In this guide
If you count calories, you've probably heard both of these sentences from people who sound equally sure: "A calorie is a calorie" and "Processed food calories are different." They can't both be fully right. On 5 October, a team publishing in Nature Metabolism added the cleanest human data yet to that argument. The answer isn't tidy, and it isn't the one either camp expected.
The researchers fed healthy adults two meals that were matched on the numbers you'd see on a nutrition label, one ultra-processed and one not, and measured what the body and brain did next. Their reaction was blunt. The first author said they were shocked that every single metabolic measure they looked at differed between the meals.
The short version: in a randomised crossover study of 57 healthy-weight adults, two ~300-calorie meals matched for macronutrients, fibre, sodium, water, energy density and glycemic index produced different responses. The ultra-processed meal triggered a larger insulin response, a more sustained rise in energy expenditure, and a weaker shift toward burning carbohydrate, alongside differences in brain-reward activity. It was a small, short-term study, so it can't say whether this causes overeating or disease. But it shows that identical label numbers don't guarantee identical biology.
The Question Every Calorie Counter Eventually Asks
Calorie counting works because of a basic principle: over time, energy in versus energy out sets the direction of your body weight. Nothing in this study breaks that principle. What it questions is a quieter assumption, that the "energy in" figure on a label tells you how a food will behave once it's inside you.
Ultra-processed foods (UPFs) already make up more than half of the calories eaten in the United States, and a similar share in the UK. Large observational studies link higher UPF intake with obesity, cardiometabolic disease and some cancers, but observational links can't prove cause. The big open question is mechanism: why would food processing matter if the calories and nutrients are the same? This study offers one possible answer.
The Study at a Glance
| Item | Details |
|---|---|
| Journal | Nature Metabolism, 5 October 2026 (Hutelin, Ahrens, Baugh and colleagues; senior author Alexandra DiFeliceantonio), with an accompanying News & Views piece on 6 October |
| Title | Metabolic and neural responses to ultraprocessed foods: a randomized, controlled, crossover study |
| Participants | 57 healthy-weight adults aged 18–45 (mean age 26, mean BMI 22.8, mostly women); their usual diets averaged about 54% of calories from UPFs |
| Subsets | 32 completed both metabolic sessions; 52 completed brain imaging (fMRI); 29 contributed to analyses linking metabolism with brain activity |
| Meals | About 300 kcal each, eaten within 10 minutes, one ultra-processed and one not, on separate days in random order |
| Matched on | Calories, macronutrients, glycemic index, energy density, available carbohydrate, fibre, sodium, water |
| Measured | Indirect calorimetry (energy burn and fuel use) and blood glucose and insulin over three hours; fMRI while viewing food pictures, plus a willingness-to-pay task |
What Was on the Plate
Classification came from the Nova system, which sorts foods by the degree of industrial processing. Group 4, ultra-processed foods, covers products made through multiple industrial steps and typically containing additives you wouldn't normally use at home. The study's UPF and non-UPF meals were carefully matched on the standard nutrition-label variables. They weren't identical in ingredients, though. The meals used different protein sources and contained different additives, and the researchers acknowledge that. That limitation comes up again below.
Participants fasted overnight, ate one meal over about ten minutes, then rested while researchers tracked their metabolism and drew blood for three hours. A few days later they repeated the process with the other meal. Because each person ate both meals, the comparison cancels out differences between individuals, which is the real strength of a crossover design.
Result One: Insulin and Blood Sugar
Total blood-glucose exposure was similar after the two meals, but the shape of the curve differed. Glucose rose faster after the non-UPF meal and was significantly higher at 20 minutes. Both meals peaked around 40 minutes, but after the UPF meal glucose tended to stay elevated for longer. By three hours, levels after the UPF meal were still above baseline, while levels after the non-UPF meal were closer to it. The differences at 120 and 180 minutes weren't statistically significant, so this is a trend, not a firm result.
