BehavioralAddictions.com

Evidence-based insights on behavioral addictions

The Patient I Didn’t Recognize

Tariq M. Ghafoor, M.D. on July 22, 2026

A few years ago, a man stopped me in a public place, somewhere ordinary, the kind of errand that fills a weekend. He was in his forties, healthy, well-dressed, easy in his manner. He called me by name and asked if I recognized him. I didn’t, and I said so.

He told me he had been my patient. For two years he had come to see me regularly. During that time he had been hospitalized more than once, diagnosed with a serious mood disorder, and treated accordingly. He had done everything asked of him. He came to his appointments, often with his wife beside him. He took what I prescribed. And he had not gotten better in the way any of us hoped.

The man standing in front of me bore little resemblance to the patient I had known. He told me he had been off all medication for several years and was, in his words, doing just fine. He had come to believe he was never mentally ill at all. His real problems, he said, had been gambling and compulsive sexual behavior. When he finally got help for those, when he finally named them, his life began to change, and the psychiatric symptoms gradually faded.

I have thought about that conversation many times since.

For two years, the things that were actually consuming his life never entered my office as the problem. We talked about mood, about sleep, about energy and despair. We did not talk about what he was doing at night, because he did not raise it and I did not know to ask in the right way. The gambling and the rest stayed outside the conversation and what stays outside the conversation, in this work, may as well not exist. I treated what was speakable. I missed what was load-bearing.

I don’t offer this as a confession of unusual carelessness. He was seen weekly, by someone experienced, with a spouse in the room who loved him and wanted him well. If the real trouble can hide under those conditions, it can hide anywhere.

What strikes me, looking back, is that this is the central difficulty with behavioral addictions and it is not the one the public worries about. We are used to thinking of addiction as something with visible signs: a substance, a smell, a slurred word. Behavioral addictions have none of that. And the behaviors underneath them are, at the harmless end, entirely normal and often admired. We praise the person who works without rest. We respect discipline that looks like exercise and ambition that looks like striving. The line between devotion and compulsion is not merely blurry; it is hidden behind approval. No one stages an intervention for a behavior the culture rewards.

There is a further problem. Because most of these behaviors have no settled place in our diagnostic language, they do not announce themselves as what they are. They borrow the appearance of conditions we do recognize. A man drowning in compulsion and its consequences does not arrive saying “I have a behavioral addiction.” He arrives anxious, sleepless, despairing, erratic, and we, trained to recognize mood and anxiety disorders, see mood and anxiety disorders. We are not wrong, exactly. We are treating the distress in front of us and missing what is generating it.

Key Insight

The most dangerous behavioral addictions are not the ones that look extreme. They are the ones that look like something else — or like nothing at all, because no one has yet found the words to call them a problem.

One thing I have learned, slowly, is to distrust the surface. The same behavior, a young man who plays games every waking hour, say – can mean entirely different things in two different people. In one, it is the cause of a life coming apart. In another, it is the response to a life that already had. In one it is a reaching toward something: absorption, mastery, relief. In another it is a flight from something he cannot sit still with. I cannot read which it is from the hours logged or the money spent. Those numbers are what the world counts, and in my experience they are precisely what mislead. Two people can behave identically, one dragged to my office by worried family, the other cushioned by circumstance and never spoken to at all, and the difference between them is often not the behavior but who around them was inconvenienced enough to say something.

So what I attend to, instead, is which of those meanings is operating in the person in front of me. And I try to stay honest that it is frequently more than one at once.

Which brings me back to the man who stopped me. He believes one thing was wrong with him, and that he cured it. My impression is different from his; though I would not claim to be certain. I suspect he did have a real vulnerability, a fault line that was always there, quiet and unstressed. The addictions and their consequences were the load placed upon it, and under that load the fault slipped, again and again, in the form of the crises I treated. When he finally lifted the weight, the fault stopped moving. From where he stands, the illness is gone. My impression is that it was unburdened, not erased. It may well hold for the rest of his life, and I hope it does, but I would not assume the fault line is gone. Only that, for now, nothing is pressing on it.

He is certain. I am not. And I have come to think certainty is the more dangerous of the two positions, because it stops a person from staying watchful.

