Orthogonal Representations of Reward Magnitude, Certainty, and Volatility in the Macaque Orbitofrontal Cortex

Abstract

Categorical knowledge about the probabilistic and volatile nature of resource availability can improve foraging strategies, yet we have little understanding of how the brain represents such knowledge. Neurons in the orbitofrontal cortex (OFC) of macaques encode several decision variables (e.g., reward magnitude, probability) that could influence choice behavior. Here we investigated whether OFC neurons also represent two aspects of reward predictability: certainty and volatility. Rhesus monkeys performed a visual stimulus-reward association task in which a set of simple shapes preceded the delivery of reward, and they learned the nature of each shapes reward association along two dimensions. One involved the certainty of a reward outcome; rewards can be either deterministic (and therefore certain) or probabilistic (uncertain). A second dimension reflected the volatility of an outcome; reward schedules can be either stable over time or volatile. During stimulus presentation, the activity of OFC neurons reflected both categorical certainty and categorical volatility, in addition to reward magnitude. These three characteristics were represented orthogonally by three distinct neural populations of similar size. These findings point to a more general role for OFC in processing reward information than one restricted to encoding parametric valuations such as reward magnitude and probability.