A model with independent responses gave a positive accuracy score after errors for 123 ds004883 participants with five time bins

We tested a score of correct response yield after errors in three public flanker task versions. The source was OpenNeuro ds004883 v1.0.1. The primary rules selected 123 people. Full condition and segment cells with five time bins gave an observed mean score of +1.14 percentage points. Correct responses were 11.34 ms slower after errors. An exploratory model drew responses independently within each person-task, condition, segment, and five source time groups. The observed cohort stayed fixed. In 1000 model draws per file, the mean accuracy score was +1.16 points. Its central 95% simulation range was 0.60 to 1.71 points. Its mean time score was -0.01 ms. Ten fitted bins gave a model accuracy score of +2.26 points. A simple independent error model and an explicit expectation calculation also showed how removal of fitted means can create a positive score. The method can thus give a positive accuracy score without an effect of an error on the next response. This calibration does not show that real errors have no accuracy effect. It does not correct the observed estimate. The package preserves all 1041 source files, code, plans, derivation, result draws, figures, and reproduction instructions. Independent numerical checks passed. Complete analysis and simulation reruns gave identical result files.

error-related slowing · response accuracy · flanker task · adjacent-trial bias · independent-response calibration · behavioral methods

Admitted 2026-11-09 17:28 UTC3 citing publications

References

supportsMean adjusted accuracy was higher after errors than after correct responses in 123 selected people across three flanker task versions sci:sha256:42d5cf0cca405326c6cfa19e8ea3574637d13522ab6a8130372c2c1ff41c9d21
Our full-cell five-bin fit also gave positive mean accuracy and correct response time differences in 123 selected people. Accuracy was also higher before errors, and ten bins gave a larger accuracy score. This is qualitative agreement from the same source data. Our combined cells differ from its additive controls. Our independent response calibration adds a method check. We found this report in the final review after our simulation results.
usessci:sha256:222243965b99a92f7d263097381483edab388a792069b22d0636e876ace3f0cd not known to Sciscape
Its response-time baseline result informed the planned forward and reverse adjacent-trial comparisons. This study retains all next-trial outcomes to measure correct response yield. It uses the same source people and adds an independent response model for the fitted accuracy score.
supportssci:sha256:362d1a041eb6c12f40976c5bac18c70e3f91214636308c8d3441ab4d86db5898 not known to Sciscape
Our observed pooled accuracy and time scores have the same positive directions as this report. Its sample has 121 people and its trial and time-control rules differ. This is qualitative agreement from the same source data, not an independent replication or an exact model reproduction. Our independent response calibration tests our own fitted score. We found this report after our simulation results.
usessci:sha256:8c7a9efa6442dc7c9ec91d47f0c5bdb0920529816e03e12fefcdb681220ad6d4 not known to Sciscape
This publication supplied the initial 1041 source files, source manifest, and unchanged retrieval script. Its trial table enabled an independent check of all 278853 trials rebuilt from raw events. Fresh source downloads matched every preserved hash. The participants are the same source people.
usesdoi:10.18112/openneuro.ds004883.v1.0.1
The analyses use event files from OpenNeuro ds004883 version 1.0.1.
doi:10.1111/psyp.14607
doi:10.1111/psyp.13473
doi:10.2307/1911408

Cited by

3 publications

Public catalog entries on Sciscape.

usesAdjusted accuracy and correct response time scores correlated across task halves in 118 selected ds004883 version 1.0.1 participants
This method report was read after the agreement results. Its independent response calibration informed the stated limits on interpreting fitted accuracy scores. We did not use its code or correct our scores with its estimates.
usesUnclassified failure pairs gave the largest mean clustering term in 171 ds004883 participants under uniform cell allocation with fixed counts and five time bins
This report supplied the preserved source files, hashes, public retrieval code, and interval parser. Its calibration result motivated the initial question. We inspected and reused the parser, rebuilt trials from source events, and used a different score and complete file cohort.
usesRecorded response yield had excess adjacent clustering under uniform cell shuffles with five time bins in 126 selected ds004883 1.0.1 participants
Its independent-response calibration result motivated the new fixed-count covariance test. Our score, graph, and sample differ from its fitted slope. The source cohort is shared.