Food Noise Explained: What It Is and What the Research Now Shows

“Food noise” started as something patients said. It is now a measured construct with two validated questionnaires, at least one controlled comparison, and an active methodological argument in the scientific literature about whether it is a genuinely new idea or a rebranding of things researchers were already measuring.

That whole arc took about three years. This page walks through where it stands.

What does food noise actually mean?

Two research groups have published formal definitions, and they are close but not identical.

The Food Noise Questionnaire team, led by Hanim Diktas at Pennington Biomedical Research Center, defines food noise as “persistent, intrusive thoughts about food that are disruptive to daily life and make healthy behaviors difficult” [1].

The group behind the RAID-FN Inventory — Emily Dhurandhar and colleagues, writing in Nutrition & Diabetes — defines it as “persistent thoughts about food that are perceived by the individual as being unwanted and/or dysphoric and may cause harm to the individual, including social, mental, or physical problems.” They add a useful distinction: food noise is separated from ordinary food-related thinking by its intensity and intrusiveness, and it resembles rumination [2].

That word — rumination — is the one that tends to land for people who have experienced it. Not “I am hungry.” More like a background process that will not close.

The patient version, as described in clinical settings and the media before any of this was formalized, is a constant inner dialogue about food: what to eat, how much, when the next opportunity is [3].

Where did the term come from?

Food noise entered the scientific literature around 2023, essentially imported from patient language after GLP-1 receptor agonists became widely used for obesity. A narrative review published in Nutrients in November 2023 was among the first to treat it seriously, proposing a conceptual model built on food cue reactivity — the Cue-Influencer-Reactivity-Outcome model — and noting that anecdotal reports from patients and clinicians alike pointed to a reduction in “what has been colloquially termed food noise” [3].

The order of events matters here. Patients described an effect. Researchers then went looking for a way to measure it. That sequence is unusual and it is the source of most of the current methodological criticism.

How is food noise measured?

Two instruments now exist.

The Food Noise Questionnaire (FNQ), published in Obesity in January 2025 and developed in a sample of 400 US adults (registered as NCT06315907). It has five items, each scored 0-4 on a five-point agree/disagree scale, for a total of 0-20. Higher is worse. The items are [1][4][5]:

1. I find myself constantly thinking about food throughout the day.
2. My thoughts about food feel uncontrollable.
3. I spend too much time thinking about food.
4. My thoughts about food have negative effects on me and/or my life.
5. My thoughts about food distract me from what I need to do.

Its psychometrics are strong: internal consistency Cronbach’s alpha 0.93, test-retest reliability r = 0.79 over about a week, a clean single-factor structure, good convergent validity against food-preoccupation measures and good discriminant validity against mood, anxiety and stress questionnaires [1].

The RAID-FN Inventory (Ro Allison Indiana Dhurandhar Food Noise Inventory), developed from a 29-item draft and validated as a seven-item instrument across three domains: preoccupation, persistence and dysphoria. It also reports good internal consistency, test-retest reliability and convergent validity [2].

The RAID-FN’s three-factor structure arguably captures more nuance than the FNQ’s single factor. Neither has yet been shown to add clinical value beyond existing scales — which is the crux of the criticism below.

Does the research show GLP-1 drugs reduce food noise?

The strongest evidence available is an observational cohort comparison presented as a poster at the 33rd European Congress on Obesity in Istanbul in May 2026. It has not yet been published as a peer-reviewed paper, and the details below come from the presenting institution’s release [5][6].

Researchers followed 417 adults in a commercial digital behavioral weight-management program — WeightWatchers — and measured one-month change on the FNQ. Ninety-two started a GLP-1 receptor agonist alongside the behavioral program (the “Med+” group); 325 received the behavioral program alone (“Core+”). Anyone already using a weight-loss medication at baseline was excluded [6]. Nobody was randomized: people who started a drug chose to, which is the central limitation of the design.

After adjusting for baseline scores, mean FNQ change was:

• GLP-1 plus behavioral treatment: -4.05
• Behavioral treatment alone: -1.15
• Adjusted between-group difference: -3.0 [6]

On a 20-point scale, that is a substantial difference — and note that behavioral treatment alone reduced food noise too, just by less. The authors suggested the reduction may serve as an early indicator of treatment response, and called for longer follow-up.