If this site has a reason to exist, it begins here; with a reluctance to accept that these problems have a single cause, a single name, or a single cure. I treated a man for two years and saw one layer of him clearly while another, underneath, did the real damage. Looking back, the real issue may be this: not that I failed to pay attention, but that I did not adequately consider how many things can be true of one person at the same time. That, more than any statistic, is what I hope to keep from repeating.

The Symptom That Won’t Sit Still

Tariq M. Ghafoor, M.D. on July 22, 2026

The study


Horváth Z, Bőthe B, Potenza MN, Stein DJ, Demirgül SA, Paksi B, Czakó A, Demetrovics Z. “Compulsive sexual behavior may be more state-like than trait-like: Findings from a three-year longitudinal representative survey.” Journal of Behavioral Addictions (2026). doi.org/10.1556/2006.2025.00419
What they did. Drawing on a representative, four-wave young-adult cohort in Budapest (782 participants, tracked across roughly three years), the authors measured compulsive sexual behavior with the short Hypersexual Behavior Inventory and applied a latent state–trait model. That method splits each person’s score into a stable, dispositional layer and a moment-to-moment, situational layer: the first formal attempt to do so for this construct.
What they found. Most of the variation, roughly 69–93% of the reliable signal, sat in the situational layer; only about 7–31% reflected a stable disposition. Scores carried over from one year to the next (moderate-to-strong autoregressive effects), and overall severity drifted modestly downward across the three years. In plain terms: in this general-population sample, the behavior behaved more like a state than a trait.

My take

What strikes me here is less the headline than the sample it comes from. This is a community cohort, and a fairly untroubled one; in any given wave, most respondents reported no symptoms at all. So when the authors say the behavior is mostly “state-like,” they are describing what moves in people who are, for the most part, not in trouble. That is exactly where I would expect situation to dominate.

I keep coming back to the same way of reading these presentations. Almost nothing in this domain has a single cause. There is a predisposition, call it the fault line: temperament, early experience, how a given nervous system handles distress. There is the load that lands on it: stress, loneliness, a relationship coming apart, easy access at 2 a.m. And there is the environment that decides how much of that load actually reaches the line. The symptom is what you see when load meets line. In a community sample, you are mostly looking at modest fault lines carrying ordinary, fluctuating loads. What you measure year to year is the weather — the load moving — not the bedrock. The study is, in effect, measuring the weather and correctly reporting that the weather changes.

That is the part worth holding onto, because it cuts both ways. The clinic is not a random sample of the community. It is a filter that concentrates the steep fault lines: the people for whom the behavior has stopped tracking circumstance and started running on its own. My impression is that if you ran this same analysis on a treatment-seeking population, the stable layer would carry far more weight. The authors say as much in their cautions; I think they are right to. A finding that something is “mostly state” in the general population is not the same as saying it is mostly state in the person sitting across from you, who found their way to a psychiatrist precisely because it would not pass.

Key Insight

“State-like in the community” and “trait-like in the clinic” are not a contradiction — they are what you get when treatment-seeking filters for the steep fault lines.

There is a clinical use for this beyond the taxonomy debate. If much of what we measure in a lower-severity patient is situational, then a snapshot, one questionnaire, one bad month, is a poor basis for a durable label. The same logic that makes the construct look unstable in a survey is the logic that should make us patient at intake: watch the course, separate the load from the line, and resist the urge to convert a hard stretch into a fixed identity. The six-month threshold in the diagnostic guidance exists for roughly this reason, and this study is a quiet argument for taking it seriously rather than diagnosing off a single elevated score.

This connects to something I wrote about in The Patient I Didn’t Recognize: the way a presentation can read as one thing on one day and as something else once you watch it move. A model that lets the situational and the dispositional be measured separately is, to me, just a formal version of what careful follow-up already teaches.

Where this sits: established vs. emerging

Established. Compulsive sexual behavior disorder is a recognized diagnosis in the ICD-11, classified as an impulse-control disorder and defined by impaired control over sexual urges or behavior that continues despite distress or clear negative consequences. That much is settled.

