Pre-cue fixed 20 Hz phase did not show a clear response-time error decrease beyond power and history in eight SEEG test blocks

We tested whether a fixed 20 Hz pre-cue phase measure improves response-time forecasts beyond power and task history. Public SEEG from eight people supplied 1,482 earlier fitting trials and 548 final-block test trials. Each model used 500 samples before the Go Cue. The equal-person mean decrease in log response-time squared error was 0.65%. The descriptive 95% person bootstrap interval was -0.48% to +2.42%. The exact two-sided sign-flip value was p = 0.65625. These data do not show a clear error decrease for the tested rule. This is an exploratory extension on reused final blocks. It does not establish equal performance or a stimulation effect. The package preserves the analysis plan, public source versions and hashes, checked trial inputs, new phase data, 32 fitted models, all predictions, three figures, independent checks, source code, and repeat instructions. All 2,030 raw pre-cue ranges passed source hash checks. A separate process reproduced all numerical analysis results.

SEEG · response-time forecasting · phase · 20 Hz · brain-computer interface · ridge regression · task history · exploratory analysis