IQ and Musical Ability: Overlap, Not Destiny

0 The Short Answer

Musicians do score slightly higher on cognitive tests than non musicians, and the best evidence says music training is not the reason. That sentence contains the entire argument of this page, and it is worth unpacking slowly because almost everything written about music and intelligence gets it wrong in one of two opposite directions. One camp promises that piano lessons will raise a child's IQ. The other dismisses the association entirely. Both are wrong, and the truth in between is more interesting than either. A third position skips the argument entirely by promoting music to an intelligence in its own right, but that idea comes from Gardner's 1983 framework, which was reasoned out from criteria such as whether an ability can be isolated by brain damage rather than from data on how abilities actually relate across thousands of people.

The association is real and reasonably well documented. People with musical training tend to perform a bit better on measures of general cognitive ability, and the gap tends to grow with years of training. The problem is that association is exactly what you would expect even if music lessons did nothing at all, because the families who provide music lessons differ systematically from the families who do not, in income, in education, in the value they place on effortful practice, and in the genes they pass on.

When researchers have designed studies capable of separating those explanations, the causal effect of training on general ability has come out far smaller than the raw association, and in the strongest designs it has often disappeared. Meanwhile, the near effects, what training does to musical skill and to auditory discrimination, are large and undisputed. The rest of this page explains how the field got here, what each type of study can and cannot establish, and what any of it means if you are deciding whether to sign a child up for lessons.

1 What the Correlation Actually Looks Like

Start with the raw finding, because it is genuine. Across many samples, children and adults with music training score modestly higher on cognitive measures than those without. Effects cluster in the range of a few IQ points, and studies frequently report a dose relationship: more years of lessons, slightly higher scores.

A dose relationship feels like strong evidence for causation, and this is the single most common reasoning error in the entire literature. Consider what else scales with years of music lessons. Family income scales with it, because lessons cost money and instruments cost more. Parental education scales with it. Household stability scales with it, since sustaining weekly lessons for a decade requires a household that can sustain anything for a decade. The child's own persistence and conscientiousness scale with it, because children who dislike practice quit, and quitting is the single largest determinant of how many years of training someone ends up with.

Every one of those variables independently predicts cognitive test performance. So a dose response between lessons and scores is precisely what a purely non causal world would produce. To get past that, you need a design that breaks the link between training and the family that provides it.

2 The Three Designs and What Each Can Prove

Research on this question comes in three flavors, and they differ enormously in what they can establish.

• Cross sectional comparisons. Recruit musicians and non musicians, test both, compare. Cheap, common, and incapable of separating training from the traits that led to training. These studies produce the largest reported effects and the weakest evidence, a pattern that should itself make you suspicious.
• Randomized controlled trials. Assign children at random to music lessons or to a control activity, then test after some months. Random assignment breaks the link between training and family background, which makes this the only design that can demonstrate causation directly. The catch is that trials are small, short and expensive, so they are underpowered to detect the few point effects at issue.
• Genetically informed designs. Compare twins who differ in how much they practiced, or model the genetic correlation between musical engagement and cognitive ability. These can absorb the shared family environment and the genetic confound at once, and in this literature they have been decisive.

A useful habit for reading any music and intelligence headline: identify which of these three produced the finding before you read the conclusion. The design tells you in advance how much the conclusion is worth.

3 What the Randomized Trials Found

The most cited experimental work in this area is Glenn Schellenberg's 2004 study published in Psychological Science, which randomly assigned six year old children to keyboard lessons, voice lessons, drama lessons or no lessons for one school year, testing IQ before and after. The music groups showed a small advantage over the control conditions, on the order of a couple of IQ points.

Schellenberg himself has been notably careful about how that result should be read, and in later writing he has emphasized how modest and fragile it is. The study is often cited as proof that music lessons raise IQ. What it actually showed is a small average difference in a single trial, in young children, over one year, with an effect size small enough that a handful of similar studies failing to replicate it would erase the finding.

That is roughly what happened next. Subsequent trials produced inconsistent results, several with no advantage at all, and the pattern in the literature became familiar to anyone who has followed cognitive training research: promising early findings, wide public uptake, and a replication record that never quite arrived.

4 The Twin Study That Changed the Picture

The most informative evidence came from a different direction. Miriam Mosing and colleagues, working with a very large sample of Swedish twins, examined whether music practice predicted cognitive ability once genetic and shared environmental factors were accounted for. Their analysis compared twins who differed in how much they had practiced, which holds family background and much of genetic background constant by design.

The association between practice and cognitive ability was present in the raw data, as it always is. Within twin pairs, it did not survive. Twins who had practiced substantially more than their co twin did not show correspondingly higher cognitive ability. The same research group found a similar pattern for musical ability itself: propensity to practice is substantially heritable, which means practice is partly a consequence of predisposition rather than purely a cause of skill.

