Preface: Following the previous #AI series and #MarvinMinsky series. This post is co-written with Claude.

While Harlow was running his experiments in Wisconsin, the British psychiatrist John Bowlby was turning similar observations into a theory in London; the two corresponded. Bowlby's starting point was children who stole. In the late 1930s, at the London Child Guidance Clinic, he compared 44 children referred for theft with 44 children who did not steal, and published the study in 1944 as "Forty-Four Juvenile Thieves." Fourteen of the thieves struck him as a distinct type, cold and seemingly without guilt, whom he called "affectionless." Twelve of those fourteen had been separated from their mothers for long periods early in life. Among the controls, only two had.

Over the following decade and more, Bowlby developed this observation into a theory. In 1951 he wrote a report for the World Health Organization on homeless and institutionalized children. In 1952 his colleague James Robertson released a short film, A Two-Year-Old Goes to Hospital, following a small girl named Laura through eight days in a hospital that allowed parents only brief visits. Audiences watched her move from protest to despair to a frozen calm. In 1958, the same year as Harlow's address, Bowlby published "The Nature of the Child's Tie to His Mother." Drawing on ethology, on Harlow's monkeys, and on these observations, he argued that attachment is a biological system in its own right: infants are built to seek closeness to a caregiver when frightened, and to use that caregiver as a secure base from which to explore. Over time, a child builds what he called internal working models. Internal working models are expectations, formed in early relationships, about whether others will be there when needed and whether oneself is worth caring for. His trilogy Attachment and Loss appeared between 1969 and 1980.

The trilogy took Bowlby more than a decade, and it was written against his own profession. He had trained as a psychoanalyst in London in the 1930s; his analyst was a Kleinian, and Melanie Klein herself supervised one of his child cases. He later recalled being annoyed that Klein treated the child's fantasies as what mattered, and paid little attention to what was actually happening in the family.

The disagreement had roots in Bowlby's own childhood. He was born in London on February 26, 1907, the fourth of six children of the surgeon Sir Anthony Bowlby. As was usual for families of their class, the children were raised by nannies in a separate nursery, and Bowlby normally saw his mother for about an hour a day, after teatime. His primary caregiver was a nursemaid named Minnie. She left the family when he was almost four. He later wrote that for a child to be looked after by a loving nanny and then lose her at that age could be almost as tragic as the loss of a mother. Her replacement was colder and sarcastic. At around seven he was sent away to boarding school, an experience he later condemned in a single line: he would not send a dog away to boarding school at seven.

His route into psychoanalysis ran through troubled children. After Cambridge, he spent six months teaching at Priory Gate, a progressive school for what were then called maladjusted children, and later described it as the most valuable six months of his life. A staff member there, John Alford, advised him to train in child psychiatry and psychoanalysis. At twenty-two Bowlby entered medical school at University College Hospital in London and at the same time began analysis with Joan Riviere, one of Klein's closest followers. He qualified in medicine in 1933, trained in adult psychiatry at the Maudsley Hospital, and qualified as a psychoanalyst in 1937. From 1937 to 1940 he worked at the London Child Guidance Clinic, which treated children's problems as rooted in what had happened to them in their families. That approach suited him.

The British psychoanalytic world he joined was split. Melanie Klein, who had moved to London in 1926, held that a child's inner world of fantasy, especially fantasy about the mother, was the proper object of analysis. When Sigmund and Anna Freud arrived as refugees from Vienna in 1938, the British Psychoanalytical Society divided into Kleinians and Anna Freud's followers, with a third, independent group forming between them. Bowlby, analyzed by a Kleinian, would end up among the independents.

The case that turned him came in 1938, when he began training in child analysis with Klein as his supervisor; his archive still holds a letter from her that year suggesting toys to use with child patients. He was assigned a boy of about three who was hyperactive, anxious, and aggressive, running in and out of the room. The boy's mother, herself anxious and troubled, sat in the waiting room during each session. After three or four months, Bowlby learned that she had been admitted to a mental hospital.

