This chapter examines the relationship between rationality, reward, and learning. In order to instil rationality into the decision-making system, actions must have meaning. When studying the learning abilities of a subject, any beneficial consequence is called reward. In order for the lamprey to learn to choose one action over the other, it is necessary to introduce a reinforcement learning mechanism and a system which informs the lamprey on the outcome of its choice. The chapter then goes on to highlight the crucial role of dopamine in these processes and shows how it is implemented in the vertebrate neural network of decision-making. The effect of dopamine on synaptic transmission is evident in many physiological and pathophysiological processes, but it is the role of dopamine in learning that is important here. The chapter then considers the substantia nigra and the ventral tegmental area, which are closely connected to the diencephalic and telencephalic loops.
Neural Network Dynamics for Model-Based Deep Reinforcement Learning with Model-Free Fine-Tuning