Reinforcement learning in Parkinson’s disease is not associated with inflammatory tone

Parkinson’s disease (PD) is associated with large variability in the development and severity of both motor and nonmotor symptoms, including depression and impulse control disorder. Neuroinflammation might contribute to this heterogeneity, given its association with dopaminergic signalling, neuropsychiatric symptoms, and reward versus punishment learning . Here, we assessed the effect of inflammatory tone on probabilistic reinforcement learning and impulse control disorders in PD. We measured computational learning model-based neural reward prediction error and expected value signals in frontostriatal circuity during reinforcement learning using functional MRI. In addition, we acquired cerebral spinal fluid of 74 PD patients and screened for 13 inflammatory factors, including our primary marker of interest IL-6, previously implicated in reward learning signaling in the ventral striatum. In contrast to our prediction, we found no association between inflammatory tone and any of the behavioural or neural reinforcement learning parameters. Furthermore, we did we not find any correlation between inflammatory tone and depressive or impulsive PD symptoms. Exploratory analyses revealed a negative association between MCP-1 and reward prediction error signals in the ventral striatum, an observation that should be replicated in future work. The null findings might reflect the fact that measurements were taken ON medication, or that our sample consists of an early disease stage cohort that may be too small to detect these effects, or that IL-6 is a suboptimal marker for inflammatory tone, or a combination of these factors.

In this collection, you can find the final codebase used for this paper. The research data is available on request only, to ensure the privacy of the participants. A data acquisition request can be sent to [e-mail].