Matched within one per cent for calories, carbohydrate, fat, protein, water and salt, a highly processed meal produced a higher insulin response and less carbohydrate burned than a minimally processed one.
Ultraprocessed foods make up more than half of the average American’s daily calories, and the epidemiological case against them is strong: people who eat more of them have worse health outcomes across a long list of measures. The persistent difficulty is that ultraprocessed foods also tend to be high in fat, high in sugar and low in fibre, so it has never been clear whether the harm comes from processing itself or from the nutritional profile that usually accompanies it.
A randomised crossover trial from the Fralin Biomedical Research Institute at Virginia Tech, published in Nature Metabolism, was designed to separate the two.
Holding nutrition constant
The method was to build two 300-calorie meals matched within one per cent for carbohydrates, fat, protein, calories, water and salt, differing only in how processed their components were.
“We know that the majority of ultraprocessed foods are high in fat, they’re high in sugar, they’re low in fiber,” said Alex DiFeliceantonio, an associate professor at the institute, whose lab led the study. “But if we artificially hold all of those things constant, is there something about the processing that leads to a different outcome?”
The highly processed meal consisted of a bite of peanut butter and jelly sandwich, deli turkey, veggie chips, a cookie, cereal, instant mashed potatoes and a little water. The minimally processed meal was sliced banana, dried cranberries, cheese and egg.
“The two meals are matched within 1 percent of carbs, fat, proteins, calories, water, and salt,” said Zach Hutelin, the study’s first author, who did the work as part of his doctoral research. “If these meals had a nutritional label, they would be practically identical.”
Fifty-seven healthy adults aged 18 to 45 enrolled; 32 completed both the brain imaging and metabolic sessions. Each participant ate both meal types in separate sessions at least three days apart, with ten minutes to eat, and gave blood immediately afterwards and six more times over the following three hours. A crossover design of this kind is a genuine strength: each person serves as their own control, which removes between-person variation from the comparison.
Every measure moved
“We were shocked when every single metabolic metric differed,” Hutelin said — “insulin response was much higher” after the ultraprocessed meal.
The pattern was consistent. The ultraprocessed meal produced a higher insulin response, greater energy expenditure, and less carbohydrate burned for fuel, with blood sugar staying somewhat higher for longer.
Those results fit together in a way that makes mechanistic sense. More insulin released alongside blood sugar that nonetheless stays elevated longer is the signature of reduced insulin efficiency — more hormone achieving less. The shift away from burning carbohydrate for fuel is the metabolic consequence of that. Over years of daily meals, a pattern of consistently larger insulin responses is the kind of thing that plausibly matters for metabolic health, though a single-meal study cannot show that it does.
The likely mechanism is physical rather than chemical. Processing breaks down food structure — the intact cell walls of a banana, the matrix of cheese — and a food whose structure has already been disassembled is digested faster, delivering its sugars to the bloodstream more quickly even when the quantity is identical. Rate of delivery, not amount, is what the insulin response tracks.
Into the brain
The imaging arm produced a finding the researchers describe carefully. In the fMRI sessions, differences in carbohydrate burning after each meal were linked to differences in brain responses to food pictures in the ventral striatum and nucleus accumbens — regions involved in learning, motivation and reward.
The structure of that claim is worth noting: it is not that the ultraprocessed meal lit up reward regions. It is that how much a given participant’s carbohydrate burning shifted predicted how much their reward-region response to food images shifted. The link is between the metabolic change and the neural change, within individuals.
That is a more interesting result than a simple difference in brain activation, because it suggests a route by which metabolism could feed back into food motivation — if what happens metabolically after a meal shapes how rewarding food images subsequently look, there is a mechanism by which processed food could influence later eating. It is also a correlation across 32 people, and the release provides no effect sizes or statistics.
One result points the other way, and the researchers report it: participants did not say they would pay more for either type of food. Whatever the brain-response differences amount to, they did not translate into a stated preference.
What the authors say is unresolved
The limitations the release lists are substantive, and the team raises them rather than leaving them to be found.
The study covered one set of meals in young, healthy adults. The protein sources differed between the two meals — deli turkey against cheese and egg — which is a real confound, since protein source affects digestion rate independently of processing. The ultraprocessed meal contained more additives, so additives and processing cannot be separated here. And food structure, the most likely mechanism, was not directly measured; total fibre was.
Thirty-two completers from 57 enrolled is also worth noting. Nearly half the sample did not finish both arms, and the release does not say why or whether completers differed from the rest.
DiFeliceantonio frames where this leads. “If you look at that population-level data, people who consume large amounts of ultraprocessed foods have higher rates of poor health outcomes,” she said. “We want to examine specific factors, such as commercial additives or specific processing steps.”
That is the right next move. This study establishes that processing has metabolic effects independent of nutritional composition — which is a meaningful finding, because it was genuinely in doubt. It does not establish which aspect of processing is responsible.
The wider reaction
Carlos A. Monteiro of the University of São Paulo, who was not an author and who originated the NOVA classification system that defines ultraprocessed food, was quoted by Nature Metabolism calling it “a novel and important contribution, as the brain ultimately regulates eating behavior.”
Readers should know that Monteiro is the leading academic proponent of the ultraprocessed-food concept, which is itself contested — critics argue the category is too heterogeneous to be a useful target for nutritional policy. His endorsement is informed and also interested, which does not diminish it but is worth knowing.
Funding came from a National Institutes of Health research grant via the National Institute of Diabetes and Digestive and Kidney Diseases, and a National Science Foundation graduate research fellowship. No conflict-of-interest statement is included in the release.
The practical reading is narrow and real: two meals with identical labels are not nutritionally equivalent, and the difference shows up within three hours. What to do with that beyond preferring whole ingredients where convenient is not something one trial in 32 young adults can say.
Notes
- Hutelin, Z., et al. “Metabolic and neural responses to ultraprocessed foods: a randomized, controlled, crossover study.” Nature Metabolism, 5 October 2026. DOI: 10.1038/s42255-026-01619-4.
- Virginia Tech. “Food processing influences metabolism and brain activity.” EurekAlert!, 5 October 2026. https://www.eurekalert.org/news-releases/1146228.