Instrumental Conditioning

1
Instrumental Conditioning

• What is?
• Situation where you have to figure out what
  response produces a reinforcer.
• Example

2
Difference between classical and instrumental
conditioning

• 1. Stimulus
• Classical - Explicit
• Instrumental - Sometimes explicit and
  sometimes not.
• 2. Response
• Classical - Involuntary response (no choice)
• Instrumental - Voluntary response

3

• 3. Reinforcer
• Classical - Automatic delivery
• Instrumental - Contingent on a correct
  response
• Seems different, yet similar

4
Similarity

• In both situations, you respond in anticipation
  of a consequence.
• Classical - S -----R
• 
  UCS
• Instrumental - S------R
• Reinforcer
• Learning is very similar

5
But, I dont agree with that..

• Classical conditioning ---- More automatic.
• Instrumental conditioning ---- Controlled.
• Recent studies show that these two processes are
  influenced by different variables

6
What is associated in instrumental conditioning?

• Two possibilities
• 1. S - R
• Reinforcer strengthens the association but not a
  part of association.
• Wrong - Less valuable reinforcer created a
  disappointment

7

• 2. S - R - R
• We learn to respond because the response predicts
  the reinforcer.
• We respond in anticipation of the reinforcer.
• Experiment

8
Colwill and Rescorla (1985)

• Organisms can learn to associate specific
  reinforcers to specific responses.
• Trained animals using two kinds of food
• Reinforcer 1 - lever pressing
• Reinforcer 2 - chain pulling
• Outside of experiment - They were fed with only
  one kind of food (e.g., Reinforcer 1).

9

• Animal got sick of it. Wanted to eat other food
  too.
• So, they produced the response that produced the
  other kind of food in the experiment.
• Animals anticipate a reinforcer based on a
  response.
• So, animals associate S-R-R

10
Colwill and Rescorla (1986)

• Devaluation of reinforcer - Reduce responding.
• Devalued a reinforcer with an injection of
  lithium chloride.

11
Other associations?

• They also form association between a response and
  a neutral stimulus.
• Similar to two phenomena in classical
  conditioning
• Sensory preconditioning
• Second-order conditioning

12
Analogous to sensory preconditioning

• Phase 1 Lever press - Noise
• Phase 2
• Experimental Noise - Food
• Control Light - Food
• Phase 3
• Lever was reintroduced

13

• Results
• Experimental group pressed the lever more
  because
• Lever press - noise - reinforcer

14

• 1. Lever - Noise
• 2. Noise - Food
• 3. Lever - Lever press
• Similar to .
• 1. CS1 - CS2
• 2. CS2- UCS
• 3. CS1 - Response

15
Analogous to Second Order conditioning

• 1. Noise - Food
• 2. Lever - Noise
• 3. Lever - Lever press
• Similar to...

16

• 1. CS1 - UCS
• 2. CS2 - CS1
• 3. CS2 - Response
• This is the way a secondary reinforcer develops.
• Response - Money - Reward

17
So, association can form..

• 1. Stimulus - Response - Reinforcer
• 2 Stimulus - Response - Neutral Stimulus which
  will anticipate a reinforcer.

18
What is stimulus?

• Complicated by
• Stimulus generalization
• Stimulus discrimination

19
Stimulus Generalization

• We respond to more than one stimulus - make it
  difficult to predict what produces response.
• Follows generalization gradient.
• Some are steep
• Some are flat

20

• Two kinds of generalization
• Excitatory - to respond
• Inhibitory - not to respond

21

• To make it more predictable - you can use
• Stimulus discrimination training
• SD S?
• But, still difficult to predict
• Response Excitatory - Inhibitory

22
Peak Shift

• Train 550 nm
• Respond 550 nm
• Train 550 nm 590 nm
• Respond 510 nm

23
Relational Responding

• Another source of difficulty.
• Instead of responding toward a specific stimulus,
  animals can respond based on the relationship
  between SD and S?.
• 550 nm 590 nm
• 510 nm 550 nm

24

• Called transposition.

25
More than one dimension

• Animals are very selective about which dimension
  to pay attention to.
• Example
• Blocking and Overshadowing

26
Miles and Jenkins (1973)Overshadowing

• SD - Bright key tone
• Condition 1 - S? Dimly lit key
• Condition 2 - S? Bright key
• Which did they pay attention?
• Light? Tone? Or both?

27
Multiple Successive Discrimination

• 1. Training
• 2. Reversal shift
• 3. Non-reversal shift
• 4. Intradimensional shift
• 5. Extradimensional shift

28
Category learning

• A stimulus can be an abstract concept.
• Experiment

29
Wasserman et al. (1988)

• Trained pigeons
• Key 1 - one of 10 cat pictures.
• Key 2- from the second set of cat pictures.
• Key 3 - one of 10 flower pictures.
• Key 4 - from the second set of flower pictures.

30

• Discriminated Key 1 and 2 from Key 3 and 4.
• Discriminating Key 1 and 2 or Key 2 and 3 was
  difficult.
• But, the stimulus can be an abstract category.

31
So, what is the stimulus?

• Very complex.
• You need to be careful as to what stimulus your
  subject is responding to.

32
Errorless discrimination

• Terrace
• - developed a technique to reduce errors
  while going through discrimination training.
• - fading
• - principle - make SD and S? very different at
  the beginning. Then, gradually make them
  similar to each other.

33

• Pigeons made no error.
• Can be applied to human.
• Egeland (1975)
• Trained children to distinguish similar
  letters.
• Acquired distinctiveness.