This paper challenges the traditional view of temporal discounting as an intrinsic preference for immediate rewards. Instead, it proposes that discounting arises from internal uncertainty in value representations of future events, where noisy mental simulations are rationally combined with prior beliefs. The authors extend this idea by incorporating adaptive modulation of simulation noise based on task demands, using principles of rational inattention. They demonstrate how optimal allocation of mental effort can explain the magnitude effect in intertemporal choice, supported by re-analyses of existing datasets and a new experiment.

Samuel J. Gershman, Rahul Bhui