For decades, physicians and researchers have explained the action of ADHD stimulant medications in the same way: the drugs work by reaching into the brain’s attention networks and turning up the volume on focus. A new study, published in Cell on December 24, 2025, says that explanation may have been wrong all along.

Researchers at Washington University School of Medicine in St. Louis analyzed resting-state brain scans from 5,795 children between the ages of eight and eleven. When they compared children who had taken a stimulant medication on the day of their scan with children who had not, the expected pattern, increased activity in the brain’s attention systems, was not there. What they found instead was elevated activity in regions tied to arousal, wakefulness, and the prediction of reward.

The study’s lead author, Dr. Benjamin Kay, an assistant professor of neurology at Washington University and a practicing pediatric neurologist, described the moment he first reviewed the results to The Washington Post.

“When I first saw the results, I thought I had just made a mistake because none of the attention systems are changing here,” Kay said.

They were not changing because, the researchers now argue, that is not where the medication acts.

The finding, if it holds up, would represent one of the most significant reframings of a class of medications in modern psychiatric pharmacology. It would also raise uncomfortable questions about why the prior consensus held for so long, what other clinical assumptions might be rooted in the same kind of methodological narrowness, and how an entire generation of patients and prescribers should now think about a treatment they have been told for half a century works on attention itself.

The Hypothesis Being Challenged

To understand why the new finding is consequential, it helps to understand what was being assumed.

The attention-network model of ADHD stimulant action did not arrive overnight. It was built up over four decades of research, primarily focused on the prefrontal cortex and the neurotransmitters that regulate its function. The model’s core claim was that ADHD reflects underactivity of the prefrontal cortex, the brain region responsible for top-down attention, working memory, and the regulation of behavior. Stimulant medications, the model held, treat ADHD by raising levels of two key neurotransmitters in the prefrontal cortex: dopamine and norepinephrine. These elevated catecholamine levels were thought to strengthen prefrontal network connectivity and improve the brain’s capacity for sustained, voluntary attention.

The model was anchored in real evidence. Decades of animal research showed that low doses of methylphenidate improved working memory and attention performance in rats and monkeys, and that this improvement could be blocked by drugs that interfered with dopamine D1 or alpha-2 adrenoceptor signaling in the prefrontal cortex. Task-based brain imaging in humans showed that stimulants appeared to normalize prefrontal activity in patients with ADHD during cognitive tasks. Genetics implicated dopamine signaling pathways. Pharmacologically, every approved ADHD medication, from methylphenidate to amphetamine to atomoxetine to guanfacine, increases catecholamine availability in the prefrontal cortex through one mechanism or another.

The model became the accepted teaching in medical schools, the framing used in patient education, and the conceptual foundation for new drug development. Pediatric neurologists and psychiatrists learned to describe stimulants to families in attention-network terms: the medication “helps the brain’s focus areas work better.” Patients learned to describe their own experience that way: “the medication helps me focus.”

The new Cell paper does not say the prior model is entirely wrong. It says the prior model may have been measuring the right brain but the wrong networks. Stimulants do act on the catecholamines. They do produce changes in the prefrontal cortex. But the change that matters for clinical effect, the change that distinguishes a medicated brain from an unmedicated one, appears to live in the arousal and reward circuits rather than in the attention networks themselves.

What the Study Did

The research drew its primary data from the Adolescent Brain Cognitive Development (ABCD) Study, a long-running, multisite project that has tracked the neurodevelopment of more than 11,000 children across the United States. ABCD is the largest long-term study of brain development and child health ever conducted in the country, and it includes a deep array of cognitive, behavioral, and brain imaging data for each participant.

The Washington University team examined resting-state functional magnetic resonance imaging, or fMRI, data from 5,795 ABCD participants aged 8 to 11. Resting-state fMRI measures patterns of brain activity when a person is not engaged in any specific task. Unlike task-based imaging, which captures the brain responding to a stimulus, resting-state imaging shows the underlying connectivity of brain networks as they cycle through their default states. It is a powerful tool for asking what a medication does to the brain’s organization rather than to its response to a particular demand.

