Early delay power gave no mean recall forecast gain in 13 SEEG patients excluded from fitting under assumed trial order

We tested recall error forecasts in public SEEG data from 13 patients. Each test patient was excluded from model fitting. Early delay power changed mean Brier loss from 0.20500 to 0.20515. The mean gain was -0.000149. Its descriptive 95 percent patient interval was -0.006370 to 0.006626. Six patients had a gain. Models fitted on each patient's earlier trials had lower mean loss, 0.17966. The test assumes that source trial IDs give task order. Past outcomes were available as forecast inputs. The study gives no clear evidence of an added mean gain from these power changes. The package preserves code, predictions, source hashes, figures, and a complete signal retrieval recipe.

SEEG · working memory · recall forecast · Brier loss · patient exclusion · intracranial EEG · signal power