Fixed tests of six posterior EEG channels and V1/V2 fMRI did not show shared spread ranks or mean pair relations for 51 videos

We tested shared video relations in separate public human EEG and fMRI studies. Development used one EEG adult, two fMRI adults and 51 odd video IDs. The reserved panel used five other EEG adults, eight other fMRI adults and 51 even IDs. The first endpoint tested within-session repeat spread ranks after strength controls from other sessions, motion controls and retained counts. Mean person correlations were -0.020 and -0.017, with pointwise 95% person intervals [-0.229,0.188] and [-0.093,0.058]. After this failed test, we added a mean pair-relation endpoint. Its own code lock preceded its outcomes. Cross-half Gram relations had motion and additive video terms removed. Mean person correlations were 0.013 and 0.017, with intervals [-0.028,0.054] and [0.002,0.032]. Its video-identity p was 0.1525. Nominal two-endpoint Holm p values were 0.855469 and 0.416534. Neither shared rule passed. The positive fMRI mean interval alone did not confirm a shared effect. Pair dependence, conditional references, adaptive question choice and public cohort reuse limit inference. Sparse scalp sampling, source processing, audio in EEG but silent MRI, finite repeats and poor spread stability also limit neural interpretation. All negative and discordant results remain. The package supplies compact trials, source versions and terms, code, plans, figures, uncertainty draws and repeat steps. Independent pair and projection calculations checked all 13 people in each endpoint. MRI precision and 48 immutable EEG range reads passed. Whole EEG neural object hashes were not verified. No positive shared human finding or common neural cause is established.

EEG · fMRI · natural videos · repeat spread · mean pair relations · negative results · reproducibility · cross-modal benchmark