Music is not enrichment. It is infrastructure.

A baby a few days old will turn her head toward a piece of music she heard before she was born. She has no words for it. She has no memory she could describe. But the brain that is doing the turning has already been keeping records. In a 2013 study from Helsinki, a team led by Eino Partanen had expectant mothers play a short melody during the last weeks of pregnancy. Months after birth, those infants showed distinct brainwave responses to the tune that other infants did not. The melody had gone in before language did, and it had stayed.

This is the kind of finding that gets filed under “amazing facts about babies” and then forgotten. It deserves better. Put it next to a handful of other findings from the same corner of developmental science and a different picture forms. Music is not something added to a child once the important work of growing is underway. It is part of how the important work gets done.

The case for calling it infrastructure

Enrichment is the word that usually attaches to music in early childhood. Enrichment is optional by definition. It is the nice thing you add when the necessities are handled, the violin lessons after the vegetables. The research from the past three decades suggests the word is wrong, or at least badly placed.

Infrastructure is the better word. Infrastructure is the load-bearing layer underneath the visible activity. You do not notice it when it is working. You notice it when it is missing. A child’s earliest exposure to song, rhythm, and shared movement appears to be doing structural work on attention, on language, on the capacity to coordinate with another human being. These are not enrichments to development. They are among the things development is made of.

The argument does not rest on any single study. Individual findings in this field are often small, and a few of the famous ones have not held up, which is worth saying plainly. The argument rests on a convergence. Three researchers in particular, working across decades and across borders, kept arriving at versions of the same conclusion from different directions.

Sandra Trehub and the infant who was listening all along

Sandra Trehub spent a career at the University of Toronto asking what infants actually perceive when they hear music, and the answer turned out to be: far more than anyone assumed. In a 2003 paper in Nature Neuroscience, Trehub laid out the developmental origins of musicality and made a claim that still reorganizes how people think about babies. Infants are not blank to music. They arrive already equipped to track melodic contour, to notice rhythm, to prefer consonance, to detect when a tune has changed. The musical mind is not built from scratch by training. A working version of it is present at the start.

The detail matters because it inverts a common assumption. Music is often imagined as a high cultural attainment, the last thing a developing mind acquires, available only after the basics of perception and language are in place. Trehub’s body of work points the other way. The perceptual skills that music draws on are among the earliest a child has. An infant who cannot sit up, who cannot reach reliably for a toy, can still notice when a melody she has been hearing is played in a new key but with its shape intact. The equipment for music is online before most of the equipment for everything else.

Trehub also documented something every parent half-knows and tends to dismiss. When a caregiver sings to an infant, the singing holds the baby’s attention in a way that recorded music and ordinary speech do not. In work with Takayuki Nakata published in 2004, infants stayed engaged longer with their mother’s live singing than with her speech. The active ingredient was not vocal skill. It was the live, responsive, face-to-face nature of the song. A parent worried they sing badly can set the worry down. The infant is not grading the performance. The infant is being held inside a shared acoustic event, and that is the thing that matters.

Singing, in this frame, is not a small charming behaviour layered on top of caregiving. It is one of the tools caregiving uses. The lullaby is not decoration. It is regulation, attention, and contact, delivered in a form the infant brain is primed to receive.

Laura Cirelli and the rhythm that builds trust

If Trehub described what the infant brings to music, Laura Cirelli described what music then does between people. Working in Laurel Trainor’s lab at McMaster University, Cirelli ran the experiment this publication has written about before: fourteen-month-olds were bounced to music either in time with a stranger across the room or out of time with them. Minutes later, the babies who had been bounced in sync were measurably more likely to help that stranger when she dropped something. The finding appeared in Developmental Science in 2014, with a companion paper in Philosophical Transactions of the Royal Society B the same year.

Read once, it is a curiosity. Read in context, it is a piece of infrastructure. The babies were too young to talk, so language was not the channel. The song was ordinary, so the music itself was not the point. What carried the social weight was moving in time with another person. Shared rhythm produced a brief, real disposition to help the person it was shared with.

This is the same mechanism that shows up later in childhood. Sebastian Kirschner and Michael Tomasello found in 2010 that four-year-olds who played a synchronized musical game cooperated more afterward than children who played a version without the synchrony. The capacity to feel briefly aligned with another person through rhythm is present in infancy and persists. Trainor and Cirelli have argued that this is, in fact, the general function of making music together across the whole human lifespan. Singing, drumming, and dancing in groups are technologies for producing low-grade affiliation between people who would otherwise have no particular reason to feel it. Babies do this at fourteen months because humans have been doing it for as long as there have been humans.

The follow-up work in Philosophical Transactions of the Royal Society B matters here, because a single striking result is easy to wave away as a fluke. When Cirelli and her colleagues returned to the question, the basic effect held. Fourteen-month-olds were using the experience of synchrony as a cue about whom to help. The infant was not simply in a better mood after being bounced. She was directing her helpfulness toward the specific person she had moved in time with, and not toward a bystander. That specificity is the part that makes it look like social information rather than general good cheer. The rhythm was telling the baby something about a relationship, and the baby acted on it.

It is worth holding onto how early this sits in a child’s life. A fourteen-month-old has a small handful of words at most. She cannot follow an explanation, cannot be reasoned with, cannot be told that the person across the room is friendly. And yet a few minutes of shared rhythm conveyed exactly that, through a channel that runs underneath language entirely. For a developing social mind, this is not a frill. It is one of the first ways a child reads the difference between a stranger and a companion.

