Integrating reward information for prospective behaviour

Abstract

Decision-making is often studied in a static context, such as deciding which option to select from those currently available. However, in everyday life we often also need to decide when to select an option to maximise reward. One possibility is that people track the latent reward of an option, updating expected changes in its value over time, to achieve appropriate selection timing. Contrary to this hypothesis, our electroencephalographic pattern analyses revealed that option properties like starting value and growth rate were translated into an estimate of when an option would become most valuable, far in advance of selecting it. The option's latent reward could not be decoded independently from neural activity. These results suggest that decisions to exploit individual options with lawful reward trajectories can be made by transforming reward information into an estimate of optimal timing, rather than actively monitoring an option's changing reward prospect.