Spectral EEG Dataset for Motor Intention Classification and Cognitive State Analysis

Published: 4 August 2026| Version 2 | DOI: 10.17632/gn8d4d4sv6.2
Contributors:
Griselda Cortés Barrera,
,
,
,
,
,
,
,
,

Description

This dataset contains spectral band-power features derived from continuous electroencephalography (EEG) recordings acquired with an Emotiv headset during two Brain–Computer Interface (BCI) paradigms, Pull/Push and Left/Right, provided as comma-separated values (CSV) files. It was developed to support the development, validation, and benchmarking of machine learning and deep learning algorithms for motor intention classification, cognitive state recognition, and adaptive BCI applications. The spectral features were extracted without prior aggressive artifact removal, retaining the frequency-domain representation of physiological (e.g., eye blinks, eye movements, and muscle activity) and non-physiological (e.g., head movement and electrode contact variations) artifacts. This enables the evaluation of robust feature engineering and classification methods under realistic acquisition conditions. In addition to motor command decoding, the dataset can be reused to investigate attention, cognitive workload, mental fatigue, and indirect stress-related responses during BCI interaction. By providing standardized spectral EEG features collected under controlled experimental protocols, this dataset promotes reproducible research and serves as a valuable benchmark for neuroengineering, biomedical signal processing, cognitive neuroscience, explainable artificial intelligence, and human–computer interaction studies.

Files

Steps to reproduce

1. Download the dataset. Download the complete dataset from the Mendeley Data repository and extract all files while preserving the original directory structure. 2. Review the documentation. Read the README.txt file before using the data. These documents describe the study, dataset organization, metadata, EEG acquisition parameters, variable definitions, and the filename convention. 3. Identify the participant. Each EEG recording is associated with a unique participant identifier using the format AASYYMMNNN, where AA indicates age, S biological sex (1 = Female, 2 = Male), YY the acquisition year, MM the acquisition month, and NNN the sequential recording number. Each participant has two EEG files corresponding to the LeftRight and PullPush experimental tasks. 4. Load the data. The recordings are stored as CSV files and can be imported into Python, MATLAB, R, Excel, or any software capable of reading comma-separated values. 5. Interpret the variables. Each row represents one observation window. The first column contains the acquisition timestamp, followed by spectral band-power features computed for the 14 EEG electrodes (AF3, F7, F3, FC5, T7, P7, O1, O2, P8, T8, FC6, F4, F8, AF4). Each electrode includes five frequency bands: Theta (4–8 Hz), Alpha (8–12 Hz), Beta Low (12–16 Hz), Beta High (16–30 Hz), and Gamma (30–45 Hz). Variables follow the Channel/Band naming convention (e.g., AF3/theta, F4/alpha). 6. Analyze the recordings. The dataset contains precomputed spectral EEG features and is ready for machine learning, statistical analysis, feature selection, brain–computer interface (BCI) development, cognitive state recognition, and benchmarking studies without requiring additional signal preprocessing.

Institutions

Categories

Data Science, Machine Learning, Cognitive Assessment, Electroencephalography, Cognitive Neuroscience, Biomedical Signal Processing, Brain-Computer Interface

Funders

Licence