 0 served as a rest condition. Optical data was acquired using an instrument with source-detector distances of 1.5 cm and 3 cm, measuring changes in oxyhemoglobin (HbO2) and deoxyhemoglobin (HbR) concentrations. Due to motion artifacts, data from only three subjects were analyzed. The researchers applied a 3-nearest neighbor classification algorithm based on the maximum amplitude of HbO2 and HbR concentration changes during the workload periods. Statistical significance was assessed using a modified t-test to control for false positives in correlated data. The results demonstrated that fNIRS signals were sensitive to mental workload, with concentration change amplitudes generally increasing with workload intensity. Classification success rates for distinguishing workloads 0, 2, and 4 ranged from 44% to 72%, significantly exceeding the calculated chance level of 19.1%. Classification based on HbR changes (55.6%–72.2%) proved more accurate than that based on HbO2 changes (44.4%–50.0%). Notably, one subject exhibited inverted hemodynamic trends (decreased HbO2 and increased HbR), consistent with negative BOLD signals reported in fMRI literature, while the other two subjects showed typical activation patterns. The study also highlighted lateralized activation, with the left prefrontal cortex providing more consistent signals than the right. The findings confirm the potential of fNIRS for non-invasive, quantitative assessment of mental workload. The authors conclude that while simple classification algorithms yield results above chance, future improvements should include motion-resistant probes, multi-parameter analysis leveraging fNIRS’s high temporal resolution, and more sophisticated machine learning techniques. These advancements could facilitate the development of real-time neurophysiological monitoring systems for human-computer interaction, allowing interfaces to adapt dynamically to a user’s cognitive state.

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

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