REINFORCEMENT (Reward and Punishment)

Any procedure that increases the strength of a conditioning or other learning process.The concept of reinforcement has different meanings in classical and operant conditioning. In the classical type, it refers to the repeated association of the conditioned stimulus (the sound of a bell, for instance) with the unconditioned stimulus (the sight of food). After the two occur together a number of times, the connection between the bell and salivation (the conditioned response) will be strengthened, and salivation will occur when the bell is sounded alone. If the conditioned stimulus is repeatedly presented without being reinforced by the unconditioned stimulus,the response will be weakened and eventually “extinguished.”In operant conditioning, reinforcement refers to the reward that is given after a correct response, or the punishment that follows an incorrect response. In training procedures it is important to apply a carefully worked out “schedule of reinforcement” instead of giving rewards or punishments haphazardly. Here, too, a weakening and eventual extinction of the response occurs if reinforcement is consistently withheld— that is, if the animal being trained is not rewarded with food.Whether we recognize it or not, most of our time is spent in doing things that give us some type of reward or keep us from receiving some kind of punishment. We are constantly striving to elicit praise and encouragement from the people we respect, and we are equally eager to avoid their disapproval. Few people work “just for the fun of it”; they expect to receive a definite reward for their efforts. In one experiment, four groups of boys were required to learn a maze. One group was paid twenty-five cents for mastering the task; the others received much smaller rewards. The first group did the best by far. Yet material rewards are by no means the only effective kinds. A gold star or a pat on the back, especially when given by an important person, can be as potent as a chocolate bar, and the feeling that we are improving can be as great a satisfaction as an increase in salary.Rewards must be appropriate to the situation. We do not give an adult a gold star when he does a job well, nor do we give a first grader a dollar bill for learning the alphabet. High scholastic standing may be a reward to one student, but a sure sign of a “greasy grind” to another. Some situations that appear to be punishing are on closer examination actually rewarding. A beginning medical student may be nauseated at the thought of cutting up a human cadaver, but this negative reaction is soon banished when he begins to reap the rewards of knowledge.In general, reward is superior to punishment as an incentive to learning. Reward reinforces positive performance; punishment only indicates what is objectionable. Punishment does not by itself modify attitudes and skills, and it frequendy produces tension and resentment. Since it is a cue as to what not to do, its chief value lies in suppressing the undesired response until the correct response can be learned.Sometimes punishment or the prospect of it actually interferes with learning the desired response. It may produce so much anxiety that the individual comes to fear the entire situation and tries to avoid it altogether. If a child is continually punished for playing with a certain toy in a particular room, after a time he may rebel at going near the room. In socializing a child, or in reforming an antisocial person, the purpose is to induce him to adopt acceptable behavior and avoid unacceptable behavior. Attempts to restrain antisocial acts merely by threats that provoke anxiety are usually ineffective. What often happens is that the person avoids the situation in which he is threatened with punishment, but acts in an antisocial fashion in a different situation. Punishment therefore fails to educate or re-educate the offender. This is one reason why reform schools and prisons do not succeed in rehabilitating the majority of their inmates.One of the most important findings in the entire field of learning is that reinforcement is most effective when it is intermittent. Both reward and punishment tend to lose their force when they are constantly applied. This is particularly true with children who are overrewarded or overpunished. Fortunately a single reinforcement may persist and have an effect for a long period of time. In a classic experimentSkinner, 1950), a pigeon pecked at a rate of six thousand times per hour for several hours even though it was only rewarded at five-minute intervals (an intermittent schedule of reinforcement). Running a gambling house is a successful operation because it takes only a few small wins to keep people in the game until they lose their whole bankroll. Unfortunately a few small losses rarely have an equally strong deterrent effect.The lasting effects of partial reinforcement are especially evident in child-rearing. If rewards and punishments are few but consistent, they will usually make a lasting impression. The mother who repeatedly reminds her child to look both ways when crossing the street can usually count on him to take this precaution when he is alone. He will, however, need occasional reminders or rebukes—that is, intermittent reinforcement. This principle can also work negatively. Many young children cry when they are put to bed in order to get extra attention and stay up longer. If the parents “reward” this behavior by giving in even once or twice, the child will try it again and again.Many superstitions are based on partial reinforcement. Indian rain dances are only occasionally followed by rain, yet this is enough to maintain belief. A superstitious person will remember the one case when he had bad luck after a black cat crossed his path, but will forget all the “negative instances” when nothing happened to him. Athletes are notoriously superstitious. After winning a game they sometimes think over every action they performed and try to repeat the same routine the next time they play. A baseball player, for example, may take the same number of steps to get to his position in the outfield, or take a drink of water at exactly the same time during every game. These are all examples of partial reinforcement.Reinforcements are frequently divided into two types. Primary reinforcements directly satisfy basic needs or punish unacceptable behavior. Examples are material rewards or expressions of approval, and punishments like a slap or a reproach (PLATE 37). Secondary reinforcements, on the other hand, are merely associated with direct reinforcements—nevertheless they can be extremely effective. In animal training, the sight of food, as opposed to the actual taste, may become a reinforcer. Restaurateurs know that good meals are their best advertisement, but they also know that a meal tastes best in an atmosphere of comfort and relaxation. The meals are the primary reinforcement, the atmosphere a secondary reinforcement. Similarly, many teachers find that the learning process can be enhanced by an occasional digression or dash of humor. As we become more sophisticated, these secondary reinforcements play a greater and greater role in our lives. A man who has long ago made enough money to satisfy all his family’s physical needs will continue to work for humanitarian causes or for such status symbols as titles and honorary degrees. Likewise, the person who does not respond to physical punishment may be deterred from bad behavior by fear of criticism or loss of face.