Correctly timed epochs show stable HFA memory effects and rising theta, with an early-half effect also present

This retest asks whether time-on-task findings from AutoSci e9c2d073 hold when word epochs use sample/fs timing rather than events.tsv onset in OpenNeuro ds004789 RAM FR1 session 0. The original analysis code was run verbatim on archived features computed from the same EDF bytes with both timings. With sample/fs timing, 24 of 26 subjects met the inclusion rule. The HFA (70–150 Hz) subsequent memory effect (SME) did not decrease from the first to second session half (+0.016 z, 95% CI −0.015 to +0.049; Holm p = 0.71), HFA did not drift (−0.023 z; Holm p = 0.71), and theta (3–8 Hz) increased (+0.138 z; p = 0.0076). These findings agree with e9c2d073. One reported pattern did not: the HFA SME was present in the early half (+0.028 z, p = 0.0085) as well as the late half (+0.044 z, p = 0.0044). The paired early-half difference was not significant (+0.032, p = 0.19, n = 20), so the change in significance pattern does not establish a difference between timings. Across paired subjects, theta drift was similar between timings (r = 0.85), while word-locked HFA SME measures had low correlations (r = 0.01–0.08). The retest includes a non-random subset of 24 patients, session 0 only, all bipolar pairs, and no region analysis. It assumes sample/fs is EDF time. The prespecified −0.05 z decision threshold is conventional; smaller decreases, down to −0.015 z, are not excluded.

intracranial EEG · subsequent memory effect · high-frequency activity · theta · time on task · epoch timing · reproducibility · free recall