The researchers compared brain connectivity patterns between children who had taken a prescription stimulant on the day of their scan and children who had not. The comparison was made possible by ABCD’s careful documentation of each participant’s medication status on the day of every assessment.

The expected finding, if decades of clinical teaching had been correct, would have been increased connectivity in classical attention systems, including the dorsal attention network and prefrontal regions associated with executive control. That finding did not appear. The children who had taken stimulants did not show meaningfully elevated activity in those networks compared to children who had not.

What the researchers found instead was a different pattern entirely. Children who had taken stimulants on the day of their scan showed increased activity in regions related to arousal or wakefulness, and in regions involved in predicting how rewarding an activity will feel.

The Validation Experiment

A finding this contrary to clinical orthodoxy needed corroboration. The team designed a small but tightly controlled validation experiment with five healthy adults who did not have ADHD and who did not normally take stimulant medication. Each participant received at least four baseline resting-state fMRI scans, then received a 40-milligram dose of methylphenidate by mouth, then was scanned again 60 to 180 minutes later, the window when the medication reaches peak effect.

The same pattern emerged. The arousal and reward networks lit up. The attention networks did not.

Two independent datasets, one observational and very large, the other controlled and small, converged on the same finding. The researchers reported their results in the December 24, 2025 issue of Cell, the flagship journal of the life sciences. The paper was accompanied by a commentary in the same issue, titled “Stimulants as agents of arousal in whole-brain functional connectivity,” which described the findings as “robust and reproducible” and noted that the multi-dataset investigation pointed toward “an arousal-based pathway linking stimulants to improved cognition.”

The Reward Prediction System

The Cell paper’s central reframing depends on a neuroscientific concept that has matured substantially over the past three decades: the brain’s reward prediction system. To understand the claim that stimulants “pre-reward” the brain, it helps to understand what that system does.

The reward prediction system is a circuit anchored in the midbrain and ventral striatum, with extensive projections into the prefrontal cortex. Its function is not to register rewards as they occur but to predict, in advance, how rewarding an upcoming activity is likely to be. Dopamine-releasing neurons in the midbrain fire in proportion to that prediction. When a predicted reward is larger than expected, the system generates a positive prediction error and dopamine release increases. When a predicted reward fails to materialize, the system generates a negative prediction error and dopamine release drops.

This prediction system is what motivates engagement. It is what tells the brain that a particular activity is worth sustaining attention on. High predicted reward feels engaging. Low predicted reward feels boring. The system runs continuously, evaluating every available option and biasing behavior toward the more rewarding choice.

In people with ADHD, the reward prediction system appears to function differently. A substantial body of research, much of it predating the new Cell paper, has documented that people with ADHD show altered reward prediction signaling, with a steeper preference for immediate over delayed rewards and a reduced capacity to sustain engagement with activities that lack short-term payoff. Whether this difference is a consequence of altered dopamine signaling, a cause of it, or both, has been debated for years. What is not debated is that the reward prediction system is involved.

The Cell paper’s contribution is to suggest that stimulant medications act directly on this prediction system. They raise the baseline of reward prediction. Activities that would have been tagged as low-reward, and therefore boring, get retagged as moderate-reward. The threshold for engagement drops. A worksheet on long division, a tedious meeting, a chapter of dense reading: the medicated brain finds these tasks rewarding enough to sustain. The unmedicated brain finds them aversive and disengages.

This is what Dosenbach means when he says stimulants “pre-reward” the brain. The medication is not handing out rewards. It is changing the brain’s prediction of how rewarding an upcoming activity will feel, before the activity begins. The pre-reward is a forecast adjustment, not a payoff.

