Public datasets are how most BCI methods are developed and compared. They let you test an idea in an afternoon, compare against published numbers, and evaluate across many more subjects than a single lab records. The catch is that reported results on the same dataset are only comparable if the evaluation protocol is the same.
Where to look
| Source | What you find |
|---|---|
| BCI Competition datasets | Classic benchmarks such as BCI Competition IV 2a (four-class motor imagery) and 2b (two-class, three channels). |
| PhysioNet | Large open datasets, including the EEG Motor Movement/Imagery dataset with 109 subjects. |
| OpenNeuro | BIDS-formatted EEG, MEG, fNIRS and MRI datasets from many labs. |
| MOABB | A Python library that downloads many public BCI datasets and runs standard evaluation protocols. |
| BCINexus dataset library | Published EEG, fNIRS and EMG datasets with paradigm metadata, provenance and citations, reusable in BCILattice. |
Pick by paradigm
Motor imagery
The largest selection. Start with BCI Competition IV 2a or PhysioNet. See motor imagery BCI.
P300 / ERP
Speller and oddball datasets with many repetitions per subject. See P300 BCI.
SSVEP
Multi-target datasets with several stimulation frequencies. See SSVEP BCI.
fNIRS and hybrid
Fewer, smaller datasets, some recorded with simultaneous EEG. See what is fNIRS.
Using a public dataset well
Read the dataset paper first: sampling rate, montage, which runs are calibration, which are evaluation, and what the labels mean. Use the dataset's own session structure for your split. Many published results use a specific train/test session division, and ignoring it makes your numbers incomparable.
On BCINexus, published datasets have their own pages and can be cited and reused, and benchmarks compare models on shared datasets with reproducible pipelines. In BCILattice, Braindecode and MOABB are on the free extension shelf.
Questions
Where can I download EEG datasets?
From PhysioNet, OpenNeuro, the BCI Competition archives, through the MOABB Python library, and from the BCINexus dataset library, which lists EEG, fNIRS and EMG datasets with paradigm metadata.
What is the best EEG dataset for motor imagery?
BCI Competition IV 2a is the standard four-class benchmark, and the PhysioNet EEG Motor Movement/Imagery dataset offers 109 subjects for cross-subject work.
Can I publish my own EEG dataset?
Yes. On BCINexus a dataset can be published with its own page, metadata and citation, separately from or alongside a study, subject to the consent your participants gave.
Read the reference
Related
EEG File Formats: EDF, BDF, GDF, FIF, VHDR, SET, SNIRF
EDF, BDF, VHDR, SET, FIF: EEG has many formats because each amplifier vendor and toolbox defined its own. Here is what each one is and what reads it.
Motor Imagery BCI: EEG Classification Explained
Imagining a movement suppresses the mu and beta rhythms over the opposite motor cortex. A motor imagery BCI detects that suppression and turns it into a command, with no movement and no external stimulus.
BCI for Beginners: How to Get Started with EEG and BCI
You do not need a headset to start. A public dataset, a laptop and a clear first question will teach you more than a new device and no plan.
Reproducible BCI Research
A methods section is a summary written from memory weeks after the run. A study that carries its own stages, parameters and fold scores is the run itself.