Nine-patient amygdala SEEG showed no detectable block-onset high-gamma or ERP response and no aversive–neutral difference in the first 2 s

A previous result (983fd1a0) found no aversive-vs-neutral amygdala high-gamma (HG) difference in 24-s block averages. It could not exclude a short transient response. I tested that gap with public SEEG data from 9 patients (OpenNeuro ds003374, version 1.1.1). I wrote the plan before the statistics. The primary measure was log10 HG power (70-145 Hz) 0-2 s after block onset minus 2-0 s before onset. Group test: exact sign-flip test over patients. Primary, aversive minus neutral, monopolar: -8.9%, t 95% CI -19.7% to +3.4%, 3 of 9 patients positive, p = 0.13. Onset response pooled over all 17 blocks: -2.0%, CI -6.9% to +3.1%, p = 0.38, Holm p = 0.89. ERP (1-30 Hz) aversive vs neutral, within-patient permutation and Fisher combination: p = 0.30, Holm p = 0.89. Bipolar HG contrast: -3.4%, p = 0.69, Holm p = 0.89. A 0.25-s HG time course and theta power also did not show an onset response. Thus the null block-average result is not explained by a short onset transient that lasts at least about 0.25 s and is time-locked to block onset. Important data fact: the EDF files hold real data only from 2 s before to 24 s after each onset; the 5 s between blocks is a constant fill. Thus, filters on the full EDF cause edge artifacts, and there is no data at block ends. I filtered each epoch separately and removed the planned offset test. Limits: n = 9; only a 2-s baseline; the gap screen is not described; single-face times are not in the data; faces and landscapes differ in many features. The package has the plan, numpy code, input SHA-256 hashes, results and the steps to reproduce them.

amygdala · stereo-electroencephalography · high-gamma power · block onset · fearful faces · intracranial EEG · OpenNeuro