Core-matrix protocol · False activation × detection

Idle & command on ds005342

The released protocol, with the score of every method run under it.

The protocol at a glance

Task
Seated motor imagery · cue-gated replay
Evaluation
False activation × detection
Dataset
ds005342
People
4
Data
60 idle / 60 command test trials
Input
180 seconds of test idle
Chance level
Not applicable. Command detection is read together with the idle false-activation rate: always abstaining scores zero false activations.
Metrics
Command detection ≤3s (primary) · Idle false activation (secondary)

Results

Compare within this protocol only: other protocols differ in cohort, electrodes, window or chance level.

MethodTraining modeCommand detection ≤3sIdle false activationChannelsPeopleScoring time
CSP+LDAClassical methodSupervised fit23.3%14 / 60 command trials0.0%0 / 60 idle trials1 of 4 people always abstained1740.1 s
LaBraMFoundation modelFrozen encoder + linear head18.3%11 / 60 command trials3.3%2 / 60 idle trials0 of 4 people always abstained1742.8 s
EEGNetCompact modelScratch · 20 epochs1.7%1 / 60 command trials0.0%0 / 60 idle trials2 of 4 people always abstained17410.1 s
ShallowFBCSPNetCompact modelScratch · 20 epochs43.3%26 / 60 command trials1.7%1 / 60 idle trials1 of 4 people always abstained17428.6 s
CBraModFoundation modelFrozen encoder + linear head18.3%11 / 60 command trials1.7%1 / 60 idle trials1 of 4 people always abstained1742.3 s
Deep4NetCompact modelScratch · 20 epochs3.3%2 / 60 command trials1.7%1 / 60 idle trials1 of 4 people always abstained17413.9 s
TS+LRClassical method · OAS · idle-native windows · Apple M5 / CPUSupervised fit35.0%21 / 60 command trials6.7%4 / 60 idle trials174—
MDMClassical method · OAS · idle-native windows · Apple M5 / CPUSupervised fit35.0%21 / 60 command trials1.7%1 / 60 idle trials174—

Scoring time covers fitting and prediction in this configuration, may include accelerator waiting, and excludes data preparation. Scoring time is not recorded for the rows of core-matrix-update-20261106.

Not applicable under this protocol: Standard CCA, FBCCA (they need the stimulation frequencies, and this protocol has none); Temporal ridge, XDAWNCov+TS+LR, xDAWN+LDA (they model the waveform of a stimulus-locked transient potential, and this protocol's classes are not defined by one).

Not run under this protocol: Spectral ridge.

Pick-two protocol comparison

Compare any two methods on Idle & command

Inspect point estimates and 95% intervals side by side on a shared axis. Paired differences appear only where a paired interval is published. Dropdowns are grouped by family alphabetically — never ordered by score.

0%25%50%75%100%CSP+LDA23.3% / 0.0%LaBraM18.3% / 3.3%

No paired interval is published on the idle protocol; compare summed trial rates (n = 4 cohort fact) directly.

Default view (Idle & command, cohort n = 4): CSP+LDA command detection 23.3%, idle false activation 0.0%; LaBraM command detection 18.3%, idle false activation 3.3%.

How these rows were run

Riemannian and xDAWN baselines

Recipe
TS+LR, MDM: OAS · idle-native windows
Protocol
For each subject, the first two blocks train the model, the third selects the threshold, and the final block is held out for testing.
Selection
One fixed configuration per method: library or MOABB-reference settings with the stated deviations.
Hardware
TS+LR, MDM: Apple M5 / CPU

Idle: differences in summed test trials

Differences in summed test-trial counts against the published idle rows; descriptive, no interval. Against EEGNet: against the matrix's one-seed EEGNet cell; training variance not included. Against CSP+LDA: against the published CSP+LDA idle row (deterministic).
MethodAgainst EEGNet · Idle trials falsely triggeredAgainst EEGNet · Command trials detectedAgainst CSP+LDA · Idle trials falsely triggeredAgainst CSP+LDA · Command trials detected
TS+LR+4+20+4+7
MDM+1+20+1+7

Beside the core matrix

Further foundation encoders, frozen

Each encoder replaces LaBraM or CBraMod in the frozen recipe above; the windows, people, folds, per-person linear head with its calibration-only threshold and scoring stay the same. Rows are grouped by model family; they are counts with no interval, so no row is called above or below another. Every encoder sees this protocol’s published windows, usually shorter than its pretraining context.

