Open multi-session and multi-task EEG cognitive Dataset for passive brain-computer Interface Applications

Hinss, Marcel F.; Jahanpour, Emilie S.; Somon, Bertille; Pluchon, Lou; Dehais, Frédéric; Roy, Raphaëlle N. · 2023 · Crossref

DOI: 10.1038/s41597-022-01898-y

archive: archived pipeline: cataloged verified

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

Summary

This paper introduces the COG-BCI database, an open-access, multi-session, and multi-task electroencephalography (EEG) dataset designed to advance research in passive Brain-Computer Interfaces (pBCI). The study addresses the scarcity of shared data in pBCI research, which hinders the development and benchmarking of algorithms for monitoring mental states such as mental workload, vigilance, and decision-making. By providing a large-scale dataset with known ground truths for cognitive states, the authors aim to facilitate the validation of transfer learning methods and improve the robustness of pBCI systems against inter- and intra-subject variability. The dataset comprises recordings from 29 participants across three separate sessions, totaling over 100 hours of EEG data. Participants performed four distinct cognitive tasks: the N-Back task (to elicit varying levels of mental workload), the Multi-Attribute Task Battery II (MATB-II) for ecologically valid workload assessment, the Psychomotor Vigilance Task (PVT) for measuring vigilance and fatigue, and the Flanker task for assessing decision-making and conflict. Data acquisition involved a 64-channel EEG system, peripheral electrocardiography (ECG), and subjective ratings using the Karolinska Sleepiness Scale and Rating Scale Mental Effort. The experimental protocol included resting states and precise electrode localization via 3D scanning. The data is provided in raw format following the Brain Imaging Data Structure (BIDS) standard, ensuring reproducibility and ease of use for the research community. Technical validation confirmed that the experimental protocol successfully elicited the targeted mental states. Behavioral and physiological analyses demonstrated expected patterns: PVT reaction times increased over trials, indicating vigilance decrement, while cardiac measures (heart rate and variability) and EEG features (alpha and theta power) varied significantly across task conditions and difficulty levels. For instance, higher mental workload in the MATB-II and N-Back tasks correlated with distinct changes in frontal alpha power and heart rate variability. Additionally, the authors provided a proof-of-concept machine learning pipeline for mental workload estimation, demonstrating the dataset’s utility for developing and evaluating pBCI classifiers. The significance of this work lies in its contribution to open science in the pBCI field. By offering a comprehensive, validated, and freely available dataset, the COG-BCI database enables researchers to test new feature extraction algorithms, benchmark classifiers, and investigate transfer learning techniques without the high costs and logistical challenges of independent data collection. This resource supports the development of more reliable pBCI systems capable of monitoring operator mental states in real-world applications, ultimately aiming to enhance human-machine teaming and reduce human error in complex operational 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 qwen3.6-27b-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, behavioral performance data