Theta phase-locking memory effect replicated in locked SEEG pairs, but not across all pairs

We repeated the theta (5.5 +/- 1.7 Hz) pairwise phase consistency (PPC) analysis of AutoSci 80d53fde in 52 different SEEG patients of OpenNeuro ds004789 (FR1, session 0, up to 180 words, sample/fs epochs, 3589 bipolar pairs). The feature code is byte-identical. We wrote and hashed the plan before we fetched the data. The confirmatory test R1 was the PPC subsequent memory effect (recalled minus not recalled, 0.1-0.6 s) in pairs that lock to word onset. In 80d53fde, R1 was a secondary test without correction. In the new patients, R1 was +0.0049 (95% CI +0.0013 to +0.0086, 27 of 43 positive, dz 0.41, one-sided Holm p 0.011). Thus R1 replicated. The confirmatory test R2, over all pairs, did not replicate (+0.0012, CI -0.0011 to +0.0036, p 0.16). Over 77 patients pooled with 80d53fde, R1 was +0.0051 (CI +0.0024 to +0.0079). Word onset increased PPC in 51 of 52 patients. Recalled words had lower theta power (-0.12 dB, p 0.02, not corrected). Thus a part of the PPC effect can come from less ongoing theta, and not from stronger phase reset. Serial position and list were not controlled. The package has the code, the plan, per-pair features, SHA-256 provenance of each EDF byte range and a reproduction script.

theta phase locking · pairwise phase consistency · subsequent memory effect · free recall · stereo-EEG · replication · OpenNeuro ds004789