Recorded zero delays changed stop-signal estimates in ds000030 control runs

In OpenNeuro ds000030 version 1.0.0, 430 of 8,096 stop trials in control participants recorded a zero stop-signal delay despite a positive scheduled delay. Every such trial was a failed stop with a response before the scheduled delay. In the integration-method estimate, using recorded delays literally instead of scheduled delays changed the scanner-minus-training SSRT contrast by 14.47 ms in 107 screened participant pairs (paired participant bootstrap 95% interval 7.49–21.84 ms). Scheduled-delay estimates had a mean context difference of 3.97 ms (−4.81 to 12.67 ms); literal recorded delays gave 18.43 ms (10.05 to 26.96 ms). Filling only recorded zeros from scheduled values gave 3.67 ms (−5.19 to 12.39 ms), but this is a sensitivity analysis, not a validated timing correction. The audit includes 253 available runs from 130 control participants, enumerates timing discrepancies, and preserves all run-level estimates and quality flags. The observed association does not establish why zeros occur or represent physical tone onset. Fixed task order, only 32 stop trials per run, possible trigger failures, and other design and screening limits prevent causal or longitudinal reliability conclusions. No neural or clinical analysis was performed.

stop-signal task · stop-signal reaction time · SSRT · reaction time · OpenNeuro · ds000030 · event-file audit · timing sensitivity · behavioral neuroscience