Reinforcement Learning: Beyond the Basal Ganglia

Abstract

In clinical psychology, reinforcement learning is one of the many forms of conditional learning that focuses on reinforcing behavior that yields beneficial results. Similar to any other structure, the process of reinforcement learning was adopted and implemented into models of machine learning with ever-evolving complexity. This chapter attempts to outline the philosophy and terminology associated with reinforcement learning in machines while also describing basic models and their connection to vertebrate neuroanatomy. Specifically, machine learning borrows from the basal ganglia, a cluster of subcortical nuclei that are responsible for a host of tasks including procedural learning, cognition, and emotion. Within this chapter, the parallels between the basal ganglia and current machine learning are clearly evaluated with thoughtful deliberation of their future relationship.