Action initiation shapes mesolimbic dopamine encoding of future rewards

Abstract

It is widely held that dopamine signaling encodes predictions of future rewards and such predictions are regularly used to drive behavior, but the relationship between these two is poorly defined. We found in rats that nucleus accumbens dopamine following a reward-predicting cue was attenuated unless movement was correctly initiated. Our results indicate that dopamine release in this region is contingent on correct action initiation and not just reward prediction.

Figure 1. Go/No-Go task in Experiment 1

Schematic of the behavioral task (left) and trial types (right). Gray shading marks the period when the auditory cues remain on.

Figure 2. NAcc dopamine signals on Go and No-Go trials in Experiment 1

(a,b) Example single-trial FCV recordings during a Go (a) and a No-Go (b) trial. Upper plots depict the dopamine level over time for the two trials types (L1 = first lever press). Inset, example cyclic voltammogram identifying the detected current as dopamine. Color plots in the lower panels show the background-subtracted cyclic voltammograms as a function of the applied voltage over time. (c,d) Average changes in dopamine levels (nA, mean ± s.e.m., n = 7 electrodes from 7 rats) recorded during different trial types aligned to cue onset (c) or to time of head exit from nose-poke (d), along with average discriminability between the Go and No-Go trial types for each time point (gray lines = maximum or minimum of 1,000 permuted sessions; line: *P < 0.05 permutation tests, corrected for multiple comparisons). Gray shading marks a 2.5-s window used to focus the analyses. As can be observed, the sustained increase in dopamine levels observed on Go trials after cue presentation was delayed on No-Go trials until after movement initiation. (e) Average absolute effect sizes (mean ± s.e.m.) from a general linear model with regressors for success rate and action initiation time. (line: *P < 0.05 permutation tests, corrected for multiple comparisons). Both factors were significant predictors of NAcc dopamine release during the cue period. a.u., arbitrary units.

Figure 3. Experiment 2: behavior and dopamine signals

(a) Schematic of the trial types. (b,d) Average changes in dopamine levels (mean ± s.e.m. n = 9 electrodes from 6 rats) recorded during different trial types aligned to cue onset (b) or to time of head exit from nose-poke (d). (c) Average absolute effect sizes (mean ± s.e.m.) from a general linear model with regressors for RPE (success rate × reward size), action initiation time and their interaction. (Line: *P < 0.05 permutation tests, corrected for multiple comparisons). a.u., arbitrary units.