Striatal and cerebellar interactions during reward-based motor performance

Abstract

Goal-directed motor performance relies on the brain's ability to distinguish between actions that lead to successful and unsuccessful outcomes. The basal ganglia (BG) and cerebellum (CBL) are integral for processing performance outcomes, yet their functional interactions remain underexplored. We scanned participants' brains with functional magnetic imaging (fMRI) while they performed a skilled motor task for monetary rewards, where outcomes depended on their motor performance and probabilistic events that were not contingent on their performance. Successful motor outcomes increased activity in the ventral striatum (VS), a functional subregion of the BG, whereas unsuccessful motor outcomes engaged the CBL. In contrast, for probabilistic outcomes unrelated to motor performance, the BG and CBL exhibited no differences in activity between successful and unsuccessful outcomes. Dynamic causal modeling revealed that VS-to-CBL connectivity was inhibitory following successful motor outcomes, suggesting that the VS may suppress CBL error processing for correct actions. Conversely, CBL-to-VS connectivity was inhibitory after unsuccessful motor outcomes, potentially preventing reinforcement of erroneous actions. Additionally, interindividual differences in task preference, assessed by having participants choose between performing the motor task or flipping a coin for monetary rewards, were related to inhibitory VS-CBL connectivity. These findings highlight a performance-mediated functional network between the VS and CBL, modulated by motivation and subjective preferences, that supports goal-directed behavior.

Keywords: cerebellum; fMRI; motor performance; ventral striatum.

Fig. 1.

Task design. (A) Reward-based motor task. At the beginning of a trial, participants were presented with the target size (Task Presentation) and the probability of the floor remaining unbroken (Floor Presentation). Participants then performed the motor task by guiding the red cursor through the target (passage in the wall) and landing on the other side (Motor Performance). If the cursor hit any part of the wall, the trial failed (Unsuccessful Motor Outcome, Upper panel). The probabilistic outcome was revealed after the successful completion of the motor task (Successful Motor Outcome, Lower panel). If the floor broke, the trial resulted in a failure, and the cursor fell off the environment (Unsuccessful Probabilistic Outcome, upper panel). If the floor remained unbroken, the trial resulted in a success (Successful Probabilistic Outcome, Lower panel). (B) Trajectory data for an exemplary participant across the three target sizes. Green lines indicate successful trials, and red lines indicate failed trials. (C) Group performance (n = 28) for the three difficulty levels in the motor task. Error bars represent SEM.

Fig. 2.

The VS and CBL encode motor performance. (A) Activity in the VS was significantly increased in successful compared to unsuccessful trials at the time of motor outcome [MNI Coordinate Peak = (−10, 6, −12), P < 0.05 FWE corrected]. Contrast is displayed at P < 0.001 uncorrected. (B) Activity in the CBL was significantly increased in unsuccessful compared to successful trials at the time of motor outcome in both anterior [MNI Coordinate Peak = (24, −36, −22), P < 0.05 _{FWE corrected}] and posterior [MNI Coordinate Peak = (−40, −70, −22), P < 0.05 _{FWE corrected}] regions. Contrast is displayed on a cerebellar flat map (91) at P < 0.001 _uncorrected. (C) Parameter estimates for contrasts of successful versus unsuccessful motor outcomes for the VS and CBL. Each point represents a single participant. The solid vertical lines represent the mean parameter estimate for the group. (D) VS and CBL activity at the time of motor outcome, for the three difficulty levels, across participants. We correlated the VS, and CBL parameter estimates for the Success – Failure contrast. There was a significant positive correlation between the VS and CBL activity for the hard (30%, Kendall’s τ = 0.33, p = 0.014) and medium (60%, Kendall’s τ = 0.42, p = 0.0015) target sizes. There was a non-significant positive trend observed for the easy target size (90%, Kendall’s τ = 0.17, p = 0.21). These results indicate that for individuals with greater VS activity following successful motor performance, the CBL exhibited reduced activity following failed motor performance.

Fig. 3.

