Dopamine’s role in learning may be broader than previously thought

A new study published in Nature Communications provides evidence that the brain chemical dopamine plays a sophisticated, dual role in how we learn, influencing both our fast, effortful thinking and our slower, habit-forming learning processes. The research indicates that a person’s natural dopamine levels may shape their reliance on mentally demanding strategies, while dopamine-boosting drugs can enhance the brain’s trial-and-error learning system.

To navigate the world, the brain uses at least two major learning systems that often work in concert. The first is reinforcement learning, a slow and steady process in which we gradually learn the value of actions based on their outcomes. Think of it like learning to ride a bike: through trial and error—getting positive feedback (staying upright) and negative feedback (wobbling or falling)—the brain gradually wires in the correct muscle movements until they become automatic. This type of learning tends to be robust but requires time and repetition.

The second system is working memory, which functions as the brain’s mental scratchpad. It allows us to instantly store and manipulate a small amount of information for a short period. For example, when someone tells you a phone number, you hold it in working memory just long enough to dial it. This system is incredibly fast and flexible, but it has limited capacity—you can’t hold dozens of numbers at once—and using it tends to require mental effort.

Both of these systems are thought to be heavily influenced by dopamine signaling in the striatum. Dopamine is a neurotransmitter, or chemical messenger, that plays a major role in motivation, reward, and movement. In the striatum, dopamine signals are believed to be essential for stamping in the lessons of reinforcement learning. At the same time, some theories suggest dopamine also helps manage working memory, possibly by reducing the mental effort it requires.

A key challenge for scientists has been to separate dopamine’s effects on reinforcement learning from its effects on working memory. If a person with higher dopamine levels learns faster, is it because their habit-forming system is more efficient, or because they’re better able to use working memory to bypass the slower process? Answering this question is important for understanding not only healthy cognition but also conditions like ADHD and schizophrenia, where dopamine signaling is believed to be disrupted.

To address this, a large international team of researchers from institutions in the Netherlands, the United States, and Sweden designed a study to isolate the contributions of these two systems. They aimed to determine how a person’s natural dopamine levels—and drugs that modify dopamine signaling—affect each system independently.

“I’ve always been interested in cognitive effort: why does thinking feel like work, why are some tasks (like working memory tasks) much more effortful than others, why are the same tasks effortless for some people, and why do I struggle to stay on task even when it is important to me to do so?” said study author Andrew Westbrook, an assistant professor of psychiatry at Rutgers University and head of the Brain Modulation & Control Lab.

“Striatal dopamine signaling has long been known to help rodents and other animals rise to the challenge when they need to exert physical effort to pursue rewards. In recent work, we built on the intuition that dopamine signaling could help shape policies about whether or not to exert cognitive effort too. We wanted to test, in part, whether drug effects and individual differences in striatal dopamine function might influence how much people rely on effortful working memory processes versus effortless reinforcement learning during an instrumental learning task.”

The researchers used a multi-faceted approach involving 100 healthy young adult participants. The study combined a specialized cognitive task with brain imaging and pharmacological interventions.

First, participants performed a task designed to pit reinforcement learning against working memory. They were shown a series of images and had to learn, through trial and error, which of three buttons corresponded to each image. The key manipulation was “set size”—the number of different images presented in a block of trials. In some blocks, there were only two images to remember (a low load on working memory). In others, there were up to five, making it much harder to keep everything straight and encouraging a shift toward the slower, more incremental reinforcement learning system.

Second, the scientists measured each participant’s baseline dopamine function. Using a brain imaging technique called positron emission tomography (PET), they measured each individual’s dopamine synthesis capacity—essentially, the rate at which their brain produces dopamine in the striatum. This provided a snapshot of each person’s natural dopamine profile.

Third, in three separate sessions, participants received either a placebo, a 20 mg dose of methylphenidate (a drug commonly known as Ritalin that boosts dopamine levels by blocking its reuptake), or a 400 mg dose of sulpiride (an antipsychotic drug that blocks a specific type of dopamine receptor known as D2). By comparing performance across these conditions, the researchers observed how boosting or dampening dopamine activity changed learning behavior.

Finally, the team used computational models to analyze participants’ choices. These models estimated hidden mental processes, such as the learning rate of the reinforcement learning system and the degree of reliance on the working memory system.

The results indicated that individuals with a higher capacity to produce dopamine tended to rely more heavily on working memory. Their performance was particularly strong in low-set-size blocks, where working memory is most effective. This suggests that a more robust dopamine system may bias a person toward faster, more flexible—but also more effortful—strategies.

The drug sulpiride had the opposite effect. Under its influence, participants’ performance declined. Modeling results suggest this was because sulpiride reduced their reliance on working memory and caused information held in memory to decay more quickly. It appears that interfering with this dopamine pathway makes working memory less reliable.

When examining the effects of methylphenidate, the researchers found a different pattern. While higher baseline dopamine was associated with stronger working memory use, methylphenidate specifically enhanced reinforcement learning. Participants on methylphenidate showed steeper learning curves, improving more with each correct response than they did on placebo.

The computational model confirmed this by showing that the drug increased the learning rate of the reinforcement learning system. This effect was strongest in individuals with high natural dopamine synthesis capacity, suggesting that the drug amplified existing learning signals. These findings indicate that dopamine not only supports fast learning through working memory but also directly boosts the slow, incremental process of habit formation.

An intriguing finding came from a surprise test phase at the end of the experiment. Participants were shown pairs of images from the learning task and asked to choose which one had been associated with more points. While participants generally performed well, the data revealed a subtle bias: people tended to devalue rewards earned during the more difficult, high-set-size blocks. It was as if the extra mental effort required to earn those rewards made them feel less satisfying.

Methylphenidate blunted this effect. When on the drug, participants were less likely to devalue rewards from harder tasks. This suggests that dopamine not only alters how we decide to expend effort but also how we learn about the cost of effort in the first place. A dopamine boost seems to make mental work feel less taxing.

The findings provide evidence “that striatal dopamine both biases us to rely more on effortful working memory when solving difficult problems and, on separate time scales, also influences how we learn about the cognitive effort required to perform those tasks in the first place,” Westbrook told PsyPost.

The computational modeling also revealed that once the contributions from working memory were accounted for, the reinforcement learning system contributed very little to rapid learning in this task. This suggests that, when faced with a new problem, the brain leans heavily on fast, effortful working memory, while reinforcement learning plays a more secondary role.

“One shocking outcome was how little RL-like processes contribute to task learning once you control for working memory-based contributions,” Westbrook explained. “Although we do find RL-like incremental learning that is moreover influenced by dopaminergic drugs, the effective learning rates were tiny once working memory was factored out. I think this has important implications for studies which aim to infer something about the link between dopamine and RL. Namely, you have to control for dopamine’s effects on working memory first before interpreting anything about RL.”

While the study offers a clearer view of dopamine’s dual role, some questions remain. The precise mechanism by which sulpiride impaired performance is still not fully resolved. Looking ahead, the researchers hope to apply these findings to better understand cognitive and motivational difficulties in psychiatric and neurological disorders. “We need to understand more about how aberrant dopamine signaling plays a role in shaping policies about cognitive effort,” Westbrook said.

The study, “Striatal dopamine can enhance both fast working memory, and slow reinforcement learning, while reducing implicit effort cost sensitivity,” was authored by Andrew Westbrook, Ruben van den Bosch, Lieke Hofmans, Danae Papadopetraki, Jessica I. Määttä, Anne G. E. Collins, Michael J. Frank, and Roshan Cools.