Failure to Find Altruistic Food Sharing in Rats

This repository is the research compendium for:

Wan, H., Kirkman, C. F., Jensen, G., & Hackenberg, T. D. (2021). Failure to find altruistic food sharing in rats. Frontiers in Psychology, 12, 696025. https://doi.org/10.3389/fpsyg.2021.696025

It brings together the Bayesian models, reproducibility workflows, figures, and representative trial videos. The data are hosted separately in the study's OSF project; the version of record remains available from the journal article.

Study at a glance

Question. Will rats leave high-value food available for a familiar partner when food quantity, food motivation, and access conditions are varied?

Design. Three focal rats made repeated choices between sucrose pellets and 30 seconds of social access to a cagemate across seven experimental conditions. The analyses use Bayesian multilevel generalized linear models to estimate food-versus-social choice, response rates, and pellet allocation.

Finding. Rats responded for both food and social access, but food sharing was rare. Sharing occurred on approximately 1% of available opportunities, including under low food-motivation conditions. These findings did not support the claim that rats were motivated to share food altruistically under the tested conditions.

Open materials

Resource	Description
Published article	Open-access version of record in Frontiers in Psychology
Data on OSF	Raw_data.csv, hosted in the OSF Electronic Supplement
CODEBOOK.md	Variables, condition definitions, and missing-value conventions
Analysis/Analysis_R.qmd	Annotated R and CmdStan workflow
Analysis/Analysis_py.ipynb	Annotated Python and CmdStanPy workflow
Analysis/archive/published-2021/	Checksummed, unmodified published analysis source
config/cmdstan-version.txt	Pinned CmdStan version shared by both workflows
figures/	Publication figures and source files
video/	Representative food-choice and social-release trials
CITATION.cff	Machine-readable citation metadata

The maintained workflows use modern CmdStan interfaces while retaining the published variables, priors, likelihoods, and sampling configuration. The repository does not distribute the data. Download Raw_data.csv from the OSF Electronic Supplement before running either workflow.

Reproduce the analyses

Download Raw_data.csv from OSF and save it as Analysis/Raw_data.csv. The file is intentionally excluded from Git. Alternatively, set the RATSHARING_DATA environment variable to its local path.

The setup commands below should be run from the repository root. Compiling the Stan models requires a working C++ toolchain. Both workflows enforce the CmdStan version recorded in config/cmdstan-version.txt, use fixed model-specific seeds, and report divergences, maximum-treedepth events, E-BFMI, R-hat, and effective sample sizes. Sampling uses four parallel chains and 24,000 total iterations per model (including warmup), so a complete run can be computationally intensive.

R

The R environment is recorded in renv.lock (R 4.4.1).

install.packages("renv")
renv::restore()
cmdstan_version <- trimws(readLines("config/cmdstan-version.txt"))
cmdstanr::install_cmdstan(version = cmdstan_version)

Then open Analysis/Analysis_R.qmd in RStudio or render it with Quarto from the Analysis/ directory.

Python

The Python dependencies are pinned in requirements.txt.

python -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt jupyterlab
python -c 'from cmdstanpy import install_cmdstan; install_cmdstan(version=open("config/cmdstan-version.txt").read().strip())'
jupyter lab Analysis/Analysis_py.ipynb

The final command can be run from the repository root. If the notebook is opened another way, set its working directory to Analysis/ before execution.

Repository structure

.
├── Analysis/           # Maintained workflows, Stan models, source archives
├── config/             # Pinned computational-tool versions
├── figures/            # Figure PDFs and editable source files
├── presentation/       # Conference posters, slides, and abstracts
├── video/              # Representative experimental trials
├── CITATION.cff        # Citation metadata
├── CODEBOOK.md         # Dataset documentation
├── LICENSE             # License for repository code
├── renv.lock           # Locked R environment
└── requirements.txt    # Locked Python dependencies

Research transparency

• Data availability: The repository does not distribute the data. Raw_data.csv is available from the study's OSF Electronic Supplement.
• Ethics: The animal study was reviewed and approved by the Reed College Institutional Care and Use Committee.
• Funding: Reed College Summer Scholarship Funds supported the research.
• Conflicts of interest: The authors reported no commercial or financial relationships that could be construed as a potential conflict of interest.
• Scope: This repository documents a completed, published study. Corrections that improve reproducibility or documentation are welcome through GitHub issues.

Citation and reuse

Please cite the article above when using these materials. Citation metadata are available in CITATION.cff and through GitHub's Cite this repository menu.

Repository code is available under the MIT License. The published article is distributed under CC BY 4.0. Article content, third-party materials, and publisher-hosted supplementary files retain their stated licenses; the MIT license does not override those terms.