Can you increase your IQ? What the research shows

Short answer: Yes, a little, and mostly the slow way. The strongest evidence is for education: a 2018 meta-analysis of over 600,000 people found about 1 to 5 IQ points for each extra year of schooling (Ritchie and Tucker-Drob, Psychological Science). Brain-training games make you better at the games, but controlled studies find no reliable gain in general ability. Retaking a test raises the score on average, and researchers still debate how much of that gain reflects real ability, so a higher retest score cannot be read as a rise in IQ. And every score carries error, so a change of a few points can be noise.

What research has found, method by method

Each line gives the method, what the best studies found, and the source with its year.

• More years of school: about 1 to 5 IQ points per extra year, across 142 effect sizes from 42 data sets (Ritchie and Tucker-Drob, 2018). The effects persisted across the life span.
• Working memory training: short-term gains on working memory measures, but "no convincing evidence" of gains in verbal or nonverbal ability when compared with a treated control group, across 87 publications (Melby-Lervåg, Redick and Hulme, 2016).
• Taking a test again, same or alternate form: an average gain of .26 standard deviations across 134,436 people, larger when the identical form was used (Hausknecht and colleagues, 2007). By our calculation that is about 3.9 points on the IQ scale. The causes of such gains "are not yet fully understood" (Jendryczko, Scharfen and Holling, 2019).
• Exercise in children: a mean gain of 4.0 IQ points across 14 randomized trials with 3,203 children aged 5 to 14 (Morales and colleagues, Pediatrics, 2024). Across all ages in healthy people, the benefit to thinking skills was negligible after correcting for publication bias (Ciria and colleagues, 2023).
• Adoption into a new home: in a Swedish study, adopted brothers scored 4.41 points higher than their brothers raised at home (Kendler and colleagues, PNAS, 2015).
• Early childhood programs: gains faded after the programs ended, across 44 randomized trials with 7,584 young children (Protzko, 2015, as reported by UC Santa Barbara).
• The Mozart effect: the original claim was about spatial tasks, and a 2010 meta-analysis by Pietschnig and colleagues traced it to low study power and bias, as a 2023 paper describes (Oberleiter and Pietschnig, Scientific Reports).

What "increasing your IQ" means

An IQ score is a rank. It does not count what you know. On the Wechsler tests, scores on each part "are standardized against age-specific norms", and the IQ scale has a standard deviation of 15 (Ramsden and colleagues, Nature, 2011). The average is set at 100, as our guide to the average IQ explains. So your score says where you stand against people your own age in the test's norm group.

That changes what "raising your IQ" can mean. Your score goes up only when you move ahead of your peers. Simply learning more is not enough, because they learn too. And when a whole population improves, the test's 100 has to be reset. Trahan and colleagues describe the rise in scores over time as one "which results in norms obsolescence" (Psychological Bulletin, 2014).

A score also carries error. No single sitting pins your ability to one number, which is why careful reports give a range. Our guide on whether IQ tests are accurate explains how big that range is. Keep it in mind for every number below: a gain smaller than the range may not be a gain at all.

Education: the most consistent way scores rise

The best evidence that IQ can rise comes from schooling. Ritchie and Tucker-Drob pooled 142 effect sizes from 42 data sets with over 600,000 participants and found gains of "approximately 1 to 5 IQ points for an additional year of education" (Psychological Science, 2018). A 2020 paper by Stankov and Lee, summarizing that meta-analysis, gives the average as 3.4 points per year (Journal of Intelligence).

These were not simple comparisons of people with more or less schooling, which could reflect who stays in school. The meta-analysis kept only stronger designs: studies that controlled for earlier intelligence, studies of changes in compulsory schooling laws, and studies of children just either side of a school-entry age cutoff (Ritchie and Tucker-Drob, 2018). The authors found the effects "persisted across the life span" and appeared on all the broad abilities studied. Their conclusion: "Education appears to be the most consistent, robust, and durable method yet to be identified for raising intelligence."

One natural experiment shows how this works. In the 1960s Norway raised compulsory schooling from 7 to 9 years, and Brinch and Galloway found the reform had "a substantial effect on IQ scores measured at the age of 19 y" (PNAS, published December 2011). Galloway told HealthDay that this came to an increase of 3.7 IQ points for each extra year of school (HealthDay, December 2011). The data came from military draft tests, so it covers men only.

