Adding eight SEEG pairs changed mean later probability error in opposite directions under four fixed penalties in six recordings of one film

Mean Brier error was 0.2469 for EOG and ECG alone and 0.2509 after adding eight SEEG contact pairs. The observed combined-minus-control change was +0.0040. The descriptive 95% person bootstrap interval was -0.0224 to +0.0261. The added SEEG inputs gave lower error in 2 of six people. The four fixed penalty checks gave combined-minus-control mean error changes of -0.0283 at C=0.01, -0.0126 at C=0.1, +0.0040 at C=1, +0.0083 at C=10. A smaller C gives a stronger penalty. These checks were fixed in the plan and remain secondary. Training used the early half of the retained film blocks; testing used the later half. The models used four fixed power bands and training-only selection and scaling. The package adds EOG/ECG voltage and features to a public six-person SEEG control set. Complete source checks reproduced every inherited SEEG feature. It includes code, source hashes, all fitted models and predictions, figures, and an exact analysis repeat. The result applies to these readers and one film. It does not establish a neural cause, speech intent, clinical benefit, or stimulation effect.

stereoelectroencephalography · SEEG · electrooculography · EOG · electrocardiography · ECG · brain-computer interfaces · Brier error · later-time prediction · artifact control