Adding mean insula SEEG activity to a fixed behavior model increased mean confidence prediction error across blocks in 12 patients
We tested whether mean insula high-gamma activity improved confidence prediction beyond behavior. We used public SEEG data from ds006253 version 1.0.3. We tested separate trial blocks in 12 patients, with 2,462 trials and 62 insula channels labeled as gray matter. The fixed least-squares model increased mean relative squared error by 1.81%. Three patients improved. The 95 percent patient bootstrap interval for error gain was -3.62% to -0.16%. The exact two-sided sign-flip p value was 0.0752. A fixed all-channel model had higher mean error. An exploratory proof shows that uniform positive shrinkage cannot give mean gain on these saved test predictions. The result concerns this model and sample. It does not show that the insula lacks confidence information. The package preserves data, source checks, code, predictions, a figure, and reproduction steps.
confidence prediction · SEEG · insula · high-gamma activity · clinical neuroscience · cross-validation
References
3- usesA fixed SEEG high-gamma model increased mean confidence prediction error across trial blocks in 17 patients sci:sha256:a82d0f18d33d80eff73f5cfc9ea22ee95b2d84706a1fa4fa872c7af43045c180
We used its compact SEEG window means, behavior and anatomy tables, pinned source maps, and extraction code. We tested a new fixed insula mean model and checked public sources again.
- usesdoi:10.18112/openneuro.ds006253.v1.0.3
Source dataset MetaRDK, OpenNeuro ds006253 version 1.0.3. This version matches the pinned release and public source objects checked in the study.
- doi:10.1038/s41467-025-61744-8
Background paper motivating the insula region and post-decision window and documenting source-pipeline context; the present study tests predictive value across trial blocks.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- A fixed insular SEEG model lost its mean confidence prediction advantage when the penalty for all channels increased from 1 to 10 in 12 patients
[Legacy relation: reproduces] Our independently coded primary model reproduces all 2,462 primary predictions and all 12 patient MSE and gain scores exactly. We found this report after our scores and source audit. The recordings and upstream features are shared, so this is a computation check with no new patient evidence. Bootstrap seeds differ.
- usesA nested model of insula SEEG channel differences increased mean confidence prediction error by 3.52% across blocks in 10 patients
This study uses direct bipolar insula anatomy matches and helper code. It adapts the source audit to the 10 patients with at least two selected channels.