Love at Goon Park: Harry Harlow and the Science of Affection

by Deborah Blum

  When Harry came home at night, he walked the baby up and down in the small Riverside Drive apartment, talking, pacing. Baby Harlow’s nighttime lullaby was a litany of the history of psychology. The cocktail hours with Clara were filled with discussion of wrongheaded science; indeed, “work was the background music of our lives,” one of Harry’s children would later recall. Harry picked his way through Goldstein’s arguments. He was out of patience finally and completely with the rat-psychology view of the world, with the simple brain and simple behaviors, with ignorance and prejudice toward other species. It seemed to him that by dismissing the abilities of other species, in the end, psychologists were dismissing the abilities of their own.

  He knew how deep the counter-arguments ran. It wasn’t just that Romanes and Kohler and other distinguished scientists had failed to persuade. It wasn’t just that Watsonian behaviorism and Pavlovian conditioning were dominant. Scientists had been insisting for centuries that animals were basically brainless. The other species could be conditioned, they could be made to respond; but think, feel, analyze, grieve—never. Back in the 1700s, French philosopher René Descartes had likened animals to machines; animals could never think as humans do, he said. They were soulless creatures, beast machines. That perception held even when Charles Darwin made his evolutionary arguments. Darwin undeniably suggested that humans and other species must share common brain structures and therefore common abilities. It was too much for Goldstein, who responded by dismissing evolution outright. But even those who believed in Darwin often could not quite accept that animals possessed the kind of complex brains that had long been reserved for humans.

  The idea of intelligent animals had a particularly rough time in the United States. One of the most famous books on the subject, Animal Intelligence, published in 1898, basically concluded that animals weren’t intelligent at all. The author, New York psychologist Edward Thorndike, tended to side with Ivan Pavlov. Animals could be trained—or conditioned—to look intelligent. But that, Thorndike said, was misleading. His most famous test involved putting cats into boxes and testing their ability to escape. The boxes were small enough to make the cats feel just a little squeezed, a little antsy to get out. Thorndike provided them with an escape mechanism. The boxes had panels that could be opened when the cats pressed a button or pulled on a string. To reward the cats for opening the panel, Thorndike placed a food treat just outside the box. To strengthen the intensity of that reward, he kept the cats hungry. His experiment involved measuring the length of time it took a cat to break free.

  After some time in the box, the cats would push, bump, and eventually trip the string or step on the button. The next time in captivity, the captives would move more directly to the button. The more often the cat went into the box, the faster it got out. After a few trial runs, some cats were pulling the string almost as soon as the box was closed.

  Some people would call those cats smart. Thorndike concluded almost the opposite. The feline behavior showed no evidence of thought, he said, merely “the accidental success of the animal’s natural impulses.” Thorndike went on to develop “laws” of animal behavior. His Law of Effect came directly out of the cat-in-the-box work. It said this: If a movement is followed by the experience of satisfaction or the removal of annoyance, that movement will be “connected” with the solution. In other words, if the cat pulls the string and the box opens, eventually it will connect string-pulling with boxopening. His second law, The Law of Exercise, said that the more this happens, the stronger the “connection” between action and result. In other words, the cat will become a string-pulling automaton. Thorndike first called this somewhat robotic turn of events “stamping in” behavior. He later came to prefer the word “reinforcing,” a term still used in psychology and animal training today. He considered his laws comparable to the laws of motion and energy in physics, another step toward making living creatures as predictable as clockwork.

  The mechanical animal—incapable of love or reason—obviously fit well with the teachings of early behaviorists such as John B. Watson. But it got an even bigger boost from Harvard-trained researcher Burrhus Frederick Skinner, perhaps the most famous psychologist of Harry’s generation. Known as B. F. Skinner to most of the world and as Fred to his friends, Skinner was adamant in his belief that animals do not have feelings. He was appalled once when, watching a squirrel gobble a nut, a friend remarked that the animal “liked” the acorn. Of course it didn’t, Skinner replied. Animals don’t like things; liking is an emotion, and squirrels don’t have those. Skinner described himself as a neobehaviorist, a builder of a more sophisticated version of the earlier science.

