Modeling and immitation training

Imitation training
• Imitation training is used to evoke new behaviours.
• Without imitation there is a low chance for a rapid acquisition of
behaviours.
• Imitative behaviour can be learned like any other behaviour:
 Reinforcement increases the occurrence of imitation
 When some behaviours receive reinforcement, other
behaviours occur without training or reinforcement
 Some children who do not imitate can learn to do so
 Imitation
Four criteria define the imitation
1. The imitative behaviour is occasioned by another person’s model of
behaviour.
2. The model and behaviour must have a formal similarity (are the
same).
3. An imitative behaviour follows the modelled behaviour closely in
time.
4. The MODEL is a controlling stimulus for imitative behaviour.
 Models
• A model as an antecedent stimulus with topographical similarity
to
the behaviour that is planned to be imitated.

• A model can be:

 A person
 Symbolic (video, picture)
 Models
Planned models
• Pre-arranged antecedent stimuli
• The model shows exactly what to do
Example: 1. A boxing instructor demonstrates how to perform “left hook punch”.
2. A video model to teach children to imitate counting in a sequence form 1 to 10.

Unplanned models
• These are any antecedent stimuli that evoke imitation.
• These occur in everyday social life
Examples??????
 Formal similarity
• The model and behaviour are the same.
 Temporal Relationship
• A close temporal relationship between the model and learner’s behaviour
it is an important feature of imitation.
• When the topography of a previous imitation occurs in the absence of the
model, this is NOT a imitative behaviour (delayed imitation).
• A delayed imitation occur as a result of different controlling variables.

E.g. An adult modelled to Child how to make a toaster (he placed a slice of bread in the
toaster and pressed the button). The next day, the child goes to the kitchen, takes a
slice of bread, places it in the toaster and presses the button.
The topography of the behaviour is the same but the model wasn’t present.

What is the SD that is controlling child’s behaviour during the next day?
 Model – Primary Controlling
VariableDoing the same – is not enough!!!
The most important property of the imitation:
The controlling relation between a model’s demonstration and
learner’s performance of topographically similar behaviour.
This controlling relation can be seen when a model evokes similar
behaviours without a prior history of reinforcement.

Holth (2003) concluded that imitation are the instances of behaviours

that do not have a direct reinforcement history.
 Model – Primary Controlling
Variable Imitative behaviour:
 Is a new behaviour.
 Behaviour that follows a novel antecedent (Model).
 The behaviour typically contacts new contingencies of
reinforcement.
 Repetition of the behaviour is controlled by the new contingency of
reinforcement.
The Behaviour with a formal similarity, a close temporal relationship
and a history of reinforcement is an example of discriminated operant.
Not
Imitative behaviour
 Imitation training
Baer et all., 1967
• Some children need special instruction to learn to imitate.
• Those who do not have imitative repertoire are trained using
prompts, shaping and reinforcement.
• Imitation it is important for socialisation in general and language
development in particular.
• The criteria for establishing the imitative behaviour is displaying a
novel response demonstrated for the first time without shaping
and prompting of that trial.
• Learning to learn phenomenon – As the individual progresses through
imitation training, he needs fewer training trials to learn new
imitative behaviours.
 Imitation training
• Imitation training Is a systematic, research-based set of steps for
teaching a non-imitative learner to imitate
models of novel behaviours.
• The objective of the training
To teach the learner to do what the model do
regardless the behaviour modelled.
• Generalised imitation
When a learner imitates a variety of
unprompted, untrained, non-reinforced
modelled behaviours in different settings.
An imitation training protocol (Striefel,1974)
Components of the imitation training protocol

1. Assess and teach pre-requisite skills for imitation training

2. Select models for training
3. Pre-test
4. Sequence models for training
5. Conduct imitation training
 1. Assess and teach pre-requisite
skills Pre-requisite Skills for imitation training
Attending skills
• Staying seated.
• Looking at the teacher.
• Keeping hands in lap.
• Looking at objects.
Absence of problem behaviours
The problem behaviours can interfere
with training process and these must be decreased
 2. Select models for training
• Start from easy models and move to complex ones.
• At the begging identify about 25 imitative behaviours.
• Involving manipulation of body parts (Examples?).
• Involving manipulation of objects (Examples?).

• Initially teach one model at a time of training.

 3. Pre-test
• The selected models should be pre-tested.
• Some models can be performed by the learner without training.
• Recommended to test models for 3 times each.
• Criterion level – 3 of 3 correct.
• Models that does not meet the criterion – goes for imitation training.
 4. Sequence models for
training
• Pre-test is used to arrange models from the easiest to the most
difficult.
• First models could be models imitated correctly on some trials.
• Next could be approximations.
• The last, models performed totally incorrect.
 5. Conduct imitation
Pre-assessmenttraining
similar with pre-test, but with first three selected
models only.
Training teaching one model until the learner responds
correctly 5 consecutive trials.
Post-assessment include 5 learned models, 5 models in training
present 3 times each. Removed from training after 3
consecutive post-assessment sessions (criterion 14
of 15).
Probe for imitation with Novel Models
five untrained models presented at the end of the
session without the verbal prompt “Do this!”
Guidelines and recommendations for
Imitation Training
• The training will likely include physical guidance to prompt the correct
response.

