# BCI Report

> Public EEG decoding results, each reported with the protocol that produced it: cohort, electrode count, evaluation split, chance level, interval and known limits. A personal, noncommercial research preview; aggregate results only, never raw EEG or per-person scores.

There is no overall ranking: a figure is comparable only with others under the same protocol. Every page states its cohort size and caveats next to the number, and every number is copied from a reviewed download listed below. Each link here is a Markdown copy; drop `index.md` for the HTML page. Chinese versions of every page except /data-use/ live under https://bci.report/zh/.

## Questions

- [Do dry EEG electrodes decode as well as wet ones?](https://bci.report/topics/dry-vs-wet/index.md): Not in this dataset. On a 12-target, two-second SSVEP task, with the same **102** people recorded on both sensor types, the training-free FBCCA reference scored **82.3%** on wet recordings and **64.6%** on dry ones. An EEGNet trained on wet recordings fell by **29.4 pp** of balanced accuracy when tested on dry recordings (paired 95% interval **−34.5 pp** to **−24.3 pp**). Wearing order and time on task are entangled with sensor type, so this is not a clean estimate of what the hardware alone costs.
- [Does a P300 decoder calibrated on a screen still work in VR?](https://bci.report/topics/screen-to-vr/index.md): It scores above chance after the switch, but this dataset cannot say how much the switch costs. Across **21** people, each calibrated on a PC screen and tested in a VR headset and also the other way round, a simple logistic regression averaged **66.0%** balanced accuracy over the two directions (95% interval **61.6%–70.2%**; chance is **50%**), with stimulus-onset timing corrected. Nobody was also tested on the display they were calibrated on, so this is the level reached after the display change, not its cost.
- [Can fewer electrodes, or electrodes in the ear, match a full scalp montage?](https://bci.report/topics/fewer-electrodes/index.md): In-ear channels did not, for sleep staging in this cohort: on the same **10** people and epochs, four in-ear channels scored **53.6%** balanced accuracy on five-stage sleep against **69.0%** for six scalp electrodes (chance **20%**), a paired gap of **+15.4 pp**. In a separate eyes-open-or-closed task on **19** other people, four posterior electrodes showed no consistent difference from all sixteen: all sixteen minus four came to **−1.6 pp** (paired 95% interval **−10.0 pp** to **+6.8 pp**). These are channel sets from specific research recordings, not headsets—the sleep gap mixes electrode position, count and hardware, and the four posterior electrodes are a software subset of one recording—so neither shows that fewer electrodes are generally better or worse.
- [Does EEG decoding still work while walking or running?](https://bci.report/topics/on-the-move/index.md): Less well. With the same **23** people on a three-target, two-second SSVEP task (eight selected scalp channels; chance **33.3%**), the training-free author-style CCA reference scored **82.3%** balanced accuracy standing, **70.7%** at a slow walk and **68.6%** at a fast walk. On an ERP task, with the same **17** people in both conditions, a decoder fit on each person's first standing session and then used unchanged saw its scalp ROC AUC change by **−0.276 AUC** from standing to running (paired 95% interval **−0.322** to **−0.227 AUC**). The two tasks are derivatives of one original mobile study, so they are related evidence rather than independent replications.
- [How much calibration data does a wearable SSVEP decoder need?](https://bci.report/topics/calibration-budget/index.md): It depends on the method: on a 12-target wearable SSVEP task, with the same **102** people and the same future test blocks, going from 24 to 48 labeled target trials lifted single-band eTRCA from **39.8%** to **64.0%** balanced accuracy on wet recordings, but only from **12.4%** to **15.1%** on dry ones. Target-only spectral ridge reached just **21.3%** (wet) and **21.8%** (dry) at 48 trials, and no calibrated eTRCA point beat the training-free CCA reference, which uses no labels (**81.1%** wet, **65.1%** dry). Trial counts are not minutes: the export records labeled trials per person and does not convert them into a setup-time claim.
- [How often does an EEG decoder fire when nobody is giving a command?](https://bci.report/topics/when-not-to-act/index.md): It depends on how often the decoder acts at all: one that never acts never fires by mistake, so its false activations mean something only beside the commands it caught. In a cue-gated idle test with **4** people, the quietest decoder, EEGNet, made **0 / 60** false activations in idle trials but detected **1 / 60** commands, and for **2** of those people its threshold rejected every command; in a separate pilot with **4** other people, a CCA decoder accepted **130 / 192** windows when a command was intended, and **9 / 48**, **20 / 48** and **11 / 96** in the states where none was. Both are small pilots, counted per trial or per window rather than per hour: neither can say how often a decoder would fire by mistake over an hour of real rest.
- [Does pretraining help EEG foundation models like LaBraM and CBraMod?](https://bci.report/topics/does-pretraining-help/index.md): Sometimes, depending on the encoder and the task: with frozen encoders and the fixed readout, pretrained CBraMod beat the mean of its three random initializations by **+6.39 pp** of balanced accuracy on MI/rest (**10** people) and by **+5.90 pp** on mental workload (**36** people), and pretrained LaBraM beat its own by **+8.12 pp** on mental workload, each with a paired 95% interval above zero. On MI/rest, LaBraM's difference was **−2.42 pp** (interval crosses zero), and **−3.56 pp** under the train-selected readout (interval wholly below zero). The comparison does not establish that these benchmarks were unseen during pretraining, nor a benefit on other tasks or adaptation budgets.
- [Can resting-state EEG separate Parkinson's disease from controls?](https://bci.report/topics/clinical-groups/index.md): Above chance, as a group-level research result from one site. Across **149** people with Parkinson's disease or in the control group, a relative-spectrum logistic regression on resting-state EEG separated the two groups at **71.7%** balanced accuracy (descriptive 95% interval **63.5%–79.3%**, AUROC **0.76**, chance **50%**); an age-and-sex-only comparator, which is not an EEG model and uses no EEG at all, reached **55.0%** (**46.9%–63.6%**) on the same folds. It is not a diagnosis: it separates groups, not individuals, and has not been repeated beyond this one site.

