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

When Love Became a Research Subject

On August 31, 1958, Harry Harlow of the University of Wisconsin delivered his presidential address to the American Psychological Association in Washington. The title was "The Nature of Love." He opened by conceding that some considered love too intimate and personal a subject for experimental research. Then he presented the experiments.

The work had not started as a study of love. Harlow's lab in Madison had been separating rhesus monkeys from their mothers a few hours after birth and raising them by hand, using techniques developed at Yale, to produce healthy, disease-free animals for learning experiments. The bottle-fed infants were in some ways healthier and heavier than mother-reared ones. But the staff noticed something else. The infants clung fiercely to the folded gauze diapers that covered the wire-mesh floors of their cages, and when the pads were removed once a day for cleaning, they threw violent tantrums. Harlow compared it to a human child's attachment to a blanket, a pillow, or a stuffed toy. The lab had also found that an infant monkey raised on a bare wire floor survived the first five days of life with difficulty, if at all.

This ran against the prevailing view. Behaviorists and psychoanalysts alike held that an infant's love for its mother was secondary: the mother feeds the child, hunger is relieved, and affection attaches itself to the source of food. The behaviorist John B. Watson had gone further and warned parents in the 1920s that hugging and kissing children would make them weak and dependent. Love, in this account, was a by-product of the bottle.

To separate food from comfort, Harlow built two kinds of surrogate mothers. The first was a block of wood covered in sponge rubber and sheathed in tan terry cloth, with a light bulb behind it radiating warmth. Harlow described her in his address as soft, warm, and tender, a mother with infinite patience who was available twenty-four hours a day and never scolded. The second had the same shape but was made of bare wire mesh. Eight newborn monkeys were each given a cage with access to both. For four of them, the feeding bottle was mounted on the cloth mother; for the other four, on the wire mother.

Within a few days the pattern was clear. Regardless of which mother held the milk, the infants spent nearly all their time on the cloth mother, by most accounts seventeen to eighteen hours a day. Those fed by the wire mother went to her only to nurse, then returned to the cloth one. Harlow named what the cloth mother provided "contact comfort," and argued it was a need as basic as food.

He then tested what the cloth mother was for. In the fear test, a mechanical teddy bear the size of the infant marched toward it beating a drum. Whichever mother had fed them, the infants ran to the cloth mother and clung to her. After a few moments of rubbing against her, many turned to look at the toy, and some left her to approach and explore it. Infants raised with only a wire mother did not seek her out; they clutched themselves, rocked, and cried.

In the open-field test, infants were placed in an unfamiliar room filled with blocks, blankets, and containers. With the cloth mother present, they used her as a base, venturing out to explore and returning to touch her. Without her, they froze, crouched, sucked their thumbs, and cried. Harlow's conclusion was that attachment rests on the comfort of contact, not only on the satisfaction of need, and that what the infant gains from that contact is the security to explore.

The cloth mother had limits Harlow did not foresee in 1958. The monkeys raised on her grew up socially impaired: they could not relate normally to other monkeys, and many of the females later neglected or abused their own infants. In the 1960s Harlow moved from studying the presence of love to studying its absence. Some infants were kept in total isolation for up to two years; others were placed in a narrow stainless-steel chamber he himself named the "pit of despair." They emerged severely disturbed. Even at the time, some colleagues considered the work cruel. Some researchers, including former students, later cited these experiments as one of the factors behind the rise of the animal liberation movement in the United States.

Harlow's results also supplied evidence for the British psychiatrist John Bowlby, with whom he corresponded, and who was then arguing that children separated from their caregivers in hospitals and orphanages suffered lasting harm. Bowlby's attachment theory, and the end of the contact-free nursery, drew on what happened in Madison.

Adult Love

Seventeen years later, adult love entered the laboratory and met immediate opposition. In March 1975, Wisconsin Senator William Proxmire issued his first Golden Fleece Award, naming research he considered a waste of taxpayer money. The recipients were Ellen Berscheid and Elaine Hatfield of the University of Minnesota: the National Science Foundation had granted them $84,000 to study why people fall in love. Proxmire's reasoning was that love is something science cannot understand and should not try to. The funding was later rescinded. Hatfield did not stop. In 1986 she and Susan Sprecher developed the Passionate Love Scale, the questionnaire used to score participants in the Leiden experiment discussed later.

