Brain, Body and Machine Seminar

The Neurodynamics of Simple Decisions: Drift-diffusion Equations as Models for Single Brains, and for Group Behaviors

Prof. Philip Holmes

Princeton University

I will describe how simple stochastic differential equations can model  evidence accumulation and decision making, sketching their derivation from biophysically-detailed models of spiking neurons, and relating them to the sequential probability ratio test from statistical decision theory. This connection yields a speed-accuracy tradeoff that optimizes rewards in a simple two-alternative perceptual decision task. I will compare the resulting model predictions with human behavior and advance explanations for failures to optimize. Finally, I will show how drift-diffusion models can be extended to describe choices in a social gambling task in which players receive limited information regarding other group members' choices and rewards.