Lecture 6: Extinction and Shaping

Operant extinction

• When a previously reinforced behaviour is emitted and not followed by reinforcement, the individual is less likely to engage in the behaviour again.

• Response was increased in frequency through reinforcement – now ceasing to reinforce the response causes it to decrease in frequency.

Factors influencing the effectiveness of operant Extinction

1. Control of reinforcers for the behaviour that is to be decreased

2. Combining extinction with positive reinforcement

3. Setting in which extinction is carried out

4. Use of instructions or rules

5. Schedule of reinforcement before extinction is carried out

6. Behaviour being extinguished might get worse before it gets better

7. Elicited Aggression

8. Spontaneous Recovery

Rule to follow for operant conditioning

• If you introduce operant extinction, keep with it.

• Things usually get worse before they get better but hang in there; doing so will pay off in the long run.

1. Control reinforcers for the behaviour that is to be decreased

• Reinforcers from others or from environment can undo extinction

• Make sure that the reinforcers you are withholding are actually the ones maintaining the behaviour

2. Combining extinction with positive reinforcement for an alternate behaviour

• differential reinforcement of alternate behaviour

3. Setting in which extinction is carried out

• Minimize the influence of alternative reinforcers on the undesirable behaviour (from others in the old setting)

• Maximize the chances of the behaviour modifier persisting with the program

4. Use of instructions or rules

• May speed up behaviour change

• “If you do X [the undesirable behaviour], then Y [the reinforcing item] will no longer occur.

5. Schedule of reinforcement before extinction is carried out

• Continuous reinforcement is quick to extinction

• Intermittent reinforcement may be slow to extinction (resistant)

6. Behaviour being extinguished might get worse before it gets better

• Extinction burst: temporary increase in responding during extinction

• In situations where extinction burst may be harmful: need to take preventative measures, or extinction should not be used

Extinction Burst

Temporary increase in responding during extinction

7. Elicited aggression

• Extinction may produce aggression that interferes with the program

• Aggression lower in programs that reinforce alternative behaviours

8. Spontaneous Recovery

Reappearance of an extinguished behaviour after a delay

Pitfalls of Operant Extinction

• Those who are unaware of extinction may unknowingly decrease a desirable behaviour!

• Desirable behaviour must be occasionally reinforced

Guidelines for the effective application of operant conditioning

• selecting the behaviour to be decreased

• preliminary considerations

• implementing the plan

• weaning from the program

Selecting the behaviour to be decreased

• Be specific

• Select proper location (behaviour might get worse before getting better)

• Select behaviour for which you can control the reinforcers maintaining it

Preliminary Considerations

• If possible, keep track of how often behaviour occurs before extinction

• Identify current reinforcer(s)

• Identify desirable alternative behaviour

• Identify effective reinforcers for the desirable alternative behaviour

• Select setting in which extinction will be successful

• Let all relevant individuals know what is going on (what is being extinguished and what is being reinforced)

Implementing the Plan

• Tell the individual about the plan before starting

• Use rules of reinforcement if rewarding alternative behaviour

• Be consistent

Weaning from the Program

• Be prepared for occasional relapses

• Three possible reasons for failure:

  1. Attention withholding is not the reinforcer

  2. Undesirable behaviour is receiving intermittent reinforcement from another source

  3. The desired alternative behaviour has not been strengthened sufficiently

Shaping

Development of a new behaviour by successive reinforcement of closer approximations and extinguishing preceding approximations of behaviour

5 aspects of behaviour that can be shaped

1. Topography

2. Frequency

3. Duration

4. Latency

5. Intensity

Topography

Specific movements involved

Frequency

number of instances that occur in a given period of time

Duration

Length of time a response lasts

Latency

time between occurrence of a stimulus and the response evoked by that stimulus

Intensity

force of a response; physical effect the response has on the environment

Factors influencing the effectiveness of shaping

• specifying the final desired behaviour

• choosing. a starting behaviour

• choosing the shaping steps

• moving along at the correct pace

Specifying the final desired behaviour

• Terminal behaviour = final desired behaviour

• Specify precise behaviour, relevant characteristics of the behaviour, and conditions under which it is to occur

Choosing a starting behaviour

• Should be a behaviour that occurs often enough to reinforce within session time

• Should be a behaviour that approximates the final desired behaviour

Choosing the shaping steps

Helpful to outline the approximations that will be reinforced

Moving along at the correct pace

• Reinforce an approximation at least several times before proceeding to the next step

• Avoid reinforcing too many times at any shaping step

• If behaviour is lost because the program was moving too fast, or the steps were too big, return to an earlier approximation

Pitfalls of shaping

• Shaping can be accidentally applied with unfortunate results

  • Harmful behaviours may be accidentally shaped

Guidelines for the effective application of shaping

• Select the terminal or final desired behaviour

• Select the starting behaviour

• Select an appropriate reinforcer

• Develop the initial plan

• Implement the plan

  • Tell learner about plan before starting

  • Begin reinforcing immediately following each occurrence of the starting behaviour

  • Never move to new approximation until the previous one is mastered

  • If learner stops working, may have moved up the steps too quickly, steps may not be the right size, or the reinforcer may be ineffective