The lullaby that crosses every border

The third strand comes from a younger lab and a larger dataset. Samuel Mehr and colleagues at what was then the Harvard Music Lab built a natural history of song, gathering recordings and ethnographic descriptions from dozens of human societies. Their 2019 paper in Science reported that certain musical behaviours appear in essentially every documented culture, and that song types are recognizable across cultural lines. Played a recording from an unfamiliar society, listeners could often tell a lullaby from a dance song from a healing song, without understanding a word.

The lullaby is the cleanest example. In a 2021 study in Nature Human Behaviour led by Constance Bainbridge in Mehr’s group, infants were played lullabies from cultures entirely foreign to their families. The infants relaxed anyway. A song built to soothe a baby in one part of the world soothes a baby on the other side of it. The acoustic signature of the lullaby, slow and soft and narrow in pitch, appears to be something the infant nervous system reads directly, regardless of the language wrapped around it.

What the universals turn out to be is specific rather than vague. The Mehr group did not find that all human music sounds alike, which would be plainly false. They found that particular functions carry particular acoustic signatures across cultural distance. A lullaby tends to be slow, soft, and narrow in pitch range, with low energy, wherever it is sung. A dance song tends to be faster and more energetic. The functions are recognizable because the signatures travel. A listener with no access to the language or the culture can still read what a song is for.

This matters for the infrastructure argument because it suggests the relevant musical behaviours are not cultural luxuries that some societies happen to have invented. They are close to universal, which is what you would expect of something that does developmental work rather than decorative work. Cultures vary enormously in their music. They do not vary in whether they sing to their babies. A behaviour that appears everywhere, that infants respond to before they could have learned it, and that arrives this early in life, looks less like an invention and more like equipment.

Where the rhythm goes next: language and reading

The strands begin to braid when the child gets older and the work turns toward language. Here the link is rhythm, and the destination is reading.

In 2013, Adam Tierney and Nina Kraus reported that the ability to tap in time to a beat was related to a cluster of cognitive, linguistic, and perceptual skills. A child who can keep a steady beat tends to do better on tasks that depend on processing sound accurately in time. Years earlier, in 2002, a study from Trainor’s group with Sima Anvari and colleagues had found that musical skills in preschoolers predicted phonological awareness and early reading ability. Phonological awareness is the capacity to hear that a spoken word is made of smaller sounds, which is the foundation that reading is built on.

The shared thread is timing. Speech is a stream of sound that the brain has to segment into rapid, precisely ordered pieces. Music is also a stream of sound organized in time, and tracking a beat exercises the same temporal machinery. A child who has spent her early years inside rhythmic, sung, shared sound has been practising, without anyone calling it practice, the exact perceptual skill that reading will later demand. Aniruddh Patel’s broader work on music and language, set out in his 2008 book, gives the theoretical scaffolding for why two systems that look so different lean on overlapping circuitry. The detailed version of that argument lives in our piece on the OPERA hypothesis, but the early-childhood version is simpler. Rhythm is foundational, and reading collects on the foundation later.

Four strands, one structure

Set the strands beside each other and the infrastructure metaphor stops being a flourish. Before birth, a melody can lay down a trace in a brain that has no other way of recording the world. In the first months, a caregiver’s singing holds and regulates an attention that has very few other anchors. In the second year, shared rhythm tells a child whom to trust, in the years before she could be told in words. And from preschool onward, the timing skills she has been exercising in song quietly become the timing skills that reading demands. Perception, regulation, attachment, language. These are not the soft extras of childhood. They are close to the whole job.

What makes this a structural argument rather than a list of nice findings is the order of operations. In each case, the musical experience comes first and the developmental capacity collects on it later. The infant is musical before she is verbal. She is rhythmic before she is social in any deliberate way. The foundation is poured early, and the house goes up on top of it. That is precisely what infrastructure means: the part that has to be in place before the visible structure can stand.

None of this requires a child to be talented, or trained, or enrolled in anything. The research is mostly about ordinary infants in ordinary homes, exposed to the ordinary music that human caregivers have always made around their young. The studies did not engineer special interventions. For the most part they observed what happens when a baby is sung to, bounced, soothed, and surrounded by song, and then measured the result. The finding is that the result is load-bearing.

What the honest version sounds like

It would be easy to overstate this, and overstating it is how the field gets into trouble. The claim is not that playing more music produces smarter children, or that a baby enrolled in a class becomes kinder or reads sooner than one who is not. Most of the studies measured something specific and bounded: a brief disposition, a perceptual skill, a longer stretch of attention, a correlation that holds across a population without promising anything about an individual child. The famous overreaches in this area, the ones that turned a small lab result into a claim that classical music makes babies brilliant, are a caution, not a model.

What the convergence supports is quieter and sturdier. The infant brain arrives ready for music. Singing holds an infant’s attention and helps regulate it. Moving in time with another person builds a real, if temporary, bond. The capacity to soothe a baby with a slow song appears in nearly every human culture. And the timing skills exercised by early musical experience feed directly into the timing skills that language and reading require. None of these is enrichment in the optional sense. Each is part of the ordinary developmental machinery, and music is one of the ways the machinery runs.

The melody that went in before words

Return to the baby turning her head. The melody she heard before birth was not making her smarter, and the Helsinki researchers did not claim it was. It was doing something plainer and stranger. It was already part of how her brain was organizing the world of sound, weeks before she had a world to organize it in.

That is the shape of the whole picture. Music in the early years is not the thing you add once the foundation is poured. It is mixed into the pour. A parent singing in a kitchen, a grandparent swaying with a bundle in their arms, a caregiver bouncing a fussy infant to whatever is on the radio: these look like ways of passing the time. They are also, in a measurable sense, the foundation being laid. The vegetables and the violin were never two different categories. For a very young child, the song was always part of the necessities. You can find more of the research on these early years in this section of the publication.

If you’d like a few of these letters delivered each month, you can subscribe here.