Ultra-processed foods account for 53.0 percent of the calories American adults consume, according to CDC data collected between August 2021 and August 2023. For children and teenagers, that figure rises to 61.9 percent. This is not a fringe dietary pattern. It is the default one.
That makes it a reasonable question to ask what this is doing to us, and the question drawing the most attention lately is about the brain. Headlines have linked ultra-processed food to dementia risk, memory loss, and cognitive decline. Some of that reporting is careful. Much of it is not.
The honest version is more interesting than the alarming one. Human research on this topic is genuinely mixed. Some large studies find a connection between ultra-processed food consumption and cognitive outcomes. Several others find no meaningful association at all when looking at total ultra-processed intake. Where the science is clearest is not in the population studies. It is in the underlying mechanism, and that mechanism runs directly through the gut.
Why "ultra-processed" Is a Slippery Category
Ultra-processed is a classification, not an ingredient. It comes from the NOVA system, which sorts food by how much industrial processing it has undergone rather than by nutrient content. That means industrial whole-grain bread and a sugary snack cake land in the same category, even though they behave very differently in the body.
This distinction matters. When researchers study total ultra-processed food intake as a single number, they are averaging together things that may have almost nothing in common biologically. That is one reason study results disagree so often. The more useful question is not whether a food is ultra-processed, but what specifically was done to it and what was removed during processing.
Two answers come up consistently:
Fiber was stripped out, and industrial additives were introduced.
Start With What Is Missing: Fiber and Fermentation
Your large intestine is home to a dense microbial community, and that community needs to eat. Its preferred food is fermentable fiber found inside whole plant foods. These are the structural carbohydrates that survive your small intestine intact and arrive in the colon still usable. For a closer look at how different fiber types behave, our post on soluble versus insoluble fiber covers the distinction well.
When bacteria ferment fiber, they produce short-chain fatty acids, primarily butyrate, propionate, and acetate. This fermentation process is one of the most consistently studied functions of the microbiome. It is also what ultra-processing quietly removes. Industrial processing strips the food matrix, leaving behind something easy to chew, easy to digest, and shelf-stable, but largely absorbed before it ever reaches the colon. The bacteria get very little to work with.
This is the least contested part of the story. A diet built on ultra-processed food gives the microbial community less to ferment.
Then What Was Added: Emulsifiers and the Mucus Layer
Emulsifiers keep the oil and water phases of a processed product from separating on the shelf. Two of the most studied are carboxymethylcellulose (CMC) and polysorbate 80 (P80).
In the gut, a mucus layer keeps the bacterial community physically separated from the cells lining the intestine. Chassaing and colleagues, publishing in Nature in 2015, measured that separation directly. Bacteria normally sit roughly 25 micrometers away from the epithelial cells, held at a consistent distance.
In controlled mouse studies, adding CMC or P80 to drinking water for 12 weeks eroded that distance. Bacteria encroached on the inner mucus layer, microbiota composition shifted, and the animals developed low-grade inflammation alongside features of metabolic syndrome. The effect followed a dose gradient. As little as 0.1 percent CMC produced modest weight increases, while 0.5 percent produced clear low-grade inflammation. When researchers transferred the altered microbiota into germ-free mice, the changes were sufficient to produce both the inflammation and the metabolic effects, suggesting the bacteria were doing the work rather than the additive acting directly on tissue. Effects persisted for at least six weeks after the emulsifiers were removed. All of that is mouse data.
The one controlled human trial is small and deserves careful description. Chassaing and colleagues published a double-blind controlled feeding study in Gastroenterology in 2022 with 16 participants: 9 on control and 7 consuming 15 grams of CMC per day for 11 days. In the treated group, microbiota diversity dropped, short-chain fatty acids and free amino acids in the stool metabolome declined, and participants reported modestly more abdominal discomfort after meals. Three findings deserve equal attention. Serum inflammatory cytokines did not change. Only 2 of the 7 treated participants showed mucus layer encroachment. And person-to-person variation was larger than the effect of the intervention itself.
The fair summary is that a high dose of one emulsifier can measurably shift the human microbiome and its metabolic output within days, that individual responses vary considerably, and that this short trial did not produce a systemic inflammatory signal.
