The mean delay difference is 30.520 ms in 266 ds000030 scanner runs.

We reproduced the planned-minus-logged stop-signal delay arithmetic in 266 public scanner runs from the UCLA Consortium for Neuropsychiatric Phenomics dataset, OpenNeuro ds000030. Each run contains 32 STOP trials. Using exact fractions of the decimal input values, the mean difference was +30.519938551924785 ms (median +17.530850937565415 ms; participant bootstrap 95% percentile interval [25.70101816098004, 35.62475534016777] ms). All participant values matched the attributed reference within 1e-9 ms; all 29 aggregate and outlier checks passed. The 611 STOP rows with zero logged delay and positive planned delay occurred in 180 runs, and all had a response before the planned delay. All 7,980 staircase transitions matched both recorded movement and response-based 50 ms update rules. Seven large positive timing residuals remain unchanged; sensitivity calculations are reported. The analyzed scanner event tables came from tag 00016. Their hashes match tag 1.0.0, but metadata differ, so this work does not claim full dataset equivalence or apply the later tag's version DOI. The logged fields' units and meanings are assumptions. This is a computational reproduction on the same source tables and participants, not an independent sample. It calculates a delay component only; it does not estimate absolute stop-signal reaction time or verify physical sound onset. The 32 STOP trials per run are below consensus guidance of at least 50 for standard group comparisons under suitable conditions.

stop-signal task · delay arithmetic · scanner event tables · computational reproduction · OpenNeuro ds000030 · staircase · reaction time · timing fields · bootstrap · data provenance