A selected response–difference mean rank test passed after spatial mean removal in SEEG and two fMRI cohorts; the primary test failed

We tested a rank link between sound response size and differences between repeats. One repeat supplied response size. Two other repeats supplied squared disagreement. We removed mean ranks for three source sound groups at each site. Separate human studies supplied electrical high-gamma SEEG and blood oxygen level dependent fMRI. A local plan fixed the score before development and reserved tests. Tests used six SEEG patients, 16 B2021 adults and eight NH2015 adults. Mean scores were 0.04223, 0.02172 and 0.01090. Their 95% person intervals were [0.03051, 0.05394], [0.01477, 0.02867] and [-0.00595, 0.02774]. SEEG and B2021 passed person tests. NH2015 did not confirm a positive mean. We kept the fixed reference screen. The single shared rule failed, with p=0.0849572. Shared folds limit the reassignment reference. Mean inference rests on person tests. No positive primary three-cohort claim is established. A specified spatial mean control was selected after outcomes. Its mean scores were 0.03279, 0.01318 and 0.01447. Their person intervals were [0.02544, 0.04015], [0.00985, 0.01651] and [0.00753, 0.02142]. Under person Student models, the mean claim passed a 46-claim Bonferroni bound, p=0.039145. Its majority claim failed. This is exploratory, with no independent confirmation. The links are small and concern selected sites and fixed sound panels. They do not identify a common neural noise source. Exact models gave positive scores with constant latent error variance after nonlinear measurement, or with fixed repeat gain drift. All negative scores and checks remain. Separate source reconstruction and scalar rank code checked every included record. The package supplies code, aggregate results, uncertainty draws, figures, source versions and terms, local locks and repeat steps. Original neural arrays stay external. Different cohorts, source selection, broad sound groups, three repeats, unequal filters and tasks limit wider transfer. No cause is established.

human auditory cortex · SEEG · fMRI · repeat disagreement · response variability · measurement