Dissociable neural representations of future reward magnitude and delay during temporal discounting

Abstract

In temporal discounting, individuals often prefer smaller immediate rewards to larger delayed rewards, implying a trade off between the magnitude and delay of future rewards. While recent functional magnetic resonance imaging (fMRI) investigations of temporal discounting have generated conflicting findings, no studies have focused on whether distinct neural substrates respond to the magnitude and delay of future rewards. Combining a novel, temporally distributed discounting task with event-related fMRI, we found that while nucleus accumbens (NAcc), mesial prefrontal cortical (MPFC), and posterior cingulate cortical (PCC) activation positively correlated with future reward magnitude, dorsolateral prefrontal cortical (DLPFC) and posterior parietal cortical (PPC) activation negatively correlated with future reward delay. Further, more impulsive individuals showed diminished NAcc activation to the magnitude of future rewards and greater deactivations to delays of future rewards in the MPFC, DLPFC, and PPC. These findings suggest that while mesolimbic dopamine projection regions show greater sensitivity to the magnitude of future rewards, lateral cortical regions show greater (negative) sensitivity to the delay of future rewards, potentially reconciling different neural accounts of temporal discounting.

Figure 1. Task Design

The temporal discounting task staggered presentation of information for future reward magnitude, future reward delay, and choice selection, providing an opportunity to isolate brain activity related to each. The first screen (2 s) presented the immediate option, which always offered a reward magnitude of $10.00 at a delay of 0 days. Next, the reward magnitude of the future option was presented (2 s), which varied between seven different amounts (ranging from $10.00-$25.00). Next, the delay of the future option was presented (2 s), which varied between 6 delays (ranging from 0 - 180 days). The last screen solicited subjects’ choice for either the immediate or future option (2 s). Each trial ended with a 2-6 s variable inter-trial interval (ITI). Trial blocks included 42 pseudorandomized trial types such that every combination of magnitude and delay was presented once, and subjects played two blocks, yielding a total of 84 trials.

Figure 2. Brain activation correlating with magnitude or delay

(a.) Brain regions significantly correlating with the magnitude of the future option (right NAcc, MPFC, PCC). (b.) Brain regions significantly correlating with the delay of the future option (left DLPFC, left TPJ, right posterior parietal cortex). (c.) Brain regions significantly correlating with the interaction of magnitude and delay (right IFG). For all images warm colors indicate positive correlations, cool colors indicate negative correlations; p<0.001, uncorrected; p<0.05 corrected.

Figure 3. Timecourses of activation from representative magnitude-sensitive (right NAcc) and delay-sensitive (DLPFC) regions (columns)

Row 1 shows timecourses of % signal change binned by high, medium, and low magnitudes, where magnitude-sensitive regions like the NAcc clearly respond to different levels of magnitude during the magnitude period, while delay-sensitive regions like the DLPFC do not. Row 2 shows timecourses of % signal change binned by high, medium, and low delays, where delay-sensitive regions like the DLPFC clearly respond to different levels of delay during the delay period, while magnitude-sensitive regions like the NAcc do not. Timecourses from all VOIs are reported in supplementary materials (Supplementary Figure 5).

Figure 4. Individual differences in neural sensitivity to future reward magnitudes and delays predict individuals’ temporal discounting rates

(a.) Neural response to magnitudes in the magnitude-sensitive NAcc negatively correlates with discounting rates (k, normalized by square-root transform). More “impulsive” individuals show less NAcc activation to large magnitudes of future rewards. (b.) Neural response to delays in delay-sensitive regions (DLPFC, posterior parietal cortex) correlates with discounting rates (k, normalized by square-root transform). More “impulsive” individuals show more DLPFC and PPC deactivations to long delays of future rewards. (c.) Neural response to delays in magnitude-sensitive regions (MPFC, PCC) also correlates with discounting rates (k, normalized by square-root transform). More “impulsive” individuals show MPFC and PCC deactivations to long delays, while more “patient” subjects do not show MPFC or PCC deactivations with delays.