conditioning and learning
August 15, 1994

This text is essentially a set of notes taken from Chapter 7 of Dennis Coon's Introduction To Psychology; Exploration to Application, 1992, West End Publishing Company.

Learning is a relatively permanent change due to experience. Learning that results from conditioning depends on reinforcement. Reinforcement increases the chances of a response recurring. Classical (respondent) conditioning associates a stimulus with another stimulus (buzzer, food); instrumental (operant) conditioning associates a consequence with a response (press button, food).

Conditioned responses can be extinguished by presenting the conditioned stimulus without the unconditioned stimulus (buzzer, no food). Sometimes the extinguished response returns - this is termed spontaneous recovery.

Stimulus can be generalised (similar sounding buzzer) or discriminated against (if similar sounding buzzer produces no food).

Conditioned responses may be emotional as well as physical; they may also be obtained vicariously (learning through the experience of others).

Complex operant responses can be taught by shaping - a series of slightly changing routines. Positive reinforcement is where reward or pleasant events follow a response, while negative reinforcement involves the removal of discomfort. Punishment involves the imposition of a negative event, or the removal of a positive event, and decreases the frequency of response. Punishment is most effective when it is immediate, consistent and intense. Punishment aids in the conditioning of fear in the subject of the punishing agent, teaches escape and avoidance techniques, and encourages aggression.

Primary reinforcers are physiological rewards, such as food or the stimulation of pleasure centres (eg, patting). Secondary reinforcers are tokens; they represent, or are associated to, primary reinforcers (eg, money). Generalised reinforcers are those which may apply to many different circumstances (money buys not only food but prestige, power, status).

Delay between response and reinforcement reduces the effectiveness of the reinforcement, but over time chains of response can be built so an occasional, but predictable, reinforcer works for many responses. Partial reinforcement, where the subject is rewarded after only selected responses, is more resistant to extinction.


Punishment reduces the probability of a response. Any consequence that does this is defined as a punisher. Different individuals react differently to similar consequences, thus a punisher to individual A may not be a punisher to individual B. A punisher may be the imposition of an unpleasant event, or the withdrawal of a positive one.

The effect of a punisher depends on its timing, consistency and intensity. For best effect it must immediately follow the undesirable response, must follow after every similarly undesirable response, and must be memorably unpleasant.

Administering a punisher after time has elapsed greatly reduces its effectiveness. "You wait until your father comes home" does more to make the father an ogre than it does to punish.

Severe punishment may actually extinguish the response, but mostly it simply suppresses it, only to have it re-emerge later. This has been established experimentally by slapping rat's paws as they pressed a response bar in a box. Slapping initially reduces the bar presses per hour, but the slapped group caught up with the bar-pressing count of the non-slapped group within 48 hours, slaps all the way. Slapping the paws of rats or children has little permanent effect on a reinforced response. Experiments have also demonstrated the effects of severe punishment; animals severely punished while eating may never eat again (Bertsch, 1976).

There are three basic methods available to control simple learning: reinforcement (strengthening of responses - laughing at a joke), non-reinforcement (which causes responses to extinguish - not laughing at unfunny jokes), and punishment (suppressing responses - verbal abuse of joker).

Punishment may be limited in effectiveness if reinforcers are still present. Punishment should be applied in conjunction with reinforcement; if a misbehaving child is reprimanded for misbehaving, the child should also be praised for behaving.

Punishment communicates the undesirability of the response but fails to communicate what was desirable. It does not teach new behavior, it simply suppresses existing ones.

Situations that pose immediate danger, such as wandering into the street or reaching for hot items, are best dealt with by a punishment that is incompatible with the response. That is, not only administer punishment to illustrate displeasure, but do it in such a way that the displeasure is associated with the danger. For example, if a child wanders into the street, screech, carry on and take the child away from the street (preferably with a degree of alarm), or when the child reaches for a candle, slap the hand, not the bottom. [(ahem :-)] Cause the danger to be withdrawn from the child, and the child will learn that danger exists from the action of withdrawal.

tips on using punishment:

  1. Don't use it at all if misbehaviour can be discouraged in other ways. First, try reinforcement of acceptable behaviour, then try negative reinforcement (extinction - ignoring, or shifting of attention).
  2. Apply immediately after undersirable behaviour. The necessity for this fact declines with the ability to communicate what the belated punishment is for. Where this communication is not possible and the time has passed, wait until the next occurence.
  3. Use the minimum amount of punishment required. Verbal rebukes are usually adequate. Avoid harsh physical punishment. (Never slap a child's face, for instance.) Taking away privileges or other positive reinforcers is usually best for older children and adults. Frequent punishment may lose its effectiveness, and harsh or excessive punishment has serious negative side effects (see below).
  4. Be consistent. Make sure it is clear what was undesirable about the response. Punish every time the response occurs. If there is a routine, stick to it. Both parents should punish for the same things and in the same way.
  5. Expect anger from those being punished. Acknowledge it, but don't reinforce it. Be willing to back down if you make a mistake, or overpunish.
  6. Punish with kindness and respect. Allow the punished person to retain their self-respect; do not punish a person in front of others. A strong, trusting relationship minimises behaviour problems. Others should want to behave well, so as to gain praise, rather than because they fear punishment.

Side-effects of punishment

The first problem with punishment is that through classical conditioning the aversive (painful or uncomfortable) nature of the punishment becomes aversively associated with the person administering the punishment and the situation in which the punishment occurs. That is, a child repeatedly punished by parents in a certain room would associate both parents and room with unpleasantness. The aversive nature of punishment makes it an especially poor tool for toilet training or social etiquette.

The second problem with punishment is the encouragement of escape learning and avoidance learning. Escape learning occurs where the punished person acts to prematurely terminate the punishment. Avoidance learning occurs where the punished person acts to prevent the onset of punishment. For example, some cars have a buzzer that sounds if the ignition key is turned while the safety belt remains unlatched. The driver who, in order to shut the buzzer up, quickly belts up is an escape learner; the driver who, in order to not hear the buzzer at all, belts up before starting up, is an avoidance learner. Avoidance, once learned, is very persistent, as the person has no opportunity to re-evaluate the situation, since it is always avoided. Punishment encourages escape learning and avoidance learning by activating desire in the punished to escape the punishment, and to avoid being punished. Children who run away from punishment (escape) may soon learn to lie or stay away from home for a period (avoidance).

The third problem with punishment is the encourages aggression. Successive studies have shown that animals subjected to pain responsed most commonly by attacking whatever else was around (Azrin et al., 1965), irrespective of who or what.

One of the most common responses to frustration is aggression. Punishment is generally painful and frustrating. When a child is spanked the child may feel angry, frustrated and hostile. If the child then goes and hits a brother or a neighbor, the release of anger and frustration may feel good, thus reinforcing the use of aggression, which will then tend to be used again in other frustrating situations.


The most common error is depending on punishment alone as a method of training. The overall emotional adjustment of a child or pet disciplined mainly for reward is usually superior to one disciplined mainly by punishment. Frequent punishment makes a person or animal unhappy, confused, anxious, aggressive, and fearful of the source of the punishment. Children who are punished often by parents or teachers learn not only to dislike parents and teachers, but also to dislike and avoid the activities associated with them (eg, schoolwork, domestic activities) (Munn, 1969).

Parents and teachers should be aware the the use of punishment can be habit-forming; when punishment is used, it often gets immediate results, this immediate gratification reinforces punishment as a worthwhile methodology, reinforcement which is not entirely eroded by the often delayed onset of side-effects.

It would seem the old adage "spare the rod and spoil the child" should be changed to simply "spare the rod."