Coding of the basic components of subjective value in primate dopamine neurons: subjectively weighted reward amount and probability
Behavioral choices of uncertain rewards suggest that agents construct subjective reward value by combining the basic value components of utility and weighted probability. Despite the general acceptance of this evaluation mechanism for explaining economic choice, knowledge about its neuronal implementation is fractionated and remains essentially unknown. We investigated in monkeys whether reward signals in dopamine neurons might represent subjective reward value based on these two fundamental value components. Despite some heterogeneity across individual neurons, the dopamine population signal reliably represented the axiomatically defined integration of utility and weighted probability into subjective value in a way that closely matched the animal-specific choice behavior. In particular, we identified a crucial contribution of subjectively weighted probability to the dopamine signal of subjective reward value. These data demonstrate a neuronal implementation of subjective value constructed from the two most basic subjective reward components.