I expected the brain scans to be the unnerving part of this study. Instead it was a plain number from its first table: the children in it got, on average, 67.7 percent of their daily calories from ultra-processed foods, and not one of them was overweight.
| Measure | Share |
|---|---|
| of daily calories, on average | 67.7 percent |
All 46 kids were ages 7 to 10, with a mean age of 8.7, and all started below the 85th percentile for BMI, the line for a healthy weight. They’re one small Southern California sample and can’t stand in for every child, but they aren’t an outlier either: the USC release puts ultra-processed food at about 67% of the calories U.S. children eat. I’d quietly filed that kind of diet under “a problem for kids who already have a weight problem.” These children didn’t have one, and two of every three calories they ate still came out of a factory formulation.
Cookies on a screen, and a year on the scale
The study, published in JAMA Network Open by first author Jasmin Alves, senior author Kathleen Page and colleagues at USC’s Keck School of Medicine and Kaiser Permanente Southern California, drew these children from the ongoing BrainChild cohort. Their diets came from two 24-hour recalls, with every food sorted by the NOVA system, whose fourth tier is the ultra-processed one. Then each child lay in an MRI scanner while photos of cookies, French fries and other appealing foods alternated with pictures of things like a pencil or a bench, so the team could see how much harder the brain worked for food. A year later, everyone was weighed and measured again.
Kids who ate more ultra-processed food had a stronger response to the food photos in the dorsal striatum (β = 0.34, 95% CI 0.07 to 0.61), and the kids with the stronger response went on to gain more. The coefficients are in standard-deviation units, so the body-fat result works out like this: a child one standard deviation above the group on dorsal striatal response gained about 0.44 of a standard deviation more in body fat percentage over the year (P = .005). The BMI link was weaker (β = 0.32, P = .04). Diet itself predicted a bigger waist a year on (β = 0.38, P = .009), while its links to BMI and body fat sat right at the edge of significance, at P = .05 for body fat. For a one-year signal in 46 kids who started out lean, that’s about what you’d expect to see.
When the researchers put the brain response into their models alongside diet, the diet–body fat association shrank by 44 percent and the diet–BMI association by 31 percent. That fits the striatum sitting somewhere on the road between the lunchbox and the waistline, but it doesn’t prove it. Attenuation like this can’t establish that the brain caused a given share of the gain, and the authors raise the reverse reading themselves: “children with inherently greater striatal reactivity may be more prone to consuming UPFs rather than the reverse.” The observational design, they write, “precludes causal inference.”
Wait, why the dorsal striatum?
This is where I got curious. The ventral striatum, the spot pop science likes to call the pleasure center, showed no meaningful link to ultra-processed intake (β = 0.16, with a confidence interval from −0.14 to 0.46). The signal sat in the dorsal striatum, which handles valuation and motivation: how much is this worth, and should I go get it? Why would a packaged-food diet track with how a kid’s brain prices food rather than how much she enjoys it?
A 2018 Yale experiment offers one candidate answer. Alexandra DiFeliceantonio, Dana Small and colleagues ran experiments with 206 adults who bid money on snacks that were mostly fat, mostly carbohydrate, or both. The foods were equally familiar, liked and caloric, yet people paid more for the fat-plus-carb combinations, and brain imaging tied that premium to reward-valuation areas including the dorsal striatum. “Our study shows that when both nutrients are combined, the brain seems to overestimate the energetic value of the food,” Small said.
Fat and starch rarely arrive together in one bite in nature, and a cookie, a fry and most of the chip aisle are built on exactly that pairing. So here’s my hypothesis, and it’s only a hypothesis: if a child’s diet keeps serving food the brain overprices, the dorsal striatum is where you’d expect the inflated price tag to show up. The USC data can’t test that. Striatal activity was measured once, so nobody yet knows whether the response grew with the diet, came before it, or would fade if the diet changed. The Yale work was in adults, and fat-carb combinations aren’t the same category as NOVA’s ultra-processed tier. Even so, I find a pricing explanation more unsettling than a pleasure one, because pleasure ends with the snack and a price can shape the next choice.
One more wrinkle. In a post hoc analysis, adjusting for added sugar pulled the ultra-processed–striatum link down to β = 0.28 (P = .07), no longer significant. Added sugar on its own wasn’t significantly tied to the striatal response either (β = 0.29, P = .06). The two travel together in these kids’ diets (r = 0.35), and with 46 children the statistics can’t cleanly separate them, so sugar neither gets off the hook nor takes all the blame.
The causal test has been run, in adults
In children the cause-and-effect question is still open, but adults have been tested directly. In Kevin Hall’s 2019 inpatient trial at the NIH Clinical Center, 20 adults ate ultra-processed and unprocessed diets for two weeks each, with the meals on offer matched for calories and macronutrients and everyone free to eat as much as they liked. On the ultra-processed diet they ate about 508 calories a day more and gained roughly 0.9 kilograms; on the unprocessed diet they lost about the same. Nobody’s striatum got scanned, so the trial can’t confirm the USC mechanism, but it does show the diet pattern itself can drive people to overeat.
Page wants to run the experiment in kids next: give children who usually eat a lot of ultra-processed food more minimally processed meals, then see whether their brain response to food photos and their weight gain change. “Children’s brains are still developing, habits are still forming and food preferences are still being made,” she told ScienceDaily. The work was funded by the American Diabetes Association and NIH’s diabetes institute, and the authors reported no conflicts of interest.
Washington still hasn’t written the definition down
HHS Secretary Robert F. Kennedy Jr. has put ultra-processed food at the center of his chronic-disease agenda. “Nearly 60% of the American diet is made up of ultra-processed foods,” he said. “We cannot reverse America’s chronic disease epidemic without transforming our food system.” HHS and FDA have sent a proposed federal definition to the White House Office of Management and Budget for final review, more than a year after FDA and USDA’s July 2025 request for information. Food Engineering reported no legal deadline for that review, and as of mid-September the definition wasn’t on the Federal Register.
Researchers can already sort a lunchbox with NOVA, as the USC team did. What’s missing is a single federal standard that labels, school meal rules or purchasing decisions could point to. Until one exists, no agency rule has to name the crackers and snack cakes that made up most of these lean children’s calories, and the longer the text sits in review, the longer the companies selling them go without one. Kennedy is right to treat this as a food-system problem rather than a willpower problem, and the system is still waiting on a sentence of federal text.
If I were packing an eight-year-old’s lunch tomorrow, I’d still glance at the sugar line, but my first move would be swapping the packaged fat-and-starch combinations (the crackers, the chips, the snack cakes) for food that still looks like where it came from, and I wouldn’t wait for Washington to finish its definition before doing it.
Sources
- JAMA Network Open: Alves et al., Ultraprocessed Food Intake, Striatal Response to Food Cues, and Adiposity Gain in Children (2026)
- ScienceDaily: Kids who eat more ultra-processed foods show a troubling brain response (USC release, 2026)
- Cell Metabolism: DiFeliceantonio et al., Supra-Additive Effects of Combining Fat and Carbohydrate on Food Reward (2018)
- Neuroscience News: Foods combining carbs and fats more rewarding than carbs or fats alone (Yale, 2018)
- NIH: NIH study finds heavily processed foods cause overeating and weight gain (Hall et al., 2019)
- Food Engineering: HHS, FDA submit proposed definition of ultra-processed foods for final review (2026)
- MedicineNet: Ultra-Processed Foods Might Rewire Kids’ Brains To Crave Junk, Study Says (2026)