Copyright © 2007-2018 Russ Dewey
Generalization and Discrimination
Generalization occurs when an organism makes the same response to different stimuli. A classically conditioned response to a slightly different signal will depend on its resemblance to the original.
If a dog receives meat powder after hearing a 500 Hz tone, it will probably salivate when hearing a 450 Hz tone also, but not as much as it would to another 500 Hz tone. It would salivate less to a 400 Hz or 600 Hz tone...and even less to a 300 Hz or 700 Hz tone.
Pavlov found that the greater the resemblance between stimuli used during training and stimuli used during testing the greater the generalization. More salivation occurred if a tone was close to the training tone. Less salivation occurred if a tone was very different from the training tone.
When does generalization occur? What determines the amount of generalization?
Generalization is often an important phenomenon in real-world settings. A therapist may work with a client to reduce anxiety states triggered by some kind of stimulus (let's say, dirt and contamination). The client might imagine the feared situation (touching dirt) while deeply relaxed on the therapist's couch.
But will the exercises performed in the therapist's office generalize to real world settings? Perhaps not completely.
The greater the resemblance of the training situation to the testing situation, the more generalization should occur. That is why in vivo (naturalistic or real life) therapies often work best.
In vivo extinction therapy with a person who is anxious about contamination could occur by plunging the arms of such a person into a can of garbage. The therapist stands by for encouragement for 20-30 minutes as extinction of anxiety occurs.
Hans Eysenck actually used this technique, and it worked well. After an initial wave of anxiety, the reaction subsided, and after 20-30 minutes the situation was boring. Anxiety was gone; extinction had occurred.
Some degree of spontaneous recovery can be expected after extinction. Therefore extinction procedures must be repeated a few times on different days before a client achieves permanent reduction of anxiety to a specific cue. If executed systematically and patiently, extinction therapies (often called exposure therapies) work very well.
A person who is anxious about flying can be taught to remain calm (or given an anxiety-reducing drug) during actual airplane flights. That helps maximize the likelihood that results of therapy will carry over (generalize) to future real world situations.
However, no two situations are identical. If an organism notices differences between situations rather than similarities, generalization will not occur. A horse that responds well to one rider may be stubborn for another.
Would the students in Landauer's class jump if one of their classmates stood up and shouted "Now!"? Maybe some would jump. If they did, they would be showing generalization.
Others might not jump. They would be showing discrimination, the opposite of generalization. Discrimination is described below.
Generalization involves our knowledge of the world. If two stimuli are interpreted as the same, a person will generalize between them.
Little children go through a phase in which they show overgeneralization in language and thinking. A one-year-old child may call all 4-legged creatures "doggies."
How do little children show overgeneralization?
Therapists often try to help their clients avoid overgeneralization. For example, if a person has a bad experience with a member of the opposite sex, that should not ruin all relationships with the opposite sex.
How can overgeneralization be a focus of therapy?
That would be overgeneralization. Reinterpreting experiences can prevent overgeneralization. For example, the therapist might encourage a person to see that "one person hurt you" but "not all men are alike" or "not all women will hurt you" therefore damaging forms of avoidance are not needed.
The opposite of generalization is discrimination. Discrimination occurs when an organism responds differently to two stimuli.
Adult knowledge of the world includes sharp distinctions between animals such as cats and dogs. We make discriminations a one-year-old might not yet make.
An adult scratched by an angry cat would not therefore develop a fear of dogs. A one-year-old might, if they do not discriminate between cats and dogs.
What is discrimination learning?
In classical conditioning, discrimination occurs when one stimulus triggers a conditional response but another does not. To set up discrimination in the laboratory, a researcher creates a situation in which two stimuli predict different things.
For example, a green light is followed by meat powder, but a red light is not. Soon the dog discriminates between green and red lights. It salivates to the green light but not to the red light.
Discrimination is evidence that an animal notices the difference between two or more stimuli. That makes the technique of discrimination learning very useful.
We cannot ask a dog whether it sees in color, but we can arrange a discrimination experiment to reveal the answer. Two stimuli are matched in shape and brightness, differing only in color.
One color is made to predict delivery of food powder. The other never predicts food powder. The positions of the two stimuli are randomized, so the dog cannot tell the stimuli apart only by their color, not by their position.
