How Attachment Forms in Childhood

In 1958, the American psychologist Harry Harlow gave infant monkeys a choice between two substitute mothers: one made of bare wire that dispensed milk, and one wrapped in soft cloth that offered no food at all. Every theory of the day said a baby should bond with whatever feeds it. The monkeys chose the cloth.

That result cracked open a question psychology still works on today. How exactly, in the first months and years of life, does a bond take shape that we then carry with us for decades? Not the warm idea of love, but the machinery underneath it.

Bowlby and the evolutionary function of attachment

While Harlow watched monkeys, the British psychiatrist John Bowlby was watching human children. Through the 1940s and 1950s he worked with orphans and with children pulled away from their parents during and after the war. He kept seeing the same thing: separation from a caregiver left marks that hunger and a shortage of toys could not explain.

In 1951 he gathered his findings in a report for the World Health Organization, then developed them into the trilogy Attachment and Loss, whose first volume appeared in 1969. His claim was radical for its time. The need for closeness is not a bit of spoiling that a child grows out of. It is an inborn system that evolved because young who stayed near an adult were more likely to survive predators, cold, and hunger.

Seen this way, even a toddler's meltdown the instant a parent slips around a corner starts to make sense. The crying, the outstretched arms, the crawling after you are not tantrums. They are an alarm system with a single job: keep the protector close. Bowlby called the parent a secure base, the spot a child ventures out from precisely because it knows it has somewhere to come back to.

This is where the internal working model is born. From thousands of tiny exchanges, a child assembles answers to three unspoken questions. Will someone come when I call? Am I worth being noticed? Can this world be trusted? The answers get filed away before the child can speak.

Harlow's monkeys

Back to those monkeys, because the detail matters. Harlow's babies clung to the cloth mother for hours at a stretch and went to the wire one only for a quick drink before hurrying back. When something frightened them, a loud mechanical toy or an unfamiliar object, they ran to the soft figure and pressed against it until they settled.

Feeding, in other words, did not manufacture the bond. Contact did. The warmth, the softness, the sense of a body to hold onto counted for more than the calories. Harlow named it contact comfort, and it dismantled the reigning idea that a baby loves its mother as a kind of reward for being fed.

The experiment was blunt, and by today's ethics it would never pass. But it made visible something Bowlby had argued from clinical observation alone: closeness itself, not food, is what the attachment system reaches for.

The Strange Situation

For a long time Bowlby's theory sat closer to philosophy than to science. There was no way to actually measure attachment. His collaborator, the psychologist Mary Ainsworth, supplied one. In the early 1970s in Baltimore she refined a procedure called the Strange Situation.

It runs surprisingly plainly. A parent and a child of roughly 12 to 18 months come into a bare room with some toys. Over about twenty minutes, eight short episodes play out. A stranger walks in. The parent quietly steps out and leaves the child on its own for a little while. Then the parent returns, and the whole sequence repeats once.

Ainsworth was not mainly watching how hard the child cried when the parent left. Almost every child does. What interested her was the reunion, what the child does the moment the parent reappears in the doorway. The response to that return reveals the strategy the child has built toward this particular person.

Three patterns emerged, and a fourth was added later by Mary Main and Judith Solomon in 1986. Roughly two thirds of children show secure attachment; the rest divide among three insecure forms.

Attachment type	When the parent leaves	When the parent returns	Rough frequency
Secure	Usually upset, looks for the parent	Settles quickly and goes back to play	about two thirds
Avoidant	Outwardly calm, keeps playing	Ignores the parent or steers around contact	roughly a fifth
Anxious (ambivalent)	Intensely distressed, hard to soothe	Reaches for the parent yet arches away in anger	a smaller group
Disorganized	No clear strategy	Contradictory movements, freezing, confusion	more common in high-risk homes

The avoidant group is the one that misleads. On the surface the child acts as though the parent's exit is no big deal, and could pass for an unusually self-sufficient, contented kid. Measures of heart rate and the stress hormone cortisol tell the opposite story. Inside, the child is as stirred up as the ones who cry, and has simply learned not to show it. The calm is a strategy, not ease.

What shapes attachment

So why does one child build a secure attachment and another an insecure one? Ainsworth's answer was not household income or the number of toys. It was sensitive responsiveness: a caregiver's ability to notice a child's signals and answer them in a fitting way.

This is worth pausing on, because a lot of needless parental guilt collects around it. Sensitive responsiveness does not mean being available every second and getting it right every time. If anything, the opposite. The child psychoanalyst Donald Winnicott spoke of the "good enough mother", a parent who meets a child's needs reliably enough for it to feel safe, but not so flawlessly that the child never runs into any frustration.