• The new and more complex imitation behaviours can be identified

from the child's environment.

• Keep training sessions active and short.

• Reinforce both prompted and independent imitative responses.

 Guidelines and recommendations for Imitation
Training
• Pair verbal praise with tangible reinforcers.

• If progress breaks down, back up and move ahead slowly.

• Fade out verbal response prompts and physical guidance.

• Collect data and make data-based decisions to terminate the training

 Modelling
• Modelling is a behaviour change strategy in which learners acquire
new skills by imitating demonstrations of the skills by live models
or symbolic models.
• Examples of symbolic videos:
 Video,
 Pictures,
 Icon,
 Photo,
 Audio
 Mixed
 Video modelling
• Video modelling is a technique that involves demonstration of the
model through the video.
• Video self-modelling – the learner has the opportunity to imitate a
behaviour performed successfully by himself.
• Point-of view perspective – it is a video where the person performing
the action cannot be seen. The action is seen as is being done by
the viewer.
 Guidelines for Effective Modelling
• Model’s similarity with learner
• Prestige (status of the model)
• Emphasize the Critical Aspect of Behaviour (Say… “I WANT”)
• Instructions (these enhance the model’s presentation)
• Context (A real situation is more likely to be imitated)
• Rehearsal and Feedback (give multiple opportunities to perform the
model’s presentation)
• Reinforcement
 Observational learning
• Observational learning - learning that results from observing the
responding of others and the consequences of such
responding (Catania, 2007).
• Requisite skills to develop observational learning:
1. Attending (observing, imitating and matching the response)
2. Imitating (imitation repertoire)
3. Discriminating (discrimination training: correct responses
reinforced, incorrect are not reinforced)
 Young et. al.,
• The goal:
1994
To examine the influence of response topography
on generalization of imitation across the response
types.
• Participants: 3 boys and 1 girl diagnosed with autism (2,11 – 4,5
years old).
• Modelling stimuli (20 sets of 27 trials each):
• Vocal (e.g. “eat cookie”)
• Vocal with toy (e.g. “crash” with car)
• Toy-play models (e.g. place man in car, drive car on and off
ramp)
• Pantomime models (movements of body that have a social
meaning, e.g. blowing a kiss)
 Young et. al., 1994
• Experimental design: A multiple baseline across responses was
used to assess generalisation of imitation
across and within responses types
• Dependent measures: The percentages of matching and
nonmatching responses emitted within 6 s
following the model.
• Independent variable:
Model alone condition
Model and prise condition
Generalisation probe trials
Young et. al., 1994
Young et. al., 1994
 Young et. al., 1994

• Conclusion: The present study demonstrated that generalized

imitation may be limited by the topographical
boundaries of response type (replicates the results
of Garcia et al., 1971).
Toy-play response class was the most slowly
acquired. Several studies identified negative
correlation between perseverative behavior and
appropriate toy play in children with autism (Koegel,
Firestone, Kramme,& Dunlap, 1974; Tryon & Keane,
1986).
 Eikeseth and Nesset, 2003
• The goal: To examine whether sufficient-response-exemplar
training of vocal imitation would result in
improved articulation in children with phonological
disorder, and
Whether improved articulation established in the
context of vocal imitation would result in improved
articulation during more natural speech.
 Eikeseth and Nesset, 2003
• Participants: 2 boys, 6 and 5,4 years old diagnosed with
Phonological Disorder (no other problem than
articulation).
• Target selection: 6 misarticulated sounds, for each sound a group
of 10 words.
• Dependent variable:
Vocal imitation of target words
• Training duration:
One 2 hours session per day. 1st participant had 15
sessions. 2nd participant had 21 sessions.
 Eikeseth and Nesset,
• Training: 2003
 Discrete-trial format.
 5 to 10 minutes sittings with 5 minutes breaks.
 Only one target was trained at a time and mixed
with mastered targets.
 Word mastered after 9 of 10 correct trials and
then presented as a part of 3-word sentence.
 Prompting, chaining and shaping were used.
 Token economy was used.
• Design: A multiple baseline design across behaviors
(target sounds).
Eikeseth and Nesset, 2003
Eikeseth and Nesset, 2003
 Eikeseth and Nesset, 2003
• Results
• It was not necessary to teach all 10 target words to achieve
mastery of the target sound.

• Findings suggest that the improved articulation established

during the vocal imitation training generalized
across verbal classes.