Chapter 4 Learning - Western Washington University

Download Report

Transcript Chapter 4 Learning - Western Washington University

Classical Conditioning
Learning: What does it mean to
learn?
• Learning is the single largest area of
Psychology second only to clinical
psychology
I. Classical Conditioning –
learning through association
• Philosophical roots: English Empiricists
• John Locke – Primary and Secondary
Qualities
• David Hume – Reflection , Cause and
Effect
Ideas
• Anything that stimulates the CNS
1. Pavlov and the conditioned
reflex
• The procedure is what distinguishes
classical conditioning from other modes of
learning
Classical Conditioning in America was
Stated by Watson and Rayner (1920)
•
Albert was set on a rug, held by Rayner
and presented with a fuzzy rabbit.
• Watson slammed two sticks together
behind Albert as he reached for the
animal.
• After several trials Albert began to cry
when the rabbit was presented
Watson, Rayner and Little
Albert
The Three Phase Process of Classical
Conditioning
Habituation
• Works at the neuron level to
organismic level.
• Habituation decreases the responses
over time both for magnitude and
frequency per unit time.
• The most basic level of learning.
Requires no reinforcement.
Habituation (cont.)
•
Habituation allows the organism to ignore
irrelevant stimuli.
• Dishabituation is the stopping of
habituation to an attended stimuli
Condition Stimulus (CS)
• A neutral stimulus, can be sound, light or
internal craving.
Uncondition Stimulus (UCS)
• The natural stimulus that drives the CNS
to elicit the reflex.
• In the vernacular of instrumental learning
this hookup of the UCS-UCR is known as
the reinforcement.
Uncondition response (UCR)
• The basic reflex that is become controlled
by procedure.
Reinforcement vs Reward
• One reinforces a response!
• One rewards the animal!
Condition Response (CR)
• The response when driven by the
Condition Stimulus
2. Major phenomena of classical
• Two types of “effective” procedures that
lead the animal to predict up coming
events.
Delay Conditioning: requires only
recognition
Critical Measures: ½ mm movement is a response, Latency to onset of
movement, Latency to peak of response.
3 Characteristics of Delay
Conditioning
• The CS starts before the UCS
• There is a time period between when the
CS in On and the start of the UCS –
known as the Inter-Stimulus-Interval (ISI).
• Both the CS and the UCS co-terminate.
The association
• The explicit pairing of the CS with the UCS
• (in eye-blink, if tone and then air-puff) sets
up the contingency that leads the animal
to predict what is to come.
• This analysis of expectation was put forth
by Rescorla.
Trace Conditioning: requires memory
3 Characteristics of Trace Conditioning
• The CS starts before the UCS
• The CS turns “OFF” before the UCS
comes “ON”.
• The time period between the CS turns
“Off” and the UCS comes “On” is the trace
interval (the inter – stimulus – interval).
• The animal must remember to delay its
response, and respond only at the end of
the trace interval.
Backward Conditioning
Backward Conditioning (cont)
• UCS comes “ON” before the CS.
• UCS goes “OFF” with “ON- set” of CS
• Ineffective for producing a conditioned
response.
Simultaneous Conditioning: very hard
to show.
Formal Classical Conditioning is a
Three Stage Process
• Habituation – Paired Training - Extinction
Pavlov’s study
• Condition stimulus, CS, was the sound of
a metronome (will become the condition
reinforcer)
• The Uncondition Stimulus (UCS) was
meat powder (the reinforcer).
• The uncondition response was drops of
salivation.
• Condition response was salivation.
Trial - By - Trial Presentation
• Trials are presented singularly and
continues until some criterion is met or a
fixed number of trials completed.
Paired trials
• When both the CS and the UCS are
presented in a fixed order and time frame
• (an explicitly paired trial).
CS Alone Trials
• Periodically only the CS is presented
• Known as CS alone trials.
• Measure the effectiveness of the CS to
drive the behavior
Statistics
• Mean and SD change in the timing or
magnitude of the paired trials.
