CBGT

This repository contains code for implementing the spiking cortico-basal ganglia-thalamus (CBGT) network and drift-diffusion model (DDM) fits described in the manuscript Reward-driven changes in striatal pathway competition shape evidence evaluation in decision-making.

The code requires several dependencies to be installed (see below for instructions). After completing the installation procedure below, the demo notebook can be downloaded and opened inside Jupyter.

Requirements

• OSX or Linux

• Anaconda with Python 3.* (for OSX, Linux)

• gcc (if Linux) or gcc-8 (if OSX, see here)

Installation Instructions

# create a new conda environment with python 3.6
# and hit 'y' to verify the install 
conda create -n cbgt_env python=3.6 anaconda ipykernel

# activate 'cbgt_env' environment
source activate cbgt_env

# use conda (not pip) to install pymc
conda install pymc=2.3.6 --no-deps

# install hddm and kabuki
pip install --upgrade kabuki hddm

# finally install numpy version 1.11.3
# (avoids hddm incompatibility with later numpy)
pip install numpy==1.11.3

# install cbgt package
pip install -U cbgt --no-cache-dir

• After installing everything, run jupyter notebook in your terminal to start Jupyter in your browser
• Drag/drop the demo notebook (CBGT_PLOSCompBio2019_Demo.ipynb) into the Jupyter browser window

Miscellaneous:

• HDDM resources: (methods paper, demo, how-to)
• Jupyter Notebook tutorial