In a talk recorded many years later, he said that what shocked him was that Klein was not interested, except that the mother's breakdown interrupted the analysis. That the mother's own problems might have something to do with the boy's seemed to him obviously relevant; to Klein it was beside the point. He said a second case later followed the same pattern. Decades afterward, colleagues noted that he still became angry when he told the story. Klein stressed the child's fantasies about the mother; Bowlby insisted on the actual history of the relationship. Attachment theory grew out of that difference: it kept much of the psychoanalytic interest in early bonds, but rejected the idea that what really happened to a child could be set aside.

A Change of Heart

From about 1951 he found an alternative in ethology, starting with Konrad Lorenz's work on goslings that follow the first moving object they see.

Ethology is the biological study of animal behavior, observed as far as possible in the animal's natural setting and treated, like anatomy, as a product of evolution. Its founders distrusted the laboratory psychology of the day, with its rats in mazes and pigeons pressing levers, and preferred to watch what animals did when left to live.

Konrad Lorenz, born in Vienna in 1903, grew up on his family's estate at Altenberg on the Danube, where his parents, he later wrote, were supremely tolerant of his love for animals. He trained as a physician to please his father, then took a second doctorate in zoology in 1933. His teacher Oskar Heinroth had already noticed that some newly hatched birds attach themselves to whoever is there; the nineteenth-century naturalist Douglas Spalding had seen it before that. It was Lorenz's 1935 study, Der Kumpan in der Umwelt des Vogels ("The Companion in the Bird's World"), that gave the phenomenon its name and its classic description. Imprinting is the process by which the young of birds that leave the nest soon after hatching, such as geese and ducks, bond with the first moving figure they encounter during a short window after birth, and from then on follow it as a parent.

The best-known images come from what Lorenz called his "goose summers" at Altenberg in the mid-1930s. In 1936 he hand-raised a greylag gosling named Martina, who imprinted on him, and photographs from those years show him walking down a path or rowing across water with a line of goslings following behind. The goslings had not attached themselves to him because he fed them; they followed him because he was the first thing that moved. Lorenz also showed that imprinting does more than secure a parent: it teaches the bird what its own kind looks like, and so directs whom it later courts. In 1936 he met the Dutch biologist Niko Tinbergen at a symposium in Leiden, and the two built ethology into a discipline. Lorenz's popular book on his animals, published in German in 1949 and in English in 1952 as King Solomon's Ring, carried the goslings to a general readership at about the time Bowlby encountered them.

The man who gave Bowlby a biology of the bond between mother and young had also put the same biology to darker use. Lorenz joined the Nazi Party in 1938, after Germany annexed Austria, and published papers arguing, in the language of domestication and degeneration, for the regime's ideas of racial hygiene. He served as a military doctor, was captured by the Red Army in 1944, and spent four years as a prisoner of war in Soviet Armenia. He later said he regretted his party membership. In 1973 he shared the Nobel Prize in Physiology or Medicine with Tinbergen and Karl von Frisch for their discoveries about individual and social behavior. Bowlby took from him the observation that a young animal's attachment does not depend on food. He did not take the politics.

The ethologist Robert Hinde became a lifelong friend and colleague, and it was Hinde who put him in touch with Harlow. Robert Hinde was sixteen years younger than Bowlby and came to animal behavior from the war. Born in Norwich in 1923, he was called up into the Royal Air Force at eighteen, trained as a pilot in Rhodesia and as a navigator in South Africa, and spent the rest of the war flying flying boats for Coastal Command over the Indian Ocean and the Atlantic. In January 1946 he went up to St John's College, Cambridge, to read zoology, then moved to Oxford to study the great tit under the ornithologist David Lack. The person who shaped him most there was not Lack but Niko Tinbergen, who arrived in Oxford from Holland in 1950. That same year Hinde became curator of the new Ornithological Field Station at Madingley, outside Cambridge, where he studied imprinting, motivation, and nest-building in birds. One of his early papers recorded how British tits had learned to open the foil caps of milk bottles left on doorsteps, and it remains a classic study of animals learning from one another.

