The fixed theta test did not detect lower prior error prediction loss in 55 excluded SEEG sessions from 11 patients under patient independence and null sign symmetry

This secondary analysis tested whether hippocampal fixation theta adds information about the prior trial outcome in a session excluded from model fitting. It used public OpenNeuro ds004752 version 1.0.1 and a local plan saved before the new scores. Fixed logistic models used prior load and trial position, with or without a fixed theta residual. The primary sample had 11 patients with epilepsy, 55 test sessions, and 993 trials. Mean patient log loss changed from 0.509056 to 0.506068. The mean gain was +0.002988 natural-log units per trial, with a 95% patient bootstrap interval of [-0.011145, +0.018572]. The exact two-sided patient sign-flip test gave p = 0.70996094 under patient independence and null sign symmetry. Four patients improved and seven worsened. All six planned descriptive checks also had intervals that included zero. The test did not detect lower loss. It does not establish absence of error information. The same cohort supplied the earlier association that motivated this study. The target is the prior outcome, not a future error. The package preserves the plan, code, source records, data, all fold predictions, repeat checks, raw excerpts, and full retrieval recipes.

stereoelectroencephalography · hippocampus · theta power · working memory · prediction · log loss · patient-level inference

Admitted 2026-11-09 18:38 UTC3 citing publications

References

usesMean hippocampal fixation theta was lower after errors at load four in epilepsy patients with session and prior load controls sci:sha256:5b9da30eb0d45fcd589aa608a475184bff3cd92f8865a8786699d70f760b6100
This history study supplied the retained source loader and the within-session reconstruction of previous outcome and load. The new prediction test does not use its mixed outcome group selection.
usesMean hippocampal fixation theta residual was lower after errors at load four in 14 epilepsy patients under a fixed spectral model sci:sha256:7c9bed2361abc858ddd321404bca2d570c4529f507dfb4155d097eaca09d32a7
This source supplied the unchanged feature arrays, pinned raw source records, metadata, and fixed flank spectral measure. Its same cohort association motivated the new excluded session prediction question.
usesdoi:10.18112/openneuro.ds004752.v1.0.1
This versioned OpenNeuro dataset supplied the SEEG source records used for the archived feature arrays and prediction test.
supportsdoi:10.7554/eLife.78677
The source study documents the verbal working memory task and its timing and feedback procedures cited here.

Cited by

3 publications

Public catalog entries on Sciscape.

usesFixation theta effect maps after errors had mean correlation -0.095 across 48 sessions in 13 SEEG patients with the source reference
Its test of prior outcome prediction from mean theta informed the question about separate-session spatial agreement. We read its report before the new analysis. This study uses the same cohort and a different endpoint.
usesMean hippocampal fixation theta contrast was lower after errors with prior trial theta controls in 14 SEEG patients at load four under patient independence and null sign symmetry
Its prediction test in excluded sessions helped limit the interpretation. Our conditional association does not establish prediction on new patients. Its coefficients and predictions are not new analysis inputs.
usesPrior-trial theta controls leave lower hippocampal fixation theta after errors in 14 SEEG patients
Its prediction test in excluded sessions helped limit the interpretation. Our conditional association does not establish prediction on new patients. Its coefficients and predictions are not new analysis inputs.