by Morton Hunt
All we need to know or can know, he said, are the external causes of behavior and the observable results of that behavior; these will yield “a comprehensive picture of the organism as a behaving system.”
Consonant with that view, he was a rigorous determinist: “We are what we are because of our history. We like to believe we can choose, we can act… [but] I don’t believe a person is either free or responsible.” The “autonomous” human being is an illusion; the good person is one who has been conditioned to behave that way, and the good society would be one based on “behavioral engineering”—the scientific control of behavior through methods of positive reinforcement.52
Skinner was a deft showman and popularizer; he was fluent, lucid, unabashedly egotistic, and charming. To demonstrate the power of his own technique of conditioning, he taught a pigeon to peck out a tune on a toy piano, and a pair of pigeons to play a kind of table tennis in which they rolled a ball back and forth with their beaks; millions who have seen these performances on TV documentaries think of Skinner as a Svengali, at least of animals. He presented his vision of the ideal, scientifically controlled society in the form of a utopian novel, Walden Two (1948), picturing a small society in which, from birth onward, children are rigorously conditioned by rewards (positive reinforcement) to be cooperative and sociable; all behavior is controlled, but for the good and the happiness of all. Despite wooden dialogue and a labored plot, it became a cult book and perennial favorite with undergraduates, and has sold well over two million copies.
But his fame with the public was greater than his standing with fellow professionals. As one admirer, the psychologist Norman Guttman, wrote in The American Psychologist some years ago:
[Skinner is] the leading figure in a myth… [the] scientist-hero, the Promethean fire-bringer, the master technologist… [the] chief iconoclast, the image-breaker who liberates our thoughts from ancient restrictions.53
Skinner was born in 1904 in a small Pennsylvania railroad town, where his father was a lawyer. As a boy, he had a great aptitude for building Rube Goldberg contraptions; later, as a psychologist, he would invent and build remarkably effective apparatuses for animal experimentation. In school and college he aspired to become a writer, and after college spent a year, much of it in Greenwich Village, trying to write. Although he closely observed the manifold forms of human behavior all around him, he discovered after a while that he had nothing to say about what he saw and, deeply dejected, gave up the effort.
But he soon found another and, for him, more practicable way to understand human behavior. In his reading he came across discussions of Watson’s and Pavlov’s work, read books by each of them, and decided that his future lay in a scientific approach to human behavior, particularly the study of conditioning. “I was very bitter about my failure in literature,” he told an interviewer in 1977, “and I was sure that writers never really understood anything. And that was why I turned to psychology.”54
He proceeded to Harvard. Introspective psychology reigned there, but he was no longer interested in what he called “the inside story,” and quietly went his own way, doing behaviorist research with rats. In his autobiography he recalls with pleasure having been something of a bad boy: “They may have thought that someone in psychology was keeping an eye on me, but the fact was that I was doing exactly as I pleased.”55 Despite the teachings of his professors, he became ever more thoroughly behaviorist, and at his dissertation examination, when asked to name some objections to behaviorism, he could not think of one.
Making good use of his mechanical aptitude, he constructed a puzzle box that was a great improvement over the Thorndike model; widely used ever since, it is known as the Skinner box. In its basic form—it has many models—it is a cage, large enough to comfortably accommodate a white rat, with a horizontal bar on one wall just above a little food tray and a water spout. When the rat, prowling about the cage, happens to rest its forepaws on the bar, pressing it down, a food pellet automatically drops into the tray. Connected equipment outside the cage automatically records the behavior by drawing a line showing the cumulative number of bar pressings minute by minute. This was a much more efficient way of gathering data than Thorndike’s puzzle box procedure, since the experimenter did not have to observe the rat or deliver the food when it pressed the bar but merely look at the record.
The box also yielded more objective data on the acquisition or extinction of behavior than anyone had gathered thus far. The rat, and it alone, determined how much time elapsed between one pressing of the bar and the next. Skinner could base his findings of learning principles on the “response rate”—the rate at which the animal’s behavior changed in response to reinforcement—uncontaminated by the experimenter’s actions.
Moreover, Skinner could program the box in ways that approximated many circumstances in the real world that either reinforce or fail to reinforce behavior. He could, for instance, study the learning of a response when it is regularly rewarded; the extinction of a learned response when the reward is abruptly discontinued; the effects on learning and on extinction when rewards are delivered intermittently at regular intervals (say, every fourth bar pressing); the effects when rewards are delivered intermittently at irregular intervals; the effects of mixed results of bar pressing (such as a reward coupled with an electric shock); and so on. In each case, the data yielded a curve showing the rate of acquisition or extinction of a behavior under those particular circumstances.
From these curves, Skinner formulated a number of principles that cast light on the behavior of rats—and human beings. An example is his discovery of an important variation of the partial reinforcement effect. After rats had been trained on a schedule in which food pellets were delivered only once in a while and at irregular intervals, the rats would persist in their bar pressing even if the food-dispensing apparatus was turned off altogether. Their learned behavior was more resistant to extinction than that of rats trained to intermittent but regular reinforcement.56 This has been likened by some to the behavior of a slot machine player in a casino: neither the rat nor the gambler has any way of predicting when the next reinforcement is to come, but, being accustomed to occasional rewards, will hang on in the hope of getting one on the next try.57
Skinner’s most important contribution, however, was his concept of “operant conditioning”; for this alone he merits a permanent place in the Hall of Fame of psychology.
