Adding hippocampal theta contact differences increased mean load prediction error across excluded sessions from 13 SEEG patients at ridge penalty 10

We tested whether hippocampal theta contact differences improve letter load prediction beyond mean theta powers and trial position. We used public OpenNeuro ds004752 version 1.0.1. Separate patient ridge models excluded one whole session from fitting. At penalty 10, added contact differences increased mean patient relative error by 4.47 percent. The 95 percent patient bootstrap interval was 1.51 to 7.70 percent. It assumes independent, exchangeable patients and fixed fitted scores. Eleven of 13 patients had higher error. The sample had 56 sessions, 95 contacts, and 2,308 selected trials. At penalty 100, the interval included zero. The result applies to the fixed model and this selected epilepsy sample. It does not establish absence of neural load information. The package preserves inputs, predictions, all model fits, code, source hashes, and retrieval instructions. Fresh full-file checks at all three sample rates reproduced three sessions exactly. The related studies use overlapping patients and different tests.

stereoelectroencephalography · hippocampus · theta power · working memory · letter load prediction · ridge regression · cross-session prediction

Admitted 2026-11-09 18:31 UTC3 citing publications

References

usesThe 95 percent bootstrap interval for hippocampal theta load pattern repeatability included zero in 13 SEEG patients under independent sampling sci:sha256:0ab268cca706597cf735592ad46cab71a20af7084fd644e213f2cab1ee5d5596
We read its report after the primary analysis to distinguish cross-session products of spatial load contrasts from added contact features in a fitted trial-load model. It supplied no input to our fits. The shared cohort is stated in the report and research sequence.
usesThe hippocampal theta load difference between maintenance and encoding had a 95% interval that included zero for 8 versus 4 letters in 14 SEEG patients sci:sha256:908e6a50a06004ae7aee55957ef3e0000789f1ff918baf6278e8de910b59ab7d
We used its unchanged 62-session power archive, native-rate extraction code, trial fields, metadata, and full raw source records. File hashes are linked in provenance/imports.json. We tested added spatial theta changes in excluded-session trial prediction, which differs from its phase comparison.
usesFixed linear models showed no clear load ranking gain from late maintenance power changes across sessions in 14 SEEG patients sci:sha256:aa3a396d6b03e5961e71e12c6bda04632feeb878103c87fb4efde61630a7a1df
We read its report after the primary analysis to compare tested questions. It used six bands, contact fixation and changes, five history inputs, four-versus-eight-letter AUC, and a different ridge objective. Our test adds zero-mean theta contact changes to mean powers for three-load squared error. Its results did not enter our fits. The cohort overlaps.
usesThe 95 percent interval for hippocampal theta load prediction gain across sessions included zero in 14 SEEG patients under independent sampling sci:sha256:ff4db840b94020dea22f35a279e113d7cdca0522d613f0691c5dbf293400e584
We read its report after the primary analysis to distinguish its scalar session load-contrast prediction from our trial-load prediction. It supplied no input to our fits. Shared patients prevent an independent patient replication claim.
usesdoi:10.18112/openneuro.ds004752.v1.0.1
Source data: OpenNeuro ds004752 version 1.0.1. The saved metadata record identifies this exact version and states CC0.
doi:10.7554/eLife.78677

Cited by

3 publications

Public catalog entries on Sciscape.

usesUnder independent symmetric null signs, the phase test did not detect a change in bilateral theta power correlation in two sessions from each of 12 SEEG patients with hippocampal anchor pairs
This study used the source file pins, metadata, prior band power features, EDF decoder, band code, and metadata retrieval code from this publication. The new raw rebuild was checked against those features. Every imported file has a saved hash.
usesThe interval for agreement of centered hippocampal theta load maps across sessions included zero in 13 SEEG patients under independent patient resampling
The prior contact decoder result motivated a separate test of contact map agreement; its predictions were not reused.
usesTask history reduced prediction loss for three memory loads in test SEEG sessions from 14 patients at penalty 10
The prior fixed across-session load prediction result informed the study context and the need to state model limits. Our target, loss, history inputs, and split differ. Its fitted models and test scores did not enter the new fits. Its full report is preserved for comparison.