Chapter 11 Correspondence Relations: Imitation and RuleGoverned Behavior

1
Chapter 11Correspondence Relations Imitation
and Rule-Governed Behavior
2
Correspondence Relations

• People often do what others do. A child who
  observes an older sibling raid the cookie jar may
  engage in similar behavior.
• This is example involves correspondence relations
  between the demonstrated behavior and the
  replicated behavior.
• Technically we may say that the behavior of one
  person sets the occasion for an equivalent
  response by the other.

3
Correspondence Relations Continued

• There are other correspondence relations
  established by our culture. People look for and
  reinforce the correspondence between saying and
  doing.
• A large part of socialization involves arranging
  social reinforcement for correspondence between
  what is said and what is done.

4
Correspondence Relations Continued

• By the time a person is an adult, people expect
  consistency between spoken words and later
  performance. A minister who preaches moral
  conduct and lives a moral life is valued when
  moral words and moral deeds do not match, people
  become upset and act to correct the
  inconsistency. In such instances, what is said
  does not correspond adequately with what is done.

5
Imitation

• Learning by observation involves doing what
  others do in which the performance of an
  observer or learner is regulated by the actions
  of a model. Although modeling can produce a
  variety of effects (e.g. social facilitation,
  stimulus enhancement, and so on), imitation
  requires that the learner emit a novel response
  that could only occur by observing a model emit a
  similar response.

6
Correspondence and Observational Learning

• McDougall (1908) indicated that, as early as 4
  months of age, his child would stick out his
  tongue when an adult did the same.
• Of course, 4-month-old infants already have a
  considerable history of interaction with their
  parents, and the observed behavior may simply be
  attributable to social conditioning. That is,
  people may smile and laugh when a young child
  imitates some adult movement.

7
Correspondence and Observational Learning

• Presumably these social consequences strengthen
  imitation by the child. Although social
  conditioning is a possibility, recent research
  with newborn infants and animals suggests that
  innate or spontaneous imitation occurs without
  reinforcement.

8
Spontaneous Imitation

• Innate or spontaneous imitation is based on
  evolution and natural selection rather than
  experiences during the lifetime of the
  individual. That is, imitation of others may be
  an important for of adaptive behavior.
• Behavior may range from instinctive actions to
  more generalized set of responses, depending on
  the species.

9
Spontaneous Imitation

• In addition, imitation may occur only when the
  model is present or it may be delayed for some
  time after the model has been removed.
• Such delayed imitation is often as a more complex
  form since it involves remembering the modeled
  stimulus, rather than direct stimulus control.

10
Spontaneous Imitation

• There are ethnological reports of imitation by
  animals and birds.
• Birds
• Japanese monkeys
• Although it seems likely that some birds, monkeys
  and a few other species can imitate the novel
  responses of a model, these naturalistic studies
  are not sufficient to establish spontaneous
  imitation or to rule out alternative processes.

11
Imitation in the Laboratory

• Thorndike (1911) conducted the earliest
  experiment on spontaneous imitation. The
  experiment involved getting out of a puzzle box
  by observing the successful performance of
  others.
• The cat that observed the successful performance
  was no better at getting out of the box than
  other naïve animals.

12
Imitation in the Laboratory

• Herbert and Harsh (1944) reported that cats could
  learn to solve manipulative problems by
  observation if they observed mistakes as well as
  successful performances.
• Cats that observed both mistakes and correct
  responses by a model did better at problems than
  ones that only watched skillful performance.
• When many alternative responses are available,
  seeing what does and does not work is necessary
  for observational learning.

13
Imitation in the Laboratory

• Warden and his associates showed imitation in
  rhesus monkeys.
• They trained monkeys by reinforcement to solve
  puzzles that opened doors to reveal hidden
  raisins. When this performance was well
  established, a naïve monkey watched a trained
  animal obtain raisins.
• Observation of the model produced instantaneous
  solutions on 76 of the test trials. However,
  only the first instance of imitation could be
  described as spontaneous since the discovery of
  the raisin would reinforce this behavior and
  increase its likelihood.

