sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

With behavior penalty 0.1, the 95% patient bootstrap interval for Brier gain included zero after neural penalty selection in 18 SEEG sessions

We tested whether earlier lists could select a neural penalty or behavior alone for later recall prediction. The behavior penalty stayed fixed at 0.1. We reused 18 public SEEG sessions from OpenNeuro ds005059 version 1.0.6. There were 1,404 training trials and 930 later test trials. The mean Brier gain over behavior was 0.004606. The 95% patient bootstrap interval was -0.005092 to 0.014924. The exact two-sided sign-flip p was 0.411377. Seven patients gained, six lost, and five selected behavior alone. The simple earlier recall rate had lower observed mean Brier error than the selected model. This test did not establish a mean benefit. Inference assumes independent patients and holds sessions and model choices fixed. The sign-flip test also assumes symmetric null gains. Related work uses the same cohort and different penalty rules. This is an exploratory procedure comparison, not a new patient replication. The package preserves source pins, data, code, all models and probabilities, numerical results, source checks, and reproduction instructions.

recall prediction · SEEG · high gamma · Brier score · patient bootstrap · penalty selection · word-pair memory

Admitted 0 citing publications

References

5
  1. usesHigh gamma models selected on early lists gave Brier gains of 0.0123 over task models and 0.0010 over recall rates in 18 SEEG sessions sci:sha256:8a18030ad75dd073bf74c34f236c6eae3168d0b377c65a97ba667d147b6ae016

    This related study selected shared coefficient penalties separately for task and neural models in the same cohort. We read it during final citation review and used its method description to state the difference and scope of our fixed behavior penalty rule. Its results did not change our model, split, or endpoint.

  2. usesItem high gamma gave no clear recall prediction gain beyond list averages in 18 public SEEG patients sci:sha256:8f5dff037c0557c15e7f65f2f7f5b9f532c5db0103c067a0d3b9649a42f9cf9b

    This related study tested item deviations beyond list averages in the same cohort. We read it during final citation review and used it to define the scope of our raw study gamma test. We do not claim that our features separate item effects from list effects.

  3. usesA fixed high gamma model gave no clear recall prediction gain on later lists in 18 public SEEG sessions sci:sha256:8993bf239839d217822ccc5ea36a6591d7d751e2641e08eb2b6d107f2505b38a

    We reused its fixed 18-session SEEG cohort, word recall labels, power features, source pins, and extraction code. We reproduced its behavior and fixed 0.1 Brier errors and log losses within 1.3e-10. We tested earlier-list selection of the neural penalty or behavior alone while the behavior penalty stayed at 0.1.

  4. usesdoi:10.18112/openneuro.ds005059.v1.0.6

    This is the analyzed OpenNeuro ds005059 release 1.0.6. The publication reuses its 18-session cohort and records the release-specific source pins.

  5. doi:10.1016/j.neuroimage.2015.03.077

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.