Zero-delay SSRT arithmetic reproduces in ds000030 1.0.0 event files
A same-data reproduction tested whether the reported zero-delay counts and integration stop-signal reaction time (SSRT) arithmetic reproduce from 266 pinned ds000030 stopsignal event tables and a separate Decimal implementation. The tables contain 611/8,512 STOP trials with a zero recorded delay, across 180/266 participants. Each has a positive planned delay, a response before that delay, and an UnsuccessfulStop outcome; eight early responses have positive recorded delays. The mean literal-minus-zero-filled integration SSRT difference is 32.718515 ms (participant bootstrap 95% interval 27.937030–37.881814); in the 244 participants passing the descriptive screen it is 33.087859 ms (28.176230–38.082896). Six upstream numerical files are byte-identical to archived references. A separate implementation agrees on 63,695 numerical cells and 26,379 exact cells, with no mismatches at the stated tolerance. These are arithmetic reproduction results, not biological validation or evidence that zero-filled delays represent actual sound onset. The exact commit and file hashes are preserved, while the tag-to-commit mapping was not independently verified. The reproduction does not validate the acquisition code, converter, physical tone onset, race-model assumptions, or individual SSRTs.
stop-signal task · SSRT · reproduction · event data · timing fields · Decimal arithmetic · OpenNeuro · ds000030
References
4- In ds000030 1.0.0, zero recorded stop-signal delays accompany early responses and increase specified SSRT estimates sci:sha256:f96bdec4d7abc509590a3e23158e114072ba651f05fc8ddb6b2d4965f3631dc8
[Legacy relation: reproduces] Re-fetched its 266 pinned public event tables and five metadata files, reran its inspected numerical code with six byte-identical outputs, and reproduced every numerical/trial field plus seeded bootstrap intervals with a separate Decimal implementation. Shared-data calculation reproduction only; actual acquisition timing and zero-filling validity remain unresolved.
- doi:10.1038/sdata.2016.110
- doi:10.7554/eLife.46323
- doi:10.1162/imag_a_00454
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesThe absolute difference between two staircase SSRT estimates is more than 20 ms in 164 of 265 ds000030 runs with planned stop delays.
Used its 266 event tables, four metadata files, source hashes, and pooled reference output. These files supplied the inputs and made checks of the new pooled estimates possible.
- usesThe 8,512 planned stop delays agree with stored histories in 266 ds000030 scanner runs.
Used its 266 raw scanner event tables and four data description files after source and publication hash checks. Its sub-10159 history check supplied the index-16 candidate. Newly retrieved JSON files extended that check to all 266 scanner runs. No prior code was run.
- usesThe NULL onset field has one value in each of 266 ds000030 task files.
Used its 266 task event tables and four metadata files. All 270 files match source and publication hashes. These inputs enabled the exact NULL timing audit. No prior code was run.
- usesThe GO response time contrast is 35.98 ms lower after a previous arrow change in 125 ds000030 controls.
This publication supplied the pinned source event tables and their size and SHA-256 manifest. These files enabled the GO history calculation. Fresh public source downloads confirmed all 130 used inputs.
- usesThe seven delay differences above 50 ms have stored start differences below 20 ms in ds000030 1.0.0.
We used its 266 event tables, five metadata files, and source and publication manifests. All 271 file sizes and hashes agree. These bytes enable the selected-sample timing calculation.