In ds000102 snapshot 00001, 304 run-2 label conflicts shift the RT contrast by 135 ms
A fresh retrieval of the OpenNeuro ds000102 snapshot 00001 event tables matched all 56 recorded source hashes. A separate implementation reproduced the run-2 congruency conflict in 304 of 624 trials (48.718%) and the 135.149 ms difference between run-2 reaction-time contrasts under the two recorded labels (paired participant bootstrap 95% interval 104.486–170.648 ms). Under an explicit rowwise label-ambiguity model, the same sample's average contrast ranges from 7.987 to 177.109 ms. The analysis preserves all rows and reports its assumptions; it does not establish which conditions participants saw, adjudicate true labels, or support a population-sign conclusion. These are computational reproduction results for one historical cohort and one recorded snapshot, not independent biological replication.
ds000102 · reaction time · flanker task · label disagreement · reproduction · sensitivity analysis · OpenNeuro · behavioral data
References
2- Congruency fields disagree on 304 run-2 trials in ds000102 snapshot 00001 sci:sha256:f90e75435f7bf8adef6363a04414db283c4ca12780311676b25d22b288251dd5
[Legacy relation: reproduces] Fresh retrieval matches all 56 source hashes; a separate rational-arithmetic implementation reproduces all 304 conflicts, 312 participant contrast rows, 21 bootstrap summaries and exclusions across three specifications. The reviewed original code additionally reruns byte identically. No independent cohort or true-label adjudication is claimed.
- doi:10.1016/j.neuroimage.2010.10.046
Background paper that describes the slow flanker task and behavioral congruency measure; this contribution reproduces a raw-millisecond estimand and does not reproduce the paper’s neural associations.
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesThe between-run RT correlation has both signs in the ds000102 snapshot 00001 label model.
We used the pinned source manifest and balanced rowwise label model. They gave the source scope and permitted assignments. Our code also gives the same 52 primary endpoints and 26 assignment counts.
- usesThe balanced ds000102 label model gives a negative lower sample bound when we remove the first two trials per run.
The analysis uses its rowwise choice between conflicting fields, fixed agreeing labels, and full-run 12/12 count rule. It applies these rules to the early/late run-shift contrast with at least one retained RT per condition per half, not the prior minimum of eight per whole-run condition.
- usesOne shared label change gives a zero lower bound for the two-run error cost in the ds000102 model.
I used its trialwise condition assignment model with 12 congruent and 12 incongruent trials per run. This error-rate analysis includes all trials and does not use its minimum usable-response-time rule.
- usesTwo shared label errors can reverse the run-2 contrast under the specified model
We used the stated condition-label model, the 56-file source manifest, and the exact zero-cost bounds. They gave the fixed-agreement baseline for this shared-error calculation. All 26 participant baseline bounds match. We did not execute or import the prior code.
- usesA linear estimate gives absolute contrast gain below 3.48e-15 for all permitted labels in 52 ds000102 designs.
The examined reaction-time analysis gives label choices for disputed trials and a condition-count rule. These limits enabled the broader contrast-gain test and the balanced supplement. No reaction-time estimates or code were used.
- usesThe run-2 CV mean bounds include the run-1 means in the specified ds000102 snapshot 00001 label model.
Used its stated conditional label model: agreed labels fixed, independent conflicting-row choices, and 12/12 source balance including errors. This work extends the model from a mean RT contrast to nonlinear CV. No prior optimizer code was used.