Transcript Trial

Lecture 5 – Chapter 4
Psych 301 Exam
Learning Spanish in High School…
Did you use flashcards? …learning the La Calavera
means skull
Let's say it took 10 trials to learn this association
When did the bulk of the learning actually take
place?
In the first or last trials?
In the middle?
Most of the associative learning occurs during
the first few trials
During the last trials, very little is added to the
strength of the association
Rescorla/Wagner Model
attempted to explain the growth of association
mathematically with the equation
Rescorla-Wagner Model
What is a model?..should be able to:
Explain behavior & also Predict it!
conditioning
blocking
Extinction
Simple mathematical account of how associative
strength – conditioning builds up over the course
of a learning experiment  learning curve
During Conditioning – CS/UCS
Learning Curves
Rescorla-Wagner Model
Mathematical Equation
Vn = c (Vmax – Vn)
= Delta (change)
V = Strength of Association
n = trial
c = constant (speed of conditioning - 0-1)
Vmax = asymptote (no more trials- max strength of association)
Vn = the strength of association at the beginning of the trial
Predicts the precise associative strength in each trial
V = Strength of Association
n = trial
c = constant (speed of conditioning – 0-1)
Vmax = asymptote (no more trials- max learning)
Vn = the strength of association at the beginning
of the trial
Amount of conditioning
for 1st trial
Vn = c(Vmax – Vn)
Add
V = Strength of Association
Can also account for extinction
n = trial
C = constant (maximal performance)
Vmax = asymptote (best learning – if a CS with UCS is paired)
Vn = the strength of association at the beginning
the trial
Vn = c(Vmax – Vn)
…no more pairing
of UCS with CS
5
??
V5 = 0.30 (0-0.76) = -0.21
First Conditioning Trial
Trial
1
c (Vmax - Vn) = ∆Vn
.5 (100 - 0) = 50
Associative Strength (V)
100
80
60
50
40
Vall
Vn
20
0
0
0
1
2
3
4
Trials
5
6
7
8
Second Conditioning Trial
Trial
2
c (Vmax - Vn) = ∆Vn
.5 (100 - 50) = 25
Associative Strength (V)
100
80
75
60
50
40
VVall
n
20
0
0
0
1
2
3
4
Trials
5
6
7
8
Third Conditioning Trial
Trial
3
c (Vmax - Vn) = ∆Vn
.5 (100 - 75) = 12.5
Associative Strength (V)
100
87.5
80
75
60
50
40
Vall
V
n
20
0
0
0
1
2
3
4
Trials
5
6
7
8
4th Conditioning Trial
Trial
4
c (Vmax - Vn) = ∆Vn
.5 (100 – 87.5) = 6.25
Associative Strength (V)
100
87.5
80
93.75
75
60
50
40
Vn
Vall
20
0
0
0
1
2
3
4
Trials
5
6
7
8
5th Conditioning Trial
Trial
5
c (Vmax - Vn) = ∆Vn
.5 (100 - 93.75) = 3.125
Associative Strength (V)
100
87.5
80
96.88
93.75
75
60
50
40
Vall
Vn
20
0
0
0
1
2
3
4
Trials
5
6
7
8
6th Conditioning Trial
Trial
6
c (Vmax - Vn) = ∆Vn
.5 (100 - 96.88) = 1.56
Associative Strength (V)
100
87.5
80
96.8898.44
93.75
75
60
50
40
Vn
Vall
20
0
0
0
1
2
3
4
Trials
5
6
7
8
1st Extinction Trial
Trial
1
c (Vmax - Vn) = ∆Vn
.5 (0 - 99.61) = -49.8
Associative Strength (V)
Acquisition
100
80
60
40
Vn
Vall
20
0
0
1
2
3
4
5
6
7
8
Trials
99.61 – 49.8 = 49.8
Associative Strength (V)
Extinction
100
99.61
Vall
V
n
80
60
49.8
40
20
0
0
1
2
3
Trials
4
5
6
2nd Extinction Trial
Trial
2
c (Vmax - Vn) = ∆Vn
.5 (0 - 49.8) = -24.9
Associative Strength (V)
100
93.75 96.88
87.5
80
98.44 99.22 99.61
75
60
50
40
Vn
Vall
20
0
0
0
1
2
3
4
Trials
5
6
7
8
Associative Strength (V)
Extinction
100
99.61
Vall
Vn
80
60
49.8
40
24.9
20
0
0
1
2
3
Trials
4
5
6
Importance of Model
•Only mathematical model of behavior
•Best Theory of Classical Conditioning
•Early trials more “learning occurs”
Kamin (1969)…The amount of conditioning on any
trial is determined not only by the UCS (shock) but also
the expectation that the CS will occur…no surprise less
learning (Kamin)…expect a shock…no learning
PS: this model only works when the CS precedes the UCS
It does not work so well for the backward sequencing of stimuli
Classical Conditioning
UCS
S ------ S
CS
Association between
CS & UCS
UCR
Why not this instead?
