Single-pulse fits gave uncertain gap-shape forecast gains in six fMRI participants after positive gains in eight ECoG patients

We tested whether one adaptation state could predict responses to repeated images from responses to single images. We used public human ECoG and fMRI data from separate studies. A fixed rule fit six single-pulse conditions and predicted six dark-gap conditions. The score measured the reduction in centered gap-shape error relative to a constant curve, divided by the pooled single-pulse mean square. ECoG gave positive gains in eight of eight patients. Mean gain was 0.01094, with a 95% person-bootstrap interval of 0.00445 to 0.01920 and exploratory one-sided p=0.00391. The independent-study fMRI test had eight records from six people. We tested all 72 possible links for repeated people. Mean gains ranged from 0.00421 to 0.01035. The interval union was -0.00377 to 0.02581, and the least favorable p was 0.26563. The cross-modal positive rule failed. Source scan-bootstrap uncertainty widened the fMRI interval. Early-visual restriction and baseline subtraction also made ECoG mean-gain evidence uncertain. This result does not show equal signals, a shared biological time constant, no adaptation, or causality. We add a condition-separated forecast benchmark, a closed-form model check, all source curves, and code. Prior work already showed temporal compression and recovery. Small reused cohorts, source selection, unknown fMRI identities, spatial sampling, finite ECoG epochs and the BOLD readout limit interpretation. The package preserves the plan, inputs, source terms, negative checks, figures, validation and repeat instructions.

human neuroscience · ECoG · fMRI · visual adaptation · condition prediction · independent cohort · negative result