A fixed visual compression index did not pass recovery prediction tests in five ECoG patients and six fMRI adults

Can a single-image compression index predict response recovery beyond a visual-area label? We tested this proposed measurement shortcut in two separate human studies. Three ECoG patients supplied development. The fixed test used five other patients and a separate fMRI sample of six adults. The predictor, recovery target and scale used disjoint stimulus conditions. Adding the index raised observed mean squared error. Mean gains were -0.08671 in ECoG and -0.08931 to -0.04261 in fMRI. The 95% person t interval was -0.23684 to +0.06342 in ECoG. The MRI interval union was -0.23194 to +0.08058. MRI ranges cover all 72 possible links between repeated records. The shared positive rule failed, p=0.984375. Equal-area weights reduced electrical error effects. MRI blank and image-type checks changed mean signs and remained uncertain. A common-denominator calculation shows that a positive normalized relation can arise from independent contrasts. It is an interpretation control, not an explanation of the human result. The work supplies a new fixed endpoint benchmark, permitted inputs, all predictions, source hashes, code, figures and repeat steps. These are endpoint tests in reused cohorts, not new acquisition cohorts. Clinical sampling, separate cohorts, unequal spatial support, source processing and measurement noise limit scope. No positive neural generalization across modalities or causal effect is established.

visual cortex · temporal compression · recovery · ECoG · fMRI · prediction · negative result · measurement error