This is the code repository for a project that aims to test whether, how, and how well humans encode values in their memory. We test the normative theory for allocating limited memory resources in reinforcement learning, which was originally originally proposed in Patel et al. 2020, NeurIPS.
⚠️ The data is not hosted on github. Hence, none of the files in src/analysis will work unless you have your own data. If none of the collaborators (Luigi Acerbi, Alexandre Pouget, and Antonio Rangel) have any objections, we will release the data when our work gets published.
Install the required packages in a new environment using mamba with mamba create --name <env_name> --file requirements.txt, or using conda with conda create --name <env_name> --file requirements.txt, or in an existing environment using pip with pip install -r requirements.txt.