The Mouse Behavior Studies, Read Correctly
This is where reporting tends to go off the rails, so the details matter.
Holder and colleagues, publishing in Scientific Reports in 2019, gave C57Bl/6J mice 1 percent CMC or 1 percent P80 in drinking water for 12 weeks and ran six behavioral tests using 5 to 6 animals per group. Emulsifier-treated males spent less time in the center of the open field, a standard index of anxiety-like behavior. CMC-treated females showed a reduced preference for a novel mouse in the social novelty test. Inflammation was indexed by shorter colons and heavier spleens rather than by a cytokine panel.
A separate 2025 paper by Arnold and colleagues in Hormones and Behavior, using male mice only, reported that dietary emulsifiers increased sensitivity to social stress. That study had different authors and a different design, and its findings should not be folded into the 2019 results.
The authors of the 2019 paper included a caveat worth quoting directly, as it is the most important sentence in the whole literature for a reader trying to decide what this means for them. They wrote that while determining the extent to which studies in mice are relevant to humans is inherently difficult, even in studies of metabolism, it is especially hard to do so for behavioral disorders, whose complexity and heterogeneity make them difficult to model in mice.
None of these studies measured memory. They measured anxiety-like and social behavior in small groups of animals at concentrations far above ordinary human exposure. For a broader look at the relationship between mental wellness and digestion, our post on gut health and anxiety is the better starting point.
What the Human Cognitive Research Actually Found
Two large cohorts found associations, and neither tells the story that usually gets repeated.
Gonçalves and colleagues, publishing in JAMA Neurology in 2023, followed 10,775 Brazilian public servants aged 35 to 74 for a median of eight years. Ultra-processed food was measured as a percent of daily calories, averaging 27.4 percent. Compared with the lowest-intake quartile, the higher-intake group showed a 28 percent faster rate of global cognitive decline and a 25 percent faster rate of executive function decline. Two caveats belong with those numbers. The confidence interval for executive function reached 0.000, and the quartile-by-quartile pattern was not a clean gradient. Global cognition decline was 21 percent faster in the second quartile, 28 percent in the third, and back down to 21 percent in the highest quartile. The memory domain showed no association at all.
Li and colleagues, publishing in Neurology in 2022, followed 72,083 UK Biobank participants aged 55 and older for a median of 10 years and recorded 518 incident dementia cases. Each 10 percent increase in ultra-processed food intake was associated with a 25 percent higher risk of all-cause dementia. Substituting that 10 percent with unprocessed or minimally processed foods was associated with a 19 percent lower risk.
These two studies are frequently cited together, but they should not be treated as equivalent. The Brazilian cohort measured ultra-processed food as a percent of calories. The UK Biobank cohort measured it as a percent of daily food intake by weight in grams. Those are different exposures, and a 10 percent shift means something different in each.
The studies that found no association rarely get reported.
Buis and colleagues, publishing in the European Journal of Nutrition in 2026, followed 1,371 Dutch adults aged 55 and older with cognition assessed four times between 2011 and 2022 and found no association with cognitive function or decline in any domain. Seago and colleagues, in the American Journal of Clinical Nutrition in 2025, followed 4,750 US adults for seven years and found no association with total ultra-processed food intake, with only two subcategories showing any signal: ultra-processed animal products and ultra-processed beverages. Weinstein and colleagues found the same shape twice, first in 568 older adults with type 2 diabetes where only ultra-processed meat and oils and spreads mattered, and again in a Framingham Offspring analysis of 1,375 participants where the association held only in people under 68 at baseline. Gauci and colleagues, studying 11,502 Australians aged 70 and over, found poorer global cognition, verbal fluency, and processing speed at high intakes but no effect on the memory test, and effect sizes were small.
Taken together, the pattern is consistent, and it is not the headline version. Associations with total ultra-processed food are inconsistent across cohorts. Where a signal appears, it tends to show up in executive function and global cognition rather than memory, and it is often carried by specific subcategories rather than the broader classification. That is why the mechanistic frame is more useful. It points at fiber, additives, and fermentation, which are factors that can actually be acted on, rather than at a label classification that groups bread and candy into the same category.
