Hippocampal high-frequency power did not differ significantly by recall contiguity in slow transitions

We tested whether hippocampal high-frequency activity (HFA, 70–150 Hz) before recall reflects temporal contiguity of recall transitions. We used OpenNeuro ds004789 RAM FR1 session 0 and hippocampal bipolar pairs. We fetched 126 subjects; 40 had at least 5 contiguous (|lag| = 1) and 5 remote (|lag| ≥ 2) transitions after applying an inter-response time (IRT) threshold of 2 s. The pre-registered test compared HFA in the 0.5 s before vocal onset. Contiguous minus remote HFA was +0.15 dB (bootstrap 95% CI [0.00, 0.35]); 26 of 40 subjects had a positive difference, and the two-sided sign-flip test gave p = 0.074. This did not meet the pre-registered significance rule. HFA did not change monotonically with |lag| (mean Spearman rho = −0.008, p = 0.82). Controlling for log IRT gave +0.15 dB, p = 0.057 (Holm 0.23). Early and post-onset windows showed no difference. In behavioral data from 126 subjects and 4,629 correct-to-correct transitions, 41% were contiguous; median IRT was 1.37 s for contiguous and 1.98 s for remote transitions. The 2 s threshold retained 29% of contiguous and 49% of remote transitions, so this test covers only slow transitions. Fast transitions, which are mostly contiguous, were excluded because the pre-onset window must not overlap the previous vocalization. Vocal onset times in events.tsv are assumed correct, and hippocampal labels come from dataset atlas labels. The bootstrap CI has a lower edge of 0.00 dB, but the pre-registered test is the sign-flip test. In this window, the data do not support a contiguity effect larger than about 0.35 dB.

hippocampus · high-frequency activity · free recall · temporal contiguity · intracranial EEG · serial position · memory retrieval · OpenNeuro ds004789