Addiction Is Not a Willpower Problem: What the Science Actually Says

No, addiction is not simply a willpower problem. Imaging research links addiction to fewer dopamine D2 receptors and reduced prefrontal activity, the brain region behind inhibitory control. Heritability of alcohol use disorder is about 49%, and childhood adversity sharply raises risk. Willpower is the faculty addiction appears to weaken.

Addiction Is Not a Willpower Problem: What the Science Actually SaysAndrew Baker's Technology Blog Posts, 09/Nov/2026
https://andrewbaker.ninja/2026/11/09/addiction-is-not-a-willpower-problem-what-the-science-actually-says/

I recently watched a show featuring Matthew Perry, and somewhere in the discussion someone made the argument you have probably heard a hundred times: that addiction is ultimately about willpower, and that people who stay addicted are, at some level, choosing to. I understand why the idea is attractive, because it is simple and it puts the world back into a shape where good choices lead to good outcomes. But it broke my heart a little, because I think it is wrong, and because I think the people who say it sincerely believe that fellow human beings are consciously authoring catastrophes in their own lives and in the lives of their families.

My view is this: addiction is the absence of meaningful choices to switch off the noise in your head. That does not mean choice vanishes entirely. It is the progressive loss of the freedom to exercise it, as the noise becomes so overwhelming that the one option offering immediate relief crowds out almost every other possibility. Telling someone they choose to be like this does not help them make a better choice; it simply layers a second tragedy on top of the first. In this post I want to set out why I believe that, what the evidence actually says, and where the honest complications are, because a position this important should be able to survive its strongest critics.

1. What “just use willpower” quietly assumes

The willpower argument rests on a hidden model of the person. It assumes there is a stable, fully functioning decision maker sitting inside each of us, weighing up options with the same equipment and the same inputs as everyone else, and that the addicted person is simply making a worse call. If that model were right, then shame, consequences and lectures would work, because they would change the inputs to a decision maker that is otherwise intact.

Anyone who has spent time building systems will recognise the flaw in that reasoning. When a production system keeps failing, you do not fix it by telling the engineers to be more careful; you look at the architecture, the feedback loops, the load it is under and the environment it is running in. “Try harder” is what we say when we have not yet understood the system. Addiction deserves at least the same engineering humility we would give a flaky service, and probably a great deal more, because the system in question is a human being.

2. You would be asking a compromised system to repair itself

The faculty we call willpower lives largely in the prefrontal cortex, the part of the brain responsible for planning, weighing consequences and inhibiting impulses. Imaging research led by Nora Volkow, who now directs the US National Institute on Drug Abuse, has repeatedly found that people with addiction show lower availability of dopamine D2 receptors, and that this is associated with reduced activity in prefrontal regions involved in salience, inhibitory control and executive function. Volkow and Rita Goldstein describe the result as a syndrome of impaired response inhibition and salience attribution: the drug becomes overwhelmingly important, and the brakes that would normally hold back the urge become weaker.

I want to be careful not to overclaim here. Most of these imaging studies compare people with and without addiction at a point in time, so they cannot fully separate what the drug did to the brain from what was already different before the first use; the most defensible reading is that these differences are consistently present in addiction and plausibly contribute to the loss of control. Even on that conservative reading, the implication for the willpower argument is uncomfortable. The very circuitry you would need to “just stop” is the circuitry that addiction appears to degrade, which means the demand for more willpower is a demand that a compromised system diagnose and repair itself using the faculty that has been compromised.

3. Nobody chooses the hand they were dealt

The second problem with the choice model is that people arrive at their first drink or first pill with very different levels of risk, and none of them chose that starting position.

Twin studies are the classic way to estimate how much of the variation in a trait across a population is linked to genetic differences, by comparing identical twins, who share all of their genes, with fraternal twins, who share about half. A 2015 meta analysis of twelve twin studies and five adoption studies estimated the heritability of alcohol use disorder at about 49%, and estimates for other substances across the twin literature land in a broadly similar range, as the chart below shows.

