Can You Improve Your IQ? What the Research Actually Shows

"Can I make myself smarter?" is one of the oldest questions in psychology, and after a century of research the answer is partially yes — but the parts of intelligence that are most malleable, the methods that actually work, and the realistic ceiling of improvement are not what the brain-training industry would have you believe. This guide separates the rigorous evidence from the marketing.

If you'd like a baseline read on your current cognitive profile first, our free 41-question IQ test scores you across five cognitive domains.

Two kinds of intelligence

Raymond Cattell's 1963 distinction between fluid and crystallized intelligence is the most useful frame for this whole question. The two have different developmental trajectories, different sensitivities to training, and different ceilings.

Fluid intelligence (Gf) is the ability to reason about novel problems — spotting patterns in unfamiliar sequences, manipulating mental structures, holding several items in working memory while operating on them. It is largely independent of acquired knowledge. Fluid intelligence peaks in the early twenties and declines gradually after.

Crystallized intelligence (Gc) is accumulated knowledge and skill — vocabulary, general information, the verbal categories you have learned to apply. It grows throughout life with reading, education, and engagement, and typically peaks in the sixties or seventies before slow decline.

This split matters because the two are differently trainable. Crystallized intelligence is the most reliably improvable component of IQ across the lifespan. Fluid intelligence is the part the brain-training industry promises to raise — and the part the rigorous evidence says responds least to deliberate training.

How stable is IQ across life?

The headline finding from longitudinal research is striking: IQ measured at age 11 correlates with IQ measured at age 77 at around r = 0.66 (Deary et al., 2004, the Lothian Birth Cohort). Two-thirds of the rank ordering established in primary school survives 66 years.

That stability has to be read carefully:

• Stability is rank order, not level. The correlation describes how well childhood scores predict adult relative position. Both can shift in absolute terms while preserving the ranking.
• Substantial individual movement is normal. A correlation of 0.66 leaves room for many people to move up or down by 10–15 IQ points over their lives. Stability at the population level is consistent with significant change at the individual level.
• Childhood-to-adult correlations rise with age at first measurement. Scores from age 5 predict adult IQ less well than scores from age 11. By late adolescence the rank ordering is fairly entrenched.

The practical takeaway: cognitive ability has a strong autocorrelative spine, but it is not fixed at birth, and the trajectory between childhood and adulthood is shaped by environment in measurable ways — education being the largest single shaper.

Education: the strongest known IQ-raiser

Stuart Ritchie and Elliot Tucker-Drob's 2018 meta-analysis of 28 studies covering more than 600,000 participants is the cleanest evidence available on the causal effect of education on IQ. The headline: each additional year of formal education raises measured IQ by approximately 1–5 points, depending on study design and which IQ component is measured.

The estimate is not just observational. The strongest evidence comes from designs that exploit policy changes — countries raising their compulsory school-leaving age, for instance — which create natural experiments holding constant the kinds of children who do or don't continue. These designs find clean causal effects of about 1–2 IQ points per additional year of education.

What does education improve? Mostly crystallized components — vocabulary, general knowledge, mathematical fluency, the verbal precision that drives much of the IQ-test signal. Fluid components rise less. But because most full-scale IQ scores are weighted averages, even targeted improvements in crystallized capacity move the headline number.

Score gains aren’t always capacity gains

Here is the awkward question buried inside the education-IQ literature: when education raises your IQ score, how much of that is real cognitive capacity and how much is just better familiarity with test formats and verbal frames?

Both effects are present and they’re hard to separate cleanly. Education trains the kinds of abstract-reasoning operations IQ tests sample, so the test does become a fairer measure of capacity for an educated person. But education also teaches the test’s formal vocabulary — "necessary and sufficient", "if and only if", "category", "analogy" — and exposure to those frames lifts scores even when underlying processing speed and working memory haven’t budged. The 1–5 point gain per year of schooling is some unknown blend of the two.

The practical implication: if your goal is doing more demanding cognitive work, chase the things that grow your capacity (reading hard material, learning new domains, sustained focused practice) rather than the things that just polish your test surface. The former produces both real gains and score gains; the latter produces score gains only.

