A bag of salty sticks disappears in ten minutes in front of a series, and you wonder where it went. A plate of potatoes and chicken does not do that. The difference has a measured cause. There is nothing mystical in the bag.
Short answer: You eat ultra-processed food faster, and the fullness signal reaches the brain tens of minutes late. In Hall's 2019 trial people ate 508 kcal (calories) a day more on an ultra-processed diet, even though both menus offered matched calories and macros. Forde's analysis put the speed at 48 kcal a minute versus 31.
The trial where both diets offered the same calories and macros
Kevin Hall and his team kept twenty adults on a ward at the US National Institutes of Health for a month (Hall 2019, Cell Metabolism, PMID 31105044). Each person spent two weeks on ultra-processed food and two weeks on unprocessed food, in random order. Both menus matched the calories on offer, energy density, macros, sugar, salt and fiber. The instruction was simple: eat as much as you want.
On the ultra-processed menu people ate 508 kcal a day more. The extra energy came from carbohydrate (+280 kcal) and fat (+230 kcal). Protein intake stayed almost the same. In two weeks on the processed menu they gained 0.9 kg. In two weeks on the unprocessed menu they lost 0.9 kg. Weight change tracked intake, with a correlation of 0.8.
Notice what did not happen. Thermodynamics worked normally that month. People gained because they ate more, and they ate more because the food let them, without them noticing. Ultra-processed food does not bypass a calorie deficit. It makes the estimate worse. The deficit itself is in calorie deficit: how much to eat.
Eating speed is the lever almost nobody mentions
Ciaran Forde, a co-author on Hall's study, then looked at how fast people ate in that trial (Forde 2020, Current Developments in Nutrition, PMID 32110771). On ultra-processed food the rate was 48 kcal a minute. On unprocessed food it was 31 kcal a minute. More than half again as fast.
The same team then pooled five studies and 327 foods. Average intake rate stepped up with the NOVA groups (NOVA sorts foods into four groups by how much industry has processed them; ultra-processed is the fourth): unprocessed 35.5 kcal/min, processed 53.7 kcal/min, ultra-processed 69.4 kcal/min. The authors add a note that almost nobody quotes: the spread inside each group is huge, and NOVA by itself offers no proven mechanism.
The fullness signal arrives tens of minutes late. If you get the same energy in half the time, "I have had enough" shows up when you are already a few hundred calories ahead.
Texture and density beat the label
Pey Sze Teo and Forde tested this directly (Teo 2022, American Journal of Clinical Nutrition, PMID 35285882). Fifty people ate four lunches in turn: soft and hard in a minimally processed version, soft and hard in an ultra-processed version. The harder versions were eaten more slowly, and people took in 26% less energy from them. The hard minimally processed lunch was the smallest (483 kcal). The soft ultra-processed lunch was the largest (789 kcal). That is about 300 kcal in one meal, and the whole day's intake moved the same way, by about 15%.
A later trial in Wageningen made it even clearer (Lasschuijt 2025, Journal of Nutrition, PMID 40516651). Sixty-nine people got sandwiches that differed only in eating speed and energy density. Slow and energy-dilute: 570 kcal. Fast and dense: 1,143 kcal. A gap of 573 kcal, a doubling. The important detail: every version got 67 to 98% of its energy from ultra-processed ingredients. Same label, double the intake.
Hardness, volume and energy density are levers you can move in a normal shop, without banning anything.
What the big observational data show
A 2024 BMJ review pooled 45 analyses covering almost ten million people (Lane 2024, PMID 38418082). The strongest findings link a higher intake of ultra-processed food with higher risk of cardiovascular death (RR 1.50) and type 2 diabetes (RR 1.12). A step weaker are all-cause death (RR 1.21) and obesity (OR 1.55).
Before anyone writes a headline, read what the authors say about quality. On the GRADE scale, 22 analyses were low quality, 19 were very low, and only four were moderate. Even the strongest number, 1.50, sits on evidence the authors themselves rank among the weaker set. These are observational data: a link, not a proven cause. Someone who eats a lot of ultra-processed food usually also differs in smoking, movement, sleep and total intake, and all of that lands in the numbers. Talk to your doctor about a diagnosis. This article does not treat one.
What I do with clients
Bans work worst of all here. What holds up:
- Put about 80% of calories on whole food. Leave the other 20% for anything that fits the day's intake.
- Keep protein at 1.6 to 2.2 g per kg of body weight a day. It fills you and protects muscle in a deficit. See how much protein.
- Get 25 to 35 g of fiber a day, mostly from vegetables, pulses and fruit.
- Pick the harder, bulkier version of the same food. A whole apple instead of juice, potatoes instead of mash, nuts instead of a bar.
- When you eat something soft and energy-dense, put it on a plate and weigh it. A guess by eye is further off than people admit. I see it at almost every first look at a food log.
The last point saves the most calories. A sloppy log is usually a bigger problem than the macro split. The plate itself is in what to eat to lose fat.
Frequently asked questions
Are ultra-processed foods unhealthy by themselves?
Processing does not turn food into poison. Yogurt, whey protein, oil and ham are processed and they have a place. The catch in NOVA group four is that it is easy to overeat, and it often brings less protein, fiber and micronutrients.
Can I lose fat if I do not cut ultra-processed food out?
Yes, if you hold a calorie deficit. In studies where people ate freely, they ate more of it. When you measure intake, the advantage of whole food is mostly fullness and ease, not a break in thermodynamics.
Do I have to eat slowly?
Slower eating reliably lowered intake in the experiments, so it makes sense. A more practical move than watching a stopwatch is to pick food that forces a slower pace by itself: harder and bulkier.
Want me to coach you?
Gold coaching in English, fully written. A weekly report from your phone (3 minutes), my written answer with your plan, WhatsApp support Monday to Friday, calories and macros adjusted every week, a plan for eating out and travel. No video calls needed.
Apply for coaching Get the free Fat Loss KitSources, so you can check:
1) Hall KD et al., Cell Metab 2019;30(1):67-77. PMID 31105044, DOI 10.1016/j.cmet.2019.05.008
2) Forde CG, Mars M, de Graaf K, Curr Dev Nutr 2020;4(3):nzaa019. PMID 32110771, DOI 10.1093/cdn/nzaa019
3) Teo PS et al., Am J Clin Nutr 2022;116(1):244-254. PMID 35285882, DOI 10.1093/ajcn/nqac068
4) Lasschuijt MP et al., J Nutr 2025;155(8):2602-2610. PMID 40516651, DOI 10.1016/j.tjnut.2025.06.006
5) Lane MM et al., BMJ 2024;384:e077310. PMID 38418082, DOI 10.1136/bmj-2023-077310
Martin Barna
Online nutrition and fitness coach. Since 2013 I have helped 600+ people lose fat and keep the result. Science, not crash diets.
Apply for coachingThis article is for information only. It is not medical advice. If you have a health condition, talk to your doctor.