A Measure of Relationships, Not of People

John Bowlby was a psychoanalyst who spent the 1940s working with delinquent and institutionalised children, and he arrived at a conclusion that the psychoanalysis of the time had no place for: that the relationship with a caregiver is a biological need in its own right, not a by-product of being fed.

The prevailing view was cupboard love. Infants attach to whoever supplies food, because feeding reduces a primary drive and the person supplying it acquires value by association. Attachment on this account is learned, secondary and essentially economic.

Harry Harlow's rhesus monkeys destroyed that view, at considerable cost to the monkeys. Infant macaques were raised with two surrogate mothers: one made of bare wire that dispensed milk, and one covered in terrycloth that dispensed nothing. If cupboard love were right, the infants should have attached to the wire mother.

They spent almost all their time on the cloth mother, going to the wire one only to feed and returning immediately. When frightened by a mechanical toy, they ran to the cloth mother and clung. With the cloth mother present they would eventually explore the frightening object; without her they froze or huddled.

Contact comfort, not calories. The studies are also among the most-cited examples in the case for ethical review, and the monkeys raised in isolation were profoundly and permanently damaged.

Mary Ainsworth then built the measure that made attachment researchable in humans, and it is the measure that has been most thoroughly misunderstood.

The Strange Situation is a twenty-minute laboratory procedure with a set sequence of separations and reunions between an infant, a caregiver and a stranger. It is scored almost entirely on the reunions.

That is the part everyone gets wrong. Whether the infant cries when the caregiver leaves carries very little information. What marks secure attachment is what happens on return: the infant seeks the caregiver, is comforted, settles, and goes back to exploring. Avoidant infants do not approach and remain focused on toys; ambivalent infants approach and cannot be settled, alternating clinging with resistance. A fourth category, disorganised, was added later for infants showing contradictory or fearful behaviour toward the caregiver, and it is the one most associated with later difficulty.

Bowlby's theoretical claim about what these patterns are is the crucial one. He proposed that a child builds an internal working model: a set of expectations about whether others will respond when needed, and about whether they are the kind of person others respond to. Expectations, note, rather than traits. Expectations can be revised when the evidence changes; traits cannot.

Which is where the popular version parted company with the research.

Attachment style, in its journey into relationship advice, became a fixed personality type acquired in infancy and carried for life: you are anxious, or avoidant, or secure, and this explains your romantic history. The longitudinal evidence does not support that. Correlations between infant classification and adult attachment measures are moderate at best, a substantial minority of people change classification, and change is predicted by life events: a parent's illness, a family's circumstances improving, a relationship that works.

The classification also describes a relationship rather than a child. The same infant can be classified differently with each parent, because it is a measure of what this pair has worked out between them. And insecure classifications are common among typically developing children; they are not disorders and none of them is a diagnosis.

So a nursery worker told that a toddler is "insecurely attached" is being given a description of how that child currently uses adults for comfort, which is useful and actionable, rather than a prognosis.

The hard evidence about how much early experience matters comes from a case nobody could ethically have designed, and which history supplied.

When Ceaușescu's regime fell in 1989, Romanian institutions were found to contain many thousands of children raised in conditions of extreme deprivation: minimal handling, no individual caregiver, little stimulation. Large numbers were adopted into British and other families, and Michael Rutter's team followed a cohort of them for decades.

The results are among the clearest timing effects in human development. Children adopted before about six months of age were, by adolescence, largely indistinguishable from within-country adoptees. Children who had spent longer showed substantially elevated rates of cognitive impairment, inattention and a striking pattern of indiscriminate friendliness toward strangers, and those differences were still present in adulthood at age twenty-five.

Two things follow, and both matter. Early deprivation has real and lasting effects, which is an argument for intervening as early as possible. And a large fraction of even severely deprived children did well, which is an argument against treating any early history as a sentence.