  In pursuit of that ideal, Skinner created a device that became known far and wide as “the Skinner box,” an updated version of Thorndike’s apparatus. The square box was soundproofed and equipped with a bar or lever. If a rat pushed the lever, a food pellet tumbled out. If a pigeon pecked the bar, it, too, received food. The rodents and birds pushed and they pecked and they ate, just as Skinner had predicted, in the most convincing way. During World War II, Skinner was able to use his box to train pigeons to peck at a target. He tried, unsuccessfully, to persuade the U.S. Army that the birds could be put into the nosecones of missiles and used to guide weapons. Of course, if the food-delivery mechanism jammed—which it sometimes did—the pigeons rapidly lost interest in pressing the bar or pecking the target. After all, what was the point of thumping on cue if no food came out? From a scientific point of view, Skinner appreciated this reluctance. It was, as he pointed out, a classic Pavlovian extinction curve. But Pavlov’s beautiful calculation of vanishing behavior made army officials doubt the reliability of pigeons as bomb-delivery systems.

  Harry Harlow was not a fan of the Skinner box. It tended to inspire him to sarcasm. “There is no other learning technique that ever did so much for the pigeon,” was his summary. It was a relief to learn, he said, that if his brain dwindled to pigeon-like dimensions, he could still be conditioned. Meanwhile, a pretty good pun occurred to him: “It is nice to know that it takes little brain to learn or think, and as I grow progressively older, I am enormously reinforced by this discovery.”

  It wasn’t that Harry denied the veracity of such experiments. He’d done classical conditioning studies. He knew they worked. He just didn’t think that such responses were everything: “Our emotional, personal and intellectual characteristics are not the mere algebraic summation of a near infinity of stimulus-response bonds.” He had his allies, especially among the young skeptics like himself. Harry was sharing his ideas with a new band of scientists that included the Canadian psychologist Donald O. Hebb, renowned today for his farsighted theories about how experience influences the brain. Back in the 1940s, Harlow and Hebb were, at best, promising outsiders. In exasperation, Hebb once declared that it would be better to be wrong about how the brain worked—to stand behind some real ideas—than to be as vague and inconclusive as psychology was at the moment.

  Measurable behavior—not the black box of the brain—was still in fashion. Thorndike and Skinner were mainstream. The most important psychology theorist of the time, Clark Hull of Yale University, was building an entire system of behavioral predictions based on the concept that stimulus and response were defining characteristics. People listened to Hull. He was a soft-voiced, articulate, and dedicated scientist, liked and respected by many of his peers. According to one analysis, 70 percent of all studies dealing with learning and motivation during the 1940s cited one or more of Hull’s books and papers.

  Hull’s theory was an almost geometric structure, crystalline in its sharply defined architecture. Often called “drive reduction theory,” it was based on the idea that behavior is created by drives or needs that we seek to satisfy or reduce. Hull’s overall concept had seventeen corollaries, seventeen postulates, and assorted theorems, proofs, and formulas. A classic Hullian equation might include stimulus (s) and drive (D) and response (R) and habit strength (sHr) and number of reinforcements (
N) and hours of food deprivation (h). All this might add up to a formula such as

  In other words, the strength of a habit equals hours of food deprivation multiplied by the number of reinforcements (amount of food) and by response to that situation. In other words, a very hungry rat, which receives food when it presses a bar, will develop a very strong bar-pressing habit. The hungrier the rat is, the stronger the habit.

