The two CNP model scripts do not show the source field for stop-signal delay.
A published stop-trigger-failure analysis reports 113 controls from legacy CNP dataset revision 1.0.5, but its public setup script loads a prepared table whose source field and preparation rules are not available in the inspected release. We audited candidate inputs from OpenNeuro ds000030 version 1.0.0, identified by exact S3 object version IDs and SHA-256 hashes; equivalence with revision 1.0.5 is unknown. Applying the paper's 17 listed participant exclusions to 130 controls yielded 113 candidates. Rules that retain omissions and successful STOP trials, and retain responses only when labeled correct and at least 150 ms, kept 10,598 GO and 3,545 STOP rows. Of 243 STOP rows with zero exported delay, 225 remained after filtering; all 243 had a response before planned LadderTime and were marked unsuccessful stops. These are candidate input counts, not recovered model inputs or estimates of a change in trigger failure. We did not fit the model or establish physical sound onset, neural trigger failure, population uncertainty, or a biological causal effect. The source release differs from the publication's named legacy revision, and a supplement reports 114 participants where the paper reports 113.
stop-signal task · stop-trigger failure · stop-signal delay · reaction time · CNP · ds000030 · candidate input audit · OpenNeuro · trial filtering · reproducibility
References
6- usesZero exported stop-signal delays co-occur with early responses in 266 CNP scan sessions sci:sha256:2526b705245e4c67d78ad1504dcdcf4b4bfbc99aa804b1547dbf61342a790b10
We used its raw scanner event files, participant table, dataset metadata, and version-pinned source manifest. These inputs enabled the 113-control candidate input audit and fresh public source checks. We did not run its programs or recover the later study model input.
- doi:10.1162/imag_a_00454
- doi:10.7554/eLife.46323
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The candidate trial counts use the CC0 raw event files and participant table identified as OpenNeuro ds000030 version 1.0.0, with exact object versions and hashes recorded in the source manifest.
- usesdoi:10.12688/f1000research.11964.2
This is the cited description of the preprocessed CNP dataset from which the recorded ds000030 source is drawn.
- doi:10.1038/sdata.2016.110
Cited by
2 publicationsPublic catalog entries on Sciscape.
- The specified filter keeps 225 STOP rows with StopSignalDelay=0 in 70 of 113 CNP runs.
[Legacy relation: reproduces] I used the input manifest, reference tables, and archive index. New public file copies gave the same trial decisions and counts. The public archive checks also agreed. This relation excludes the study model fits.
- usesThe exported delay gives higher mean trigger-failure estimates in two race models for 112 CNP runs.
We used the 113-run candidate selection, pinned public-file manifest, and response-filter rule. Fresh downloads agreed with all 117 hashes and object versions. Its missing original input table remains unresolved.