Escape And Avoidance

Although aversive stimuli generate large emotional effects in which broad segments of a repertoire may be depressed or strengthened, a major effect of aversive stimuli is the reinforcement (strengthening, increased frequency) of specific responses which terminate or avoid them. Examples of performances maintained by escape have already been presented. Another example is opening or closing a window to regulate the temperature of a room. The aversive stimulus is defined by the effect of the stimulus on the behavior of the organism. Those events where removal strengthens behavior are aversive stimuli (negative rein-forcers) and those stimuli which strengthen the behavior they follow are positive reinforcers.

Avoidance behavior may occur and be maintained without any explicit warning stimulus. The rat, in the example given above, will continue to press the lever, even though the shock is no longer preceded by a warning stimulus. Such a procedure may be programmed explicitly by arranging for the aversive event to occur periodically. Each response of the animal postpones the shock by a specified period, e.g., one minute. Under these conditions, the rate of responding will vary closely with the intensity of the electric shock and the interval by which each response postpones the electric shock. The explicit reinforcement for this kind of avoidance behavior is difficult to specify in the sense of the careful use of reinforcement which we have followed to date. Other than postponing the next shock, each avoidance response has no effect on the external environment. The reason for the continued maintenance of the avoidance behavior without any explicit reinforcement lies in the probable punishment of almost any activity the organism engages in other than the one postponing the shock. Over a long period of time almost every form of activity will be punished (followed by shock) except the specific response which postpones the aversive stimulus. The actual reinforcement for the avoidance response is the escape from the various behaviors which are characteristically punished and which are incompatible with the avoidance response. This same paradigm will occur repeatedly as we discuss other processes involving punishment, such as Freudian mechanisms of denial, reaction formation, repression, and projection.

It seems highly probable that many clinically observed obsessions and compulsions, e.g., hand wringing, tics, and repeated gestures, represent avoidance responses which are very strongly maintained because they prevent the occurrence of other behaviors which have been associated with very strong aversive control. The compulsive housewife may be engaging in continuous housecleaning because any cessation in these activities will permit other behaviors to occur which lead to conditioned aversive stimuli (anxiousness).

Much of human behavior is maintained by its effect in avoiding an aversive stimulus. More often than not, we get the child to pick up its clothes or replace its toys by presenting an aversive situation when it does not clean up, rather than some positive consequence for picking up the toys or clothes. Payment of taxes to governmental agencies occurs because it postpones or avoids fines, disapproval, or incarceration. Many behaviors of many students are maintained because they postpone or avoid the aversive consequence involved in dismissal from school or censure by the school authorities or the adverse effects at home of low grades. Many devices of ordinary conversation, for example stuttering or stalling remarks such as "uh" or "ah," are essentially avoidance responses postponing a pause in speech which is, in general, an aversive situation.

The clinical syndrome, denial, is another example of behaviors maintained by their effectiveness in removing or escaping from aversive stimuli. The woman who sets two places at the table for many weeks after her husband dies and who continues to look out the window at the hour when he had customarily returned home is essentially avoiding conditioned aversive stimuli. She is engaging in performances which are incompatible with the verbal behavior and other performances associated with the death of her husband. These performances are not reinforced by the normal practices of the verbal community, but the net reinforcement, which occurs by avoiding stimuli which define the death of the spouse, strengthens behavior which asserts the arrival of the husband.

To extinguish avoidance behavior, it is necessary both that the aversive event no longer occur and that the avoidance response not occur for an interval somewhat longer than the customary avoidance interval. In the case of a rat pressing a lever because the response postpones the electric shock for 60 seconds, the discontinuation of the electric shock will have no effect on the behavior of the animal until at some point it pauses for over 60 seconds. With well-established avoidance behavior, however, the rate of responding is so high and sustained that long pauses seldom occur. The history of avoidance conditioning increases the probability that an avoidance response will occur with every second that an avoidance response is not emitted. With shock of a sufficiently high level, a monkey will continue responding in a sustained manner so that shocks will occur only rarely, perhaps every 20 or 30 hours. As a result, when the electric shock is first disconnected, it will take many hours before the animal pauses sufficiently long enough for it to be exposed to the condition of not responding and not being shocked. Until recently, it was believed by some that the strongly maintained avoidance behavior was irreversibly conditioned. Eventually, however, it is possible to extinguish avoidance behavior under the proper conditions. Figure 12 shows an experimental recording of an extremely prolonged extinction curve after avoidance conditioning. If only the initial part of this performance were recorded, it might be assumed that the avoidance behavior was permanently in the animal's repertoire.