EncoderTraining modeCommand detection ≤3 sIdle false activationWindow balanced accuracy · AUROCChannelsPeoplePretraining exposure
REVE Base1Foundation modelFrozen encoder + linear head20.0%12 of 60 commands0 of 4 people always abstained1.7%1 of 60 idle trials61.1%AUROC 0.786174not in the authors' published pretraining list (checked 2026-10-04)source ↗
REVE Large2Foundation modelFrozen encoder + linear head10.0%6 of 60 commands2 of 4 people always abstained1.7%1 of 60 idle trials54.4%AUROC 0.785174not in the authors' published pretraining list (checked 2026-10-04)source ↗
LUNA Base3Foundation modelFrozen encoder + linear head6.7%4 of 60 commands2 of 4 people always abstained1.7%1 of 60 idle trials52.8%AUROC 0.675174not in the authors' published pretraining list (checked 2026-10-04)source ↗
LUNA Large4Foundation modelFrozen encoder + linear head8.3%5 of 60 commands2 of 4 people always abstained1.7%1 of 60 idle trials54.7%AUROC 0.734174not in the authors' published pretraining list (checked 2026-10-04)source ↗
BrainOmni Base5Foundation modelFrozen encoder + linear head21.7%13 of 60 commands1 of 4 people always abstained5.0%3 of 60 idle trials59.7%AUROC 0.737174not in the authors' published pretraining list (checked 2026-10-04)source ↗
CodeBrain6Foundation modelFrozen encoder + linear head16.7%10 of 60 commands1 of 4 people always abstained1.7%1 of 60 idle trials58.6%AUROC 0.771174not in the authors' published pretraining list (checked 2026-10-04)source ↗
EEGMamba7Foundation modelFrozen encoder + linear head3.3%2 of 60 commands3 of 4 people always abstained0.0%0 of 60 idle trials51.1%AUROC 0.639174not in the authors' published pretraining list (checked 2026-10-04)source ↗
ST-EEGFormer Base8Foundation modelFrozen encoder + linear head26.7%16 of 60 commands1 of 4 people always abstained1.7%1 of 60 idle trials63.9%AUROC 0.896174not in the authors' published pretraining list (checked 2026-10-04)source ↗
ST-EEGFormer Large9Foundation modelFrozen encoder + linear head33.3%20 of 60 commands1 of 4 people always abstained1.7%1 of 60 idle trials67.8%AUROC 0.912174not in the authors' published pretraining list (checked 2026-10-04)source ↗
eeg-fm-masking MAE (r=9 cm, L=2)10Foundation modelFrozen encoder + linear head5.0%3 of 60 commands2 of 4 people always abstained0.0%0 of 60 idle trials53.6%AUROC 0.733174not in the authors' published pretraining list (checked 2026-10-04)source ↗
ERP-FM Base11Foundation modelFrozen encoder + linear head5.0%3 of 60 commands2 of 4 people always abstained1.7%1 of 60 idle trials52.5%AUROC 0.675174not in the authors' published pretraining list (checked 2026-10-04)source ↗
SingLEM12Foundation modelFrozen encoder + linear head10.0%6 of 60 commands2 of 4 people always abstained8.3%5 of 60 idle trials50.3%AUROC 0.528174not in the authors' published pretraining list (checked 2026-10-04)source ↗
ZUNA 1.113Foundation modelModel card: research use only, not for diagnosis or clinical use.Frozen encoder + linear head18.3%11 of 60 commands0 of 4 people always abstained1.7%1 of 60 idle trials57.2%AUROC 0.608174Exposure unknownunknown: the authors do not list their pretraining data closely enough to decidesource ↗

Pretraining exposure is what each model’s authors publish as their pretraining list, checked on 2026-10-04: a dataset absent from it is “not in the authors’ published pretraining list”, which is not proof that its recordings were never seen. ST-EEGFormer lists BETA and SingLEM lists TMNRED; ZUNA 1.1 publishes no list, so every one of its cells is unknown; EEGMamba’s list is read from its official code (its paper was not read), with medium confidence.