  The message was a familiar one. The bar-pressing rat was a conditioned animal, responding to a hunger drive rather than exhibiting intelligent thought. It was Thorndike and Watson and Skinner all over again. But Hull’s theory seemed to his colleagues to take psychology to the next level. It integrated the experiments, pulled together the results, in a systematic way. It also followed the classical notions of science: putting forth theories that could then be tested. Leading experimental psychologists, such as Kenneth Spence of the University of Iowa, turned their labs and their students over to testing those elaborate calculations. “Hull’s theory was truly scientific and so I became a Hullian,” explains William Verplanck, an emeritus professor of psychology at the University of Tennessee, looking back toward his student years in the mid-twentieth century.

  Did Harry Harlow go for this? Not in the least. Spence and Hull used to exchange exasperated letters about Professor Harlow, who didn’t believe in the theory and appeared, in their opinions, to be needlessly outspoken on the point. Harry didn’t step back. He made sarcastic remarks about people who thought the “Hull truth” was the whole truth. “Harry was no theorist,” Verplanck says. “He was simply a hardheaded empiricist. He just followed his nose and published what he found out.” Over the years, Verplanck—now in his nineties—has come to agree with the Harry Harlow perspective on theorizing, and perhaps even goes beyond it: “I think that theory is the curse of psychology—if we get rid of the theory, we might know something.”

  Actually, Harry was less troubled by the theories in general than this one in particular. He thought Hull’s idea depended on studies that made animals look simpler and stupider than they really were. He had come to think that popular experiments, even the Skinner box, achieved little in understanding how an animal’s brain handles complex situations. Did psychologists really believe that they could define behavior with the discovery of the “empty organism,” simply managed by simple training techniques? Surely there was more to the brain—and to us—than that?

  It wasn’t that Harry opposed testing for animals’ abilities. It wasn’t even that he was opposed to building devices that could be used to look for a particular behavior. He was—like Skinner or Watson or Thorndike—a dedicated experimentalist. He believed in the power of evidence gathered in the laboratory. But he also believed that too many psychologists were setting artificial limits on their subjects. How much intelligence did it really take to press a bar or to push a button? One of the standard mazes of the time was a T-shape. A rat could hurry down a long straight arm and then turn either right or left. How much of a challenge was that?, Harry asked. How much did one really learn from watching a rat run forward? Was there value in proving that rats can move fast? A device was needed that would really challenge animals to think. He didn’t want a one-time challenge such as Kohler’s box-and-banana problem. He wanted a systematic way to push monkeys to achieve, beyond the techniques of the scientific community. He wanted to take the way Goldstein tested his human patients and apply those standards to monkey intelligence.

  When he and Clara and Robert returned to Madison, they had two different projects planned, one personal and one professional. The Harlows were going to build a new house. Clara was going to oversee that project. And Harry was going to hasten over to his shabby laboratory and design a thought-provoking device, one that met all the scientific specifications. It’s worth noting that the first paper out of Harry’s lab, describing a “Test Apparatus for Monkeys,” was recommended for publication by that device-loving psychologist, B. F. Skinner himself.

  Eventually, the test apparatus was formally named the Wisconsin General Test Apparatus, or WGTA. In primate research, it became a genuinely famous design. Copies of the WGTA can still be found, mechanized and modernized, at primate centers around the world. When one of Harry’s later graduate students, Allan Schrier, took a job at Brown University, he was still proud of having worked with the original. Schrier purchased vanity plates for his 1966 Volkswagen bug that said 66WGTA. When he bought a new car two years later, he updated his license plates to 68WGTA. The older plates went to Harry with a note: “Rhode Island recognizes a good apparatus when it sees one.”

  Here’s how the first WGTA worked: There was a cube-shaped cage, two feet in all dimensions, with a solid oak floor, three-quarters of an inch thick (hardwood floors were cheap in those days). A sliding panel that could be raised and lowered with a rope-and-pulley system was fitted to one side of the cage. A monkey would sit in the cage, waiting, presumably a little curious about what was going to happen. When the panel rolled up, the animal could see a table edged with brass rods. The researchers could slide trays onto the rods and ease them right up to the monkey. The trays were packed with test objects and treats. The monkey could reach through the bars of the cage to grab, discard, and puzzle over test objects; and, of course, pick up treats. When a monkey had worked through one set of challenges, the scientist could replace the tray. At the opposite end of the table from the monkey was a small observation post. There, tucked behind a one-way screen, the scientist could watch the monkey without being seen himself. The WGTA was a good design and that was one of the few things that B. F. Skinner and Harry Harlow agreed upon.

