Value encoding in the globus pallidus: fMRI reveals an interaction effect between reward and dopamine drive

Abstract

The external part of the globus pallidus (GPe) is a core nucleus of the basal ganglia (BG) whose activity is disrupted under conditions of low dopamine release, as in Parkinson's disease. Current models assume decreased dopamine release in the dorsal striatum results in deactivation of dorsal GPe, which in turn affects motor expression via a regulatory effect on other nuclei of the BG. However, recent studies in healthy and pathological animal models have reported neural dynamics that do not match with this view of the GPe as a relay in the BG circuit. Thus, the computational role of the GPe in the BG is still to be determined. We previously proposed a neural model that revisits the functions of the nuclei of the BG, and this model predicts that GPe encodes values which are amplified under a condition of low striatal dopaminergic drive. To test this prediction, we used an fMRI paradigm involving a within-subject placebo-controlled design, using the dopamine antagonist risperidone, wherein healthy volunteers performed a motor selection and maintenance task under low and high reward conditions. ROI-based fMRI analysis revealed an interaction between reward and dopamine drive manipulations, with increased BOLD activity in GPe in a high compared to low reward condition, and under risperidone compared to placebo. These results confirm the core prediction of our computational model, and provide a new perspective on neural dynamics in the BG and their effects on motor selection and cognitive disorders.

Keywords: Basal ganglia; Dopamine; Globus pallidus; Indirect pathway; Parkinson's disease.

Fig. 1

Neural model and simulated neural dynamics. A) Schematic representation of the anatomic neural areas discussed in the paper (left) and graph illustrating the simplified neural architecture used to simulate BG circuit dynamics (right). Four simulated sensorimotor or action-state values (B), putatively encoded in cortical signals reaching the striatum, are constant during the whole simulation. Tonic DA release targeting the striatum (C) slowly increases during the simulation, affecting the way the input is encoded in the striatum. Striatal D1-enriched units (D) amplify the differences among the inputs as a direct function of DA release. Conversely, striatal D2-enriched units (E) show the higher differentiation under low DA release, as the input signals are compressed towards the end of the simulated time, in association with high DA release. GPe (F) and GPi/SNr (G) receive the input after it is processed by D2 and D1 enriched striatum, respectively. Due to the inhibitory afferent connections, the GPe mirrors the signal received from the D2 striatum. Finally, GPi/SNr receive conflicting inhibitory information from D1 striatum (direct pathway) and GPe (indirect pathway), resulting in the compression of signal differences, at low dopamine release. The key prediction of this computational hypothesis is further illustrated in the heatmaps for each BG nucleus, where we represent the differences in simulated neural activity between the encoding of high and low action-state values. This difference is tested in our within-subject fMRI paradigm, as the model predicted increased High-Low differentiation in the GPe under a condition characterised by low dopaminergic release, in comparison with basal or high dopaminergic conditions.

Fig. 2

Schematic representation of the computer task. A fixation cross is followed by a symbol indicating the starting trial is characterised by either high reward (“gold coins”) or low reward (“iron coins”), with equal probabilities. Each condition is followed by the motor part of the task, where a red and green scale-like image is presented to the participant on a black screen. This indicates the moment the “grip action” has to be performed, applying a sustained force on the apparatus, with the right hand, until the end of the trial, for 2–4 s. The force applied by a participant is constantly recorded and reported on screen by means of a horizontal blue line that moves vertically on the scale. The participants collect points proportionally to the reward of the trial, per each second spent applying the correct force, i.e. having the feedback blue line positioned in the “green zone” of the scale.

Fig. 3

Boxplot representation of the distribution of reaction times under the conditions of high and low reward (HR and LR, respectively), combined with the DA drive condition of placebo (pla) or risperidone (ris). Participant 5 is highlighted as outlier in terms of RTs recorded under the condition of risperidone.

Fig. 4

Whole brain activity reported with a threshold of p < .005 and a 50 voxel extent. (A) The first contrast (coordinates for the image: 22, 24, −6), between placebo and risperidone condition, shows BOLD activity in the putamen (bilateral), caudate (right hemisphere) and orbitofrontal cortex (right hemisphere). (B) The second contrast (coordinates for the image: −2, 56, 10), between high and low reward presentation, reveals BOLD activity in the prefrontal cortex.

Fig. 5

Illustration of the maps used for the globus pallidus ROIs in the left hemisphere (coordinates for the image: −22, −4, 0) and extracted beta values for the left dorsal GPe. In the map, dorsal and ventral GPe are highlighted in red and yellow, respectively, whereas dorsal and ventral GPi are highlighted in blue and cyan, respectively. Extracted values are reported as bars with mean and standard error for the four conditions characterising the experimental design: high vs low reward (HR - LR) and placebo vs risperidone (pla - ris). Single data points are reported (orange) per each condition, linking values extracted under HR and LR conditions. Repeated measures ANOVA shows a significant interaction effect (F = 6.53, p = .02), as the mean within subject difference changes from HR-LR = −0.13, under risperidone condition, to HR-LR = +0.14, under placebo condition. No main effect is reported for either variable. The beta values are reported after the exclusion of the behavioural outlier (participant 5).

Fig. 6

Illustration of the map used for the SNr ROI, right hemisphere (coordinates for the image: 11, −13, −15), and extracted beta values for this mask under the four conditions characterising the experimental design: high vs low reward (HR - LR) and placebo vs risperidone (pla - ris). Bars with mean and standard error are overlaid with single data points (orange) per each condition, where we have linked values extracted under HR and LR conditions. A significant interaction effect was found in the right SNr (F = 4.85, p = .04), which was caused by a change of direction in the mean of the within subject difference: HR-LR = +0.21, under risperidone condition, HR-LR = −0.13, under placebo condition. No main effect is reported for either variable. The beta values are reported after the exclusion of the behavioural outlier (participant 5). Two further outliers for the beta values (participants 7 and 11) are also marked with a red circle.