A dose-response relationship — more of a thing producing more of an effect — is one of the stronger signals available in nutrition science, because it's much harder to explain away than a single positive trial. For omega-3 and cognitive function, a 2025 meta-analysis went looking for exactly that. It found one for several domains, no relationship for others, and one domain where the curve initially went the wrong way.
The evidence base
A systematic review and dose-response meta-analysis in Scientific Reports pooled 58 randomized controlled trials, with daily omega-3 doses varying from 230 to 4,950 mg/day and intervention durations from 4 to 160 weeks.1 Nine of the included studies used multi-component interventions combining omega-3 with vitamins or antioxidants, which is a real limitation on attributing effects to omega-3 alone.1
What the dose-response curve showed — domain by domain
The model estimated the change in cognitive function per 2,000 mg/day increment in omega-3. Reporting this properly means naming the effect sizes and the GRADE rating, because "improvements across several domains" is doing a lot of work otherwise:
- Global cognitive abilities: SMD 1.08 (95% CI 0.73–1.44), GRADE low
- Attention: SMD 0.98 (0.41–1.54), GRADE low
- Language: SMD 0.98 (0.41–1.54), GRADE low
- Primary memory: SMD 0.87 (0.17–1.56), GRADE moderate
- Visuospatial function: SMD 0.86 (0.46–1.27), GRADE moderate
- Perceptual speed: SMD 0.50 (0.05–0.95), GRADE moderate
That is a genuine dose-response finding across dozens of independently run trials, which is exactly the pattern that makes a nutrition claim harder to dismiss as noise. It is also graded low-to-moderate certainty, which means real signal rather than proof at the highest evidentiary bar.
Why omega-3 is thought to affect cognition at all
A dose-response curve tells you a pattern exists. It doesn't by itself explain why. The mechanistic story for omega-3 and the brain runs through two related but distinct fatty acids — EPA (eicosapentaenoic acid) and DHA (docosahexaenoic acid) — and they are not interchangeable.
DHA is structurally built into neuronal membranes. It's one of the most abundant fatty acids in the phospholipid bilayer of brain cell membranes, particularly at synapses, where it's thought to increase membrane fluidity — the physical flexibility that lets membrane-embedded receptors and ion channels function properly, and that supports the vesicle trafficking involved in forming new synaptic connections.2 Low brain DHA status has been associated with reduced synaptic density and altered neurotransmission in animal models, which is part of why DHA in particular is framed as a structural, rather than purely signaling, nutrient for the nervous system.2
EPA, by contrast, is metabolized into a different set of signaling molecules — eicosanoids and specialized pro-resolving mediators that shape the inflammatory response.3 Chronic low-grade neuroinflammation is one of the more consistent findings in cognitive aging research, and EPA's anti-inflammatory signaling is the proposed route by which it could support cognitive function even though it makes up a smaller share of brain membrane fatty acids than DHA does. That mechanism is also one of the most over-marketed in the category, and we've separated the real biology from the sales pitch in inflammation and the brain.
Put simply: DHA is more about the hardware, EPA is more about calming the environment the hardware sits in. A meta-analysis measuring "omega-3" as one pooled variable, as the 2025 dose-response study does, is necessarily blending both mechanisms together.
DHA is built into the membrane. EPA changes the signaling environment around it. "Omega-3" is really two ingredients wearing one name.
Does the EPA:DHA ratio matter for cognition specifically?
This is a genuinely well-studied distinction in omega-3 research generally, and the honest answer for cognition is: it's not settled, and the two outcome areas where omega-3 is best studied — heart health and brain health — don't point the same direction. We've written up the cardiovascular side, where two major reviews reached opposite conclusions, in two major reviews, two different conclusions.
The clearest head-to-head test for cognition came from a 2021 randomized controlled trial in the American Journal of Clinical Nutrition. 310 healthy adults aged 25–49 took either an EPA-rich oil (900mg EPA + 360mg DHA/day), a DHA-rich oil (900mg DHA + 270mg EPA/day), or 3,000mg/day refined olive oil as placebo, for 26 weeks. Both global accuracy (p = 0.044) and global speed (p = 0.003) improved with the EPA-rich oil compared with placebo, and memory accuracy improved with EPA compared with DHA (p = 0.034) — the DHA-rich oil did not reach those results.4 Worth noting the honest limits: the prefrontal-oxygenation findings the authors also report were trends rather than significant (p = 0.060 and p = 0.082), the participants were young and healthy rather than cognitively ageing, and the paper pools more than one registered trial. Still, it cuts against the common assumption that DHA is automatically the "brain" fatty acid and EPA the "heart" one.
The 2025 dose-response meta-analysis doesn't resolve this either way — it pooled trials using DHA alone, EPA alone, and combined formulations together under a single "omega-3" dose variable, so it can't tell us whether a given ratio outperforms another.1 The fair summary is that both fatty acids show real, independently plausible mechanisms and some independently positive trials, and the field hasn't produced a confident answer on optimal ratio for cognitive outcomes specifically. We're not going to claim otherwise.
