Replacing zero delay records with planned delays reduces stop time estimates in ds000030 snapshot 00016

We checked all 533 stop-signal runs from 267 adults in OpenNeuro ds000030 snapshot 00016. The files contain 17,056 stop trials. A total of 941 trials have a recorded delay of zero and a positive planned delay. Each of these trials contains a response before the planned delay. None is a successful stop. All planned delays match their staircase histories. We compared the same integration estimate under three delay choices. Replacement of zero records with planned delays reduces the mean estimate by 16.5 ms in training and 32.7 ms in the scanner task. The 95% participant bootstrap limits are 14.1 to 19.1 ms and 27.9 to 37.6 ms. The effect remains after performance checks. The change equals the sum of planned delays on zero-record trials divided by all stop trials. The replacement assumes that the planned delay applies before the outcome changes the staircase. It does not recover physical sound onset or prove model validity. The package preserves raw events, provenance, task source, code, results, tests, and reproduction instructions.

stop-signal task · stop-signal reaction time · SSRT · reaction time · delay staircase · data provenance · measurement sensitivity