The framing has an immediate clinical consequence. Under the attention-network model, stimulants were thought to give patients more focus. Under the reward-prediction model, they give patients more motivation to engage with tasks the brain would otherwise abandon. These are different mechanisms, and the second one is closer to what many adults with ADHD describe experiencing on medication: “It’s not that I can suddenly focus. It’s that I can finally make myself care enough to start the thing.”

The Hyperactivity Reframe

The reward-prediction reframing has implications for the second core symptom dimension of ADHD: hyperactivity. The condition’s name pairs attention deficits with hyperactivity for good reason. The two symptoms cluster together. They tend to respond to the same medications. They have been understood, traditionally, as different expressions of the same underlying problem with executive control.

Dosenbach offered an alternative reading in the Washington University statement that accompanied the paper.

“Whatever kids can’t focus on, those tasks that make them fidgety, are tasks that they find unrewarding,” he said. “On a stimulant, they can sit still because they’re not getting up to find something better to do.”

This is a substantial reconceptualization. The traditional model of hyperactivity has framed it as a motor expression of executive dysregulation: the child cannot inhibit the impulse to move, so they fidget, climb, run, leave the seat. The new framing reverses the causal direction. The child is not failing to inhibit motor impulses. The child is generating motor activity in pursuit of stimulation, because the available activity is not rewarding enough to sustain.

Under this model, hyperactivity is not impulsivity. It is reward seeking. The child who cannot stay in their seat during long division is not failing to control themselves. They are leaving long division because their brain has correctly assessed that long division is not rewarding enough to sustain, and is searching for something more rewarding to engage with.

The reframing matters because it implies a different relationship between the child and the symptom. In the inhibition model, hyperactivity is a deficit, a failure of a system that should be working better. In the reward-seeking model, hyperactivity is a strategy, a sensible response to an environment that is not adequately rewarding for the brain that has to navigate it. The strategy may be socially costly. It may interfere with school. But it is not a failure of self-control; it is an adaptive response to an unrewarding situation.

For neuroaffirming practitioners who have argued for years that ADHD should be reframed as a difference in motivation and reward processing rather than as a deficit of attention or inhibition, the new findings provide neuroscientific support for what was previously a primarily phenomenological argument.

The Sleep Problem

One of the most consequential findings in the paper concerns sleep, and it is the part of the research most likely to change clinical practice.

The researchers found that stimulant medications produced brain activity patterns that mimicked the brain signature of adequate sleep. In children who had not slept enough, taking a stimulant erased the neuroimaging markers of sleep deprivation, along with the associated behavioral and cognitive impairments.

“We saw that if a participant didn’t sleep enough, but they took a stimulant, the brain signature of insufficient sleep was erased, as were the associated behavioral and cognitive decrements,” Dosenbach said.

The ABCD data showed this effect at the level of academic performance. Children who slept less than the recommended nine or more hours per night and took a stimulant received better grades in school than children with insufficient sleep who did not take a stimulant. The medicated, sleep-deprived children received roughly the same grades as well-rested children who did not take a stimulant.

This is where the study turns sober. Sleep-deprived children exhibit symptoms that look very much like ADHD. They have trouble paying attention. They struggle to sit still. They perform poorly academically. They appear restless and dysregulated. A clinician evaluating such a child, working through the standard ADHD diagnostic checklist, may reasonably conclude that ADHD is present and prescribe a stimulant. The stimulant, the new data suggests, will work. The child will appear to improve. The clinician will feel confirmed in the diagnosis.

The underlying problem, though, may have been sleep all along.

“Not getting enough sleep is always bad for you, and it’s especially bad for kids,” Kay told the Washington University news service. He urged clinicians evaluating children for ADHD to consider sleep deprivation as a factor and to explore strategies and treatments to address inadequate sleep before, or alongside, stimulant medication.

The mechanism the researchers propose is that stimulant medications produce a brain state that resembles having slept well, even when the child has not. The chronic underlying sleep deficit continues. The child is still not getting the developmental and health benefits of adequate sleep. But the surface symptoms, the ones a teacher or parent would notice, are masked by the medication. The treatment works, in the limited sense that the child performs better, while the actual physiological problem goes unaddressed.