The effect may shrink with age. A Danish study found 4.3 IQ points per year of education in young adulthood and 1.3 points in mid-life (Hegelund and colleagues, 2020, as reported by Stankov and Lee, 2020). It also found that people with lower childhood scores gained the most. These studies measured years of formal schooling, so they say little about whether one short course in adulthood would move a score.

Brain training and working memory games

Brain training is the claim most people meet first, and it has the weakest support once studies use proper controls.

The idea took off in 2008. Jaeggi and colleagues reported that participants who practiced a hard working memory game called dual n-back improved on fluid intelligence tests, and that "the training effect is dosage-dependent" (PNAS, 2008). A 2015 meta-analysis by Jaeggi's group pooled 20 n-back studies of healthy adults aged 18 to 50 and found "a small but significant positive effect of n-back training" on fluid intelligence (Au and colleagues, Psychonomic Bulletin and Review).

Larger and stricter studies did not hold this up:

• Melby-Lervåg and Hulme reviewed 23 studies and found training "produced reliable short-term improvements in working memory skills", but no convincing evidence that the gains spread to other skills (Developmental Psychology, 2013).
• Their 2016 follow-up, with Redick, covered 87 publications and 145 comparisons. When training was compared with a treated control group, there was "no convincing evidence of any reliable improvements" in verbal or nonverbal ability (Perspectives on Psychological Science, 2016).
• Redick and colleagues gave young adults 20 sessions of dual n-back against a placebo task. Despite "a high level of statistical power, there was no positive transfer to any of the cognitive ability tests" (Journal of Experimental Psychology: General, 2013).
• A 2019 meta-analysis of meta-analyses found that once placebo effects and publication bias were controlled, the overall effect "equaled zero" (Sala and colleagues, Collabra: Psychology).

Commercial apps fare no better. Simons and colleagues read every peer-reviewed study that leading brain-training companies cited and found "little evidence that training enhances performance on distantly related tasks or that training improves everyday cognitive performance" (Psychological Science in the Public Interest, 2016). A study of over 1000 people who had used brain-training programs for up to 5 years found that time spent training "showed no relationship with any cognitive performance measure" (Stojanoski and colleagues, Journal of Experimental Psychology: General, 2020).

Regulators reached a similar view. In January 2016 the FTC settled with Lumos Labs, the company behind Lumosity, which agreed to pay $2 million (FTC press release of January 5, 2016, as reprinted by Training Industry). The FTC's Jessica Rich said "Lumosity simply did not have the science to back up its ads".

There is one notable counter-case. In former Yugoslavia, Kvashchev ran a creative problem-solving program for over three years with close to 300 students and random assignment of school classes. Stankov's 1986 report on the study found an average gain of seven points across 28 intelligence tests. A 2020 re-analysis by Stankov and Lee pooled the standard deviations and put the trained group 10 points ahead of the control group (Journal of Intelligence). It is one historical study, reanalyzed by the researcher who reported on it in 1986, and it looks more like years of intensive schooling than a few minutes of games a day.

Why a retake score goes up

Taking a test again usually raises your score. Researchers call it the retest effect: better performance on the same or an equally hard version of a test the second time (Jendryczko, Scharfen and Holling, Journal of Intelligence, 2019).

The size is well measured. A 2007 meta-analysis of 107 samples and 134,436 people found an adjusted average gain of .26 standard deviations, larger with coaching and with the identical test form (Hausknecht and colleagues, Journal of Applied Psychology). On the IQ scale, with its standard deviation of 15 (Ramsden and colleagues, 2011), that works out to about 3.9 points by our calculation. A 2018 meta-analysis of 153,185 people found significant retest gains and "no further score gains after the third test administration" (Scharfen, Peters and Holling, Intelligence).

