“Relapse is a part of recovery”: That’s a common refrain among professionals who treat substance use disorders. Many people who have completed treatment programs return to substance use and reenter treatment multiple times, after days, weeks or even years of sobriety.
Marina Wolf , a behavioral neuroscientist at the Oregon Health & Science University, studies how cells in the brain respond to drug exposure in ways that can lead people to develop powerful cravings even months after they stop using drugs such as cocaine, opioids or alcohol. Specifically, she has focused on an aspect of this problem called cue-induced craving, in which people’s brains come to associate a cue — such as seeing a certain location where they previously used drugs — with the desire to use that drug. These learned associations, as she described in the 2025 Annual Review of Pharmacology and Toxicology , are caused by structural changes to the brain — neuroplasticity — as a result of drug use, including the strengthening of connections, called synapses, between specific nerve cells.
These changes don’t disappear as soon as a person, or animal, stops using a drug. Cravings, in fact, can strengthen after abstinence, leaving a person vulnerable to resume using.
Wolf talked with Knowable Magazine about drug-induced neuroplasticity and its implications for treatment. This conversation has been edited for length and clarity.
How did you become interested in neuroplasticity and addiction?
I never had any formal training in synaptic plasticity or addiction. As a graduate student and then a postdoctoral fellow, I worked on how neurons are regulated by the neurotransmitter dopamine, but we studied dopamine’s role in antipsychotic drug effects, not addiction. But when I was setting up my own lab in the early 1990s, I had a friend from graduate school who was involved in groundbreaking studies to work out synaptic plasticity mechanisms in the brain’s hippocampus, a region of the brain responsible for encoding memories. This was fascinating work that helped demonstrate a critical role for a neurotransmitter called glutamate in synaptic plasticity, so I followed it closely.
At the same time, I was in a department where there were a lot of people who were using animal models of addiction. Most addiction work at that time focused on dopamine, which everyone’s probably heard of as the neurotransmitter associated with reward and addiction.
But I thought that dopamine adaptations alone were unlikely to be sufficient to lead to addiction and that there had to be an important role for glutamate and synaptic plasticity. The brain is obviously changing during addiction — people sometimes describe that as maladaptive learning — and synaptic plasticity is the way that experience changes the brain.
And there are glutamate synapses throughout the circuits that connect different brain regions that are important for addiction.
So I started doing experiments to test the role of glutamate in a super-simple rat model of addiction called behavioral sensitization. I was very excited when we got positive results, so I just kept going with this line of work, incorporating better and better animal models and more sophisticated techniques as time went on.
It’s hard for people to believe now, but there was a lot of opposition to this line of work at the time. Still, the body of evidence that supported a critical role for glutamate just kept growing. And now the idea that synaptic plasticity at glutamate synapses is important for addiction is the dogma.
What exactly is synaptic plasticity?
The word “plasticity” just means change, and synaptic plasticity refers to a change in the strength of connections between neurons as a result of an experience. On a cellular level, that's the basis for learning. But to understand this, we have to back up and consider how information is normally transmitted between neurons.
This occurs at structures called synapses that are the connections between neurons. The simplest example involves two neurons — one is sending the signal and the other neuron is receiving it. In the case of a glutamate synapse, the sending neuron will release its chemical transmitter — glutamate molecules — which diffuse across the synapse and bind to glutamate receptors on the receiving neuron. This is translated into an increase in the electrical activity of the receiving neuron.
This, in turn, influences other neurons downstream of the receiving neuron. In this way, you can get changes in activity throughout very complicated circuits. Depending on the circuit, glutamate transmission can be involved in just about every aspect of behavior.
How does this activity at synapses relate to plasticity?
Well, the idea is that the strength of these connections can be changed depending on the history of the neurons’ experience. The classic example of this is long-term potentiation , and this is something scientists have known about for a very long time. Scientists showed that if they stimulated a neuronal pathway with high-frequency electrical stimulation, the receiving neuron subsequently became more sensitive to the effects of glutamate and this increased sensitivity persisted for a long time.
To make a long story short, it was found that the strengthening of these synapses by long-term potentiation was due to the insertion of additional glutamate receptors into the synapse. This made it stronger, and able to respond better when glutamate was released again.
There’s a converse of this, where different patterns of activity will weaken synapses. That’s called long-term depression.
Lastly, there are very interesting but completely different forms of synaptic plasticity that you don’t hear so much about — called homeostatic plasticity — in which a neuron changes synaptic strength to compensate for a long-term change in the level of activity it’s seeing. If, say, there’s a long-term reduction in activity, the neuron will increase the level of glutamate receptors at its synapses. If there’s a significant period of overstimulation, the neuron will reduce the level of receptors.
“Synaptic plasticity research is a very promising way to identify new medications that can serve as a useful partner with behavioral interventions, to help people maintain abstinence for longer periods of time and avoid relapse.”
Homeostatic plasticity is probably very important in addiction, because this is a disorder where there are long-term changes in the activity of brain pathways that are very important for motivated behavior. In fact, in the rat model used in my lab to study cue-induced drug craving — in other words, craving triggered by cues the rat previously learned to associate with the drug — this is the type of plasticity that maintains high levels of craving even months after the last time the rat experienced the drug.