## Datasets

- [EEG datasets with results on BCI Report](https://bci.report/datasets/index.md): Every public EEG dataset BCI Report has published a result on, with the task, the licence, and each figure measured on it — plus the datasets that are listed but not scored.
- [EEG Alpha Waves dataset: EEG decoding results](https://bci.report/datasets/alpha-waves/index.md): Published EEG decoding results on EEG Alpha Waves dataset (Eyes open / eyes closed) for Relative band-power logistic regression, each with its cohort, interval and the page it comes from.
- [BETA: EEG decoding results](https://bci.report/datasets/beta/index.md): Published EEG decoding results on BETA (40-target SSVEP) for Standard CCA, Spectral ridge, LaBraM, CBraMod, EEGNet, each with its cohort, interval and the page it comes from.
- [ds003810: EEG decoding results](https://bci.report/datasets/ds003810/index.md): Published EEG decoding results on ds003810 (Motor imagery / rest) for Spectral ridge, CSP+LDA, LaBraM, CBraMod, EEGNet …, each with its cohort, interval and the page it comes from.
- [OpenNeuro ds004584 · Parkinson's disease, rest eyes open: EEG decoding results](https://bci.report/datasets/ds004584/index.md): Published EEG decoding results on OpenNeuro ds004584 · Parkinson's disease, rest eyes open (Resting-state EEG, Parkinson's disease and controls) for Relative-spectrum logistic regression, each with its cohort, interval and the page it comes from.
- [ds005342: EEG decoding results](https://bci.report/datasets/ds005342/index.md): Published EEG decoding results on ds005342 (Cue-gated idle / command) for CSP+LDA, LaBraM, EEGNet, ShallowFBCSPNet, CBraMod …, each with its cohort, interval and the page it comes from.
- [ds006593: EEG decoding results](https://bci.report/datasets/ds006593/index.md): Published EEG decoding results on ds006593 (P300 target ERP) for Spectral ridge, Temporal ridge, LaBraM, CBraMod, EEGNet, each with its cohort, interval and the page it comes from.
- [EEGMAT: EEG decoding results](https://bci.report/datasets/eegmat/index.md): Published EEG decoding results on EEGMAT (Mental arithmetic / rest) for Spectral ridge, LaBraM, CBraMod, EEGNet, Log-covariance ridge, each with its cohort, interval and the page it comes from.
- [EESM19 scalp subset: EEG decoding results](https://bci.report/datasets/eesm19/index.md): Published EEG decoding results on EESM19 scalp subset (Five-stage sleep) for Spectral ridge, LaBraM, CBraMod, EEGNet, each with its cohort, interval and the page it comes from.
- [EESM23 · Ear-EEG Sleep Monitoring 2023: EEG decoding results](https://bci.report/datasets/eesm23/index.md): Published EEG decoding results on EESM23 · Ear-EEG Sleep Monitoring 2023 (Five-stage sleep, in-ear and scalp) for Log-bandpower logistic regression, each with its cohort, interval and the page it comes from.
- [Multimodal gait dataset · treadmill walking: EEG decoding results](https://bci.report/datasets/gait-eeg/index.md): Published EEG decoding results on Multimodal gait dataset · treadmill walking (Treadmill walking at three speeds) for Relative spectral bands, each with its cohort, interval and the page it comes from.
- [Mobile BCI dataset (SSVEP and ERP paradigms): EEG decoding results](https://bci.report/datasets/mobile-bci/index.md): Published EEG decoding results on Mobile BCI dataset (SSVEP and ERP paradigms) (SSVEP and ERP while standing, walking and running) for Author-style CCA, Spectral ridge, LaBraM, CBraMod, Temporal-feature L2 logistic regression, each with its cohort, interval and the page it comes from.
- [Phantom EEG with motion, muscle and eye artifacts: EEG decoding results](https://bci.report/datasets/phantom/index.md): Published EEG decoding results on Phantom EEG with motion, muscle and eye artifacts (Physical head phantom with motion, muscle and eye artifacts) for Fixed multi-output ridge regression, each with its cohort, interval and the page it comes from.
- [TMNRED / ds005383: EEG decoding results](https://bci.report/datasets/tmnred/index.md): Published EEG decoding results on TMNRED / ds005383 (Semantic target ERP) for Spectral ridge, Temporal ridge, LaBraM, CBraMod, EEGNet, each with its cohort, interval and the page it comes from.
- [Cattan PC/VR P300 dataset: EEG decoding results](https://bci.report/datasets/vr-pc-p300/index.md): Published EEG decoding results on Cattan PC/VR P300 dataset (P300 on a PC screen and in a VR headset) for Mean-window logistic regression, Spatiotemporal shrinkage LDA, each with its cohort, interval and the page it comes from.
- [Wearable SSVEP BCI dataset (dry and wet electrodes): EEG decoding results](https://bci.report/datasets/wearable-ssvep-102/index.md): Published EEG decoding results on Wearable SSVEP BCI dataset (dry and wet electrodes) (12-target SSVEP, dry and wet electrodes) for CCA, FBCCA, Spectral ridge, EEGNet, Single-band eTRCA, each with its cohort, interval and the page it comes from.
- [YSU asynchronous SSVEP-BCI dataset: EEG decoding results](https://bci.report/datasets/ysu-async-ssvep/index.md): Published EEG decoding results on YSU asynchronous SSVEP-BCI dataset (Asynchronous SSVEP, control and non-control states) for Fixed CCA with rejection, each with its cohort, interval and the page it comes from.

