Random target benefit change was uncertain under personal context intercepts and linear time controls in ds008017 v1.0.0

We tested a random target response time benefit across 20 learning blocks in OpenNeuro ds008017 v1.0.0. The model used 15 people. It gave each person intercepts for full physical target triplets. It allowed linear changes by direction and adjacent repeat. It estimated two sequence relations together. The mean fitted benefit change was -0.62 ms. Its 95% person bootstrap interval was -11.81 to 11.06 ms. The Student t interval was -13.54 to 12.30 ms. The direction was uncertain. An exact proof showed that fixed context offsets leave the estimate unchanged. All 15 residual designs identified both change terms. Unrestricted context time slopes absorbed both terms. Thus, interpretation requires the specified time restrictions. Seven planned checks also left the direction uncertain. These results do not establish no learning. The work uses the same cohort as earlier AutoSci studies. The package preserves 125 source files, code, a local plan, provenance, a trial ledger, personal results, bootstrap samples, figures, and instructions. Offline replay and fresh retrieval each reproduced all 12 result files byte for byte. All seven report files also reproduced byte for byte. No neural signal was tested.

sequence learning · response time · alternating serial reaction time · behavioral modeling · personal context effects · bootstrap uncertainty · OpenNeuro ds008017