A simulated 4 ms delay gave nonzero time reversal contrasts at 20 ms forecasts in independent contact models from two SEEG rest records

We tested whether a relative channel delay changes a time reversal contrast between shared bipolar SEEG inputs. A simulated 4 ms delay gave nonzero contrasts in all 64 shared pair-period tests in independent contact models. The source was OpenNeuro ds005670 version 1.0.0, with two people and two 120 second periods per person. The primary test used 100 ms acquired histories and 20 ms forecasts of 4 ms voltage means. Mean absolute contrasts ranged from 0.107107 to 0.130553. All zero-delay contrasts were zero. Equal normalized contact time scales did not remove the delay effect. Separate-contact pairs remained at zero. Known buffering restored the independent model's zero contrast. An exact white-noise example and 1,920 separate projection checks support the calculation. New complete source downloads matched both pinned EDF hashes and all 23,040,000 selected native values. A full numeric replay reproduced nine files with exact byte equality. The package preserves selected native inputs, code, source metadata, results, a proof, three figures, and repeat instructions. These are fixed stationary covariance models. The delay was simulated. The result does not identify a brain pathway or an actual source timing fault. No live decoder or stimulation response was tested. The contribution is a public SEEG timing benchmark for directional brain-interface analyses.

time reversal · relative delay · bipolar signals · SEEG · signal processing · brain-computer interfaces · finite-history prediction