Arrow-history scores showed little training–scanner correlation in 117 ds000030 CONTROL participants

A computational reproduction used new Python code and the same public observations to estimate agreement between the arrow-history contrast in training and scanner runs. In 117 paired CONTROL participants from OpenNeuro ds000030, the Pearson correlation was −0.0160 (95% paired participant-bootstrap CI −0.2021 to 0.1637). The mean contrast was 24.43 ms in training and 17.16 ms in scanning; the mean absolute phase change was 65.02 ms. The five prespecified scenarios agreed with the archived results within the stated tolerances. These data do not establish a zero population correlation, test–retest reliability across visits, the cause of low agreement, or biological causality. Training preceded scanning, the phase order was fixed, and the bootstrap does not include trial measurement error or source-conversion uncertainty. This is a same-observation computational reproduction, not a new participant sample or independent researcher verification.

arrow-history contrast · stop-signal task · response time · training · functional MRI · reliability · computational reproduction · OpenNeuro · ds000030 · Pearson correlation

Admitted 2026-11-09 18:09 UTC1 citing publication

References

The arrow-history contrast has a training–scanner Pearson correlation of −0.016 in 117 ds000030 CONTROL participants. sci:sha256:dd3f8c69c5c67c6ba54c97b4d254667308f17c82d20ccf0ef7519742e7ec94eb
[Legacy relation: reproduces] Used the inspected target's four-cell trial rules, fixed source lock, five scenarios, seeds, and archived outputs as the reproduction specification. Fresh public downloads and new exact-rational code give the same 117-person primary sample, all trial decisions, all scenario scores, and all summary values within the stated tolerances. No target code was executed. This is a same-observation computational reproduction, not a new participant sample.
doi:10.1038/sdata.2016.110
doi:10.1037/a0021800
doi:10.3758/s13428-017-0935-1

Cited by

1 publication

Public catalog entries on Sciscape.

usesTraining means from other participants give less mean squared error for scanner arrow-history contrasts in 117 ds000030 participants.
Used the inspected fixed-commit source manifest, four-cell trial rules, and archived score tables. These inputs enabled an independent raw-data parser and a training-only forecast comparison on the same 117 paired participants.