{
  "type": "bar",
  "data": {
    "labels": ["Nicotine dependence", "Alcohol dependence", "Cannabis addiction", "Cocaine use disorder"],
    "datasets": [{
      "label": "Range of heritability estimates from twin studies (%)",
      "data": [[33, 71], [48, 66], [51, 59], [42, 79]],
      "backgroundColor": "rgba(54, 162, 235, 0.6)",
      "borderColor": "rgba(54, 162, 235, 1)",
      "borderWidth": 1
    }]
  },
  "options": {
    "indexAxis": "y",
    "plugins": {
      "title": { "display": true, "text": "How much of the variation in addiction risk is linked to genes (twin study ranges)" },
      "legend": { "display": false }
    },
    "scales": {
      "x": { "min": 0, "max": 100, "title": { "display": true, "text": "Heritability (%)" } }
    }
  }
}

Ranges as summarised by Agrawal and colleagues (2012). Each bar spans the estimates reported by different twin studies, using different populations and designs; they are not confidence intervals. The same review reports a consistent estimate of about 49% for gambling disorder across two major twin studies, which matters because no substance is involved at all.

Two caveats matter, and they make the argument stronger rather than weaker. Heritability describes variation across a population, not a percentage of any one person’s fate, so it does not mean half of your addiction is genetic; it means that across a population, roughly half of the differences in who becomes addicted track genetic differences. And it is not destiny, because genes load the gun while environment and experience do much of the rest.

That environment starts early. The Adverse Childhood Experiences (ACE) study, run by Kaiser Permanente and the US Centers for Disease Control with more than 17,000 adults in total, counted categories of childhood adversity such as abuse, neglect and household dysfunction. In an analysis of 8,613 of those adults focused on illicit drug use, Shanta Dube and colleagues found that each category of adversity increased the likelihood of early drug initiation two to four fold, and that the ACE score had a strong graded relationship with later drug problems, addiction and injection drug use; people reporting five categories of adversity were 7 to 10 times more likely to report these problems than people reporting none. They estimated that between half and two thirds of serious drug problems in their sample were attributable to these childhood experiences. That figure is a population estimate based on adults recalling their childhoods, so it is not proof of cause in any individual case, but the pattern is consistent and dose dependent: the more adversity, the higher the risk.

Put those two findings together and the choice argument starts to look very strange. A child does not choose their genes, and a child does not choose to grow up in a house with violence, neglect or a parent who is drinking. Yet the willpower model holds the adult that child becomes fully responsible for the predictable consequences of both.

4. Addiction is usually a solution before it becomes a problem

This is the part that the willpower argument almost always misses. Most people who become addicted are not chasing pleasure; they are chasing relief. The physician Gabor Maté puts it well when he says the question is not why the addiction, but why the pain. Substances are extraordinarily effective, at least in the short term, at quieting anxiety, numbing trauma, filling emptiness and switching off the relentless commentary that some people live with every waking hour.

Matthew Perry described this with painful honesty. In his memoir and the interviews around it, he spoke about living with anxiety and depression, about a hole that fame, money and attention could never fill, and about how he had been in therapy since he was eighteen. He also said that he detoxed around 65 times, went to rehab at least fifteen times, attended roughly 6,000 AA meetings and estimated in interviews that he had spent around $9 million trying to get sober (the book itself puts it at $7 million). Whatever you believe about addiction, that is not a description of a man who lacked the will to stop. It is a description of someone who tried, again and again, with every resource money can buy, against something that was stronger than his ability to choose in the moment. He died in October 2023, at 54.

This is what I mean by the absence of meaningful choices. If the noise in your head is unbearable, and the only thing that has ever reliably turned it down is a drink or a pill, then the “choice” to abstain is a choice to sit in that noise with no alternative. For many people it is not a choice between a good life and a bad one; it is a choice between pain right now and pain later, and the brain is built to pick relief right now.

5. Change the environment and the “choice” changes with it

Some of the most powerful evidence that addiction is not purely about individual character comes from the Vietnam War. In 1971, US officials feared a wave of heroin addiction as soldiers came home, because heroin use in Vietnam was widespread. The psychiatric researcher Lee Robins led a study of returning soldiers, and a review of that work in the journal Addiction summarises the findings: 34% of soldiers had used heroin in Vietnam and 20% had symptoms of dependence, yet in the first year back home only 1% became addicted to heroin again, even though 10% tried it after their return. These figures describe one population in very unusual circumstances, so they should not be stretched to cover everyone with an opioid addiction, but they are striking.

The conventional expectation at the time was that most treated heroin users would relapse within a year, so the result was a genuine shock. The people were the same; their character had not changed on the flight home. What changed was the environment, the cues, the stress, the availability and the reasons for using. The lesson I take from Robins is not that addiction is easy to leave behind, because many of those soldiers carried trauma home with them, but that context shapes behaviour so powerfully that framing addiction as a pure act of individual will simply does not fit the data.

