SWDB 2025 Day 3

Welcome to the Summer Workshop on the Dynamic Brain (SWDB) 2025 repository focusing on Dynamical Systems and Reinforcement Learning approaches for understanding neural behavior and decision-making.

🎯 Overview

This repository contains educational materials and computational tools for understanding dynamical systems in neuroscience, with a particular focus on dynamic foraging task. This capsule provides hands-on experience with:

• Basic Reinforcement learning models for decision-making
• Model fitting and parameter recovery for behavioral data
• Recurrent neural network analysis for understanding neural dynamics

📚 Workshop Structure

Workshop 1: Model Fitting (Workshop-1-Model-Fitting.ipynb)

Learn how to fit computational models to behavioral data, including:

• Foraging behavior models (Q-learning, Loss-counting, etc.)
• Parameter recovery techniques under the same model architecture
• Model comparison of different model architectures using AIC/BIC

Workshop 2: RNNs for Dynamic Foraging (Workshop-2-RNNs for Dynamic Foraging.ipynb)

Explore recurrent neural networks and their dynamics:

• Actor-critic models for decision-making
• Neural trajectory analysis in hidden state space
• Dimensionality reduction (PCA) of neural activity
• Fixed point analysis and dynamical systems theory

Helper Functions

utils_model_recovery.py

Comprehensive toolkit for model fitting and analysis:

• Foraging agent classes: ForagerQLearning, ForagerLossCounting, BanditModel
• Parameter fitting: Differential evolution optimization
• Model comparison: BanditModelComparison class
• Visualization tools: Parameter recovery plots, confusion matrices

utils.py

Utilities for neural data visualization and analysis:

• 3D trajectory plotting for neural hidden states
• Animation tools for dynamical systems visualization
• Data processing helpers

🚀 Getting Started

Prerequisites

• Python 3.9+
• Jupyter Lab/Notebook
• Basic knowledge of neuroscience and machine learning

📁 Repository Structure

SWDB_2025_Dynamical_Systems/
├── code/
│   ├── Workshop-1-Model-Fitting.ipynb      # Model fitting tutorial
│   ├── Workshop-2-RNNs for Dynamic Foraging.ipynb  # RNN analysis tutorial
│   ├── utils_model_recovery.py             # Core modeling toolkit
│   ├── utils.py                            # Visualization utilities
│   ├── data/                               # Pre-processed datasets
│   └── resources/                          # Images and diagrams
├── environment/                            # Docker configuration
├── environment.yml                         # Conda environment
└── README.md                              # This file

📚 Further Links

• SWDB 2025 Dynamical Systems GitHub Repo

---

Happy learning and exploring the fascinating world of RL and dynamical systems in neuroscience! 🧠⚡