By then the two had already spent a decade on attraction experiments. Hatfield was still Elaine Walster at the time. In the fall of 1966, the Welcome Week handbook distributed to incoming University of Minnesota freshmen listed a "Computer Dance." It was not a regular event. It had been set up for the experiment. 752 freshmen bought tickets and filled out questionnaires, believing a computer would match them by similarity. In fact the pairing was random, with one constraint: no man was paired with a taller woman. The four "staff" selling tickets were hired sophomores whose real job was to rate each buyer's physical attractiveness. At intermission, participants were asked to evaluate their dates. Personality tests, scholastic aptitude scores, and high school rank predicted nothing about who liked whom. The only significant factor was the date's looks.

In 1972, Karen Dion, Berscheid, and Walster published "What Is Beautiful Is Good." Participants viewed photographs of people of varying attractiveness and judged the better-looking ones to have better personalities and better prospects. In 1973, Walster and colleagues ran five experiments testing the folk wisdom about playing hard to get. The woman male participants most wanted to date was neither the one warm to everyone nor the one cold to everyone, but the one warm to him alone and cold to every other man.

The theory most directly concerned with love itself came next. In 1974, Berscheid and Walster applied the two-factor theory of emotion to romantic love. The two-factor theory of emotion holds that an emotion consists of two parts: the body's physiological arousal and the mind's interpretation of that arousal. By this logic, a person whose heart is racing who happens to meet someone attractive may read that arousal as love, even if its real source is fear.

The theory came from Stanley Schachter and Jerome Singer at Columbia University in 1962. Participants were told they would receive an injection of a vitamin called "Suproxin" to test its effect on vision. In fact they received either epinephrine or a saline placebo. Epinephrine raises the heart rate, makes the hands tremble, and flushes the face. One group was told about these effects accurately. One group was deliberately misinformed and told to expect numb feet, itching, and a headache. A third group was told nothing.

After the injection, each participant was led into a waiting room where another "participant" already sat, in fact a confederate working for the researchers. In one condition, the confederate was euphoric: he crumpled scrap paper into balls and shot baskets, and played with a hula hoop. In the other, the two filled out an increasingly insulting questionnaire together while the confederate grew angrier with each page. Those who had no correct explanation for their racing hearts were more easily swept along. The misinformed group was the most euphoric, the uninformed group the angriest, and the group that knew the drug's effects the calmest. Schachter and Singer concluded that the body supplies only undirected agitation; people search their surroundings for cues and label it "happiness" or "anger."

The experiment proved hard to replicate. In 1979, Gary Marshall and Philip Zimbardo repeated the euphoria condition and did not obtain the same result. The same year, Christina Maslach used hypnosis instead of injection to induce arousal and found that unexplained arousal tended to turn into negative emotion, even when the person nearby was cheerful. Other scholars reexamined the original data and pointed out that many key comparisons in the 1962 paper were not statistically significant. The two-factor theory was enormously influential as an idea, but its founding experimental evidence was weak.

The same year Berscheid and Walster carried the idea into love, Donald Dutton and Arthur Aron, two postdoctoral researchers at the University of British Columbia, carried the experiment outdoors. The site was the Capilano Canyon Suspension Bridge in North Vancouver: five feet wide and 450 feet long, wooden boards laid across wire cables, tilting and swaying underfoot, with low cable handrails and a 230-foot drop to rocks and rapids below. The control bridge stood upriver, built of heavy cedar, only ten feet above a shallow stream, with high handrails. It did not sway.

A young female interviewer stopped men between 18 and 35 who were crossing alone and asked them to fill out a questionnaire and write a short story about an ambiguous picture. When they finished, she wrote her name and phone number on a slip of paper and said they could call if they wanted to hear more about the study. On the suspension bridge, 18 of 23 men took the number and 9 later called. On the solid bridge, 16 of 22 took it and only 2 called. The stories written on the suspension bridge also scored significantly higher for sexual imagery. When the interviewer was a man, the difference between the two bridges disappeared. In a later interview, Aron put the lesson plainly: when you feel an intense attraction to someone, do not assume that person is right for you.

The study is a textbook classic, but it has problems. The samples were small, and the men being compared had chosen their own bridge; those willing to cross the suspension bridge may have been more inclined to take risks to begin with. In 2012, the Polish psychologist Szczucka reexamined the original data and identified statistical errors.