6. The honest complication: people do recover, and they make choices to get there

If I stopped here, I would be presenting only half of the evidence, and a technical reader would rightly push back. So here are the strongest counterarguments, taken seriously.

First, many people do recover, and many do it without formal treatment. The psychologist Gene Heyman has pointed out that across large US population surveys, most people who met the criteria for dependence on an illicit drug no longer did by around the age of thirty, that addiction has one of the highest remission rates of any psychiatric disorder, and that most of those people quit without professional help. Heyman uses this to argue that addiction is better understood as a disorder of choice than as a chronic disease.

Second, the brain disease model itself is contested. The American Society of Addiction Medicine defines addiction as a treatable, chronic medical disease involving complex interactions among brain circuits, genetics, the environment and a person’s life experiences. But the neuroscientist Marc Lewis, who was himself addicted to opioids, argues in The Biology of Desire that addiction is better understood as a form of deep learning: the brain doing exactly what it is designed to do, seeking reward and relief, and restructuring itself around the most powerful reward available. Lewis is equally critical of the pure choice model, and he acknowledges that the disease framing has helped many people.

I think both critiques make my position more precise rather than undermining it. People who recover do make real choices, but those choices become available when circumstances change: when the pain is addressed, when support and connection arrive, when something more meaningful than the drug comes into view, or simply when life moves on from the setting that drove the use. Population samples also include many milder cases, while the people who end up in treatment, like Perry, tend to sit at the severe end where natural recovery is far less common. Choice in addiction is not absent; it is narrowed, sometimes to almost nothing, and it widens again when the conditions around the person change. That is exactly why blaming someone for lacking willpower is so misguided. Willpower does not come from nowhere; it is built from circumstances, and blaming someone for lacking it is, in practice, blaming them for their circumstances.

7. Blame does not just fail; it makes things worse

There is a common belief that shame is a useful deterrent, that if people feel bad enough about their behaviour they will change it. The evidence points the other way. In a study published in Clinical Psychological Science in 2013, Daniel Randles and Jessica Tracy recruited around a hundred newly sober members of Alcoholics Anonymous, recorded them describing the last time they drank and felt bad about it, and followed up the 46 who returned, typically about four months later. Nonverbal displays of shame, such as slumped shoulders and a narrowed chest, predicted whether participants relapsed and how severe the relapse was, while self reported shame and guilt did not predict relapse. It is one small study and should be treated as one, and it shows an association rather than proof that shame causes relapse, but its conclusion is consistent with what clinicians have long observed: shame tends to drive hiding and avoidance, and for someone whose primary coping tool is a substance, hiding and avoidance usually mean using.

Relapse itself is also widely misunderstood. A frequently cited review in JAMA in 2000 by Thomas McLellan and colleagues compared drug dependence with type 2 diabetes, hypertension and asthma, and found that heritability, the role of personal choice and environment, and rates of medication adherence and relapse were broadly comparable across all of them. We do not tell a person with hypertension that their returning blood pressure proves they never wanted to get better; we adjust the treatment. Perry made a similar point in his own words when he said that losing your sobriety does not mean you lose all the time and learning that came before it.

So the willpower narrative is not a harmless opinion. If shame is associated with a greater risk of relapse, then telling someone “you choose to be like this” is not tough love backed by evidence; it may well be adding another obstacle to recovery.

8. What we should do instead

If addiction is the narrowing of meaningful choice, then the job of everyone around the person is to widen it. That applies to families and friends, but I think it applies with particular force to leaders, because organisations shape a large part of the environment people live in.

None of this is a substitute for treatment. Not everyone recovers because their circumstances improve; many people need medication, structured psychological therapy and long term clinical support, and withdrawal from some substances, particularly alcohol and benzodiazepines, can be medically dangerous without supervision. For opioid addiction the evidence for medication is especially strong: a 2017 BMJ meta analysis of 19 cohorts found that all cause mortality among people on methadone was 11.3 deaths per 1,000 person years in treatment against 36.1 after leaving it. That is an observational comparison, so some of the gap may reflect who stays in treatment, but the direction is consistent across studies. Environment and compassion widen the door; treatment is often what gets people through it.

