The ultra-processed-food debate is often framed as a fight between two slogans. One side says processing itself is the problem. The other says it is all just calories and personal choice.
Controlled feeding studies make both slogans too simple. Calories remain central to short-term body-weight change, yet the structure of a diet can change how many calories people spontaneously consume. The unresolved part is which properties — energy density, eating rate, texture, palatability, food form, processing or combinations of them — explain the effect in each setting.
That distinction matters because “ultra-processed” is a broad classification category, not one biological mechanism shared equally by every product.
Quick answer
Is ultra-processed food only a problem because people simply choose to eat more calories?
No — that wording is too simple. In randomized feeding studies, diet conditions differing in processing level changed spontaneous energy intake and weight even when participants were not instructed to restrict calories. But calories still mediated the short-term weight change, and the trials do not prove that processing itself is one isolated mechanism or that every ultra-processed food has the same effect.
Key takeaways
- In the 2019 NIH inpatient crossover trial, 20 adults consumed about 508 kcal/day more on the ultra-processed diet and gained about 0.9 kg, while losing about 0.9 kg during the unprocessed phase.
- That trial supports a causal difference between the tested diet conditions, but it was designed to compare diets — not to identify one universal mechanism behind the intake difference.
- In the 2025 UPDATE trial, both healthy-guideline diets produced weight loss, but weight loss was greater on the minimally processed diet: −2.06% versus −1.05%.
- The 2025 trial is important but not mechanistically final: Nature Medicine published methodological critiques in 2026, including concern that the diets differed in relevant food properties beyond processing classification.
- A 2025 randomized meal experiment shows that energy density and eating rate can substantially change ad-libitum intake, supporting plausible mechanisms without proving they explain every UPF effect.
Calories matter — but that is not the whole question
Short-term body-weight change depends strongly on energy balance. In the NIH trial, participants ate more energy during the ultra-processed condition, and body-weight change closely tracked the difference in energy intake.
So the controlled evidence does not support the slogan “calories do not matter”. The more interesting scientific question is why people consumed different amounts when food environments and food properties changed.
The 2019 NIH metabolic-ward trial changed the debate
Hall and colleagues admitted 20 weight-stable adults to the NIH Clinical Center and randomized the order of two 14-day diet periods. Participants could eat as much or as little as they wanted.
Energy intake was 508 ± 106 kcal per day higher during the ultra-processed phase. Participants gained 0.9 ± 0.3 kg during that phase and lost 0.9 ± 0.3 kg during the unprocessed phase. Because the feeding conditions were randomized within the same participants, the study provides strong evidence that the tested diet environments caused different intake and short-term weight outcomes.
What that trial did not prove
A randomized diet comparison can show that two diet conditions produce different outcomes without identifying the one feature responsible. The NIH diets differed in many real-world food properties even though they were designed to match several presented nutrients and calories.
The authors themselves discussed possible contributors such as eating rate, food texture and non-beverage energy density. That means the result should not be translated into “NOVA category 4 is a single biological toxin mechanism”.
The 2025 UPDATE trial made the story more realistic — and more complicated
The UPDATE trial tested two eight-week ad-libitum diets in a free-living crossover design. Both followed the UK Eatwell Guide, but one emphasized minimally processed food and the other ultra-processed food. Fifty participants contributed primary-outcome data.
Both diet periods produced weight loss. Estimated weight change was −2.06% on the minimally processed diet and −1.05% on the UPF diet, a within-person difference of about one percentage point. Self-reported energy intake was also 327.3 kcal per day lower during the minimally processed phase.
Why the exact UPDATE result should not be generalized to everyone
The participants had overweight or obesity and high habitual UPF intake at eligibility, and 90.9% of those randomized were female. The trial therefore does not provide a universal one-month or eight-week weight-loss forecast for every population.
It was also a free-living study. Energy intake was self-reported rather than measured with the same precision as an inpatient metabolic ward. Nature Medicine later corrected supplementary energy-density values, which reinforces the need to use the current record rather than older summaries.
The 2026 methodological debate matters
In January 2026, Nature Medicine published three Matters Arising commentaries about the UPDATE trial and a formal reply from the trial authors. One major criticism was that the diet conditions differed in potentially important nutritional and physical food properties as well as processing classification.
That criticism does not erase the observed difference between the diets. It changes what can be claimed about mechanism. TwinPare should report the trial result while remaining cautious about saying that “processing itself” was isolated as the sole causal factor.
Energy density and eating rate can change how much people eat
A separate 2025 randomized crossover meal study manipulated energy density and eating rate in 69 healthy adults. The slow-eating, low-energy-density meal averaged 570 kcal, while the fast-eating, high-energy-density condition averaged 1,143 kcal — a 573 kcal difference in that experimental comparison.
This is strong evidence that food properties can alter spontaneous energy intake. It does not prove that energy density and eating rate completely explain every UPF result, but it shows why “people just decide to eat more” is an incomplete model.