14
Spontaneous and Delayed Imitation in Pigeons

• In fact, it is difficult to find an experiment
  that reliably demonstrates spontaneous imitation.
  This is because reinforcement of the observers
  behavior always confounds the results.
• Epstein, 1984.
• Birds that strongly imitated the model were found
  to continue this imitation even when the model
  was removed (i.e. model-absent imitation). The
  data suggested that delayed-spontaneous imitation
  can occur in laboratory pigeons, but the results
  were inconsistent over subjects.
• Spontaneous imitation occurred even after 24
  hours had elapsed between watching the model and
  the test for imitation.

15
Analysis of Epsteins Experiments

• It therefore appears that spontaneous imitation
  is a real effect and is a form of phylogenetic
  behavior. That is, imitative behavior occurs
  because it has been important to the survival and
  reproduction of the species. In other words,
  organisms that imitated others were more likely
  to find food, avoid predators, and eventually
  produce offspring.

16
Analysis of Epsteins Experiments

• The phylogenetic basis of spontaneous imitation
  is a reasonable hypothesis.
• Pigeons who are isolated from birth may show
  smaller effects of exposure to a model. In
  addition, the effects of food reinforcement may
  have contributed to the results. Although
  observers were never directly reinforced with
  food for imitation, they did see the models eat
  from the feeder.

17
Spontaneous Imitation by Human Infants

• There is evidence that spontaneous imitation
  occurs in human infants, almost from the moment
  of birth.
• Meltzoff and Moore (1977) were the first to
  report that 12-to-21-day-old infants can imitate
  the facial and hand movements of adult models.
  In these experiments, the imitative responses
  were tongue protrusion, mouth opening, lip
  protrusion, and sequential finger movements.

18
Difficulties with Infant Imitation Research

• The results of Meltzoff and Moores experiments
  remain controversial.
• Infant imitation may simply be an artifact of the
  experimental procedures.
• Other problems concern measurement, restricted
  range of responses, and infant attention to the
  modeled stimulus.

19
Difficulties with Infant Imitation Research

• Meltzoff and Moore noted that infants responses
  are not discrete or well formed each time they
  occur.
• This means that the response class is not
  clearly defined and it is therefore difficult to
  obtain a reliable measure. Because the response
  class is unclear, coders who observe the modeled
  gesture may see the imitative response more
  frequently than other response forms.

20
Delayed Imitation by Human Infants

• A series of studies by Meltzoff (1988a, 1988b,
  1988c) indicate that infants ranging in age from
  9 to 24 months will imitate significantly more
  modeled actions than a control group over delays
  ranging from 24 hr in the youngest infants to 4
  months in the oldest infants.
• Additional research indicates that 14-month-old
  infants will show delayed imitation of behavior
  modeled on television after 24 hr day.
• In the same study, the researchers found delayed
  imitation by infants of behavior modeled by an
  expert toddler performing a novel response
  after a 48 hour and a change in content from the
  experimental situation to the home setting.

21
Analysis of Spontaneous and Delayed Imitation by
Infants

• Spontaneous imitation in human newborns involves
  the infant observing a modeled gesture and
  responding with a set of muscle movements that
  correspond to the visual stimulus.
• The correspondence between the modeled stimulus
  and the form of response is a remarkable
  achievement because the infant is unable to see
  its own face when it reproduces the facial
  gestures of the adult model.

22
Operant and Generalized Imitation

• It is possible to train imitation as an operant
  in a social contingency of reinforcement.
• The discriminative stimulus is the behavior of
  the model (SDmodel), the operant is a response
  that matches the modeled stimulus (Rmatch), and
  reinforcement is verbal praise (Srsocial).
• Matching the model is reinforced, while
  noncorrespondent responses are extinguished.

23
Operant and Generalized Imitation

• If imitation is reinforced and nonimitation is
  extinguished, imitation of the model will
  increase.
• On the other hand, nonimitation will occur if
  imitation is extinguished and nonimitation is
  reinforced.

