Download Punishment

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Conditioned place preference wikipedia , lookup

Operant conditioning wikipedia , lookup

Transcript
Punishment
Definitions?
• Learning from the consequences that produce pain or
discomfort
• The loss of reinforcers: This has survival value for the
individual and for the species.
• Punishment teaches us not to repeat responses that
cause us harm
• Any stimulus or event, when used contingently,
decreases the probability of a response.
Nature of Punishment
• Punishment is defined neither by
– the actions of the person delivering the
consequences
– nor by the nature of those consequences.
• A decrease in the future frequency of the
occurrence of the behavior must be
observed before a consequence qualifies
as punishment.
Definitions of Punishment
• Positive or Type I Punishment :
– Presentation of a stimulus (or an increase in the intensity of an
already present stimulus) immediately following a behavior
– results in a decrease in the frequency of the behavior.
– Spanking, electric shock, etc.
• Negative or Type II Punishment:
– termination of an already present stimulus (or a decrease in the
intensity of an already present stimulus) immediately following a
behavior
– results in a decrease in the future frequency of the behavior.
– Response cost, overcorrection
Discriminative Effects
• Stimulus condition in the presence of which a response
has a lower probability of occurrence than it does in its
absence
• Response-contingent punishment:
– Delivery occurs in the presence of a stimulus cue
– Punishment only occurs when emitting contingent response.
• Stimulus control is important:
– If punishment occurs only in some stimulus conditions and not in others: the
suppressive effects of punishment will be most prevalent under those
conditions
– Organism learns the setting conditions as to when punishment will occur
– E.g.: Teacher has to see you misbehave
Punishment is also Known as
Aversive Control
• Note that aversive events are associated with
both:
– positive punishment
– negative reinforcement
• Term aversive control is often used to describe
intervention involving either or both of these two
principles.
Aversive but not contingent?
• Aversive stimuli can also affect operant
behavior when given noncontingently
– That is, a targeted behavior neither produces
nor prevents the punisher
– when aversive stimuli occur independently of
responding.
• Most famous example is conditioned
emotional response (CER)
Conditioned Emotional
Response (CER)
• Basic set up:
– Rat must lever press to obtain food.
– Rat receives periodic pairings of tone with electric
shock. (tone signals the shock)
– Rat eventually press lever at a lower rate when
tone is on than when it is off.
• Phenomenon is called conditioned
suppression or conditioned emotional
response (CER; Estes & Skinner, 1941).
Why are CER’s important?
• The “threat” of an upcoming aversive event can
decrease responding
– Even when noncontingent
– Unpredictability produces GREAT CERs
• Because the aversive event is likely, organism
“prepares” or “gets ready” for the event
– Can be cued or uncued
– Really gets in the way of ongoing responses
• E.g., in dog training: if dog is afraid of being punished
– Rate of overall responding goes down
– Animal is afraid of punisher, so doesn’t ‘risk’ behaving
Parameters of Punishment
• Is all we learned about positive reinforcement
true, in mirror-image form, of punishment?
– Yes and no.
– Are some additional characteristics of punishment
• Any operant punishment situation is really a
punishment plus reinforcement situation.
– Just like any reinforcement situation =Sr + P
– If get punished, don’t get the reinforcer.
– If get the reinforcer, don’t get punished.
Punishment + Reinforcement?
• For punishment to suppress operant responding,
responses must already be occurring with some
frequency.
• For responses to occur, they must be producing
reinforcement.
• So, effect of punishment reflects interaction of two
contingencies--reinforcement and punishment.
• Jointly operate in most situations.
Punishment Effectiveness
• Punishing only reinforced response is often not an
effective procedure.
– Suppresses responses
– Doesn’t provide a “replacement” response
• If you give organism an alternative, unpunished route
to reinforcement, then effects of punishment are
enhanced.
• Always shape an incompatible response!
– A response that is the opposite of the inappropriate
response
– E.g., punish getting out of seat but reinforce sitting in seat
Punishment Intensity:
• As intensity of punishing stimulus
increases, degree of suppression
increases.
• If very intense shock is used, then
suppression may be virtually complete.
• Partially due to physiological effects;
partially due to contingency effects
Role of past experience!
• Experience with the punisher is important
– If never experienced shock, don’t know it hurts!
– Punishment effects are relative: What has been prior
experience and how is the punisher COMPARED to
previous punishers?
• Lots of research showing this:
– Suppressive effect of intermediate shock intensity depends
on animal’s past experience with shock.
– If animal has experienced intensities going from mild to
intermediate, then there will be little suppression.
– If animal has experienced intensities going from severe to
