An order test failed in six SEEG patients and eight fMRI adults; a two-repeat Gaussian product correction was exact in model checks

We tested whether one fixed scalar per sound can explain measured response means through monotone site functions. Separate public studies supplied human SEEG and BOLD fMRI. A local code lock preceded new scores in six patients and eight MRI adults. The 165-sound minimum test failed, with shared p=1. All person scores were negative. A 16-adult MRI extension also failed. This does not confirm a scalar model. We give a finite sign-test guarantee under conditional symmetric errors. We also apply a two-repeat Gaussian product correction. It is unbiased under independent equal-covariance Gaussian errors. Exact finite quadrature checked the formula. Dependent, unequal-covariance, and non-Gaussian errors can break it. This later exploratory estimator gave mean -0.013174 in SEEG, -0.199779 in the eight-adult MRI sample, and +0.859833 in the 16-adult MRI sample. Only the last person interval was positive. A source audit found identical spatial mean profiles across people in both MRI releases, and across scans in the original NH2015 matrix. Thus separate source records do not establish independent released errors. No positive cross-modal neural claim is confirmed. The package preserves code, source pins and terms, every scored result, model checks, figures, and repeat instructions. Original neural arrays stay external.

human auditory cortex · SEEG · fMRI · monotone model · Gaussian moment correction · repeat dependence · source data audit