Why the gain happens is less settled. Jendryczko, Scharfen and Holling describe three explanations: a real gain in the ability the test measures, test-specific strategies often called test-wiseness, and the fading of things such as test anxiety and unfamiliarity that held back the first score (Journal of Intelligence, 2019). In their own study, a matrix test was given seven times. A substantial reduction in how much anxiety interfered with performance between the first two sittings went with the largest retest gain. A smaller reduction went with a smaller gain at the third sitting. Scores improved up to the fourth sitting and then leveled off. The reduced interference did not explain that last gain. So the authors conclude that a reduction in anxiety-induced measurement bias cannot completely explain retest effects. They write that the causes "are not yet fully understood", and Hausknecht and colleagues also call for more research on what retest scores mean (2007).

So a higher second score cannot be read as a rise in ability. Because a retest score reads high on average, it is not directly comparable to a first score. When you compare results from two tests, compare a first sitting with a first sitting.

Does IQ change with age?

For most people, rank holds fairly steady across a lifetime, but individual scores can move, especially in the teenage years.

Ramsden and colleagues tested 33 adolescents at an average age of 14.1 and again at 17.7 (Nature, 2011). Changes in full scale IQ ranged from minus 18 to plus 21 points. Even allowing for measurement error with 90% confidence intervals, 33% of the group showed a clear change in full scale IQ. The changes matched changes in gray matter: "verbal IQ changed with grey matter in an area that was activated by speech". Yet the group as a whole stayed in much the same order, with a correlation of 0.79 between the two sittings.

Over a whole life, the link between early and late scores is strong but far from perfect. In the Lothian Birth Cohort 1921, scores on the same test at age 11 and age 90 correlated at .54, or .67 when corrected for range restriction (Deary, Pattie and Starr, Psychological Science, 2013). Across two Lothian cohorts, the correlations from age 11 to ages 70, 79 and 87 were .67, .66 and .51 (Gow and colleagues, Psychology and Aging, 2011). In plain terms, a high scorer at 11 tends to score high at 80, with plenty of people moving up or down.

Different abilities also change at different ages. A study of 48,537 online participants put it this way: "Some abilities peak and begin to decline around high school graduation", while "others do not peak until subjects reach their 40s or later" (Hartshorne and Germine, Psychological Science, 2015). Our guide to fluid intelligence covers which abilities tend to fall first. Because IQ is scored against people your own age, slowing down with age at the same pace as your peers does not lower your IQ score.

Exercise

Exercise has a promising result in children and a weak one in adults.

A 2024 meta-analysis of 14 randomized trials with 3,203 children aged 5 to 14 found exercise programs improved general intelligence, a mean gain of 4.0 IQ points with a 95% confidence interval of 1.44 to 6.64 (Morales and colleagues, Pediatrics). The authors word it carefully: exercise is "associated with improvements in intelligence", and the programs and children studied varied widely.

For healthy people of all ages, an umbrella review of 24 meta-analyses found a small benefit to thinking skills, d = 0.22, that became "negligible after correcting for publication bias" at d = 0.05 (Ciria and colleagues, Nature Human Behaviour, 2023). That review measured general cognition rather than IQ scores. The evidence that exercise raises an adult's IQ is thin.

If IQ is heritable, is it fixed?

No. Heritability describes how much of the difference between people in a group is linked to genes. It says nothing about whether one person's score can change.

Plomin and Deary put the heritability of intelligence at about 20% in infancy, 40% in adolescence and 60% in adulthood, and perhaps 80% in later adulthood (Molecular Psychiatry, 2015). Yet a 2017 review argues that "Intelligence can have an extremely high heritability, but also be malleable", pointing to gains after adoption and the rise in scores across generations (Sauce and Matzel, Psychological Bulletin).

Three kinds of evidence back this up:

• Adoption. In France, Duyme and colleagues followed 65 children who were neglected or abused as infants, adopted between ages 4 and 6, and scored under 86 before adoption. By about age 13.5 they had gained 7.7 points on average in adoptive families of lower socioeconomic status (SES) and 19.5 points in high-SES families (PNAS, 1999). In Sweden, adopted brothers scored 4.41 points higher than their brothers raised at home (Kendler and colleagues, PNAS, 2015).
• Generations. Across 285 studies since 1951, IQ scores rose 2.31 points per decade on average, and 2.93 points on modern Wechsler and Stanford-Binet tests (Trahan and colleagues, Psychological Bulletin, 2014). Our guide to the average IQ explains this Flynn effect.
• Reversal within families. In Norway the rise later turned into a decline. Bratsberg and Rogeberg found that the rise, its turning point and the decline "can all be fully recovered from within-family variation", which points to the environment rather than genes (PNAS, 2018).

