For the roughly one million Americans living with Parkinson’s disease, levodopa is the closest thing to an indispensable drug. Nothing else matches its ability to restore fluid, voluntary movement. But the longer patients take it, the more likely they are to develop a cruel tradeoff: involuntary twitching, writhing, and jerking known as levodopa-induced dyskinesia, or LID. Within a decade of starting the medication, as many as 80 percent of patients experience some form of it.
A study published in May 2026 in Science Advances now offers a sharper explanation of why that happens and, potentially, how to stop it. A team led by neuroscientist D. James Surmeier at Northwestern University traced the problem to a specific pattern of “aberrant learning” in the striatum, the brain region that coordinates movement. Their experiments in mice showed that the same dopamine surge meant to help patients move normally can, under certain conditions, rewire neurons to produce uncontrollable movements instead. Critically, the researchers demonstrated they could block that rewiring without sacrificing levodopa’s therapeutic benefit.
How the striatum learns the wrong lesson
The striatum relies on two populations of spiny projection neurons (SPNs) that act like competing throttles for movement: one set (the direct pathway) promotes action, while the other (the indirect pathway) suppresses it. In a healthy brain, dopamine keeps these pathways balanced. Parkinson’s disease strips away dopamine-producing cells, collapsing that balance and making voluntary movement difficult.
Levodopa restores dopamine and, with it, the ability to move. But the Northwestern team found that the drug’s effect depends heavily on the precise state of those spiny projection neurons at the moment dopamine floods in. When neurons are in a particular vulnerable configuration, the dopamine surge does not just restore normal signaling. It trains the circuits to fire in patterns that produce involuntary movement. Over repeated doses, this aberrant learning becomes entrenched, and dyskinesia emerges.
The key experimental advance: using cell-type-specific tools, the researchers shifted neuronal activity out of that vulnerable state during levodopa dosing. Mice that received this intervention retained improved motor function but did not develop the characteristic dyskinetic twisting and jerking. In effect, the team decoupled the drug’s benefit from its most disabling side effect.
This work builds on a body of research linking abnormal striatal activity to LID. A 2019 review cataloging aberrant activity in the parkinsonian striatum summarized competing models of how dyskinesia develops, including imbalances between the direct and indirect pathways, altered firing patterns, and changes in gene expression. The Northwestern paper moves beyond correlation to demonstrate a causal, state-dependent mechanism, a meaningful step forward in a field that has long debated which circuit-level changes actually drive LID.
Why current treatments fall short
The clinical urgency behind this research is hard to overstate. Levodopa has been the gold-standard Parkinson’s therapy for more than 50 years, and no replacement is on the horizon. Yet the only widely approved medication specifically targeting LID is amantadine, available in an extended-release formulation (marketed as Gocovri) that the FDA approved for this indication. Its effects are modest and often fade over time.
A 2025 review of LID treatment strategies published in Frontiers in Aging Neuroscience documents the breadth of the unmet need. Deep brain stimulation (DBS) can reduce dyskinesia in some patients, but it requires surgery, careful programming, and is not suitable for everyone. Experimental compounds in development target various receptor systems, but none directly address the state-dependent learning mechanism the Northwestern group described. The gap between what patients need and what medicine can currently offer remains wide.
What the study does not yet answer
The research was conducted entirely in mice. No human clinical trial data exist to confirm whether the same state-dependent mechanism operates in people, and the human striatum is considerably more complex than its rodent counterpart. Translating circuit-level findings from animal models to patients has historically been a slow, uncertain process. The paper does not propose a specific drug or device that could target these circuits in a clinical setting.
Competing mechanistic accounts also complicate the picture. Some researchers emphasize changes in thalamostriatal connections and broader basal-ganglia networks rather than focusing primarily on spiny projection neurons within the striatum. The Northwestern paper does not rule out contributions from those other circuits; it argues that the striatal mechanism is a necessary link in the chain. Resolving how these different circuit-level accounts fit together will require experiments that can track and manipulate multiple brain regions simultaneously.
There is also the question of reversibility. The concept of aberrant learning implies that once maladaptive patterns are established, simply normalizing dopamine levels may not undo them. The mouse experiments suggest that preventing the vulnerable neuronal state during levodopa dosing can block dyskinesia from forming in the first place. Whether similar interventions could unlearn or overwrite entrenched LID circuits in patients who have taken levodopa for years is unknown. Answering that question will require longitudinal studies and, eventually, early-phase clinical trials.
What this means for patients and caregivers
Nothing in this study should prompt anyone to change their medication. Levodopa remains the most effective treatment for Parkinson’s motor symptoms, and the risks of stopping or adjusting it without medical guidance far outweigh any speculative benefit from a mouse study. Patients experiencing dyskinesia should discuss their concerns with a neurologist or movement disorder specialist, who can explain how findings like these fit into the broader evidence base and whether any related clinical trials are recruiting.
The practical significance of the Northwestern work is narrower but real: scientists now have a more precise target for intervention. Instead of trying to dampen dyskinesia after it appears, future therapies might aim to prevent the aberrant learning process that creates it. That precision could eventually shorten the trial-and-error cycle that defines current dyskinesia management, whether through new drugs, refined DBS protocols, or approaches not yet imagined. The path from a mouse experiment to a clinic is long, but it starts with knowing exactly which circuits to aim for. This study brings that target into considerably sharper focus.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
