Acoustic maps fitted to other sound classes reduced mean error for fixed music clips in human iEEG and fMRI

We tested whether an acoustic map fitted to other sound classes predicts differences among Music clips. Separate public studies supplied signed mean iEEG local field potential and BOLD fMRI. Five patients and four adults supplied development. Fixed tests used 20 other patients and 16 other adults, with 18 and 21 music clips. Each person had a separate map. Source labels defined Music. No response from that class set a new fitted coefficient, scale or penalty. We removed sound class means and spatial response means. Mean within-music error reductions over a constant music-mean predictor were 1.176% and 10.589%. Their 95% person t intervals were [0.582%,1.770%] and [8.456%,12.722%]. Scores were positive in 17/20 and 16/16 people. The fixed shared mean-and-majority rule passed, p=0.001288414. Joint person-and-sound ranges were [0.283%,2.085%] and [4.936%,16.994%], conditional on fitted maps. All negative people and planned controls remain. A direct source rebuild matched all 110,682 electrical means. Independent regression and score checks passed. A one-scalar source can also give positive transfer. The finding concerns partial prediction for fixed source panels. Separate cohorts, different acoustic features, clinical sampling, signed time averaging, BOLD pooling, upstream site selection and descriptor row assumptions limit scope. No common neural generator or cause is established. Code, permitted inputs, fitted maps, predictions, provenance, local plans, results, figures and repeat instructions are preserved.

human neuroscience · intracranial EEG · fMRI · music · acoustic encoding · category transfer · prediction · independent cohorts