A fixed area test did not confirm shared load direction changes in 12 SEEG patients and two fMRI samples

We tested whether memory-load responses change direction between encoding and delay beyond arbitrary changes in strength. One data block chose a plane. Two other blocks supplied a signed area. Its expectation is zero for one stable response direction under independent block errors. Public, separate studies supplied 12 SEEG patients with 50 eligible sessions, 16 fMRI adults and 11 other fMRI adults. Development used two eligible patients and three MRI people. Fixed test mean areas were +0.004334, +0.090675 and -0.017008. Their 95% person t intervals were [-0.001679,+0.010347], [+0.054251,+0.127099] and [-0.052561,+0.018546]. The single shared positive rule failed, with p=0.84423. A positive digital-bound electrical check did not replace that failure. A weaker phase-distance reference was positive in all groups but permits pure strength changes. Independent pair-coordinate arithmetic matched all 77 primary records. Simulations separated known strength and direction changes; shared block errors caused false positives. These are new-endpoint tests in reused cohorts, not new-cohort confirmation. The package gives the proof, code, source versions and terms, compact permitted inputs, plans, all failures, figures and repeat instructions. No shared neural direction change, equivalence, or causality is claimed.

working memory · SEEG · fMRI · response geometry · cross-validation · negative result · measurement error