One added 400 ms pulse gives a test that cancels baseline error in human ECoG and fMRI designs under a stationary pair model

We studied a visual timing design shared by separate human ECoG and fMRI studies. An exact calculation shows that all twelve active conditions can fit an arbitrary stationary signed pair-interaction model if the response baseline is free. The design has rank twelve. Adding one 400 ms single-image condition leaves rank twelve among thirteen means. Its unique test contrast cancels a common baseline. Thus one added condition is sufficient and necessary for this active-only scalar test under the stated model. A continuous signed kernel can realize the finite pair coefficients. Interpolation matched all active means at 94 ECoG contacts and 48 fMRI region records. These exact fits are a design limit, not model validation. An initial fixed negative-direction test of an earlier identity failed in five reserved ECoG patients and six fMRI adults under all 72 possible repeated-person links. Shared p was 1.0. All controls and two failed exploratory monotonicity tests remain. The new assay can miss higher-order terms that cancel. Finite tails and nonlinear measurement can also break its neural meaning. Noise factors are conditional and do not show a power gain for a common effect. The added pulse was not measured. No new positive shared neural finding is claimed. Code, permitted inputs, source pins and terms, plans, all results, figures, an assay tool, and repeat steps are supplied.

visual cortex · temporal summation · ECoG · fMRI · identifiability · baseline · experimental design · quadratic model