dockerHDDM: A user-friendly environment for Bayesian Hierarchical Drift-Diffusion Modeling

Drift diffusion models (DDMs) are pivotal in understanding evidence accumulation processes during decision-making across psychology, behavioral economics, neuroscience, and psychiatry. Hierarchical drift diffusion models (HDDM), a Python library for hierarchical Bayesian estimation of DDMs (Wiecki et al., 2013), has been widely used among researchers, including those with limited coding proficiency, in fitting DDMs and other sequential sampling models. However, issues of compatibility in installation and lack of support for more recently Bayesian modeling functionalities poses serious challenges for new users, limiting broader adaptation and reproducibility of HDDM. To address these issues, we created dockerHDDM, a user-friend computational environment for HDDM with new features. dockerHDDM brings three improvements: (1) easy-to-install once docker is installed, ensuring reproducibility and saving time for researchers; (2) compatible with machine with apple chips; (3) seamlessly integration with ArviZ, a state-of-the-art Bayesian modeling library. This tutorial serves as a practical, hands-on guide for researchers to leverage dockerHDDM’s capabilities in conducting efficient Bayesian hierarchical analysis of DDMs. The notebook presented here and within the docker image will enable researchers with various programming levels to model their data with HDDM.