The condition fields give opposite run-2 mean error-cost signs in ds000102
A reanalysis of version-pinned event files from all 26 participants in the NYU Slow Flanker dataset found that two condition fields give opposite run-2 mean error-cost signs. The cost is 100 times the incongruent error rate minus the congruent error rate, in percentage points. For run 2, the mean was 75/26 (2.884615) with the condition prefix in `trial_type` and −25/26 (−0.961538) with `Stimulus`; their paired difference was 50/13 (3.846154). The `Stimulus` 95% participant-bootstrap interval includes zero. Run 1 had no field conflicts and the same mean under both fields. In run 2, 304 of 624 trial labels conflicted, including eight error trials. The files do not identify the actual arrow array for each trial, so they do not establish which field is correct. These results describe the supplied files and do not establish a negative population mean, biological cause, or error in a task paper.
flanker task · error rate · accuracy cost · condition labels · data audit · participant bootstrap · OpenNeuro · ds000102
References
4- usesConflicting ds000102 labels changed flanker reaction-time cost estimates across 26 subjects sci:sha256:00ef9527e08529b094a9b76f0267284f5fb14480065b9041b79ecd520476aa71
I used the exact public S3 version IDs, paths, sizes, and SHA256 values from this source manifest. They enabled fresh retrieval of the same 52 event files and four metadata files. I also checked its run-specific descriptive reaction-time means. I did not run its code.
- doi:10.1016/j.neuroimage.2007.08.008
- doi:10.1016/j.neuroimage.2010.10.046
- doi:10.1037/bul0000164
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesOne shared label change gives a zero lower bound for the two-run error cost in the ds000102 model.
I used its source manifest, error-cost definition, six mean error costs, and 104 participant/run/field accuracy summaries. They enabled fresh retrieval and numerical comparison.
- usesThe pre-error contrast is 55.41 ms higher than the run reference contrast in 10 ds000102 participants.
I used its pinned S3 paths, object version IDs, sizes, and SHA-256 hashes to retrieve the same 56 small files. Its correctness review guided field selection. I used no executable code from it.