Can you increase your IQ? What practice changes, and what it does not

What the practice studies found

The clearest evidence comes from experiments that taught people the rules behind one kind of puzzle and then tested them on it. In two experiments, Schneider, Becker, Krieger, Spinath and Sparfeldt (Intelligence, 2020) randomly assigned 112 psychology students and 229 teacher-education students to a video of under 14 minutes on the rules of figural matrices or to an unrelated video. The group that learned the rules scored higher on matrices with effect sizes of d ≥ 1.19.

An effect size of 1.19 is more than one standard deviation, which is large for a fourteen-minute intervention. As arithmetic only: if that matrix test were scored on the familiar IQ scale, where one standard deviation is 15 points, a gap of 1.19 standard deviations would be about 18 points on that one test. Nobody became 18 points more intelligent in fourteen minutes. They learned what the puzzles reward.

Number series move less. Schneider and Sparfeldt (Intelligence, 2021) randomly assigned 192 teacher-education students to a number series tutorial, an unrelated tutorial, or working through number series items alone. The tutorial group scored higher than both, with effect sizes of d = 0.44 against the unrelated tutorial and d = 0.30 against working through the items alone. That is still a real gain, and a smaller one than for matrices.

Plain repetition helps too, mainly with speed. A 2018 meta-analysis of retesting on mental speed tests by Scharfen, Blum and Holling found response times fell by 0.237 standard deviations from a first test to a second, 0.367 by the third and 0.499 by the fourth. On a timed test, being faster on a familiar format is worth points by itself.

Why a higher score is not more intelligence

The same 2020 study carries a check on that. Everyone in it had taken a separate intelligence test beforehand, and the correlation of matrix scores with that earlier test did not differ between the groups that watched the rules video and the unrelated one (about .55 in the first experiment, .40 in the second). The video moved matrix scores up without changing what they had in common with the other test. The gain belongs to the practised kind of puzzle, and the study did not measure any gain in intelligence.

There is a second reason, and it is about how an IQ number is made. A standardized score compares your raw score with a norm group of people who met the questions fresh. When you have learned the rules first, you are no longer being compared like with like: the same number of correct answers means something different. That is why a result from a test someone has met before is read with care, and why a practised score reads higher than the same person's unpractised one without saying anything new about the person.

Learning how a type of puzzle works is useful in itself, and if a test stands between you and a course or a job, being familiar with its formats removes a handicap that has nothing to do with ability. A practised score reflects that familiarity.

What does change scores more broadly

One influence has good evidence of broader effects, and it works over years rather than weeks: schooling. Ritchie and Tucker-Drob's 2018 meta-analysis in Psychological Science, of 142 effect sizes from 42 datasets with over 600,000 participants, found effects of education on cognitive abilities of approximately 1 to 5 IQ points for an additional year of education. The studies it pooled were chosen because they could separate the effect of extra schooling from the fact that brighter children tend to stay in school longer, for example by using changes in the school-leaving age.

That is a slow, population-level effect of a year of full-time learning across many subjects. It is not evidence that a few weeks of puzzles, an app, or a supplement will do the same, and nothing on this site claims that it will.

Brain-training games, and the 2016 settlement

Brain-training products have promised far more than score gains on their own games, and the evidence has not kept up. A 2016 review in Psychological Science in the Public Interest by Simons and colleagues found that brain-training tasks improve performance on the trained tasks themselves, with less evidence for closely related tasks and very little for distantly related tasks or everyday cognitive performance. That is the same pattern as the practice studies: you get better at what you practise.

Regulators have acted on the bigger promises. On 5 January 2016 the U.S. Federal Trade Commission announced that Lumos Labs would pay $2 million in redress to settle charges that it advertised, without the science to back it, that Lumosity training would improve performance at work, in school and in everyday tasks and delay age-related cognitive decline. Any page that promises to raise your intelligence is making the kind of claim that settlement was about.

How to practise, if you want to

If your goal is a better score on a particular kind of test, the studies above point to a plain method. It is the method the practice questions on this site are built around, and it works the same with a book or a pencil.

• Learn the rules first. Most matrix puzzles combine a small set of rules: shading or counting that steps along a row, one of each shape per row, shapes that turn, and lines that combine or cancel when two tiles are laid together. Number series use differences, ratios, two series taking turns and terms built from earlier terms.
• Use hints in order, not the answer. Try where to look, then the rule, then the worked solution, and only then the answer. Each step you take yourself is the part that transfers to the next puzzle of the same kind.
• Mix the types once each one makes sense, so that the hard part becomes recognising which rule a new puzzle uses.
• Add a clock last. Speed comes with familiarity, and a timed set before you know the rules mostly practises guessing.
• If you want a baseline, take a fixed test once before you practise, and read any later result as a practised score.

What this site will and will not tell you

The practice questions here end with counts: how many you solved on your own, with a hint, or were shown. No practice answer feeds the IQ-style tests on this site. Those tests show an IQ-style practice estimate, and an estimate after practice may be higher for the reasons above.

The answer is more specific than yes or no. You can raise your score on the kinds of puzzle you practise, sometimes by a lot. Years of education appear to raise scores more broadly, by a few points a year. Nobody has shown that puzzles, games or apps raise intelligence itself, and this page will not tell you that they do.