The Science of Cravings: Food Noise on a GLP-1, and What Happens When You Stop

The fortnight I thought something was wrong with me

I stopped taking retatrutide on 20 June 2026. I'd taken it since January, and Mounjaro before that, and every dose is in my case study. My appetite came back once I stopped. I expected that. I didn't expect the noise.

For about two weeks I thought something had gone properly wrong with me. Not hungry, exactly. Obsessed. I'd be doing something else entirely and my brain would produce a fully rendered picture of one particular food, then keep producing it, like a browser tab I couldn't close.

My weekly check-in in Healthcount caught it. In the week of 9 August, about seven weeks after my last dose, its question about food cravings hit the top score.

What unsettled me most was how convincing it felt. It didn't feel like a symptom. It felt like a personality, as though the quiet months had been the fake bit and this was me going back to my factory settings.

One thing before I go on. Retatrutide isn't licensed anywhere. Lilly, which makes it, says it is "legally available only to participants in Lilly's clinical trials" (Lilly, May 2026), and the UK's medicines regulator warns that with products sold as retatrutide "there's no guarantee that they even contain retatrutide" (MHRA, July 2026). Nothing here is a reason to take it.

So I read about cravings, the way I read about everything now. The research turned out to be more specific than I expected, and more honest about what it doesn't know.

Food noise, cravings and hunger aren't the same thing

Food noise didn't start as a scientific term. It was made popular by people managing their weight, and researchers have only lately tried to pin it down. One group defines it as persistent thoughts about food that feel unwanted or upsetting, and that are more intense and intrusive than ordinary thoughts about what to eat (Dhurandhar et al., 2025). The phrase took off on TikTok and Reddit, usually in posts about GLP-1 medicines (Wylie et al., 2026).

It's common. In a survey of 1,000 US adults in May 2026, about half said they had food noise at least sometimes, and nearly 1 in 5 had it often or always. Having it often was more common in people who felt a loss of control over eating, or were restricting what they ate. It was also more common in people taking a GLP-1, or who had taken one: about 1 in 4 of them, against about 1 in 6 of people who never had (Hayashi and Masterson, 2026). A survey like that can't tell you what causes what.

A craving is narrower. It's an intense wanting for one particular food (Hill, 2007). In my experience, hunger builds slowly and almost any food settles it. A craving arrives in your head, it's about one thing, and an apple does nothing for it. You eat the apple and the craving's still there, now with an apple on top.

Two findings from the craving research surprised me. The old idea that a craving is your body asking for a nutrient it's short of hasn't held up. And trying to cut out one particular food goes with craving it more (Hill, 2007). Hold on to that second one. It comes back later.

What GLP-1 medicines do to cravings while you take them

For many people they turn cravings down. Not for everyone, though, and maybe not for good.

In a 12-week trial of semaglutide, the medicine in Wegovy and Ozempic, people had less hunger and fewer food cravings, and felt more in control of their eating, than on a placebo (Blundell et al., 2017). A 20-week trial of the 2.4 mg dose found fewer and weaker cravings (Friedrichsen et al., 2021). Novo Nordisk, which makes semaglutide, sponsored the 2021 trial, and its staff co-wrote both papers. For tirzepatide, the medicine in Mounjaro and Zepbound, a 6-week trial funded by Lilly found lower appetite, fewer food cravings and less reaction to food around them than on a placebo (Martin et al., 2025).

Researchers have started measuring food noise itself. In a 2026 study of 79 adults, people who'd been on a GLP-1 medicine for six months scored lower on a food noise questionnaire than people who were eligible for one but hadn't started (Rutigliani and Mattes, 2026). That's a snapshot of different people, not the same people followed over time. And when 30 people in the US who were taking these medicines, or had taken them, were interviewed, less food noise, or less appetite, was one of the main things they described (de Vere Hunt et al., 2026).

The quiet may not last, even on the medicine. In a 60-week trial, people on semaglutide reported less hunger and less preoccupation with food than the placebo group at week 20. By weeks 40 and 60 the two groups no longer differed on those, although the semaglutide group was still eating less at a test lunch (Tronieri et al., 2026). Two of its authors report a research grant from Novo Nordisk to their university. Reviewing the field in September 2026, Wylie and colleagues put it plainly: "little is known about what exactly food noise is, what causes it" (Wylie et al., 2026).

Why food noise can get louder when you stop

The honest answer starts with this: nobody has measured it. On 25 October 2026 we searched PubMed for studies that mention food noise alongside stopping or cutting down a GLP-1. There were 8. None measured food noise before and after someone stopped. A 2026 review suggests watching for early changes in appetite and preoccupation with food after stopping, and says these ideas "require prospective validation" (van Bruggen et al., 2026).

