A Bag of Chips a Day Took a Bite Out of Attention
Key Findings
Each 10 percentage-point bump in ultra-processed food (UPF) was tied to a small drop in attention performance, ฮฒ = โ0.05 z-score (a small but real attention dip; 95% CI โ0.09 to โ0.01; p = 0.012). The translation the authors themselves use: a 10% UPF increase is roughly one bag of chips a day on top of a 1.5 kg/day food intake.
The signal held up after the researchers controlled for Mediterranean diet adherence and BMI. So the effect isn't just "ultra-processed eaters have worse diets overall." Something about the processing itself looks like it's doing work.
The top fifth of UPF eaters (28% to 73% of food by weight from UPF) scored noticeably lower on attention than the bottom fifth (4% to 13%): ฮฒ = โ0.21 z-score (a meaningful gap; 95% CI โ0.32 to โ0.09; p-trend = 0.006).
The middle three groups didn't separate cleanly from the bottom group. So the effect is loaded into the heaviest UPF eaters, not a smooth slide across everyone.
Memory scores didn't budge. ฮฒ = 0.01 (basically zero; 95% CI โ0.02 to 0.04; p = 0.502). Top quintile vs. bottom quintile? Also null (ฮฒ = 0.04, p-trend = 0.427).
This kills the "UPF causes Alzheimer's" headline before it starts. The hit in this study landed on attention, not memory. Anyone telling you ultra-processed food is causing Alzheimer's is going further than this paper supports.
The paper reports two dementia risk scores. The original CAIDE score (the one validated as a 20-year dementia predictor) showed a weak link to UPF that lost statistical significance once Mediterranean diet was added to the model (p went from 0.037 to 0.061, just past the cutoff).
The score that survived the adjustment is a Modified CAIDE that the authors built from the modifiable factors only. It rose by 0.24 points per 10% UPF increase (the largest effect in the paper). The catch: the modified version isn't a validated dementia-prediction tool. The authors call it "exploratory" themselves. We're flagging it the same way.
Why It Matters (For You, Not the Kid)
Most Loot Drop issues are about your kid. This one is about you. The study looked at adults 40 to 70, not children. Stay with us anyway, because the bridge is uncomfortably tight.
Attention is the cognitive domain that took the hit here. Attention is also the domain a lot of ND adults are already managing on a thin battery. If you're an AuDHD or ADHD parent running a household where executive function is the bottleneck, the food you grab to survive the day might be quietly draining the same system you need to run the day.
The honest version goes both ways. It's also plausible that subtle attention struggles push people toward UPF in the first place, because cooking a meal from scratch when your prefrontal cortex is fried is a tall ask. This study can't tell which direction the arrow points (more on that in the Fine Print). Either way, the food category and the attention system are tangled up together, and that's worth knowing.
The Fine Print
The +0.24 risk-score increase per 10% UPF is the largest effect in the paper, and it's the number most likely to end up in a Twitter thread. It comes from a "Modified CAIDE" score the authors built from the modifiable risk factors only. That modified version has not been validated to predict dementia.
The original CAIDE (the actual validated 20-year predictor) showed a weaker effect that just missed statistical significance (p = 0.061) after the researchers controlled for Mediterranean diet. The paper is honest that the modified score is "exploratory." A lot of the coverage of this study won't be.
โ๏ธ Our take: The clean finding is the attention drop. Lead with that. The dementia-score finding is suggestive on a tweaked instrument, and the validated instrument barely missed. Don't quote the +0.24 number without that asterisk.
This is a cross-sectional study. The researchers measured diet and cognition at the same moment. They did not follow people over years to see whose attention got worse. That means the data is consistent with two opposite stories.
Story A: more UPF nudges attention down. Story B: people whose attention is already a little taxed reach for more UPF, because convenience food is what executive function reaches for when it's running low. Both fit the numbers. This is especially worth saying out loud for ND adults, where Story B is very plausible.
The one true causal experiment we have is a small NIH inpatient trial (Hall 2019) where calorie-matched UPF diets still led people to eat about 500 extra calories per day. That's a different endpoint than cognition, but it tells us UPF can change behavior, not just correlate with it.
โ๏ธ Our take: The five large prospective cohorts that follow people over time (UK Biobank, ELSA-Brasil, REGARDS, Framingham, plus meta-analyses) point the same direction. So Story A has support outside this paper. But this paper alone can't tell you which way the arrow runs.
