Bayesian Foraging Toolbox

A scientifically rigorous Python toolbox for Bayesian reinforcement learning in foraging contexts, with initial focus on multi-armed bandit problems.

Features

• Bayesian RL Agents: Reward maximization policies with belief state tracking and uncertainty quantification
• Foraging Environments: Custom multi-armed bandit environments (extensible to other foraging paradigms)
• Scientific Validation: Parameter recovery and cross-validated model fitting
• Modular Design: Object-oriented architecture for easy extension

Project Status

🚧 Under Development 🚧

This project is currently in active development following a multi-agent test-driven development workflow.

Installation

# Clone the repository
git clone <repository-url>
cd foraging

# Install in development mode
pip install -e .

Quick Start

Coming soon - awaiting initial implementation

Documentation

Coming soon

Development

This project follows a structured multi-agent development workflow with emphasis on:

• Test-driven development (TDD)
• Scientific rigor (parameter recovery, validation)
• Modular, object-oriented design
• Comprehensive documentation

See .agents/WORKFLOW.md for detailed development protocols.

Project Structure

foraging/
├── src/foraging/          # Core library code
│   ├── agents/            # Bayesian RL agents
│   ├── environments/      # Foraging environments
│   ├── belief/            # Belief state tracking
│   ├── policies/          # Reward maximization policies
│   └── utils/             # Utility functions
├── tests/                 # Test suite
├── docs/                  # Documentation
├── examples/              # Tutorial notebooks
└── scripts/               # Analysis scripts

Contributing

This is a research project following strict development protocols. See .agents/WORKFLOW.md for contribution guidelines.

License

To be determined

Citation

Coming soon

Contact

Project maintainer information