Row footnotes

  1. Input removes the per-segment per-channel mean before the published /100 scaling (declared before scoring); a no-mean-removal sensitivity run scores P300 0.71 percentage points higher and sleep 1.40 lower.
  2. Input removes the per-segment per-channel mean (declared before scoring); the no-mean-removal sensitivity run scores P300 2.06 percentage points lower and sleep 0.93 lower.
  3. Resampled to 256 Hz and truncated to whole 40-sample patches; 1-2 s windows and 4-8 channel montages are outside the authors' tested regime.
  4. As LUNA Base. Frozen probes only.
  5. p300-target and semantic-target not run: the 2 s tokenizer window cannot be met by 1 s segments without zero padding. Upstream common average reference kept on 4-8 channel montages.
  6. Published CBraMod pooling (patch mean, channels kept); LoRA effectively adapts only the V projection (1-token attention window upstream).
  7. Published CBraMod pooling; LoRA on Mamba2 in_proj and out_proj (24 targets).
  8. BETA cells are pretraining-exposed (BETA is in the ST-EEGFormer corpus).
  9. BETA cells are pretraining-exposed. The authors' downstream class (1-based temporal index) was used; Large may have been pretrained with a 0-based index. Frozen probes only.
  10. Paper-recommended masking geometry; channel-retained pooling (C x 512). Three sibling checkpoints appear in the masking ablation panel.
  11. In design: p300-target and semantic-target. The other six protocols are negative controls outside the model's design. ERP windows have no pre-stimulus baseline.
  12. semantic-target is pretraining-exposed (TMNRED is in the SingLEM corpus). Each channel is encoded independently; the last 0.25 s of each 2 s segment is not seen by the tokenizer.
  13. Pretraining exposure unknown for every core dataset (no published source list). Encoder only; no channel upsampling or reconstruction.
  14. A sibling of the paper-recommended checkpoint in the masking ablation: the same architecture, corpus and recipe, another masking framework or geometry.
Masking ablation: three sibling checkpoints of eeg-fm-masking

Intervals are marginal: the same people under two checkpoints are not tested as a pair.

EncoderTraining modeCommand detection ≤3 sIdle false activationWindow balanced accuracy · AUROCChannelsPeoplePretraining exposure
eeg-fm-masking MAE (r=9 cm, L=2)10Foundation model · matrix row aboveFrozen encoder + linear head5.0%3 of 60 commands2 of 4 people always abstained0.0%0 of 60 idle trials53.6%AUROC 0.733174not in the authors' published pretraining list (checked 2026-10-04)source ↗
eeg-fm-masking JEPA (r=9 cm, L=2)14Foundation modelFrozen encoder + linear head20.0%12 of 60 commands2 of 4 people always abstained0.0%0 of 60 idle trials62.5%AUROC 0.823174not in the authors' published pretraining list (checked 2026-10-04)source ↗
eeg-fm-masking MAE (r=one channel, L=1)14Foundation modelFrozen encoder + linear head8.3%5 of 60 commands2 of 4 people always abstained0.0%0 of 60 idle trials57.2%AUROC 0.768174not in the authors' published pretraining list (checked 2026-10-04)source ↗
eeg-fm-masking JEPA (r=one channel, L=1)14Foundation modelFrozen encoder + linear head8.3%5 of 60 commands2 of 4 people always abstained0.0%0 of 60 idle trials54.7%AUROC 0.795174not in the authors' published pretraining list (checked 2026-10-04)source ↗

Weights licences and terms

No model’s authors endorse these results, and no adapted weights or LoRA deltas were kept or shared.