UCS
CS
UCR
Test the hypothesis
that the association is
between CS & UCS vs
CS & UCR
food
UCS
tone
CS
UCR
Try and affect association between CS & UCS
then measure conditioning
Post Conditioning Devaluation: alter value
of UCS after conditioning
Test the hypothesis
that the association is
between CS & UCS vs
CS & UCR
Holland & Straub (1979)
1. Noise + Food activity
(CS)
(UCS) (UCR)
Noise  activity
food
UCS
noise
CS
UCR
Activity
PCD
No Activity
2. Post Cond. Devaluation
Food + Rotation
3. Noise  ??….
Significant decrease in movement!
Pavlov: CS becomes the UCS  Stimulus Substitution
“Substitution Theory”
Jenkins & Moore (1973)
Pigeons & autoshaping
(Key LT + food)
• food: open beak/closed eyes
• H20: closed beak/eyes open
Key Light + Food
Key Light + Water
Just Key Light (water)
…”drink key light”
Just Key Light (Food)
…eat the Key Light”
Classical Conditioning
S ------ S
Association between
CS & UCS
types of associative learning
Operant Conditioning
(instrumental learning)
S ------ R
Direct link between CS & Response
Operant Conditioning vs Classical Conditioning
In Classical Conditioning, the dog’s response has no
consequences; it produces no change in the
environment.
The dog gets the food after the bell is rung whether
or not he salivates to the bell. His behavior
…response doesn’t matter.
In Operant Conditioning, the dog has to pull the
latch to get the food. His behavior does matter.
Operant Conditioning
The nature of behavior is determined by its
consequences whether positive or negative
Learning occurs through reinforcements
S--------R
So, in OC the reinforcer is contingent of the response
=
So, in CC the reinforcer is paired with a stimulus
+
Classical Conditioning:
A theory that attempts to explain how organisms
develop involuntary responses to particular
stimuli in the environment
B.F.Skinner: (respondent behavior elicited by a
known stimulusUCS)
Operant Conditioning:
A theory that explains how
people develop voluntary behavior
B.F.Skinner: (operant behaviornot elicited by a stimulus)
VH1 “Behind the Science”
Edward Lee Thorndike
(1874-1949)
-born Willamsburg, MA
- mental telepathy in children
- studied at Harvard (James)
- hatched chicks in bedroom
then James’ basement
- Ph. D. Colombia University
“Animal Intelligence” - Cats
- ”Puzzle Box”
- The Law of Effect
- First experimental study
of rewards
-1912 Prez of APA
507 pubs
Do animals posses intelligence?
Applied animal to human education
Intelligence tests
Romanes: Animal learn just like we do….
Naturalistic Observation - Anecdotal Evidence
Thorndike did not buy this!..just coincidence
Anthropomorphizing
“How long would it take an animal to discover
how to get out of a box?”
Would this time decrease with exposure to
task?
Large box (crate)
Place hungry cat in
Placed food just outside box
Cat could see out
Could get out by pulling string attached to latch
Some of Thorndike’s actual cat puzzle boxes
Puzzle Box
- Scramble
-Claw & bite
-Stumble on string
-Latency shorter
-2nd trial…
again scramble…
Thorndikes Conclusions
1. Behavior is highly variable
2. Course of learning is erratic
3. Gradual improvement over trials
4. No suggestions of “sudden insight”
5. No suggestion of any “mental process”
6. Not a rational understanding but an association
with the escape with the food reward
Thorndike
theory of connectionism:
Through experience neural bonds (connections) are
formed between stimuli & responses
Intellect facilitated neural bonds
higher intellect = forms more bonds
lower intellect = forms less bonds
Genetic
Content of intellect (f) experience
Law of Effect (1911)
The presentation of a reward
strengthens the connection between
the response that precedes it and
the stimuli presented at the time
Behavior that produces satisfying consequences
will be repeated, and behavior that produces
dissatisfying consequences will not be
repeated
VH1 “Behind the Science”
Burrhus Fredric Skinner
(1904-1990)
- BA in English
-Wanted to be a writer
-Master’s In Psych – 1930
-Ph.D – 1931 – Harvard
-Dept. Chair, Indiana 1945
-Harvard, 1948
-Father of Operant Conditioning
-Radical Behaviorism
Skinner strongly championed Thorndikes Law of Effect
Invented a number of devices for studying OC (“operating
on the environment”)– Skinner Box, Teaching Machine
Radical Behavorism:
Not a science but a philosophy…
Did not rule out mentation,
introspection, self-knowledge
“…things that take place in the
private world of the skin…” are
observable…but cannot infer causes
of behavior
“The organism becomes a person [i.e., a unique individual] as it acquires a
repertoire of behavior under the contingencies of reinforcement to which
it is exposed in its lifetime. The behavior it exhibits at any moment is
under the control of a current setting. It is able to acquire such a
repertoire because of processes of conditioning, to which it is susceptible
because of its genetic endowment. “
(Skinner, 1974)