Hinde was an ethologist who argued with ethology. In the mid-1950s he publicly criticized Lorenz's "energy model" of instinct, the idea that a drive builds up inside an animal like water behind a dam until it is released. Lorenz, Hinde later said, regarded him as a traitor for the rest of his life, until a Christmas card finally arrived in the year Lorenz died. The skepticism mattered for Bowlby. He took ethology's method, close observation of behavior as an evolved function, from Lorenz and Tinbergen, but in building his own theory he also left behind the hydraulic picture of drives that psychoanalysis and early ethology shared, and described attachment instead as a control system that monitors a goal. Hinde went on to become a Royal Society Research Professor, turned increasingly from animal behavior to the study of human relationships, served as Master of St John's College from 1989 to 1994, and chaired the British branch of the Pugwash movement against war. He died in 2016, at ninety-three.

1957, "The Nature of the Child's Tie to His Mother."

On June 19, 1957, Bowlby presented the first of his new papers to the British Psychoanalytical Society; it became "The Nature of the Child's Tie to His Mother." Over the next few years he read further papers on separation anxiety and on grief in infancy. They received a mainly hostile reception. Anna Freud, who had argued that infants could not grieve for more than moments, wrote that Bowlby was too valuable a person to get lost to psychoanalysis. Donald Winnicott confessed in a letter that the papers were building up in him a kind of revulsion. Bowlby never resigned from the Society, but he rarely attended its meetings.

The first volume, Attachment (1969), set out what the child's tie to its mother is, how it develops, and what it is for. Bowlby borrowed from evolutionary biology the idea that the tie exists because it kept infants alive, and from control-systems theory the idea of a system that monitors a goal and corrects for deviations, the way a thermostat holds a temperature. The attachment behavioral system, in Bowlby's account, is a built-in system whose goal is to keep a young child close enough to a caregiver; when the child is frightened, ill, or threatened, it switches on, exploration stops, and the child seeks the caregiver until it feels safe again. He treated it as a motivational force on the same level as the drives psychoanalysis had built itself around, sex and aggression. He revised the volume in 1982.

The idea had been forming since 1958. In "The Nature of the Child's Tie to His Mother," Bowlby proposed that the attachment visible in a twelve-month-old is built from a handful of simpler responses that mature at different times in the first year. In three of them, sucking, clinging, and following, the baby is the active partner. The other two, crying and smiling, are signals: they work by drawing the mother close. Only sucking is tied to feeding. The rest have nothing to do with food, which was exactly the point Harlow's cloth mother made in monkeys.

By 1969 he had replaced that list with a single organizing idea. He took the term "behavioral system" from the ethologists, who used it in place of "instinct" because the older word explained nothing. A behavioral system, in this sense, is a set of behaviors that are organized around one goal, switched on and off by particular conditions, and preserved by evolution because they helped the species survive. For attachment, the goal is not food or pleasure but a degree of closeness to, or access to, a particular caregiver. The system behaves like a thermostat. It keeps track of where the caregiver is, what state the child is in, and what is happening around them, and when the distance or the danger exceeds what the goal allows, it sets off behaviors that close the gap: crying, calling, approaching, following, clinging.

What switches the system on falls into two groups. Some triggers come from inside the child: illness, hunger, pain, exhaustion. Others come from outside: a stranger, darkness, a sudden noise, something approaching fast, being left alone, the caregiver moving away. Bowlby explained why these, rather than other things, should be alarming by reference to what he called the environment of evolutionary adaptedness, the conditions in which humans evolved, which he pictured as resembling the life of hunter-gatherers. For an infant in that world, being alone, in the dark, or near something unfamiliar was dangerous, above all because of predators. A child who protested and moved toward its caregiver at such moments was more likely to survive. What switches the system off depends on how strongly it was triggered. A mildly uneasy child may be calmed by the sight or sound of the caregiver; a frightened one needs to be held.

The system does not work alone. Bowlby set it against a second system, exploration, which draws a child out toward the new and unfamiliar. The two pull in opposite directions. When the attachment system is quiet, the child explores; when it is activated, exploration stops and the child returns to the caregiver. This is the secure base that Harlow saw in his open-field test and that Ainsworth would later measure in the Strange Situation. A child who is confident that the caregiver is available can turn its attention outward to the world. A child who is not must keep its attention on the caregiver.

