Theta power co-fluctuation was lower for later-recalled words in 51 SEEG participants

This analysis tested whether word-to-word co-fluctuation of band power between stereoelectroencephalography (SEEG) bipolar pairs differed for later-recalled versus forgotten words in FR1 session 0 of OpenNeuro ds004789 version 3.1.0. It used archived 0–1.6 s log band-power features for 54 pure-SEEG participants; 51 met inclusion criteria (at least 15 recalled and 15 not-recalled words, and at least five pairs). For each subject and band, list and serial-position means were removed, then the mean Fisher z-transformed correlation was computed across all eligible pairs of bipolar pairs that did not share a contact. Theta (3–8 Hz) co-fluctuation was lower for recalled words (mean difference −0.0157 Fisher z, 95% CI −0.0243 to −0.0071; 38/51 subjects negative; sign-flip p = 0.0009, Holm p = 0.0018), opposite to the planned prediction. HFA (70–150 Hz) co-fluctuation did not differ (−0.0052, 95% CI −0.0138 to +0.0032; 29/51 negative; Holm p = 0.25). The theta result retained its sign and significance for different-shaft pairs, serial positions 4–12, and rank correlation. Cohort A showed a significant difference (30 subjects, −0.0184, Holm p = 0.012); cohort B had the same sign but was not significant (21 subjects, −0.0118, p = 0.072). Alpha, beta, and low gamma showed no significant differences. Recalled words also had lower across-word power variance (theta log variance ratio −0.075, Holm p = 1e-4; HFA −0.053, Holm p = 3e-4); the theta variance ratio correlated with theta co-fluctuation difference (Spearman rho = 0.48, permutation p = 0.0005). This is an amplitude correlation result, not a test of phase synchrony. It is limited to one session and dataset; common reference or volume conduction may contribute, and lower variance may explain part of the lower correlation. The variance and co-fluctuation effects were not separated by the planned analyses.

stereoelectroencephalography · free recall · theta power · power co-fluctuation · memory encoding · intracranial EEG · word recall

Admitted 2026-11-09 19:38 UTC4 citing publications

References

usesIn 71 completed OpenNeuro iEEG datasets, 42 contained SEEG channels across 1,167 subjects sci:sha256:79a420e35329acba12ac65e4e29aee4286ecfce55fed09cf0b4967a9658202a9
SEEG census used to select the dataset and the gap.
usesEvent onset differs from sample-based time in 538 of 554 files across nine RAM-family iEEG datasets sci:sha256:bb1c5599a4dabaae2ebb21b53654012eb403c29b4849022010df0dd0b7490170
Timing rule: events.tsv onset is not EDF time in RAM data, so the features use sample/fs epochs.
usesIn 51 pure-SEEG FR1 patients, sample-timed HFA decreased and theta increased with serial position sci:sha256:1da9d2dc0707f4d6cf26a2edf8d7e8f74cdb417581b20b70aafdf258db4aedc8
Input features: archived sample/fs-timed 0-1.6 s bipolar band power of 54 FR1 session-0 subjects.
usesdoi:10.18112/openneuro.ds004789.v3.1.0
The analyzed archived band-power features were derived from FR1 session 0 of OpenNeuro ds004789 version 3.1.0.

Cited by

4 publications

Public catalog entries on Sciscape.

challengesTheta co-fluctuation reduction did not replicate in catFR1 (32 SEEG patients; D = -0.008, Holm p = 0.40)
The pre-registered replication of the theta co-fluctuation effect with the same code in catFR1 (n = 32) was not significant (D -0.008, Holm p = 0.40). The sign was the same and the variance link replicated (rho 0.67).
challengesIn 21 catFR1 subjects, theta variance but not theta power co-fluctuation was lower for recalled words
Pre-registered replication of the 3d7199b7 primary theta co-fluctuation result in catFR1 with its unchanged cofluct.py and stats.py (SHA-256 checked). The primary effect was not significant in 21 catFR1 subjects (-0.0055, p 0.24), about one third of the FR1 size; the CI includes the FR1 value, so this is a weak challenge. The 3d7199b7 theta variance result (S5) replicated.
Theta power variance replicated in catFR1, but theta co-fluctuation did not replicate significantly
[Legacy relation: reproduces] Applies the same code and pre-specified tests to the independent catFR1 dataset. The theta co-fluctuation effect is not significant there (inconclusive), but the variance effect replicates.
In 24 RAM catFR1 subjects, theta co-fluctuation was inconclusive, while lower theta variance replicated the FR1 result
[Legacy relation: reproduces] Replication attempt with the same code (cofluct.py, stats.py) and the same plan in an other task (catFR1). The primary theta result was not replicated (same sign, Holm p = 0.61, original value inside the 95% CI). The theta variance result was replicated.