What reliably improves cognitive performance

Beyond formal education, several interventions have replicated effects on either real cognitive performance or test scores.

Reading and substantive engagement

Reading widely is the most reliable everyday IQ-raiser available to adults. Vocabulary growth from reading drives crystallized IQ directly; engagement with complex argumentative text trains the verbal-reasoning machinery; exposure to varied domain knowledge expands the conceptual space the mind has to work with.

Cardiovascular exercise

Aerobic exercise produces small-to-moderate improvements in executive function and processing speed in randomised trials (Smith et al., 2010, meta-analysis of 29 studies). The mechanism is partly direct (BDNF, neurogenesis) and partly indirect (better sleep, lower inflammation, better mood). Effects on full-scale IQ are smaller but real.

Sleep

Sleep loss is one of the most rapid and reversible suppressors of measured cognitive performance. A single night of total sleep deprivation reduces working-memory and reasoning scores by 10–15% on average. Chronic mild deprivation accumulates similar deficits without subjective awareness.

Avoiding cognitive depressors

Heavy alcohol use (Topiwala et al., 2017), chronic anxiety, and untreated attention disorders all reduce measured IQ. Treating these does not raise underlying capacity, but removes a depressor — in some cases by 5–10 measured IQ points.

Doing hard cognitive work in your domain

Domain expertise developed through deliberate practice doesn't raise general intelligence, but it produces transferable thinking habits. The chess player gets better at chess, not at unrelated reasoning, but the discipline of sustained attention and rule-following carries over more than rote training does.

The brain-training mirage

The commercial brain-training industry — Lumosity, CogniFit, BrainHQ, and a long tail of n-back apps — sells the promise that targeted cognitive exercises raise general intelligence. The rigorous evidence does not support this claim.

Three findings, repeated across multiple meta-analyses:

• Near transfer is real. Training a task makes you better at that task and at very similar tasks. This is not surprising and not a useful claim.
• Far transfer is not. Melby-Lervåg and Hulme's 2013 meta-analysis and Simons et al.'s 2016 comprehensive review of brain-training research found no reliable far transfer from training games to fluid intelligence, academic performance, or real-world cognitive function.
• Methodologically clean studies show smaller effects. Studies using active control groups (where the control does some other engaging task) consistently show smaller training effects than studies using passive controls. Many of the original brain-training studies used passive controls and conflated practice effects with capacity changes.

In 2016 the U.S. Federal Trade Commission imposed a $2 million settlement on Lumos Labs, the makers of Lumosity, for "deceptive advertising of its games' ability to forestall memory loss, dementia, and Alzheimer's disease". The FTC's complaint was specific: the company's claims went well beyond what the evidence supported. Most of the brain-training industry sells a similar bill of goods.

The honest summary: if you find brain-training games enjoyable, do them. They will not raise your IQ in any meaningful real-world sense, but they will not harm you either. The opportunity cost of doing them instead of reading a substantive book or having a hard conversation is the more interesting question.

The Flynn effect and what it means

James Flynn's 1984 paper documented something striking: average IQ scores in many countries had risen by approximately 3 points per decade through the 20th century. Children in 1990 were scoring well above the standardisation samples from 1940. The effect was largest on fluid-reasoning subtests — the very components considered least trainable.

What caused it? Probably a combination: improved nutrition (early childhood is sensitive to micronutrients), more universal education, smaller family sizes giving children more parental attention, and a modern environment increasingly demanding abstract-reasoning skills (literacy, hypothetical reasoning, formal categorisation). The exact weighting of each cause is contested.

In the past two decades the picture has become more mixed. Several Western countries (Norway, Denmark, the UK, Finland) have shown reversals or plateaus since the 1990s (Bratsberg & Rogeberg, 2018). The drivers of those reversals are not yet settled — possibilities range from environmental factors (changes in education quality, screen time, attention demands) to demographic shifts (aging populations, immigration patterns).

The takeaway: average IQ is not fixed across generations. It is responsive to large-scale environmental change in measurable ways. That responsiveness is the cleanest evidence we have that intelligence is not biologically immutable — even though, at any given moment, individual rank ordering remains relatively stable.