Specifically, during drug abstinence, an atypical but very potent type of glutamate receptor, called a calcium-permeable AMPA receptor, is inserted into glutamate synapses in a part of the brain called the nucleus accumbens. The nucleus accumbens is a key center where the input of glutamate from many regions are integrated to drive motivated behavior, including drug-seeking.
Homeostatic strengthening of these synapses during abstinence probably occurs to compensate for lower glutamate release during the abstinence phase compared to the drug-taking phase of the experiment. The problem is that drug-seeking depends on glutamate transmission at these synapses. So this homeostatic strengthening of glutamate synapses becomes a liability if the rat is exposed to a cue that reminds them of the drug — now the nucleus accumbens neurons have a stronger response to glutamate released by that cue, and this will cause the rat to more strongly seek the drug.
Can you talk a little bit about the models you use to study neuroplasticity in substance use disorders? How closely do they reflect what we see in humans?
The gold standard is to use drug self-administration, in which the animal — usually a rat — decides when to take the drug, rather than the experimenter just giving the rat injections of drug. This is very important, because you can only study motivation for drugs when the animal is choosing to take drugs.
In self-administration experiments, the animal is placed in a box where they learn that performing a behavior, such as poking their nose in a specific hole, will deliver an intravenous infusion of drug. Typically, that infusion is paired with some kind of a cue — a light that comes on in the nose-poke hole, for example — so that the animal learns the relationship between that cue and the availability of drug. This parallels the human situation where people, places and things associated with drug use can become powerful triggers for craving and relapse.
In our experiments, we train the rats for 6 hours per day for 10 days so that they learn the relationship between the nose-poke, the cue and drug infusion really well. Then they go back to their home cage for a period of abstinence, during which they resume their normal rat life — no drugs, no cues. After different periods of abstinence, they can be tested for how much craving the light, or other cue, evokes.
During the test, which typically lasts 30-60 minutes, rats are put back in the box where they learned to self-administer the drug, but now the rules have changed — when they poke their nose in the hole that previously delivered the cue and the drug, they only get the cue, not the drug. The number of times they’ll keep nose-poking under this condition is our measure of how strongly the cue alone is able to maintain drug-seeking behavior. In rats, drug-seeking is our measure of motivation for drug — the closest we can come to assessing craving.
Years ago, researchers used this type of experiment to show something surprising. When they tested rats after different periods of abstinence, they found that cue-induced craving progressively increased, or “incubated,” over the first few weeks of an abstinence period. You might think craving would fall off as rats get further away from the last drug self-administration session, but it’s the opposite. This phenomenon was termed “incubation of drug craving.”
After this initial period when craving increases, the rat reaches a plateau phase in which high craving is sustained, although ultimately it does decline. So the level of craving follows an inverted U-shaped curve. For cocaine, which has been best studied, the highest levels of craving are observed between one and three months of abstinence. While the amount of research on incubation in humans is limited compared to rodent research, the work that has been done mirrors this timeline. This is chilling if you think about the fact that addiction treatment programs often last only a month — so people are returning to the real world just at the point when cue-induced craving is peaking.
Unfortunately, the majority of people who are recovering from substance use disorders will ultimately relapse. And that doesn’t always happen right away. It can happen months after they achieve abstinence. This persistence of vulnerability to craving and relapse is one of the major reasons that addiction is so hard to treat. So a big question for the field, and the one my lab is focused on, is, What is happening in the brain to maintain a high level of vulnerability to craving and relapse for such a long period of time? The incubation of craving model gives us a way to attack this question.
As I mentioned earlier, we’ve shown that a critical mechanism in maintaining the plateau phase of incubated cocaine craving is strengthening of glutamate synapses in the nucleus accumbens through the insertion into glutamate synapses of those very powerful calcium-permeable AMPA receptors. Interestingly it takes a few weeks for these receptors to be inserted but once this happens, their levels in the synapse remain high for months. Using pharmacological or other approaches to remove them from synapses, or prevent glutamate from activating them, will prevent the expression of incubated cocaine craving. The review article goes into a lot of detail on how we might translate this knowledge into therapeutic approaches to reduce craving and help people maintain abstinence.
A lot of your work, particularly in the review, focused on cocaine. Do we have reason to believe that we see similar neurological changes with other drugs, like opioids, alcohol and nicotine?
Incubation of craving in rats — in other words, an inverted U-shaped curve where craving rises, plateaus and then declines — holds across drug classes. So although it was initially demonstrated for cocaine, incubation of craving occurs in rats after self-administration of methamphetamine, opioids, nicotine and ethanol. Incubation of cue-induced craving has also been demonstrated in humans — so far, this has been shown during abstinence from cocaine, methamphetamine, nicotine and alcohol.
So that long plateau phase that we see in the animals is a relevant model for the persistent vulnerability to craving and relapse in humans who are trying to recover from substance use disorder.