• Mean and SD of the CS alone trials
Results
• In the beginning trials the CS does not
elicit the CR
• Across trials there is a change in the ability
of the CS to drive the CR. The change
increase as the trials progress.
• The CR looks like the UCR, but micro
analysis of the two behaviors will show
clear difference
Results of Explicit Pairing CSUCS
• By the end of the trials, both for paired and
CS alone trials, the CS is able to elicit the
CR.
• Magnitude increases, time shortens or
frequency changes for the behavior
How does one know that one has
stimulus control?
• Use Extinction trials.
Present the trials in the manner of
Habituation (but is not habituation)
• Repeated presentation of the CS alone will
drive the changes that were learned during
CS presentation to zero.
EXTINCTION
• Note at the beginning of day 2, the first
response is larger than last response on
day one. Called spontaneous recovery.
Generalization vs discrimination
• Generalization is the tendency to respond
to like stimuli the same way.
• If a rabbit is trained on a 1200Hz tones, to
a critiereon, his percent correct for a 800,
or a 1600Hz tone will be less than the
1000Hz tones
• See next slide
Generalization gradient
Discrimination training
• The correct stimulus is explicitly paired
with a given reinforcement (see Mauk and
Ruiz slides 65- 71 below).
• Or two different responses are paired with
two explicitly different stimuli.
The vast array of stimuli and condition
responses
Eyeblink Studies
• First done by Ernest Hilgard in humans,
dogs and monkeys 1928 – 1936).
• Modern Studies were started by Isidore
Gormenzano
• A Model Animal – The Rabbit, its blink rate
is 1 or 2 time a minute.
Eye lid Conditioning
• In the U.S. Ernest Hilgard studied eyelid
conditioning in three major papers, 1929,
1932, 1936. He used dogs, monkeys and
humans. He set the procedures for all
neurosciences and their use of CC
Modern Studies started in 1964
Gormaenzano – modern study of eyelid
conditioning
Mauk studies as representational of
eyelid conditioning in the rabbit.
• Demonstrated, by brain removal studies in
the rabbit, that classical conditioning could
be mediated by the cerebellum alone.
Cerebellar brain circuit for mediating
classical conditioning.
What Is The Behavior Measured
• A very sharp distinguishable criterion of
what constitutes the movement of an
eyelid.
Eyelid Movement Changed to an
Electrical Potential
• An Stainless Steel Wire (0.007 in. by
0.018 cm) Extended From the suture loop
in the eyelid to a phototransistor
potentiometer.
Optional slide
• An infrared emitter is bounced off the eyelid.
• The reflected beam passes through two small
Polaroid filters, one fixed and one operated by
the eyelid closure.
• The two Polaroid filters are set so that infrared
light passes through both filters to an infrared
receiver when the eyelid is up.
The receiver turns the infrared light into volts,
depending on the amount of infrared light
received.
When the rabbit blinks its eye the wire is pulled
and rotates the moveable Polaroid filter. This
movement reduces the amount of light
transmitted, lowering the voltage.
• The movement of the eyelid, closure (distance in
mm), was calibrated to voltage drop by the
potentiometer.
• A movement of ½ mm of eyelid was defined as
an eyeblink.
CS
• Sine Waves of Various Frequencies
presented at an intensity of 80dB (SPL)
UCS Airpuff
• UCS was calibrated to present to the eye
and airpuff of 2 N/cm2 delivered to the
cornea of the rabbit through, a 1 mm
tuberculin syringe, positioned
approximately 1 cm from the cornea.
Criterion Learning
• How Many Trials to Learn to a Fixed
Criterion: Eight Correct Closures Out of
Ten.
