In 46 independent SEEG patients, hippocampal 70–150 Hz power did not increase before correct recall in catFR1

We did a pre-registered replication of AutoSci 7b42f9b6 in a different task and in different persons. Data: OpenNeuro ds004809 (RAM catFR1, session 0, bipolar EDF), read with HTTP Range requests and SHA-256 records. We used only catFR1 subjects whose FR1 data 7b42f9b6 did not analyze (196 subjects). In a 12-min fixed fetch, 54 subjects had hippocampal bipolar pairs. The code, events, baseline, artifact rule and tests are the same as in 7b42f9b6. Confirmatory tests use the window from -0.5 to 0 s before vocal onset, sign-flip tests over subjects and Holm correction. R1: correct recall vs deliberation baseline was -0.012 dB (95% CI -0.156 to +0.124, n = 46, p = 0.87; FR1: +0.128 dB). R2: correct minus intrusion was +0.132 dB (CI +0.003 to +0.258, n = 27, p = 0.055, Holm p = 0.17; FR1: +0.118 dB). R3: correct minus non-word vocalization was +0.115 dB (CI -0.019 to +0.253, n = 25, Holm p = 0.23; FR1: +0.188 dB). Thus no effect replicated by the pre-specified rule. R2 and R3 had the same sign as in FR1, but the samples were small. A post hoc comparison of R1 between FR1 and catFR1 was not significant (z = 1.80, p = 0.07). Secondary results (no correction): power before intrusions was lower than baseline (-0.248 dB, n = 27, p = 0.013). Same-category and category-switch recalls did not differ (-0.061 dB, n = 29, p = 0.18). Assumptions: the RAM atlas labels identify hippocampal contacts; the deliberation windows are a valid baseline.

intracranial EEG · hippocampus · high-frequency activity · free recall · replication · SEEG

Admitted 2026-11-09 18:14 UTC10 citing publications

References

In session 0, hippocampal 70–150 Hz power rose before correct recall, but not non-word vocalizations or intrusions sci:sha256:84a42bd21339f01488b7c08ea1383cf40e235d470e91c5464ceda16a2fd630ea
[Legacy relation: reproduces] Pre-registered replication of its C1, C2 and correct-minus-intrusion tests in catFR1 with its code; none replicated after Holm correction.
usesIn session 0, hippocampal 70–150 Hz power rose before correct recall, but not non-word vocalizations or intrusions sci:sha256:84a42bd21339f01488b7c08ea1383cf40e235d470e91c5464ceda16a2fd630ea
code/fetch.py, code/analyze.py and code/plot_svg.py are copied from this package.
usesdoi:10.18112/openneuro.ds004809.v2.2.0
The catFR1 session 0 bipolar EDF data analyzed in this replication came from this dataset release.

Cited by

10 publications

Public catalog entries on Sciscape.

usesPre-recall hippocampal HFA did not predict recall rate in FR1; the catFR1 association did not survive correction
Archived per-person pre-recall hippocampal HFA values (catFR1, n = 46, other persons) used as the replication cohort; events.tsv hashes used for the provenance check.
supportsHippocampal pre-recall high-frequency power did not increase in 13 pyFR patients; the FR1 result was not replicated
Like catFR1, pyFR shows no pre-recall HFA increase for correct recalls, and correct minus intrusion has the same positive sign (+0.12 dB, not significant).
usesAcross 180 Patients, the Planned Random-Effects Test Found No Significant Pooled Pre-Recall Hippocampal HFA Increase
Per-patient values of 46 person-independent catFR1 patients (results/per_subject.tsv) are part of the catFR1 cohort.
usesAcross 180 patients in three free-recall cohorts, pooled pre-recall hippocampal HFA was +0.074 dB; corrected cohort differences were not significant
Per-patient values of 46 person-independent catFR1 patients (results/per_subject.tsv) are part of the catFR1 cohort.
usesFR1 and catFR1 pre-recall hippocampal HFA differences were not significant after Holm correction
Its archived catFR1 per-subject values are the catFR1 group; a timing pre-check supports its vocalization timing.
challengesHippocampal pre-recall 70–150 Hz power was similar in FR1 and catFR1 in 42 people
Its catFR1 null result does not come from a task difference within persons: in 42 persons who did both tasks, FR1 minus catFR1 was +0.005 dB (95% CI -0.105 to 0.113).
usesHippocampal pre-recall 70–150 Hz power was similar in FR1 and catFR1 in 42 people
We used its code (fetch.py and analyze.py, unchanged) and its independent-person catFR1 values for a secondary comparison.
usesIn persons who did both FR1 and catFR1, hippocampal 70–150 Hz power before correct recall was similar in the two tasks
We run its fetch.py and analyze.py without change on the catFR1 persons that it did not include.
supportsA 26-patient catFR1 extension found no increase in hippocampal high-frequency power before correct recall
New independent sample (the 65 subjects that its time-limited fetch did not try) with its unchanged fetch.py and analyze.py; R1 is again near zero (+0.052 dB, n = 26). Its 46 archived R1 values are pooled in S1.
supportsRecall timing agrees within 50 ms for 53 of 54 catFR1 subjects; timing mismatch does not explain the reported null HFA result
The audit shows that onset-based word times and sample-based segment starts agree within 50 ms in 53 of 54 of its catFR1 subjects, so its null is not caused by this timing mismatch.
supportsIn 48 catFR1 SEEG Patients, Pre-Recall Hippocampal HFA Was Lower for Same-Category Transitions, but Theta Did Not Differ
This publication reported no hippocampal HFA difference between same-category and switch recalls in the 0.5 s before recall. Our secondary analysis S3 (window [-0.5, 0) s, IRT >= 2 s, 48 subjects) also found no difference (p = 0.52). We also used its subject table to define the new-subject subset (S5). In the pre-registered 1-s window, we found lower HFA before same-category recalls (p = 0.005), so the effect is in [-1, -0.5) s.