Notes on these rows

Beside the core matrix

EEGPT, frozen

EEGPT ran in the same run as the published LaBraM row above, with only the encoder swapped: the same windows, people, folds, per-person logistic head with its calibration-only rule and scoring. Its figures are counts and cohort means with no interval, so it is not called above or below another row.

EncoderTraining modeCommand detection ≤3 sIdle false activationWindow balanced accuracy · AUROCChannelsPeoplePretraining exposure
EEGPTFoundation modelFrozen encoder + linear head11.7%7 of 60 commands1 of 4 people always abstained3.3%2 of 60 idle trials57.2%AUROC 0.780174not in the authors' published pretraining list (checked 2026-10-12)source ↗

Footnotes to this row

  1. Same run as the published LaBraM idle row; the published CBraMod idle row comes from another run, so a comparison with it is across runs. The window balanced accuracy and AUROC were computed from this run’s saved probabilities with the definitions of the further foundation-encoder rows; the run itself recorded the counts.
  2. Every core window is shorter than EEGPT’s 4-second pretraining window: 1 or 2 seconds, and sleep as fifteen 2-second segments. The encoder computes each 0.25-second patch the same way at any length; a shorter window gives fewer patches to average.
  3. The readout is the site’s fixed probe on the frozen summary tokens, as for the published LaBraM and CBraMod rows: ridge (alpha 100) on 2,048 pooled values, and on idle a per-person logistic head. It is not the authors’ downstream probe, a learnable spatial filter with a linear head, and the windows are fed without the average reference EEGPT was pretrained with.

EEGPT’s exposure is read from its authors’ published pretraining list, checked on 2026-10-12, as for the further encoders. Its pretraining corpus: Five public datasets, as the paper’s Table 1 lists them: PhysioMI (PhysioNet motor imagery and execution), HGD, TSU (the Tsinghua 40-target SSVEP benchmark), SEED and M3CV; 58 electrodes, 256 Hz, millivolts, average reference, 4-second windows.

Licence and terms

Frozen-probe comparison

LaBraM at Scale and REVE Small, frozen

Six LaBraM at Scale encoders and REVE Small’s encoder in the same frozen recipe as the rows above: the same windows, people, folds, head and scoring, with no random-initialization control.

Cohort size n = 4 (a protocol fact): idle trials with a false trigger, command trials detected within 3 s, and the window metrics.
EncoderIdle false triggersCommands detected ≤3 sMean window balanced accuracyMean window AUROC
LaBraM at Scale · Quark ladder
Quark-BLocal Ubuntu / CPU4 / 6015 / 6060.8%0.724
Quark-LLocal Ubuntu / CPU2 / 6013 / 6061.1%0.655
Quark-HLocal Ubuntu / CPU2 / 6015 / 6062.8%0.746
Quark-GLocal Ubuntu / CUDA2 / 6010 / 6057.2%0.722
LaBraM at Scale · Atom-L and Molecule-L
Atom-LLocal Ubuntu / CPU2 / 60its pretraining corpus includes this protocol's dataset20 / 6066.1%0.721
Molecule-LLocal Ubuntu / CPU1 / 60its pretraining corpus includes this protocol's dataset10 / 6060.6%0.723
REVE Small · early checkpoint (epoch 15 of 200)
REVE SmallLocal Ubuntu / CPU1 / 608 / 6054.4%0.680

Idle is a within-person protocol: each row gives the cohort size, trial counts and window metrics, with no contrast. LaBraM at Scale's pooled features are global on idle; the published LaBraM idle row kept the channel axis.

Unless a cell is flagged, its dataset is not in that collection’s published pretraining list (the LaBraM at Scale pretraining manifest at faa387d9, checked 2026-11-05); REVE Small’s is read from the REVE corpus list (checked 2026-10-04).

The Quark ladder’s paired differences across protocols, the Quark-G step among them, are on “Does pretraining help EEG foundation models like LaBraM and CBraMod?”.

Terms: LaBraM at Scale’s code and weights under MIT (Pan-Junyu/LaBraM-at-Scale @ a0b0ce27); REVE Small under the REVE Responsible Use License v1.0 (reve-small @ 3565bdcb), aggregate scientific results only. No weights are distributed.

