Two factors that influence how valuable a reward seems to be, that is, the subjective value of the reward, are the reward size and the amount of time that passes before the reward's delivery. Larger rewards are seen as more valuable than smaller ones, and rewards that are delivered immediately are more valuable that rewards delivered after a delay. After bilateral damage to the ventral striatum both rodents and primates appear to lose motivation, perhaps being unable to sense the value of a reward. We studied whether ventral striatal neurons encode information about the two factors, the reward size and the delay to receiving a reward, or whether the neurons might encode the subjective value itself. Encoding the subjective value would be seen as a neuron firing the same way to different combinations of the size and delay that are of equivalent value. Thus, if the subjective value is encoded we expect to see similar neuronal activity when a large reward is delivered after long delay and when a small reward is delivered after a short delay. We recorded neuronal responses in the ventral striatum from two monkeys while they performed a task in which we offered 9 combinations of reward by mixing 3 sizes (2, 4 or 6 drops of water) and 3 delays (1, 5 or 10s). A cue indicating the combination being offered was presented throughout the trial. The probability of accepting the offered reward was highest for the largest reward with the shortest delay, and became progressively smaller as the reward became smaller and/or the delay became longer. The behavioral performance for large reward-long delay trials was similar to that for small reward-short delay trials showing that the monkey was sensitive to the subjective value as suggested above. The probability of accepting was modeled using a logistic regression model with two continuous explanatory variables, reward size and delay. We analyzed the firing rate of the ventral striatum neurons in the initial period after the cue appeared on the screen, but before any action was required. In this task the firing of individual ventral striatal neurons was related to reward size, or delay, and in a few cases both. Overall, the individual neurons seemed to be coding for reward size and/or delay, but very few coded for the subjective value itself. Given these results, we asked whether the coding of the reward size and the delay has an aspect in common across all of the recorded ventral striatal neurons. To determine whether the coding across neurons was consistent we considered all of the neurons as a single group in a modern statistical model, a generalized linear mixed-effects model (GLMM), with neurons and the number of trials as random factors, and linear and quadratic terms of reward size and delay as fixed factors. The neural responses were well predicted by the GLMM when both linear and quadratic terms were included for the fixed effects for both monkeys. Thus, it appears that the population of neurons has strong encoding for the reward size and delay to reward. If the encoding of reward size and delay to reward in the ventral striatum is important for the monkeys behavior, inactivation of the ventral striatum should affect the manner in which the monkeys value the trials. We carried out two types of temporary inactivations. First we used a conventional neuronal silencing agent, muscimol, to inactivate the ventral striatum following unilateral injection, Second, we used a new chemogenetic technique, a Designer Receptor Exclusively Activated by a Designer Drug (DREADD), that we have developed for monkeys, to achieve neuronal silencing. The DREADD is a gene carried into neurons using specialized viral constructs. When the DREADD receptor is expressed about 6 weeks after being introduced, it can be activated by systemic injection of a drug, Clozapine N-oxide (CNO). For the amount of time the CNO is circulating the neurons expressing the gene are silenced. With both types of silencing, muscimol and DREADD, the monkey performed a stimulus-reward association task in which stimuli representing both reward amounts and delay-to-reward times were varied. In each trial, the monkey either accepted or rejected the offer presented, choosing to receive the reward represented by the stimulus after the proposed wait time, or choosing to immediately move on to a new trial with the potential for a new stimulus and different outcome.. On muscimol treatment days, the monkey exhibited significantly more early errors, releasing the lever after the presentation of the stimulus but prior to the presentation of the accept or reject cues.There was a main effect (ANOVA; p < 0.05) of treatment and delay-to-reward, and an interaction between treatment and delay-to-reward. The distribution of early errors by stimulus type was affected by muscimol-induced striatal inactivation; a greater proportion of early errors occurred with longer delays-to-reward when the striatum was inactivated, compared with a more even distribution of early errors among the delays-to-reward during control sessions. These results reflect an increased sensitivity to delay-to-reward times following muscimol-induced unilateral ventral striatal inactivation, possibly produced by a decreased sensitivity to reward size. In the second phase of the study, hM4Di DREADDs were targeted to the ventral striatum, and systemic CNO injection (10 mg/kg) was used to unilaterally inactivate the ventral striatum. On CNO treatment days, the monkey exhibited significantly more early errors, consistent with the effects of muscimol inactivation. No interaction of CNO treatment and delay-to-reward was observed. Our current interpretation is that the DREADDs as used here yield a weaker inactivation of the ventral striatum than is seen with the muscimol-induced inactivation. Both results show that even a unilateral inactivation of the ventral striatum interferes with the monkeys ability to wait for a delayed imperative signal. We are working to try to connect the single neuron results with the behavioral findings.
Showing the most recent 10 out of 18 publications
