A fixed test of repeat distances did not confirm shared tree rejection in six SEEG patients and 24 fMRI adults

We tested whether fixed relations between four sounds reject an additive tree in separate human SEEG and fMRI studies. Independent repeats supplied squared response distances. The score kept a minimum within each person and quartet before averaging. Under stable repeat means and zero cross-repeat error covariance, concavity makes its expected value nonpositive for a tree. Two patients and four MRI adults supplied development. A local code and list lock preceded tests in six other patients, sixteen B2021 adults and eight NH2015 adults. Each modality had a separate list of 128 quartets. The MRI list transferred by audited sound ID and clip name to the second MRI study. Mean scores were -0.066239, +0.220456 and +0.138247 in squared calibration site SD units. Their 95% person Student intervals were [-0.211332,+0.078853], [+0.173437,+0.267475] and [+0.093962,+0.182531]. All 24 MRI adults and one of six patients had positive scores. The fixed shared rule failed, p=0.853286. Alternate SEEG calibration blocks gave positive means. Keeping spatial means left NH2015 uncertainty across zero. These checks do not replace the primary result. Exact examples show that channel mixing, stable nuisance and one scalar can give positive squared-distance scores. Selecting an order from test noise also gave a positive simulation mean for a true tree. The contribution is an audited assay, a repeated-response benchmark and a negative cross-modal test. Clinical sampling, source selection, different cohorts and filters, fixed sounds, few electrical patients and repeat assumptions limit scope. The package preserves code, source versions and terms, local plans, all results, figures and repeat instructions. Original neural arrays remain external. No common neural geometry or cause is established.

human auditory cortex · SEEG · fMRI · representational geometry · tree metrics · independent repeats · negative result · public data