Why did my interest in alcohol disappear alongside the food noise, and what does that overlap tell us?

I started semaglutide for weight and something I did not expect happened: I stopped drinking, more or less by accident. I was on four or five drinks most evenings for years and had made several deliberate and unsuccessful attempts to change that. Now I open a bottle, have half a glass, and forget about it. It is the same quality of experience as the food thing, an absence of pull rather than an act of restraint.

Two other things went quiet at the same time: compulsive online shopping, and a phone-scrolling habit that had been genuinely out of hand. Those two might be coincidence or reduced stress, I cannot tell.

What I want to know: is the alcohol effect documented, or is it anecdote? Is the mechanism the same one as the food effect, and if so what does that imply about what food noise actually is? And is anyone studying whether this holds up, because if it does it seems like a much bigger deal than the weight.

The alcohol effect is documented, including in a randomised trial, and the mechanistic overlap is the most likely explanation. That overlap is genuinely informative about what food noise is: it suggests the drug is acting on a general appetitive-motivation system rather than on anything food-specific.

The evidence, in order of strength

1. Randomised clinical trial. Low-dose semaglutide in adults with alcohol use disorder reduced drinking in a laboratory self-administration paradigm and reduced craving, with effects on drinks per drinking day [1]. Small and short, but randomised and controlled, which is a different evidential category from everything that preceded it.
2. Preclinical work across multiple laboratories and species showing that GLP-1 receptor agonists reduce alcohol intake, reduce alcohol-induced dopamine release in the nucleus accumbens, and reduce relapse-like drinking in animal models. Consistent, replicated, and mechanistically specific.
3. Observational and pharmacoepidemiological signals in large prescription datasets, suggesting lower rates of alcohol-related outcomes in people receiving these drugs. Suggestive, heavily confounded, and the kind of evidence that has been wrong before.
4. Self-report, which is where this began and which described the effect accurately years before the trials caught up. Worth noting as a rare case where community observation led the literature.

Reported effects on nicotine, and to a much weaker evidential standard on gambling, compulsive shopping and other appetitive behaviours, exist but sit firmly in categories 3 and 4. Your shopping and scrolling observations belong there: interesting, plausible, unverified, and vulnerable to the obvious confound that people who are losing weight and drinking less feel better generally.

Why the mechanism plausibly overlaps

GLP-1 receptors are present in mesolimbic structures including the ventral tegmental area and nucleus accumbens, and receptor activation modulates dopaminergic signalling there. That circuit is not a food circuit. It is the general-purpose system that assigns incentive salience to any learned reward cue: food, ethanol, nicotine, a slot machine, a notification. If a drug turns down the gain on that system, you would predict exactly the pattern you are reporting, and you would predict it to generalise unevenly depending on how much each behaviour depended on cue-driven wanting versus on habit, physical dependence or social context.

Which is a substantive claim about what food noise is: not a food phenomenon with an incidental resemblance to craving, but the same process applied to food. The subjective similarity you noticed between the alcohol effect and the food effect is not a coincidence of description. It is the same reduction reported through two channels.

What this does and does not imply

• It does not mean these are treatments for addiction. One small randomised trial in alcohol use disorder is a starting point, not an indication. Established treatments for alcohol use disorder have far more evidence behind them and remain the appropriate first line.
• It does not mean the effect will persist. The durability question is completely open for alcohol, and if the effect fades on a similar timeline to the partial fading people describe for food, someone who has restructured their life around not drinking could be blindsided.
• Stopping the drug is the risk point. If your reduced drinking is pharmacologically maintained rather than habit-maintained, then discontinuation carries a relapse risk that nobody is currently warning people about. That is worth taking seriously and worth building actual non-pharmacological structure around while you have the window.
• There is a genuine safety intersection. Reduced alcohol intake is good for you. But alcohol on a very low food intake with delayed gastric emptying behaves differently: intoxication comes on faster and hypoglycaemia risk rises, particularly for anyone also on insulin or a sulfonylurea. Several people report much lower tolerance and unpleasant reactions rather than reduced interest.

The honest summary

You have experienced something real that has randomised evidence behind it, arising from a mechanism that makes sense, in a research area that is early and moving fast. The appropriate stance is interest without over-claiming. If the reduced drinking matters to you, and it sounds like it does, treat the current period as an unusually favourable window in which to build the habits and supports that will hold when the pharmacology changes, rather than as a solved problem. That is the same advice as for the weight, for the same reason.

If alcohol has been a significant problem, this is also worth discussing with a clinician rather than managing alone, both because there are better-established treatments and because unsupervised reduction from heavy daily drinking carries its own risks.

edited 23 Apr 2026 by tobias_maartens — expanded the table to cover the lower concentration

Pushing back slightly on the strength of the generalisation, not because I think it is wrong but because the confounds here are unusually severe and the topic attracts over-claiming.

Alternative explanations for the cluster of behaviour changes OP describes, which are not mutually exclusive with the mesolimbic account:

• Alcohol is caloric and appetitive at once. Reduced intake of a caloric beverage is partly explained by the same satiety and volume-tolerance mechanisms that reduce food intake. Nothing reward-specific is needed for part of the effect.
• Physical unpleasantness. Many people report that alcohol simply feels worse: faster intoxication, worse reflux, worse nausea, worse next day. That is aversive conditioning via gastrointestinal effects, not reduced incentive salience, and it predicts a different durability profile.
• Behavioural cascade. Losing weight, feeling better, sleeping better and gaining a sense of agency reduces all kinds of coping behaviours. Drinking less improves sleep, which reduces scrolling and impulse spending. A single upstream change can produce a cluster of downstream ones without any shared receptor.
• Attention and expectancy. Someone who has noticed a dramatic change in food-related thoughts is primed to notice reductions elsewhere, and reporting is not blinded. This is exactly the mechanism that produces long lists of claimed effects for any drug with a striking primary effect.

The reason this matters practically: the four explanations have different implications for what happens when you change dose or stop. Reduced salience should scale with dose and persist. Aversion should fade with tachyphylaxis. A behavioural cascade should persist independently of the drug. Expectancy should not survive an unblinded change of circumstance.

So the useful question for OP is not which story is true in general but which is true for them, and the way to find out is to notice what happens to the drinking across the dosing week and after any dose change. A drinking pattern that tracks day 1-3 versus day 5-7 is telling you something different from one that is flat.

Adding a mechanistic detail that is relevant to why the effect might be stronger for some substances than others, and to why the food and alcohol reports are so much more consistent than the gambling and shopping ones.

Ethanol and food share something that gambling and scrolling do not: both are ingested, both are sensed by the gut, and both produce post-ingestive signals that travel to the brainstem by vagal and humoral routes. So for those two, a drug acting at gut receptors, vagal afferents, the area postrema and the nucleus tractus solitarius has direct access to the interoceptive signal that reinforces the behaviour, quite apart from any mesolimbic effect. There is a peripheral route as well as a central one.

For a purely cue-and-outcome behaviour like gambling or compulsive scrolling, only the central route is available. If the central mesolimbic effect is the weaker of the two, which the receptor densities suggest is plausible, you would predict exactly the observed gradient: strong and consistent effects on eating, strong effects on drinking, weaker and less consistent effects on smoking, and mostly anecdote for the non-ingested behaviours.

That gradient is itself a useful piece of evidence about the mechanism, and it argues against the most expansive framing of these drugs as general anti-compulsion agents. It also generates a testable prediction: if the gradient is driven by peripheral interoceptive access, then compounds in this class with different peripheral-to-central distribution should show different profiles across behaviours. As far as I know nobody has tested that directly.