Behavioral Learning 1

1
Behavioral Learning (1)

• ...any technology of control will be bought,
  adapted, and monopolized by the powerful for
  their own not always benevolent purposes.

2
Why Study Behavioral Learning?

• Train a pet
• Clinical significance
• Control your friends (?)

3
Habituation

• The simplest of all behavioral learning forms
• A decline in the tendency to respond to familiar
  stimuli
• The faster we habituate the faster we recover and
  the slower we habituate the slower we recover
• Why habituate? Darwin would say...It narrows the
  range of stimuli that elicit escape reactions,
  thus freeing resources to focus on new stimuli
  that might be dangerous

4
Habituation to Conditioning

• In habituation, one learns a familiar event but
  doesnt learn to associate it with anything
• Association is the basis for Conditioning
• Ivan Pavlov (1849-1936) is credited with the
  advancement of conditioning

5
The Conditioned Reflex

• Repeated bell-food pairings led to salivation
  when the bell was presented alone
• Pavlov described this phenomena in terms of
  unconditioned and conditioned reflexes
• Unconditioned stimuli (US) is normally associated
  with an unconditioned response (UR)
• A conditioned stimulus (CS) is paired with a (US)
• Giving a number of presentations the (CS) will be
  perceived as a (US) and lead to a conditioned
  response (CR)

6
Extinction

• The (CS) will gradually disappear of repeatedly
  presented without being enforced
• Reconditioning can occur with fewer reinforcement
  trials
• Thus, the learning wasnt completely gone but
  rather masked

7
Generalization

• Will a response be given only if the stimulus is
  identical?
• A dog conditioned to a tone of 1000 Hz will give
  a response when the tone is 900 or 1100 Hz
• However, there is a generalization decrement such
  that a weaker (CR) will be elicited if not the
  original (CS)

8
Discrimination

• Even if a stimulus is similar we may not want to
  generalize our behavior to it (may be dangerous)
• Thus, we need to discriminate
• A dog is conditioned to a black square. The dog
  will also salivate to a gray square
  (generalization). After withholding
  reinforcement to the gray square the dog will
  gradually discriminate and only respond to the
  black square

9
Blocking

• When a stimulus is redundant (information that an
  organism already has) it will not become
  connected to the CS
• Group I - (stage 1) noise-shock, (stage 2)
  noiselight-shock, (test) light alone, (result)
  no conditioned fear
• Group II - (stage 1) no stimulus, (stage 2)
  noiselight-shock, (test) light alone, (result)
  conditioned fear
• Once theyve discovered a stimulus that signals
  the US they no longer connect other stimuli to
  the US
• Proactive Interference?
• Bases for Prejudice

10
The Drug Effect

• Opiates (morphine or heroin) have effects such as
  euphoria, relief from pain, and relaxation (UR)
• The needle becomes the CS
• The sight of the needle produces depression,
  restlessness, and increased sensitivity to pain
  thus increasing the likelihood of an overdose
• The CR is actually the opposite of the UR
• Given a different environment the Drug Effect is
  weaker

11
Classical to Instrumental Conditioning

• Instrumental conditioning is also called operant
  conditioning
• Differences
• Instrumental - reinforcement (reward) depends
  upon the proper response. The response must also
  be selected from a sometimes large set of
  alternatives
• Classical - the US is presented regardless of
  behavior

12
Instrumental Conditioning

• Edward Thorndike (1874-1949) is credited with
  Instrumental (Operant) Conditioning
• The idea was developed by observing hungry cats
  in a Puzzle Box
• Through random movements the cat would hit the
  trigger to open the door
• Give enough trials and the cat would trigger the
  door right away

13
Law of Effect

• According to Thorndike, some responses get
  strengthened and other weakened as learning
  proceeds
• The response determines whether the tendency to
  perform is strengthened or weakened
• If followed by a reward it will be strengthened
  if followed by a the absence of reward (or
  punishment) it will be weakened
• No intellectual process required

14
Skinners Distinction

• In Classical, the animals behavior is elicited
  by the CS (from the outside)
• In Instrumental, reactions are voluntary (emitted
  from within)
• Tendencies for reactions are strengthened or
  weakened
• Measured response rate with the Skinner box

15
Skinners Terminology

• Appetitive Stimulus Positive Reinforcement
• Aversive Stimulus Negative Reinforcement

16
Shaping

• Method of Successive Approximations
• Start by reinforcing a general behavior
• Change the reward placement so there needs to be
  a more specific behavior
• Change the reward placement so there needs to be
  another more specific behavior
• Your dog puts the paper on the table