24
Operant and Generalized Imitation

• Although, operant imitation provides a
  straightforward account to observational
  learning, Bandura (1969) noted that the operant
  account may be limited to situations in which the
  observer sees the model, an imitative response
  immediately occurs, and reinforcement follows.
  In everyday life, there are occasions when
  imitation does not conform to this sequence.

25
Operant and Generalized Imitation

• Although Bandura (1969, 1977, 1986) has argued
  against an operant account based on these
  difficulties, Donald Baer and his associates
  provided a behavior analysis of imitation that
  handles each of the apparent challenges to the
  operant paradigm.
• The approach is called generalized imitation and
  is based on operant principles of discrimination
  and generalization.

26
Generalized Imitation

• The procedures of generalized imitation begin
  with simple reinforcement of correspondence
  between the modeled performance (SD model) and
  the imitative operant (Rmatch).
• The actual discrimination procedures involve
  several modeled stimuli (SDs) and multiple
  operants (Rmatch).
• In each case what the model does sets the
  occasion for reinforcement of a similar response
  by the child all other responses are
  extinguished. This training results in a
  stimulus class of modeled actions and an
  imitative response class. The child now imitates
  whichever of the three responses that the model
  performs.

27
Generalized Imitation

• The next step is to test for generalization of
  the stimulus and response class.
• Baer and Sherman (1964) showed that a new-modeled
  stimulus would set the occasion for a novel
  imitative response, without any further
  reinforcement.
• If the puppet began pressing a lever, the child
  also imitated this performance even though this
  response was never reinforced with praise.
• Generalized imitation account for the appearance
  of novel imitative acts in children- even when
  these responses were never reinforced.

28
The Bobo Doll Experiment

• Albert Bandura (1965) designed an experiment to
  show a form of learning by observation more
  complex than generalized imitation.
• Children participated in this experiment on the
  imitation of aggressive behavior.
• Each child watched a short film in which an adult
  demonstrated four distinctive aggressive actions
  toward an inflated Bobo doll.

29
The Bobo Doll Experiment

• Every aggressive action was accompanied by a
  unique verbal response. While sitting on the
  Bobo doll, the adult punched it in the face and
  said, Pow, right in the nose, boom, boom.
• In another sequence, the adult hit the doll with
  a mallet saying Sockeroo, stay down.
• Also, the model kicked the Bobo doll and said,
  Fly away, and threw rubber balls at the doll
  while saying, Bang.

30
The Bobo Doll Experiment

• Some of the children saw that model rewarded by
  another adult, who supplied soda, snack, and
  candies while saying, strong champion.
• Other children saw the model receive negative
  consequences. The adult scolded and spanked the
  model for picking on that clown and warned him
  or her not to act that way again.
• A third group saw that modeled aggression, but no
  consequences for the aggressive behavior were
  portrayed.

31
The Bobo Doll Experiment

• When the film ended, a child was taken to a room
  that contained many toys, including a Bobo doll.
  The child was encouraged to play with the toys
  and then was left alone.
• Generally, there was a high frequency of
  imitative aggressive behavior toward the Bobo,
  and boys were more aggressive than girls.
• Bandura (1965) also found that reward and
  punishment of the models actions affected the
  imitation of aggression.

32
The Bobo Doll Experiment

• Children who saw that model punished were less
  likely to imitate aggression than those who saw
  the model rewarded.
• Children who saw the model rewarded did not
  differ in imitative aggression from those who
  watched the model perform the actions without
  receiving social consequences.
• Importantly, this means that just seeing modeled
  aggression (no consequences) had about as much
  impact on the imitation as observing violence
  being rewarded.

33
The Bobo Doll Experiment

• Finally, Bandura offered an incentive to all the
  children if they could remember the actions of
  the model in the film.
• With this incentive, all three groups recalled
  the modeled aggression at about the same level.
• It seemed that all children had learned equally
  from the modeled aggression, but those who
  witnessed punishment of the model were less
  inclined to perform the aggressive sequences.