Emerging / debated. Almost everything about its course and category is still open. Whether it is best understood as addictive, impulsive, or obsessive–compulsive in nature remains unresolved. The state-versus-trait question is newer still: this is the first study to partition the construct this way, it rests on a single representative young-adult sample in one country, it relied on a screening measure rather than a full diagnostic interview, and its own authors flag that higher-severity or clinical populations may look quite different. Useful, genuinely novel, and not the last word.

The Algorithm Didn’t Need to Aim

Tariq M. Ghafoor, M.D. on July 9, 2026

The piece

Petrovskaya E, Khoo N, Xiao LY, Leahy D, Roberts A. “Gambling adverts on social media reach 2.3 times more men than women: Using the Meta Ad Library to assess gambling advertising in Ireland.” Journal of Behavioral Addictions (2026). doi.org/10.1556/2006.2025.00484

Researchers used the Meta Ad Library, a public ad repository the EU’s Digital Services Act now requires large platforms to maintain, to examine 411 gambling adverts from 88 licensed operators in Ireland. The question was not only who the ads were aimed at, but who actually saw them.

My take

The usual worry about gambling advertising is intent: that operators deliberately aim at the people least able to absorb the harm. What makes this study worth sitting with is that it found the harm-shaped pattern without the intent. Only about a fifth of the ads explicitly targeted men, and none targeted women alone; most were simply set to reach everyone. Yet men were reached more than twice as often as women, and most of that gap came from ads that never asked to be shown to men. The skew was produced by the platform’s delivery algorithm, not by the advertiser’s targeting.

That distinction matters more than it first appears. The system was not handed a list of the vulnerable; it assembled one as a by-product of chasing engagement. In Ireland, men aged 25 to 34 carry the highest rate of problem gambling of any group, and that is almost exactly the group these ads reached most. What is striking is that the delivery concentrated exposure in the very demographic already known to be at highest risk, without any operator having asked it to.

Key Insight

An engagement-optimizing system doesn’t need to target the vulnerable; it finds them on its own, because to the algorithm, vulnerability and engagement look identical.

I would resist the cleanest version of this story, though, because the mechanism is a loop rather than a push. The algorithm is not seeking out vulnerable men out of malice; it is optimizing for engagement, and men engage more with gambling ads, so the system learns to send them more. The vulnerability and the exposure feed each other. That is, in a way, worse than deliberate targeting, because there is no single decision to regulate away — and it is exactly why the authors are right that banning explicit gender targeting would not fix this. A skew the algorithm generates on its own survives a rule aimed only at what operators consciously ask for. The delivery mechanism itself is the thing that needs looking at. The limits are worth stating plainly. This is one country, a small slice of all the gambling ads in circulation, and platform data that researchers cannot independently verify. The numbers also predate Ireland’s 2024 gambling law, which since March 2025 restricts these ads to users who opt in, so the snapshot is of a pre-regulation moment rather than the present. And the data cannot tell us whether the men reached were new recruits or existing customers being pulled back in. None of that softens the observation that travels well beyond Ireland: when exposure follows engagement, the people who get the most of it are often the ones who can least afford it.

When the Clock Is the Wrong Instrument

Tariq M. Ghafoor, M.D. on July 8, 2026

The study

Bonnaire, C., & Lambert, L. (2026). From healthy to problematic gaming use: Considering the role of age. Journal of Behavioral Addictions.

This commentary makes two arguments about the line between healthy and disordered gaming. First, time spent playing is a poor diagnostic signal: hours alone fail to separate the passionate gamer, the professional, and the person in real trouble, and treating duration as the marker leads to both over- and under-recognition. Second, age matters — gaming begins in childhood, is legal across the lifespan, can benefit development, and accompanies people through stages when the brain’s control systems are still maturing. The authors argue that diagnostic criteria should bend to developmental stage rather than apply uniformly.

My take

They are right, and the first point is one I keep arriving at from the other direction — the clinic. The count is the seduction; the function is the finding. What a person plays away from, and what the playing costs them, tells me far more than the number of hours ever will. But I would offer three refinements from practice, because the paper, in its proper caution against overpathologizing, lets go of more than I would.

First, time is the wrong metric — until it isn’t. At the extreme, quantity becomes its own quality. An adolescent, or an adult, who plays fifteen or more hours a day, day after day, is not a case where hours are merely uninformative; at that magnitude nothing else in a life survives, and the number has become the finding. The field has corrected so hard against time that it risks waving away the outlier it should not.

