Chance, chapter 8

Operant Learning: Punishment


Punishment, like reinforcement, is defined by its measurable effects on behavior. Reinforcement means an increase in the strength of behavior due to its consequences. Punishment means a decrease in the strength of behavior due to its consequences. An experience must have three characteristics to qualify as punishment:

  1. a behavior must have a consequence

  2. the behavior must decrease in strength (e.g., occur less often)

  3. the reduction in strength must be the result of the consequence


There are two types of punishment

  1. In positive punishment, the consequence of a behavior is in the appearance of, or an increase in the intensity of, a stimulus.

  2. In negative punishment, a behavior is weakened by the removal of, or a decrease in the intensity of, a stimulus.


Variables affecting punishment

Many of the same variables that are important in reinforcement are also important in punishment.


Contingency

The degree to which a punishment weakens a behavior varies with the degree to which a punishing event is dependent on that behavior. The greater the degree of contingency between a behavior and a punishing event, the faster the behavior changes.

If you use a punishment, you would do well to remember that the more consistently a behavior is followed by a punishing event, the less likely the behavior is to occur in the future.


Contiguity

The interval between a behavior and a punishing consequence is also very important: the longer the delay, the less effective the punisher is.

Perhaps delays reduce the effectiveness of punishment because during the delay interval, other behaviors occur, and these may be suppressed rather than the indented behavior. Thus, a delayed punisher may suppress the same amount of behavior, but immediate punishment is more likely to act on the targeted behavior.


Punisher intensity

Several studies have shown a clear relationship between the intensity of a punisher and its effects.


Introductory level of punisher

Azrin and Holz argue that using an effective level of punishment from the very beginning is extremely important. The problem with beginning with a weak punisher and gradually increasing its intensity is that the punished behavior will tend to persist during these increases, and in the end a far greater level of punisher may be required to suppress the behavior.

Unfortunately, the idea of beginning with a strong aversive is also problematic. For one thing, hit is not obvious at the outset what level of punisher will be effective.


Reinforcement of the punished behavior

In considering punishment, remember that the unwanted behavior almost certainly is reinforced. If this were not the case, the behavior would not occur or would occur very infrequently. It follows that the effectiveness of a punishment procedure depends on the frequency, amount, and quality of reinforcers the behavior produces.


When punishing an unwanted behavior, be sure to provide an alternative means of obtaining the reinforcers that maintain that behavior.


Motivating operations

The effectiveness of a reinforcer can be increased by performing an establishing operation. The same is true of punishment.

In general, the higher the level of reinforcer deprivation, the more effective a punisher is.


Qualitative features of the punisher can be an important variable affecting punishment as well.


Theories of punishment

Early theories of punishment proposed that response suppression was due to the disruptive effects of aversive stimuli. Research on punishment undermined this explanation by producing two key findings: first, the effects of punishment are not as transient as Skinner thought if sufficiently strong aversive are used. Second, punishment has a greater suppressive effect on behavior than does aversive stimulation that is independent of behavior.

Today, the two leading explanations of punishment are the two-process and one-process theory.


Two-process theory

Two-process theory says that punishment involves both Pavlovian and operant procedures. The theory is applied to punishment much in the same way as it is applied to avoidance. If a rat presses a lever and receives a shock, the lever is paired with the shock. Through Pavlovian conditioning, the lever then becomes a conditioned stimulus for the same behavior aroused by the shock, including fear. Put another way, if shock is aversive, then the lever becomes aversive. The rat may escape the lever by moving away from it. Moving away from the lever is reinforced by a reduction of fear. Of course, moving away from the lever necessarily reduces the rate of lever pressing.

Critics of the two-process theory charge that the theory has the same flaws when applied to punishment as it does when explaining avoidance. For instance, the theory predicts that punishment would reduce responding in proportion to its proximity to the punished behavior.


One-process theory

The one-process theory of punishment is similar to the one-process theory of avoidance. It says that only one process, operant learning, is involved. Punishment, this theory argues, weakens behavior in the same manner that reinforcement strengthens it.

The Premack principle states that high-probability behavior reinforces low-probability behavior. If the one-process theory is correct, then the opposite of Premack’s reinforcement rule should apply to punishment: low-probability behavior should punish high-probability behavior. This is, in fact, what happens. If, for example, a hungry rat is made to run following eating, it will eat less. The low-probability behavior (running) suppresses the high-probability behavior (eating).