Why Ultra-Processed Food Doesn't Fill You Up

In the spring of 2018, Kevin Hall sealed 20 adults inside the metabolic ward of the NIH Clinical Center in Bethesda, Maryland, and offered them three meals a day for four weeks. Half the time the meals were ultra-processed — packaged cereals, sandwich meats, canned ravioli, the kind of food that fills American supermarket aisles. Half the time the meals were prepared from minimally processed ingredients: grilled chicken, fresh vegetables, whole grains, plain yogurt. The two diets had been carefully matched for total calories presented, protein, fat, carbohydrate, sugar, fibre, and sodium. Participants were told to eat as much or as little as they wanted.

The result, published in Cell Metabolism the following year, was one of the more unexpected findings in modern nutrition. On the ultra-processed arm, participants spontaneously consumed an average of 508 more calories per day than on the unprocessed arm. They gained weight. On the unprocessed arm, they lost it. The participants reported similar levels of meal pleasantness on both diets. They weren't choosing ultra-processed food because they preferred it. They were eating more of it without noticing.

The Hall study was the first randomised controlled trial to isolate the effect of ultra-processing itself, holding nutrient composition constant. It demonstrated something that observational research had long suggested but couldn't prove: ultra-processed food appears to bypass the body's satiety signalling in a way that produces overconsumption almost regardless of intent.

What "ultra-processed" actually means

The terminology comes from the NOVA classification system, developed by Carlos Monteiro and his group at the University of São Paulo in the early 2010s. NOVA divides foods into four categories: unprocessed or minimally processed (group 1), processed culinary ingredients like oils and salt (group 2), processed foods such as canned beans and cheese (group 3), and ultra-processed foods (group 4).

The ultra-processed category is defined less by any single ingredient than by industrial manufacturing processes and the presence of substances rarely used in home cooking — high-fructose corn syrup, hydrogenated oils, protein isolates, emulsifiers, artificial flavours, colour stabilisers. Soft drinks, packaged snacks, breakfast cereals, instant noodles, ready meals, reconstituted meat products, mass-produced baked goods. The hallmark is not processing per se but industrial reformulation that breaks foods down to their components and rebuilds them with additives that wouldn't appear in a kitchen.

By NOVA estimates, ultra-processed foods now account for roughly 58% of total caloric intake in the United States, 50% in the United Kingdom, and rising shares almost everywhere these foods have been introduced. Their share of the food supply is the public-health context in which the Hall trial landed.

The mechanisms that explain the calorie gap

Hall's study showed that ultra-processed food drives overconsumption. It did not, on its own, fully explain why. The subsequent years of research have pointed to several converging mechanisms.

Eating rate. One of the most consistent observations across feeding studies is that ultra-processed foods are consumed faster. They are typically softer, less fibrous, easier to swallow. In Hall's trial, participants on the ultra-processed arm ate at a meaningfully higher rate — roughly 17 calories per minute compared with 7 on the unprocessed arm. The satiety hormones — CCK, PYY, GLP-1 — need 15 to 20 minutes to rise to levels that signal fullness. When more food enters in the same window, meals end at higher caloric loads before the brain has the information to stop.

Hyperpalatability. Tera Fazzino at the University of Kansas developed a quantitative definition of hyperpalatability in 2019, identifying three combinations — fat and sodium, fat and simple sugars, carbohydrate and sodium — that produce reward responses exceeding what any single nutrient generates alone. By her group's analysis, roughly 62% of the US food supply meets the hyperpalatability criteria, and almost all of those products are ultra-processed. The engineered combinations activate the brain's reward circuitry — the nucleus accumbens, the ventral tegmental area — in ways that whole foods at the same caloric content typically don't.

Lower volume and fibre per calorie. Ultra-processed foods tend to be energy-dense. Removing water, fibre, and structural complexity packs more calories into smaller portions. Stomach stretch receptors register volume rather than calories; the same fullness signal arrives at a larger calorie cost. Across the Hall trial's meals, the ultra-processed versions were measurably more energy-dense even after the team's careful matching effort.