Insulin was the clearer signal. Even with similar overall glucose, the UPF meal produced significantly more insulin at 40, 60, 90 and 120 minutes, returning to baseline by three hours. A separate randomised crossover analysis published in August 2026 also found higher post-meal insulin after ultra-processed meals regardless of their energy density, which adds weight to the pattern.
Why might that happen? One idea is that processing can change how quickly nutrients are released. For example, grinding legumes or breaking down their cell walls can raise glucose and insulin responses compared with eating them intact, and extruded starches can be easier to digest. This study didn't pin down which feature of processing was responsible, so that explanation is a hypothesis.
Result Two: Energy Burn and Fuel Choice
After eating, your metabolic rate rises as your body digests and stores food. Both meals pushed it up to a peak at about 40 minutes, but after the UPF meal the rise was more sustained, so overall post-meal energy expenditure was higher.
Fuel choice shifted too. Normally, after a meal your body increases how much carbohydrate it burns. That increase was stronger after the non-UPF meal. After the UPF meal, carbohydrate oxidation rose less and fat oxidation was relatively greater, which also showed up in the respiratory exchange ratio (RER), a measure of how much carbohydrate versus fat you're burning. The authors noted that this blunted carbohydrate use resembles features seen in insulin resistance and type 2 diabetes. They were careful to add that they didn't measure insulin sensitivity directly, and that nothing here shows UPFs cause either condition.
Result Three: The Brain Scans
On a separate day, participants lay in an MRI scanner and looked at 28 pictures of familiar foods, 14 ultra-processed and 14 not, then bid up to five dollars for each in a standard auction task. Self-rated liking, familiarity and expected fullness were broadly similar across the two categories, and so were the bids. People didn't say they valued one type more.
The scans told a subtler story. Participants whose carbohydrate burning differed more between the two meals also showed different brain responses to the two groups of food pictures, in regions linked to sensory processing, learning and reward: the left superior temporal gyrus, right caudate and left ventral striatum. Other metabolic measures, such as peak metabolic rate, glucose and insulin, weren't significantly linked to the brain data.
This is correlation, not cause, and the sample for this analysis was just 29 people. But it fits a theory researchers have been testing in animals: the brain learns the value of foods partly from the metabolic signals that follow eating, not only from taste. If processing changes those signals, it could change how the brain tags the food.
The Counterintuitive Part: More Calories Burned
Here's something that surprises calorie counters. The ultra-processed meal led to higher total post-meal energy expenditure. If you only read that line, it sounds like a point in UPFs' favour. It isn't, for three reasons.
- It's one meal. A modest, short-lived difference in energy burn over three hours says little about weight over months.
- Burning more isn't the same as being better off. The extra energy cost came alongside higher insulin and a weaker carbohydrate response, features researchers watch for as possible early warning signs, though this study can't say whether they matter long term.
- It adds to, rather than replaces, the thermic effect we already know. As we explain in our piece on negative-calorie foods, digestion always has a cost, and it varies with what you eat. Here the point is that processing may nudge that cost too.
Why Calorie Labels Can't See Food Structure
Calorie labels rely on a system developed in the late 1800s that assigns average energy values to protein, carbohydrate and fat: roughly 4, 4 and 9 kcal per gram. It's an excellent approximation, and it's why calorie counting works at all. But it assumes your body extracts the same energy from a gram of fat or starch regardless of the food it came in.
Real foods bend that rule. Research on whole almonds, for instance, has shown that people absorb noticeably fewer calories than the standard factors predict, because intact cell walls trap some of the fat. Cooking and grinding do the opposite and release more. Food structure, in other words, is a hidden variable that labels can't see. Ultra-processing usually breaks structure down, which could plausibly change how quickly and completely nutrients arrive. That's the idea this study starts to test, and the same logic helps explain why published label values are only accurate within a margin even before you account for portion errors.
How This Fits With Earlier Research
This isn't the first signal. In 2019, Kevin Hall and colleagues at the US National Institutes of Health ran a tightly controlled inpatient trial in which adults lived in a clinical centre and were offered either an ultra-processed or an unprocessed diet matched for presented calories, sugar, fat, fibre and macronutrients. On the ultra-processed diet, participants ate about 500 more calories per day on their own and gained roughly 0.9 kg in two weeks, while on the unprocessed diet they lost about the same amount.
