Post-update training values match 530 of 532 scanner staircase starts in ds000030
This audit compared the first scanner LadderTime with two candidate paired training endpoints in public ds000030 stop-signal event tables. Of 532 ladder pairs from 266 scanner participants, 530 matched the training value after its final 50 ms update, covering both ladders for 265 participants. Two pairs from sub-10855 were discrepant: both training updates predict 50 ms, while scanner starts were 200 and 250 ms. Only 2/532 pairs matched the last presented training value. Each of 533 runs had 96 GO, 32 STOP, and 128 NULL rows; all 15,990 checkable within-run transitions met the logged update rule and zero floor. RT, numeric code, stop outcome, and movement checks had no disagreements, although 1,300 active rows had numeric responses alongside the text label n/a. This finite event-field audit does not establish an acquisition error, stopping-latency estimate, behavioral performance change, neural result, or biological cause. Exact analyzed bytes are SHA-256 pinned. The live mirror metadata name v1.0.0, but mixed CHANGES labels prevent confirmation that the full analyzed mirror equals that immutable snapshot.
stop-signal task · staircase · event tables · training-to-scanner continuity · OpenNeuro · ds000030 · data provenance
References
3- usesIn ds000030 scanner files, zero logged stop delays raised the fixed-quantile SSRT subtraction by a mean 30.520 ms sci:sha256:31421430de8ed67ff78fad03fb2f828c94f067358264a73a64976837943452f6
Reused its 266 bundled raw scanner event tables and four metadata files, independently verified every input SHA-256, and paired them with all 267 newly retrieved public training tables to test training-to-scanner staircase continuity. Its estimator and executable scripts were not reused.
- doi:10.1038/sdata.2016.110
- doi:10.7554/eLife.46323
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesThe training and scanner delay histories do not agree for 1 of 266 ds000030 participants.
Its two training-to-scanner failures for sub-10855 supplied the question for the complete history comparison. The present check repeats the last-value count and locates earlier record differences.
- usesThe Pearson correlation difference is less than zero after adjustment for training GO time in 241 ds000030 participants.
The archived public training event tables supply the 266 paired training GO reaction-time records used for the adjustment. Their bytes and Git blob identifiers agree with the pinned public source version.
- usesFixed-response replay keeps a +50 ms start change through 16 stop trials in 495 of 532 ds000030 scanner staircases.
The analysis uses its 533 event tables, four metadata files, manifests, and endpoint table. Its 530 of 532 start agreement motivates the changed-start replay.
- supportsThe starting delay term is 0.5301 in the scheduled SSRT correlation for 107 ds000030 CONTROL pairs.
Fresh ds000030 1.0.0 CONTROL data show post-final training values matching 250 of 252 scanner starts, with the same two sub-10855 exceptions. All 214 starts in the 107-pair primary sample match. This supports the reported transfer pattern only in the shared CONTROL subset and connects it to a descriptive covariance calculation.
- usesInitial delay dominated the SSRT agreement covariance term in 232 ds000030 pairs
The 267 training event tables and four source metadata files were reused after checksum and pinned-source checks. Its state-continuity observation motivated the initial-delay diagnostic.
- supportsTraining and scanner SSRT residuals have a correlation of 0.454902 in 232 ds000030 pairs.
New code gives the same 530/532 scanner-start agreements with final training updates. All 464 starts agree in the primary sample. These are the same source observations.