A fixed test did not confirm a response level and repeat spread relation in human SEEG and two fMRI groups

We tested whether larger measured responses mark greater repeat spread within fixed nonvocal sound panels. Separate public human studies supplied SEEG high-gamma activity and two fMRI cohorts. Different repeats supplied response level and squared response difference. Disjoint sounds supplied site scales. Development used two patients and four MRI adults. Fixed tests used six other patients, sixteen other B2021 adults and eight NH2015 adults. Mean dimensionless covariances were 0.090441, 0.019761 and 0.016203. Pointwise 95% person t intervals were [0.039348,0.141535], [0.010382,0.029139] and [-0.008269,0.040675]. The declared three-cohort rule failed, conjunction p=0.080705. Global repeat-gain alignment retained a positive SEEG interval but left both fMRI intervals across zero. A ten-site fMRI check was discordant. An exact gain-only model gave a positive score with no additive noise. An added estimator check showed why skew can bias a same-repeat mean-variance covariance; human differences were not uniform across studies. All negative outcomes remain. An independent sound-pair calculation checked all 36 development and test people within 5.56e-17. Source selection, shared normalization, separate cohorts, finite repeats and measurement filters limit neural interpretation. The contribution is an audited assay, narrow raw measurement evidence, and a failed broad confirmation test. The package preserves code, source terms and hashes, plans, all results, figures and repeat instructions. Original neural arrays remain external. No shared neural noise law or cause is established.

human neuroscience · SEEG · fMRI · auditory cortex · measurement variability · repeat gain · sample skew · reproducibility