Topic 8: Shaping and Chaining Behavior Flashcards
Shaping
-Differential reinforcement (4 types) of successive approximations of a target behavior
-Behavior that person typically does not exhibit (novel)
-Reinstating behavior that previously, but no longer occurred
-Change in dimension of existing behavior
Behavioral “Selection”
-“Natural selection”
-Reinforcing certain variations (extinction bursts) in behavior that are closer to our goal than we are presently
-Reinforced variations are selected for: More likely to occur again
-Reinforced variations are selected against” Become less likely to occur again
Changed behavior
-Behavior changes as a consequence they produced in environment
-Behavior affects environment and therefore affect future behavior
-Behaviors become more or less likely to occur in future based on how well they work when we use them
Door knob example
-Turning key and handle don’t work
-Operant extinction
-Variation in key turning and knob turning
-Results in new behavior
-Environment provides reinforcement (door opening)
Rat pressing lever example
1) Reinforce facing correct direction
2) Reinforce approaching lever
3) Reinforce sniffing lever
4) Reinforcing touching with paw
5) Reinforce full depression of lever
-Small baby steps until step 5) (final goal in this scenario)
Successive approximations
-Steps between starting behavior and the target behavior
-Either relearning behavior, variation of behavior or behavior we are unable to do
Differential reinforcement
-Reinforcing approximations that get us closer to the target behaviors while not reinforcing behaviors that don’t
-Previous approximations no longer reinforced (extinction –> Extinction burst –> Variations)
Shaping procedure
1) Define target behavior
2) Determine whether shaping is the most appropriate procedure
3) Identify starting behavior: What behavior that they are capable of is closest to target behavior (Baseline)
4) Select successive approximations (Make steps)
5) choose reinforcer
6) Differentially reinforce successive approximations
-extinction to move to next step
-Do not reinforce previous approximations
Acquisition vs maintenance
-During acquisition (learning) use continuous reinforcement and then for maintenance use intermittent reinforcement
Marking behavior
-Clickers used so that an immediate conditionally reinforcing consequence is provided
-Aids learning by providing immediate reinforcement and preventing accidental reinforcement of other behaviors
Shaping undesired behaviors
-Unknowingly using differential reinforcement of successive approximations
-Ignoring kid crying turns into temper tantrum (Burst)
-Reinforcing tantrum = new behavior shaped
-Next time the extreme behavior (Tantrum) will be baseline
-Escalation in extreme behaviors due to failed extinction
Chaining behavior
-Complex behavior consisting of two or more component behaviors that occur together in a sequence
-Each component of chain has a S^D and corresponding response to that S^D
-Each response creates a new situation that functions as an S^D for the next response
Getting piece of gum example
1) S^D1 (Gum in pocket) –> R1 (Reach into pocket)
2) SD2 (Hand in pocket) –> R2 (pull pack out)
3) SD3 (gum in hand) –> R3 (Pull out stick of gum)
4) SD4 (Gum in hand) –> R4 (Unwrap gum)
5) SD5 (Unwrapped gum in hand) –> R5 (Gum in mouth)
Task analysis
-Breaking down task into component elements
-Establishing discriminative stimuli (S^D), responses and reinforcers
-Prior to chaining, subject must be capable of each component
-Revise as necessary (Adjust steps, include shaping)
3 Types of chaining procedures
1) Forward chaining
2) Backward chaining
3) Total task presentation
1) Forward chaining
-Requires reinforcement of each component which can be impractical
-Issues can arise because at first, step one is reinforced but it won’t always be
-Extinction of previous behavior
-Prompting and fading may need to elicit new components
Forward chaining steps
1) Reinforce performance of first component and repeat until behavior formed reliably
2) Reinforce performance of first 2 components until reliable
3) Continue until whole chain completed before reinforcement delivered after terminal (end) behavior
2) Backward chaining
-Components are added from back-to-front
-Only required terminal reinforcement
-Avoids issues of extinction pointed out in forward chaining procedure
-Prompting and fading may be needed to elicit new components
Backward chaining steps
1) Reinforce performance of last component and repeat until reliable
2) Reinforce behavior of last 2 components
3) Continue until whole chain is completed before reinforcement is delivered after terminal behavior
3) Total task presentation
-Use prompting to get learner to perform whole chain start to finish
-Uses graduated guidance
-Consider: Task complexity, learners ability, teachers ability
