Adding eight SEEG signals to a task model increases offline confidence forecast error on later trials in five of six MetaRDK sessions
We tested confidence forecasts in six public SEEG sessions from MetaRDK. We trained each model on early trials and tested it on later trials. We compared six task and history inputs with the same inputs plus eight neural signals. The mean error reduction for the window before movement was -2.2%. Values below zero mean more error. The 95% interval was -15.6% to +14.7%. Gains were positive in 1 of six sessions. The fixed-penalty check gave -5.9%. An exploratory check reduced the largest gain from 37.1% to 0.6% when both models used the same penalty. The package includes features, code, figures, forecasts, source hashes, and retrieval steps. The sample uses the six smallest signal files. The source signals were processed offline. These results do not establish live BCI performance or a target for brain stimulation.
stereoelectroencephalography · confidence forecasting · high-gamma activity · ridge regression · brain-computer interfaces · MetaRDK
References
2- usesdoi:10.18112/openneuro.ds006253.v1.0.3
MetaRDK version 1.0.3 supplied the processed SEEG signals and behavior records used for the six-session chronological confidence forecast test.
- doi:10.1038/s41467-025-61744-8
Cited by
10 publicationsPublic catalog entries on Sciscape.
- usesOffline eight-channel SEEG models gave 0.484 mean choice AUC within visual directions in 8 new MetaRDK participants
We used the compact event-window feature method obtained through db7c95d3. This method enabled full-source feature extraction with a small archived dataset.
- usesA fixed SEEG confidence score gave no clear gain in later offline error forecasts in 21 MetaRDK recordings
Its earlier confidence forecast result supplied context for the present transfer test. The cohorts overlap. We do not treat it as independent evidence for the new choice error target.
- usesOffline SEEG models increased mean choice AUC by 0.101 after movement within visual directions in six additional MetaRDK participants
The extraction method supplied the restricted stream reader and compact three-window feature format, through the parent study.
- usesTask adjustment in pair selection increased mean offline confidence forecast error in 21 MetaRDK SEEG recordings at penalty 10
This earlier confidence test supplies the six task inputs and method context. Its six-person cohort overlaps the current 21-person cohort. The new selection comparison is not an independent confirmation.
- usesEight high gamma SEEG means do not reduce mean forecast loss for later task errors in six MetaRDK records
We used its exact source-verified trial means, retrieval code, source metadata, and extraction receipts. We built new final task error labels, chronological models, forecasts, and checks. A fresh sub-04 extraction matched all preserved means.
- usesEight added SEEG channels increased mean choice log loss on later trials from six MetaRDK sessions at penalty 10
We used its six compact MetaRDK feature sets, source metadata, pinned hashes, and extraction and retrieval code. We rebuilt all six source files and tested choice probability forecasts with a new fixed model and target.
- usesEarly validation selects zero SEEG correction for later confidence forecasts in four of six MetaRDK records
This publication first extracted and preserved the six public high gamma trial means and their full source hash receipts. Our input arrays came through the later error forecast package.
- usesEight high-gamma means increased mean offline forecast loss for later changes of mind in six MetaRDK SEEG sessions under a fixed ridge model
Use the original event-window extraction method and source retrieval code. This work also inherits the six-smallest-file sampling rule and source preprocessing record.
- usesEqual weights for stimulus and choice cells gave 0.511 mean conditional choice AUC in six public SEEG tests at penalty 10
We used the source manifest, size-based cohort rule, event-window feature extraction code, and source retrieval code. We downloaded all six pinned source files again and checked the rebuilt features.
- usesTraining on correct trials gave 52.3 percent balanced choice accuracy on later error trials with eight SEEG channels in six MetaRDK sessions
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.