Fixed window tests did not confirm shared variance reduction in 7 iEEG and 8 fMRI patients after stimulation of the amygdala region

We tested whether amygdala-region stimulation reduces residual signal variance in two separate human cohorts. Four iEEG patients supplied development. Fixed tests used 7 other iEEG patients and 8 fMRI patients with different IDs. Electrical variance was measured across repeated pulses at 70–200 ms. BOLD variance was measured in matched 18 s ON and OFF windows after removal of a flexible mean cycle response, motion and drift. Patients were the test units. The geometric summaries of patient ratios were 1.0246 and 0.8949. Their pointwise 95% person t intervals were [0.9817, 1.0695] and [0.7786, 1.0286]. Variance fell in 2/7 and 5/8 patients. Shared decrease, shared increase and a development-led 80% retention claim all failed Holm correction. This does not establish absent neural quenching or a causal modality difference. A terminal-train input repair followed the first fMRI result; its unchanged output and the repair are preserved. Separate SVD fits checked all primary BOLD parcel scores. The package gives independent code, new aggregate data, all outcomes, source versions and terms, fixed plans, figures, audits and repeat steps. Original epochs, coordinates and licensed BOLD time series remain external. Sparse clinical sampling, different pulse protocols and sensing scales, finite repeats, source processing, motion and signal loss limit scope. Cross-modal generalization was not established.

human neuroscience · intracranial EEG · functional MRI · electrical stimulation · neural variability · reproducibility