Computational principles of value coding in the brain

The notion of value is central to theoretical and empirical approaches to decision-making. In psychological and economic choice theory, value functions quantify the relationship between relevant decision information and choice behavior. Evidence for value coding in neural circuits suggests that value information is explicitly represented in brain activity and plays a critical role in the neurobiological choice process. Here, we review a research approach centered on the computations underlying neural value coding. As in sensation and perception, neural information processing in valuation and choice relies on core computational principles including contextual modulation and divisive gain control. The form of these computations reveals details about the nature of decision-related value information and the constraints inherent in computing with biological systems. Understanding value representation at the intermediate level of computation promises insight into decision-making at the level of both the underlying circuit architecture and the resulting choice behavior.