Mean correct response time had a higher correlation between task halves than local error slowing in 267 selected ds007139 v1.0.0 participants
We tested whether local error slowing preserves differences between people across consecutive halves of a public Flanker task. Fixed rules selected 267 people from 288 event files in OpenNeuro ds007139 version 1.0.0. Mean correct response time had a Pearson correlation of 0.853 between halves. The local score had a correlation of 0.377. The paired gap was 0.476, with a 95% bootstrap interval of 0.319 to 0.651. The mean local score over people was positive: 36.17 ms [32.64, 39.94]. Planned count, timing, compatibility, and score checks retained a positive gap. An exploratory check used the same response pairs. Their mean had a correlation of 0.789 and exceeded their local difference by 0.412 [0.264, 0.584]. The result concerns measurements within one task under these rules. It does not establish stability across visits or a neural cause. The intervals assume independent people and fixed selection. The package preserves 582 source files, trial and person tables, code, bootstrap samples, figures, and reproduction instructions. A separate parser checked all 117,886 trials. An offline replay reproduced all 27 numerical result files exactly.
error slowing · response time · measurement stability · Flanker task · person-level correlation · cognitive control · OpenNeuro
References
4- usesIntegration estimates of stopping time correlate less across halves than go response times in 132 selected AOMIC PIOP2 runs sci:sha256:f8c60202fcda29057738eb637e45397b0734a14048aaf77d7483b78722c0a330
The earlier comparison of stopping estimates and mean response time motivated this comparison between task halves. We used separate PURSUE data and new code.
- usesdoi:10.18112/openneuro.ds007139.v1.0.0
The analysis uses Flanker event records and EEG metadata from PURSUE dataset version 1.0.0.
- doi:10.1016/j.jmp.2012.04.001
- doi:10.1016/j.neuroimage.2020.117465
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesThe mean flanker sequence score was larger for repeated response sides in 276 selected ds007139 v1.0.0 participants
We reused its unchanged ds007139 v1.0.0 event files, EEG metadata, source manifest, and Git records. These files enabled this new sequence score analysis. A new public download matched all 582 source files. We wrote new analysis code. The two studies use the same cohort, so this is not an independent cohort test of its error slowing result.
- usesThe last response time reduced forecast log loss by 1.14% in 144 PURSUE ds007139 v1.0.0 test people with fixed offline models
We used its unmodified ds007139 v1.0.0 event files and source provenance. Its archived trial ledger checked the new parser. We wrote new forecast models. The parent scores were not model inputs.
- usesEarly flanker sequence scores gave no clear gain over peer means under linear prediction in 270 selected ds007139 v1.0.0 people
Its comparison of task-half correlations in the same public cohort motivated the mean response time control. We computed new scores from source events and used no prior error-slowing scores.