Separate code reproduces scanner results for 266 ds000030 participants

A separately written program reproduced the reported scanner calculations for 266 participants in OpenNeuro ds000030 snapshot 00016, using the same observations. It matched 6,379 participant numeric cells, 85,120 STOP-trial numeric cells, 118 summary values, 13 structural-check values, and all eight bootstrap endpoints; the maximum participant difference was 3.41 × 10⁻¹³ in the corresponding field unit. The program compared recorded stop-signal delays with a rule that replaces recorded zeros by positive nominal delays. The mean delay increase was 32.7185 ms across all participants (95% participant-bootstrap interval 27.9721–37.7585) and 33.2988 ms in the 241-participant exploratory screen (28.3516–38.4406). The two rules had Spearman rank correlations of 0.7807 for 265 estimable participants and 0.7585 for the 241 screened participants. This is a software replication using the same observations, not independent biological replication. It does not establish actual cue delivery, identify a correct delay rule, or test biological causality. The analysis used the public snapshot at the recorded Git commit; the saved version-specific dataset DOI refers to a different release and is not assigned here.

computational replication · stop-signal task · scanner event data · stop-signal reaction time · bootstrap · OpenNeuro · ds000030 · software verification