Bowlby also described how the system develops in four phases. In the first two months or so, the infant orients and signals to anyone, with little discrimination. From about two to six months, it begins to respond differently to familiar people, smiling and vocalizing more readily for the caregiver. From about six months, once it can crawl, the system takes its clear form: the child protests when the caregiver leaves, greets her on return, clings when frightened, follows when it can, and grows wary of strangers. From around the third year, the child begins to understand the caregiver as a separate person with her own goals and plans, and the relationship becomes what Bowlby called a goal-corrected partnership: the child can accept a separation it understands, negotiate, and wait. By three or four, a brief physical absence no longer threatens the bond in the same way. Bowlby held that a child usually has one principal attachment figure at the top of a small hierarchy of others, an idea, called monotropy, that later researchers have disputed.

Separation: Anxiety and Anger (1973)

The second volume, Separation: Anxiety and Anger (1973), applied the scheme to what happens when the tie is broken. It drew directly on the hospital observations behind Robertson's film. Bowlby described a sequence in young children separated from their mothers: protest, despair, and detachment. First the child cries, calls, and searches. Then it grows quiet and listless. Finally it seems to recover, takes an interest in the nurses, and on the mother's return may barely acknowledge her. Against the common view of the time, he argued that young children grieve with no less intensity than adults. He also argued that fear of separation is not irrational: for a young primate, being alone is a natural cue to danger. Anger, in this account, is often a bid to restore the bond, a protest meant to make the absence less likely to happen again. Conditions such as school refusal and agoraphobia, he suggested, were often forms of anxious attachment rather than fears of the school or the street.

The book opens with a chapter Bowlby called "Prototypes of Human Sorrow." The evidence was a decade and a half of watching small children in hospitals and residential nurseries: James Robertson's notes from 1948 onward and his film A Two-Year-Old Goes to Hospital (1952); Christoph Heinicke and Ilse Westheimer's Brief Separations (1965), which followed toddlers placed in residential nurseries for two weeks or more and compared them with children who stayed home; and the Young Children in Brief Separation films that James and Joyce Robertson made between 1967 and 1973. The most painful of these, John, Nine Days in a Residential Nursery (1969), follows a seventeen-month-old left in a nursery while his mother has a second baby. The nurses are kind but overstretched and rotate in shifts, and over nine days John moves from searching for them, to trying to attach to whichever adult passes, to refusing food and lying face down with a toy bear. When his mother comes back, he will not look at her. Bowlby also recorded what happened afterwards at home: for weeks or months, children who had been away were clingy, refused to let the mother out of sight, or turned on her with hostility, and often did both by turns.

The Robertsons' films also complicated his argument. In the same series they took four children into their own home as foster parents, kept to each child's routines, and talked constantly about the absent mother; those children were sad but did not collapse the way John did. The Robertsons concluded that much of what Bowlby called the response to separation was really the response to separation with no substitute care, and the two men's relationship cooled over how much weight the finding deserved. Bowlby acknowledged the result in Separation but held that the loss of the particular person still mattered, even when the care around the child was good.

The middle third of the book is an ethological theory of fear. Psychoanalysis had treated separation anxiety as a puzzle to be explained by something else: in Freud's later account, fear of losing the mother was really fear of the unmet needs her absence would bring. Bowlby turned the question around. Humans, like other animals, are frightened by a small set of what he called natural clues to danger: strangeness, sudden noise or movement, rapid approach, heights, darkness, and being alone. None of these is harmful in itself, but over evolutionary time each has gone along often enough with real danger that reacting to it early paid off, a logic he summed up under the heading "better safe than sorry." Being alone is the clue that matters most, because it compounds every other one: a strange noise in the dark is far more frightening with no one there, and a familiar companion takes much of the fear out of the same situation. On this view the child who screams when the door closes is not being unreasonable. It is responding correctly to a signal that, for most of human history, did mean danger. Bowlby proposed keeping two words apart: alarm for fear of something present and threatening, and anxiety for fear arising from the absence or uncertain availability of an attachment figure.