Who the effect looks strongest for
Dose matters, but so does who's taking it. Two trials outside the 2025 meta-analysis illustrate the range. In the MIDAS trial, 485 adults aged 55 and older with age-related cognitive decline and subjective memory complaints took 900mg/day of algal DHA for 24 weeks and showed measurable improvement in memory and learning compared with placebo.5 At the other end, the AREDS2 cognitive-function study followed 3,741 older adults (mean age 72.7) at risk of late age-related macular degeneration over five years, testing 1g/day of long-chain omega-3 (650mg EPA plus 350mg DHA) in a factorial design. It found no effect whatsoever: yearly change in composite cognitive score was −0.19 with omega-3 versus −0.18 without, a difference of −0.03 (99% CI −0.20 to 0.13), p = 0.63.6
Those two results aren't necessarily a contradiction. MIDAS enrolled people with documented age-related cognitive decline, used a higher DHA dose, and measured memory improvement. AREDS2 enrolled people without diagnosed cognitive impairment, dosed lower on DHA, and asked a harder question — can omega-3 prevent decline that hasn't started — over a much longer timeframe. That AREDS2 null is the single most important result on this page for anyone hoping a fish oil capsule will protect their future memory: in the largest and longest trial of the question, it didn't. Nothing here supports a claim that omega-3 treats or prevents dementia — it doesn't, and no trial here says it does. For the factors that actually carry population-level evidence, see modifiable dementia risk.
The caveats that matter
Three, and they belong together. Certainty of evidence was graded low to moderate — real signal, not proof. The non-linear analyses are messier than a clean upward curve: global cognitive ability rose with dose and then fell (P for non-linearity = 0.008), while episodic memory fell with rising dose before curving back up (P for non-linearity = 0.01), so "more is better" does not hold indefinitely or uniformly. And in the non-linear modelling, several domains' confidence intervals widen enormously at higher doses — attention's estimate at 5,000 mg/day spans −10.4 to 7.23 — which is a way of saying the data at the top of the dose range is thin enough that the model can't say much.1
What the researchers themselves concluded
The authors' own language was that omega-3 "may lead to a modest improvement in cognitive function among adults," with an explicit call for "well-designed randomized trials with long-term follow-up" to confirm it. That's a considered, moderate conclusion, and it's the one we're working from.
What a realistic dose actually looks like — including ours
The 2,000 mg/day increment used in the dose-response finding is a research unit, not a serving size — it's the step size researchers used to model the curve. Most commercial fish oil capsules deliver far less than that per serving, and it's worth checking the actual EPA+DHA figure on the Supplement Facts panel rather than assuming "1,000mg fish oil" means 1,000mg of omega-3, because it usually doesn't. The method for reading that panel properly is in how to read a supplement label.
Applying that to our own product rather than pointing you elsewhere: Proco's Omega-3 provides 1,000mg of fish oil per softgel, delivering 180mg EPA and 120mg DHA — 300mg of combined EPA+DHA. That is near the floor of the 230–4,950 mg range the meta-analysis covers, and roughly a seventh of the 2,000 mg increment its effect sizes are expressed per. Anyone wanting to approach the doses in the trials above would be taking several softgels a day or choosing a higher-concentration oil, and that's a conversation worth having with a clinician rather than an upsell from us. The wider dosage picture is in the omega-3 dosage guide.
What we take from this
Omega-3 has the deepest evidence base of the ingredients we carry for cognitive health, and this dose-response data is part of why. But depth is not the same as strength: the certainty grading is low to moderate, the curve is non-monotonic for two domains, and the largest long-term prevention trial was flatly negative. We describe the effect the way the meta-analysis itself does — a modest, real improvement in some domains — and we tell you plainly that a single 300mg EPA+DHA softgel sits at the bottom of the studied range rather than in the middle of it. That's a less appealing sentence than the one we used to have, and it's the accurate one.
This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease, including cognitive decline or dementia. Speak with a healthcare provider before starting any new supplement, especially if you take medication or have an existing health condition.
References
- Systematic review and dose-response meta-analysis of omega-3 supplementation on cognitive function. Scientific Reports, 2025. pmc.ncbi.nlm.nih.gov/articles/PMC12368174
- Bazinet RP, Layé S (2014). Polyunsaturated fatty acids and their metabolites in brain function and disease. Nature Reviews Neuroscience, 15(12), 771–785. nature.com/articles/nrn3820
- Calder PC (2010). Omega-3 fatty acids and inflammatory processes. Nutrients, 2(3), 355–374. mdpi.com/2072-6643/2/3/355
- Patan MJ, Kennedy DO, Husberg C, et al. (2021). Supplementation with oil rich in eicosapentaenoic acid, but not in docosahexaenoic acid, improves global cognitive function in healthy, young adults: results from randomized controlled trials. American Journal of Clinical Nutrition, 114(3), 914–924. pmc.ncbi.nlm.nih.gov/articles/PMC8408864
- Yurko-Mauro K, McCarthy D, Rom D, et al. (2010). Beneficial effects of docosahexaenoic acid on cognition in age-related cognitive decline. Alzheimer's & Dementia, 6(6), 456–464. alz-journals.onlinelibrary.wiley.com/doi/10.1016/j.jalz.2010.01.013
- Chew EY, Clemons TE, Agrón E, et al., Age-Related Eye Disease Study 2 Research Group (2015). Effect of omega-3 fatty acids, lutein/zeaxanthin, or other nutrient supplementation on cognitive function: the AREDS2 randomized clinical trial. JAMA, 314(8), 791–801. pmc.ncbi.nlm.nih.gov/articles/PMC5369607
Related reading
- Omega-3 and Cognitive Function: EPA, DHA, and What Dose Actually Matters
- What the Research Actually Says About Nutrition and Attention
Want to know what's actually in what you take? Join the list — no spam, just it straight.