You might know a tomato when you see one, but for the federal government, identifying one is harder. Is a tomato a fruit or a vegetable? In 1893, this question went all the way to the Supreme Court, which decided that, for tax purposes, at least, a tomato was a vegetable. What about milk? For a bureaucrat setting standards, a straightforward explanation such as “liquid from a cow’s udders” won’t quite cut it; according to the FDA, milk is “the lacteal secretion, practically free from colostrum, obtained by the complete milking of one or more healthy cows.”Even for the government, however, tomatoes and milk are relatively simple to define compared with ultra-processed foods. Although many researchers agree that these creations are negatively affecting people’s health, they cannot agree on what foods do and don’t belong in that category. That’s not keeping Robert F. Kennedy Jr. from trying: After months of promising the federal government’s first formal definition of ultra-processed foods, the secretary of Health and Human Services announced on Monday that the department is finalizing one.He stopped short, though, of sharing the exact text in either the public announcement or at a “food policy celebration” held Monday at HHS’s Washington headquarters; the definition is still being reviewed by the White House. This effort may have real upsides. Having a regulatory definition of ultra-processed foods (or UPF, as they’re often referred to) could at the very least advance inquiries into the harms of these products. But the effort also raises the question: Can the government adequately define a problem before researchers know its true cause?So far, scientific efforts to define UPF have revealed only how fuzzy the category is. The most commonly used definition, which was developed by Brazilian researchers, takes the most expansive approach: It lumps all foods with industrial ingredients, such as sweeteners and emulsifiers, into the ultra-processed category. A classification system created by academics at the University of North Carolina at Chapel Hill, meanwhile, defines “highly processed foods” as products “processed to the extent that they are no longer recognizable as their original plant/animal source.” Yet another well-respected system, this one designed by a French analytics company, attempts to incorporate additional factors, such as a food’s salt, sugar, and fat content, as well as the “function, number, and potential health risk of additives.”These rubrics are distinct enough from one another that flavored yogurt is an ultra-processed food under the Brazilian system and a moderately processed food under the UNC system. Under the French system, the classification depends entirely on the yogurt itself (even though a paper from the group found that 95 percent of flavored yogurts in French grocery stores were indeed ultra-processed).Researchers disagree on how to define these foods, because the science around how and why they may be harmful is ambiguous. Epidemiological studies have found that UPF-heavy diets have consistently been associated with a range of maladies including cardiovascular disease and type 2 diabetes. But although researchers suspect that certain additives are causing dietary issues, the number of industrial ingredients or additives in Americans’ food is so large that many more may also be causing trouble. Nor do researchers know what processing steps cause everyday food ingredients to go from healthful to harmful. That means Kennedy and his department are essentially trying to define a food group that no one really understands.Kennedy is under no illusions about how little we know about the American food supply. He has frequently criticized a decades-old loophole that allows food companies to put new additives in their food without even telling the FDA. On Monday, he announced a new proposal to fix this issue; if finalized, the rule would require companies to let the FDA know when they’re introducing a new additive. But even that won’t fully solve America’s additive problem.Researchers at the Pew Health Group estimated in 2011 that America’s food supply already features more than 10,000 additives. The new rule would require manufacturers that quietly introduced such compounds to tell the FDA what those additives are but not to submit any scientific proof of the additives’ safety. And although the FDA can do its own review of such ingredients, that process is often glacially slow. (The agency still hasn’t completed its rereview of several chemicals nearly a year after announcing that it was taking a second look at their safety.) Plus, certain additives may be safe when consumed on their own but harmful when consumed alongside other additives. A study published last year found, for example, that consuming a cocktail of emulsifiers, a preservative, and a dye was associated with higher rates of type 2 diabetes but that several other industrial concoctions were not. If anything, researchers know less about how processing changes the ways humans digest certain foods. As the term ultra-processed implies, researchers believe that the ways food companies process individual ingredients has had a profound impact on our health. But because that concept is so new, very little research has been done on it. The Brazilian researchers, for example, have cited pre-frying as a mark of ultra-processing, even though it’s hard to imagine that precooking chicken nuggets could meaningfully affect how they’re digested. And although research does suggest that extrusion, which is used to make products such as cereal by pushing ingredients through a heated tube with rotating screws, makes it easier for the body to break down starches, the evidence of any ill effect on humans is limited.All of this makes Kennedy’s push to define UPF difficult. Any earnest effort to “Make America Healthy Again” cannot ignore the fact that a majority of the calories Americans consume come from ultra-processed foods, and that something in those foods seems to be making people ill. Creating a definition is an obvious first step toward addressing the issue, and the previous definitions for ultra-processed foods have undoubtedly advanced researchers’ understanding of those foods’ impact on our health. A 2019 study using the Brazilian definition found, for example, that people ate faster and gained more weight when fed an utlra-processed diet compared with an unprocessed one.But for researchers, the definition of UPF serves more as a hypothesis—an idea that they can test and adjust about how to think about this category of food. A government definition of UPF has a different purpose—it can prompt research, but it can also be used to make rules about how foods are tested, advertised, and sold. An overbroad definition could cost food companies millions in unnecessary costs and stoke unsubstantiated fear in consumers; a narrower definition, based on a hunch, could cause the same problems.Government definitions are not written in stone: In 2024, the FDA overhauled an outdated definition of healthy foods, which was issued amid the low-fat craze of the 1990s and excluded foods such as nuts and fatty fish. But the process of revising the definition took eight years. As Kennedy said Monday, “If you can’t define a problem, you can’t solve it.” But if you define a problem in the wrong way, you probably can’t solve it either.