Replacing zero stop delays with scheduled delays lowers mean stopping time estimates in ds000030 version 00016

We tested how the choice of a delay field changes an estimate of stopping time. The source was OpenNeuro ds000030, version 00016. We retrieved all 533 stop-signal event files in this snapshot. We excluded the discovery participant from the main analysis. All 932 suspect zero delays in the remaining files had a response before a positive scheduled delay. We then replaced these zero values with the scheduled values. The mean estimate decreased by 33.03 ms in 240 MRI runs and 16.77 ms in 253 training runs that passed the stated checks. The 95% bootstrap intervals were 28.06 to 38.11 ms and 14.25 to 19.45 ms. The package includes the source files, a delay audit, the analysis code, and a tool that requires an explicit delay policy. These results show sensitivity to the delay field. They do not establish when the sound played or the true time needed to stop a response.

stop-signal task · stop-signal reaction time · delay fields · event data · OpenNeuro · measurement audit