Source notes

Can the kinds of foods available change how much people eat without anyone telling them to restrict calories? Twenty adults lived in a research unit for four weeks, spending two weeks on each of two menus in randomly assigned order. They could eat as much or as little as they wanted. On the ultra-processed menu, they ate about 508 more calories per day on average than on the unprocessed menu. They gained about 0.9 kilograms during the ultra-processed period and lost about 0.9 kilograms during the unprocessed period. These were short-term changes under tightly controlled conditions.

The experiment supports a causal effect of the menus tested on food intake. It does not isolate processing itself from every other food characteristic. The meals offered were matched for several nutrients, but differed in features such as the calorie density of foods excluding beverages. Participants also ate faster on the ultra-processed menu. Their ratings of hunger and how pleasant the food tasted did not explain the difference in intake.

How Does This Show Up in Real Life?

The study changes how a choice between meals can be understood. Two menus can look similar on selected nutrition totals yet lead people to consume different amounts when portions are unrestricted. That makes the eating environment relevant, rather than placing all the explanation on deliberate calorie decisions. The authors suggest limiting ultra-processed foods as a possible approach to preventing or treating obesity, while recognizing the practical burden of preparing alternatives. Their participants received meals without having to buy ingredients, cook, or manage time. A household choosing dinner has to handle those constraints. The findings therefore support considering the overall foods that make up a diet, but do not supply a test for judging every packaged product. Nor do they establish that one particular ingredient caused the effect. The comparison involved complete menus, not a single snack or an isolated additive.

The sample was small, and each diet lasted only two weeks. There was no washout period between menus, although the intake difference remained when researchers compared the final week of each period. Longer-term weight change and persistence after leaving the research unit were not tested. The study also found no significant difference in glucose tolerance between the two diets over this short interval.

This is unusually direct evidence about eating behavior because intake was measured while the food supply was controlled. Its precision should not obscure its boundaries: a short inpatient experiment can demonstrate an effect while leaving the reasons and everyday durability unresolved.