Removal of one STOP response can change 16 of 17 failed arrow group mean time checks in 124 ds000030 control runs.
In 124 OpenNeuro ds000030 control runs that passed a pooled basic screen, 17 arrow-group mean-time checks failed. Exhaustive removal of each GO or STOP record, with all other observations and cohort membership fixed, found that at least one STOP-response removal changed 16 of those 17 groups to a passing check; the remaining group stayed negative after every single-record removal. Nineteen initially passing groups could fail after one removal. These deterministic sensitivity results use the same observations as a prior report, not a new sample. The removal ranges are not confidence intervals or population probabilities, and deleting a table row does not represent behavior in a task without that trial. The results do not establish a model violation, biological cause, or reason to discard a response.
stop-signal task · response inhibition · mean reaction time · sensitivity analysis · OpenNeuro · reproducibility
References
5- usesRemoval of one response changes the result of the check of mean reaction time in 13 of 266 ds000030 files. sci:sha256:a16d31fbe42052ea69182ec444a4b05382719c5a76199a256ef6589fb204ebb6
Its pooled mean-check sensitivity study supplied the preceding one-record removal method and the limit that later recorded behavior stays fixed. This study applies that method to each arrow group in a fixed cohort. It does not use its omission screen or execute its code.
- An arrow group fails the mean response time check in 17 of 124 ds000030 control runs that pass the pooled screen. sci:sha256:ef5eef0e3017f5be1245a0d453b63113c1463445ff9e43f0cc33253c52f17c2b
[Legacy relation: reproduces] New code reproduces all 378 baseline rows, with 5,292 numeric, count, and flag comparisons. Its manifest and 130 source files enable the same 124-run cohort. This link does not claim reproduction of its random partitions or bootstrap intervals.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
Source observations are from the version 1.0.0 dataset; the retained source manifest records file URLs and SHA-256 checksums.
- doi:10.7554/eLife.46323
- doi:10.1038/sdata.2016.110
Cited by
1 publicationPublic catalog entries on Sciscape.
- In 124 ds000030 control runs, one removal can make 14 of 17 failed arrow groups pass the basic screen.
[Legacy relation: reproduces] New code agrees with every compared field in 378 baseline rows and 16,128 removal rows. Counts and flags agree exactly. Numerical tolerance is 1e-9. The target mean result remains 16 of 17 groups in the fixed 124-run sample. The new probability calculation gives 14 of 17 passing groups.