That design is exactly what the question needed. It explains the raw correlation without requiring a causal effect, and it explains why the experimental literature has struggled to produce one reliably. The people who accumulate thousands of hours of music practice differ from those who do not before the first lesson.

5 Near Transfer, Far Transfer, and Why the Distinction Decides Everything

Cognitive scientists separate two kinds of improvement, and holding them apart resolves most of the confusion in this field. Outside music, the same line sorts the memory claims, where competition mnemonics, eidetic imagery and unusually detailed autobiographical recall all get folded into the single phrase photographic memory despite differing in materials, retention intervals and scoring standards.

Near transfer is improvement on tasks closely related to what was trained. Music training produces near transfer abundantly and uncontroversially. Trained musicians are better at pitch discrimination, at parsing rhythm, at detecting fine timing differences, at reading notation, at auditory stream segregation. Some of these carry into related non musical domains, particularly speech in noise perception, where the auditory system is doing recognizably similar work.

Far transfer is improvement in general capacities such as fluid reasoning or working memory, capacities that would raise scores on tests bearing no resemblance to music. Far transfer is what the "music makes you smarter" claim requires, and far transfer is what decades of research across every training domain has struggled to demonstrate.

Giovanni Sala and Fernand Gobet have made this point repeatedly through meta analyses of music training and of cognitive training generally, finding that once studies with active control groups and adequate designs are isolated, effects on general cognitive ability shrink toward zero. Their broader conclusion, drawn across chess training, working memory training, video game training and music training alike, is that far transfer is rare to the point of being an exception when it appears at all.

This is the honest summary: music training reliably makes you better at music and at hearing. It does not reliably make you better at reasoning about matrices.

It is worth asking why far transfer is so hard to produce, because the answer explains a great deal beyond music. Skilled performance in any domain is built largely from domain specific knowledge: patterns recognized, sequences automated, solutions retrieved rather than derived. Those structures are precisely what does not generalize, because they are representations of that domain and nothing else. A chess master's memory advantage vanishes on randomly arranged boards. A musician's auditory precision does not help with spatial rotation. What would have to transfer is the general machinery underneath, and that machinery appears far less malleable by training than the intuitive model of the brain as a muscle suggests.

This is also why the burden of proof sits where it does. Claiming near transfer is claiming something ordinary and expected. Claiming far transfer is claiming that a specific activity moved a general capacity that decades of deliberate attempts have barely budged, and that claim needs evidence proportional to how surprising it would be.

6 The Mozart Effect and How a Small Finding Became a Product Category

No discussion of this topic is complete without the cautionary tale that shaped public belief more than any actual research.

In 1993, Frances Rauscher, Gordon Shaw and Katherine Ky published a brief report in Nature describing college students who performed slightly better on a spatial reasoning task immediately after listening to a Mozart sonata compared to after silence or relaxation instructions. The effect was small, it was measured minutes after listening, it involved a narrow spatial task rather than general intelligence, and it concerned listening rather than training. The authors did not claim it made anyone permanently smarter.

What followed had almost nothing to do with the study. An industry of infant recordings appeared. A state government distributed classical music to new parents. The phrase "the Mozart effect" entered common usage as shorthand for a permanent intelligence boost from passive listening, which is not what was found.

Careful replication work later suggested that the modest short lived effect was better explained by arousal and mood, that any stimulus a listener enjoyed produced comparable bumps, and that nothing about Mozart specifically was involved. The episode is the clearest illustration available of how a small, narrow, temporary laboratory finding becomes a permanent public belief, and it is worth remembering whenever the next study is announced. Whether the music somebody chooses to listen to tracks ability at all is a separate question with a separate literature, and the preference studies find education predicting genre taste more strongly than measured ability does.

7 The Numbers in Context

Because this field trades in small effects, it helps to see what small means on the standard scale, where the mean is 100 and the standard deviation is 15. A much larger literature makes the same point about that scale: the breastfeeding advantage of 2.62 points, which is what survives once maternal cognitive ability is in the model, still leaves the formula fed child scoring higher in roughly 45 out of every 100 random pairings. The standard scale only means something when a defined comparison group sits behind it, which is one of the items on a checklist for judging whether an IQ test is accurate, alongside reliability and validity.

Put plainly: the largest honestly estimated cognitive effect of music training is smaller than the noise in a single administration of a good test. That does not make it nonexistent, but it does mean no individual could ever detect it in their own scores.

8 Which Cognitive Domains Are Even Plausible Candidates

If music training did influence measured ability, where would it show up? Thinking through the mechanism narrows the field considerably, and it is a useful exercise because it applies to any claimed cognitive intervention. Answering that requires domain-level scores rather than one composite, and a normed online IQ test reports a separate result for each broad ability it samples instead of a single headline figure.

