History

1
History

• Why do we need to know?
• Because scientific knowledge is cumulative.
• Whatever we are doing now is an extension of what
  people were doing in the past.
• History justifies the current state.
• How far back?
• Ancient Greek

2
But.

• Scientific endeavor did not start until late 19th
  century.
• Who started it?
• Three pioneers
• Ebbinghaus
• Pavlov
• Thorndike

3
Hermann Ebbinghaus

• What did he do?
• He was influenced by British empiricists and
  associationists.
• He was convinced that learning is based on
  association.
• So, decided to conduct scientific investigations
  of how association is formed and how association
  is retained.

4
Scientific method he developed

• Nonsense syllables (CVC, CCC)
• e.g., BOK, BLV
• No prior experience
• You can observe how new associations are formed.
• He used himself as the only subject.
• These experiments are too inhumane.

5
Phenomena

• Retention Curve (Forgetting curve)
• Used percentage saving measure
• original learning - relearning
• saving ----------------------------------- x
  100
• original learning
• Found - negatively accelerating curve.
• Learning curve also showed negatively
  accelerating function.

6
Ivan Pavlov

• He was a physiologist. Studied digestive system.
• Knew philosophers debate about association.
• Accidentally came across scientific method to
  investigate how association is formed.
• Dogs associated environmental cues with food.

7
Method

• Before conditioning
• Conditional stimulus (CS) - Orienting reflex
• Unconditional stimulus (UCS) - Unconditional
  response (UCR)
• During conditioning
• CS - UCS
• After conditioning
• CS elicits Conditional response (CR)

8
Found

• Learning curve - S-shaped
• Slow - fast - slow
• Unlearning
• Presentation of CS alone
• Extinction took place

9
Edward L. Thorndike

• Agreed - learning is based on association
• It is formed based on the consequence of
  behavior.
• Experiment
• A cat was placed into a puzzle box.
• Escape required pulling of a string.
• Did it learn? Yes.
• Became faster as it became more experienced.

10
Explanation

• Phase I
• Trial and error
• The animal emits all kinds of response.
• Phase II
• Law of effect
• A satisfying consequence increases association
  between a stimulus and a response.
• An annoying consequence weakens association
  between a stimulus and a response.

11
They all differed, but...

• They all agreed that learning is based on
  association.
• Three distinct areas were developed.
• Classical conditioning
• Instrumental conditioning
• Verbal learning (Human memory)

12
What followed?

• Two movements
• Behaviorism
• Cognitivism

13
J. B. Watson

• Serious about making psychology a science.
• To do so
• follow positivism.
• use a scientific observation - public
  observation.
• We should only deal with observable events.

14
But, what can be observed?

• Stimulus - Antecedent condition
• Response - Consequent condition.
• As in conditioning, a response (behavior) can be
  predicted from the stimulus.
• Psychologists job
• Cataloging what stimulus condition leads to what
  response.

15
Clark Hull

• Went to U of M and U of Wisconsin
• Became one of the most famous behaviorists.
• Followed Watsons lead
• Pushed the idea that human and animal behavior
  can be explained based on S-R association.

16
But...

• Realized its not all that straight forward.
• A number of variables mediate S-R association.
• Antecedent condition - a glass of water
• Consequent condition - drink
• But, the antecedent condition does not guarantee
  the consequent condition.

17
So..

• Invented intervening variables.
• S - (H, D, K, I) - R
• Behavior can be predicted if we know how these
  variables interact.
• E (H x D x K) I
• Developed a grand scale model of behavior.
• But, still it does not predict much.

18
Edward Tolman

• Deviated from Watsons behaviorism.
• We need to include non-observable variables.
• Experiments
• Learning cannot be observed (latent learning)
• We dont learn response (Place learning)

19
Latent Learning Experiment

• Question
• Is reinforcement necessary for learning?
• Used a complex maze
• Compared rats
• (1) Reinforced from Day 1
• (2) Received no reinforcement.
• Which did better? (1) of course.

20
But why?

• Two possibilities
• Non reinforced rats did not learn anything.
• Non reinforced rats learned about the maze by
  wandering around but did not have any reason to
  complete the maze.
• Which one is correct?
• Include another condition.
• Rats were reinforced only after 10 days of
  nonreinforced trials.

21
Predictions

• If they dont learn anything by wandering
• Performance would gradually improve from Day 11.
• If they learn by wandering but no reason to
  complete the maze.
• Performance would suddenly improve because they
  now have a reason.

22
Results

• They improved performance rapidly.
• Important distinction
• Learning Performance
• Only performance can be observed.

23
Place learning

• T-maze with two starting boxes.
• Condition 1 - Rats turned either left or right
  depending on which starting box they were placed.
• Condition 2 - Rats always turned in one direction
  no matter which starting box they were placed.

24
Which is easier?

• If they are learning response - Condition 2.
• If they are learning where they can find a food -
  Condition 1.
• Results - Condition 1 was easier.
• Conclusion
• We need to study non-observable events.

25
B. F. Skinner

• Thats not the wrong direction.
• We dont need these theories to predict behavior.
• What we need to know - How to control behavior.
• To do so, we need to know how to manipulate
  reinforcement and punishment.

26
Method

• Invented Skinner box.
• Developed the way to create and maintain
  behavior.
• Shaping - based on successive approximation
• Schedule of reinforcement
• continuous schedule
• partial schedule
• Quite successful.

27
But,

• Didnt explain much - especially complex
  behavior.
• E.g., Language learning
• Chomsky
• It cant be based on S-R association.
• Its too generative.
• Many dissatisfied with his approach.

28
Newell and Simon

• Cognitive approach.
• Complex behavior can be modeled by computers.
• Developed a computer program
• General Problem Solver
• Simulate human problem solving
• Impressed other psychologists - It worked.

29
Atkinson Shiffrin

• The same way.
• Adopted information processing approach.
• Assumed humans are computers.
• How do we process information?
• Two types of memory store
• Short term memory
• Long term memory

30
Opened New Era

• We dont have to worry about non-observable
  events.
• We can rigorously specify everything.
• Cognitive approach is the main approach.
• But, still a number of behavioral theories
  remains.