A repeat area test gave positive means for nine vocal sounds in eight SEEG patients and two fMRI cohorts

We tested whether responses within a nonspeech human voice group require more than one stable spatial pattern. One training repeat selected two patterns. Two other repeats supplied a signed determinant score. Under stable conditional means and independent zero-mean test errors, its mean is zero for any rank-one response matrix. A local plan and code lock preceded all new human scores. Separate public studies supplied eight SEEG patients, 20 B2021 fMRI adults and eight NH2015 fMRI adults. Each source supplied nine vocal sounds and three repeats. Mean scores were 0.10042, 0.07838 and 0.04207. Their 95% person Student intervals were [0.02830,0.17253], [0.04558,0.11117] and [0.00795,0.07619]. The one fixed shared mean test passed, p=0.01125. Two patients and one NH2015 adult had negative scores. The SEEG positive-person sign reference did not pass. Every one-sound deletion kept positive cohort means. Intervals for panels of up to 28 sites included zero in all cohorts. A separate matrix implementation matched all 36 scores within 3.4e-16. An exact model showed why the neural rank interpretation is conditional: selecting the two repeat-agreement cases among eight independent noise-sign cases changed the mean score from zero to 12/13, although unconditional response means had rank one. This gate was not the source fMRI gate. Source site screening used repeated measurements; we cannot verify the conditional error assumptions in the human release. Thus the result supports a shared processed-measurement assay, not an unconditional neural rank claim, equal latent axes, semantic coding or causality. The package gives code, source versions and terms, fixed plans, all outcomes, model cases, figures and repeat instructions. Original neural arrays stay external.

human neuroscience · SEEG · fMRI · auditory cortex · vocal sounds · response rank · cross-validation · source selection · reproducibility