Do ultraprocessed meals change post-ingestive metabolic and brain responses beyond calories?

A meal comparison built to isolate processing

The question is whether the degree of food processing changes what happens after eating, independent of what the meal contains. To test that, the study used a randomized, controlled, crossover design: each participant ate both a UPF meal and a non-UPF meal, so comparisons were made within the same person rather than between different groups. The meals were matched on energy, macronutrients and glycaemic index, and each was about 300 kcal and fully consumed.

That matching is the study's central design choice. Earlier work comparing meals that differed in processing also differed in total energy and macronutrient composition, which made processing hard to separate from nutrient content. Here, the authors assembled each meal from a diverse set of foods so that the measured responses would not hinge on a single item. Processing level was classified by Nova, the system that sorts foods by the extent and purpose of industrial processing.

The participants were 57 healthy-weight adults with a mean BMI of 22.75 kg/m2 and a mean age of 26.21 years. They were not a clinical population, and the meal was a single fixed load rather than a habitual diet. Those two facts bound what the results can say: they describe acute post-ingestive physiology in lean adults after one standardized meal.

Insulin and energy responses rose without a glucose difference

The most decision-relevant result is a split between two measures readers often treat as interchangeable. Blood glucose area under the curve was comparable between the UPF and non-UPF conditions, yet the UPF meal elicited a larger insulinaemic response. In other words, similar circulating glucose was accompanied by more insulin release after the processed meal.

The UPF meal also produced a greater energetic response, and carbohydrate oxidation was attenuated relative to the non-UPF meal. Carbohydrate oxidation is the rate at which the body burns carbohydrate for energy, and the paper reports it in grams per minute. A lower rate after the UPF meal means the body shifted away from burning carbohydrate at that point in the post-meal window.

The authors read this pattern as consistent with a relative reduction in insulin sensitivity, and they note that a delayed shift in substrate oxidation is also seen in insulin resistance and type 2 diabetes. That is the authors' interpretation of an acute physiological pattern, not a demonstration that a single meal causes insulin resistance. The study was not designed to assess chronic outcomes.

Carbohydrate oxidation differences tracked striatal food-cue responses

The study then asked whether these metabolic differences related to how the brain responded to food cues. Between-condition differences in peak carbohydrate oxidation were regressed on participant-level brain contrast estimates, meaning the analysis tested whether people with a larger metabolic gap between meals also showed a different neural gap.

A whole-brain analysis found an association in the left superior temporal gyrus. Within an a priori striatal mask, small volume correction identified associations in the right caudate and the left ventral striatum. The direction matters: a larger increase in peak carbohydrate oxidation for non-UPF relative to UPF was negatively associated with the neural response to non-UPF versus UPF cues. Peak respiratory exchange ratio, a closely related measure, showed a similar relationship in the left superior temporal gyrus and right caudate.

Between-condition changes in peak metabolic rate, fat oxidation, blood glucose and blood insulin were not significantly associated with neural responses. So the link the paper reports is specific to the carbohydrate-oxidation and respiratory-exchange measures, not a general coupling of every metabolic change to brain activity. The authors describe this as consistent with the possibility that a processing-related shift in carbohydrate oxidation may be associated with food-cue reactivity learned through exposure, and they connect it to prior findings that carbohydrate metabolism matters for neural response.

Bids were equal but value encoding reversed

Subjective food value was measured with a Becker-DeGroot-Marschak auction during fMRI: participants viewed 28 food images and bid their willingness to pay from $0 to $5 for each item. Willingness to pay did not differ between non-UPF and UPF foods, even after adjusting for perceived healthiness and frequency of consumption. This was contrary to the authors' hypothesis.

The neural representation of value told a different story. When willingness to pay was entered as a parametric modulator, the expected value-related regions appeared overall, including the left orbitofrontal cortex and left pregenual anterior cingulate cortex, plus higher-order visual regions such as the right fusiform gyrus and right lateral occipital cortex. That served as a check that the task was engaging known valuation circuitry.

Comparing conditions, the left fusiform gyrus and two peaks in the left lingual gyrus showed opposing associations: positive with value in the non-UPF condition and negative in the UPF condition. Within the striatal mask, the right putamen and right caudate showed the same reversal. Notably, the direct UPF versus non-UPF contrast produced no significant voxels at either the whole-brain or small-volume level. The authors suggest this could reflect individual-level differences in how subjective value is encoded, despite no overall group difference in bidding.

The food sets were matched on most rated attributes

Because valuation analyses can be confounded by how appealing or familiar the foods seem, the study checked the image set directly. Participants rated estimated calories, estimated energy density, estimated price, liking, expected satiety and familiarity similarly for non-UPF and UPF foods. Two attributes did differ: participants rated non-UPF foods as healthier and reported consuming them more often.

Those two differences were handled by adding healthiness and frequency of consumption as trial-by-trial parametric modulators or covariates in the models comparing UPF with non-UPF. The authors also report that the pattern of fMRI results was unchanged when hunger, fullness and thirst were entered as covariates, and remained similar after adjustment for BMI. Age and sex were included as covariates in all models.

One behavioural difference is worth keeping separate from the metabolic findings: the UPF meal was eaten more slowly than the non-UPF meal, by a modest margin. The authors argue that a roughly 1.5-minute difference in eating time is unlikely to produce a physiologically meaningful effect. This runs opposite to the faster ad libitum UPF eating reported elsewhere, so it should not be read as evidence about real-world eating speed.

A single 300 kcal meal in lean adults

Several limits sit close to the results they qualify. Total protein was matched, but the protein sources differed between meals and could have influenced the insulin response. The UPF meal contained more additives, and the study did not explicitly measure the physical and chemical structures of the food matrix beyond total fibre, even though additives and matrix alteration are inherent to the Nova definition.

The meal load was also fixed at about 300 kcal, and the authors note that responses may differ with larger caloric loads. The sample was healthy-weight adults, so the acute pattern here does not establish what happens in people with obesity, insulin resistance or type 2 diabetes, nor what repeated exposure would do over time. Willingness-to-pay results should also be read with the caveat that local grocery prices changed rapidly during data collection.

The insulin-glucose split is the finding to carry forward

The paper's distinctive contribution is not that processed food is unhealthy in the abstract. It is that two meals matched on energy, macronutrients and glycaemic index still diverged on insulin, energy expenditure and carbohydrate oxidation, and that the size of the carbohydrate-oxidation gap tracked striatal food-cue responses. That makes post-ingestive nutrient availability, not calories or macronutrient labels alone, the candidate mechanism the authors put forward.

The practical priority for readers is therefore to keep the measures apart. Comparable blood glucose after the two meals does not mean comparable post-ingestive physiology: the insulin and carbohydrate-oxidation results moved in a different direction from glucose. Likewise, equal willingness to pay does not mean equal neural valuation, since the fusiform, lingual and striatal value signals reversed sign between conditions.

What remains open is whether these acute differences persist with larger meals, in metabolically impaired populations, or under repeated exposure. The authors frame the data as a plausible physiological route by which repeated exposures could contribute to the metabolic dysfunction reported in epidemiological studies, while stating that the study was not designed to test chronic outcomes. The finding to carry forward is the dissociation itself: processing level changed post-ingestive and neural responses that calorie and macronutrient matching did not equalize.

Sources used in this answer