17
Partial Reinforcement (Schedules)

• Fixed-Ratio Schedule - produce a specified number
  of responses for every reward
• Variable-Ratio Schedule - reinforcement comes
  after a certain number of responses, but that
  number varies irregularly
• Fixed-Interval Schedule - Reinforced after a
  certain interval (i.e. every 2 min)
• Variable-Interval Schedule - Interval varies
  irregularly around some average period (i.e. 4
  min)

18
Question

• What kind of reinforcement is better? Partial or
  Continuous
• Both receive the same number of trials but no the
  same number of reinforcements
• Partial seems to be better, but which schedule?
• Ratio engender greater resistance to extinction
  if they are variable

19
Aversive Conditioning

• What is the reward in aversive reinforcement?
• It is not absence of the aversive condition that
  is the reward, but rather the absence of the
  aversive condition when it is expected

20
Behavioral Learning (2)

• Cognitive Interpretations

21
Question

• What are test animals actually learning? Simple
  associations or causal links?
• Behaviorists - learning is in terms of
  associations, either between events, or action
  and consequences. Learning can only be achieved
  by doing, thus performance is indispensable
• Cognitive psychologists - learning is in terms of
  causal links

22
Cognitive View - Classical Conditioning

• Hilgard (1938) - Two lights (left right), when
  one light became brighter there as an airpuff
  delivered to one eye causing a blink.
  Conditioning occurred.
• When subjects were told that no airpuff would
  result given a right light the conditioning in
  that right eye was immediately extinguished
• Instruction works thus, subjects need to be
  aware of the stimulus-response relationships

23
Support for Cognitive View (2)

• Gleitman (1963)
• Rats were ferried from one end of a large room to
  another in transparent trolley cars
• Latter tests showed that they had learned
  something about the general features of the room
  even thought they had not performed any relevant
  responses

24
Relation Between CR and UR

• Are the two the same?
• Pavlov believed that they are the same
• However, his own data show that they look the
  same but under are not identical
• There was a greater amount of salivation measured
  under UR conditions compared to CR conditions
• Thus, the CS serves as a preparatory signal but
  does not become a substitute for it (cognitive
  link)

25
Cognitive View - Instrumental Conditioning

• Evidence for Act-Outcome Associations
• Rats were run in an enclosed maze that had a
  black end-box on one side and a white end-box on
  the other side, both had food
• Later, rats were placed into each box w/out
  running the maze. In the black box they found
  food and in the white box they were shocked
• When put back in the maze they ran to the black
  box
• The rats combined two experiences this doesnt
  fit into the instrumental learning framework

26
Clinical Considerations

• Maladaptive Learning Application of the
  Conditioning Theories

27
Learned Helplessness

• How do people reach the state of helplessness and
  depression?
• Experiment - Two groups of dogs who received
  electric shocks. Dogs were linked such that
  Group 1 could turn off the shock by pushing a
  panel. Group 2 had no power but rather received
  the same shocks as Group 1 (same fate).
• Given a new situation Group 1 dogs learned to
  push the button quickly. Group 2 dogs ended up
  becoming passive. They lay down and took
  whatever shocks were delivered
• Group 2 dogs had learned helplessness

28
Learning Fear

• While fear and phobias may be generated by
  conditioning in some cases, there seems to be
  more at work
• Fear of the Dentist may be conditioned but fear
  of snakes and spiders may be cause by a cognitive
  association
• Someone tells you a story of a tick that ate
  someones brain will make you fear ticks
• Overcoming fear and depression are difficult

29
Treatment Methods

• Desensitization - brief approaches
• Modeling - Therapist approaches object then
  subject
• Flooding - Rapid prolonged approach to the
  stimulus
• Operant Shaping - Rewarded for a steady approach
  to object

30
Treating Panic Attacks

• Behavior modification has been helpful in
  treating panic attacks
• The problem is a feedback loop between the
  initial chance symptom and the patients fearful
  interpretation of it
• Teaching relaxation and breathing exercises
  alleviate the physical symptoms

31
Treating Self-Injurious Behavior

• Maladaptive behavior is often associated with
  anxiety (caused by need for attention, feeling of
  vulnerability, confronting difficulty...)
• Replace self-injurious behavior with relaxation
• Relaxation drugs are often used (nitrous oxide or
  anesthetic)
• Thus, relaxation becomes associated with anxiety

32
When Conditioning Fails

• Token economies have been experimented with for a
  number of years
• Rewards for good behavior are tokens which can be
  exchanged for things
• Often subjects will not perform unless directly
  rewarded. Thus, behavior is conditional
• In real life rewards are not always given for
  being good

33
Question

• Why dont we use conditioning more often in
  humans?
• Why does conditioning often fail in human
  subjects?