Punishment

The process of punishment is dealt with here because it represents an interaction between an aversive stimulus and a behavior of an organism maintained by some process of negative or positive reinforcement.

The effect of punishment in reducing the frequency of an item in an individual's repertoire is more correctly described as suppression of the behavior rather than its elimination. All of the experimental work on animals supports the finding that punishment suppresses the behavior, more or less temporarily, rather than eliminates it (Skinner, 1938; Estes, 1944). So long as the response is still being reinforced by a durable reinforcer, the behavior will occur again as soon as the punishment is discontinued. "When the cat's away, the mice will play." Thus, punishment (negative reinforcement) is not simply the algebraic opposite of reinforcement. With very severe punishment, however, the result may appear to be a complete elimination of the behavior. The effectiveness of the severe punishment in preventing the occurrence of the behavior is due not so much to the elimination of the punished behavior from the individual's repertoire as it is to the conditioning of durable and strongly conditioned avoidance behavior. The mechanism is the same as that described for the repression of "punished verbal behavior." Responses which avoid the conditioned aversive stimuli associated with any tendency to emit the punished behavior generate such strong avoidance and escape behavior.

An animal experiment will more easily present the salient features of the main effects of punishment. The first experiment showing the temporary effect of punishment used a rat, reinforced with food for pressing a bar (Skinner, 1938). The delivery of food was discontinued and an attempt was made to reduce further responding on the lever by punishing any subsequent responses. While the punishment was being applied, the rat practically stopped pressing the bar. As soon as the punishment was discontinued, however, the rat began responding again and emitted approximately the same amount of responses as he would have had no punishment been delivered at all. The effect of the punishment was simply the postponement of the emission of the behavior which had been conditioned as a result of the prior reinforcement with food. With more severe punishment, as for example with very intense electric shocks, it would be possible to suppress the behavior so severely that it would not occur at all. But in this latter case, we would be creating conditions in which another behavioral process would be responsible for the cessation of behavior. A very severe aversive stimulus would condition avoidance behavior in respect to the lever and behaviors associated with it, the reinforcement of which would strengthen any performances that will take the animal away from the punished situation. In spite of the very severe depression of the behavior, in every case it should be theoretically possible to reinstate the original behavior. By the careful application of procedures, we could extinguish the aversive effects of the electric shock so that the avoidance behavior would no longer be prepotent over the tendency to press the lever as a result of the previous positive reinforcement.

Punishment may also be used to bring behavior under stimulus control by following incorrect responses with an aversive event. The employer criticizes a worker for a mistake; the student is graded poorly for errors in arithmetic; and even the physical environment punishes when we don't look where we walk. Except in the case of the physical environment, where the punishment of mistakes occurs naturally and continuously, the effects of punishment of mistakes are temporary, and a performance will deteriorate as soon as the punishment is discontinued. An example is a pigeon experiment in which a bird has identified colors with good accuracy thousands of times under conditions where occasional mistakes are punished by shutting off the experiment for a brief period. As soon as punishment is discontinued, however, the bird stops paying attention altogether and pecks the various keys randomly, being reinforced intermittently rather than paying attention to the stimuli.

Repression And The Secondary Effectsof Preaversive Stimuli

In some cases the incidental effects of aversive stimuli in generating unexpected avoidance or escape responses are more subtle. Responses may be reinforced because they pre-empt behaviors which are conditioned aversive stimuli due to a punishment history. Any response which prevents the occurrence of conditioned aversive stimuli is reinforced because it terminates an aversive stimulus. The tendency to "forget" an unpleasant appointment is such an example. All available experimental evidence leads us to believe that there is little forgetting or weakening of behavior simply through passage of time. Changes in response strength (forgetting) are due to secondary processes, interfering or competing with behavior or extinction. All of the verbal responses associated with the appointment lead to conditioned aversive stimuli which may reinforce any performance which terminates these stimuli. Such performances are all of the prepotent incompatible responses which make it impossible to keep the appointment or lead to forgetting it. This is the classical Freudian mechanism of "repression." The repressed responses usually are verbal ones which are never emitted because the verbal behavior generates conditioned aversive stimuli that strengthen some incompatible behavior which is so strong that it is prepotent over the "repressed" behavior. In most cases the overt behavior never occurs because the early, incomplete, or related forms of behavior generate aversive effects. The functional analysis of behavior makes it clear that the phenomenon of repression occurs because punishment and aversive control do not eliminate, but simply suppress, behavior. Thus, these "punished" responses are still in the individual's repertoire but cannot be emitted until the incompatible avoidance responses are eliminated. These in turn cannot be eliminated until procedures are carried out which reduce the aversive and emotional effects elicited by the suppressed performances. Many clinical therapeutic procedures appear to be methods of extinguishing the conditioned aversive properties of punished verbal behavior by getting the patient to emit the responses to a listener, the therapist, who never punishes. The general procedure is to strengthen as much verbal behavior as possible in the patient so that the therapist can lead the patient to verbal responses which are slightly related to the repressed behavior.

If the aversive effects of emitting responses similar to the repressed behavior are not so severe as to lead to avoidance responses, then some of the conditioned aversive effects of the repressed behavior will have been extinguished. Each time the patient emits performances related to the repressed ones in the presence of a listener who provides no further aversive consequences, the repressed performances lose some of their conditioned aversive properties. The degree of the loss of aversive properties of the repressed behavior would be proportional to how closely related the therapy performances are to the repressed behavior. As more and more extinction occurs, the patient is more and more likely to emit the suppressed (clinically repressed) verbal responses or very similar ones. The possibility of carrying out such an extinction procedure depends, of course, upon the strength of the verbal response. Part of the therapeutic procedure must be designed to strengthen the patient's verbal behavior sufficiently for these verbal responses to compete with the avoidance behavior generated by the conditioned aversive stimuli. In the extreme case, an unskilled therapist may strengthen verbal behavior too rapidly and the resulting conditioned aversive stimuli may reinforce the behavior of avoiding the therapist.

Punishment leads to the same complex interaction of the processes of aversive control as escape and avoidance. Punishment is likely to occur under characteristic circumstances and is contingent upon specific responses. The repeated circumstances under which punishment has occurred will therefore lead to the same general emotional effects of the preaversive stimuli described above. The intended specificity of punishment is not always borne out in its actual effects on the individual. It is more likely that the punishment of a specific response will lower the frequency of many related behaviors. A parent who punishes a child for not picking up its toys may, as a consequence, effectively punish all of the behaviors involved in playing with toys as well as other activities which may happen to be going on at the same time. Punishment also creates potential for escape and avoidance responses. The rat who is punished in a typical laboratory situation remains in the box only because of physical restraint. In a more normal ethnologic situation, the conditioned aversive stimuli generated by the punishment will reinforce behaviors such as running away from the entire experimental situation.