A limbic response-shape difference in five SEEG patients was driven mainly by one patient

We tested whether hippocampus or amygdala stimulation sites produce ventral temporal cortex (VTC) response waveforms more similar to each other than to waveforms from nonlimbic sites at the same distance. In five patients from OpenNeuro ds004457, we analyzed 26 VTC bipolar recording pairs using a pre-analysis plan. The primary statistic was limbic–limbic minus limbic–nonlimbic waveform similarity after linear regression on stimulation-site distance. Within-patient permutation tests combined by Fisher’s method gave p = 0.0088. This result does not establish a shared effect across patients: sub-1 contributed most, the Fisher p was 0.20 without sub-1, and the five-patient sign-flip test was not significant (4/5 positive, p = 0.25). At split-half reliability r ≥ 0.7, Fisher p = 0.18. Limbic responses were not more similar to each other than nonlimbic responses were (limbic–limbic minus nonlimbic–nonlimbic, Fisher p = 0.44). Exploratory response size was greater for limbic sites in all five patients (mean 0.205 log10 units, about 1.6 times; two-sided sign-flip p = 0.0625). The analysis includes only five patients; sub-4 had three limbic sites and only 275 of 5,000 valid primary permutations. Anatomical labels came from dataset sidecars, distance control was linear in centroid distance, and stimulation contacts in the seizure onset zone were not excluded. The live OpenNeuro S3 files were identified by recorded keys and hashes; equivalence to an immutable dataset snapshot was not verified.

SEEG · ventral temporal cortex · limbic stimulation · response waveform · intracranial EEG · permutation test · OpenNeuro ds004457