The study found nothing on memory. That matters. It bounds the claim. UPF, in this dataset, is connected to attention, not to the memory loss most people picture when they hear "dementia."
Half the popular coverage of UPF-and-brain studies blurs this distinction. Memory being clean here means the headline "UPF causes Alzheimer's" overshoots what the paper actually found.
โ๏ธ Our take: Cite the null. It's evidence, not absence of evidence.
The Healthy Brain Project sample was 75% female, 83% White or European, and 62% had 15+ years of education. It was also recruited specifically to be enriched for family history of dementia. And it's Australian.
So this is a sample of mostly highly educated, mostly female, mostly White Australians worried enough about their brain health to sign up for a study about it. That's not the average overwhelmed parent in a Houston suburb.
โ๏ธ Our take: Useful evidence, narrow window. Don't extrapolate the exact effect sizes to your own demographic without a grain of salt.
The 2024 Lancet Dementia Commission (Livingston et al.) updated its list of 14 modifiable risk factors for dementia. UPF wasn't on it. The evidence didn't clear the bar yet. Cardoso frames her paper as helping build that case. The disagreement is real, just quieter than the news cycle makes it sound.
โ๏ธ Our take: Smart researchers are still split on whether UPF rises to the level of "actionable dementia-prevention advice." Treat this paper as one more brick in the wall, not a settled case.
The NOVA system (used by this paper to classify UPF) was created by the same Sรฃo Paulo lab one of the co-authors works in. That's not a conflict, but it's a methodological allegiance. NOVA-based UPF definitions also get pushback from researchers who think the category is too broad. Greek yogurt, frozen pizza, and packaged bread can all end up in "UPF" together, and that lumping makes some nutrition scientists wince.
โ๏ธ Our take: The signal direction in this paper agrees with five large prospective cohorts that used various definitions, so the result isn't pure NOVA artifact. But "UPF" as a single bucket is doing a lot of work, and the bucket is debated.
What To Do With This
Swap one ultra-processed item a day for a whole-food version. One. Not all of them. The 10% UPF reduction in this paper is roughly one bag of chips. That's the size of intervention worth aiming for, especially if convenience food is a survival tool in your house. Small, repeatable, low decision cost.
The biggest UPF contributors in this study were sweetened drinks, packaged snacks, ready meals, dairy desserts, and reconstituted processed meats. If you want a target, the swappable wins live there first. A sparkling water instead of a soda. A piece of fruit instead of a packaged dessert. Boring on purpose.
If your attention is already running thin, watch what you reach for. ND adults often eat for executive function relief, not for hunger. That's not shameful. It's worth noticing, because the food category and the attention system are connected somehow, even if this study can't tell us which way.
If logging your own patterns helps (food, focus, sleep, mood), that's the kind of thing Brainloot was built for. Log meals alongside daily symptoms or focus check-ins so you can see your own patterns instead of guessing at them.
Diet quality and food processing aren't the same variable. This paper says the processing matters even after Mediterranean adherence is in the model. That's worth bringing into nutrition counseling for adult cognitive health, especially for patients with neurodivergent profiles where attention is a clinical concern.
Don't run with the dementia headline. The attention finding is the defensible one. The validated CAIDE score didn't survive the diet-quality adjustment. Patients reading this in the news may overshoot what's actually shown.
Reverse causation is clinically relevant. If a patient is reaching heavily for UPF, that may be a downstream signal of cognitive load, not just a risk factor. Asking why before counseling against can land better.
Prospective designs in younger and ND populations are the gap. Cardoso's sample is highly educated, mostly female, family-history-enriched. The clearest test of a UPF-attention causal arrow needs longitudinal designs with broader demographic sampling and ideally subgroups stratified for ADHD/autism status. The reverse-causation question won't resolve in cross-sectional data.
The Modified CAIDE needs validation work or retirement. If it's going to keep getting used as a dementia-risk readout, it needs to be tested against actual dementia outcomes. Otherwise it's a composite of the easier-to-move risk factors masquerading as a prediction tool.
About a bag of chips a day worth of ultra-processed food was tied to a small but real dip in attention scores in 2,192 middle-aged Australian adults, even after diet quality and BMI were accounted for. Memory was untouched. The dementia-risk piece of the headline rests on a tweaked score that hasn't been validated for predicting dementia, and the study can't prove cause from a single snapshot in time.
If you're a parent (especially an ND parent) whose attention is already a managed resource, the quiet read is that the action item is small. Swap one UPF item a day for a whole-food version. That's it. No food shame required.