A six-cycle wavelet and linear background rule did not resolve a theta duration change across a three-second memory delay in 12 SEEG patients

We tested theta duration in public hippocampal SEEG data from OpenNeuro ds004752, version 1.0.1. The sample had 12 patients, 44 bipolar pairs, and 1085 trials from two sessions per patient. We removed the mean trial waveform within each memory load. We used one power threshold per session and pair. The threshold came from a linear fit of the log background spectrum. An event needed at least three cycles at 4 to 7 Hz. Mean occupancy was 0.76% in the first delay second and 1.23% in the last second. The mean change was 0.47 percentage points (95% t interval -0.15 to 1.10). The exact two-sided patient sign-flip test gave p=0.13916. This test used sign flips of patient contrasts and assumed symmetric null contrasts. All nine fixed checks had a positive mean contrast. The fixed primary rule did not resolve the mean duration change. The result does not establish no change. The wavelet can mix postprobe activity into the last delay second. The measure depends on its signal and threshold rules. It does not prove neural oscillations or a causal memory process. The sample was used in an earlier power study. This is a new endpoint in that sample. The package contains source code, fixed source hashes, filtered epochs, trial features, and repeat instructions.

hippocampus · theta activity · SEEG · verbal working memory · oscillatory bursts · event duration · intracranial EEG