The last third deals with why some people are far more prone to this fear than others, and Bowlby's answer was the experience of how available the attachment figure had been. The prevailing explanation of clingy, "overdependent" children was the theory of spoiling: too much comfort and attention produced a child who could not let go. Bowlby argued the reverse. What he renamed anxious attachment grew from uncertainty, in children reared without one permanent mother figure, in children who had been through repeated separations or long days of rotating substitute care, and in children whose parents used the threat of leaving as discipline. He gave particular weight to threats: telling a child that you will go away, send them away, stop loving them, or kill yourself if they misbehave. Such threats were common, he noted, and rarely mentioned by parents to clinicians. He also described the inverted relationship, in which a parent leans on the child for care and the child, outwardly mature and dutiful, is in fact anxious and tied to home. Against the spoiling theory, he pointed to evidence that children who are confident their parents will be there when needed explore more and are more self-reliant, not less. Self-reliance, in his account, rests on dependable dependence rather than being its opposite.

The chapter on anger gives the book the second half of its title. Bowlby treated anger as a normal part of the response to separation, and in its ordinary form as functional. The child who hits or ignores the mother on her return, or the adult who rages at a partner who came home late, is protesting the absence in a way that makes the other person less likely to repeat it. The reproach serves the bond. It turns dysfunctional when it grows too intense or lasts too long, or when it cannot be expressed: a child who has been told that bad behaviour will drive a parent away learns that anger itself puts the bond at risk, and the result is a tangle of resentment, guilt, and fear that Bowlby called ambivalence. In such cases, he argued, anger and anxiety feed each other, since each outburst brings the dreaded loss closer.

He then reread a set of clinical labels in these terms. "School phobia," he argued, was usually not fear of school but fear of leaving home, most often because the child was worried about what might happen to a parent while away, or because a parent, consciously or not, wanted the child at home. He reread Freud's famous case of Little Hans, the boy afraid of horses, and pointed to something Freud had passed over: Hans's mother had threatened to leave the family. He made a similar case for many adult cases of agoraphobia, which he saw less as fear of open spaces than as fear of being away from an attachment figure. In the chapter titled "Omission, Suppression, and Falsification of Family Context" he described how these histories go missing from the clinical record: parents leave out the threats they made, children are told that what they saw or heard did not happen, and the patient arrives with a fear that seems to have no source. That thread became his 1979 paper "On Knowing What You Are Not Supposed to Know and Feeling What You Are Not Supposed to Feel."

The book is also, at moments, close to autobiography. Bowlby was sent to boarding school at seven, and the psychiatrist Arturo Ezquerro points to a line in Separation about how long a school term feels to a small child: "The thirteen weeks of a term might just as well be thirteen years."

Loss: Sadness and Depression (1980)

The third volume, Loss: Sadness and Depression (1980), turned to bereavement, much of it drawn from research on widows and widowers, including the work of his colleague Colin Murray Parkes. Bowlby traced a path through grief that ran from numbness and disbelief, through yearning and searching for the lost person, through disorganization and despair, toward some reorganization of life. He linked the adult forms of failed mourning to the childhood sequence of Volume 2: protest that is never expressed becomes displaced anger, despair that does not lift becomes chronic mourning, and detachment that persists becomes the defensive shutting-out of feeling. He argued that the way a person mourns as an adult is shaped by the separations they lived through as a child. He also replaced the psychoanalytic idea of repression with what he called defensive exclusion: the mind keeps certain painful information out of awareness, which is, in the end, a description of where attention is not allowed to go.

The book was twenty years in the making, and it opened on an old wound. In 1960 Bowlby had published "Grief and Mourning in Infancy and Early Childhood" in The Psychoanalytic Study of the Child, arguing that the despair of a toddler separated from its mother was mourning in the full sense, the same process an adult goes through after a death. The same issue carried replies from Anna Freud, Max Schur, and René Spitz, all of whom rejected it. The orthodox position was that a small child's ego was too immature to mourn; what looked like grief was something more primitive, and it passed. The first chapter of Loss, "The Trauma of Loss," goes back over that dispute under the heading "Do young children mourn? a controversy," and the rest of the book is, among other things, his answer.

