Ultra-Processed Food and Attention: What a Study of 2,100 Adults Found
Poorer attention showed up in people eating more ultra-processed food even when their weight and bloodwork looked fine. Here is what that link means, and what it can't yet prove.
The afternoon fog is not a personality trait. Sometimes it is a receipt.
You know the version of yourself that shows up around three in the afternoon. Reads the same paragraph three times and retains none of it. Opens a tab, forgets why, closes it, opens it again forty seconds later. For years I filed that under sleep debt, or too many meetings, or just being a person in the back half of a long day. The thing I never seriously interrogated was breakfast.
A study of more than 2,100 adults has put that assumption under some pressure. Researchers measured what share of each person's diet came from ultra-processed food, then ran them through standard cognitive testing — the kind that times how fast you can sort symbols, hold a sequence in working memory, or track one thing while ignoring another. Higher ultra-processed intake tracked with poorer attention and slower mental processing speed.
The finding itself is not the surprising part. Where it held is. The association persisted in people whose diets were otherwise reasonable, and in people at a normal weight — which is inconvenient for the tidiest explanation we had. The comfortable story was that ultra-processed food hurts the brain because it makes you heavier and drags your blood sugar around all day. If the effect shows up in people it did not make heavier, that story is at best incomplete.
What "Ultra-Processed" Actually Means
The term gets used as a synonym for junk food, which is why the conversation around it stays muddy. It isn't one.
The definition comes from the NOVA classification, developed by Carlos Monteiro and colleagues at the University of São Paulo in 2009. It sorts food by how much industrial transformation it has undergone, not by how many calories or grams of fat it contains. Group one is unprocessed and minimally processed: an apple, rice, milk, a chicken thigh. Group two is culinary ingredients: oil, butter, salt, sugar. Group three is processed food — group one plus group two, essentially, made by methods a home cook would recognise. Canned beans. Cheese. Bread from a bakery with four ingredients in it.
Group four is the one under scrutiny. Ultra-processed foods are industrial formulations built substantially from substances you would never find in a home kitchen: protein isolates, modified starches, invert sugars, hydrogenated oils, plus the cosmetic additives that hold the whole thing together and make it palatable — emulsifiers, thickeners, flavourings, colourings, non-nutritive sweeteners.
Read that definition carefully and you notice something uncomfortable. It has almost nothing to do with whether a food is "healthy" in the way we usually mean. Most supermarket bread is ultra-processed. So is most breakfast cereal, most flavoured yoghurt, most plant milk, most protein bar, most deli meat, most low-calorie ready meal marketed at exactly the person trying to eat better. Meanwhile butter, cream and lard sail through as group two. This is not an oversight; it's the point. The classification is asking a different question than the nutrition label is.
It's also a category with genuinely fuzzy edges, and its critics are not being unreasonable when they say so. A wholegrain bread with an emulsifier in it and a supermarket sponge cake land in the same bucket, which is obviously crude. Keep that looseness in mind — it matters when we get to what the research can and cannot support.
What the Study Found, Precisely
Precision matters here, because the headline version of this study is broader than the study.
The deficits clustered in two specific domains: attention and processing speed. Not memory across the board, not general intelligence, not anything resembling dementia. In practical terms, the sort of thing these tests capture is how long it takes you to pick up a task, how easily a competing signal knocks you off it, and how quickly you can move through material that requires steady low-level effort. Which is a fairly good description of ordinary work.
The relationship was graded rather than binary. There was no cliff edge where a person crossed into "bad diet" territory and their scores collapsed. More ultra-processed food, somewhat worse performance, across the range. That gradient is one of the more informative details in the whole thing — it suggests something continuous rather than a threshold effect.
And the association survived adjustment for body weight and for the overall quality of the rest of the diet. That's the finding driving the new scrutiny. It means the effect is not simply riding along on obesity, and it's not simply that people eating a lot of ultra-processed food are also eating no vegetables. Something appears to attach to this category of food specifically.
Four Explanations, None Yet Confirmed
Nobody has established the mechanism. There are four serious candidates, and they are not mutually exclusive — the honest expectation is that several are partly true.
Low-grade inflammation. Diets high in ultra-processed food are associated with elevated inflammatory markers, and inflammation has a well-documented relationship with cognitive performance. The brain is not walled off from the rest of the body's immune signalling; when you feel mentally sluggish during a mild illness, you are experiencing a version of this. The theory is that a chronically low hum of the same process produces a chronically low hum of the same fog.
The gut. Emulsifiers and certain non-nutritive sweeteners have been shown in animal work to alter the gut microbiome and the integrity of the intestinal lining. The gut communicates with the brain constantly — through the vagus nerve, through immune signalling, through metabolites produced by bacteria. If you change what lives in the gut, you change part of that conversation. This is the most exciting theory and also the one where the evidence in humans is thinnest.
Displacement. The least glamorous explanation and possibly the strongest. Every ultra-processed meal is a meal that isn't something else. Fibre, polyphenols, omega-3 fats, magnesium, choline — the brain runs on this stuff, and it comes overwhelmingly from foods in group one. Under this theory, ultra-processed food isn't doing anything to you. It's occupying the slot where something useful would have been. Notably, the study adjusted for overall diet quality, which pushes back on this — but adjustment is a statistical instrument, not a wall.
The additives themselves. The most direct hypothesis: some of the compounds engineered into these products have neurological effects at the doses people actually consume. This is plausible and largely untested. Individual additives are safety-tested in isolation, at doses set decades ago, mostly for endpoints like organ toxicity rather than reaction time on a symbol-sorting task. The interaction of thirty of them in a day is nobody's study.
