A fixed test did not confirm shared face and house coding in panels selected for decreases in six ECoG patients and six fMRI adults

We tested whether sites selected for response decreases to both faces and houses retain those decreases and support later category decoding. Separate public studies supplied surface intracranial EEG and BOLD fMRI. One ECoG patient supplied development. A local plan and code lock preceded tests in six other ECoG patients and all six fMRI adults. Earlier data selected sites and fitted a spatially centered centroid reader. Later data tested decoding and both response signs. Mean AUC was 0.53773 and 0.81352. The 95% person Student intervals were [0.48646,0.58901] and [0.62228,1.00476]. Four of six and six of six people were above chance. The fMRI part met its five planned one-sided criteria. The shared rule failed, p=0.957961. All five analyzable ECoG panels had positive face means in the test data. One patient had only one selected contact and kept the planned chance and zero-score fallback. An equal-window electrical check gave mean AUC 0.49627. All negative and discordant results remain. A later stationary-noise example showed that unequal power windows can give a positive mean log ratio without an event response. It was not fitted to people. The result does not confirm a common decrease code or establish a modality effect. Clinical and source region selection, separate stimuli and tasks, baseline choice, unequal sampling, small cohorts and missing nuisance traces limit scope. The package preserves code, source versions and terms, all outcomes, plans, figures, audits and repeat instructions. Original neural arrays stay external.

intracranial EEG · ECoG · fMRI · response decreases · negative BOLD · category decoding · independent validation · baseline choice