The researchers note this could produce long-term harm if stimulants are used as a substitute for sleep rather than as a treatment for an underlying attention disorder. They call for more research on the long-term effects of stimulant use when the medication is functioning as a sleep workaround rather than as a treatment for ADHD.

The Pushback

A finding this contrary to consensus invites scrutiny, and the most substantive pushback has come from Dr. Amy F. T. Arnsten, a professor of neuroscience at Yale University School of Medicine and one of the most cited researchers in the field of prefrontal cortex pharmacology. Arnsten has spent decades demonstrating that catecholamine signaling in the prefrontal cortex is central to both ADHD and the action of the medications used to treat it. Her published work is heavily referenced in the Cell paper’s own introduction.

In a commentary published in Medscape on February 19, 2026, Arnsten raised methodological objections to the resting-state approach the Cell authors used. Her core argument is that resting-state fMRI cannot capture what medications do when the relevant brain circuits are actively engaged in cognitive tasks.

“Neurons in dorsolateral prefrontal cortex are not activated during rest,” Arnsten said, “and so changes in the neurochemical state of those neurons during rest tell us little about how medications affect them when those neurons are actually doing their work.”

“The whole point of this field is trying to learn how stimulants improve symptoms,” Arnsten continued. “Thus, we need to see how they alter brain activity when they are improving performance of tasks linked to symptoms.”

Arnsten’s critique points to a real tension in the methodology. Task-based fMRI has consistently shown that stimulants normalize prefrontal and striatal activity in patients with ADHD during cognitive tasks. The Cell authors acknowledge this in their paper but argue that task-based findings are confounded by drug-induced improvements in task performance itself: if a child does better on the task after taking the medication, you cannot cleanly separate the brain change caused by the drug from the brain change caused by the changed performance.

Dosenbach responded to Arnsten’s critique directly.

“That’s unlikely to be true given that we found clear and strong effects of stimulants during the resting state, just not in the attention networks,” he said.

Kay added that the brain spends about 95 percent of its metabolic energy at rest, so prefrontal neurons are certainly active during resting-state imaging. He also noted that in supplemental analyses, the team examined functional connectivity during an attention-demanding n-back task and found patterns in children that mirrored those at rest, with no dramatic stimulant-related changes in canonical attention networks.

The disagreement is unlikely to be resolved by either side asserting its position. It is the kind of dispute the field resolves through replication, additional experiments, and the slow accumulation of converging evidence from different methodologies. What is striking is that Arnsten, Kay, Dosenbach, and the field generally all agree on one point: stimulants work. They reduce symptoms in patients with ADHD. The clinical efficacy data, drawn from hundreds of randomized trials over decades, is among the most robust in psychiatry. The disagreement is entirely about how they work, not whether.

“We’re not saying don’t give these drugs,” Dosenbach emphasized. “The clinical data show they work great.”

Why It Took So Long to Find

One question the paper inevitably raises is how a class of medications prescribed to roughly 3.5 million American children, and to many more adults, could have been mischaracterized for so long. The answer is partly methodological and partly historical.

Most prior research on stimulant medication has used task-based brain imaging. In a task-based design, participants are asked to perform an attention task, like watching for a target on a screen, while their brain is being scanned. Researchers measure how the brain responds to the demands of the task. This kind of design is excellent for measuring how the brain handles attention, but it builds an assumption into the measurement itself: the researcher is looking for changes in the brain’s response to an attention task, so changes elsewhere in the brain can be missed.

Resting-state fMRI does the opposite. It measures the brain at rest, when no specific task is being performed, and asks how the underlying connectivity of brain networks is organized. A medication that changes the brain’s overall state, by elevating arousal, by changing the baseline activity of reward circuits, will show up in resting-state data even if it does not produce a measurable change in any particular task.