What This Study Cannot Prove
It's observational. It cannot establish that ultra-processed food causes anything, and I'd rather say that plainly than bury it.
Three specific limits are worth holding on to. First, dietary intake in studies this size is almost always collected through food frequency questionnaires — people reporting from memory what they generally eat. This is a famously noisy instrument. People underestimate the things they feel bad about eating, and the underestimating is not random.
Second, residual confounding. Ultra-processed intake correlates with income, education, working two jobs, shift work, sleep quality, chronic stress, neighbourhood food access. Every one of those independently affects attention. Statistical adjustment attempts to strip those out; it never fully succeeds, because you can only adjust for what you measured, and you measured it imperfectly.
Third — and this one is underrated — causation may point backwards. A tired, depleted, cognitively overloaded person reaches for food that requires no decision, no preparation and no cleanup. That is not a moral failing; it's an entirely rational response to having nothing left. If low attention drives ultra-processed intake rather than the reverse, the correlation looks identical.
What would settle it is a randomised controlled trial. The closest thing we have is Kevin Hall's 2019 NIH metabolic ward study, where twenty adults lived in a research facility and ate ultra-processed or minimally processed diets, matched for calories, sugar, fat, fibre and sodium. On the ultra-processed diet, people spontaneously ate about 500 more calories a day and gained weight. That's a genuinely causal finding about eating behaviour. It was four weeks, twenty people, and it did not measure cognition. Nobody has yet run its equivalent for attention.
So: a real signal, an unproven mechanism, and a plausible alternative explanation that hasn't been ruled out. That's the actual state of things.
Trimming It Without the All-or-Nothing Spiral
Here is what I've watched happen — to myself and to nearly everyone I know who has read a study like this one. You go home, open the pantry, read labels for ninety minutes, conclude that essentially everything you own is poison, throw out sixty dollars of food, eat impeccably for eleven days, and then have a hard Wednesday and abandon the whole project. Nothing about that sequence improved anyone's attention span.
The framework that survives contact with a real week looks different.
Count before you cut. Spend one week changing nothing and simply noticing. Not logging calories — just a rough tally of which ultra-processed items actually recur. Most people's list is shorter than they expect and different from what they'd have guessed. You cannot fix a pattern you haven't seen.
Go after volume, not villains. The instinct is to eliminate the food with the worst reputation, which is usually something you eat once a fortnight. The leverage is in the thing you consume five days a week without registering it — the bread, the cereal, the flavoured coffee drink, the afternoon bar. One swap in the daily slot beats four swaps in the occasional slot.
Upgrade the default; leave the treat alone. This is the whole trick. Change what happens automatically — the breakfast you don't think about, the lunch you buy without deciding — and leave the deliberate pleasures completely untouched. The ice cream on Friday night is not the problem, and treating it as one is exactly what makes people quit.
Add before you subtract. Put one whole-food item into a meal rather than removing something. Eggs alongside the cereal. Nuts in the yoghurt. A pear next to the sandwich. Addition doesn't trigger the deprivation reflex, and it does the displacement work in the direction you want.
Protect the meal before your hardest work. If any of this matters for attention, it matters most in the two hours you actually need your attention. I've become fairly indifferent about what I eat at 8 p.m. and fairly deliberate about what I eat before a morning of writing code. That's not a rule anyone gave me; it's just where the return seems to be.
Watch the language. "Clean" and "dirty" are the vocabulary of a spiral. A food is more or less processed. It's a dial, not a verdict on your character, and the moment it becomes a verdict you are three weeks from quitting.
What I keep returning to is how modest the ask is. Nobody in this research is suggesting that a life of unprocessed food will make you brilliant. The claim on the table is smaller and stranger: that some part of the mental fog we've accepted as the standard texture of a Tuesday afternoon might be coming from somewhere, and that somewhere might be reachable. Even at a fraction of the effect the study suggests, that would be worth knowing.
My meditation practice has taught me one thing that transfers here directly. Attention is not a fixed quantity you're issued at birth. It's a condition — it has inputs, it responds to how you've treated yourself, and it is more negotiable than it feels at three in the afternoon.
Questions People Actually Ask
Is ultra-processed just a fancy way of saying junk food?
No, and this is the most common misreading. The classification is about industrial formulation, not nutritional quality. Butter and lard are not ultra-processed. Most supermarket wholegrain bread, most protein bars and most low-fat yoghurts are. A food can be ultra-processed and still be a reasonable thing to eat, and plenty of ultra-processed foods — fortified cereals, for instance — deliver nutrients people genuinely need.
How much is too much?
No threshold has been established, and be suspicious of anyone who gives you a number. The relationship in this study was graded, meaning less appears to be better across the range rather than there being a line you cross. Practically, that's freeing: any reduction is a real reduction, and there is no score you have to hit.
Will cutting back actually make me sharper?
Unknown. This study shows an association, not a causal effect, and no trial has yet tested whether reducing intake improves attention. What can be said is that the changes involved are low-cost and low-risk, and that the same swaps have decent evidence behind them for other reasons. That's a reasonable basis for acting under uncertainty — just don't expect a transformation, and don't conclude the science was wrong if you don't feel one.
I have almost no time to cook. Is this realistic for me?
Partly, and the honest version of this answer matters more than an encouraging one. Ultra-processed food solves a genuine problem — time, money, energy, decision-making capacity — and a person working two jobs is not going to fix their diet with a framework from a blog. The realistic move is one default meal, usually breakfast, because it's repetitive and low-stakes. That's it. Anything more is a plan for someone with a different Tuesday than yours.