intermediate, then there will be substantial suppression.
Immediate is Best!
• For punishment to be maximally effective, it
must immediately follow operant response.
• As delay interval between response and
punishment increases, amount of
suppression decreases.
• Too many other behaviors can occur
between R and P; contingency not as
effective when initially using P
Probability of a punisher
• Initially: should be certain and follow each operant
response.
– Probability of punishment should be 1.0
– When responses are punished intermittently,
effectiveness of punishment procedure is reduced.
• Can shape towards partial schedule of punishment
• Is this different than what observe with
reinforcement? Think about it!
Recovery from Punishment:
Extinction
• When punishment is discontinued, suppressive
effects on responding ARE not permanent
• The rate of responding after punishment is
discontinued will
– not only recover
– But briefly exceed level at which it was occurring prior
to punishment
– Opposite of extinction burst
– Is this really surprising?
Unconditioned Punishers
• Unconditioned punisher: stimulus whose presentation
functions as punishment without having been paired with
any other punishers.
– Innate
– Biologically relevant
– Still, all organisms not respond the same!
• Unconditioned punishers will suppress any behavior that
precedes their onset.
• Again, is this similar to reinforcement?
How are Conditioned
Punishers different?
• Conditioned punisher is a stimulus that
functions as punishment as a result of a
person’s conditioning history.
– E.g., the word “no” is a conditioned stimulus
– A child’s name can become a punisher!!!!
• Acquires capability to function as punisher
through stimulus-stimulus pairing with one or
more unconditioned or conditioned punishers.
Extinction of Conditioned Punishers
• If responses occur in absence of punisher, the
response will return, potentially to prepunishment levels
• If the conditioned punisher is repeatedly
presented without the punisher with which it was
initially paired, effectiveness as punishment will
diminish until it is no longer a punisher.
Generalization and Discrimination
• Stimulus that has been paired with numerous forms of
unconditioned and conditioned punishers becomes a
generalized conditioned punisher.
• Stimulus that has been paired with only one specific
conditioned or unconditioned punisher becomes a
discriminated punisher.
Other factors influencing
effectiveness of punishment:
• Schedule or frequency of punishment
– Continuous punishment schedules knock down behavior more
quickly
– Partial punishment schedules keep behavior suppressed more
effectively
• Availability of reinforcement for the target behavior
– Must eliminate inadvertent sources of reinforcement for your
behavior targeted for punishment
– Teacher may punish, but the other kids may keep reinforcing “class
clown” behavior
• Availability of reinforcement for an alternative behavior.
– Punishment more effective if reinforce the opposite behavior
– Again, must give organism alternative path to the reinforcer that was
maintaining the unwanted behavior
Punishment can lead to
Aggression
• Reflexive aggression:
– When punished, act out aggressively
– Is called reflexive because appears to be
innate
– When in pain, you bite!
• Operant Aggression
– Learn to be aggressive contingently
– retaliatory
Punishment and Aggression
• Aggression may model aggression
– That is, aggression breeds aggression
– Use of aggression as punishment may provide model of
how to “solve problems”
• Social Disruption
– The person who delivers the punishment/situation in which
punishment occurs become tainted
– Poisoning cues
– Also called social disruption:
• Behavior is disrupted in presence of cues that predict
punishment
• May get freezing, reduced effort, etc., in presence of punisher
Is Time Out a Punisher?
• Yes, by definition it is a negative punisher
– Losing the opportunity to get reinforcement from
many other sources
– Again, to be effective, must really isolate so can’t
get reinforced.
• Time out is not isolation
– Time out is brief, focused and contingent
– Isolation is of long duration, often noncontingent,
and often not effective.
Rules for Using Time-Out
• 1 minute per year of age
– Not really effective for children under 6-9 mos
– For really little ones, VERY brief withdrawal of
attention
• Must be quiet to get the timer to start
• Cannot use for dangerous, disruptive or selfstimulatory behavior
• Must really be “time out” from other rewards
Negative Punishment
• Response cost: your response costs you
something or some behavior
• OVERCORRECTION: two parts
– Restitution: reinstatement of environment
(clean up)
– Positive practice: practice better response for
situation
– Can also use satiation/habituation
Guidelines for using
positive punishment
• Behavior must be (immediately) dangerous to
person or others
• Rate of responding is so high that there is NO
chance to interrupt and reinforce “good” behavior
• Must have tried other alternatives
• An example:
http://www.youtube.com/watch?v=13UcT1FVVts
Examples:
• Severe self injurious behavior (SIB)
• Highly aggressive behavior to others
• Behavior which creates in immediate
danger for self or others
– Running into the street
– Pulling a hot pan off the stove