1⃣ Replace judgement with curiosity. Ask what the person is trying to quiet before you ask why they will not stop. You will learn more, and they are far more likely to keep talking to you.

2⃣ Make help easy to reach and safe to ask for. If admitting a problem at work is career ending, people will hide it until it becomes a crisis. Confidential support, medical cover that treats addiction like any other chronic illness, and a clear path back to work all widen the set of available choices.

3⃣ Treat relapse as information, not a verdict. In any other chronic condition, a relapse triggers a change in treatment. It should do the same here.

4⃣ Change the environment, not just the person. The Vietnam evidence is a reminder that context matters enormously. Cultures built around heavy drinking, relentless stress and permanent availability are part of the system, and they are within our control.

5⃣ Watch your language. Words like “junkie” and “choosing to use” carry the shame that the evidence says makes things worse. Talking about a person with an addiction, rather than “an addict”, is a small change that keeps the human being at the centre.

Closing thought

The willpower argument survives because it is comforting. It tells those of us who have not been addicted that we earned our good fortune through strength of character, and it tells us that the people who are struggling could fix it if they only tried harder. The evidence tells a more humbling story: genetic differences account for a substantial share of the variation in who becomes addicted, childhood experiences nobody chooses shape that risk further, and addiction can impair the brain systems involved in self control, while shame is associated with worse recovery rather than better.

None of this means people with addiction have no agency. It means their agency has been squeezed by forces they did not choose, and that it grows back when the rest of us help change the conditions around them. Matthew Perry spent decades and millions of dollars fighting for his choices. The least the rest of us can do is stop telling people like him that they simply chose wrong.

If you or someone close to you is struggling with addiction, please reach out to a doctor or a local support service. In South Africa, the SADAG Substance Abuse Helpline is available on 0800 12 13 14 (or SMS 32312).

References

1. Agrawal A, Verweij KJH, Gillespie NA, et al. The genetics of addiction: a translational perspective. Translational Psychiatry, 2012. https://pmc.ncbi.nlm.nih.gov/articles/PMC3410620
2. Verhulst B, Neale MC, Kendler KS. The heritability of alcohol use disorders: a meta analysis of twin and adoption studies. Psychological Medicine, 2015. https://arc.vcu.edu/publications-2014-2016/the-heritability-of-alcohol-use-disorders-a-meta-analysis-of-twin-and-adoption-studies.html
3. Volkow ND, Fowler JS, Wang GJ, et al. Imaging dopamine’s role in drug abuse and addiction. Neuropharmacology, 2009.
4. Goldstein RZ, Volkow ND. Dysfunction of the prefrontal cortex in addiction: neuroimaging findings and clinical implications. Nature Reviews Neuroscience, 2011. https://pmc.ncbi.nlm.nih.gov/articles/PMC3462342/
5. Dube SR, Felitti VJ, Dong M, et al. Childhood abuse, neglect, and household dysfunction and the risk of illicit drug use: the Adverse Childhood Experiences Study. Pediatrics, 2003.
6. Hall W, Weier M. Lee Robins’ studies of heroin use among US Vietnam veterans. Addiction, 2017. https://pubmed.ncbi.nlm.nih.gov/27650054/
7. Heyman GM. Addiction and choice: theory and new data. Frontiers in Psychiatry, 2013. https://www.frontiersin.org/articles/10.3389/fpsyt.2013.00031/full
8. American Society of Addiction Medicine. Definition of Addiction, 2019. https://www.asam.org/Quality-Science/definition-of-addiction
9. Lewis M. The Biology of Desire: Why Addiction Is Not a Disease. PublicAffairs, 2015.
10. Randles D, Tracy JL. Nonverbal displays of shame predict relapse and declining health in recovering alcoholics. Clinical Psychological Science, 2013. https://www.psychologicalscience.org/news/releases/shame-about-past-alcoholism-predicts-relapse-and-declining-health-in-recovering-alcoholics.html
11. McLellan AT, Lewis DC, O’Brien CP, Kleber HD. Drug dependence, a chronic medical illness: implications for treatment, insurance, and outcomes evaluation. JAMA, 2000.
12. Perry M. Friends, Lovers, and the Big Terrible Thing. Flatiron Books, 2022.
13. Sordo L, Barrio G, Bravo MJ, et al. Mortality risk during and after opioid substitution treatment: systematic review and meta analysis of cohort studies. BMJ, 2017. https://doi.org/10.1136/bmj.j1550