The ABCD Study, with its large resting-state imaging dataset and its well-documented medication histories, made the comparison possible at scale for the first time. The Washington University team also developed open-source software, available on GitLab, to extract traits of interest from the ABCD data, which they have made available for other researchers to replicate and extend their analysis.

There is also a historical answer. The attention-network model was built primarily in the 1990s and early 2000s, when neuroimaging was still task-based and resting-state imaging was an emerging technique. By the time resting-state methods matured into a standard tool, the attention-network model had become consensus, and the questions researchers were asking of new data tended to be questions about how the data supported the existing model rather than questions about what other model the data might fit. The Cell paper is, in part, the result of researchers asking a different question of a dataset large enough to answer it.

What the Study Does Not Say

Several misreadings of the research are predictable, and the researchers have been careful to head them off.

The study does not say that ADHD stimulants do not work. They do work. Children with ADHD who take stimulants get better grades and perform better on cognitive tests, a pattern documented in the ABCD data and in decades of prior research. The paper is not a critique of stimulant medication. It is a re-explanation of how stimulant medication helps.

“The paper clearly shows that they help,” Dosenbach told the Washington Post. “They help kids who have a diagnosis of ADHD do better in school and do better on tests, and they help kids who don’t sleep enough, and a lot of Americans don’t sleep enough.”

The study also does not say that ADHD is just sleep deprivation in disguise, or that the disorder is overdiagnosed. The data does not support either claim. The study says that one mechanism by which stimulants help, the masking of sleep deficit, may explain some apparent treatment successes in a subset of children whose underlying problem is sleep rather than executive function. Distinguishing that subset from children with primary ADHD is a clinical question the study does not resolve.

The study does not say that the attention network model of ADHD is wrong. Many other lines of research, including task-based fMRI studies, behavioral genetics, and clinical trial data, continue to implicate attention networks in ADHD itself. The study’s contribution is specific to the action of the medication, not to the nature of the disorder.

And the study does not say that anyone should stop taking their medication. The clinical recommendation it generates is the opposite of an abandonment recommendation. It is a recommendation that sleep be more carefully assessed in children being evaluated for ADHD, and that treatment plans for diagnosed children include attention to sleep alongside medication.

What This Means for People With ADHD

The reward-system reframing has implications beyond clinical practice. For many adults with ADHD, the new model offers neuroscientific support for an experience they have been articulating to themselves and to skeptics for years.

The dominant cultural framing of ADHD, anchored in the attention-network model, has been a deficit framing. The brain’s attention systems are weaker. The patient cannot focus. The medication corrects the deficiency. This framing has produced its own kind of harm. People with ADHD have been told, implicitly or explicitly, that they are failing at something the rest of the population is succeeding at, and that the medication is a corrective for that failure. The internalized version of this message, repeated over years, is the well-documented ADHD adult experience of shame: I should be able to do this. Everyone else can. Why can’t I?

The reward-system reframing changes the story. Under the new model, the brain with ADHD is not failing to attend. It is doing exactly what an unmedicated brain should do when the predicted reward is too low to justify the cognitive cost: it is disengaging. The behavior that looks like attention failure is, neurologically, an entirely rational response to an unrewarding situation. The medication does not fix a broken attention system. It changes the brain’s reward predictions, raising the baseline rewardingness of activities the brain would otherwise correctly identify as not worth the effort.

“It’s not that I can suddenly focus,” is something many adults with ADHD have said about medication for years. “It’s that I can finally make myself care enough to start the thing.” The reward-system reframing matches that lived experience more cleanly than the focus-pill framing ever did.

The neuroaffirming community, which has argued for years that ADHD should be understood as a difference in motivation and reward processing rather than as a deficit of attention or self-control, has reason to take the new findings seriously. The Cell paper is not a manifesto for neuroaffirming framing. It is a careful piece of empirical neuroscience. But its findings, if they hold up, provide an empirical foundation for a way of understanding ADHD that has, until now, been articulated primarily in first-person terms and dismissed by some practitioners as a softer alternative to a real diagnosis.