Realistic expectations

Three frames worth holding in mind.

The ceiling is not zero, and it is not infinite. An adult committing seriously to reading widely, sustaining cognitively demanding work, exercising consistently, and sleeping well can expect measurable improvements in measured IQ over years, particularly on crystallized components. A reasonable upper estimate for a self-directed adult intervention is on the order of 5–10 points on a full-scale IQ measure over a multi-year period — modest but real. Claims of larger effects, particularly from short interventions, are usually a combination of practice effects and regression to the mean.

The biggest gains come from things you'd want to do anyway. Reading more, sleeping better, exercising regularly, drinking less, learning new domains: these are good for almost every health and cognitive outcome. The IQ benefit is one of many returns on the same investment.

The score is not the goal. An IQ score is a measurement, not an outcome. Raising it for its own sake is a strange ambition. Doing the things that raise it — reading, learning, sustained difficult work — produces a person who is more interesting to talk to, more capable, and more useful to others, regardless of what number falls out at the end.

A practical plan

1. Read substantively, daily, for years. Forty pages of demanding non-fiction or literary fiction per day, sustained over years, reliably grows vocabulary, conceptual range, and verbal reasoning — the most trainable components of IQ.
2. Sleep seven to nine hours. Most adults are mildly sleep-deprived; correcting this raises measured cognitive performance more than any supplement or app.
3. Get moderate aerobic exercise three to five times per week. The cognitive benefits replicate; the metabolic and mood benefits make consistency easier.
4. Pick a hard domain and stay with it. Mathematics, programming, music, a foreign language, a craft. Sustained engagement in a difficult domain trains the underlying machinery of focus, working memory, and abstract reasoning more reliably than rotating through brain-training apps.
5. Avoid heavy alcohol use. The dose-response relationship between alcohol and cognitive performance is steep above moderate drinking. Removing this depressor is one of the largest single moves available.
6. Treat untreated attention or mood disorders. ADHD, depression, and chronic anxiety all measurably suppress IQ-test performance. Treating them does not raise underlying capacity, but unmasks it.
7. Don't bother with brain-training apps. They train themselves. Spend the time on the items above.

Frequently asked questions

Related reading

• What Your IQ Score Actually Means — how the standard IQ scale works, and what your number does and doesn't predict.
• IQ Classification Ranges — the standard score bands and what each predicts.
• IQ vs EQ — how cognitive intelligence and emotional intelligence differ.
• Working Memory Test — the contested case of cognitive training and far transfer.
• IQ Test Sample Questions — six worked examples, one per cognitive domain.

References

1. Cattell, R. B. (1963). Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational Psychology, 54(1), 1–22.
2. Deary, I. J., Whiteman, M. C., Starr, J. M., Whalley, L. J., & Fox, H. C. (2004). The impact of childhood intelligence on later life: Following up the Scottish Mental Surveys of 1932 and 1947. Journal of Personality and Social Psychology, 86(1), 130–147.
3. Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science, 29(8), 1358–1369.
4. Smith, P. J., et al. (2010). Aerobic exercise and neurocognitive performance: A meta-analytic review of randomized controlled trials. Psychosomatic Medicine, 72(3), 239–252.
5. Topiwala, A., et al. (2017). Moderate alcohol consumption as risk factor for adverse brain outcomes and cognitive decline. BMJ, 357, j2353.
6. Melby-Lervåg, M., & Hulme, C. (2013). Is working memory training effective? A meta-analytic review. Developmental Psychology, 49(2), 270–291.
7. Simons, D. J., et al. (2016). Do "brain-training" programs work? Psychological Science in the Public Interest, 17(3), 103–186.
8. Flynn, J. R. (1984). The mean IQ of Americans: Massive gains 1932 to 1978. Psychological Bulletin, 95(1), 29–51.
9. Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. Proceedings of the National Academy of Sciences, 115(26), 6674–6678.
10. Federal Trade Commission. (2016). Lumosity to pay $2 million to settle FTC deceptive advertising charges for its "brain training" program. FTC press release, January 5, 2016.
11. Heckman, J. J. (2006). Skill formation and the economics of investing in disadvantaged children. Science, 312(5782), 1900–1902.