A second calculation gives the reported delay results for 126 ds000030 control runs.

This computational reproduction uses the same public observations as an earlier AutoSci report. A separate calculation analyzed 126 of 130 ds000030 controls with a stop-signal task flag. Each selected run had 96 GO, 32 STOP, and 128 NULL trials. It compared estimates based on recorded StopSignalDelay values with hybrid estimates that replaced zero recorded delays with planned LadderTime values. The recorded-minus-hybrid mean contrast was 30.059524 ms (95% participant-bootstrap interval 23.561198 to 37.016679) across all 126 controls and 30.456149 ms (24.067540 to 37.676411) in the target's 124-person basic screen. The independent calculation agreed with all 5,171 reference comparisons within 1e-8 ms for time values and 1e-8 for probabilities; counts, identities, and flags agreed exactly. Among 4,032 STOP trials, 274 had zero recorded delay and positive planned delay, in 84 participants. The findings establish agreement between calculations on the same files. They do not establish actual sound onset or a corrected stop-signal reaction time. Task and conversion versions remain unresolved; GO errors, omissions, 32 STOP trials per participant, and model assumptions limit biological interpretation. No biological cause is established.

stop-signal task · reaction time · computational reproduction · OpenNeuro · ds000030 · bootstrap · timing fields