Transcript STIMULUS CONTROL OF BEHAVIOR

STIMULUS CONTROL OF
BEHAVIOR
Chapter 10
Stimulus Control of Behavior
2

Generalization


Responding in the same manner to similar stimuli.
Discrimination Learning

Responding in different ways to different stimuli.
The Generalization Process
3



Stimulus generalization/discrimination occurs
frequently in the real world
Sometimes it is undesirable
 Black Americans hold negative stereotypes of their
own group (e.g., racial prejudice).
 Children over generalize rules of language (e.g.,
Christine “goed” to the hockey game)
Sometimes it is undesirable
 Republican talk show hosts are idiots.
 It “snowed” today.
Generalization Gradient
4
A visual representation of the response strength
produced by stimuli of varying degrees of similarity to
the training stimulus.
 Excitatory generalization gradients




A graph showing the level of generalization from an
excitatory conditioned stimulus (S+) to other stimuli.
Trained on S+, and then tested in extinction on S+ and other
values.
Many studies of these gradients employ pigeons, which have
excellent color vision.
Gutman and Kalish (1956)
5
Group 530
Group 550
Group 580
Group 600
Inhibitory Generalization Gradients
7

Weisman and Palmer (1969) illustrate the
inhibitory-conditioning generalization gradient.
 Green disk (S+) rewarded with food on a VI-1 min.
schedule.
 White vertical line (S-) on green disk no food.
 Tested with a series of white lines ranging from 090 degrees off the vertical on green disk
Results from 5 pigeons
8
Smartest Pigeon
Dumbest Pigeon
Hull-Spence Theory
9
 According
to the Hull-Spence view, discrimination
learning develops in 3 stages.
 First,
conditioned excitation develops to the S+ as a result of
reinforcement.
 Second, nonreinforcement in the presence of the S- results in
the development of conditioned inhibition to the S-.
 Finally, the excitation and inhibition generalize to other
stimuli.
Predictions
10


The Hull-Spence model predicts a steeper
generalization gradient with discrimination training
than with nondiscrimination training.
The maximum response occurs not to the S+, but
rather to a stimulus other than the S+, and in the
stimulus direction opposite that of the S+.
The Peak Shift Phenomenon
12

Hanson (1959) reported responding might be shifted
from S+.
The greatest response for discrimination-training subjects
was not to the S+ but to a stimulus away from S+ opposite
the direction of the S His results support Spence-Hull

Peak shift
13
 The
shift in the maximum response, which occurs to
a stimulus other than S+ and in the stimulus
direction opposite that of the S-
 In
contrast, pigeons receiving nondiscrimination training
responded maximally to the S+
 The overall level of response was higher with
discrimination training than with nondiscrimination
training, which the Hull—Spence model did not predict.
14
Is Generalization Passive or Active?
15

Both!
 Passive generalization: Perceptual confusion does
happens because the stimuli can’t be distinguished.
 Active generalization: In other situations, animals or
people may generalize to stimuli that are very
different to the conditioning stimulus
Passive Generalization Process
16

Lashley-Wade theory

Animal is unable to distinguish between the
generalization test stimulus and the trained stimulus.
 Thus,
failure to discriminate between training and test stimuli
is responsible for stimulus generalization.
Lashley and Wade Predictions
17




1. Perceptual experience should influence the amount
of stimulus generalization.
2. Discrimination training results in generalization only
to stimuli very similar to the conditioning stimulus
3. Generalization to stimuli dissimilar to the training
stimulus should occur when nondifferential
reinforcement training is used.
4. Preexposure to a common element may reduce
generalization
Ducklings raised in mono-light
18
Monochromatic-Light
(Evidence for #2)
Normal-Light
(Evidence for #1)
Jenkins and Harrison (1960)
19
Nondifferential
(Tone always “On”)
Differential
(Tone, VI, No Tone: EXT
(Evidence for #3)
Mackintosh, Kaye, Bennett (1991)
20

Grp Preexposed:


Lemon Drink

Grp Novel:

No Lemon Drink

Test


Lemon + Saline
Lemon + Sucrose
Conditioning
Lemon + Saline
followed by LiCl illness

Example of Active Generalization
21
Acquired Equivalence: Generalization between
physically different stimuli trained alike


Group 1
 Noise  No Food
 Tone  Food
 Click  Food
Group 2
 Noise  No Food
 Tone  No Food
 Click  Food

Testing
 Click  Shock



Generalization Test:
Tone (more fear) vs. Noise
Testing
 Click  Shock


Generalization Test:
Tone and Noise (equal fear)