Stimulus Control

1
Stimulus Control

• Determining the conditions under which reinforced
  behavior occurs

2
Definition and Terminology

• Stimulus Control is
• The degree to which a particular behavior occurs
  in the presence of a stimulus and does not occur
  in the absence of the stimulus.
• Discriminative stimulus
• SD the stimulus in the presence of which S?
  the stimulus in the presence of which
  reinforcment does not occur
• reinforcment occurs

3
Stimulus Generalization

• Training to respond in the presence of one
  stimulus will result in responding to similar
  stimuli
• If stimuli can be arranged on a continuum, this
  creates a generalization gradient

Responses
SD
wavelength
4
How to Improve Stimulus Control

• Discrimination Training
• Training with an S? will lead to differential
  responding to the reinforced and nonreinforced
  stimulus
• Sharpens generalization gradient

5
Types of Discrimination Training

• Intradimensional training
• S and S- are on the same dimension
• red light , orange light -
• Interdimensional training
• S and S- are on different dimensions
• red light , tone

6
Basis of Generalization

• Thorndikes Common Elements
• Stimulus is thought of as composed of elements
• Similarity is based on number of shared elements,
  relative to number of elements
• Problems
• Discrimination leads to perfect differentiation
  of similar stimuli

7
Shepards Universal Law of Generalization
8
What Happens with Compounds?

• Train with a stimulus compound of AB
• Responding may be controlled by A
• Responding may be controlled by B
• Responding may be controlled by both

9
Easy-to-Hard Transfer

• Difficult discriminations can be solved by
  starting with an easy discrimination and
  gradually making it more difficult
• Example
• Fading
• Errorless discrimination training (Terrace)
• Easy-to-hard transfer along a continuum

10
Simultaneous Discriminations Choice Behavior

• The Matching Law
• B1 R1
• B1 B2 R1 R2

11
Calculating Matching

• Ratio Schedules
• Ratio is the inverse of the rate of reinforcement
  / response
• FR 6 vs. FR 12
• 1/6 / (1/6 1/12)
• .17 / (.17 .08) .67
• Interval Schedules
• Ratio is the inverse of the intervals
• VI 30 vs. VI 90
• 1/30 / (1/30 1/90)
• .03 / (.03 .01) .75

12
Probability Matching

• Choice of an alternative matches the probability
  that it is the correct alternative
• Consider a 2-choice situation
• Choosing A leads to reward with probability p
  .70
• Choosing B leads to reward with probability p
  .30
• Then, probability of choosing A is
• p(A) .70 / (.70 .30) .70

13
Maladaptive Nature of Probability Matching

• Maximizing is to choose A 100 of the time. Then
  p(reward) .70
• Matching is to choose A 70 of the time. Then
  p(reward) .70 .70 .30 .30
• .49 .09 .58

14
Optimal Foraging

• One problem for a foraging animal is that food is
  distributed in patches, and the animal must
  find a patch before it can get food.
• A second problem is that the amount of food in a
  patch varies, and the animal must decide if it
  has found all the food in a patch.

15
Optimal Foraging

• The time to switch from one patch to another
  depends on the rate of extraction of food from
  the patch and on the amount of travel time to the
  next patch.
• Optimal foraging means switching when the energy
  gain from switching exceeds the energy gain from
  staying.

16
Choice Commitment

• Rachlin Green Experiment
• Immediate Choice -
• Red gt immediate small reward
• Green gt delayed large reward
• Delayed Choice
• white/left gt Immediate Choice
• white/right gt Delayed Reward