A fixed test did not confirm shared variation directions in late MEG and IT fMRI maps for 118 images

We tested whether directions of variation among people carry between neural measurements after removal of image offsets and two calibration mean patterns. Public MEG and inferior temporal (IT) fMRI targets came from two human studies. Development used 92 images. A local plan and code lock preceded a separate 118-image test in 15 adults. Five rows per modality set four contrast directions. Unknown cross-modal person links left at least five independent targets outside both calibration groups. We bounded every possible five-person group. In the fixed test, possible mean score ranges were -0.0272 to +0.0270% for MEG targets and -0.0102 to +0.1973% for fMRI targets. Outer 95% person limits were -0.0749 to +0.0800% and -0.1910 to +0.4236%. Only 4/10 and 7/10 candidate scores were positive. The shared rule failed. All 20 development scores were positive, but development mean limits also included zero. A separate model used valid correlation distances, identical true neural patterns for all people, and independent transient error coefficients in four shared error templates. It passed the same rule. Thus a positive score need not identify stable neural traits. The contribution is a checked assay, a negative human benchmark and an explicit error example. Source selection, anonymous people, small calibration groups, dependent distance errors, unequal measurement filters and different image libraries limit scope. Code, source pins and terms, every score, uncertainty, figures and repeat instructions are supplied. Original neural matrices remain external.

MEG · fMRI · human neuroscience · representational geometry · individual differences · measurement error · negative result · anonymous person links