Ten category contrasts supplied 98% of covariance between photos in later MEG and IT fMRI maps, under assumed image order

We tested whether fine object relations persist between different photos in released human MEG and fMRI maps. One study supplied two photo sets and 15 adults. Plans preceded new neural outcomes. Five rows formed templates. Ten rows tested transfer. After image and category controls, later MEG transfer was 0.01313. Its 95% conditional person interval was 0.00238 to 0.02303, with Holm p=0.1875. IT fMRI transfer was -0.00049. Its interval was -0.00606 to 0.00525, with p=0.875. Both fixed tests failed. A later exploratory split assigned 98.43% of later MEG covariance and 98.11% of IT covariance between photos to ten category contrasts. These percentages describe observed group maps. They are not population fractions or neural variance. Larger category than residual covariance passed a conservative check for participant overlap in each modality. There is no independent confirmation. Limited image-feature controls retained about 98% shares. The hidden target has no image key. An order change within categories could preserve category agreement and remove fine transfer. All claims assume matched image indices. Source reliability also differed. The package preserves code, plans, source hashes and terms, all person and negative results, figures, numerical checks, and repeat instructions. Source arrays and photographs remain external. No causal or semantic mechanism is established.

MEG · fMRI · visual representation · photo transfer · category structure · representational similarity · negative result · participant dependence