Analysis code and group summaries for Neural responses during motor statistical learning and gradual sequence replacement

Analysis source code and group-level imaging and behavioral summaries for an interleaved dual-structure serial reaction time fMRI study.

Version 1.2.2 adds the continuing-sequence benchmark and early participant-paired new-minus-continuing summaries, with the source code that produced them. The article title now uses gradual sequence replacement. Continuing contrasts have separate four-parcel headline and 48-parcel families per phase and cohort; early new and paired companions retain their 48-parcel families. The slice-timing-aligned sensitivity is retained. The 22-participant unaligned uncensored reference analysis with at least 80% observed regional support remains primary. The overlapping learner-selected 20-participant sample appears only in the article supplement. Findings are exploratory; confidence intervals are pointwise. Analysis history is disclosed in the article Methods.

Participant-level observations, fitted models, MRI images, raw permutation arrays and recoverable individual values are withheld. Complementary suppression applies across added and inherited summaries, overlapping cohorts, coverage rules and related components. Numerical ROI reports that could restore suppressed statistics are withheld. Code labels and paths refer to restricted inputs; approved data access and separate dependencies are required to re-estimate models. The complete original run_saved.py remains withheld; its generic direct function and necessary imports are distributed.

This version supersedes version 1.2.1 for citation. Earlier archives remain unaltered: version 1.2.1 at doi:10.5281/zenodo.23484681, version 1.2.0 at doi:10.5281/zenodo.23474855, version 1.1.0 at doi:10.5281/zenodo.23471923, and version 1.0.0 at doi:10.5281/zenodo.23438462. Original software is MIT; group summaries, figures and documentation are CC BY 4.0.