In “classical” (Pavlov’s) conditioning, the animal’s unconditioned response (salivating) to food is made into a conditioned response to a formerly neutral stimulus (the sound of the metronome or bell); the crucial element in the behavior change is the new stimulus.
In “instrumental” (Thorndike’s) conditioning, the crucial element in behavior change is the response, not the stimulus. A neutral response— the accidental stepping on the treadle during random efforts to get to the food—is rewarded and becomes a learned bit of behavior serving an end it formerly did not have.
Skinner’s operant conditioning is an important development in instrumental conditioning. Any random movement the animal makes, for whatever reason, can be thought of as “operating” on the environment in some way and therefore, in Skinner’s terms, is an “operant”; rewarding the movement produces operant conditioning. By rewarding a series of little random movements, one by one, the experimenter can “shape” the behavior of the animal until it acts in ways that were not part of its original or natural repertoire.
Here is how Skinner shaped the behavior of a pigeon to peck at a small colored plastic disk set flush in one wall of the Skinner box:
We first give the bird food when it turns slightly in the direction of the spot [i.e., the disk] from any part of the cage. This increases the frequency of such behavior. We then withhold reinforcement until a slight movement is made toward the spot. This again alters the general distribution of behavior without producing a new unit. We continue by reinforcing positions successively closer to the spot, then by reinforcing
only when the head is moved slightly forward, and finally only when the beak actually makes contact with the spot.
In this way we can build complicated operants which would never appear in the repertoire of the organism otherwise. By reinforcing a series of successive approximations, we bring a rare response to a very high probability in a short time… The total act of turning toward the spot from any point in the box, walking toward it, raising the head, and striking the spot may seem to be a functionally coherent unit of behavior, but it is constructed by a continual process of differential reinforcement from undifferentiated behavior.58
(Other experimenters, using Skinner’s technique, have constructed far more peculiar behaviors. One taught a rabbit to pick up a coin in its mouth and drop it into a piggy bank; another taught a pig named Priscilla to turn on a TV set, pick up dirty clothes and put them in a hamper, and run a vacuum cleaner over the floor.59)
Skinner likened the operant training of his pigeons to a child’s learning to talk, sing, dance, play games, and in time acquire the entire repertoire of adult behavior. All, in his view, is due to the assembling of long chains of behavior out of tiny links of simple behaviors by operant conditioning. One might call it an Erector-set view of the human being (Homo erectorus?)—a mindless robot assembled by operant conditioning from a multitude of meaningless bits.
Skinner was more or less ignored by the psychological establishment for a long while but slowly acquired a number of devotees—enough, finally, to result in the publication of four journals of Skinner behaviorist research and theory and the creation of a special section of Skinner-type studies within the American Psychological Association (Division 25, Experimental Analysis of Behavior, since renamed Behavior Analysis). Skinner boxes and the techniques of operant conditioning have long been widely used by experimental psychologists. In recent years Skinner’s name and work have been cited in hundreds of behavioral science publications each year (though far less often than Freud’s).60
Still, it was outside of mainstream psychology that Skinner had his major impact.
During a visit to his daughter’s fourth-grade class in 1953, it occurred to him that operant techniques similar to those by which he had taught pigeons to play the piano would make for more efficient teaching than traditional methods. Complicated subjects could be broken down into simple steps in a logical sequence; the students would be presented with questions, and immediately told whether their answers were correct. Two principles would be at work here: the knowledge that one has answered correctly is a powerful reinforcer (reward) of behavior; and immediate reinforcement works better than delayed reinforcement. The result is known as “programmed instruction.”
But since one teacher cannot simultaneously provide reinforcement to a roomful of children, new textbooks would have to be written in which questions and answers were presented one by one, each taking a short step toward mastery of the subject and each permitting children to reward themselves immediately by uncovering the answer. Skinner also developed teaching machines for operant self-instruction by comparable means; the mechanical models were a fad for a time, then dropped out of use, but today, computer-based self-instruction with immediate reinforcement is widely used by schools, businesses, and elder-care centers, among others.
For some years the programmed learning movement had a major influence on teaching; courses and course materials designed to teach through operant conditioning were in use in a large proportion of grade schools and colleges in America, and in many schools in dozens of other countries. But eventually educators recognized that the atomistic methods of programmed instruction provide only part of what human beings need; they also need holistic, hierarchical thought structures. And later research showed that in human beings delayed reinforcement often has better results than immediate reinforcement; thinking about one’s responses may lead to more learning than quickly responding and getting an answer.61 Finally, the observation of other people’s behavior, a highly effective form of learning for humans, even if not for cats, involves no immediate reinforcement. Still and all, Skinner’s doctrine of immediate reinforcement has proven useful, is familiar to most teachers, and is incorporated into many curricula and grade school textbooks.
