Can you increase your IQ?

This page separates the interventions with real causal evidence behind them from the very much larger set that has none, and explains the distinction — between raising a score and raising an ability — that makes the whole literature comprehensible once you see it.

The distinction that makes sense of everything else

Almost every argument about raising IQ is really two arguments wearing one coat.

The first question is whether you can improve someone's score on a test. That is straightforwardly yes, and always has been. Practice the format, arrive rested, take it in a quiet room, and the number moves. So does correcting the things that were dragging it down.

The second question is whether you can improve general cognitive ability itself — the underlying capacity that IQ tests are attempting to estimate, usually written as g. That is a much harder claim, and the evidence for it is thinner than the headlines suggest even for the interventions that clearly work on scores.

Keeping the two apart is not pedantry. It is the difference between an intervention worth your time and a subscription worth cancelling.

What the evidence actually supports

Education, by a distance

Schooling is the one intervention with strong causal evidence behind it. In 2018, Stuart Ritchie and Elliot Tucker-Drob published a meta-analysis pooling 142 effect sizes from 42 datasets covering more than 600,000 participants. Crucially, they restricted it to designs capable of separating cause from correlation: studies controlling for earlier intelligence, studies exploiting changes in compulsory schooling laws, and regression-discontinuity designs using school-entry age cutoffs, where children born days apart are assigned to different amounts of schooling by an arbitrary date.

Across all three design types, the answer converged: an additional year of education is associated with a gain of roughly 1 to 5 IQ points.

That is a large effect by the standards of this field. It is also durable in the sense that it did not appear to fade with age in their analysis. Two caveats keep it honest. Most of the underlying evidence concerns schooling during childhood and adolescence rather than education taken up in later life, so the result does not directly license the claim that an evening course at 45 will do the same thing. And the gains show up on tests, which loops us back to the distinction above — researchers still disagree about how much of the effect is broad ability and how much is specific skills that tests happen to measure.

Correcting a deficiency, where one exists

The largest IQ gains ever recorded from a single intervention come from iodine, and they occur only in populations that were deficient in the first place. Iodine is required to produce the thyroid hormones that govern brain development, and its absence during gestation and early childhood does lasting damage.

Three separate meta-analyses have estimated the gap between iodine-deficient and iodine-replete populations at somewhere between roughly 8 and 13.5 IQ points, with intervention studies pooling to around 8 points recovered. That is close to a full standard deviation, and it is why salt iodisation is regarded as one of the highest-return public health measures ever implemented.

The critical qualifier: this is a deficiency correction, not an enhancement. If you are iodine-replete, which most people in countries with iodised salt are, supplementing further does nothing for you. The same logic applies to the other well-evidenced environmental factors — removing lead exposure, adequate early nutrition. They remove a brake. They do not add an engine.

Getting a fair reading in the first place

This one is unglamorous and often the largest single effect available to an individual adult. A tired, stressed, distracted or rushed sitting can cost ten points or more against the same person's performance on a good day. Working memory, which matrix reasoning leans on heavily, is the first thing fatigue takes.

Being unfamiliar with the format costs points too. A person who has never seen a matrix puzzle spends their first few items working out what is being asked rather than answering it. This is a real and well-documented practice effect — and note carefully what it means: it raises your score without raising your ability by a single unit. If you retake a test days later and score higher, that is the most likely explanation, which is why any honest retest should be separated by weeks rather than hours.

What does not work

Commercial brain training

This is the clearest negative result in the field, and it came from the largest test anyone has run. A 2010 study published in Nature recruited 11,430 participants and trained them on cognitive tasks over six weeks. Participants improved substantially at the tasks they practised. Transfer to untrained cognitive abilities, including ones closely related to what they had been drilling, was absent.

Practice makes you better at the thing you practised. That result is robust, unsurprising once stated plainly, and inconvenient for an industry built on the opposite claim.

Working memory and n-back training

Working memory training deserves separate treatment because it had a stronger theoretical case than the rest. Working memory capacity correlates well with fluid intelligence, so training it ought to help. For a while, some studies suggested it did.

It did not survive scrutiny. A 2016 meta-analysis by Melby-Lervåg, Redick and Hulme covered 87 publications and 145 experimental comparisons. Training reliably improved performance on working memory tasks — the thing being trained, and things resembling it. On far transfer to nonverbal ability, verbal ability, reading comprehension and arithmetic, there was no convincing evidence of any reliable improvement when compared against actively treated control groups.

That last clause is where the earlier positive findings went. Comparing trainees against people who did nothing measures expectation and engagement as much as cognition. Comparing them against people who did something equally demanding but different removes that, and removes the effect with it.

Supplements, audio tricks and puzzle books

No supplement has credible evidence for raising IQ in a person who is not deficient in something. The Mozart effect, widely reported in the 1990s, turned out to be a short-lived arousal effect that had little to do with Mozart and nothing to do with intelligence. Crosswords and sudoku make you better at crosswords and sudoku.

A reliable heuristic: if a product promises IQ points, that promise is itself the strongest available evidence that it has not been tested against an active control.

The evidence at a glance

The fadeout problem

Early-childhood interventions complicate the picture in an interesting way. Intensive preschool programmes reliably produce IQ gains in participating children, sometimes substantial ones. Follow the same children for several years and the cognitive gains typically shrink, often to nothing.

What makes this genuinely puzzling rather than simply disappointing is that the same programmes frequently show lasting benefits on outcomes that matter — educational attainment, employment, contact with the criminal justice system — long after the IQ advantage has disappeared. Something durable was produced. It was not the thing the test was measuring.

Whatever the explanation, fadeout is a standing warning against reading too much into short-term score gains from any intervention. A gain measured immediately after training is the weakest form of the claim.

The Flynn effect: the awkward counterexample

Here is the tension that keeps this question alive. Individual interventions mostly fail. And yet, across the twentieth century, average measured IQ rose in country after country by roughly three points per decade — the Flynn effect, named after James Flynn's 1987 analysis of fourteen nations. Test publishers have had to re-norm repeatedly, resetting each generation's average back to 100.

That is an enormous shift. A person scoring at the average in 1950 would look distinctly below average against today's norms. So something in the environment plainly does move measured intelligence at population scale, even though almost nothing we can hand an individual does.

The candidate explanations — more schooling, better nutrition, smaller families, work and leisure that became steadily more abstract — share a shape. They are slow, cumulative and structural, operating across whole childhoods and whole societies. None of them resembles a six-week training programme. Whatever raises intelligence, the Flynn effect suggests it works on the timescale of an upbringing.

Read more on how average IQ changes with age, where the Flynn effect turns up again in a stranger form: the age at which people reach peak performance appears itself to be getting later.

What to do with all of this

If the aim is a higher number on a test, take it rested, take it once properly rather than repeatedly, and if you are still in education, stay in it — that is the only lever with real evidence attached.

If the aim is doing better at work, at study, or at the things a score is standing in for, the honest answer is that the score was never the main constraint. Conscientiousness predicts academic and occupational outcomes about as well as measured ability does, and unlike ability it responds to habit, structure and environment. Study technique — spaced repetition, active recall, deliberate practice, working from worked examples — is the mechanism that converts ability into results, and it is entirely learnable at any level of the scale.

The uncomfortable part of the research is that IQ is harder to change than the self-improvement industry claims. The genuinely encouraging part is that the outcomes people actually want turn out to depend on it less than they fear.

Frequently asked questions