## Methods

- [EEG decoding methods with results on BCI Report](https://bci.report/methods/index.md): EEGNet, LaBraM, CBraMod, CCA, FBCCA, eTRCA and other EEG decoders: every published BCI Report figure for each, grouped by dataset, with cohort and interval — and the models not yet measured.
- [CBraMod: results on public EEG datasets](https://bci.report/methods/cbramod/index.md): Every published BCI Report result for CBraMod, grouped by dataset (ds003810, EEGMAT, BETA, ds006593, TMNRED / ds005383, EESM19 scalp subset, ds005342, Mobile BCI dataset (SSVEP and ERP paradigms)), each with its cohort, interval and source page. Not an overall ranking.
- [CCA: results on public EEG datasets](https://bci.report/methods/cca/index.md): Every published BCI Report result for CCA, grouped by dataset (BETA, Wearable SSVEP BCI dataset (dry and wet electrodes), Mobile BCI dataset (SSVEP and ERP paradigms)), each with its cohort, interval and source page. Not an overall ranking.
- [CSP+LDA: results on public EEG datasets](https://bci.report/methods/csp-lda/index.md): Every published BCI Report result for CSP+LDA, grouped by dataset (ds003810, ds005342), each with its cohort, interval and source page. Not an overall ranking.
- [Deep4Net: results on public EEG datasets](https://bci.report/methods/deep4net/index.md): Every published BCI Report result for Deep4Net, grouped by dataset (ds005342), each with its cohort, interval and source page. Not an overall ranking.
- [EEGNet: results on public EEG datasets](https://bci.report/methods/eegnet/index.md): Every published BCI Report result for EEGNet, grouped by dataset (ds003810, EEGMAT, BETA, ds006593, TMNRED / ds005383, EESM19 scalp subset, ds005342, Wearable SSVEP BCI dataset (dry and wet electrodes)), each with its cohort, interval and source page. Not an overall ranking.
- [eTRCA: results on public EEG datasets](https://bci.report/methods/etrca/index.md): Every published BCI Report result for eTRCA, grouped by dataset (Wearable SSVEP BCI dataset (dry and wet electrodes)), each with its cohort, interval and source page. Not an overall ranking.
- [FBCCA: results on public EEG datasets](https://bci.report/methods/fbcca/index.md): Every published BCI Report result for FBCCA, grouped by dataset (Wearable SSVEP BCI dataset (dry and wet electrodes)), each with its cohort, interval and source page. Not an overall ranking.
- [LaBraM: results on public EEG datasets](https://bci.report/methods/labram/index.md): Every published BCI Report result for LaBraM, grouped by dataset (ds003810, EEGMAT, BETA, ds006593, TMNRED / ds005383, EESM19 scalp subset, ds005342, Mobile BCI dataset (SSVEP and ERP paradigms)), each with its cohort, interval and source page. Not an overall ranking.
- [ShallowFBCSPNet: results on public EEG datasets](https://bci.report/methods/shallowfbcspnet/index.md): Every published BCI Report result for ShallowFBCSPNet, grouped by dataset (ds005342), each with its cohort, interval and source page. Not an overall ranking.