If the dog learns to salivate to one light and not the other, then we know it can tell the difference between them. (And, yes, experiments like this show that dogs can see in color.)
How can discrimination learning be used to investigate sensory capabilities of an animal?
In developmental psychology research, very young babies can provide research data through discrimination learning techniques. In one procedure, a baby sucks on a nonnutritive nipple: a pacifier connected to a computer.
How is discrimination learning used to study young babies?
If the baby sucks faster when it sees the mother's face rather than a stranger's, this shows the baby can tell them apart. It can discriminate between them. Such experiments show that babies can detect their mother's faces, voices, and odors.
Release from Habituation
One very useful research technique is called release from habituation. Habituation is gradual cessation (stopping) of a response when the same stimulus is repeated many times.
Habituation happens reliably (it is a robust phenomenon, as scientists say). Habituation is adaptive for all organisms, because organisms do not waste energy on stimuli that are constantly around but cause no harm.
If there is a loud ticking in the room, soon you will not hear it. If you live near traffic, soon you tune it out. If you repeat a tone, soon a baby will ignore it.
But if a different sound is played, the baby might perk up, or stop sucking on a non-nutritive nipple. If that happens, then obviously the baby can tell the difference between the new tone and the earlier tones.
This phenomenon, called release from habituation or dishabituation, is proof of an ability to discriminate. It shows an organism reacts differently to two categories of stimuli.
What is release from habituation? What else is it called?
The great advantage of dishabituation is that it works in creatures without language. Babies cannot tell you if they discriminate between two forms of music, but release from habituation shows that they can.
We ran into an example of dishabituation in Chapter 4 when discussing blindsight as an unusual form of sensitivity. The "second visual system" (the one passing through the superior colliculi of the brain) directs eye movements. It is unconscious.
Normally the collicular visual system directs eye movements toward moving stimuli. But if the same stimulus is repeated often enough, the eye movements stop. This is habituation.
If the stimulus in peripheral vision changes, the eye movement returns. That shows the collicular system is capable of pattern recognition of a simple variety.
Experiments using release from habituation have shown remarkable discrimination abilities in babies. For example, babies have a primitive sense of number.
A baby can be shown a series of slides in which there are always three objects in various different positions. Each picture shows different objects, but there are always three.
Soon the baby gets bored and stops responding to the pictures. But if a picture with four objects appears, the baby perks up and pays attention to it.
Similar studies have shown that babies have the ability to distinguish between language sounds. The can discriminate the phonemes used in all different human languages, before the age of one.
What are examples of discrimination abilities in babies, shown by dishabituation?
After the age of one, the ability of babies to discriminate some of these speech sounds disappears. Eventually they only respond to language sounds they hear in their own environment.
Pavlov believed that being pressured to make excessively fine discriminations could trigger neurosis or mental disturbance. In one experiment, Pavlov taught a dog to discriminate between a circle and an oval.
The location of the circle and the oval was changed randomly, so the dog had to discriminate between them on the basis of shape, not location. When the dog pointed its nose at a circle, it received food. When it pointed at the oval shape, it received an electric shock.
Gradually Pavlov made the oval rounder and rounder. Soon it was hard to tell the oval from the circle. The dog began showing signs of distress, whining and defecating. Pavlov said this showed an experimental neurosis.
How did Pavlov produce an "experimental neurosis"?
To modern sensibilities, the situation was rather pathetic. Dogs desperately want to please their human masters, and Pavlov created a situation in which failure was inevitable and painful. No wonder the dog was distressed.
As mentioned earlier, the Soviet Union endorsed Pavlov's psychology. In effect, it became the officially approved version of psychology in the country, and every psychology student knew Pavlov's theories.
I heard a Soviet psychologist who was visiting the U.S. use Pavlov's ideas about neurosis. He was explaining the strange behavior of hippies in the late 1960s.
Observing longhaired young men tossing Frisbees on the University of Michigan campus, the psychologist said–with a confident gleam in his eye–that he knew what caused this odd behavior. "It is your supermarkets. You have 20 different brands of soap; who can tell which is best? The result is neurosis."
Write to Dr. Dewey at email@example.com.