Fig. 12. Sustained responding by a cat during avoidance conditioning with a warning signal. (From Sidman, J. Abnorm. dr Social Psychol., 50:219, 1955.) Another difficulty in eliminating avoidance behavior from the human repertoire occurs because it is often impossible to keep the individual from making the avoidance response, and hence, exposing himself to a situation where the nonoccurrence of the aversive stimulus can weaken the avoidance behavior. In many obsessive compulsive behaviors observed clinically, the failure to "step on every crack on the sidewalk" or "touch every door knob" produces such great anxiety that these responses always occur. The elimination of these responses requires that reality be tested. Another factor preventing the extinction of avoidance behavior in human actions is that the individual will also avoid the entire situation in which the avoidance behavior is relevant. Consider, for example, the case of the individual who has a phobia of buses which has developed as a result of a past history of aversive consequences in similar vehicles. To eliminate the phobia it is necessary that the individual be in a bus and not be exposed to any strong aversive stimuli. It is difficult to arrange this deconditioning because the bus is a conditioned aversive and preaversive stimulus which supports very strong avoidance and escape behavior. The problem referred to here is similar to that referred to in traditional psychiatric terminology as "reality testing." To eliminate the phobia, a situation similar enough would have to be arranged so that the situation would share some of the aversive properties of the bus, while not being so similar that the avoidance behavior produced by it would be so strong that the individual would immediately leave the situation. When the avoidance behavior extinguishes in the situation related to the bus, the patient might be exposed to closer and closer approximations to the actual bus. At all times it would be necessary to be careful that the situation does not become aversive enough to produce a complete avoidance response. Such procedures have been successfully used in actual clinical treatment of phobias (Wolpe, 1958; Jones, 1924) and in laboratory demonstrations. Another technic for hastening the extinction of avoidance response would be the use, concurrently, of a strongly maintained positively reinforced behavior to keep the person exposed to the situation in spite of aversive stimuli that might appear temporarily. The strong positive reinforcer would keep the person in the aversive situation in spite of the strong behavior which would be reinforced by escape from the aversive situation. Such technics are easily arranged in the laboratory but are more difficult to establish and maintain clinically.

In the control of behavior by avoidance of aversive stimuli, as in many of the other behavioral processes we have discussed, the verbal behavior of the person controlled has only indirect effects. Explaining to a patient that nothing bad will happen in this situation any more will have little effect in lessening the avoidance repertoire of the individual. Extinction must occur by the individual being in the situation, not responding, and not being exposed to the aversive event. On the other hand, when the conditioned aversive stimuli consist of the individual's own verbal behavior, the verbal repertoire is highly relevant. Verbal responses which have been punished or which because of their thematic content derive aversive consequences as a result of generalization from other behaviors that have been punished, or which have simply occurred in close proximity to aversive stimuli, may be markedly reduced in strength to the point that the individual is no longer consciously aware that they are in his repertoire. Once these verbal responses come to function as conditioned aversive stimuli, any tendency to emit them will strengthen incompatible responses because they terminate the aversive stimuli. This is the Freudian concept of "repression," and many psychiatric technics, for example, free association and catharsis, are directed toward arranging conditions which will make it possible for a patient to talk about things which have a history of strong aversive control. Theoretically, the treatment is essentially directed toward eliminating the conditioned aversive stimuli that reinforce avoidance responses which are pre-potent over the repressed behavior. The patient is directed to emit verbal responses which are closely related to those repertoires which are "repressed" because their emission would produce strong conditioned aversive stimuli. The clinical technics are first designed to maintain a lot of verbal behavior which is far enough removed from the repressed behavior so as not to produce extreme anxiety. As the thematic and formal content of the patient's speech comes closer to the previously punished behavior and no further punishment occurs, there is partial extinction of the conditioned aversive stimuli through induction.' As more and more behavior is emitted without additional punishment, the patient is able to speak about matters closer and closer to the punished behavior, continuing the extinction process. Wolpe has formalized a procedure of this kind in a special therapeutic method (Wolpe, 1958; Miller and Dollard, 1950).

As with many of the processes of positive reinforcement, the actual history by which the individual is exposed to the avoidance contingencies has bearing on whether avoidance behavior will be generated and how strongly. The direct emotional effects of the aversive stimulus may interfere with the original development of avoidance behavior. If the avoidance response which was to be established occurred infrequently, then the high initial number of aversive stimuli would suppress the behavior of the individual, including the avoidance response. It would also be very difficult to develop performances in an individual by avoidance of aversive stimuli if something approximating these performances is not already in the repertoire of the organism. The emotional effects of the aversive stimulus will interfere with the subtle behavioral development required when a new form of behavior is successively approximated. This characteristic of aversive control will be discussed in more detail later when we compare the use of positive and negative reinforcement as general technics of control.