Gains can also fade. Protzko's meta-analysis of 44 randomized trials with 7,584 young children found that children in early intervention programs "lost their IQ gains" after the programs ended (UC Santa Barbara news release, December 2015). He added: "This paper doesn’t argue that intelligence cannot be raised". The studies covered young children only.

What does not have good evidence

• The Mozart effect. The claim began with better spatial task scores after hearing a Mozart sonata. A 2023 paper says that result was "ultimately demonstrated to be a consequence of low study power and bias-related measurement artifacts" (Oberleiter and Pietschnig, Scientific Reports).
• Short courses of brain games. The controlled studies above found no reliable gain in general ability, and the FTC's 2016 Lumosity settlement turned on claims that were not backed by science.
• Big, fast jumps. None of the controlled studies above reported large gains from a few weeks of games or exercises. The largest gains came from years of school or a change of home in childhood.

How to read a change in your own score

If you have two scores and want to know whether you improved, these checks help:

• Compare like with like. Two different tests can put the same person at different numbers, so compare scores from the same kind of test.
• Count the retest effect. A second sitting reads high by a few points on average, and how much of that is real change is still debated, so a small rise on a retake does not show that your ability rose.
• Read the range as well as the point. If the two ranges overlap, the scores may not differ at all.
• Ask a psychologist about a clinical result. Whether a change on a clinical test such as the WAIS is meaningful is a question for the psychologist who gave it.

WhatIQ's test has 24 timed picture puzzles and takes about 15 minutes (whatiq.com, 17 October 2026). It gives a score on the standard IQ scale with the range it most likely sits in, and the site says the result is an estimate, not a clinical assessment.

Frequently asked questions

Can IQ be increased after 18?

Some studies say yes, by a little. A Danish study found 4.3 IQ points per year of education in young adulthood and 1.3 points in mid-life (Hegelund and colleagues, 2020, as reported by Stankov and Lee, 2020). That study counted total years of education, much of it before 18, so it does not show what one course would do for an adult. Brain-training games have not shown reliable gains in adults when compared with a proper control group.

Is IQ genetic or learned?

Both. Plomin and Deary put heritability at about 40% in adolescence and 60% in adulthood (Molecular Psychiatry, 2015). Heritability describes differences within a group and sets no limit for one person, and school, adoption and the rise in scores across generations all show that the environment moves IQ scores too.

Can you increase your IQ by 20 points?

Not through anything a reader could choose to do, as far as controlled studies show. Gains near 20 points appeared in a small study in France of neglected children adopted between ages 4 and 6 into families of higher socioeconomic status (Duyme and colleagues, PNAS, 1999). In one study of 33 teenagers, verbal IQ changed by between minus 20 and plus 23 points over about three and a half years (Ramsden and colleagues, Nature, 2011).

Do brain-training apps raise IQ?

Not as far as the strongest studies show. A 2016 review of the studies cited by brain-training companies found little evidence of gains beyond the trained tasks (Simons and colleagues). In 2016 the FTC settled with the maker of Lumosity, which agreed to pay $2 million over its advertising claims.

Does IQ go down with age?

Your IQ score is compared with people your own age, so it does not drop just because you get older. Some abilities do decline earlier than others: one study of 48,537 people found some peak around high school graduation and others in the 40s or later (Hartshorne and Germine, Psychological Science, 2015).

Does retaking an IQ test raise your IQ?

It raises the score on average, but that does not show a rise in ability. A 2007 meta-analysis of 134,436 people found an average retest gain of .26 standard deviations, about 3.9 IQ points by our calculation (Hausknecht and colleagues). Gains stop after about the third sitting (Scharfen, Peters and Holling, 2018). Jendryczko, Scharfen and Holling (2019) write that the causes of retest effects "are not yet fully understood", so a retake score is not directly comparable to a first score.