The 100 ms GO rule allows both contrast signs in two tested ds000030 control samples.

A trial-matching analysis of OpenNeuro ds000030 control data found that the STOP-minus-GO contrast can be positive or negative under the tested 100 ms GO rule in both the expanded sample and a fixed-coverage follow-up. Under the original 50 ms rule, every permitted assignment in its tested sample remained negative. The result limits the scope of a general sign claim; it does not reject the original 50 ms finding. The main sample included 36 participants and 200 failed STOP trials. The nearest-cost contrast was -76.9430 ms, while permitted assignments ranged from -153.4746 to +22.0429 ms. A minimum-cost assignment reaching +0.01494 ms increased total matching cost by 14.3090%. In a follow-up that held the original 29 participants and 132 STOP trials fixed, the 100 ms rule ranged from -194.6480 to +28.7365 ms. This exploratory follow-up shows that additional observations were not required for a positive endpoint. Positive assignments selected GO donors using reaction time after the center; they are sensitivity witnesses, not unbiased effect estimates, null tests, or evidence of a causal effect. The nearest-cost estimate remained negative. The analysis does not establish biological causes, and its numerical optimization is not a formal proof of least cost. Source-file hashes, exact object versions, assumptions, methods, code, results, and limitations are preserved in the package.

stop-signal task · reaction time · GO choice · trial matching · assignment bounds · post-error slowing · OpenNeuro ds000030