Fixed Trial Procedure
• Percent of Closures in X Number of Trials
Most Frequently Used Procedure
• Total of 108 Trials Divided in 12 Blocks
• First Trial of each block is a CS Alone Trial
• ACROSS 108 TRIALS: 12 CS ALONE
TRIALS AND 12 BLOCKS OF PAIRED
CS/UCS TRIALS
Critical Measures: ½ mm movement is a response, Latency to onset of
movement, Latency to peak of response
What IS MEASURED
• NUMBER OF TRIALS IN WHICH AN
EYEBLINK WAS MADE.
• LATENCY to ONSET: Time it takes from
CS Onset to beginning of eye-blink
(previously defined in terms of mm of
movement).
Mauk & Ruiz
• The differential learning (conditioning) with
two different ISIs.
Hypothesis:
• The rabbit can learn to discriminate
between two different ISIs.
• The rabbit is learning to time its eye-blink
to two different inter – stimulus – intervals.
• The rabbit is learning to time its behavior
of eyelid closing!
Delay Conditioning: Two or more ISIs
• Manipulating the ISI is one of the
independent variables.
The Independent Variables: CS1 and
CS2 and the position of the US (UCS)
Note
• The US, in Mauk’s terms, in the above
slides, must be manipulated to coterminate with the CS. The UCS (air-puff)
is always the in duration, 100ms. It occurs
with the last 100 ms of the CS to coterminate.
The Dependent Variable
• Getting the eye closed at the appropriate
time
The Dependent Variable: The wiggly
lines above CS1 and CS2
Acquisition of Conditioned Eyelid
Closure (two different ISIs)
Latency To Onset of Eyelid Closure
Mauk & Ruiz: Six Different ISI to
Six Different tones
Note the Difference in Behavior of
the Eyelid to Different ISIs
Three Conditioned Eyelid Closures
Elicited by Three Different ISI
Associated Tones
TRACE CONDITIOING
1. Acquisition of conditioned
• What is measured or counted
• Number of blinks across trials for CS alone
• Number of blinks across trials for paired
CS/UCS
• Latency to Onset.
Neural Response
Berger et al
High-Order Conditioning
• One can use the CS as the UCS
CS USED AS A UCS
OVER SHADOWING
• The salience of one cue is greater than
that of the other cue.
• The more salient cue overshadows the
less salient cue. The less salient cue can
not be used to predict the contingency.
TEST INDIVIDUAL CUE
TEST INDIVIDUAL CUE, LIGHT
BLOCKING
The behavior in Phase III
• In Blocking, when tested in Phase III,
Group I does not demonstrate a CS to the
sound. The light has control over the
behavior in Phase II and the sound
becomes blocked.
• When tested in Phase III, Group II shows
CSs to both sound and light.
Test for Sound Group I
TEST GROUP II
Condition fear in a dog. Sound of the horn comes to
signal the shock. The dog jumps over the divider.
• Panel 1 shock occurs with no cue. Dog
jumps over barrier only when the shock
appears. Non-cued shock increases fear
and anxiety.
• Panel 2 a loud tone is presented before
shock appears. A contigency is set up
The Condition Emotional Response CER
• The sequence described above is how
emotional responses get tied to external
cues.
• The reduction of anxiety is a positive state
of affairs for the animal, i.e., a positive
reward.
• In the above dog case the dog avoids the
negative state off affairs, fear, leading to
anxiety.
• between the shock and the tone. The tone
comes to predict the shock. Anxiety is high
before tone is on.
• Panel 2 the dog comes to jump, before the
shock when it hears the tone. Anxiety
starts to drop as the dog starts to jump.
Note the shock has not appeared.
• Panel 3 Dog is jumping, shock appears,
anxiety goes down.
• Drug habits in humans leads to the
paraphernalia of the habit to become
condition stimuli for the high response,
coupled with avoidance behavior of the
withdrawal symptoms. The withdrawal
symptoms leading to anxiety coupled to
fear.
What is learned in Classical
Conditioning?
• The animal learns an “if – then”
contingency relationship between the CS
and the UCS.
• The CS increases the probability that the
UCS will follow.
• The animal has increasing predictability
that the UCS will follow the CS, thus
gaining control off the situation!