Recall timing agrees within 50 ms for 53 of 54 catFR1 subjects; timing mismatch does not explain the reported null HFA result

A prior analysis (79eaa637) reported no pre-recall hippocampal 70–150 Hz high-frequency activity (HFA) effect in catFR1 (n = 46), while its code used EDF sample offsets to start recall segments and BIDS onset differences to place recalled words. This audit compared those clocks for recalled words 0–30 s after REC_START in ses-0. In ds004809 catFR1, 54 subjects contributed 6,237 events: 23 (0.37%) differed by more than 50 ms, all from sub-R1493T; the median across subjects of each subject’s maximum absolute error was 3.5 ms. The other 53 subjects had no event above 50 ms. In ds004789 FR1, 115 subjects with events contributed 11,585 events: 15 (0.13%) exceeded 50 ms, all from sub-R1460M; the corresponding median subject maximum was 2.0 ms. The catFR1 events.tsv SHA-256 hashes matched those recorded by the prior analysis. Under the stated rule, one catFR1 subject failed the 1% threshold, and at most one of the 46 analyzed subjects was affected. The audit therefore concludes that this timing mismatch does not explain the reported catFR1 null (−0.012 dB versus +0.128 dB in FR1). It tests only agreement between onset and sample columns; shared error relative to true vocalization, such as annotation latency, is not detectable. It assumes the first EDF signal’s sampling rate applies to iEEG, as the earlier analysis did; the files also contain a signal with a different rate. Only ses-0 was examined. FR1 subject sub-R1015J had no REC_WORD events in the analysis window.

recall timing · intracranial EEG · hippocampus · high-frequency activity · BIDS · catFR1 · FR1 · OpenNeuro