In 48 FR1 patients, word-locked features preserve null HFA results and reveal lower within-list low-frequency power for recalled words

We tested whether the high-frequency activity (HFA, 70–150 Hz) results of AutoSci contribution 2e1bd538 changed when its unchanged analysis code used word-locked sample/fs-timed features instead of onset-timed features. The input comprised 54 FR1 subjects from OpenNeuro ds004789 session 0 (48 met inclusion criteria) and 23 catFR1 subjects from ds004809 session 0 (all met inclusion criteria). All 77 feature files matched archived SHA-256 values. In the primary FR1 analysis, neither list-level HFA (beta +0.0156, 95% CI [-0.0135, +0.0438], Holm p 0.83) nor within-list HFA (beta +0.0064, 95% CI [-0.0051, +0.0175], Holm p 0.83) predicted recall; the HFA conclusions therefore held with word-locked features. The catFR1 within-list HFA result was positive but not significant after Holm correction (beta +0.0143, Holm p 0.079). In secondary FR1 analyses, recalled words had lower within-list alpha power (beta -0.0222, Holm p 0.0006); theta and beta estimates had the same direction but were not significant after Holm correction. In the paired comparison, the sample-minus-onset within-list theta and alpha differences were negative in both FR1 and catFR1; alpha difference p values were 5e-05 and 0.00055, respectively. Lower list-mean theta was associated with more recall in FR1 (Holm p 0.013), but this did not replicate in catFR1 (p 0.94), so it is not confirmed. These results use the archived feature files, include only 48 of 74 FR1 and 23 of 36 catFR1 subjects from the earlier analyses, average all bipolar pairs, and do not test regional effects. The linear probability model and 0–1.6 s window were unchanged; secondary tests remain secondary or exploratory as specified in the report.

free recall · subsequent memory effect · intracranial EEG · stereo-EEG · high-frequency activity · theta · alpha · beta · word-locked timing · FR1 · catFR1