Skinner also had a measurable effect on the treatment of mental and emotional disorders. It occurred to him that a system of tiny rewards for tiny changes from sick acts toward healthy ones might reshape the patient’s behavior. Beginning in the late 1940s, he and two of his graduate students made the first experimental trials of what came to be known as behavior modification. They set up lever-pressing stations at a state mental hospital near Boston; psychotic patients received candy or cigarettes for operating the machines in an orderly fashion. Once that worked, the therapists gave tokens to patients for appropriate behavior, such as voluntarily attending meals, grooming themselves, and helping with housekeeping tasks. The tokens could be exchanged for candy, cigarettes, or privileges like choosing a dining companion, talking to a physician, or watching TV.62
The rewarding of desired behavior in deeply disturbed people often worked. One depressed woman would not eat and was in danger of dying of starvation, but she seemed to enjoy visitors and the TV set, radio, books and magazines, and flowers in her room. The therapists moved her into a room devoid of all these comforts, and put a light meal in front of her; if she ate anything at all, one of the comforts was temporarily restored. The therapists gradually withheld the rewards unless she ate more and more. Her eating improved, she gained weight, and within two months she was released from the hospital. A follow-up eighteen months later found her leading a normal life.63
The behavior modification movement spread to a number of mental hospitals and reform schools. Psychiatrists and psychologists now consider it a useful component of their therapies for severely disordered patients, though a costly one in terms of time and staff effort. Behavior modification is also used by many psychotherapists in the treatment of less severe problems, like smoking, obesity, shyness, tics, and speech problems. It is a specialized technique within the field of behavior therapy, most of which is based on Pavlov-type conditioning rather than on Skinner’s behavior modification.
Skinner’s best-known work, Walden Two, has not remade American society or even part of it, but it undoubtedly has influenced the thinking and social concepts of its millions of readers. Only one effort has been made to create an actual utopia on the Walden Two model: Twin Oaks Community in Louisa, Virginia, a commune founded by eight people in 1967. After surviving many rocky years, it has grown to a population of eighty-five adults and fifteen children. While still modeled administratively on Walden Two, the commune long ago gave up the effort to define ideal behavior and to shape one another’s behavior through methods of Skinner reinforcement.64
Skinner was sometimes self-deprecating about his impact on the world. “In general,” he once said, “my effects on other people have been far less important than my effects on rats and pigeons—or on people as experimental subjects.” That was probably not meant to be taken seriously. What he did mean seriously was the following remark: “I was never in any doubt as to [my work’s] importance.” And he added, on a characteristically perverse note: “When it began to attract attention, I was wary of the effect rather than pleased. Many notes in my files comment on the fact that I have been frightened or depressed by so-called honors. I forgo honors which would take time away from my work or unduly reinforce specific aspects of it.”65
The Impending Paradigm Shift
As behaviorist research accumulated, it became evident to all but the most dedicated adherents of the theory that rats and other laboratory animals frequently acted in ways that the theory could not explain.
For one thing, their behavior often failed to conform to supposedly universal principles of conditioning. “Pigeon, rat, monkey, which is which? It doesn’t matter,” Skinner had written66—but it did matter. Researchers could easily train a pigeon to peck at a disk or a key for food but fo
und it almost impossible to train the bird to flap its wings for the same reward. They could easily teach a rat to press a bar for food but could get a cat to do so only with great difficulty. A rat given sour blue water to drink, followed by a nauseating drug, would thereafter shun sour water but willingly drink blue water; a quail, given the same treatment, would shun blue water but drink sour water. These and scores of comparable findings forced behaviorists to admit that each species has its own built-in circuitry that enables it to learn some things easily and instinctively, others with difficulty, and still others not at all. The laws of learning were far from universally applicable.67
A more serious flaw in behaviorist psychology was that experimental animals kept acting in ways that could not be explained by the neat rate-of-response curves. Many researchers had found, for instance, that at the beginning of an extinction trial an animal would respond to the stimulus with greater vigor than it had during a long series of reinforcements. A rat that had been getting a food pellet each time it pressed a bar would, if no pellet emerged, press the bar with extra force again and again, although according to strict behaviorist theory the absence of the reward should have weakened the response, not strengthened it.68
But of course human beings do the same thing. When a vending machine fails to deliver, the customer pulls or pushes the lever harder a few times, or even hits or kicks the machine, either expressing frustration or acting on the thought that something is jammed and needs an extra jolt. There was no place in behaviorist theory for such internal processes, particularly not for thinking about a problem, yet a number of behaviorists noticed that their rats sometimes behaved as if they were indeed doing rudimentary purposive thinking.
One leading researcher who was aware of this was Edward Chace Tolman (1886–1959), an eminent contemporary of Hull’s and a leading neobehaviorist of the 1930s and 1940s. He observed that after a rat had run a maze a few times, it would pause at a point of decision, look this way and that, take a few steps, and perhaps turn back, all before making its choice and going on. In his presidential address to the APA in 1938, Tolman said it seemed clear that the rat was performing “vicarious trial and error” in its head. “Anthropomorphically speaking,” he added, “it appears to be a ‘looking before you leap’ sort of affair.”69