Drift diffusion model and beyond for linking brain to behavioral data in patient populations

Thursday, March 4, 2021 – 4:00 – 5:00 pm

Dr. Michael J. Frank, Cognitive, Linguistic & Psychological Sciences; Neuroscience Graduate Program; Director,Carney Center for Computational Brain Science; Carney Institute for Brain Science, Psychiatry and Human Behavior, Brown University

What?

Computational modeling can formally adjudicate between theories and affords quantitative fits to behavioral/brain data. It can help reduce the dimensionality of large datasets, in principle by 'carving nature at its joints'.
This talk has two parts: in the first I describe an example of how simplified models of brain function like the drift diffusion model (DDM) can be useful for linking neural circuits to decision making functions. The models provide a mechanistic interpretation of changes in decision making in patient populations.

In the second part of the talk I will present a new method using artificial neural networks that learn approximate likelihoods for arbitrary generative models, allowing fast posterior sampling with only a one-off cost for model simulations that is amortized for future inference.

Why?

For quantitative fitting purposes, the space of plausible generative models considered is dramatically limited by the set of models with known likelihood functions. For many models, the lack of a closed-form likelihood typically impedes Bayesian inference methods. As a result, standard models like the DDM are often fit to data for convenience, even when other models might be better suited to describing how the brain generates cognition and behavior. Relying on the likelihood approximation networks can accurately carry out Bayesian inference for a variety of neurocognitive process models.

How?

Prerequisites:

• basic coding skills (in any language, but most toolboxes are available in matlab, python and R).
• understanding of the key concepts of computational modeling and what models can and cannot do (and how to tell the difference)
• sometimes high-performance compute clusters can help for intensive model explorations across a wide variety of models. But often fitting can be achieved on a laptop.

Tutorial: http://ski.clps.brown.edu/hddm_docs is a free tutorial for basic HDDM, on HDDM with RL (RL-DDM), and on novel models that can be fit using neural networks. We provide code, a toolbox, and a tutorial allowing users to easily deploy these methods for arbitrary models linking brain to behavioral data and for interrogating alterations in patient populations.

Reference publications:

• Wiecki TV, Sofer I and Frank MJ (2013). HDDM: Hierarchical Bayesian estimation of the Drift-Diffusion Model in Python. Front. Neuroinform. 7:14. doi: 10.3389/fninf.2013.00014 https://www.frontiersin.org/articles/10.3389/fninf.2013.00014/full
• Pedersen ML and Frank MJ (2020). Simultaneous hierarchical Bayesian parameter estimation for reinforcement learning and drift diffusion models: a tutorial and links to neural data. Computational Brain & Behavior. doi: 10.1007/s42113-020-00084-w https://doi.org/10.1007/s42113-020-00084-w
• Fengler A, Govindarajan LN, Chen T, and Frank MJ (2020). Likelihood Approximation Networks (LANs) for Fast Inference of Simulation Models in Cognitive Neuroscience. bioRxiv https://www.biorxiv.org/content/10.1101/2020.11.20.392274v1

Datasets: Any dataset in which behavioral data are collected, with or without neural data.

Registration Information

You must subscribe to the Workshop series using your email to receive Zoom information for the virtual workshop: https://medicine.yale.edu/psychiatry/map/