Second, the signal is rarely time itself — it is time measured against what the time costs. Four hours on a school night and eight on a summer afternoon are not the same behavior scaled up and down. The summer eight may displace little; the school-night four can cost sleep, study, and the next morning. The same hours read as benign or alarming depending entirely on what they take the place of. That is the reading the raw count obscures.

Third, developmental stage matters more than the paper’s own instrument can capture — because chronological age is a poor proxy for developmental maturity. The argument to adapt criteria by stage is correct in spirit, but the variable that matters is the maturity of the regulatory system, and that is not reliably indexed by the birthday. It is not uncommon to see a sixteen-year-old more developmentally advanced than a twenty-two-year-old. “Impaired control” has to be judged against the control the person actually has, not the control their age is presumed to confer.

Key Insight

Hours are not the measure. What the hours cost — and what they take the place of — is the measure.

None of this contradicts the paper; it sharpens it. The clock is the wrong instrument for the ordinary case, a necessary one at the extreme, and useful in between only when read against context and against the person’s real, rather than assumed, developmental stage. As ever, the behavior is not the finding. What it costs, and who is doing it, is.

The Dose Was the Point

Tariq M. Ghafoor, M.D. on July 8, 2026

The piece

Camacho-Barcia L, Jimenez-Murcia S, Granero R, … Fernández-Aranda F. “Comprehensive analysis of the relationship between ultra-processed food consumption and food addiction at one-year follow-up in older adults with metabolic syndrome.” Journal of Behavioral Addictions 15(1), 471–487 (2026). doi.org/10.1556/2006.2025.00363

A year-long look at 429 older Mediterranean adults with overweight or obesity and metabolic syndrome, asking whether eating fewer ultra-processed foods tracks with less food addiction on the Yale Food Addiction Scale. The intervention sat inside the PREDIMED-Plus-Cognition study; ultra-processed intake was scored with the NOVA system and split into thirds.

My take

The headline most people will take from this, that ultra-processed food is addictive, is not quite what the study shows, and the authors are careful about that. Two findings struck me as more useful than the slogan. First, over the year, the people who cut ultra-processed food the most were the ones whose food-addiction scores actually came down; a modest trim did little. Second, that improvement showed up independent of weight loss. The grip loosened even when the scale had not moved much.

What I appreciate is that the study does not oversell itself. The pull toward these foods, in my experience, is rarely about the food in isolation; it sits somewhere between how a particular person is wired to respond to engineered combinations of sugar, fat, and salt, and how relentlessly the surrounding environment keeps putting those combinations within reach. What a year of reduced intake seemed to do was ease that pull for the people in whom it ran strongest, without any claim to have changed the person underneath.

Key Insight

Cutting ultra-processed food seemed to loosen food addiction’s grip independent of the weight lost — but only for those who cut deeply, not cosmetically.

I would temper the enthusiasm in two places. The sample barely had the condition in question: only about 5.6% met the threshold for food addiction, below what is reported in the general population, so this is a low-signal group and the effect was modest. And the design is observational, with both study arms already being steered toward better eating, so some of the improvement belongs to the intervention itself rather than to the ultra-processed cut specifically. The authors say plainly that ultra-processed intake alone did not explain the total addiction score; the food is part of the picture, not the whole of it.

Where it lands clinically, for me, is that weight-independent finding. One thing I have noticed over the years is how often patients treat the scale as the sole measure of progress; when the weight stalls, they conclude nothing has changed. Yet the relationship to food may already be shifting in ways that matter. This study is a small piece of evidence that the grip can loosen before the weight has visibly moved, provided the change is real rather than cosmetic. The translation I would offer a patient is narrow and concrete: aim at the most engineered foods specifically, and cut deeply enough to matter, because the data here only rewarded the largest reductions.

A New Surface for an Old Vulnerability

Tariq M. Ghafoor, M.D. on July 5, 2026

A patient forms a fixed, false belief around a chatbot — that someone they love is hidden inside it, that it knows their thoughts before they type them, that its responses are a private channel meant only for them. There is a phrase for this now: “AI psychosis.” The implication is that a new technology has produced a new illness. I am skeptical of that framing, and the skepticism is the reason I wanted to write this.

