For Juice or Money: The Neural Response to Intertemporal Choice of Primary and Secondary Rewards

Studies of reward processing in humans often focus on secondary rewards or reinforcers such as money. In contrast, studies in nonhuman primates typically focus on primary rewards such as food or juice. Both approaches implicate the mesolimbic dopamine system and associated structures (termed the “limbic reward system”) for the processing of primary and secondary rewards (Schultz, 2002; Ramnani et al., 2004). However, the temporal discounting functions of primary and secondary rewards remains an area open for investigation. Delay discounting paradigms are used to measure the value of delayed rewards relative to immediate rewards, and can be used to examine how the temporal context of reward delivery affects decision making in both human and animal models (Ainslie, 1975; Richards et al., 1999). In their recent study in the Journal of Neuroscience, McClure et al. (2007) used a delay discounting paradigm to investigate the neural and behavioral response to the delivery of early and delayed fluids in thirsty subjects. The authors compared their results to a previous study using a similar paradigm with monetary rewards.

McClure et al. (2007) designed two experiments using a delay discounting paradigm to determine the neural and behavioral responses to intertemporal choices between fluid deliveries. Subjects were instructed to refrain from drinking 3 h before the functional magnetic resonance imaging session to ensure that they were thirsty. On each trial, subjects had a choice between smaller amounts of water or juice at an early time point, or greater amounts of fluid at a later time point. In experiment 1, the time difference between early or late reward was either 1 or 5 min, and the time of the earliest fluid delivery was 0, 10, or 20 min. In experiment 2, delays to fluid delivery were divided into three categories: early (10 min), middle (20 min), and late (30 min).

In experiment 1, subjects demonstrated nonexponential delay discounting in their behavioral choices to certain temporal conditions. In other words, for primary rewards, subjects were more likely to select the early reward option in the immediate (0 min) condition than in either of the delayed conditions. This effect appeared to be driven primarily by trials with 5 min differences between immediate and delayed fluid delivery rather than the trials with 1 min differences. When the earliest possible fluid delivery was delayed by 10 min in experiment 2, subjects did not treat the “early” fluid delivery as being immediate, and were equally likely to chose the lesser reward amount when it was available after a 10 min delay or after a 20 min delay. Thus, for primary rewards, the subjects only showed delay discounting for delivery of fluid rewards in <10 min. The authors hypothesize that, for primary rewards which fulfill appetitive needs, the temporal characteristics of the “early” and “delayed” reward options may be more immediate and temporally stable, whereas the temporal characteristics of secondary rewards may be more variable and context dependent.

Similar to their previous study which used gift certificates as secondary rewards (McClure et al., 2004), the “limbic reward system,” including the ventral striatum, medial prefrontal cortex, and orbitofrontal cortex, showed greater activity when an intertemporal choice included an immediate reward than if choices were delayed. However, there was little voxel-level replication of brain regions that were more active to immediate reward choices for primary and secondary rewards. This suggests that limbic reward areas process both primary and secondary rewards, but reward-related brain regions may be stimulus specific. In parallel with the behavioral results, when the earliest available fluid delivery was delayed to 10 min, the “limbic reward system” did not respond differentially to early versus delayed fluid delivery. Other regions of the brain that are thought to be involved in executive control such as the lateral prefrontal cortex and posterior parietal cortex responded similarly to intertemporal choices, regardless of whether an immediate reward choice was present, or whether the experiment focused on primary or secondary rewards. This result suggests that areas related to executive control may be important for evaluating intertemporal choices regardless of the delay interval or stimulus type.

This study suggests that, although limbic reward-related brain areas process both primary rewards (fluids) and secondary rewards (money), these areas are influenced by stimulus type and temporal characteristics. This could reflect stimulus specific processing of early reward choices. However, unlike secondary rewards, fluid rewards allow for precise control of the timing between delivery and reward. Thus, the possibility also exists that the differences in reward related activation arise from temporal characteristics of the stimulus delivery between primary and secondary rewards, rather than a function of the stimulus type itself. The authors did not include a regressor that was specifically associated with only those trials involving delayed reward in the later delay conditions. This weakens the argument that certain areas of the limbic reward system show temporal discounting effects only for trials involving immediate reward, because distinctions between successive nonimmediate delays were not explicitly modeled.

McClure et al. (2007) hypothesize that the neural response to intertemporal choice of early primary rewards such as juice is similar to that of secondary rewards such as money, although this is dependent on the behavioral observation of delay discounting. They hypothesize that primary rewards are less susceptible to contextual framing than secondary rewards and, therefore, have more rapid, stable temporal characteristics. The investigation of primary reward discounting may be less susceptible to contextual factors, and more related to internal states including satiety and temperature. Importantly, investigations of primary rewards in humans allows for more direct comparison of reward-related experiments in animal models. Further understanding the relationship between reward-related processing of primary and secondary rewards could provide a mechanism for understanding how evolutionarily evolved reward systems for physiological needs such as thirst have adapted to process more abstract secondary rewards such as money.