Discover datasets

EEG recordings from an online brain-computer interface study with twelve healthy participants using a visual matrix P300 speller. Three fully unsupervised decoding methods were compared: learning from label proportions (LLP), expectation-maximization (EM), and their mixture (MIX). Participants copy-spelled a German sentence across three blocks, one per algorithm. The dataset supports research on calibration-free ERP-based BCIs.

Participants

12

Channels

31 (10-10)

Citations

27

HED

v8.4.0

Size

4.81 GB

Version

v1.0.3

Updated

Oct 8, 2026

Magnetoencephalography (MEG) recordings from 13 participants who viewed and heard audiovisual sequences while performing causality judgment and temporal order judgment tasks. The same six audiovisual sequences were used in both tasks, which allowed endogenous task demands to be manipulated while sensory input stayed constant. Auditory-only and visual-only localizer blocks are also included, along with anatomical MRI, EOG, ECG and empty-room recordings. The dataset supports research on multisensory correlation computations in the human brain.

Participants

14

Citations

28

Size

75.7 GB

Version

v1.0.0

Updated

Jun 22, 2026

The "Podcast" electrocorticography (ECoG) dataset captures intracranial neural activity from 9 participants (1,330 grid, depth, and strip electrodes) who listened to a 30-minute naturalistic story containing over 5,000 words. It includes raw recordings, a minimally preprocessed high-gamma version, the auditory stimuli, an aligned word-level transcript, and linguistic features ranging from acoustic properties to large language model embeddings. Tutorials replicating prior findings are provided to support research on natural language comprehension.

Participants

9

Size

14.4 GB

Version

v1.0.0

Updated

Jun 27, 2026

This dataset contains 32-channel EEG recordings from 8 healthy participants performing a combined steady-state visual evoked potential (SSVEP) brain-computer interface task. Spatially overlapping arrow stimuli at a single location, flickering at distinct frequencies derived from an 85 Hz CRT refresh rate, encode four directions (up, down, left, right). Participants covertly attended to one direction per trial, and the data cover three stimulation and attention schemes across two experiments.

Participants

8

Channels

31 (10-10)

Citations

3

HED

v8.4.0

Size

5.30 GB

Version

v1.0.2

Updated

Aug 18, 2026

Study D of the bigP3BCI collection, comprising 32-channel EEG recordings from 17 healthy participants using a P300-based speller with a 6x6 character grid and dynamic row-column stimulus presentation. Data were recorded in a single session per subject with online feedback, and events are labeled as target or non-target flashes. The dataset is a BIDS-formatted derivative of the PhysioNet bigP3BCI release, intended for machine learning and BCI benchmarking.

Participants

17

Channels

32 (10-10)

Citations

2

HED

v8.4.0

Size

2.41 GB

Version

v1.0.3

Updated

Aug 18, 2026

Beetl2021-B is a preprocessed 32-channel EEG dataset from the NeurIPS 2021 BEETL competition, built to benchmark transfer learning for motor imagery decoding. It contains recordings from 2 healthy subjects performing four-class motor imagery (left hand, right hand, feet, rest) at 200 Hz, totaling 1590 trials. The dataset supports evaluation of cross-subject and cross-dataset generalization and domain adaptation in brain-computer interfaces.

Participants

2

Channels

32 (10-10)

Citations

1

HED

v8.4.0

Size

1.99 GB

Version

v1.0.5

Updated

Aug 20, 2026

Raw and preprocessed magnetoencephalography (MEG) recordings from participants viewing objects that moved along trajectories and were either occluded or disappeared. The dataset supports investigation of how the brain represents objects during dynamic occlusion. It includes behavioural and eye-tracking data as source data, and preprocessed epochs in CoSMoMVPA format.

Participants

22

Citations

12

Size

258 GB

Version

v1.0.0

Updated

Jun 18, 2026

This dataset contains electroencephalography (EEG) recordings from 16 participants viewing object images presented in eight different 2-D rotations. Images were shown in rapid serial visual streams at two presentation rates (5 Hz and 20 Hz). The data support investigation of the time course of rotation-tolerant, high-level object representations in the human brain.

Participants

16

Citations

4

Size

18.3 GB

Version

v1.0.0

Updated

Jun 23, 2026

MEG recordings from a single infant subject in the ILABS amnoise dataset, collected in the context of research on infants at risk for dyslexia. The recordings relate to auditory processing during the linguistic sensitive period for native phoneme learning. The data are organized in BIDS format and were converted using MNE-BIDS.

Participants

2

Citations

21

Size

1.78 GB

Version

v1.0.0

Updated

Jun 23, 2026

A multi-session dataset of simultaneous EEG-fMRI recorded from 24 healthy adults aged 18-35 while they performed a gradual-onset continuous performance task (GradCPT) and a resting-state condition with intermittent multidimensional experience sampling. The same protocol was repeated on two days. Carbon wire loop sensors in the EEG cap support artifact removal. The dataset also includes ECG, behavioral responses, and mental health surveys. It supports study of attentional fluctuations, mind-wandering, and the test-retest reliability of EEG-fMRI relationships.

Participants

24

Channels

31 (10-10)

Citations

1

Size

115 GB

Version

v1.0.0

Updated

Jul 2, 2026