Computational Models of Cognitive and Motor Control

Most of the earliest work in both experimental and theoretical/computational system neuroscience focused on sensory systems and the peripheral (spinal) control of movement. However, over the last three decades, attention has turned increasingly toward higher functions related to cognition, decision making and voluntary behavior. Experimental studies have shown that specific brain structures – the prefrontal cortex, the premotor and motor cortices, and the basal ganglia – play a central role in these functions, as does the dopamine system that signals reward during reinforcement learning. Because of the complexity of the issues involved and the difficulty of direct observation in deep brain structures, computational modeling has been crucial in elucidating the neural basis of cognitive control, decision making, reinforcement learning, working memory, and motor control. The resulting computational models are also very useful in engineering domains such as robotics, intelligent agents, and adaptive control. While it is impossible to encompass the totality of such modeling work, this chapter provides an overview of significant efforts in the last 20 years. It also outlines many of the theoretical issues underlying this work, and discusses significant experimental results that motivated the computational models.