Performance outcomes modulate VS-CBL connectivity. (A) Regions of interest and modulating conditions for the DCM analysis. Regions were the VS (red) and CBL (blue), and the modulating conditions were successful (MS, green) and unsuccessful (MF, orange) motor outcomes. VS activity was extracted using an a priori mask and the CBL activity was extracted using a leave-one-subject-out (LOSO) ROI based on the contrast in Fig. 2C. (B) DCM specification. We specified four potential models that varied according to how successful (MS) and unsuccessful (MF) motor outcomes modulated VS-CBL effective connectivity (black arrows). (C) A Bayesian model comparison showed that Model 3 was the winning model (Pp = 88%) for describing the commonalities (average connectivity) across participants’ VS-CBL connectivity. (D) Bayesian Model Average (BMA) of Parametric Empirical Bayes (PEB) parameters highlighting the impact of successful and failed motor outcomes on VS-CBL connectivity. The VS inhibits CBL following successful motor outcomes, whereas the CBL inhibits VS following unsuccessful motor outcomes. **indicates that PEB parameters have very strong positive Bayesian evidence (Pp > 0.99).

Fig. 4.

VS-CBL connectivity predicts participants’ preference for performing the motor task. (A) Prior to performing the reward-based motor task in the scanner, participants made a series of choices to either perform motor tasks of varying difficulty or accept potential probabilistic outcomes (weighted coin flip) for a potential monetary reward of $25. The target size associated with the motor task was sampled from the three target sizes used in the reward-based motor task (Fig. 1 B and C). The coin flip had a probability of success varying from 0.10 to 0.90. Participants’ subjective preference was computed as the mean acceptance rate of the motor task over the coin flip. (B) A Bayesian model comparison showed that Model 3 was the winning model (Pp = 85%) for describing the effect of participants’ subjective preference for performing the motor task on their VS-CBL connectivity. (C) BMA of PEB parameters illustrating the impact of participants’ subjective preference for performing the motor task on VS-CBL connectivity. Individuals with a greater preference to perform the task exhibited increased inhibitory connections from the VS to the CBL following successful motor outcomes and from the CBL to the VS following failed motor outcomes. **indicates that PEB parameters have very strong positive Bayesian evidence (Pp > 0.99). (D) Out-of-sample estimation of participants’ subjective preferences for performing the task using the degree of inhibition of the CBL by the VS following successful motor outcomes (PEB parameter, Left). (E) Out-of-sample estimation of participants’ subjective preference using the degree of inhibition of the VS by the CBL following failed motor outcomes (PEB parameter, Left).

Fig. 5.

Motor outcomes modulate VS-CBL connectivity for both the sensorimotor and cognitive CBL. (A) Regions of interest and modulating conditions for the DCM analysis were specified in a similar fashion as the original DCM analysis (Fig. 3A), with the exception that the CBL was divided into the sensorimotor (Mtr. CBL) and cognitive (Cog. CBL) subregions. The Mtr. CBL consisted of activations in lobules IV, V, and VI, whereas the Cog. CBL consisted of Crus I and II. (B) Candidate PEB models were specified according to the two factors. The open blue-yellow circles indicate that factor 1 (modulation by motor success/failure) did not consider the specificity of the Mtr. (yellow) or Cog. CBL (blue). The filled green-orange circle indicates Factor 2 (modulation of Mtr. or Cog. CBL) did not consider the modulation of regions due to task success (green) or failure (orange). (C) A Bayesian model comparison showed that models 1 (Pp = 44%) and 9 (Pp = 38%) best described the commonalities (average connectivity) across participants’ VS-Mtr. CBL and VS-Cog. CBL connectivity. (D) The VS inhibits Mtr. CBL and excites Cog. CBL following successful motor outcomes. The Mtr. and Cog. CBL inhibits VS following failed motor outcomes. **PEB parameters showing very strong positive Bayesian evidence (Pp > 0.99). *indicates that PEB parameters show positive Bayesian evidence (Pp > 0.73).