A fixed determinant test did not confirm visual map transfer between MEG and fMRI for 118 images

We tested changes in visual image relations between MEG and fMRI. We used a two by two determinant. An exact scalar model gives zero, even with a positive ordinary contrast. One human study supplied 92 images for development. A separate study supplied 118 images for a fixed test. MEG windows were 70–90 and 140–160 ms in both studies. The first five people supplied each reference. Cross-measurement person links were absent. We bounded results over all possible groups of ten target people. Fixed-test means over all people were 0.000252 in MEG and 0.000239 in fMRI. Smallest group means were -0.000098 and +0.000094. Outer 95% person Student limits were [-0.000428, 0.000996] and [-0.000060, 0.000582]. MEG had eight positive people. fMRI had thirteen. Each modality had fifteen people. The largest test p value was 0.910156. The full rule failed. Development failed its interval rule despite positive means and small p values. Controls kept all negative and discordant results. Unequal isotropic noise gave a positive ordinary contrast but a determinant near zero. A shared measurement term gave positive determinants with one neural pattern. Thus the score does not identify neural rank or causality. Cohort differences, fixed times and source processing limit the result. All person limits condition on the references and images. The package gives code, source versions and terms, all results, figures, audits and repeat instructions. The evidence did not confirm neural generalization between the two modalities.

MEG · fMRI · representational similarity · determinant · visual processing · replication · measurement noise