An affine detector showed an uncertain hippocampal theta time change during memory delay in 12 SEEG patients

We tested time classified as sustained theta during a three-second verbal memory delay in public SEEG data. The sample had 12 patients, two sessions each, 44 hippocampal pairs, and 1085 trials with a zero source artifact flag. We removed the mean waveform within each load and session. The detector used 4 to 8 Hz wavelets, a fixed affine background, and a three-cycle duration rule. The estimated change was 1.845 percentage points. The interval permits both a fall and a rise. Early theta time was 6.870 percent. Late theta time was 8.715 percent. The 95 percent patient t interval for change was -0.939 to 4.629 percentage points. The exact two-sided test of patient sign assignments gave p=0.194336. This test requires independent patients and symmetric patient contrasts under the null. The quadratic background gave a change of 2.118 points, with an interval of -0.001 to 4.237. Four cycles, total signal, and no digital bounds gave positive pointwise intervals. These fixed checks are descriptive and have no correction for multiple comparisons. The result applies to the specified detector. It does not prove a change in pure neural oscillations. The package preserves full trial bipolar signals, raw examples, source identities, code, and all fixed checks. All 24 whole-source hashes, independent EDF checks, and exact numerical replay checks passed. This is a new measure in a previously studied cohort. It is not an independent replication.

intracranial EEG · hippocampus · theta oscillations · verbal working memory · burst detection · SEEG