Value Learning through Reinforcement: The Basics of Dopamine and Reinforcement Learning

This chapter provides an overview of reinforcement learning and temporal difference learning and relates these topics to the firing properties of midbrain dopamine neurons. First, we review the Rescorla Wagner learning rule and basic learning phenomena, such as blocking, which the rule explains. Then we introduce the basic functional anatomy of the dopamine system and review studies that reveal a close correspondence between responses emitted by dopamine neurons and signals predicted by reinforcement learning. Finally, we introduce the generalization of the Rescorla Wagner rule to sequential predictions as provided by temporal difference learning, and discuss its application to phasic activation changes of dopamine neurons. Subsequent chapters in this section deal with more advanced topics in reinforcement learning and presume that the reader is familiar with material covered in this chapter. LEARNING: PREDICTION AND PREDICTION ERRORS