What the Research Says About Music
A synthesis of 6 studies on music — what actually works, what doesn't, and how to test it yourself.
Six months of piano practice grew grey matter in the brains of 68-year-olds — but the listening group got the same benefit
Here is the finding that should make you reconsider how you think about music and the brain: In a 2023 randomized trial of 132 healthy older adults (mean age 68.4), six months of either piano practice or active music listening increased grey matter volume in the cerebellum and improved auditory working memory, despite participants experiencing the normal age-related brain shrinkage expected over that period. The piano group practiced at least 30 minutes per day, five days per week. The listening group just attended a weekly one-hour session of music theory and listening exercises. Both groups improved. This matters because it suggests that the cognitive benefits of music may not require years of instrument practice — they might be accessible through structured, attentive listening. But as you'll see, the devil is in the dose, the population, and the outcome you care about.
What the research actually shows
The strongest evidence for music's cognitive effects comes from randomized controlled trials (RCTs) and meta-analyses, not case reports or correlational studies. Let's start with the most rigorous.
The 2023 RCT on older adults (n=132, 114 completers) randomly assigned healthy, musically naive 60-to-79-year-olds to six months of either group piano lessons (with 30+ minutes of daily home practice) or group music culture classes (active listening, theory, no instrument). Both groups showed increased grey matter in the cerebellum and improved accuracy on a tonal working memory task. The effect was not trivial: both groups improved despite general brain atrophy over the same period. The study controlled for sleep quality and practice intensity, and the dropout rate was balanced (9 per group).
Now, the dyslexia RCT is even more striking. A 7-month music training program in 48 children (aged 8–11) with developmental dyslexia produced a ~1.5 standard deviation improvement in phonological awareness and a ~0.8 standard deviation improvement in reading accuracy compared to an active control group that received painting training. That is a large effect. The music program focused on rhythm, melody, and sensorimotor synchronization — not just passive listening. The control group got equal attention from teachers and group interaction, so the effect is unlikely to be a placebo.
The meta-analysis on music training and literacy in typically developing children (13 studies, 901 children aged 4–11) found a smaller but reliable effect: music training produced a ~0.2 standard deviation gain in phonological awareness. Reading fluency did not reliably improve. Importantly, the effect on rhyming skills grew stronger with more hours of practice, suggesting a dose-response relationship.
The meta-analysis on self-efficacy and music performance (46 studies, 5,847 participants) found that believing you can perform well has a medium-sized positive effect on actual music achievement and a medium-sized negative effect on music performance anxiety. Targeted interventions (mastery experiences, verbal persuasion) substantially boosted self-efficacy, but the evidence was weaker for vocalists than instrumentalists.
The original 1964 case report on Melodic Intonation Therapy (MIT) for aphasia is the weakest link in this chain — it's a single patient with no control. But modern meta-analyses (e.g., Popescu et al., 2022) have since found that MIT produces moderate improvements in speech output for chronic nonfluent aphasia patients. The mechanism is thought to involve right-hemisphere language networks being recruited through melodic and rhythmic cues.
Finally, the systematic review on physical activity and academic performance (dozens of studies, ages 5–18) found that adding up to one hour per day of physical activity to the school curriculum does not harm academic performance and may produce small improvements in GPA and classroom behaviour. This is not a music study per se, but it's relevant because it shows that time spent on an enriching activity (like music) does not necessarily trade off against cognitive productivity.
The nuance most people miss
The biggest nuance is that the effects are domain-specific and population-dependent. The ~1.5 SD gain in phonological awareness for dyslexic children is huge, but those children had a diagnosed deficit. The ~0.2 SD gain for typically developing children is much smaller — and reading fluency didn't improve at all. If you're a neurotypical adult hoping music will make you a better reader, the evidence is weak.
The second nuance is the active ingredient. The 2023 older adult RCT found that both piano practice and active listening improved cerebellar grey matter and working memory. This suggests that the benefit may come from sustained, focused auditory attention and pattern recognition — not from motor skill acquisition per se. But the listening group only attended one hour per week of structured sessions; they didn't do daily listening homework. So the dose was lower, yet the effect was comparable. This is either a ceiling effect or a signal that the active ingredient is simpler than we think.