Later research changed the understanding of the mechanism. In 1981, Gregory White and colleagues had male students run in place for either 15 or 120 seconds, then watch a video of a woman introducing herself. She had been made up to look either attractive or unattractive. The men who had run for 120 seconds liked the attractive version more and the unattractive version less. Arousal did not simply increase liking; it amplified the judgment already being formed. In 1998, Craig Foster and colleagues pooled 33 experiments with 1,232 participants and found the same pattern. They also found that arousal affected attraction even when people knew perfectly well their heart was pounding from exercise, though the effect was stronger when the source was ambiguous. Misreading, in other words, may not be the whole explanation. Arousal may amplify the impression a person makes before any interpretation takes place.

In 2003, Cindy Meston and Penny Frohlich of the University of Texas ran a variation at the exits of amusement park roller coasters, in a paper titled "Love at First Fright." People who had just gotten off the ride rated a photograph of an opposite-sex stranger as more attractive than people still waiting in line. But if the person seated beside them on the ride was their own partner, the effect disappeared. Arousal needs somewhere to go, and where it lands depends on who is standing in one's attention at that moment.

Different Kinds of Love

The common account says the ancient Greeks divided love into four kinds, each with its own word: eros for erotic love, philia for the love between friends, storge for the affection within families, and agape for love that asks nothing in return. All four words do exist in Greek. But the people who arranged them into a neat taxonomy were not the ancient Greeks.

In 416 BCE, the tragic poet Agathon won the prize at the Athenian dramatic festival and the next day hosted a banquet at his house. Plato set his Symposium there: the guests agree to drink lightly and take turns delivering speeches in praise of Eros, the god of love. Socrates relays what he says he learned from a priestess named Diotima. Eros is a longing for what one lacks. It begins with love for one beautiful body and climbs, step by step, to love of all beautiful bodies, of beautiful souls, of beautiful knowledge, and finally of beauty itself. For Plato, eros is not merely desire. It is a force that leads a person toward philosophy.

Aristotle devoted two books of the Nicomachean Ethics to philia. The word covers far more than friendship: it includes the bonds between parents and children, husbands and wives, and fellow citizens. He divided it into three kinds: friendship for utility, friendship for pleasure, and friendship based on admiration of the other's character. The first two love what the other person provides and dissolve when circumstances change. Only the third loves the other person for who they are, and it lasts longest. Storge appears rarely in classical texts and usually refers to the natural affection of parents for their children.

Agape was an unremarkable word in classical Greek. Translation and religion made it important. In the last centuries BCE, Jewish scholars in Alexandria translated the Hebrew Bible into Greek, producing the Septuagint, which used agape for the love between a man and a woman in the Song of Songs, a love that did not exclude desire. The authors of the New Testament then used the word extensively for God's love of humanity and for love of one's neighbor; the "love is patient" of 1 Corinthians 13 is agape. Its meaning of unconditional, self-giving love is a product of Christian theology.

The arrangement of these words into opposing categories came from two twentieth-century books. In the 1930s, the Swedish theologian Anders Nygren published Agape and Eros, which set the two against each other: agape as God's downward, unmotivated giving; eros as humanity's upward, self-centered longing. In 1960, C. S. Lewis published The Four Loves, placing affection, friendship, eros, and charity side by side and matching them to the four Greek words. The "four Greek words for love" circulating today are largely Lewis's version. Several of the modern psychological taxonomies that follow borrowed this vocabulary directly.

In 1973, John Alan Lee, a sociologist at the University of Toronto, published Colours of Love. He first read widely in the literature and philosophy of love, then asked more than two hundred people to describe in detail one romantic relationship they had lived through. From these accounts he derived six "styles of loving," which he named with Greek and Latin words.

He borrowed the metaphor of the color wheel: three primary colors mixing into three secondary ones. The first primary is eros: love at first sight, strong attention to physical appearance, an urge toward quick intimacy. The second is ludus: love as a game, no plans for the future, the possibility of seeing several people at once, and wariness of the other's dependence. The third is storge: love that grows slowly out of friendship, calm and without strong emotion. The three secondaries are mania, a mix of eros and ludus, possessive and emotionally volatile, close to what Tennov called limerence; pragma, a mix of ludus and storge, like choosing a partner with a checklist, weighing background, education, and suitability; and agape, a mix of eros and storge, placing the other's good before one's own.

Lee stressed that the six styles describe a relationship, not a fixed personality. The same person may love in different colors with different partners. Nor did he rank them. His view was that many relationships fail not because one person loves too little, but because the two are loving in different colors. Pair a ludus with a mania, and the more relaxed one becomes, the more anxious the other grows.