## Data

- [BCI Report data API: static JSON and CSV at stable URLs](https://bci.report/api/index.md): Every BCI Report result as static JSON and CSV files at stable URLs — no key, no account. What each file holds, how to load it with curl, Python or Hugging Face datasets, and how to cite a release.
- [context-update.json](https://bci.report/data/context-update.json): Screen-to-VR P300 transfer, treadmill walking speed beside a movement-nuisance comparator, and the asynchronous SSVEP non-control pilot, with audits.
- [clinical-update.json](https://bci.report/data/clinical-update.json): Parkinson's disease vs. controls from resting-state EEG (ds004584) beside an age-and-sex-only comparator, the claim boundary, and status-only sources.
- [evidence-update.json](https://bci.report/data/evidence-update.json): In-ear vs. scalp sleep staging (EESM23), four vs. sixteen electrodes (Alpha Waves), and the physical head phantom, each with its rights record.
- [deployment-topics.json](https://bci.report/data/deployment-topics.json): Deployment topics: dry vs. wet sensor transfer, calibration budget, movement (SSVEP and ERP) and pretraining controls — measurements, paired contrasts, seed sensitivity and dataset citations.
- [experiments.json](https://bci.report/data/experiments.json): The core matrix: every protocol with each method’s score, interval, cohort, electrode count, training mode and limitations, plus the model and dataset directories.
- [mi-rest-results.csv](https://bci.report/data/mi-rest-results.csv): Motor imagery & rest: one row per method — score, interval, secondary metric, cohort and compute time.
- [mi-rest-protocol.json](https://bci.report/data/mi-rest-protocol.json): Motor imagery & rest: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [idle-results.csv](https://bci.report/data/idle-results.csv): Idle & command: one row per method — score, interval, secondary metric, cohort and compute time.
- [idle-protocol.json](https://bci.report/data/idle-protocol.json): Idle & command: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [beta-8ch-results.csv](https://bci.report/data/beta-8ch-results.csv): SSVEP · 8 channels: one row per method — score, interval, secondary metric, cohort and compute time.
- [beta-8ch-protocol.json](https://bci.report/data/beta-8ch-protocol.json): SSVEP · 8 channels: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [beta-4ch-results.csv](https://bci.report/data/beta-4ch-results.csv): SSVEP · 4 channels: one row per method — score, interval, secondary metric, cohort and compute time.
- [beta-4ch-protocol.json](https://bci.report/data/beta-4ch-protocol.json): SSVEP · 4 channels: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [arithmetic-rest-results.csv](https://bci.report/data/arithmetic-rest-results.csv): Arithmetic & rest: one row per method — score, interval, secondary metric, cohort and compute time.
- [arithmetic-rest-protocol.json](https://bci.report/data/arithmetic-rest-protocol.json): Arithmetic & rest: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [p300-target-results.csv](https://bci.report/data/p300-target-results.csv): P300 target ERP: one row per method — score, interval, secondary metric, cohort and compute time.
- [p300-target-protocol.json](https://bci.report/data/p300-target-protocol.json): P300 target ERP: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [semantic-target-results.csv](https://bci.report/data/semantic-target-results.csv): Semantic target ERP: one row per method — score, interval, secondary metric, cohort and compute time.
- [semantic-target-protocol.json](https://bci.report/data/semantic-target-protocol.json): Semantic target ERP: the protocol — cohort, channels, windows, split, training budget, rights and limitations.
- [sleep-scalp-results.csv](https://bci.report/data/sleep-scalp-results.csv): Sleep staging: one row per method — score, interval, secondary metric, cohort and compute time.
- [sleep-scalp-protocol.json](https://bci.report/data/sleep-scalp-protocol.json): Sleep staging: the protocol — cohort, channels, windows, split, training budget, rights and limitations.

## Optional

- [Every EEG score, with the protocol that produced it.](https://bci.report/index.md): Public EEG decoding results reported with their protocol: motor imagery, SSVEP with 4 and 8 electrodes, P300 and semantic ERP, cognitive load, sleep staging, and idle false activations.
- [Releases](https://bci.report/releases/index.md): Every reviewed BCI Report release: what it added, what it held back, and the SHA-256 of every file it ships.
- [Evidence you can inspect. Recordings we do not redistribute.](https://bci.report/data-use/index.md): How BCI Report selects sources, credits datasets and shares aggregate EEG benchmark results.