The third nuance is the self-efficacy finding. The meta-analysis found that self-efficacy interventions work, but the effect was weaker for vocalists than instrumentalists. If you're a singer, believing you can perform well may not translate as strongly into actual performance quality. The authors speculate this might be because vocal performance is harder to objectively measure or because singers face different psychological pressures.
The fourth nuance is the aphasia evidence. MIT works for chronic nonfluent aphasia patients, but the original case report is from 1964, and the modern meta-analyses show only moderate effects. This is not a miracle cure. It's a tool that works for a specific population with a specific deficit.
Practical implications
If you want to improve cognitive function in later life, try structured active listening, not just instrument practice. The 2023 RCT found comparable grey matter and working memory benefits from six months of weekly music culture classes (listening, theory, analysis) as from daily piano practice. You don't need to buy an instrument. You need to listen attentively, identify instruments, rhythms, and emotions in pieces, and discuss musical structure. One hour per week, for six months.
If you have a child with dyslexia, consider rhythm-based music training as a supplement to reading instruction. The 7-month RCT showed a ~1.5 SD improvement in phonological awareness and ~0.8 SD in reading accuracy. The program focused on rhythm, melody, and sensorimotor synchronization — not just singing songs. Look for programs that involve tapping to a beat, rhythm discrimination, and melody reproduction. The effect size is large enough that it's worth trying, but it's not a replacement for evidence-based reading interventions.
If you want to improve your own music performance, work on self-efficacy directly, not just technique. The meta-analysis found a medium-sized positive effect of self-efficacy on actual achievement. Interventions that work include mastery experiences (successfully performing a slightly easier piece), verbal persuasion (specific, credible encouragement from a teacher), and managing physiological state (breathing exercises before performance). The effect is weaker for vocalists, so singers may need additional strategies.
If you're recovering from a stroke that caused nonfluent aphasia, ask about Melodic Intonation Therapy. Modern meta-analyses show moderate improvements in speech output. The therapy involves singing phrases with exaggerated prosody and rhythm. It's not a quick fix, but it's one of the few evidence-based approaches for this condition.
Design your own experiment
What to test: Whether 6 weeks of daily active music listening (20 minutes per day) improves your auditory working memory or verbal working memory compared to your baseline.
The intervention: Each day, listen to one piece of unfamiliar instrumental music (classical, jazz, or world music — no lyrics). While listening, complete a structured listening task: identify the instruments, tap along with the beat, predict the next phrase, and describe the emotional arc. Do this for 20 minutes. Use a different piece each day. No background multitasking.
How long to run it: 6 weeks minimum. The 2023 RCT used 6 months, but pilot data suggests changes in working memory can appear in 4–6 weeks. Run a 2-week baseline period first (measure your working memory daily without any intervention), then 6 weeks of the intervention, then a 2-week follow-up.
What to measure: Your primary metric is auditory working memory, measured by a tonal working memory task (you can find free versions online — e.g., hear a sequence of tones and identify whether a probe tone was present). Secondary metric: verbal working memory, measured by Digit Span Backward (repeat a sequence of numbers in reverse order). Measure both at the same time each day, before your listening session.
What confound to watch for: The biggest confound is attention. If you're just listening passively while scrolling your phone, you're not doing the intervention. The active ingredient is sustained, focused auditory attention. Track your subjective focus level (1–10) after each session. Also watch for sleep quality changes — the 2023 RCT measured it as a moderator. If your sleep changes during the experiment, it could confound the results.
What a positive result looks like: A consistent upward trend in your tonal working memory accuracy (percentage correct) over the 6 weeks, with a minimum improvement of 10 percentage points from your baseline mean. For verbal working memory, an increase of at least 1 digit in your maximum backward span. If you see no change after 4 weeks, the intervention likely isn't working for you — try increasing the dose to 30 minutes or switching to a rhythm-based tapping task instead.