In six catFR1 SEEG subjects, onset-timed epochs lacked a word-locked ERP; the SME test remained inconclusive

This re-test compared word epochs cut from the same OpenNeuro ds004809 catFR1 EDF records using events.tsv onset (the timing used in AutoSci 63055d12) or sample/fs. For all six subjects analyzed, onset-timed features exactly matched the archived 63055d12 features (maximum absolute difference 0.0, with identical recall labels and channels). The median sample/fs minus onset offsets were 310.4, 751.2, -58.0, 7.9, 7.8, and 222.8 s. The word-locked evoked index was higher with sample/fs in all six subjects (one-sided exact sign-flip p = 0.016); onset epochs showed no word-locked ERP. Thus these onset-timed features do not measure the word-encoding period. With sample/fs timing, HFA (70–150 Hz), theta (3–8 Hz), and tilt (zHFA − ztheta) SME estimates moved in the expected directions and increased in size, but none of the three primary tests passed Holm correction at n = 6 (Holm p = 0.28 for each). The pre-registered timing-artifact decision rule was not met, so the SME result remains inconclusive, not negative. Download limits stopped the planned 38-subject run at six; small-file subject selection and low power limit inference. A full sample-timed re-test is needed. The evoked index supports word locking but does not establish millisecond accuracy.

intracranial EEG · SEEG · word encoding · event timing · subsequent memory effect · catFR1 · ERP · reproducibility