That trial pointed to overeating but couldn't explain why. The 2026 Nature Metabolism study looks at what might sit upstream of the extra eating: different metabolic and brain responses to a single matched meal. Put together, they form a plausible but still unproven chain: processing changes how a food behaves, behaviour changes signalling, signalling shapes how much you eat. The chain has missing links, which is why researchers are careful with their wording.
The Limits You Should Know About
- Small and specific. Fifty-seven people were enrolled, only 32 finished both metabolic sessions, and just 29 contributed to the brain-metabolism analysis. Participants were young, healthy-weight and mostly women.
- Single, small meals. About 300 calories, one time. Responses to larger meals or to months of eating may differ.
- Processing wasn't isolated. The meals differed in protein sources, ingredients and additives, and the researchers didn't measure food structure beyond fibre. We can't say whether processing itself, a specific ingredient or an additive drove the results.
- No weight or overeating outcomes. The study didn't measure how much people ate afterwards or whether anyone gained weight.
- Short-term biology isn't disease. Insulin sensitivity wasn't directly measured, so the study can't show UPFs cause insulin resistance or diabetes.
- Peer review is new. An earlier version circulated as a preprint in April 2026; the final peer-reviewed paper appeared on 5 October, so independent replication is still to come.
What It Means for Counting Calories
It doesn't mean counting calories is pointless. It means calories are a very good first-order tool with some noise attached, and that noise may be larger for heavily processed foods. Three practical shifts follow.
- Keep tracking, but lean on food quality to make it easier. Meals built around minimally processed foods are generally more filling for the same calories, so staying in a deficit takes less willpower. Use the calorie calculator to set your target, then spend your calories on foods that keep you satisfied.
- Count home-cooked meals properly. When you cook from scratch you control the ingredients. Paste a recipe link into our recipe nutrition calculator to see calories, macros, fibre and sodium per serving, then compare it with the packaged version you'd otherwise buy.
- Anchor your meals in protein and fibre. Both are linked to fullness, and they're easier to hit with whole foods. Our macro calculator and the macro calculator guide show how to set targets.
If you track and still can't explain your results, remember that reporting error is bigger than most of us think. We explain why in why most people are eating the wrong number of calories, and processing may be one more hidden variable layered on top of it.
A Realistic "Upgrade Ladder" (No Guilt Required)
Ultra-processed foods are cheap, convenient and often tasty, and many people rely on them for good reasons: time, budget, access, disability. You don't need to purge your kitchen. A better approach is to upgrade a few meals at a time, starting wherever the swap is easiest.
| Nova group | What it means | Everyday examples |
|---|---|---|
| 1 — Unprocessed or minimally processed | Whole foods, perhaps dried, frozen or ground without additives | Fruit, vegetables, eggs, plain meat and fish, plain oats, dried beans, plain yoghurt |
| 2 — Processed culinary ingredients | Substances extracted from group 1 for cooking | Olive oil, butter, sugar, salt, flour |
| 3 — Processed foods | Group 1 foods plus group 2 ingredients, made simply | Canned beans, simple bread, cheese, tinned fish |
| 4 — Ultra-processed | Industrial formulations with additives and refined ingredients | Many breakfast cereals, packaged snacks, soft drinks, instant noodles, reconstituted meats |
Pick one meal a day and move it one rung up. Swap a sweetened breakfast cereal for plain oats with fruit and nuts. Replace a packaged sandwich with one you assemble from bread, cheese and salad. Keep a couple of convenient favourites, since the goal isn't purity, it's shifting the average. It's also fine to keep UPFs that help you hit protein or fibre targets, such as certain protein yoghurts or wholegrain breads, while you work on the rest.
The Bottom Line
"A calorie is a calorie" is a good rule of thumb for balancing energy. This study suggests it's not the whole story of what a meal does to you. Two meals that look identical on a label produced different insulin curves, different fuel use and different brain-activity patterns in healthy young adults. We don't yet know whether those short-term differences translate into overeating or disease, and the study is too small and too short to tell.