Disrupted food matrix. Whole foods deliver their macronutrients within an intact structural matrix — fibre, cell walls, proteins folded into recognisable physical configurations. Ultra-processing breaks this matrix down. Even when reconstituted with similar macronutrient content, the food is digested differently: faster glucose release, weaker satiety hormone response, and less of the slow, sustained signalling that arrives when nutrients have to be released from intact cellular structures.

The reward-circuit angle

Ashley Gearhardt at the University of Michigan, whose work has explored the addiction-like properties of ultra-processed food, has argued that the relevant comparison may not be other foods but other addictive substances. Her group's research on the Yale Food Addiction Scale finds that the foods scoring highest on addiction-like criteria — those that produce loss-of-control eating, withdrawal-like symptoms when stopped, persistent cravings — are almost entirely ultra-processed: pizza, ice cream, chips, chocolate, cookies. No one, in Gearhardt's studies, reports addiction-like patterns around broccoli or grilled fish.

This is not loose metaphor. Nora Volkow's brain imaging work at the National Institute on Drug Abuse has documented overlapping dopaminergic activation patterns between palatable food cues and drug cues in vulnerable populations. The same reward system, recruited by the same neurochemistry, responds to engineered food much as it does to engineered drugs. The food was designed, in many cases, to do exactly this.

Why the satiety arithmetic breaks down

The body's satiety system evolved to manage a food environment where calorie density was constrained by physical properties — water content, fibre content, the structural challenges of breaking food down. Hunger turned off when the stomach filled and when hormones released by stretched-and-digested whole food signalled adequacy. The system worked because food, in evolutionary terms, was almost never engineered to defeat it.

Ultra-processed food removes those constraints. The water is gone or controlled; the fibre is stripped; the structural matrix is dismantled; the fat-sugar-salt combinations are tuned for maximum palatability. The result is food that delivers more calories per unit of fullness signal, faster than the satiety system can process, in formulations the reward circuitry treats as urgent.

None of this is a moral failing on the part of the people eating it. The Hall participants were not weak-willed; they were in a tightly controlled metabolic ward and still consumed 508 extra calories per day they didn't notice. The food was doing the work.

What this means in practice

The Hall trial doesn't argue that every ultra-processed food must be eliminated. It argues that the share of the diet they occupy is a meaningful variable for body weight regulation, independent of macronutrient totals or calorie counting. Shifting some portion of weekly intake toward minimally processed foods — whole grains, plain meats, vegetables, fruits, dairy without engineered formulations — tends to lower spontaneous caloric intake without any explicit restriction.

For people whose appetite biology is already dysregulated — leptin resistance, elevated post-diet ghrelin, the broader pattern that drives clinical obesity — food choices help but often will not be enough. Cravings driven by the reward circuitry respond poorly to behavioural override, which is why the obesity medicine field has shifted toward pharmacological intervention. GLP-1 receptor agonists reduce reward-circuit activation by food cues, slowing the specific pull that ultra-processed food exerts. Combined with shifting the food environment, the effect is multiplicative rather than additive.

For someone considering where to start, the practical move is rarely elimination. It is replacement: substituting ultra-processed items with minimally processed equivalents at the meals where it is easiest to do so, and accepting that the satiety arithmetic will quietly start working in a different direction.

Key takeaways

• Kevin Hall's 2019 NIH randomised trial matched ultra-processed and unprocessed diets for calories, protein, fat, sugar, fibre, and sodium; participants on the ultra-processed arm spontaneously ate 508 more calories per day.
• The NOVA classification, developed by Carlos Monteiro, defines ultra-processed foods by industrial reformulation and additives rarely used in home cooking.
• Ultra-processed foods drive overconsumption through faster eating rate, hyperpalatability (engineered fat-sugar-salt combinations), higher energy density, and disrupted food matrix.
• Tera Fazzino's analysis estimates roughly 62% of the US food supply meets hyperpalatability criteria; almost all are ultra-processed.
• Ashley Gearhardt's research finds that the foods producing addiction-like eating patterns are nearly entirely ultra-processed; no participants report these patterns around whole foods.
• The satiety system evolved for an environment where calorie density was physically constrained; ultra-processed food removes those constraints by design.
• Replacing rather than eliminating ultra-processed foods at the meals where it's easiest tends to lower spontaneous calorie intake without explicit restriction.