Biology subjects

Gerraty, R. T.

Publications and source records attributed to Gerraty, R. T..

4 recordsLinked to original sources

bioRxiv2026

The Periaqueductal Gray Selectively Supports Reversal Learning During a Flexible Discrimination Task in Mice

Flexible, goal-directed behavior depends on the ability to update value representations in response to changing contingencies. This ability depends on distributed brain networks, calling for use of whole-brain imaging. Widely used in human research, whole-brain imaging in rodents has been a major challenge that has only recently been properly addressed. Using functional MRI in behaving mice performing a go/no-go odor discrimination task, we compared neural activity during initial cue-reward learning (acquisition) and subsequent contingency reversal. To link neural activity to underlying learning processes, we modeled value updating using a model-free reinforcement-learning algorithm. Trial-by-trial estimates of state-action values allowed us to dissociate acquisition from reversal-related signals, revealing that ventral striatal responses tracked expected value during acquisition, whereas reversal learning additionally recruited the periaqueductal gray (PAG), a midbrain structure classically linked to threat processing and aversive learning. PAG activity closely followed model-derived signatures of reversal learning, implicating it in the suppression of previously rewarded actions and in updating behavior in the absence of explicit punishment. These findings reveal a previously unrecognized computational role for the PAG in value-based decision-making and cognitive flexibility, and substantiate task-fMRI as a powerful tool to study the rodent brain at a mesoscale resolution.

neuroscience↗

bioRxiv2020

Hierarchical Bayesian Models of Reinforcement Learning: Introduction and comparison to alternative methods

Reinforcement learning models have been used extensively to capture learning and decision-making processes in humans and other organisms. One essential goal of these computational models is the generalization to new sets of observations. Extracting parameters that can reliably predict out-of-sample data can be difficult, however. The use of prior distributions to regularize parameter estimates has been shown to help remedy this issue. While previous research has suggested that empirical priors estimated from a separate dataset improve predictive accuracy, this paper outlines an alternate method for the derivation of empirical priors: hierarchical Bayesian modeling. We provide a detailed introduction to this method, and show that using hierarchical models to simultaneously extract and impose empirical priors leads to better out-of-sample prediction while being more data efficient.

neuroscience↗

bioRxiv2019

Dopamine modulates learning-related changes in dynamic striatal-cortical connectivity in Parkinson’s disease

Learning from reinforcement is thought to depend on striatal dopamine inputs, which serve to update the value of actions by modifying connections in widespread cortico-striatal circuits. While considerable research has described the activity of individual striatal and midbrain regions in reinforcement learning, the broader role for dopamine in modulating network-level processes has been difficult to decipher. To examine whether dopamine modulates circuit-level dynamic connectivity during learning, we characterized the effects of dopamine on learning-related dynamic functional connectivity estimated from fMRI data acquired in patients with Parkinsons disease. Patients with Parkinsons disease have severe dopamine depletion in the striatum and are treated with dopamine replacement drugs, providing an opportunity to compare learning and network dynamics when patients are in a low dopamine state (off drugs) versus a high dopamine state (on drugs). We assessed the relationship between dopamine and dynamic connectivity while patients performed a probabilistic reversal learning task. We found that reversal learning altered dynamic network flexibility in the striatum and that this effect was dependent on dopaminergic state. We also found that dopamine modulated changes in connectivity between the striatum and specific task-relevant visual areas of inferior temporal cortex, providing empirical support for theories stipulating that value is updated through changes in cortico-striatal circuits. These results suggest that dopamine exerts a widespread effect on neural circuitry and network dynamics during reinforcement learning.

neuroscience↗

bioRxiv2016

Dynamic flexibility in striatal-cortical circuits supports reinforcement learning

Complex learned behaviors must involve the integrated action of distributed brain circuits. While the contributions of individual regions to learning have been extensively investigated, understanding how distributed brain networks orchestrate their activity over the course of learning remains elusive. To address this gap, we used fMRI combined with tools from dynamic network neuroscience to obtain time-resolved descriptions of network coordination during reinforcement learning. We found that learning to associate visual cues with reward involves dynamic changes in network coupling between the striatum and distributed brain regions, including visual, orbitofrontal, and ventromedial prefrontal cortex. Moreover, we found that flexibility in striatal network dynamics correlates with participants learning rate and inverse temperature, two parameters derived from reinforcement learning models. Finally, we found that not all forms of learning relate to this circuit: episodic memory, measured in the same participants at the same time, was related to dynamic connectivity in distinct brain networks. These results suggest that dynamic changes in striatal-centered networks provide a mechanism for information integration during reinforcement learning.\n\nSignificance StatementLearning from the outcomes of actions-referred to as reinforcement learning-is an essential part of life. The roles of individual brain regions in reinforcement learning have been well characterized in terms of the updating of values for actions or sensory stimuli. Missing from this account, however, is a description of the manner in which different brain areas interact during learning to integrate sensory and value information. Here we characterize flexible striatal-cortical network dynamics that relate to reinforcement learning behavior.

neuroscience↗