I. Stimulus Generalization

A. Lab Methods

Classical conditioning paradigm

Instrumental/Operant conditioning paradigms

(1) stimulus generalization tests

(2) Probe trials

In both paradigms, the typical finding is _________________________?

Guttman & Kalish (1956)

This figure shows the amount of responding to both the original stimulus (600 nm light) and other similar (nontrained) stimuli.

The shape of the generalization gradient reflects the degree of stimulus generalization that an organism exhibits.

What is the typical shape of a stimulus generalization gradient?

steep gradient---responding decreases substantially even when minor changes in the graining stimulus occur.

How much generalization does a steep gradient indicate?

shallow gradient—substantial responding occurs even to stimuli that are highly DISSIMILAR to the training stimulus.

How much generalization does a shallow gradient indicate?

What kind of information is represented on the X-axis?

What kind of information is represented on the Y-axis?

Absolute gradients vs. Relative gradients

Excitatory vs. Inhibitory gradients

B. Important Variables

1. Degree of Original Learning

Hearst & Koresko (1968)

The more training pigeons received with the vertical-line stimulus, the greater was the level of responding to this stimulus on test trials.

Also, more training REDUCED generalization

**This effect of degree of training on absolute generalization gradients is reasonably common.

There is disagreement as to whether increased training on the original stimulus (S+) reduces generalization in a RELATIVE SENSE.

2. Motivational Level

Kalish & Haber (1964)—Before being trained to peck a colored key, pigeons were deprived of food to 70%, 80%, or 90% of their free feeding body weight.

Which group is considered to be the most motivated?

Kalish & Haber found that increased levels of motivation resulted in greater responding to the S+ during the generalization test.

They also found that higher levels of motivation tended to produce STEEPER absolute-generalization gradients.

*Opposite results (shallow gradients) with relative generalization gradients.

*Variables affecting generalization gradients are complex and depend on the manner in which the gradients are measured.

3. Schedules of Reinforcement

Hearst, Koresko, and Popper (1964)

All pigeons were exposed to a VI schedule, but the average length of the inter-reinforcement interval was 30 seconds or 1, 2, 3, or 4 minutes in the five groups, respectively.

*Schedules of reinforcement that provide more frequent reinforcement tend to reduce generalization.

4. Duration of Training-Test Interval --the length of time that separates training on S+ and the subsequent test.

Several studies have shown that the tendency to generalize increases as the training-test interval is lengthened.

Thomas & Lopez (1962) trained pigeons to peck a colored key and later tested them using keys of different colors. Pigeons were tested 1 minute, 1 day, or 1 week after training.

Pigeons tested after 1 minute exhibited very steep generalization gradients (little generalization). Those tested 1 day or 1 week after training showed somewhat flatter gradients.

Why is this effect often interpreted as an example of forgetting?

5. Previous Discrimination Training

Usually results in less generalization

Also, compare the wavelength at which the peak occurs for the two groups

Peak-shift phenomenon