Modulation of impulsivity and reward sensitivity in intertemporal choice by striatal and midbrain dopamine synthesis in healthy adults

Abstract

Converging evidence links individual differences in mesolimbic and mesocortical dopamine (DA) to variation in the tendency to choose immediate rewards ("Now") over larger, delayed rewards ("Later"), or "Now bias." However, to date, no study of healthy young adults has evaluated the relationship between Now bias and DA with positron emission tomography (PET). Sixteen healthy adults (ages 24-34 yr; 50% women) completed a delay-discounting task that quantified aspects of intertemporal reward choice, including Now bias and reward magnitude sensitivity. Participants also underwent PET scanning with 6-[(18)F]fluoro-l-m-tyrosine (FMT), a radiotracer that measures DA synthesis capacity. Lower putamen FMT signal predicted elevated Now bias, a more rapidly declining discount rate with increasing delay time, and reduced willingness to accept low-interest-rate delayed rewards. In contrast, lower FMT signal in the midbrain predicted greater sensitivity to increasing magnitude of the Later reward. These data demonstrate that intertemporal reward choice in healthy humans varies with region-specific measures of DA processing, with regionally distinct associations with sensitivity to delay and to reward magnitude.

Keywords: delay discounting; immediate reward bias; impulsive choice; putamen; ventral tegmental area.

Fig. 1.

Illustration of delay-discounting paradigm. A: depiction of the 4 trial types. The 4 trial types included WANT, DON'T WANT, and 2 objective choice types: SOONER and LARGER. The trial ratio was 1:2 WANT trials and 1:6 each for the other 3 trial types. B: the temporal sequence of trial events is shown for an example WANT trial. Illumination of a fixation cross (“Ready”) marked each trial onset. An instruction cue informed the subject of the upcoming trial type. Two options were then presented while the Trial type cue remained on the screen. Options remained on the screen for 2 s, but subjects had 6 s to indicate their choice. Instruction cues for each trial type were depicted in a distinct color. ITI, intertrial interval.

Fig. 2.

Definition of regions of interest (ROIs). A: putamen ROIs. Left: the manually defined bilateral dorsal putamen ROIs superimposed on a high-resolution structural image for 1 participant in the study. Each participant's structural image was subsequently coregistered to a positron emission tomography (PET) image depicting 6-[¹⁸F]fluoro-l-m-tyrosine (FMT) uptake, and binding potential (K_i) values (relative to the cerebellum) were extracted from the ROIs. Center: FMT uptake in a PET image at the level of the dorsal putamen in a Low putamen FMT subject. Right: putamen FMT uptake in a High putamen FMT subject. Both PET images are shown overlaid on the individual structural images. B: midbrain ROI. Left: the manually defined bilateral midbrain ROI superimposed on a high-resolution structural image of 1 study participant. Center and right: FMT uptake in example PET images overlaid on the individual structural image, from which binding potential values (relative to the cerebellum) were extracted within the ROI: low midbrain FMT subject (center) and High midbrain FMT subject (right).

Fig. 3.

Now bias is elevated in those with lower putamen FMT. A: covariate-adjusted impulsive choice ratio (ICR) in participants with below-median FMT signal in bilateral putamen (LO-FMT putamen) or above-median FMT signal in bilateral putamen (HI-FMT putamen). ICR differed significantly between groups [F_(1,12) = 5.27, *P = 0.041]. B: covariate-adjusted ICR as a function of delayed reward time in the Low putamen FMT group and High putamen FMT group. C: covariate-adjusted ICR as a function of delayed reward amount in the Low putamen FMT group and High putamen FMT group. Values reflect estimated marginal means ± SE. Lines represent logarithmic fit of the group averaged data, with regression terms shown for each group.

Fig. 4.

Lower putamen FMT signal is associated with substantially elevated criterion interest rate threshold: covariate-adjusted criterion interest rate acceptance threshold in participants with below-median FMT signal in bilateral putamen (LO-FMT putamen) and above-median FMT signal in bilateral putamen (HI-FMT putamen). The criterion interest rate significantly differed between groups [F_(1,12) = 12.44, *P = 0.004]. Conventions as for Fig. 3.

Fig. 5.

Lower putamen FMT signal is specifically associated with more impulsive intertemporal choice, not greater inconsistency in intertemporal choices. A: covariate-adjusted log₁₀k_q (impulsivity) by putamen FMT group. Less negative log₁₀k_q values indicate more impulsive intertemporal choice at delay D = 0. Log₁₀k_q values differed significantly between groups [F_(1,9) = 10.59, *P = 0.01]. B: covariate-adjusted q (consistency). Less negative q values indicate more consistent intertemporal choice across delay times. The average q value did not differ significantly between groups [F_(1,9) = 1.93, P = 0.198]. Values reflect estimated marginal means ± SE. Conventions as for Fig. 3.

Fig. 6.

Decreasing impatience (DI_q) as a function of delay time. DI_q is elevated in the Low putamen FMT group relative to the High putamen FMT group [F_(1,9) = 8.71, P = 0.016], and while DI_q declines as a function of delay in both groups, DI_q declines more steeply in the Low putamen FMT group, indicating more irrational intertemporal choice. Conventions as for Fig. 3.

Fig. 7.

Steeper reward magnitude discounting associated with Low midbrain FMT signal. A: Now bias does not differ in those with lower midbrain FMT. Plot depicts covariate-adjusted ICR in participants with below-median FMT signal in the midbrain (LO-FMT midbrain) and above-median FMT signal in the midbrain (HI-FMT midbrain). B: no significant differences are found for covariate-adjusted ICR as a function of delayed reward time in the Low midbrain FMT group vs. the High midbrain FMT group. C: semilog plots of covariate-adjusted ICR as a function of Later amount in the Low midbrain FMT group and the High midbrain FMT group. The average slope of the magnitude discounting curves was significantly greater in the Low midbrain FMT group [F_(1,11) = 5.75, P = 0.035]. Values reflect estimated marginal means ± SE. Lines represent logarithmic fit of the group averaged data, with regression terms shown for each group.