  What made the WGTA look brilliant was something else. Harry still didn’t have enough monkeys in his lab. There were no domestic breeding colonies. Monkeys were hard to find, expensive, and often, after being trapped and shipped in less-than-nurturing conditions, half-dead when they arrived. He considered his few dozen healthy animals solid primate gold. He hoarded them. Out of simple preservation, he had to throw out the standard rules of animal testing, which were based on rodent work. Rat research worked on the principle of unlimited supply. When psychologists were testing conditioned responses, they often wanted inexperienced rats for each study. If an animal was already conditioned in one experiment, it was hard to separate the effect for the next study. So rats were rarely recycled. Harry once described the standard psychology experiment as a Blitzkrieg involving forty-eight rats: “The controls are perfect, the results are important, and the rats are dead.”

  Harry Harlow might have taken the same approach if he’d had a similar river of monkeys flowing through his lab. But he had only a small pool, one he couldn’t afford to drain. He was a psychologist who had a finite number of test subjects and an infinite number of WGTA tests that he wanted to conduct. One forty-eight-monkey “do and die” study would have left him with a lab full of empty cages. He never even considered it. So instead of conducting a Blitzkrieg, he rotated his monkeys, four at a time, through those countless studies. Over and over again, first one problem set, then something harder, then something harder yet. Unlike the rats—and even the cats in previous studies, the monkeys couldn’t avoid building on previous experience. The result was that the WGTA didn’t just make monkeys look smart, it made them look like small geniuses. To the surprise of everyone, including the Wisconsin psychologists running the tests, the monkeys began making educated decisions, fast and savvy. “Had we run many monkeys on just a few problems, as was the custom of the rodentologists with rats, we never would have realized that animals could learn how to learn,” Harry said.

  That gave him an opportunity, which he took, to argue against the rapid killing of research animals. Harry reminded his fellow psychologists that they could benefit by keeping their animals alive: Even if you did believe that all human behavior could be worked out in rats, he said, the “do or die” design of standard experiments was just flawed. It couldn’t explain people very well because people don’t generally do one task at on
e time and then fall dead. The practice of psychology might open up a new understanding of very short-lived rodents, he suggested, but it didn’t do much for anyone, human or other animal, who lived long enough to gain a little experience.

  The scarcity and cost of monkeys also forced Harry to rely on the cheapest primate possible, the abundant Indian rhesus macaque. That, too, turned out to be an extraordinary piece of good luck. Rhesus macaques are not the most beautiful of monkeys. They lack the elegant ballet dancer build of a squirrel monkey or the endearing fluffiness of a titi monkey. Rather, rhesus macaques look competent and tough, steelworker monkeys on their way to the factory. Wiry golden gray fur frames a squared face dominated by a long snout and a pair of close-set eyes, coffee-dark and alert. Primarily forest dwellers, they can live almost anywhere. They have the rare ability to take advantage of wherever they land; they skitter through city streets, colonize old temples, raid farms, whatever it takes.

  Rhesus macaques thus are resilient, adaptable, and, perhaps even more important to researchers, accessible. Scattered across such a range of habitats, they are easy to find and therefore an ideal research monkey from the collecting point of view. Even today, rhesus macaques are the most used monkey in research. Modern medicine owes them much. The Rh factor in blood (positive, negative) was worked out in these monkeys. The Rh, in fact, comes directly from the term Rhesus. During the desperate 1950s race to find a vaccine for polio, scientists tested rhesus macaques by the boatload and by the cargo plane–full. So many monkeys were scooped out of India—well over a million—that the country banned their export in the early 1960s, fearing soon none would be left.