Working memory is the most plausible candidate. Reading ahead in a score while executing the current bar, holding a phrase in mind while transposing it, tracking multiple voices at once: these are working memory operations under time pressure, repeated for thousands of hours.

Processing speed has a plausible route through motor sequencing and timing precision, though the transfer from finely trained motor timing to symbol matching on a test is far from obvious.

Verbal comprehension has an indirect route through auditory processing and phonological discrimination, which is why the literature on music and reading development is livelier than the literature on music and general reasoning.

Fluid reasoning is the least plausible, and it is also the domain most central to the popular claim. Nothing in music practice resembles inferring a rule from an unfamiliar matrix. This is precisely the gap that far transfer would have to bridge, and it is exactly where the evidence is weakest.

Notice how this framework flips the usual conversation. Instead of asking whether music raises IQ, ask which specific capacity would change and by what mechanism. Most claims dissolve at that question, which is why it is a good one to keep.

9 The Brain Imaging Claims and What They Actually Establish

A parallel literature reports structural and functional differences between the brains of trained musicians and non musicians. These findings are frequently presented as the biological proof that music training builds a better brain, and they deserve a careful look because the reasoning error involved is the same one this whole page is about, dressed in more impressive clothing. Music is a favorite host for that error, because scan findings slot neatly into the older story about a creative right hemisphere, when in fact hemispheric specialization is graded and function-specific and even language, the most strongly lateralized function of all, recruits both sides.

The findings themselves are reasonably solid. Musicians show differences in regions associated with auditory processing and motor control, and in the connective tissue linking the two hemispheres, with differences often larger in those who began training early. Functional imaging shows more precise neural responses to sound. None of this is in serious dispute.

The interpretive problem arrives immediately. Almost all of this work is cross sectional: it scans musicians and non musicians and compares them. That design cannot tell you whether training sculpted the brain or whether brains that started out differently made training more rewarding and therefore more likely to continue. The handful of longitudinal studies that scan children before and after training periods are small, and their results are more modest than the cross sectional literature suggests.

There is a second problem that gets less attention. Even where training clearly does change brain structure, and it plausibly does, a structural change in auditory and motor regions is exactly what you would predict from a decade of auditory and motor practice. It is evidence of skill acquisition, which nobody doubts. It is not evidence that general reasoning improved, because the regions involved are not where general reasoning lives. Brain images are persuasive in a way that effect sizes are not, and that persuasive power is precisely why they should be read slowly.

10 Music, Aging and the Cognitive Reserve Argument

A different version of the claim appears in the literature on aging, and it is the version with the strongest practical case, though it still requires care.

The cognitive reserve hypothesis holds that a lifetime of complex mental and social activity builds resilience against age related decline, so that people with richer histories of engagement maintain function longer for a given amount of underlying change. Music making is often cited as a model activity for reserve because it combines complex motor demands, auditory processing, memory, emotional engagement and, in ensemble settings, sustained social interaction. Observational studies do report associations between musical activity in later life and better cognitive outcomes.

The confounding problem is, if anything, worse here than in the childhood literature. Older adults who play music are healthier, wealthier, more socially connected and better educated on average than those who do not, and every one of those factors independently predicts cognitive aging. Reverse causation is also live: early cognitive decline makes demanding hobbies harder to sustain, so people quit playing because they are declining rather than declining because they quit.

What can be said honestly is this. Sustained, complex, socially embedded activity is associated with better cognitive aging, music qualifies as such an activity, and there is no evidence that music is uniquely privileged over other demanding pursuits. Taking up an instrument at sixty is a good idea for many reasons. Presenting it as a proven intervention against decline goes past what the evidence supports. It is also worth being precise about what actually declines, because rank order among peers holds up across a lifetime while absolute performance falls, and any reserve effect has to be visible against that background rather than instead of it.

11 Reverse Causation and Selection

The arrow may point the other way, and there is a decent case that it partly does.

Learning an instrument is cognitively demanding from the first week. It requires decoding a symbolic notation system, coordinating hands independently, holding corrective feedback in mind across repetitions, and tolerating months of sounding bad before sounding adequate. Children who find those demands manageable are more likely to continue. Children who find them punishing quit, and quitting is common.

So the population of people with ten years of training has been filtered, repeatedly, on exactly the traits that cognitive tests measure, plus persistence and tolerance for delayed reward. By the time you compare trained musicians to untrained adults, you are comparing survivors of a decade long selection process to everyone else. Some of the gap you observe was there at the start and some of it was created by the filter, and neither has anything to do with lessons changing brains.

The same logic applies to nearly every "activity X correlates with intelligence" finding you will encounter, from chess to reading for pleasure to learning languages. Ask who stays, and why. It applies to the substance literature too, where higher mental ability at age 10 predicts more problem drinking at 30 rather than less, which is the opposite of the direction the headlines assume.

