Discover datasets

This dataset comprises multi-channel EEG recordings from 27 healthy adults performing a sustained-attention driving task in a virtual-reality simulator across 62 sessions. The task involved event-related lane-departure paradigms where participants maintained vehicle position on a simulated highway, with recordings capturing 32-channel EEG (30 scalp + 2 mastoid references) sampled at 500 Hz alongside vehicle position data. The dataset includes approximately 82 hours of raw, unfiltered EEG data with over 27,000 lane-departure trials and associated behavioral markers, providing a resource for investigating fatigue, drowsiness, and sustained attention mechanisms.

Participants

27

Channels

32 (10-10)

Citations

68

HED

v8.2.0

Size

36.4 GB

Version

v1.0.0

Updated

Jul 10, 2026

Beetl2021-B is a preprocessed EEG dataset from the NeurIPS 2021 BEETL competition, focused on transfer learning for motor imagery decoding. It contains 32-channel EEG recordings from 2 healthy subjects performing a 4-class motor imagery task (left hand, right hand, feet, rest) sampled at 200 Hz. The dataset is designed to benchmark transfer learning and domain adaptation algorithms addressing cross-subject and cross-dataset generalization challenges in brain-computer interfaces.

Participants

2

Channels

32 (10-10)

Citations

2

HED

v8.4.0

Size

1.99 GB

Version

v1.0.5

Updated

Aug 20, 2026

A BIDS-formatted derivative dataset of steady-state visually evoked potential (SSVEP) EEG recordings from 54 subjects across 2 sessions, containing 21,600 trials recorded at 1000 Hz with 62 channels. This dataset, derived from the Lee et al. 2019 source dataset, has been standardized using the Mother of All BCI Benchmarks (MOABB) framework to enable reproducible benchmarking and analysis of brain-computer interface paradigms based on SSVEP stimulation.

Participants

54

Channels

62 (10-05)

Citations

100

Size

72.4 GB

Version

v1.0.7

Updated

Aug 18, 2026

BigP3BCI Study A is a derivative P300 brain-computer interface dataset containing 32-channel EEG recordings from 13 healthy subjects performing a visual speller task using a 6×6 checkerboard stimulus paradigm. The dataset comprises EEG data sampled at 256 Hz with binary classification of target and non-target events, annotated using HED 8.4.0 schema and processed through the MOABB framework.

Participants

13

Channels

32 (10-10)

HED

v8.4.0

Size

3.86 GB

Version

v1.0.3

Updated

Aug 18, 2026

A high-density 124-channel EEG dataset comprising event-related potentials (ERPs) from 10 healthy participants during a visual object recognition task. Participants viewed 5,184 photographs from six object categories (human body, human face, animal body, animal face, fruit/vegetable, and inanimate objects), with 72 photographs per category, presented for 500 ms each. The dataset is suitable for investigating neural representations of object categories through single-trial EEG classification and representational similarity analysis.

Participants

10

Channels

124 (egi-geodesic)

Citations

93

HED

v8.4.0

Size

3.40 GB

Version

v1.0.3

Updated

Aug 18, 2026

BCIComp2020UpperLimb is a preprocessed EEG dataset from BCI Competition 2020 Track 4 containing motor imagery recordings of three grasping tasks (cylindrical, spherical, lumbrical) from 15 healthy subjects across three sessions. The dataset comprises 60-channel EEG data sampled at 250 Hz with 450 trials per subject (150 trials per session across 3 sessions), designed to evaluate session-to-session transfer learning in brain-computer interface applications. Data were preprocessed with 60 Hz notch filtering and cue-aligned epoching, with the 4-second motor imagery window extracted for analysis.

Participants

14

Channels

60 (10-10)

Citations

73

HED

v8.4.0

Size

2.76 GB

Version

v1.0.3

Updated

Aug 18, 2026

Beetl2021-A is a preprocessed motor imagery EEG dataset derived from the BEETL Competition 2021 (NeurIPS Task 2), comprising 63-channel, 500 Hz recordings from healthy subjects performing a four-class motor imagery task (rest, left hand, right hand, feet) during an online BCI racing game (Cybathlon2020IC). The dataset was designed to benchmark transfer learning and domain adaptation methods for subject-independent, cross-dataset EEG-based brain-computer interfacing.

Participants

3

Channels

63 (10-10)

Citations

2

HED

v8.4.0

Size

2.17 GB

Version

v1.0.6

Updated

Aug 20, 2026

This dataset comprises EEG recordings from 15 healthy participants performing a visual P300 oddball task during treadmill walking at 1.6 m/s. The study includes simultaneous scalp-EEG (46 channels), ear-EEG (14 channels), electrooculography (4 channels), and inertial measurement unit recordings (6 channels) sampled at 100 Hz. The dataset was collected as part of BCI Competition 2020 Track 5 to evaluate brain-computer interface performance under ambulatory conditions. This is a BIDS-formatted derivative dataset converted from the original data using the MOABB (Mother of All BCI Benchmarks) pipeline.

Participants

15

Channels

46 (10-10)

Citations

71

HED

v8.4.0

Size

224 MB

Version

v1.0.3

Updated

Aug 18, 2026

This dataset comprises EEG recordings from 15 healthy participants performing reach-and-grasp motor imagery tasks using three different electrode systems: gel-based laboratory equipment, water-based mobile EEG, and dry-electrode mobile EEG. Data were acquired at 256 Hz from 58 EEG channels plus 6 EOG channels across three sessions with 7200 total trials. Participants executed palmar grasp (toward glass) and lateral grasp (toward spoon) actions. The study investigates the feasibility of decoding natural reach-and-grasp neural correlates across different EEG acquisition modalities for brain-computer interface applications.

Participants

45

Channels

58 (10-05)

Citations

51

HED

v8.4.0

Size

8.67 GB

Version

v1.0.3

Updated

Aug 18, 2026

This dataset comprises longitudinal motor imagery EEG recordings from 18 BCI-naive subjects across six sessions (one offline, five online), designed to study transfer learning and skill acquisition in brain-computer interfaces (BCIs). It compares two domain adaptation frameworks—Generic Recentering and Personally Assisted Recentering—for calibration-free BCI training using left/right hand motor imagery with visual feedback. Data were collected at 512 Hz with 22 EEG channels and analyzed using Riemannian geometry-based classifiers.

Participants

18

Channels

22 (10-10)

Citations

49

HED

v8.4.0

Size

14.4 GB

Version

v1.0.4

Updated

Aug 20, 2026