34
Social Cognitive Interpretation

• Bandura (1986) argued that the difference between
  learning and performing modeled aggression
  requires a cognitive theory of observational
  learning.
• The observer pays attention to the modeled
  sequence, noting the arrangement of each action.
• The general information in the sequence must be
  coded and rehearsed as when the child says,
  First sit on the Bobo, and then say the words
  pow. Once this abstract information is
  retained in memory, imitation is a matter of
  reproducing the component responses in the
  correct sequences.

35
Social Cognitive Interpretation

• Complex behavior patterns, however, cannot be
  learned by observation until the component skills
  have been mastered.
• Finally, the anticipated consequences of
  imitation determine whether an imitative response
  will occur.
• People who expect positive outcomes are likely to
  perform actions they have witnessed, and those
  who expect negative consequences are less likely
  to imitate such actions.

36
Behavioral Interpretation

• A behavioral interpretation for complex
  observational learning is that it may build on
  the processes of generalized imitation.
• Generalized imitation provides an account of
  novel instances of imitation.
• An operant perspective, imitation, is most likely
  to occur in situations in which It was reinforced
  previously.
• Behavior is unlikely in situations in which it
  was extinguished, or in settings in which it was
  punished.

37
Behavioral Interpretation

• This kind of conditioning history provides a
  plausible account of Banduras results concerning
  complex observational learning.
• The learning and performance differences of the
  Bobo doll research may also be due to previous
  conditioning.
• When Bandura offered an incentive for recalling
  the modeled action, he presented a discriminative
  stimulus that increased in probability of this
  verbal behavior.
• For most children, it is likely that being
  promised a reward for recalling some action is a
  situation that has accompanied reinforcement in
  the past.

38
Rules, Observational Learning, and Self-Efficacy

• Bandura has noted that observational learning in
  humans involves the discovery and use of abstract
  rules.
• I began to develop the notion of modeling as a
  broad phenomenon that serves several functions.
  This conceptualization of modeling is concerned
  more with the observers extracting the rules and
  structure of behavior, rather than copying
  particular examples they had observed. Once they
  acquire that structure and the rules, they can
  use that knowledge to generate new patterns of
  behavior. Modeling is a much more complex
  abstract process than a simple process of
  response mimicry.

39
Rules, Observational Learning, and Self-Efficacy

• From a behavioral perspective, extracting the
  rules is verbal operant behavior that describes
  the contingencies of reinforcement.
• Skinner and Bandura agree about the importance of
  rules from human behavior, but they differ in
  terms of interpretation and philosophy.

40
Rules, Observational Learning, and Self-Efficacy

• Bandura talks about rules as cognitive events,
  and Skinner views them as verbal descriptions.
• For Skinner, following the rules is behavior
  under the control of verbal stimuli. That is,
  statements of rules, advice, maxims, or laws are
  discriminative stimuli that set the occasion for
  behavior.
• Rules, as verbal descriptions, may affect
  observational learning.

41
Rules, Observational Learning, and Self-Efficacy

• From a behavior analysis view, statements of
  self-efficacy, as a class of verbal stimuli, can
  affect subsequent behavior.
• Based on conditioning for compliance, statements
  of self-efficacy often predict how a person will
  act in subsequent (similar) situations.

42
Rule-Governed Behavior

• A large part of human behavior is regulated by
  verbal stimuli.
• Stimuli are the products of speaking, writing,
  signing, and other forms.
• The common property of these kinds of stimuli is
  that they describe the operating contingencies of
  reinforcement.
• Formally, rules, instructions, advice, and laws
  are contingency-specifying stimuli, describing
  the SDR? Sr relations of everyday life.
• The term rule-governed behavior is used when the
  listeners (readers) performance is regulated by
  contingency-specifying stimuli.

43
Constructing and Following Rules

• In solving a problem, people often make up or
  construct their own discriminative stimuli.
• A person who has an important, early morning
  appointment may set an alarm clock for six
  oclock a.m. Technically, setting the alarm is
  precurrent behavior, or an operant that precedes
  some other response. This behavior produces a
  discriminative stimulus that sets the occasion
  for getting up and going to the meeting.
• Thus, a major function of precurrent behavior is
  the construction of SDs that regulate subsequent
  action.