What we do know points one way. These medicines work partly by turning appetite down (NHS, Tirzepatide), and once you stop they leave your body over a few weeks. Tirzepatide's half-life is about 5 days (Mounjaro SmPC, 2026) and retatrutide's about 6 (Urva et al., 2022). Our guide to how long Mounjaro stays in your body works through the arithmetic.

In the trials where people stopped semaglutide or tirzepatide, they regained, on average, much of the weight they had lost within a year (Wilding et al., 2022) (Aronne et al., 2024). Weight was what those trials measured, not cravings, but I find it hard to see how that happens without appetite coming back. The 2026 review describes it as possibly "the re-emergence of biological pressures favouring weight restoration rather than treatment failure" (van Bruggen et al., 2026). And in the US survey, people who had stopped a GLP-1 reported frequent food noise about as often as people still taking one (Hayashi and Masterson, 2026). Again, a survey can't say why.

So if you stop and the noise comes back loud, it isn't a sign you've failed. You've stopped taking a medicine that was turning the volume down. Our guides to what happens when you stop and coming off a GLP-1 cover the practical side.

How a craving works: an automatic thought, then a picture

One of the best-known theories of craving is the elaborated intrusion theory. It comes from a 2005 paper with the best title in the field, "Imaginary relish and exquisite torture" (Kavanagh, Andrade and May, 2005). It splits a craving into two steps.

Step one is the intrusion. A thought about food pops up on its own, set off by something around you, something in your body or another thought (Kavanagh, Andrade and May, 2005). You don't choose it.

Step two is the elaboration. If the thought is a pleasant one, you start building on it. You picture the food, how it tastes, where it is in the cupboard. Those pictures feel good for a moment, but they also make you more aware of what you're missing (Kavanagh, Andrade and May, 2005). That's the exquisite torture.

Ten years on, the theory's authors looked back at the evidence. The split between automatic intrusions and their elaboration is "well established", and tasks that bring in competing sights or sounds get in the way of the elaboration, especially when they use the same limited mental space (May, Kavanagh and Andrade, 2015).

That changed more for me than any willpower advice. The thought arriving isn't a verdict on me. What happens in the next half a minute is where anything gets decided.

Why Tetris works better than the television

If cravings run on mental pictures, anything that keeps your mind's eye busy should crowd them out. Researchers at Plymouth tested that in everyday life. Thirty-one students carried iPods for a week, and a text message asked them 7 times a day what they were craving and how strongly. Fifteen of them then played Tetris for 3 minutes and rated the craving again. Their cravings fell by about 14 points on a 0 to 100 scale. That held for food and drink, for alcohol, nicotine and caffeine, and for things like exercise and gaming (Skorka-Brown et al., 2015).

Three minutes of falling blocks, outside a lab. I still find that surprising. It also explains something I'd noticed and misread. Sitting in front of the television never did anything for my cravings, and I took that to mean distraction didn't work for me. The study didn't test television. My guess, from the theory, is that it asks too little of your mind's eye: you can watch a whole episode and picture a bag of crisps the entire time.

What I do now

None of this is a treatment plan. It's what I landed on after that fortnight, built on the two-step idea.

I name it. Out loud sometimes, which is ridiculous, but it works. "That's a craving." Calling it an event rather than a fact about me stops the second step about half the time.

I give my hands and eyes something to do. Not the television. A game, a puzzle, my language app, anything I have to look at and press. Three minutes is usually enough, which I wouldn't have believed before I read the Tetris study.

I say yes, in 25 minutes. Not no. Yes, in 25 minutes, if I still want it. Saying no starts an argument, and arguing means picturing the food. It also makes the food forbidden, and cutting out a food goes with craving it more (Hill, 2007). Most of the time the 25 minutes passes and I've forgotten. When I haven't, I eat the thing and it's fine. I use the same timer at the end of meals, which I wrote about in how I learned to eat more slowly.

I write it down. Partly because that's what I built Healthcount for, and partly because a craving that's written down is already a bit outside my head. Mine cluster in ways I'd never have guessed from memory.

I still get them. That's the honest ending. What changed is that they stopped feeling like evidence of something wrong with me, and started feeling like weather.

When it's more than noise

Food noise can sit close to something more serious. In the US survey, the strongest link with frequent food noise was feeling a loss of control over eating (Hayashi and Masterson, 2026). A 2026 article in the International Journal of Eating Disorders warns that the same effects that quieten appetite and preoccupation with food may, in some people, feed restriction, skipped meals and relapse, and asks for plans for stopping (Škudar, 2026).

If cravings come with bingeing, feeling out of control, or eating very little to make up for them, that's worth talking to a doctor about. Our piece on why regular eating works looks at how regular meals can help break a binge and restrict cycle, and our review of Overcoming Binge Eating covers a self-help programme with trials behind it.

What's solid and what isn't

Questions people ask

This is one person's story and general information, not medical advice. Decisions about your medicine belong with your prescriber.