Combined EEG and immersive virtual reality unveil dopaminergic modulation of error monitoring in Parkinson's Disease

Abstract

Detecting errors in your own and others' actions is associated with discrepancies between intended and expected outcomes. The processing of salient events is associated with dopamine release, the balance of which is altered in Parkinson's disease (PD). Errors in observed actions trigger various electrocortical indices (e.g. mid-frontal theta, error-related delta, and error positivity [oPe]). However, the impact of dopamine depletion to observed errors in the same individual remains unclear. Healthy controls (HCs) and PD patients observed ecological reach-to-grasp-a-glass actions performed by a virtual arm from a first-person perspective. PD patients were tested under their dopaminergic medication (on-condition) and after dopaminergic withdrawal (off-condition). Analyses of oPe, delta, and theta-power increases indicate that while the formers were elicited after incorrect vs. correct actions in all groups, the latter were observed in on-condition but altered in off-condition PD. Therefore, different EEG error signatures may index the activity of distinct mechanisms, and error-related theta power is selectively modulated by dopamine depletion. Our findings may facilitate discovering dopamine-related biomarkers for error-monitoring dysfunctions that may have crucial theoretical and clinical implications.

Fig. 1. Cluster-based permutations in the time domain for each group.

A Scalp representation of the cluster-based permutations (dependent sample t-tests with cluster-correction p < 0.05) of incorrect vs correct action extracted at two representative time points inside the window of interest. B Channel (y-axis) vs time (x-axis) representation of the cluster-based permutation for incorrect vs correct actions in the three groups; 0 ms corresponds to the avatar’s arm-path deviation, and the movements end at ~300 ms. For a larger visualisation of (B), please see https://osf.io/z9rbu/files/osfstorage (Figures section).

Fig. 2. Electrophysiological results in the time domain for each group (ERPs).

A Grand average oPe waveforms at electrode Pz. The end of the avatar’s movement is at 0 ms. Lighter colours denote the standard error of the mean. The light grey rectangle demarks the analysis interval window. B Graphical representation of the voltage distribution. Values represent incorrect minus correct actions.

Fig. 3. TF representation of relative power change (in %) with respect to baseline for incorrect and correct actions.

The end of the avatar’s arm-path deviation is at 0 ms. Incorrect and correct plots for electrode FCz between 1 and 50 Hz are shown for each group. Differential plots (erroneous − correct actions) are provided in the third column. The white rectangles demark the a priori chosen window of interest between 300–700 ms and 4–8.1 Hz, which were the values used for statistical analyses.

Fig. 4. Graphical representation of theta power (4–8.1 Hz) in the three groups on electrode FCz.

A Violin plots represent theta activity with correct and incorrect actions. Y-axes represent theta power expressed in relative power change (in %). Grey diamonds in the violin plots denote the mean value; black lines connect individual subject observations (i.e. black points) in the two conditions. B Graphical representation of voltage distribution. The values indicate the difference between incorrect and correct actions.

Fig. 5. Cluster-based permutation in the TF domain for each group.

Scalp representation of the cluster-based permutation (dependent sample t-test with cluster-correction at p < 0.05) of incorrect vs correct action. For each frequency and group, the topography shows the time-point at which the cluster reached the maximal spatial extension in the interval of interest. The intervals with a significant effect are shown below the topography. The bottom line shows cluster-based comparisons of the differential activity (incorrect minus correct) in the frequency bands of interest between HCs and off-condition PD; only intervals with significant clusters are shown.

Fig. 6. Schematic representation of the experimental paradigm and setup.

A Images (c) and (d) show the participant immersed in a virtual scenario in the CAVE system while observing the real-size virtual arm from a 1PP during a correct (a) or erroneous (b) grasping action. B The timeline of a single trial. The avatar’s action lasted ~1000 ms, with the reaching phase equal for both outcomes. The onset of the avatar’s arm-path deviation is at 0 ms, and the end of the avatar’s action occurs at 300 ms. The time windows for event-related potential (ERPs) and time-frequency (TF) analyses have been chosen a priori based on previous studies^,.