Delay-field choice reorders calculated SSRTs in ds000030 control pairs
In a computational sensitivity analysis of OpenNeuro ds000030 version 1.0.0 CONTROL event files, we compared stop-signal reaction time (SSRT) estimates based on scheduled LadderTime, literal recorded StopSignalDelay (including zeros), and recorded delays with only zeros replaced by scheduled values. The analysis reimplemented a prior 253-run audit and extended it to participant ordering and paired training–scanner correlation contrasts. Among 107 screened participant pairs, literal versus scheduled SSRT ranks had Spearman correlations of 0.8854 (training) and 0.7264 (scanner), with 728/5,671 (12.84%) and 1,172/5,671 (20.67%) strict pair reversals. Training–scanner Spearman agreement was 0.6065 scheduled and 0.5146 literal; the paired difference was −0.0918 (95% participant-bootstrap interval −0.2339 to 0.0409), which includes zero. Zero-filled values nearly preserved scheduled ordering, but this is not a validated timing correction. The results describe stored observations and calculated estimates; they do not establish delivered audio timing, true inhibitory ability, causation, or population-level change. The selected cohort has 32 STOP trials per run, fixed training-before-scanner order, possible trigger failures, and uncertain delay-field semantics.
stop-signal task · stop-signal reaction time · SSRT · participant ranking · measurement sensitivity · ds000030 · training–scanner agreement · bootstrap correlation
References
6- Recorded zero delays changed stop-signal estimates in ds000030 control runs sci:sha256:5ede3d0ea1dc1723230475afb6813476effef6e389d019e311526bfb92def4ed
[Legacy relation: reproduces] Independently reimplemented its 253 CONTROL-run integration estimates and paired screens using the pinned upstream source and its manifest; reproduced 4,807 numerical run cells, 759 flag/pass cells, and 45 paired point summaries. Preserved its tables as comparison evidence and extended the audit to participant ordering and directly paired bootstrap correlation contrasts. Prior analysis code was inspected but not executed or imported.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The analyzed event files are from the version-pinned OpenNeuro ds000030 1.0.0 source; source hashes and the upstream commit are preserved.
- doi:10.1038/sdata.2016.110
- doi:10.7554/eLife.46323
- doi:10.3389/fpsyg.2012.00037
- doi:10.1162/imag_a_00454
Cited by
3 publicationsPublic catalog entries on Sciscape.
- Two calculations agree on score correlations in 118 ds000030 CONTROL pairs.
[Legacy relation: reproduces] The separately written calculation uses the same 118-pair sensitivity sample and 126 complete pairs. Exact identifiers and 68 summary cells agree within 1e-9. This includes selected 118-pair intervals and point values for both samples. The 107-pair primary sample and 126-pair intervals are outside the reproduction scope.
- usesThe starting delay term is 0.5301 in the scheduled SSRT correlation for 107 ds000030 CONTROL pairs.
Its inspected source manifest, exact 107/118/126-pair screens, estimator convention, and frozen run table define this reanalysis. Fresh pinned source files match its hashes. New code matches 4,807 numeric run cells and 759 flag cells, then calculates scanner starting-delay covariance terms. Its rank-reversal analysis is outside this check.
- usesFive examined ds000030 task programs share nominal tone values, but the deployed stop tone remains unknown
The delay-field evidence gap motivated this task-stimulus question. The prior metadata provenance enabled the current check of the same frozen dataset version.