Does Coding Make You Smarter? What Learning to Program Does to IQ | Riot IQ
Oct 20, 2026·Advanced Topics & Research
Does Coding Make You Smarter? What Learning to Program Does to IQ
Does coding make you smarter? Learning to program improves related skills, but against active control groups the gain on other tests is only about 0.16 SD.
Dr. Russell T. WarneChief Scientist
Share
Learning to code makes you better at coding and at closely related tasks, but there is little evidence that it raises intelligence. The largest meta-analysis found an average benefit of about half a standard deviation on other cognitive tests, yet when the comparison group also did some new activity, the benefit fell to about 0.16 standard deviations, roughly 2 IQ points if the tests had been IQ tests, and most of them were not.
The idea that programming teaches children how to think is about as old as school computers. This article covers what the transfer studies show, why the early Logo results disappointed, what actually predicts who learns to program well, and why the link between coding and intelligence mostly runs the other way.
What the best evidence says about coding and intelligence
The main source on this question is a 2019 meta-analysis by Ronny Scherer, Fazilat Siddiq and Bárbara Sánchez Viveros, who pooled 105 studies and 539 effect sizes on whether learning to program transfers to other cognitive skills.
• The headline numbers: The overall transfer effect was g = 0.49, where "g" here is Hedges' g, a standardized difference between groups measured in standard deviations (and not the general intelligence factor). "Near transfer," to tasks close to programming, was large (g = 0.75). "Far transfer," to unrelated skills, was moderate (g = 0.47).
• Which skills moved: The biggest far-transfer effects were on creative thinking (0.73) and mathematical skills (0.57). Reasoning (0.37) and spatial skills (0.37) were smaller. School achievement (0.28) gained little, and literacy did not change at all (about −0.02).
• The control-group problem: This is the finding that matters most. Studies in which the comparison group did nothing new showed an average effect of 0.65. Studies with "active" control groups, which did some other new activity for the same time, showed only 0.16, and for far transfer alone the estimate was 0.15 with a confidence interval that included zero. Published studies also showed larger effects (0.60) than unpublished ones (0.34).
That control-group gap is the pattern Giovanni Sala and Fernand Gobet found across chess, music and working memory training. When a study compares trained children with children who simply carried on as usual, any new and engaging activity looks good. When the comparison group gets an equally novel activity, most of the far-transfer benefit disappears.
The outcome measures also matter for anyone asking about IQ. The closest category to IQ in the meta-analysis lumped together "intelligence, attention, perception and memory" tests, and its average effect was 0.25, about 4 IQ points if taken at face value, though it rested mostly on comparisons with untreated groups. About 55% of the effect sizes came from studies published in the 1980s, many of them small. No large, preregistered trial has shown that learning to code raises full-scale IQ.
The Logo era: big promises, small results
In the early 1980s, Seymour Papert's Logo language, with its screen "turtle" that children steered with typed commands, was promoted as a way to build general thinking skills. Researchers tested the claim quickly.
• The planning studies: Roy Pea and Midian Kurland ran two year-long studies at a Manhattan school with 8- to 12-year-olds. After a year of Logo, the programming groups planned no better than classmates who had not programmed, first on a classroom chore-scheduling task and then on a computer version of it.
• The critique: In a 1984 review, Pea and Kurland argued that the claim of general mental benefits from programming rested on weak evidence. They argued that any transfer should depend on how far a learner's programming skill develops and on how it is taught, so a few months of beginner programming should not be expected to change general thinking.
• A positive small study: Douglas Clements and Dominic Gullo randomly assigned 18 six-year-olds to 12 weeks of Logo or of computer-assisted instruction. The Logo group did better on measures of reflectivity, divergent thinking and describing directions, but there were no differences on measures of cognitive development.
Studies like the last one, with 18 children split between two groups, are why the overall effect in the meta-analysis looks bigger than the active-control effect. Small samples produce noisy, often inflated estimates, and the striking ones are the ones that get published.
Coding, computational thinking and executive functions
Today the claim is usually made about "computational thinking," Jeannette Wing's term for approaching problems the way a computer scientist does (breaking them into parts, spotting patterns, writing step-by-step procedures). Wing argued that it is a fundamental skill for everyone, and the idea is now a common justification for teaching coding in schools.
The newest evidence focuses on "executive functions," the mental control skills that let a person plan, hold information in mind and stop an unhelpful response. A 2024 meta-analysis by Chiara Montuori and colleagues reviewed 19 experimental studies with 1,523 children aged 4 to 16, and pooled 11 of them. Coding instruction improved problem solving the most (d = 0.89), followed by planning (0.36), with small effects on working memory (0.20) and inhibition (0.17). These are executive-function tasks, not IQ tests, and the problem-solving tasks often look a lot like programming puzzles, which puts much of that effect closer to near transfer.
The practical reading is that coding builds the skills coding uses. That is a good reason to learn it. It is a weaker reason to expect higher scores on a general intelligence test. For a wider look at what does and does not raise scores, see whether you can increase your IQ.
