Skip to main content

How the Brain Makes Decisions: Modeling the Dynamics of Neurons that Drive Choice (DSI-SRP)

Posted by on Saturday, August 1, 2020 in College of Arts and Science, Completed Research, DSI-SRP, Natural and Life Sciences, Social and Behavioral Sciences.

This DSI-SRP fellowship funded Claire Hanson to work in the laboratory of Professor Thomas Palmeri in the Department of Psychology during the summer of 2020. Claire graduated in May 2021 with a major in Neuroscience and minor in French.

The project funded by this fellowship aimed to understand how the brain makes decisions. Historically, the canonical model of firing rates of decision-making neurons has been the accumulation of evidence model; accumulation of evidence is also the canonical model used to explain human and non-human primate decision-making behavior. This model assumes that neural activity gradually ramps up until a threshold is reached and then a decision is made. A relatively recent publication challenged this notion, introducing evidence for the possibility that an immediate transition in firing rates – a step rather than a ramp – is a better model to describe the dynamics of decision-making neurons when those dynamics are measured on a trial-to-trial basis rather than averaged across trials. The question of whether the firing rates of decision-making neurons are better characterized by ramping vs. stepping dynamics is foundational for our theoretical understanding how the brain makes decisions. Claire’s project involved conducting Monte Carlo simulations of model neurons with known dynamics and using a Bayesian statistical analysis program testing for ramping vs. stepping dynamics (an adaptation of same analysis program used by the authors who proposed stepping dynamics as the preferred model). These simulated model neurons included those with simple steps or with simple ramps, as well as the diffusion model of decision making and the leaky competing accumulator (LCA) model of decision making (both of which are members of the class of accumulator models); each of these models produced simulated spike trains across multiple trials that could be analyzed in the same way that real neural data is analyzed. Claire’s simulations have shown that while simple steps are classified as simple steps by the analysis program and that simple ramps are characterized as simple ramps by the analysis program, the more complex dynamics of the diffusion and LCA are actually characterized as *steps* by the analysis program. Even though models like the diffusion model and LCA are clearly accumulating evidence over time, the analysis program characterizes them as steps. While real neurons might look “step-like” on a trial-by-trial basis, the computations being performed by these neurons may still be best characterized as an accumulation of evidence over time.

Claire continued this project through her senior year for her honor’s thesis, graduating with Highest Honors in Neuroscience. Claire joined the NIH baccalaureate program in the summer of 2021 and plans to then go on to an MD-PhD program.

In addition to receiving support through a DSI-SRP fellowship, this project was supported and facilitated by the DSI Data Science Team through their regular summer workshops and demo sessions.

Tags: , , , , ,