Four things are worth holding onto before treating this as settled. Baseline food-noise scores were significantly different between the two groups (unadjusted means 13.1 in the medication group versus 10.7 in the behavioral-only group), which is why the headline figures are baseline-adjusted. The sample was 94% white, 93% female, with a mean age of 59 and a mean BMI of 34 — not a cross-section of people starting these drugs. The follow-up was one month, chosen because it was the first time point available. And the program that generated the data is a company that sells both the behavioral program and access to the medications [6].

The largest descriptive dataset is the INFORM survey: 550 US adults using injectable semaglutide for weight management, 86% women, mean age 53, 81% on treatment for at least four months. Median recalled FNQ score before starting semaglutide was 13 of 20 (interquartile range 10-16). After starting, the median was 6 (IQR 3-10) — a median change of -5. Agreement with individual food-noise statements dropped from 47-63% before to 15-20% after. Eighty-three percent reported treatment satisfaction, and most agreed their mental health, self-confidence, lifestyle and habits had improved [7].

The important caveat, which the INFORM authors state themselves: the survey is cross-sectional and asks people to recall a “before” state after the fact. That design cannot establish causation, and recall of a prior state is not a reliable measurement. They call for prospective longitudinal research.

Two further things to know about the source. Respondents were recruited from Numerator’s consumer opinion panel rather than from a clinical population, and several of the authors are employees of Novo Nordisk, which makes semaglutide. Neither fact makes the numbers wrong; both are reasons to weight a recalled before-and-after less heavily than a prospective measurement.

What is the mechanism?

The clearest mechanistic anchor is not a food-noise study at all — it is a 20-week randomized trial of 72 participants that measured how much people actually ate.

At week 20, the semaglutide 2.4 mg group consumed 35% fewer calories than placebo at a lunch they served themselves. They reported significantly less hunger and less prospective food consumption, significantly more fullness and satiety, better control of eating, and fewer and weaker food cravings [10]. A smaller 12-week trial of 30 participants on semaglutide 1 mg found similar results, as summarized in the food-noise review literature [2].

For tirzepatide, a six-week phase 1 randomized trial that included the Food Craving Questionnaire-State, the Food Craving Inventory, the Eating Inventory, the Power of Food Scale and an appetite visual analog scale found reductions in appetite, hunger, cravings, tendency to overeat and reactivity to environmental food cues versus placebo, as summarized in the food-noise review literature [2].

The proposed biology, as described in the food noise literature, is that these drugs act on appetite pathways in the hypothalamus and hindbrain both directly and indirectly. Food noise itself is framed within a neurocognitive model of obesity, where dysregulated appetite and food-reward processing — including altered dopaminergic signaling and heightened food cue reactivity — contribute to the frequency and persistence of food-related thoughts [2][3].

Bariatric surgery produces a similar effect, which is consistent with the mechanism: surgery increases postprandial GLP-1 secretion and is associated with fewer and less intense food cravings [2].

What are researchers arguing about?

A 2026 paper in Appetite raised a serious methodological objection, and it deserves a fair hearing because it affects how much weight to put on everything above [8].

The core points:

• Both the FNQ and RAID-FN drew heavily on media reports, anecdotes and expert opinion for item development rather than starting from structured qualitative research with people who experience food noise.
• The items are semantically very similar to existing validated tools measuring food preoccupation, food cravings and food cue reactivity.
• The FNQ correlates 0.87 with the Preoccupation with Food subscale of the Food Cravings Questionnaire-Trait. The RAID-FN correlates 0.79 with the overall FCQ-T score and 0.68 with a food-cue responsivity scale.
• An instrument designed to capture an under-researched phenomenon should ideally demonstrate utility beyond already-validated measures. The authors argue that has not yet been shown [8].

Their constructive suggestion is that the field establish a theoretical foundation inductively first, using rich qualitative data about lived experience, and that future work should test agreement between the new instruments and the older ones.

This does not mean food noise is not real. It means the measurement question is genuinely open, and consumer articles that treat the FNQ as a settled diagnostic tool are overreaching.

Is food noise the same as taste changes?

No, and a 2026 Frontiers in Nutrition review argues these get conflated far too often [9].

That paper proposes separating what people report into three levels:

• Sensory — food genuinely tastes different.
• Liking — food is less enjoyable while you are eating it.
• Wanting — food is less tempting, even when it is in front of you.

