For 48 images, mean human versus animal matching across faces and bodies was negative in EEG and positive in late MEG and IT fMRI

A fixed positive test failed across EEG, MEG and fMRI. We tested whether the human versus animal distinction carries across faces and bodies. The fixed panel had 48 images in four source groups. Separate studies supplied eight EEG people and a 15-person MEG and fMRI cohort. Raw mean matching scores were -0.053426, +0.019900 and +0.009378. Pointwise 95% person intervals were [-0.082234,-0.024618], [+0.010178,+0.029621] and [+0.004754,+0.014002]. Appearance control kept each mean sign. All eight EEG scores were negative. A local plan and a recorded extension preceded all new neural scores. The three-measurement positive rule failed, with Holm p=1. EEG signs remained in both repeat halves, every image deletion and all twelve fixed half-size panels. An image-and-person resample range included zero for IT. A direct increase test also failed. The related raw face/body contrast reproduced prior SCI values; a rectangle-adjusted MEG and IT test passed, with Holm p=0.021515. Its nuisance fit differs from the prior test. An exact model gave positive matching with a flat true map through image reliability differences alone. It was not fitted to people. The result concerns measured relations in fixed images and source windows. It does not establish a common neural category code, an independent new-cohort confirmation, or a causal modality effect. The package gives code, permitted EEG maps, source versions and terms, all results, figures, local plans, audits and repeat steps.

visual categories · EEG · MEG · fMRI · human images · pattern distances · cross-context matching · negative result · reproducibility