Discover datasets

A brain-computer interface dataset implementing a gaze-independent rapid serial visual presentation (RSVP) P300 speller paradigm. Twelve healthy participants performed mental typewriting tasks using visual attention to discriminate target symbols from non-targets presented sequentially at screen center. The dataset includes preprocessed EEG recordings (63 channels, 200 Hz sampling rate after downsampling from 1000 Hz acquisition) with event-related potential markers, achieving 94.8% mean accuracy and 1.43 symbols/minute spelling rate.

Participants

12

Channels

63 (10-10)

Citations

95

HED

v8.4.0

Size

5.70 GB

Version

v1.0.2

Updated

Aug 18, 2026

This dataset contains EEG recordings from 11 healthy participants performing a multi-streamed musical oddball paradigm for auditory brain-computer interface (BCI) research. Participants selectively attended to one of three concurrent instruments in polyphonic music clips (Synth-Pop or Jazz) while counting deviant patterns, enabling classification of attended instrument from EEG signals including P300 and early auditory ERP components. The dataset is a BIDS-formatted derivative converted using MOABB from the original BNCI Horizon 2020 BNCI2015-006 dataset.

Participants

11

Channels

64 (10-10)

Citations

83

HED

v8.4.0

Size

11.8 GB

Version

v1.0.1

Updated

Aug 18, 2026

This dataset comprises preprocessed EEG recordings from a novel 9-class auditory event-related potential (ERP) paradigm designed for brain-computer interface (BCI) applications. Ten healthy participants performed a spelling task using a predictive text entry system (PASS2D) that exploited spatial auditory stimuli varying in pitch and direction. The study demonstrates competitive performance for auditory ERP-based BCIs, achieving 0.8 characters per minute and 3.4 bits/min information transfer rate, with potential applications for communication in patients with severe motor impairment.

Participants

10

Channels

63 (10-10)

Citations

84

HED

v8.4.0

Size

7.62 GB

Version

v1.0.2

Updated

Aug 18, 2026

An auditory brain-computer interface dataset implementing the AMUSE (Auditory Multi-class Spatial ERP) paradigm for spelling applications. The dataset comprises EEG recordings from 10 healthy subjects performing an auditory oddball task with spatial cues from six speaker locations, achieving online spelling performance of up to 1.41 characters per minute. Data includes preprocessed EEG signals with P300 and N200 event-related potentials, suitable for BCI research and benchmarking.

Participants

10

Channels

8 (10-10)

Citations

81

HED

v8.4.0

Size

74.4 MB

Version

v1.0.3

Updated

Aug 18, 2026

This dataset comprises P300 evoked potentials recorded from 10 healthy female subjects using two complementary paradigms: the P300 Speller (overt attention) and GeoSpell (covert attention) interfaces. Scalp EEG was acquired from 16 electrodes at 256 Hz while subjects focused on one of 36 alphanumeric characters across multiple sessions. The dataset includes preprocessed data with bandpass filtering (0.1-20 Hz) and is designed to investigate P300 latency jitter effects on brain-computer interface performance.

Participants

10

Channels

16 (10-10)

Citations

79

HED

v8.4.0

Size

247 MB

Version

v1.0.2

Updated

Aug 18, 2026

The BNCI 2014-002 Motor Imagery dataset comprises EEG recordings from 14 healthy subjects performing two-class motor imagery tasks (right hand and feet imagination) in a cue-guided Graz-BCI paradigm. Data were acquired at 512 Hz using 15 EEG channels with online Butterworth filtering and Laplacian montage, yielding 160 trials per subject with continuous visual feedback. This minimally preprocessed dataset has been benchmarked for brain-computer interface applications using machine learning classifiers including random forests and regularized linear discriminant analysis.

Participants

14

Channels

15 (other)

Citations

89

HED

v8.4.0

Size

1.37 GB

Version

v1.0.4

Updated

Aug 18, 2026

This dataset comprises EEG recordings from 8 patients with amyotrophic lateral sclerosis (ALS) performing a P300-based brain-computer interface (BCI) speller task. Participants completed a copy-spelling paradigm using a 6×6 matrix with row-column intensification, generating 35 trials per subject across a single session. The preprocessed data, acquired at 256 Hz from 8 electrodes using active g.Ladybird sensors, demonstrates high classification accuracy (97.5%) and represents a valuable resource for developing and benchmarking BCI systems for communication assistance in severely paralyzed populations.

Participants

8

Channels

8 (10-10)

Citations

88

HED

v8.4.0

Size

241 MB

Version

v1.0.2

Updated

Aug 18, 2026

This dataset comprises EEG recordings from 6 healthy subjects performing an error-related potential (ErrP) monitoring task where participants observe an autonomous cursor moving toward targets with controlled error rates (20-40%). The study demonstrates that error-related brain potentials can be detected and decoded during passive monitoring of external agents without direct user control, providing foundational evidence for error monitoring in brain-computer interface applications.

Participants

6

Channels

64 (10-10)

Citations

83

HED

v8.4.0

Size

5.21 GB

Version

v1.0.2

Updated

Aug 18, 2026

A multi-joint upper-limb motor imagery EEG dataset comprising 18 healthy subjects performing eight distinct imagery tasks involving hand, wrist, elbow, and shoulder movements. The dataset contains 320 trials per subject acquired at 1000 Hz using 62-channel EEG with visual cue-based paradigm, designed for brain-computer interface research and motor rehabilitation applications.

Participants

18

Channels

62 (10-10)

Citations

4

HED

v8.4.0

Size

50.0 GB

Version

v1.0.2

Updated

Aug 18, 2026

This EEG dataset comprises recordings from 9 individuals with CNS tissue damage (stroke and spinal cord injury) performing five mental imagery tasks: word association, mental subtraction, spatial navigation, and motor imagery of the right hand and feet. Data were collected across two sessions using 30-channel EEG at 256 Hz with visual cue-guided paradigm. The dataset includes preprocessed signals with artifact rejection and is intended for brain-computer interface research and motor imagery classification studies.

Participants

9

Channels

30 (10-10)

Citations

74

HED

v8.4.0

Size

2.74 GB

Version

v1.0.2

Updated

Aug 18, 2026