Adding pooled insula high-gamma after movement increased mean error prediction loss in fixed block tests for 14 SEEG patients

This study tested whether insula activity after movement added useful error prediction. It used public MetaRDK SEEG data from 14 patients and 2,971 trials. The baseline included reported confidence, behavior, and insula activity before movement. Tests used five consecutive trial blocks. Adding the pooled signal increased mean log loss from 0.58938 to 0.59657 nats per trial. The mean gain was -0.00719. Its 95 percent patient bootstrap interval was [-0.01471,-0.00035]. The exact sign-flip p value was 0.0852. Twelve patients had higher loss. A stronger model penalty gave a gain close to zero. Both primary models had higher mean loss than a constant predictor. The result applies to the tested models and reused cohort. It does not show that the insula lacks error information. The package preserves code, compact inputs, source versions, trial predictions, model records, figures, and reproduction instructions.

insula · error prediction · SEEG · high-gamma activity · log loss · MetaRDK · patient-level analysis