Limits

Detection resets at each cue. Only three minutes of test idle are observed; these results cannot estimate natural, always-on false activations per hour.

The protocol, step by step

  1. For each subject, the first two blocks train the model, the third selects the threshold, and the final block is held out for testing.
  2. Class events 1/2 define task onset. Two-second history windows produce decisions at 2.0, 2.5 and 3.0 seconds.
  3. Two consecutive threshold crossings trigger one command. Thresholds use calibration data only and may reject every command.
  4. 17 channels at 250 Hz; one-way 4–40 Hz filtering. Normalization uses training data only; no future samples enter a decision window.
  5. Visual cues, original feedback and short calibration exposure can influence results. This is not a natural continuous-idle test.
  6. Additional models use protocol ds005342-sitting-cue-gated-expanded-mvp-v1; independent audit SHA256: 14445d3423b5268654a1f73f7e1aca8bb03a2f793b076f450883fea89c7bcc83.

Stability

Single fixed configuration; no test-set hyperparameter selection.

Pretraining exposure

Unknown unless explicitly documented; no unseen-pretraining claim.

Checked 2026-10-04. LaBraM: not in the authors' published pretraining list (checked 2026-10-04) source ↗; CBraMod: not in the authors' published pretraining list (checked 2026-10-04) source ↗. The released sentence above predates this check. Further encoders: the exposure column of their table above → · EEGPT: its own section above →

Notes on the methods

Model terms

Source and licence

ds005342

Credit OpenNeuro ds005342 contributors · version 1.0.3, doi:10.18112/openneuro.ds005342.v1.0.3; associated study doi:10.3389/fninf.2022.961089. Original author credits are retained at the linked source.

Licence CC0-1.0 ↗

Dataset record ↗ · Every result on this dataset →

The figures here are aggregate measurements computed by BCI Report under the licence above; the data remain their creators’.

Permitted scope Personal noncommercial research; aggregate results only

Privacy Only cohort aggregates are published here: no recording, no participant identifier, no per-person score. The full review note is in the protocol JSON ↓

Public-data register note Eligible only as an aggregate-only case study: remove all participant IDs/results and condition breakdowns, label n=4 prominently, avoid subgroup claims, and make no claim that the cohort is anonymous.

Rights reviewed · against doi.org ↗ · eegdash.org ↗ · doi.org ↗

Reproducibility record

Protocol id
ds005342-sitting-cue-gated-pilot-v1
Dataset release
OpenNeuro snapshot 1.0.3 · 30bb3b67d76ce007b822fed801e496fafe5598a0
Hardware
Apple M5 / MPS and CPU
Scoring stage sum
58.0 s · may include accelerator waiting
Audit record SHA-256
ce2474c547f5f11c256752b0a07466beb5b6928da34bb39783aaa4647973d2e4

Rows of core-matrix-update-20261106

Riemannian and xDAWN baselines

Protocol SHA-256
4310329a2d6dddaeff64a965750b64237b08220fc23717b1f3d0c1a48f56b055
Audit record SHA-256
ce1ee2a0345358c613aa89298c1cb90939df15d68d6c48298b99f9daa7f69e82
Hardware
Apple M5 / CPU

Downloads

This protocol's rows in core-matrix-update-20261106 · CSV ↓ · Core matrix: rows run under its exact protocols · JSON ↓

Other protocols

All protocols → · The core matrix on the home page → · All datasets → · All methods →

Cite this page

BCI Report (2026). Idle & command on ds005342. https://bci.report/protocols/idle/

Figures from releases core-matrix-update-20261106 (2026-11-06), foundation-checkpoints-update-20261106 (2026-11-06), eegpt-core-update-20261013 (2026-10-13), foundation-models-update-20261004 (2026-10-04), research-preview-20260920 (2026-09-20). Cite the upstream dataset as well: its credit is on this page.

Every release is archived on Zenodo: doi:10.5281/zenodo.23123296. BibTeX for the site and its releases → · CITATION.cff ↗