NOVA is a classification, not one mechanism
The NOVA system groups foods by the nature and purpose of processing. That can be useful for population research and food-environment questions, but the category contains products with different nutrient profiles, textures, energy densities and eating rates.
A classification can identify a pattern without explaining every biological step. Research still has to test which food properties, formulations and contexts actually drive intake or health outcomes.
Not every ultra-processed food should be treated as identical
The evidence does not justify calling every UPF “toxic” or assuming that two foods with the same NOVA label produce identical metabolic effects. It also does not justify dismissing the category simply because calories mediate weight change.
A more useful interpretation is that processing level can act as a marker for bundles of food properties that sometimes make rapid, high-energy intake easier. Which properties matter most remains an active research question.
TwinPare boundary
TwinPare Research can explain controlled diet evidence, uncertainty and methodological debate without turning a food classification into a diagnosis or moral judgement.
TwinPare Health & Fitness must not diagnose obesity, diabetes, food addiction or gastrointestinal disease from UPF intake, and this article should not become an individualized elimination diet or treatment plan. People with eating disorders, medically prescribed diets or metabolic disease may need professional dietary guidance.
Source notes
The sources have been verified and editorially reviewed for this article. The limitations below show which level of conclusion the sources support.
- [hall-upf-metabolic-ward-2019] Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake Kevin D Hall et al.. Cell Metabolism, 2019. Evidence type: Inpatient randomized crossover feeding trial in 20 weight-stable adults, with two weeks per diet and ad-libitum intake Limitation: Energy intake was 508 ± 106 kcal/day higher during the ultra-processed diet; participants gained 0.9 ± 0.3 kg during the UPF phase and lost 0.9 ± 0.3 kg during the unprocessed phase. The trial establishes a causal difference between the tested diet conditions, not one universal mechanism for every NOVA-classified food. Published errata exist; numerical claims here use the current corrected record. PubMed PMC DOI
- [dicken-update-upf-mpf-2025] Ultraprocessed or minimally processed diets following healthy dietary guidelines on weight and cardiometabolic health: a randomized, crossover trial Samuel J Dicken et al.. Nature Medicine, 2025. Evidence type: Eight-week free-living randomized crossover feeding trial in adults with overweight or obesity and high habitual UPF intake Limitation: Both Eatwell-aligned diets produced weight loss, but loss was greater during the minimally processed phase (−2.06% versus −1.05%; within-person difference −1.01 percentage point). Self-reported energy intake was 327.3 kcal/day lower on MPF. The cohort was 90.9% female, energy-density values in supplementary material were corrected in December 2025, and later methodological debate cautions against treating processing level as the only between-diet difference. PubMed PMC DOI
- [lasschuijt-energy-density-eating-rate-2025] The Effect of Energy Density and Eating Rate on Ad Libitum Energy Intake in Healthy Adults—A Randomized Controlled Study Marlou P Lasschuijt et al.. The Journal of Nutrition, 2025. Evidence type: Randomized crossover meal experiment in 69 healthy adults manipulating meal energy density and eating rate Limitation: The slow-eating/low-energy-density condition averaged 570 kcal versus 1,143 kcal in the fast-eating/high-energy-density condition, a 573 kcal difference in that experimental comparison. This supports plausible intake mechanisms but does not prove energy density and eating rate fully explain all UPF effects. PubMed DOI
- [robinson-forde-update-critique-2026] Concerns around evidence that food processing should be included in dietary guidance Eric Robinson; Ciarán G Forde. Nature Medicine, 2026. Evidence type: Matters Arising methodological critique of the 2025 UPDATE trial Limitation: The authors argue that the trial diets differed in nutritional and food-property characteristics as well as processing level, complicating attribution of the weight difference to processing itself. This is a methodological critique, not a refutation of the observed trial outcome. PubMed DOI
- [dicken-update-reply-2026] Reply to Ludwig, D. S. et al.; Robinson, E. & Forde, C. G.; Wang, Z. & Peng, C. Samuel J Dicken; Adrian Brown; Rachel L Batterham. Nature Medicine, 2026. Evidence type: Author reply to three Matters Arising critiques of the 2025 UPDATE trial Limitation: The reply documents the authors’ response to methodological criticisms. TwinPare should present the existence of the debate rather than treating either the critique or the reply as a final mechanistic verdict. PubMed DOI
Editorial source review
This section shows how the article's key factual claims are linked to the source.
Phrasings that require caution
- Do not say all ultra-processed food is toxic.
- Do not say calories do not matter for body-weight change.
- Do not claim processing alone causes obesity regardless of energy intake.
- Do not treat NOVA as a biological mechanism.
- Do not claim one additive is the cause of the observed trial effects.
- Do not generalize the exact UPDATE weight effects to all populations.
- Do not hide the 2026 methodological debate around the UPDATE trial.
- Do not equate minimally processed with nutritionally optimal in every product or dietary context.
- Do not turn population evidence into individualized treatment for obesity, diabetes, gastrointestinal disease or eating disorders.
| ID | Claim | Source support | Caution |
|---|