A fundamental feature of neurobiological systems is adaptive coding, where neural activity reflects the recent history of information processing. Such adaptation is thought to be critical to efficient coding by neural systems under finite capacity constraints and occurs prominently in sensory circuits, where neurons adjust their responses to characteristics of the recent stimulus distribution. Importantly, recent evidence suggests that adaptive coding extends to reward processing and decision-making circuits. We are currently examining how a temporal form of divisive normalization can mediate adaptive value coding and the implications of this computation for both coding efficiency and choice behavior.