Food noise as defined by both questionnaires sits mostly in the “wanting” category, and partly in an upstream cognitive layer above it. The review’s conclusion is that current evidence does not support a consistent primary impairment of basic taste function on these drugs, and that reports are more consistent with state-dependent food revaluation and reduced reward-driven motivation [9].

What about food noise coming back?

This is one of the most searched questions on the topic, and it is the one with the least research behind it.

There is no published study measuring the return of food noise over a dose interval, after months at a stable dose, or after a dose increase. Patient communities describe all three patterns frequently. What the food-noise literature does say is that the construct sits on a continuum rather than being binary — FNQ scores are numbers on a 0-20 scale, not on/off states — and the INFORM data shows median post-treatment scores of 6, not 0 [7].

What is documented is the broader trajectory: stop the drug, and the physiology that produced the appetite change goes with it. A 2026 meta-regression found about 60% of lost weight regained within a year of stopping across six trials [see the plateau article on this site for the full picture].

If food noise returns, the useful move is to bring it to a prescriber, who can look at dose, timing, other medications and what else has changed. It is not a reason to change anything on your own.

Why does any of this matter beyond word choice?

Three reasons come up in the literature.

It names something patients could not previously name. The clinical value of the term, whatever its measurement problems, is that it gives people language for an experience that was previously described as a lack of willpower.

It may be an early marker of response. The ECO 2026 researchers suggested the one-month FNQ change could serve as an early indicator of treatment response — potentially useful given that weight itself moves slowly in the first months [6].

It changes the framing of obesity. If persistent intrusive food thoughts are a measurable feature of the condition that responds to a drug acting on appetite pathways, the “just eat less” framing becomes harder to defend. The Nutrition & Diabetes authors explicitly list the stigma associated with food noise and its potential influence on public health policy among the research priorities [2].

Where the evidence stands

Question	Status
Is there a formal definition?	Yes, two, published 2025
Are there validated measures?	Yes — FNQ (5 items) and RAID-FN (7 items)
Do GLP-1s reduce measured food noise more than behavioral treatment?	One observational cohort, one month, n=417, conference poster
How large is the effect?	-4.05 vs -1.15 on a 0-20 scale at one month
How fast does it happen?	Not measured; earliest published point is one month
Does it come back, and when?	Not studied
Is it distinct from food craving and preoccupation?	Contested
Is it a diagnosis?	No

Nothing here is medical advice. If persistent thoughts about food are affecting your life — with or without medication — that is worth raising with a healthcare provider, and particularly worth raising if you have any history of disordered eating, where a 2026 scoping review found only nine studies in total and essentially no clinical guidance.

Sources

1. Diktas HE et al. Development and validation of the Food Noise Questionnaire. Obesity 2025. https://doi.org/10.1002/oby.24216
2. Dhurandhar EJ et al. Food noise: definition, measurement, and future research directions. Nutrition & Diabetes 2025. https://www.nature.com/articles/s41387-025-00382-x
3. Hayashi D et al. What Is Food Noise? A Conceptual Model of Food Cue Reactivity. Nutrients 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10674813/
4. Development and Validation of a Brief Food Noise Questionnaire. ClinicalTrials.gov NCT06315907. https://clinicaltrials.gov/study/NCT06315907
5. Adding GLP-1 drugs to behavioral interventions silences food noise (ECO 2026 coverage, with full FNQ item list). News-Medical, May 2026. https://www.news-medical.net/news/20260513/Adding-GLP-1-drugs-to-behavioral-interventions-silences-food-noise.aspx
6. Diktas HE et al., presented at the European Congress on Obesity 2026 (as above).
7. Retrospective Assessment of Food Noise Changes After Initiation of Injectable Semaglutide: The INFORM Survey. Advances in Therapy 2026. https://pmc.ncbi.nlm.nih.gov/articles/PMC13415491/
8. Brewis A, Hayashi D et al. Food noise: conceptual, methodological, and ethical considerations. Appetite 2026. https://doi.org/10.1016/j.appet.2026.108700
9. Du Y et al. Altered eating experience during GLP-1 receptor agonist therapy: a sensory-liking-wanting framework for food preference. Frontiers in Nutrition, July 2026. https://doi.org/10.3389/fnut.2026.1870484
10. Friedrichsen M et al. The effect of semaglutide 2.4 mg once weekly on energy intake, appetite, control of eating, and gastric emptying in adults with obesity. Diabetes, Obesity and Metabolism 2021;23(3):754-762. https://doi.org/10.1111/dom.14280