For now, the sensible takeaway is simple: keep counting calories, build most of your plate from minimally processed foods, and treat processing as one more reason, not the only one, to cook more of your own meals when you can.
Frequently Asked Questions
Are calories from ultra-processed food different from other calories?
Chemically a calorie is a calorie, but the body's response to a meal depends on more than its calorie count. In this 2026 study, two meals matched for calories and macronutrients produced different insulin responses, energy expenditure and fuel use. Whether that changes weight over time still needs longer studies.
Did the ultra-processed meal cause more insulin release?
Yes, in this study. Insulin was significantly higher after the UPF meal at 40, 60, 90 and 120 minutes, even though overall glucose exposure was similar. A separate August 2026 crossover analysis also found higher post-meal insulin after ultra-processed meals.
Does this prove ultra-processed food causes insulin resistance or diabetes?
No. The researchers did not measure insulin sensitivity and the study looked at single meals. The pattern resembles aspects of insulin-resistant metabolism, but the study can't show UPFs cause either condition.
Why did the ultra-processed meal burn more energy?
Post-meal energy expenditure stayed elevated longer after the UPF meal, producing a higher overall total. The study doesn't explain the mechanism, and a small difference over three hours doesn't mean UPFs help with weight loss. It shows that digestion and metabolism differ with food form.
What is the Nova classification?
Nova groups foods by degree of industrial processing: unprocessed or minimally processed, processed culinary ingredients, processed foods and ultra-processed foods. Group 4 includes industrial formulations usually containing additives not found in home cooking.
Should I stop counting calories?
No. Calorie counting still works well for managing weight. This research suggests food quality and structure add extra variation on top of the numbers, which is a reason to favour minimally processed foods, not to abandon tracking.
Do I have to cut out all ultra-processed foods?
The evidence doesn't require perfection. Many people rely on UPFs for cost, time or access. Upgrading a few meals at a time, and keeping UPFs that help meet protein or fibre goals, is a practical middle path.
Was the study big enough to trust?
It's a well-designed randomised crossover trial, but small: 57 enrolled, 32 completed both metabolic sessions, and 29 contributed to brain-metabolism links. Participants were young, healthy-weight and mostly women, so the results need confirming in larger and more diverse groups.
How can I check the nutrition in my own cooking?
Weigh ingredients and use a recipe calculator. Our free recipe nutrition calculator accepts recipe links or manual entries and shows calories, macros, fibre and sodium per serving.
References & Sources
- Hutelin, Z., Ahrens, M., Baugh, M. E. et al. (2026). Metabolic and neural responses to ultraprocessed foods: a randomized, controlled, crossover study. Nature Metabolism. doi:10.1038/s42255-026-01619-4.
- Nature Metabolism (2026). Research Briefing: Ultraprocessed foods evoke distinct metabolic and neural responses (doi:10.1038/s42255-026-01618-5); News & Views by J. Penney (6 October 2026).
- Hutelin, Z. et al. (2026). Preprint version, medRxiv, posted 11 April 2026 (doi:10.64898/2026.04.10.26350599).
- Hall, K. D. et al. (2019). Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metabolism, 30(1), 67–77.
- Monteiro, C. A. et al. (2018). The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing. Public Health Nutrition, 21(1), 5–17.
- Novotny, J. A. et al. (2012). Discrepancy between the Atwater factor predicted and empirically measured energy values of almonds in human diets. American Journal of Clinical Nutrition, 96(2), 296–301.
- Randomized crossover trial of ultra-processed meals with differing energy density on postprandial glyco-insulinemic and appetite-related hormonal responses (published 4 August 2026).
- News-Medical (8 October 2026). Nutritionally similar meals might not affect your body the same way.
- MacrosMeasure (2026). Free Recipe Nutrition Calculator.
- MacrosMeasure (2026). Free Calorie Calculator.
Calculate your own numbers — free.
18 science-backed calculators covering everything in this article and more.