Pooled and delay-cell rejection rates varied across four simulated stop-signal models
We simulated four stop-signal models using 125 public ds000030 task schedules and synthetic response times. Each model had 250,000 runs. Pooled rejection rates ranged from 0.0104% to 0.1004%; delay-cell flag rates ranged from 0.0152% to 0.0812%. In the stationary 300 ms STOP model, the pooled rate was 0.1004% (251 rejections) and the delay-cell flag rate was 0.0812% (203 flagged runs). The continuous equal-law rank reference had an exact probability under independent continuous equal-law samples, but its use with adaptive observations requires a separate check. These conditional model results do not establish human-task error limits. Intervals describe approximate Monte Carlo uncertainty and omit fit, model, and human sampling uncertainty. The study used the R1.0.5 source files identified by pinned URLs and hashes; the cached dataset DOI applies to a different release and is not assigned here.
stop-signal task · rank test · adaptive delay · Monte Carlo simulation · distribution order · false rejection · response time
References
6- usesExact permutation probabilities are more than 0.31 for all 102 distribution crossings in 122 selected ds000030 controls. sci:sha256:8ae3a83a153b600b06b6361503afe80db95a11d0e60b811efa18d60d9c34adfa
Its exact one-sided pooled CDF rank reference supplied the statistic definition and continuous equal-law label assumptions. New integer-count code tests that reference with adaptive simulated responses.
- usesSSRT rank agreement is lower with recorded delays in two splits of 102 ds000030 controls sci:sha256:24a9839d28305476eb83ac828e48e86be3f108a54649b4d9aeec053b1c975882
Its pinned R1.0.5 source manifest supplied the 126 CONTROL event paths, source hashes, and source selection. All source bytes were downloaded again. Its analysis code and split-rank results were not reused.
- usesThe rank checks do not give a task error limit for each planned delay in 121 selected ds000030 control runs. sci:sha256:d69995dd6b2cf9a9504219239f01cec02de26b0059e4482f6d6124cfe3f3ae8e
Its unresolved delay-cell error-limit question specified adaptive-task simulation as a next check. This method supplies four conditional model benchmarks with the 16 declared delays and preserves the remaining serial-dependence question.
- usesdoi:10.1038/sdata.2016.110
Provides the ds000030 stop-signal task design and adaptive-ladder mechanics used to interpret the public schedules.
- usesdoi:10.7554/eLife.46323
Describes the mean-response check used in the repeated behavior screen; it does not provide a delay-cell rank-test error bound.
- usesdoi:10.1007/s42113-023-00184-3
Provides the distribution-order assumptions discussed for the GO and failed STOP distributions. A violation does not identify a unique biological cause.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.