At 20 calibration sounds, matched class examples lowered association forecast error in 14 iEEG patients and 16 fMRI adults

We tested whether class examples improve auditory response forecasts after a fixed acoustic match. Separate public studies supplied signed iEEG field means and BOLD fMRI. Four people per study supplied development. A local code lock preceded new scores in 14 other patients and 16 other adults. Each forecast used 16 common calibration sounds and four examples. The control used four non-class examples matched on ten acoustic axes. Mean association gains were 6.640 and 17.346 reference-error points. Their 95% person Student limits were [3.075,10.205] and [16.665,18.028]. All reserved people had positive gains. A shared gain and skill test passed two-claim Holm correction, p=0.00144670. Eight adults from another MRI study gave a 16.156-point mean gain on the same MRI sounds. The original acoustic-fit hierarchy test failed, p=0.43139648. Development prompted the added mean-forecast test. Pre-event subtraction retained a positive electrical mean gain. The early electrical window was uncertain. Exact acoustic-only examples gave either gain sign. They were not fit to human responses. Matching was limited and did not give equal acoustic geometry. Separate cohorts, two sound libraries, source screens, sparse clinical coverage and different signal filters limit scope. No semantic code or common neural cause was established. Code, operators, permitted inputs, all losses, source checks, negative results and repeat steps are included.

human neuroscience · intracranial EEG · fMRI · auditory cortex · calibration · matched examples · prediction error · cross-modal generalization