The book is built in three parts: a theoretical opening, the mourning of adults, and the mourning of children. For adults, the core evidence came from Colin Murray Parkes's study of London widows in the 1960s and from the Harvard Bereavement Study, in which Parkes worked with Ira Glick and Robert Weiss on widows and widowers in Boston. From these Bowlby set out four phases of mourning. Numbing comes first and lasts from a few hours to a week, a stunned unreality broken by bursts of intense distress or anger. Then comes yearning and searching: the bereaved person scans crowds for the lost face, hears the key in the door, keeps a place at the table, and is often furious with doctors, relatives, or the dead person themselves. Bowlby read this searching as the attachment system doing exactly what it was built to do, trying to recover a figure who has gone missing, except that this time the figure cannot be found. Disorganization and despair follow once the search fails, and only then, gradually, reorganization, in which the person rebuilds a picture of the world and of themselves that the loss no longer contradicts. He insisted the phases blur and overlap, and that a person swings back and forth between them for a long time.

One of his observations ran against the clinical advice of the day. Many widows in the studies reported a continuing sense of their husband's presence, talked to him, or consulted him over decisions. The prevailing view was that healthy grieving required letting go of the dead. Bowlby argued that a continuing relationship with the dead person, held in a way that does not deny the death, is common in healthy mourning and is often part of how reorganization happens. It is one of the ideas from Loss that later grief research took up most directly.

The adult section also includes chapters on the loss of a child, including stillbirth and the death of a newborn, on mourning customs in other cultures, and on what he called the disordered variants. These fall into two broad patterns. In chronic mourning, the grief does not lift; the person stays fixed on the loss for years, often with persistent anger and self-reproach. In what he called prolonged absence of conscious grieving, the person seems to carry on as before, busy and composed, until the unexpressed grief surfaces later as depression, illness, or a sudden collapse, sometimes on an anniversary. He then asked who is most at risk, and his answer pointed back to childhood: people whose attachments had been anxious and clinging, people given to compulsive caregiving, who look after others in place of being looked after, and people who had built a show of fierce independence. All three, he argued, carried patterns laid down by earlier separations and losses.

The chapter "Sadness, Depression and Depressive Disorder" draws the line the book's title implies. Sadness, for Bowlby, is the normal and even healthy response to loss, part of the work of reorganizing. Depression in the clinical sense is what happens when that work gets stuck and the person comes to feel helpless to rebuild anything. Here he leaned on George Brown and Tirril Harris's study of women in Camberwell, Social Origins of Depression (1978), which found that losing a mother before the age of eleven made a woman markedly more vulnerable to depression in adulthood when later losses came. He named three kinds of childhood experience he thought set a person up for depression: the death of a parent, repeated failed attempts to win a stable and secure relationship with parents, and being told, explicitly or implicitly, that one is unlovable or incompetent.

The children's section turns on a practical claim. Young children can and do mourn healthily, Bowlby argued, provided three things are in place: they are told the truth about the death promptly and in terms they can grasp, they are allowed to ask questions and to grieve openly alongside the surviving adults, and they have a trusted person who stays. He gave cases where these conditions held, including two four-year-olds whose mourning ran its course, and cases where they did not: children told that a dead parent had "gone away" or would come back, children kept from the funeral, and, most disturbing, children who had witnessed or discovered a parent's suicide and were then told it had been an accident or had not happened. In those cases, he argued, the child is caught between what they know and what they are allowed to know, and the conflict can shape the rest of their life.

The theoretical spine of the book is chapter four, "An Information Processing Approach to Defence." Bowlby wanted to replace the Freudian machinery of drives and repression with something closer to the cognitive psychology of the 1970s. He drew on experiments in subliminal perception and "perceptual defence," which suggested that the mind assesses incoming information before it reaches awareness and can screen some of it out. Defensive exclusion, in his account, is that screening applied to attachment: information that would activate the attachment system, such as a loss, a rejection, or a parent's cruelty, is kept from further processing because processing it would hurt too much or has been forbidden. The cost is that the excluded material keeps operating outside awareness. A person can end up with two incompatible models of the same parent, the loving one they were told about and the frightening one they experienced, held in separate systems that never meet, a split he discussed under the heading "Self or selves." It is the closest the trilogy comes to a theory of attention: grief, on this account, is resolved only when attention is allowed to go where it has been kept from going.

Academic psychologists criticized the trilogy for resting on case material, and the psychoanalytic establishment largely ostracized Bowlby for decades. Attachment theory has since become the dominant framework for understanding early social development.

(writing)