The striatum is critical for making decisions based on predictions of future reward. These predictions can be updated by brief pulses of dopamine, encoding reward prediction errors. However, it is unclear how this mechanism handles the need to generate predictions over multiple time horizons: from seconds or less (if singing a song) to potentially hours or more (if hunting for food). Here we monitor and model dopamine pulses across distinct striatal subregions, and find that these reflect predictions over distinct temporal scales. Dopamine dynamics systematically accelerated from ventral to dorsal-medial to dorsal-lateral striatum, in the tempo of their spontaneous fluctuations, their integration of prior rewards, and their discounting of future rewards. In this way parallel striatal circuits can achieve a more comprehensive set of value computations, to guide a broad range of reward-related behaviors.

https://www.biorxiv.org/content/biorxiv/early/2021/11/07/2021.10.31.466705.full.pdf

Striatal dopamine pulses follow a temporal discounting spectrum

The learning of prospective and retrospective cognitive maps within neural circuits

Relative salience signaling within a thalamo-orbitofrontal circuit governs learning rate.

A key concept in reinforcement learning (RL) is that of a state space. A state space is an abstract representation of the world using which statistical relations in the world can be described. The simplest form of RL, model free RL, is widely applied to explain animal behavior in numerous neuroscientific studies. More complex RL versions assume that animals build and store an explicit model of the world in memory. To apply these approaches to explain animal behavior, typical neuroscientific RL models make assumptions about the underlying state space formed by animals, especially regarding the representation of time. Here, we explicitly list these assumptions and show that they have several problematic implications. We propose a solution for these problems by using a continuous time Markov renewal process model of the state space. We hope that our explicit treatment results in a serious consideration of these issues when applying RL models to real animals.

https://biorxiv.org/content/10.1101/2021.02.07.430001v1.full.pdf

K Namboodiri

Papers

What is the state space of the world for real animals