In 1986, Clyde and Susan Hendrick, psychologists at Texas Tech University, turned Lee's qualitative typology into a 42-item questionnaire, the Love Attitudes Scale. Factor analysis yielded six clear dimensions with low correlations among them, broadly matching Lee's scheme. The scale has since been used in more than a dozen countries. A common finding is that men tend to score higher than women on ludus, and that ludus correlates negatively with relationship satisfaction.

Hatfield's distinction is the simplest. In A New Look at Love (1978), written with her husband William Walster, she divided adult love into two kinds. Passionate love is a state of intense longing for union with another. Reciprocated, it brings fulfillment and ecstasy; unreciprocated, emptiness, anxiety, and despair. Companionate love is the affection and tenderness felt for someone with whom one's life is deeply intertwined.

To make the first measurable, Hatfield and Susan Sprecher published the Passionate Love Scale in 1986. It breaks passionate love into three dimensions. The cognitive: intrusive thoughts, idealization of the partner and the relationship, the desire to know and be known. The emotional and physical: attraction, longing, and arousal such as a racing heart. The behavioral: doing things for the other, finding ways to stay close. The scale remains one of the most widely used instruments in the field; the Leiden and New York experiments discussed later used it to score participants.

The distinction carries a time dimension. Passionate love tends to concentrate at the start of a relationship, arriving fast and fading easily; companionate love grows slowly but holds steadier. What people call "the honeymoon being over" is the stage when the first has receded and the second has not yet matured. Langeslag later found that the scale actually measured both kinds of love at once, and developed separate scales for infatuation and attachment.

In 1986, Robert Sternberg of Yale University published "A Triangular Theory of Love" in Psychological Review. The triangular theory of love divides love into three components, intimacy, passion, and commitment, whose combinations produce different kinds of love. Intimacy is the feeling of closeness, connection, and bondedness. Passion is the drive toward romance, physical attraction, and sex. Commitment has two layers: in the short term, the decision that one loves this person; in the long term, the decision to maintain that love. Sternberg described them by temperature: intimacy is warm, passion hot, commitment cold.

The presence or absence of each component yields eight cases. None of the three is nonlove. Intimacy alone is liking, or friendship. Passion alone is infatuation; love at first sight belongs here. Commitment alone is empty love, common in long marriages where passion and intimacy have gone, and at the start of arranged marriages. Intimacy plus passion is romantic love; intimacy plus commitment is companionate love. Passion plus commitment without intimacy he called fatuous love: the marriage after a few weeks' acquaintance. All three together form consummate love.

The triangle metaphor does two further things. First, the triangle's size represents the amount of love and its shape the balance among the components; each partner has a triangle, and the more they overlap, the more satisfying the relationship. Second, the three components follow different curves over time: passion rises and falls quickly, intimacy accumulates slowly, and commitment grows as the relationship takes shape. The same relationship is therefore a different triangle in different years. Sternberg later proposed that love is a story: each person carries a script of love learned from experience and culture, and measures the relationship at hand against it. Some scholars have noted inconsistencies in the theory between how the three components are defined, how they are measured, and how they change over time.

In 1998, Fisher proposed in the journal Human Nature that mammals have three distinct but interrelated systems for mating and reproduction. The first is lust, the craving for sexual gratification, associated mainly with testosterone and estrogen. The second is attraction, which in humans is romantic love. She described it as heightened energy, attention focused on a specific partner, obsessive following, and possessiveness, associated with elevated dopamine and norepinephrine and lowered serotonin. The third is attachment, which brings calm, security, and a sense of union with a long-term partner, associated mainly with oxytocin and vasopressin.

In her evolutionary account, each system handles one stage of reproduction. Lust sends a person looking for any possible partner; attraction concentrates limited time and energy on one; attachment keeps two people together long enough to raise a child. Her central claim is that over the course of human evolution the three systems became increasingly independent of one another. A person can be deeply attached to a spouse, attracted to someone else, and sexually drawn to a third. Fisher uses this separation to explain marriage, adultery, divorce, and remarriage, as well as extremes such as stalking and crimes of passion.

The model was for a long time the dominant framework, but it has been criticized as too simple. The Australian researcher Adam Bode has recently proposed dividing it into five systems, and argues that the core of romantic love is borrowed from the brain machinery of mother-infant attachment. For this essay, though, one feature of Fisher's definitions is worth noting: of her three systems, only attraction is defined in terms of attention.

(writing)