The implications for how patients are counseled are real. A child told that medication will “help your focus” learns one thing about themselves. A child told that medication will “help your brain feel like the work is worth doing” learns something different, and arguably more accurate. The first framing implies that without the medication the child is broken. The second framing implies that without the medication the child is making sensible decisions about an environment that does not adequately reward their effort.

What the Study Does Not Yet Resolve

The Cell paper opens more questions than it answers, which is true of most consequential research.

It does not resolve whether the long-term effects of stimulants on children’s brain function are restorative or harmful. The researchers have noted that wakefulness has its own restorative effects through the brain’s glymphatic waste-clearing system, which is most active during specific physiological states. Stimulants might support brain health by promoting these states, the authors suggest, or they might cause lasting harm if used to mask chronic sleep deficits. The data does not yet distinguish between those possibilities.

It does not address the question of whether the reward-network mechanism explains the action of stimulants in adults with ADHD. The validation experiment with five adults supports the basic mechanism, but the developmental implications for children whose reward systems are still maturing are likely different from the implications for adults whose neural architecture is established.

It does not address the comparative mechanism of different stimulant medications. Methylphenidate, the active ingredient in Ritalin and Concerta, and amphetamine salts, the active ingredients in Adderall, work somewhat differently at the pharmacological level. The study examined both classes through the ABCD data, but the mechanistic question of whether the same arousal-reward effect drives both classes equally is unresolved.

It does not yet resolve the debate with researchers like Arnsten. The methodological question of whether resting-state imaging can adequately speak to medication effects during task performance is genuine, and will require further work to settle. Arnsten’s argument that task-based studies show stimulant-induced normalization of prefrontal activity is real, and the Cell authors’ counterargument about performance confounds is also real. The next generation of studies will likely involve task-based designs that explicitly control for performance changes, allowing researchers to separate the medication’s effect on the brain from the medication’s effect on what the brain is doing.

The researchers also note, candidly, that the ABCD data has limitations for pharmacokinetic precision. The study could not perfectly account for the timing of medication doses relative to scans, or for the difference between immediate-release and extended-release formulations. These limitations may have led to an underestimation, not an overestimation, of the medication’s effects on fMRI connectivity.

What Clinicians and Families Should Take From This

For clinicians, the practical implication is twofold. First, sleep assessment should be part of any ADHD evaluation, particularly for children. The risk of misclassifying chronic sleep deprivation as ADHD is real, and the stimulant treatment that follows such a misclassification, while it may produce surface improvement, leaves the underlying problem in place. Second, for children with diagnosed ADHD, attention to sleep is not optional alongside medication. The medication will help, but it will help differently and perhaps more durably if sleep is also being addressed.

For families of children currently taking ADHD medications, the practical implication is more measured. The medication is still working. The child is still receiving real benefit. The story has not changed at the level of “should my child take this.” The story has changed at the level of how the medication is doing what it does, and that change matters for how families think about complementary support: making sure the child gets adequate sleep, structuring the day so that demanding tasks come when the child has the most reward-system fuel, recognizing that “boring” tasks are pharmacologically harder for the medicated brain than they are intrinsically aversive.

For adults with ADHD, the implications are similar. The medication helps you persist on tasks your reward system does not naturally embrace. It does not give you a sharper focus the way a pair of glasses gives sharper vision. It changes the relative cost of staying engaged with cognitively demanding work. Understanding the mechanism may help adult patients design their workdays around it.

For the field of ADHD research, the implications are larger and slower. The reward-network mechanism, if it holds up under further investigation, will reshape how new ADHD medications are developed and tested. It will likely also reshape how the disorder itself is conceptualized: not as a deficit of attention per se, but as a configuration of the reward system that makes ordinary cognitive tasks feel more aversive than they need to. Treatments that act on that configuration, whether pharmacological or behavioral, may prove more effective than treatments aimed at attention systems that may not be the relevant target.

The Study at a Glance