In practice it looks unremarkable. A one-year-old drops her spoon off the high chair and starts to cry. Her mother, assuming hunger, first lifts a piece of food toward her, and the crying only gets louder. Then she spots the spoon on the floor, hands it back, and the tears stop. It was never about food; it was about the lost spoon, and the parent got there on the second try. Dozens of these micro-moments happen in a single day.

The developmental psychologist Edward Tronick pushed this further. His "still-face" experiment (1975) is still one of the most striking demonstrations of how early a baby actively co-creates a relationship. The setup is simple. A mother plays normally with her few-month-old for a couple of minutes. She coos, smiles, echoes the baby's sounds. Then, on cue, she goes blank, still looking at the baby but with a perfectly motionless face, no smile, no reply.

What follows is hard to watch even on video. The infant first trots out everything that has ever worked. A smile, a squeal, a pointed finger, arms reaching up. When none of it lands, the baby grows visibly distressed, turning away, hunching, sometimes bursting into tears. Two or three minutes of no response is enough to undo a baby who otherwise wants for nothing. The moment the mother "thaws" and reconnects, most babies pull themselves back together within seconds.

Tronick drew two lessons from it. A baby is not a passive recipient of care but an active partner who expects something from the exchange and comes apart without it. His recordings also showed that ordinary parent-child interaction is slightly out of tune most of the time. The parent misses a cue, the baby looks away, something does not fit.

What decides the outcome is not whether these lapses happen. They happen to everyone. What decides it is the repair: the parent notices the mismatch, reconnects, and the child calms. Out of that endless run of small ruptures and reconciliations grows a belief that relationships can be mended.

How much childhood decides adulthood

Here comes the part that worries a lot of parents. If attachment forms this early, does the first year or two settle every relationship to come? The short answer is no.

A 2002 meta-analysis by R. Chris Fraley, pooling long-term studies that followed people from infancy into adulthood, found only a modest link between early attachment and later attachment. Fraley set two models against each other. In one, a person carries the early pattern along more or less unchanged; in the other, the original model keeps being rewritten by new experience. The data land somewhere in between. Early attachment hints at something. It settles almost nothing.

Psychologists even have a term for the hopeful side of this: earned security. It describes people who grew up insecurely attached and still built a secure style in adulthood. Attachment is not concrete that sets hard at eighteen months.

What does that rewriting look like day to day? Most often it comes through a long relationship with someone who reacts predictably. A partner who does not turn cold when you say what you need. A friend who picks up the phone after midnight. The old working model insists that no one can be relied on. The new experience keeps contradicting it, and once it repeats for long enough, the newer version starts to win.

The second route runs through therapy. For a while the therapist takes on the secure-base role, someone with whom you can try the things that were not safe to try at home. Saying out loud what bothers you. Admitting a fear without bracing for a penalty. Neither route is fast. It is not one powerful experience that fixes everything, but hundreds of small ones that stack up, much the way the originals once did.

Can you recall a moment from your own childhood when you simply knew there was somewhere to come back to? It need not be anything dramatic. For most people it was enough to sense that if something went wrong, someone would be home. And if no such memory comes to mind, that is not a sentence handed down. The word "earned" is the whole point: security can be built much later, too.

How the very same attachment plays out in an adult partnership is its own subject, which we cover in the guide to attachment in adult relationships. If you want to see where you land, you can discover your own adult attachment style in a short test.

"Attachment parenting" is not attachment theory

One thing confuses parents perhaps more than anything else. The word "attachment" turns up in two completely different places. Alongside the scientific theory of attachment there is a popular parenting movement, attachment parenting, made famous in the 1990s by the pediatrician William Sears. It rests on specific practices: carrying a baby in a sling, sharing a bed, extended breastfeeding.

It sounds logical, except the two "attachments" have less to do with each other scientifically than the shared word suggests. The name itself was more marketing than research, and plenty of attachment scientists bristle at seeing it mistaken for the theory. Nothing in Bowlby's or Ainsworth's work ever said a secure bond is produced by a sling or a shared mattress.

The distinction is subtle, and it matters. What the research predicts secure attachment from is sensitive responsiveness, whether a parent reads a child and answers. Not whether the child sleeps in the parents' bed or in a cot beside it. A sling can make responsiveness easier, since the baby is closer and its signals get noticed sooner. On its own it guarantees no particular bond, and a parent who skips the sling loses nothing.

For a lot of worn-out parents that difference lands as a relief. You do not give a child safety by ticking off a list of practices. You give it by paying attention, imperfectly and day after day, and by understanding the child a little better as you go. The bond Bowlby described grows out of plain, ordinary noticing, not out of a parent getting everything right.