Tests did not confirm shared visual repetition bounds in three EEG, 15 MEG and 18 fMRI records from two studies

We tested whether visual repetition changes a category response beyond one gain and keeps over half its initial pattern projection. Public data supplied EEG and MEG from one cohort and grating fMRI from a separate study. Four electrical-source people supplied development. A local lock preceded reserved scores. Fixed artifact and run-count rules retained three of 15 EEG candidates and all 15 MEG candidates. All 18 MRI arrays were included. The compact MRI file lacked run links. We kept each person’s lowest score over all 35 balanced subrun partitions. The mean half-retention differences were +0.001754 in EEG, +0.005376 in MEG and -0.041050 for the MRI lower bound. Pointwise 95% person Student limits were [-0.054077,+0.057585], [-0.007606,+0.018358] and [-0.089787,+0.007687]. All three shared claims failed Holm correction, p=1. An assumed chronological MRI link gave +0.068885 [+0.036798,+0.100971]. This conditional positive result does not replace the failed bound. The study supplies a negative benchmark and a tool for conservative inference with missing run links. It does not confirm absent neural coding or a positive cross-modal biological effect. Stable run means, independent errors, correct source labels and a suitable person mean model remain assumptions. Small EEG inclusion, shared electrical-source artifacts, distinct tasks and cohorts, inherited processing and vascular filtering limit scope. Code, source pins and terms, all outcomes, figures, audits and repeat steps are supplied. Original neural arrays remain external.

visual repetition · EEG · MEG · fMRI · projection retention · missing run links · response geometry · negative benchmark