It is also worth being clear about where the evidence stops. The connection between a low-fiber, additive-heavy diet and changes in the gut is well documented. The connection between those gut changes and measurable changes in the human brain is inference, drawn from animal research and observational cohorts rather than from controlled human trials with cognitive endpoints. Those trials do not yet exist. Our post on the gut-brain connection walks through how the microbiome and brain are thought to communicate and where the open questions remain.
Where Your Stool pH Fits In
There is one part of this chain you can observe at home, and it is worth understanding what it does and does not tell you.
An ultra-processed diet is low in fermentable fiber. Less fermentable fiber means less fermentation in your colon, fewer short-chain fatty acids produced, and less acid overall. That shows up as a higher stool pH.
The Coprata Microbiome Activity Tracker measures that pH as an indicator of fermentation activity. It reflects whether your microbial community is actively working on what you feed it. It is not a dementia risk score. No study has linked a specific stool pH value to a specific score on a cognitive test. If you want to understand why pH is used as a practical stand-in for direct short-chain fatty acid measurement, our post on how SCFAs are actually measured explains the tradeoffs.
The MAT kit is not a medical device. It is for informational and educational purposes only and cannot diagnose, treat, cure, or prevent any disease, including dementia, anxiety, or cognitive decline.
The Bottom Line
The case against ultra-processed food does not rest on a mouse maze or a dementia headline. It rests on something simpler and better established. Industrial processing removes the fermentable fiber gut bacteria depend on and adds compounds that, in animal studies and one small human trial, disrupt the microbial community and its output.
The human cognitive research is real but unsettled. The signal sits in executive function and global cognition rather than memory, and it is likely driven by specific components rather than the ultra-processed category as a whole. That is a less dramatic story! And a more actionable one. The fix is not avoiding a label. It is putting back what the processing took out.
References
Chassaing B, Koren O, Goodrich JK, Poole AC, Srinivasan S, Ley RE, Gewirtz AT. Dietary emulsifiers impact the mouse gut microbiota promoting colitis and metabolic syndrome. Nature. 2015;519(7541):92-96. Corrigendum: Nature. 2016;536:238.
Chassaing B, et al. Randomized Controlled-Feeding Study of Dietary Emulsifier Carboxymethylcellulose Reveals Detrimental Impacts on the Gut Microbiota and Metabolome. Gastroenterology. 2022;162(3):743-756.
Holder MK, Peters NV, Whylings J, Fields CT, Gewirtz AT, Chassaing B, de Vries GJ. Dietary emulsifiers consumption alters anxiety-like and social-related behaviors in mice in a sex-dependent manner. Scientific Reports. 2019;9:172.
Arnold AR, Chassaing B, Lakhani K, et al. Consumption of dietary emulsifiers increases sensitivity to social stress in mice: A potential role for the COX molecular pathway. Hormones and Behavior. 2025;172:105750. Male mice only.
Gonçalves NG, Ferreira NV, Khandpur N, et al. Association Between Consumption of Ultraprocessed Foods and Cognitive Decline. JAMA Neurology. 2023;80(2):142-150.
Li H, et al. Association of Ultraprocessed Food Consumption With Risk of Dementia. Neurology. 2022;99(10):e1056-e1066.
Buis C, Nicolaou M, Visser M, Olthof MR, Wijnhoven HAH. Ultra-processed food intake and cognitive decline in older adults. European Journal of Nutrition. 2026;65(2):66.
Seago ER, Rego MLM, Davy BM, Katz B. Ultra-processed food intake and cognitive impairment in older US adults. American Journal of Clinical Nutrition. 2025;121(5):965-971.
Weinstein G, et al. Ultra-Processed Food Consumption and Cognitive Performance in Older Adults With Type 2 Diabetes. Journal of Gerontology: Series A. 2023;78(1):134-142.
Weinstein G, et al. Ultra-processed food consumption and cognitive decline in the Framingham Offspring Study. Journal of Prevention of Alzheimer's Disease. 2025;12(2):100042.
Gauci S, et al. Ultra-processed food consumption and cognition in older adults: ASPREE. GeroScience. 2026.
Williams AM, Couch CA, Emmerich SD, Ogburn DF. Ultra-processed Food Consumption in Youth and Adults: United States, August 2021 to August 2023. NCHS Data Brief No. 536. August 2025.
