Analysis materials for Older Adults Show Altered Default Mode and Executive Control Network Connectivity during Fairness Decisions (bioRxiv DOI 10.1101/2025.08.13.670194).
The repository contains behavioral/task data, analysis code, FSL templates,
network and ROI masks, and compact revision outputs. MRI images are distributed
through OpenNeuro dataset ds003745, not through this Git repository.
This DVS-Lab/srndna-ug repository and its main branch are the repository of
record for this paper. The older srndna and srndna-datapaper repositories
remain historical/upstream sources, while the server directory
srndna-ultimatum stores legacy production derivatives. See
docs/REPOSITORY_OF_RECORD.md; do not merge those Git histories or treat a
partial derivative mirror as authoritative.
From the repository root:
# Static checks, synthetic tests, and read-only audits
make test
# Reviewer-requested behavioral models, source tables, and figures
make reviewer-behavior
# Read-only local checks of tracked masks, L3 templates, and ROI influence
make reviewer-imaging-auditThe behavioral workflow requires R with lme4, lmerTest, ggplot2, and the
scales dependency used by ggplot2. The image-header audit requires FSL's
fslhd and fslstats; validation skips that check cleanly when FSL is absent.
Exact behavioral package versions are written to
results/reviewer/tables/behavior_software_versions.tsv.
source_data/bids/: curated Ultimatum events TSVs in a BIDS-like layout.source_data/partner_ratings/: coded Bargaining partner-rating inputs.behavioral_analyses/data/: submitted cleaned behavioral and covariate inputs.code/: active revision audits plus legacy preprocessing/FEAT scripts.templates/: submitted and later FSL templates; these are provenance evidence, not all interchangeable production entry points.masks_SANS/andimaging_plots_SANS/: focal submitted masks and plotting extracts used for low-cost verification.results/reviewer/: compact aggregate revision tables and figures.logs/records/: the manuscript-result manifest and durable audit records.docs/SERVER_IMAGING_AUDIT.md: read-only handoff for production imaging provenance and the gate before any image-level rerun.docs/REPOSITORY_OF_RECORD.md: canonical ownership and boundaries among the paper repository, OpenNeuro source data, and legacy production derivatives.
Start with code/WORKFLOW_AUDIT.md. It distinguishes established submitted
analyses, revision analyses, provisional reconstructions, and unresolved
provenance. The machine-readable result index is
logs/records/manuscript-result-manifest.tsv.
Reviewer scripts write participant-level derived rows under
results/reviewer/private/. That directory is ignored and remains local.
Only aggregate reviewer tables and figures belong in the active repository.
Raw DICOMs, MRI data, the full BIDS dataset, and other directly identifying
source material are not included. A local bids/ directory is ignored. See
source_data/README.md for the narrow set of analysis inputs retained here.
The analyses were not preregistered. Legacy text claiming otherwise was incorrect and has been removed. Reviewer-requested work is labeled as revision analysis rather than presented as part of the submitted workflow.
The local checkout is suitable for behavioral reanalysis, template inspection,
mask/header checks, and preparation of server commands. It is not an
authoritative imaging-derivative store. Do not run the legacy L3 wrappers or
launch FEAT/FLAME/randomise from this checkout without first completing the
production provenance audit.
On the designated Linux host, follow docs/SERVER_IMAGING_AUDIT.md. The
collector copies only small metadata/design artifacts and never starts an
analysis. Any approved robustness run must use a new versioned output directory
and preserve the submitted result.
The initial checkout was approximately 8 GB, including a 3.70 GiB Git object
database dominated by generated RL/HBDM fits and diagnostic figures. On
2026-09-07, 4,218 generated files (4.03 GiB in the working tree) were hashed,
recorded in logs/records/generated-artifact-archive.tsv, and moved to a
verified local archive. Git history was then rewritten to remove those files,
older generated contents of behavioral_analyses/fits/, unrelated RL/DDM
source and tutorial code, and the legacy root bids/ tree. The resulting Git
object database is approximately 142 MiB. See docs/HISTORY_REWRITE.md before
resynchronizing a clone made before the rewrite.
Many older shell scripts and notebooks retain lab-specific absolute paths and
represent exploratory or superseded analyses. They remain for provenance while
the resubmission audit identifies the exact submitted result chain. Unrelated
DDM/RL experiments, generated model fits, and a generic RL tutorial were removed
from both the active tree and Git history. Do not assume that a file is active
merely because it is tracked. The modern entry points above are path-portable
and covered by make test.
This work was supported in part by NIH awards R21-MH113917 and R03-DA046733 to David V. Smith, R15-MH122927 to Dominic S. Fareri, and a Scientific Research Network on Decision Neuroscience and Aging pilot award (NIH R24-AG054355, Gregory Samanez-Larkin, PI). See the manuscript for the full contributor list.