Can You Improve Your IQ Score? What the Evidence Says

"Can I raise my IQ?" is really two questions wearing one coat. Can you raise the number a test reports? Often, yes, and sometimes easily. Can you raise the underlying general reasoning ability the number is trying to estimate? That is much harder, and the honest answer is more cautious than most of the internet suggests.

The two questions

Score and ability come apart constantly. Take the same test twice and your score rises — not because your reasoning improved overnight, but because you had seen the items. Take it after a bad night's sleep and it falls, without your capability having changed at all.

So when evaluating any claim about raising IQ, the question to ask is: did performance improve on untrained tasks the person has never seen? That is the test of genuine transfer, and it is where most interventions fail.

What reliably moves the number

Practice effects — real, but limited

Retaking a cognitive test produces a measurable gain, largest between the first and second attempt, shrinking with each subsequent one. You learn the instructions, the item formats and the strategies. It is a genuine and well-documented effect, which is why clinicians allow substantial intervals between administrations and note prior exposure in reports.

What it is not is an improvement in reasoning. It is improvement at this test.

Familiarity with the format

Someone who has never seen a matrix reasoning item spends part of their first attempts working out what is even being asked. Removing that overhead by trying a few practice items produces a real score gain — and unlike raw retaking, it is legitimate: it strips out a source of measurement error that had nothing to do with your reasoning.

Test conditions

These are undramatic and genuinely worth attending to, because they move scores more than most people assume:

• Sleep. Sleep deprivation impairs attention, working memory and processing speed. Taking a timed reasoning test while short on sleep is one of the easiest ways to depress your own score.
• Time of day. Take it when you are normally alert, not at the end of a long day.
• Interruptions. On a timed test, every interruption costs you twice — the time lost and the cost of reloading the problem into working memory.
• Anxiety. Test anxiety consumes working-memory capacity that would otherwise go to the problem. Treating it as a low-stakes puzzle rather than a verdict measurably helps.
• Not stalling on hard items. With a fixed overall limit, minutes spent on one stubborn item are minutes not spent on several solvable ones. Move on and come back.

What the evidence does not support

Brain training apps

This is the big one, and the evidence is considerably less encouraging than the marketing.

Working-memory training — the n-back task and its commercial descendants — reliably improves performance on the trained task. It also shows some near transfer: improvement on tasks closely resembling the trained one. But far transfer, the thing that would actually matter, is where it breaks down. Meta-analyses of working-memory training have generally found little to no durable improvement in fluid intelligence or in unrelated cognitive tasks.

The pattern that keeps recurring: the better controlled the study — active control groups rather than no-contact controls, outcome measures unlike the trained task, follow-up months later — the smaller the effect gets. Studies with weak controls report the largest gains. That is the signature of an effect driven by expectancy and task-specific learning rather than genuine cognitive improvement.

The commercial claims in this area have also drawn regulatory attention: in 2016 the US Federal Trade Commission settled with the maker of Lumosity over deceptive advertising about real-world cognitive benefits.

You will get better at the game. That is the honest summary.

Supplements and nootropics

Correcting an actual deficiency — iron, iodine, B12 — improves cognition, sometimes markedly. That is treating a deficit, not enhancing a healthy baseline. For people who are not deficient, evidence that supplements raise reasoning ability in healthy adults is weak and inconsistent. Stimulants such as caffeine improve alertness, which can help a timed test, but that is a state effect on the day, not a change in ability.

Listening to classical music

The "Mozart effect" is a durable myth. The original 1993 finding was a small, brief improvement on a specific spatial task, and it did not replicate robustly. The prevailing explanation for what was seen is arousal and mood, not music-induced intelligence. Nothing supports the popular version.

What plausibly helps in the long run

Two things have reasonable support, and neither is a shortcut.

Education. This is the best-evidenced intervention there is. Studies exploiting natural experiments — changes in compulsory schooling laws, for example — find that additional years of schooling are associated with measurable gains in test performance. Whether this reflects a change in general ability or in the specific skills tests sample is debated, but the effect itself is consistently found.

Cardiovascular exercise. Aerobic fitness shows modest associations with cognitive performance, particularly in older adults and for executive function. Effect sizes are moderate and the literature has its own replication problems, so this is a reasonable bet rather than a guarantee. It is also good for you regardless.

Both work slowly, over months and years, and neither will change a score you take next week.

A more useful frame

It is worth asking what raising the number would actually get you. An IQ score is an estimate of a fairly narrow slice of cognition. The things that determine what people accomplish — knowing a domain deeply, working at something past the point of boredom, judgement about which problems are worth attempting, the ability to collaborate — are almost entirely outside what the test samples, and they are far more improvable.

Learning something difficult and specific will do more for your thinking than any amount of training on abstract puzzles. It just will not show up as a bigger number on a matrix test.