44
Constructing and Following Rules

• People also may construct discriminative stimuli
  through written words or spoken sounds.
• For example, a person may make a shopping list
  before going to the supermarket. Making a list
  is precurrent behavior, and the list is a
  discriminative stimulus for choosing groceries.
• Similarly, economical shoppers may say to
  themselves, only buy products that are on sale.
  This verbal stimulus acts something like the
  grocery list in the previous example.
• As a rule, the verbal expression points to the
  relation between the stimuli, behavior, and
  reinforcement in the marketplace.

45
Rule-Governed and Contingency-Shaped Behavior

• People are said to solve problems either by
  discovery or by instruction.
• From a behavioral perspective the difference is
  between the direct effects of contingencies
  (discovery) and the indirect effects of rules
  (instruction).
• When performance is attributed to direct exposure
  to reinforcement contingencies, behavior is said
  to be contingency-shaped.
• As previously noted, performance set up by
  constructing and following instructions (and
  other verbal stimuli) is termed rule-governed
  behavior.

46
Rule-Governed and Contingency-Shaped Behavior

• Although behavior attributed to rules and
  contingencies occasionally may look the same, the
  variables that affect performance are in fact
  quite different.
• Reinforcement or following the advice of others
  in various situations may establish a general
  tendency to do what others recommend.
• When directions are backed up with social
  punishment rather than natural consequences, they
  are called orders and commands. Individuals
  follow orders because they have been punished for
  disobedience.

47
Rule-Governed and Contingency-Shaped Behavior

• Generalized obedience, however, may be a problem.
  Governments can induce blind obedience in which
  a person harms another without regard for moral
  consequences. In many countries, Amnesty
  International has documented the torture of
  political prisoners by guards and police. In
  these cases, obedience to authority is
  unquestioned and obviously results in serious
  harm or death to the victims.

48
Rule-Governed and Contingency-Shaped Behavior

• The importance of reinforcement contingencies in
  establishing and maintaining rule-following is
  clearly seen with ineffective rules and
  instructions.
• When rules describe delayed and improbable
  events, it is necessary to find other reasons to
  follow them.

49
Instructions and Contingencies

• In his discussion of rule-governed and
  contingency-shaped behavior, Skinner (1969)
  speculated that instructions may affect
  performance differently than the actual
  contingencies of reinforcement.
• One way to test this idea is to expose humans to
  reinforcement procedures that are accurately or
  inaccurately described by the experimenters
  instructions.
• If behavior varies with the instructions while
  the actual contingencies remain the same, this
  would be evidence for Skinners assertion.

50
Instructions and Contingencies

• An early study by Lippman and Meyer (1967) showed
  that human performance on a fixed-interval
  schedule varied with instructions.
• In a similar study, Kaufman, Baron, and Kopp
  (1966) placed subjects on a variable-interval
  (VI) schedule of reinforcement and told them that
  points were available on either a fixed-interval
  or variable-ratio basis.
• Performance was more in accord with the
  experimental instructions than with the actual VI
  contingencies.

51
Instructions and Contingencies

• The fact that instructions, in these experiments,
  seem to override the actual contingencies has
  been used to argue against a reinforcement
  analysis of human behavior.
• Bandura linked instructions to modeling and
  argued that both of these procedures activate
  subjects expectancies which, in turn, affect
  subsequent behavior. This means that expected
  reinforcement, rather than actual contingencies,
  is the stronger determinant.

52
Instructions and Contingencies

• In addition, Spielberger and Denike (1966)
  disputed the claim that instructions were complex
  discriminative stimuli.
• Both objections were addressed by Mark Galizio,
  the results of Galizios experiments provide
  strong support for the view that instructional
  control is a form of rule-governed behavior.

53
Rules as Function-Altering EventsAlerting
Discriminative Relations

• Rules can act as function-altering events,
  altering the function of other stimuli and,
  thereby, the strength of relations among these
  stimuli and behavior.