Transcript PSY 402

PSY 402
Theories of Learning
Chapter 10 – A Synthetic Perspective on
Instrumental Learning
Avoidance Learning

Avoidance learning is puzzling because the
reinforcer is not obvious.



What sustains running to the other side in a
shuttlebox?
How can something that does not happen be a
reinforcer?
Mowrer’s two-factor theory was the first
explanation.
10.1 The shuttlebox—an apparatus used in studies of avoidance learning in rats
Side with
shock
Safe side
Two-Factor Theory

The first factor is Pavlovian fear conditioning.



A light is a CS that warns of a shock.
The CR is fear.
The second factor is reinforcement of
avoidance behavior.

The avoidance response terminates the fear (CR)
and the experience of relief (negative reinforcer)
rewards the behavior.
10.2 The two factors in the two-factor theory of avoidance learning
Acquired Drive Experiment

Miller presented a shock in the presence of a
white chamber.




Later, rats could exit from the white side to a
black side through a door.
Later rats had to run in a wheel to open the door
to the black side.
Later rats had to press a lever to open the door to
the black side.
The later responses were reinforced by relief.
Clinical Applications




Drug-taking and alcohol use may be
reinforced by escape from negative affect.
People with panic disorders such as
agoraphobia avoid the places that evoke
anxiety.
OCD hand-washing reduces fear of
contamination.
Bulimic purging reduces fear of getting fat.
Problems with Two-Factor Theory

The strength of the avoidance response should
be correlated with the strength of the fear, but
it is not.




Dogs avoiding shock don’t look worried.
Maybe after extensive training, habit takes over.
Learning that turns off the CS but not the
shock should reinforce behavior, but doesn’t
unless it is related to an innate behavior.
Learning varies with the type of response.
10.3 Fear of the warning signal declines after extensive avoidance training
Fear goes
away with
more training
More fear
10.4 A Sidman, or “free-operant,” avoidance procedure
These are the
responses
10.5 The effect of turning off the warning signal depends on the avoidance behavior being trained
Rats find it easier to learn
wheel running as an
avoidance behavior
Terminating the warning reinforces avoidance in
both situations, but more for running the wheel
10.6 Avoidance behaviors are not equally learnable—another example of preparedness
Species Specific Defense Reactions

An avoidance behavior is easy to learn in the
lab if it is similar to a SSDR.



Thigmotaxis – wall-seeking behavior.
Rats learn to jump to a platform along the wall
but not in the middle of the space.
Mice from Eastern vs Western Washington
freeze in response to different predators.

Neither freeze to garter snakes or squirrels.
10.7 Annual rainfall versus deer mice adaptation in eastern Washington
Freezing is a CR (Respondent)

Rats rapidly learn to freeze in order to avoid
shock, but the behavior is a CR not operant.




100% of rats learned to freeze to avoid shock.
Rats punished for freezing never learned not to
freeze.
Freezing was elicited by the punishment – the
box became a CS eliciting the freezing as CR.
Pavlovian learning is important to avoidance.
10.8 Freezing is actually a respondent rather than an operant (Part 1)
Punishment of freezing doesn’t
complete suppress the
behavior.
10.8 Freezing is actually a respondent rather than an operant (Part 2)
Rats stay frozen for a long
time after a shock.
Endorphins

CS’s associated with shock also elicit
endorphins (natural pain killers).



These inhibit recuperative behaviors (licking a
wound) that would interfere with evading a
predator.
The actual SSDR elicited depends on the situation
as well as the species (flight, freezing, burying).
Fear also inhibits pain.
10.9 Perceptual-defensive-recuperative model of fear and pain
CS Termination is Important


Animals are able to learn non-SSDR
behaviors under certain circumstances.
Termination of the CS aids such learning
because it provides feedback about the correct
behavior for avoidance.



Feedback stimuli become conditioned inhibitors.
The avoidance response becomes an inhibitor too.
Not performing the response predicts the US.
Expectancies About Shock

Seligman & Johnston proposed that an
expectancy is formed that response=no shock.


This idea is inconsistent with learning theory.
DeHouwer found that changing an expectancy
to response=shock (when no signal is present)
does not change the avoidance behavior.

The idea that avoidance behavior is a negative
occasion setter (response-no shock) fits better.
10.10 Shock expectancy ratings: Avoidance testing
R1 or R2 mean shock when A or B
are off, but no-shock when they
are on. The expectancy is not
associated with the key, but with
the signal (A or B).
Learned Helplessness




Overmeier & Selgman found that dogs have
difficulty learning avoidance behavior after
being exposed to inescapable shock.
Step 1 – expose dogs to 3 conditions:
escapable shock, inescapable shock (yoke), no
shock
Step 2 – escape training
Results – yoked dogs did not learn to escape.
10.11 The learned helplessness effect