Different classes of drugs do have different initial targets in the brain. For example, cocaine interacts with proteins that regulate dopamine and related neurotransmitters, not glutamate. Downstream of these initial targets, other neurotransmitters and circuits become recruited, and to some degree this depends on the specific drug class. For example, circuits involved in depressive and anxiety-like behaviors seem to be recruited to a greater degree during opioid abstinence as compared to stimulants.
It’s important not to downplay these differences. But there are some circuits, such as those involving the nucleus accumbens, where synaptic plasticity at glutamate synapses is important across drug classes. This makes sense, because glutamate synapses in the nucleus accumbens are generally important for translating cues into motivated behavior.
For example, calcium-permeable AMPA receptors are important for incubation of methamphetamine and oxycodone craving, as well as for cocaine.
That said, there are some interesting differences between drug classes in the cell types and input pathways that undergo this plasticity, which we are currently exploring.
How could this information help doctors, people in recovery, families, anybody dealing with substance use disorders?
Before I answer that, there are a few other things I should say to put all this in context. The first is that we have discovered a form of synaptic plasticity in the nucleus accumbens that is essential for the expression of incubated drug craving. However, plasticity contributing to incubation of craving has also been discovered in other brain regions. Because the brain is so highly interconnected, no behavior can be attributed to a single brain region — it always reflects the activity of complex neuronal circuits.
“This persistence of vulnerability to craving and relapse is one of the major reasons that addiction is so hard to treat.”
The trick is finding places to interrupt pathological circuit activity, and the nucleus accumbens is obviously a strong target. However, if this is going to be accomplished by giving a drug, that treatment has to have the desired effect in the targeted brain region without causing problems in other regions.
The other caveat is that these incubation studies are just modeling one aspect of addiction: high reactivity to drug cues. There are so many other critical factors, like ramping up of brain systems that mediate stress responses or brain systems that mediate anxiety and depressive states. Changes in these neuronal systems can persist for long periods of abstinence too.
It’s also important to realize that substance use disorder is often associated with erosion of personal relationships, employment and financial security. These are the aspects of life that help most of us cope with stress. Without long-term support, it’s very unlikely that people with substance use disorder will be able to maintain abstinence. So while it’s important to discover plasticity mechanisms and develop treatments based on those mechanisms, it’s also important to see the big picture of these human beings out in the world, and the many challenges that they cope with.
That said, the research discussed in my review article offers a way to identify potential new treatments for addiction including, but not limited to, medications. For cocaine and methamphetamine use disorder, there’s currently no FDA-approved pharmacotherapy, so the standard of care is behavioral interventions like motivational interviews, contingency management, community reinforcement and cognitive behavioral therapy.
These things work, but there are barriers relating to cost, availability of providers, social stigma and more. Medications that would help people stay abstinent might give these sorts of interventions time to take hold.
There are drugs in the pipeline for cocaine and methamphetamine use disorder that work by blocking the rewarding effects of these drugs if a person takes them, but I would argue that we’d be better off targeting the plasticity that supports long-term vulnerability to craving and relapse during abstinence. The good news is that the mechanism we’ve discovered — the strengthening of nucleus accumbens glutamate synapses — but also other plasticity mechanisms that other researchers have identified, have given us insight into targets for design of anti-craving medications that might be able to reverse the plasticity and thus reverse the addiction-related behavior.
But even though we have starting points for the development of such medications, we all know that developing a compound that is safe enough to give repeatedly to a human being, and advancing that compound through clinical trials, is very, very difficult. Nonetheless, I think that a combination of medications and behavioral interventions is going to be the key to lasting recovery for stimulant users.
We already know this from opioid use disorder, where decades of research have shown us that giving drugs like methadone can reduce opioid use and help people improve their quality of life.
Synaptic plasticity research is a very promising way to identify new medications that can serve as a useful partner with behavioral interventions, to help people maintain abstinence for longer periods of time and avoid relapse.
Stay in the Know
Sign up
for the Knowable Magazine
newsletter today
OK, and the million-dollar question: Do these synapses ever go back to what they were like prior to substance use?
There is a lot of controversy about the extent to which these changes recover. This can be very complicated to study in humans because humans experiencing substance use disorder can differ widely in their history of drug use but also in other aspects of mental health, such as comorbid depression or anxiety, and environmental circumstances. Still, brain-imaging studies in both humans and nonhuman primates suggest that there’s some recovery of both structure and function over the first year of abstinence.
This suggests that plasticity is going on in both directions. There’s plasticity associated with the induction of the behavioral and underlying brain changes linked with substance use disorder, and also, presumably, in the recovery. But to get at the underlying plasticity, we need to use animal models.
We can use them in two ways: to figure out how to disrupt plasticity that promotes addictive behaviors, and how to enhance plasticity that promotes recovery. I've talked more about the first kind of work, namely the strengthening of glutamate synapses that underlies the strengthening of craving in the incubation model. But you can also use rodents and plasticity research to identify behavioral interventions to promote recovery.
So, for example, other groups have shown that putting rats in an enriched environment or eliciting improvements in their quality of sleep can promote behavioral improvements, and in some cases this has been linked to opposing the synaptic plasticity underlying drug craving.
This article originally appeared in Knowable Magazine , an independent journalistic endeavor from Annual Reviews. Sign up for the newsletter .
