Reliable population signal of subjective economic value from unreliable neurons in primate orbitofrontal cortex

Abstract:

ABSTRACT Behavior-related neuronal signals often vary between neurons. Despite the unreliability of individual neurons, brains are able to accurately represent and drive behavior. The notion may also apply to economic (‘value-based’) choices and the underlying reward signals. Reward value is subjective and can be defined by nonlinear weighting of magnitude (utility) and probability. Using a wide variety of reward magnitude and probability, we assessed subjective reward value at choice indifference between safe and risky rewards as prescribed by the continuity axiom that provides stringent criteria for meaningful choice. We found that individual neurons in the orbitofrontal cortex (OFC) of monkeys carry unreliable and heterogeneous neuronal signals for subjective value that largely fails to match the animal’s choice. However, the averaged neuronal signals matched well the animals’ choices, suggesting reliable subjective economic value encoding by the observed population of unreliable neurons. Highlights Different from widely held views, reliable neuronal information processing may not require reliable processors. Neurons in monkey orbitofrontal cortex (OFC) process reward magnitude and probability heterogeneously and unreliably. Despite unreliable neuronal processing, OFC population activity codes choices reliably. Reliability systems performance from unreliable elements seems to be a broad feature of neuronal reward coding. In brief Using stringent concepts of behavioral choice, Ferrari-Toniolo and Schultz describe unreliable individual reward neurons in monkey orbitofrontal cortex whose activity combines to a reliable population code for economic choice.