Training on correct trials gave 52.3 percent balanced choice accuracy on later error trials with eight SEEG channels in six MetaRDK sessions
We tested whether a SEEG decoder trained on correct early trials followed recorded choice on later error trials. The model used eight selected bipolar channels, fixed time windows, and logistic penalty 10. We used public high gamma signals from six MetaRDK sessions in OpenNeuro ds006253 version 1.0.3. Mean balanced choice accuracy was 52.3 percent on 123 later errors. The descriptive person bootstrap interval was 45.5 to 60.1 percent. Accuracy was 52.6 percent on 324 later correct trials. The error-minus-correct difference was -0.23 percentage points. The planned test did not establish accuracy above 50 percent. It did not establish a selective loss on errors. The package adds a fixed error-trial assay with a proof and ideal-rule score checks. We rebuilt all six pinned source files and verified exact feature equality. Independent code checked 90 fits and 270 score groups. A full repeat gave identical numerical files. The archive includes code, compact features, trial predictions, source records, model states, and three figures. Choice labels are inferred from accuracy. The sample is small and reused. The source is processed offline signal. The study does not test a live interface or brain stimulation.
stereoelectroencephalography · choice decoding · error trials · brain-computer interface · balanced accuracy · MetaRDK
References
4- usesAdding eight SEEG signals to a task model increases offline confidence forecast error on later trials in five of six MetaRDK sessions sci:sha256:f153e0af57732bd1a666747e35778c555a9091f0c980875e17ea5b2460f566ee
We used its public SEEG feature extraction and source provenance through the later choice study. We independently rebuilt the six source files. We did not reuse its confidence scores as new results.
- usesEight added SEEG channels increased mean choice log loss on later trials from six MetaRDK sessions at penalty 10 sci:sha256:97e13fb0a793ade1295f735fbb0fc7a8d15bb94c768e56c76e84f4f65ce2055d
We used its six MetaRDK feature files, source metadata, retrieval code, label rules, channel selection, and logistic solver. The new test fits correct early trials and scores choice on later errors.
- usesdoi:10.18112/openneuro.ds006253.v1.0.3
The analysis used processed high gamma SEEG derivatives from OpenNeuro ds006253 version 1.0.3; the source URL, hashes, and retrieval recipe are preserved.
- doi:10.1038/s41467-025-61744-8
Cited by
3 publicationsPublic catalog entries on Sciscape.
- Offline SEEG readers trained on correct trials gave 48.6 percent balanced accuracy for inferred responses before movement on later errors in 15 MetaRDK records
[Legacy relation: reproduces] Earlier six-record correct-trial response-transfer test; exact numeric replay and unchanged numerical core.
- usesSignal subtraction reduced mean AUC on later errors by 0.1205 in four offline SEEG readers fit on correct trials with eight channels at penalty 10
Its correct-trial fitting rule and later-error assay guided the label-conflict test. We compared the signal after movement with the signal change across movement.
- usesSignal subtraction reduced mean AUC on later errors by 0.1205 in four offline SEEG readers fit on correct trials with eight channels at penalty 10
Its correct-trial fitting rule and later-error assay guided the label-conflict test. We compared the signal after movement with the signal change across movement.