Why programmers tend to score high: selection, not training
The belief that coding makes people smart survives partly because good programmers often are very able. The causal arrow mostly points from ability to coding.
• Ability predicts learning to code: Chantel Prat and colleagues gave 36 adults ten 45-minute Python lessons and measured cognitive skills beforehand. The models explained 50 to 72% of the differences in how well people learned. Fluid reasoning and working memory together explained the most (34%), followed by language aptitude (17%), while numeracy explained only 2%. Programming, at least at a beginner's level, seems to draw on general reasoning and on the skills used in learning a second language more than on arithmetic.
• Computational thinking overlaps with intelligence: Marcos Román-González and colleagues gave a computational thinking test to 1,251 Spanish students in grades 5 to 10 and found that its scores correlated r = .44 with spatial ability, r = .44 with reasoning and r = .67 with problem solving. A correlation of that size means the test measures a good deal of what reasoning tests measure.
People who find reasoning easy are more likely to enjoy programming, to stick with it, and to choose computing courses and careers. Any comparison of programmers with non-programmers therefore mixes the effect of coding with prior ability, years of education and family background. The same selection runs through math, which is covered in the article on whether math makes you smarter. The pattern also resembles the research on video games and IQ: players improve at the games, and the hoped-for spread to general intelligence is small.
Frequently asked questions
Does coding make you smarter?
Coding makes you better at coding and at related problem-solving tasks. Against active control groups, the average benefit on other cognitive tests is only about 0.16 standard deviations, and no large trial has shown a rise in IQ.
Does learning to code improve math skills?
Programming studies showed one of their larger transfer effects on math (g = 0.57). Most of those studies compared coders with children who did nothing new, so the true effect is probably smaller.
Do programmers have high IQs?
People who learn to code well tend to score high on reasoning tests, mainly because reasoning ability predicts who learns programming quickly. In one study, fluid reasoning and working memory explained about a third of the differences in how fast adults learned Python.
Is coding good for children's brains?
Coding lessons appear to improve children's problem solving and planning, with smaller effects on working memory. Those are useful gains, but they are not the same as higher intelligence.
Do you need to be good at math to learn to code?
Less than many people assume. In the Python study, numeracy explained only 2% of the differences in learning, while reasoning and language aptitude explained much more.
The takeaway
Learning to program is worth doing for its own sake, and it does sharpen problem solving that resembles programming. The best evidence does not show that it raises intelligence: transfer effects shrink sharply in studies with fair comparison groups, and the strong link between coding and high scores mostly reflects who chooses to code and who learns it quickly. If you want an accurate measure of the reasoning ability that coding draws on, take a full-length online IQ test.
References
1. Scherer, R., Siddiq, F., & Sánchez Viveros, B. (2019). The cognitive benefits of learning computer programming: A meta-analysis of transfer effects. Journal of Educational Psychology, 111(5), 764-792. [doi.org/10.1037/edu0000314](. doi.org
2. Sala, G., & Gobet, F. (2017). Does far transfer exist? Negative evidence from chess, music, and working memory training. Current Directions in Psychological Science, 26(6), 515-520. [doi.org/10.1177/0963721417712760](. doi.org
3. Pea, R. D., & Kurland, D. M. (1984). Logo programming and the development of planning skills (Technical Report No. 16). Bank Street College of Education, Center for Children and Technology. [eric.ed.gov/?id=ED249930](. eric.ed.gov
4. Pea, R. D., & Kurland, D. M. (1984). On the cognitive effects of learning computer programming. New Ideas in Psychology, 2(2), 137-168. [doi.org/10.1016/0732-118X(84)90018-7](. doi.org
5. Clements, D. H., & Gullo, D. F. (1984). Effects of computer programming on young children's cognition. Journal of Educational Psychology, 76(6), 1051-1058. [doi.org/10.1037/0022-0663.76.6.1051](. doi.org
6. Wing, J. M. (2006). Computational thinking. Communications of the ACM, 49(3), 33-35. [doi.org/10.1145/1118178.1118215](. doi.org
7. Montuori, C., Gambarota, F., Altoé, G., & Arfé, B. (2024). The cognitive effects of computational thinking: A systematic review and meta-analytic study. Computers & Education, 210, 104961. [doi.org/10.1016/j.compedu.2023.104961](. doi.org
8. Prat, C. S., Madhyastha, T. M., Mottarella, M. J., & Kuo, C.-H. (2020). Relating natural language aptitude to individual differences in learning programming languages. Scientific Reports, 10, 3817. [doi.org/10.1038/s41598-020-60661-8](. doi.org
9. Román-González, M., Pérez-González, J.-C., & Jiménez-Fernández, C. (2017). Which cognitive abilities underlie computational thinking? Criterion validity of the Computational Thinking Test. Computers in Human Behavior, 72, 678-691. [doi.org/10.1016/j.chb.2016.08.047](. doi.org
Hero image: Fortran punch card (ESA), by Till Niermann, CC BY-SA 3.0. Via Wikimedia Commons.
Take our professional IQ test
Want to know your IQ? Try the first ever professional online IQ test.
