Why Confrontation Fails and Connection Works: What the Science of Craving Actually Says About Relapse

In January, actress Natasha Lyonne — sober since 2006, after a heroin addiction serious enough to require open-heart surgery — posted something most public figures spend enormous effort avoiding: she told her own followers she’d relapsed. “Recovery is a lifelong process,” she wrote. “Anyone out there struggling, remember you’re not alone.” Two months later, she said she was back on her feet, crediting the people around her, and asked, simply, to keep the details private. No press tour. No redemption arc staged for cameras. Just a plain, public admission that the thing she’d already beaten once wasn’t actually finished with her — and it might never fully be.

If you have ever relapsed and believed, even for an hour, that it erased every day you’d put together before it: it didn’t, and a brain-imaging study published this year can tell you exactly why the belief is wrong.

The craving isn’t a moral failure. It’s a measurable change in how your brain weighs a decision.

Researchers at Yale, led by Xiaosi Gu, built a computational model of how people with addiction actually learn from reward and punishment — the basic feedback loop underneath every decision anyone makes, from what to eat to whether to pick up. What they found, published this year, is that craving doesn’t just make a drug feel more appealing in the moment. It changes the math your brain runs to decide anything — and it changes it differently depending on what you’re craving. In people with alcohol use disorder, craving sped up how fast they updated their beliefs based on new information, for better or worse. In people with cannabis use disorder, it slowed that same learning down. Same mechanism, opposite direction, depending on the substance. “Addiction is very hard to treat,” Gu’s team wrote, “and one of the reasons is that we don’t fully understand the craving linked with it.”

Here’s the plain-language version, because the finding matters more than the jargon around it: craving isn’t a request your brain politely submits and you either grant or deny through strength of character. It’s closer to your brain temporarily recalibrating how it processes evidence, in a specific, substance-dependent way, without asking your permission first. That recalibration is why “just don’t” has never worked as addiction treatment for anyone, ever, at scale — you’re not arguing with a preference. You’re arguing with a system that has, for the duration of the craving, changed how it does arithmetic.

What actually moves the number: not confrontation. Connection.

This is where Maia Szalavitz’s reporting lands with more weight than a single opinion piece usually carries. Szalavitz has spent decades writing about addiction from a position almost no addiction journalist occupies: she lived through her own, cocaine and heroin, in the 1980s, and has spent the years since turning that experience into some of the most cited addiction journalism in the country. Her argument, made most recently in an essay on why compassion outperforms confrontation, is not sentimental — it’s closer to a clinical claim: “Feeling valued by others is what tends to prompt recovery.” Not shame. Not an intervention staged to make someone feel cornered. Not a family member listing every way addiction has disappointed them, however true each item on that list might be. Being treated, consistently, like you are still worth someone’s effort.

It’s also why the word “recovery” itself has become a live argument inside the field, not just a label. Helen Redmond — a clinical social worker and longtime addiction journalist — has pushed back hard against conflating recovery with total abstinence, arguing that defining recovery that narrowly writes off everyone whose progress doesn’t look like a clean, permanent line. If connection is what predicts getting better, a definition of “better” that excludes anyone who slips even once is working against the exact mechanism the Yale data points to — it turns a relapse into a disqualification instead of the data point it actually is.

There’s a mechanism behind why connection specifically works, and it has a name researchers have used for decades: vicarious mastery. Watching someone who was exactly where you are get out is more convincing than any doctor, any pamphlet, any lecture about consequences — because it’s proof the thing is survivable, delivered by someone who has no professional stake in convincing you, coming from inside the experience instead of above it. A systematic review of peer recovery support services published this year, drawing on more than two decades of studies, found consistent improvement in treatment engagement and retention wherever peer support was built into care — someone further down the same road, showing up not as an authority but as evidence.

There’s a mechanism behind why connection specifically works, and it has a name researchers have used for decades: vicarious mastery.

If you are the person still in it, here’s what this actually means for you, not in the abstract: the research is not telling you to wait for willpower to arrive. It’s telling you that the thing most likely to move you isn’t a threat. It’s a person who has been exactly where you are and is still standing close enough for you to see it.

What this means the next time someone you love relapses

If you’re the family member, the caseworker, the friend on the other end of a call that starts with “I messed up” — this is the part worth carrying into that conversation. The instinct to confront, to list consequences, to make the moment feel as bad as the choice that caused it, is understandable and almost always counterproductive. It treats a measurable change in brain-level decision-making as if it were a lapse in effort that shame could correct. It isn’t. What the Yale data and Szalavitz’s reporting both point toward, from completely different directions — one a brain-imaging lab, one a lifetime of reporting from inside the disease — is the same intervention: stay reachable. Don’t disappear the moment things get hard. The connection you maintain through a relapse is doing more clinical work than the confrontation you’re tempted to reach for instead.

Natasha Lyonne didn’t owe anyone an explanation for her relapse, and she gave almost none. What she gave instead was proof of the exact thing the research describes: that the story doesn’t end at the first slip, that staying connected to people — publicly, in her case, though it rarely needs to be — is what the return actually looks like. Not a straight line. A person, still reachable, still worth reaching for.