12 What Music Training Genuinely Does Change

A page this skeptical risks leaving a false impression, so here is the other side stated as strongly as the evidence supports.

Musical training produces some of the most robust and well replicated skill effects in all of cognitive science. Trained musicians show measurable differences in auditory processing, in the precision of neural responses to sound, and in the ability to extract a signal from noise. Their pitch and rhythm discrimination is not marginally better, it is dramatically better. Their ability to sight read, to anticipate harmonic structure, to memorize long passages of music, and to coordinate with other players in real time reflects years of genuine cognitive development in a specific domain.

There is also a defensible case that sustained music training builds habits that matter well beyond music: deliberate practice on difficult material, tolerance for slow progress, self monitoring against an external standard, and performing under pressure. Those are real and valuable. They are also not what an IQ test measures, and that is the point. The gap between "music training develops something important" and "music training raises IQ" is the entire subject of this article.

13 If You Are Deciding About Lessons

Practical guidance follows directly from the evidence, and it is short.

Do not buy lessons as a cognitive intervention. The expected return on that specific investment is somewhere between very small and zero, and it is unmeasurable at the individual level regardless. Any program marketed on IQ gains is selling a claim the research does not support.

Do consider lessons for what they reliably deliver. A skill worth having for life, a social activity with other players, a domain where effort visibly produces improvement, and a demanding discipline pursued for its own sake. That list justifies the cost without any help from psychometrics.

Watch for the enjoyment variable. A child who likes practicing will accumulate the hours that produce genuine musical ability. A child compelled through years of resented lessons gets the cost without the skill, and no cognitive bonus arrives to compensate. If your goal is any of the benefits above, enjoyment is the mechanism through which all of them operate.

14 How to Read the Next Study on This Topic

New research on music and cognition appears constantly, and most coverage of it is unreliable in predictable ways. Four questions will let you evaluate a headline faster than reading the article. The checklist transfers directly to the exercise research, where a genuine randomized effect coexists with the same overclaiming problem this field has.

• Was assignment random, or did participants choose? Self selected musicians tell you about who chooses music, not about what music does.
• Was the control group active? Comparing lessons against nothing at all confounds training with attention, structure and expectation. Drama lessons or sports make a fair comparison; an empty control group does not.
• Is the outcome near or far? Improved rhythm discrimination is near transfer and unsurprising. Improved fluid reasoning is far transfer and extraordinary, and extraordinary claims need the strongest designs.
• How large is the effect, in points? Translate whatever standardized measure is reported onto the familiar scale. Effects that sound impressive in standard deviations often turn out to be two or three points, well inside measurement error.

If you would like to see what a structured cognitive profile actually looks like, measured across separate domains rather than summarized as one number, that is something you can do directly rather than infer from your hobbies. It will not tell you whether your piano lessons helped, because no test can answer that for an individual, but it will tell you where your measured strengths currently sit.

15 Frequently Asked Questions

16 Sources Behind This Page

The relationship discussed here comes from published research, and the honest reading includes its limits. These are the primary sources behind the numbers on this page.

• Schmidt, F.L. & Hunter, J.E. (1998). The validity and utility of selection methods in personnel psychology. Psychological Bulletin, 124(2). The meta-analysis behind general ability as the strongest single predictor of job performance.
• Strenze, T. (2007). Intelligence and socioeconomic success: a meta-analytic review of longitudinal research. Intelligence, 35(5).
• Deary, I.J., Strand, S., Smith, P. & Fernandes, C. (2007). Intelligence and educational achievement. Intelligence, 35(1). A five year cohort of over 70,000 English schoolchildren.
• Kuncel, N.R., Hezlett, S.A. & Ones, D.S. (2004). Academic performance, career potential, creativity, and job performance: can one construct predict them all? Journal of Personality and Social Psychology, 86(1).
• Whalley, L.J. & Deary, I.J. (2001). Longitudinal cohort study of childhood IQ and survival up to age 76. BMJ, 322(7280). Open access.
• Batty, G.D. et al. (2007). IQ and mortality: a population based cohort study. International Journal of Epidemiology, 36(3), 538-553.
• Zagorsky, J.L. (2007). Do you have to be smart to be rich? The impact of IQ on wealth, income and financial distress. Intelligence, 35(5).
• Judge, T.A., Colbert, A.E. & Ilies, R. (2004). Intelligence and leadership: a quantitative review and test of theoretical propositions. Journal of Applied Psychology, 89(3).
• Nisbett, R.E. et al. (2012). Intelligence: new findings and theoretical developments. American Psychologist, 67(2). The broad APA review of what moves measured intelligence and what does not.
• Duckworth, A.L., Peterson, C., Matthews, M.D. & Kelly, D.R. (2007). Grit: perseverance and passion for long term goals. Journal of Personality and Social Psychology, 92(6). The reference study on the non cognitive side of achievement.