In 42 amygdala units from 5 patients, firing was not reliably higher in 24-s fearful-face than landscape film blocks

We tested if human amygdala neurons fire more in aversive film blocks (fearful faces) than in neutral film blocks (landscapes). We used the microwire spike times in the NIX files of OpenNeuro ds003374 v1.1.1. Only 5 of 9 subjects have units (42 units). We wrote the plan before we read the spike data. The unit effect is the mean log rate (0-24 s) in aversive blocks minus that in neutral blocks. The subject mean effect was +0.070 (rate ratio 1.07). 3 of 5 subjects were positive (exact sign-flip p = 0.56). The unit mean effect was +0.036, with a 95% subject-cluster bootstrap CI of -0.10 to +0.21. 5 of 42 units had p < 0.05 in an exact within-unit block-label permutation (4 increases, 1 decrease; 2.1 expected; binomial p = 0.057). The sensitivity tests (raw Hz, offset 1 Hz) and the exploratory onset response gave the same result. Important: with 5 subjects, the smallest possible exact p is 0.0625. Thus the primary test could not reject, and the result is not evidence of absence. The file-size matches of sub-05/08 and sub-06/09 are not duplicates (different hashes and metadata; no units). The result agrees with earlier macro SEEG band-power nulls on the same dataset. The package has the code, input hashes, a fetch script, per-unit and per-block tables, and all statistics.

amygdala · single-unit firing · fearful faces · intracranial EEG · spike trains · sign-flip test · OpenNeuro ds003374