What strikes me, reading the early clinical reports, is how little of the underlying mechanism is actually new. I have spent a long time watching people build beliefs they will not put down, and the most instructive cases are rarely the dramatic ones. They are the quiet ones, in which a vulnerable person and a responsive other slowly assemble a false structure together, with no ill intent anywhere in the room.

We have a clinical precedent for this, and it has nothing to do with machines. Think of the inexperienced or dogmatic therapist who, meaning only to help, inadvertently feeds a patient’s psychopathology — the era of recovered “memories” is the cautionary example, but milder versions happen constantly. A clinician who validates without ever introducing friction, who completes the patient’s interpretation instead of testing it, can help a suggestible person construct a vivid, fixed, and entirely false account of their own life. The mechanism that “AI psychosis” describes is, in its bones, this old one: a responsive other who reinforces a belief in one direction and never supplies the resistance that would let reality back in.

So if the mechanism is old, what has changed? My impression is that the answer is not a new capability but the removal of the limits that used to make the old one rare. Consider what kept even a misguided therapist’s influence bounded. It cost money and required showing up, so exposure was rationed. The therapist was another mind, with their own fatigue, their own doubt, their own occasional sense that something did not add up. A third party was always possible — a colleague, a spouse, someone who could say this is making you worse. And the therapist went home; the session ended; the patient had hours alone in which a distorted belief could quietly decay. A conversational system removes all four. It is tireless, it has no countervailing doubt, no supervisor sits in the loop, and it never goes home. The nightly pause that once let a belief lose its grip is gone.

There is, though, one element that does seem genuinely new in kind, and it is the part I find most worth sitting with. The dogmatic therapist at least held still. A rigid belief is stable; it pushes in one direction and stops where its own conviction stops. A system optimized to keep you engaged has no fixed point of its own. It does not hold a position you must move toward — it moves toward you. Each exchange, the person adjusts slightly to the machine, and the machine adjusts to the person, and with no external reference anchoring either of them, the two can drift together into a belief that neither would have reached alone. This is not the transmission of one party’s conviction to another. It is something closer to mutual entrainment — a belief that ends up co-authored, held jointly, and for that reason almost impossible to dislodge from inside the loop.

Key Insight

The right question to ask about these systems is not whether they push back, but whether they are anchored to anything outside their relationship with you. A good clinician is anchored to the patient’s welfare, which is sometimes at war with the patient’s wishes — and that war is the treatment. A system anchored only to your engagement has no such war to offer.

I should be honest that this is not a danger I can place entirely on the far side of the desk, in the patient. The same dynamic operates, in milder form, in how any of us now use these tools — and I include myself. I have thought through difficult ideas in dialogue with such systems. What kept that useful rather than distorting was not a property of the machine. It was what I brought to it: a specific question rather than an unmet need, the habit of reality-testing worn in over decades, and enough stability that I was reaching to sharpen a thought, not to be soothed. The identical tool, met by the same person in a lonelier or more fragile hour, tilts toward the consulting room. The machine does not change. What changes is the load the person is carrying when they sit down with it.

That is the part the public conversation keeps getting wrong, and it is the same error I find myself returning to again and again. We want to make the object safe — to regulate it, to tune it, to design the danger out of it. But the safety was never really located in the object. It is a property of the person and their circumstances: whether they come carrying a wound or a question, and whether anyone in their life — including themselves — is positioned to notice the drift before it sets. A system can, in principle, be anchored to something outside the loop. But that anchor has to be supplied by the user, and it asks for exactly the capacities that loneliness and vulnerability erode. The people best able to keep these tools safe are often the ones who least need protecting; the people most at risk are precisely those least equipped to install the brake.

I do not think conversational AI is making people ill in any simple sense, any more than I think the man I once treated was simply mentally ill or simply addicted. These things are layered. A predisposition, a load, an environment that happens to reinforce the wrong interpretation — braided together until pulling one thread moves all of them. What I distrust is the confident version of the story in either direction: that the technology is harmless, or that it is the cause. My impression is that it is neither. It is a new and unusually frictionless surface for a very old human vulnerability — and the work, as ever, is to keep watching the person, and the load, rather than the tool.