Nominal scanner delay starts matched training values after the last update in 265 of 266 people in ds000030 version 1.0.0

This study tested the nominal delay settings in all 266 training and scanner pairs in OpenNeuro ds000030 version 1.0.0. Both scanner ladders matched the training values after the last update in 265 people. There were 530 matches among 532 ladders. One person had two exceptions. All 267 training files followed the final mean formula 1100 - 50R + 25B milliseconds, where R counts failed stops and B counts failures at zero delay. A training count thus enters both the training integration rank and the scanner start. In 232 pairs that passed the data checks, the nominal stopping-time correlation was 0.550. Its exact covariance sum contained a delay-start term of 0.582. These terms do not identify a causal effect. The dataset paper already describes transfer of the settings. This work adds the full file audit, the count proof, and tools to inspect the estimate arithmetic. The package preserves source events, hashes, code, checks, bootstrap draws, and reproduction steps. A fresh environment reproduced all tables exactly and JSON numbers within 1e-12. Actual sound timing and changed human responses were not tested.

stop-signal task · stopping-time estimate · nominal delay · scanner delay · training-to-scanner transfer · SSRT · OpenNeuro · ds000030 · covariance decomposition · reproducibility