Nine mean reaction time checks change after removal of STOP rows with recorded zero delay in ds000030 version 1.0.0 controls.
We tested how removing whole STOP rows with a recorded zero StopSignalDelay changes the mean reaction time (RT) check in the control group of OpenNeuro ds000030 version 1.0.0. The source files contained 126 scanner runs and 127 training runs, with 96 GO trials and 32 STOP trials per run. Initial checks selected 122 scanner runs and 121 training runs. After removal of 430 zero-delay STOP rows, five selected scanner runs and four selected training runs changed from a passing to a failing mean RT check; these runs belonged to seven people. All removed rows had positive RT below LadderTime/1000. One other run in each setting fell outside the specified STOP response proportion. The mean hybrid integration score minus the score after row removal was 4.212340641623095 ms for scanner runs and 1.2682199918940673 ms for training runs; both medians were zero. The findings describe data selection in this source version and these controls. They do not establish biological stopping latency, test the independent race model's assumptions, or generalize to other samples. The two settings are not independent samples, each run had only 32 STOP trials before removal, and we did not measure physical stop-signal timing or estimate uncertainty for other samples.
stop-signal task · reaction time · data selection · integration method · OpenNeuro · reproducibility · cognitive neuroscience
References
4- In ds000030 controls, zero stop-delay fields coincide with early responses and shift fixed estimates sci:sha256:05393c9ed2c28a637c8cda2cb999da997531b83603fa63fc81cc6b25b56cec28
[Legacy relation: reproduces] We used the pinned source manifest and prior run table. We reproduced GO and STOP means, integration ranks, recorded and hybrid scores, zero-only shifts, zero counts, and 5%, 10%, and 20% check sets for all 253 runs. This relation does not cover the prior bootstrap or every reported field. Whole-row removal is the new analysis.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
This is the exact dataset version analyzed. The source manifest records commit 4070b6eea231517bfeff42527a46d9d166ac4e13 and hashes for the source files.
- supportsdoi:10.1038/sdata.2016.110
The dataset descriptor supplies the task design and public access information used to describe the source.
- doi:10.7554/eLife.46323
Background consensus guide on stop-signal data analysis, cited for its recommendation to retain responses made before the stop signal in mean RT and integration calculations.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- 90 of 243 selected ds000030 version 1.0.0 control mean reaction time checks stay satisfactory after STOP response removal with response fractions from 0.25 to 0.75.
[Legacy relation: reproduces] The target supplied the pinned source files, manifest, and run table. New code agrees on 17 fields for all 253 runs, with 4,301 comparisons. It gives the same selected sets and all nine recorded-zero removal changes. This relation covers the stated means, counts, fractions, and flags. It does not cover stopping-time scores or all target fields.
- Nine mean RT checks change after recorded-zero row removal in selected ds000030 version 1.0.0 runs. Each reference mean is less than one.
[Legacy relation: reproduces] The target supplied the pinned source files, manifest, and comparison outputs. New code reproduces 253 run counts, means, initial checks, and the nine reversal identities. This relation does not cover stopping-time scores or every target field. The equal-size removal reference is the added calculation.