Transcript Experimental Analysis of Behavior

What is “Analysis of Behavior”?
Functional Analysis of Behavior
• Two ways to classify behavior:
– Structurally:
• what are the components of the behavior; what is its
structure?
• What is the topography of the behavior
– Functionally:
• what is the function of the behavior? What is its purpose?
• What is the behavior gaining the organism?
• What is the behavior’s reinforcement history/
Response Functions
• Think of behavior as a performance that follows a specific stimulus
and results in a particular consequence:
• S+: R C (Sr or P)
–
–
–
–
S+ = predictive stimulus (S- predicts the contingency won’t happen)
R = response
Sr = reinforcer
P = punishment
• Can analyze behavior using 3-step analysis: the ABCs of Behavior!
– A: antecedents: what are the setting conditions/stimuli
– B: Behavior: what is the structure/topography of the response
– C: Consequences: what reinforcer or punisher follows the behavior
• Determining the ABCs of behavior help determine the function of
the response
But what is a Response?
• Response (R)= an integrated set of movements
or a behavioral performance.
• Two basic classes of responses
– Respondent: really classical conditioning
– Operant: instrumental or operant conditioning
But what is a Response?
Functional Responses Classes
Response Class
Respondent
Operant
Function
Controlling Event
Elicited
Stimulus(event) preceding the
response
Emitted
Stimulus or event FOLLOWING
the response
Responses
• Elicited Responses:
– No consequence controls this behavior
– Classical conditioning
– Often reflexive or innate behaviors (but not always)
• Emitted Responses:
– Operant behavior: consequence controls this behavior
– Will have a reinforcement function (what does it get you?)
– Discriminative function: Can come under discriminative
control by a discriminative stimulus or stimuli
Stimulus Classes
• Stimuli have different functions, as well
– Defined by common effect on behavior
– Not defined by the similarity of the stimuli, but on their FUNCTION
• Discriminative stimuli:
– Serve as a cue for particular behaviors
– Bring that behavior under stimulus control
– E.g.: Stop signs result in a particular behavior
• Reinforcing stimuli
– Again defined by their function, not their similarity
– Here we get a four-square of behavior
• Reinforcement: positive and negative
• Punishment: positive and negative
Motivation of Behavior
• Context of behavior matters:
– Stop signs:
• if walking, typically don’t STOP at a stop sign
• Driving: do STOP at a stop sign
• Using Normal Theater as lecture hall: any problems with context of behavior?
– Deals with setting conditions:
– What conditions “set” the contingency?
• Motivational operation (MO):
– any environmental change that has a motivational operation has 2
major effects
– any event that alters the reinforcement effectiveness of behavioral
consequences and
– changes the frequency of the behavior maintained by those
consequences
Two basic kinds of MOs
• Establishing operation: EO:
– Change increases the momentary effectiveness of
reinforcement supporting operant behavior
– Change increases momentarily the responses that had in
the past produced such reinforcement
• Abolishing operation: AO:
– Decreases momentary effectiveness of reinforcement and
– Momentarily decreases rate of response
Conducting Behavioral Research from
a Behaviorist Perspective
• Make use of basic tactics of research:
– Independent variable; Dependent variable
– Correlational research or experimental method
– But: use SMALL N Designs rather than group designs!
• Interested in behavior change in an individual
– Use individual as own control
– In ABA, important that behavior change is functional and clinically
significant
– Allows evaluation of small groups and individuals
– NOT interested in the mean, but in individual behavior change
Small N designs
• Small N designs use a small number of subjects:
– Also called single- case designs
– Research designs that use the results from a single participant or small
number of subjects to establish the existence of cause- and- effect
relationships.
• Does not provide researchers with a set of scores from a group of
subjects, but examines change in behavior of an individual across a
set of conditions.
• Presentation and interpretation of results from a small N design
experiment are based on
– Visual inspection of a simple graph of the data
– In experimental analysis, also statistical analysis
11
Example
12
Limitation
• With just a baseline/intervention phase, the
results as presented do not represent a true
experiment because there is no control over
extraneous variables.
• Showed a change, but cannot conclude
causation just yet…..
13
Phases and phase changes
• A phase: series of
observations of the same
individual under the same
conditions.
• Baseline: observations
when no treatment is
being administered
• Treatment: Observations
when the treatment IS
administered
14
3 types of baseline
• Stable level
– Data generally flat, no high/low
– May bounce around a bit, but can show no gains or losses
• Stable trend:
– Data show a stable upward or downward trend
– Importance is in stability of the trend.
• Unstable data:
– No consistency
– This is undesirable for a baseline!
15
3 types of baseline
16
Dealing With Unstable Data
• The researcher can simply wait; occasionally, a participant
reacts unpredictably to the novelty of being observed.
• Consider the average of a set of two (or more)
observations.
• Look for patterns within the inconsistency.
– For example, a researcher examining disruptive classroom
behavior may find that a student exhibits very high levels
of disruption on some days and very low levels on other
days.
– E.g., days she has a swimming lesson
17
Next: Apply an Intervention
• Implement for several observations
• Again, establish stability in your data to
demonstrate change
Length of a Phase
• To establish a pattern ( level or trend) within a
phase and to determine the stability of the
data within a phase, a phase must consist of a
minimum of three observations.
• Why? Three points make a line!
– Allows you to (hopefully) determine the direction
of behavior change
– May need more if data are unstable
19
When to Change Phases
• Wait:
• When the data in a baseline phase show a trend indicating
improvement in the client’s behavior
• a researcher should not intervene by introducing a treatment phase.
• Don’t Wait:
• If baseline data indicate a seriously high level of dangerous or
threatening behavior.
• Researcher probably should not wait for the full set of five or six
observations necessary to establish a clear pattern.
• Stopping Treatment:
– If a treatment appears to produce an immediate and severe
deterioration in behavior,
– Stop the treatment IMMEDIATELY
20
Visual Inspection Techniques
• Unfortunately, there are no absolute,
objective standards for determining how
much of a change in pattern is sufficient to
provide a convincing demonstration of a
treatment effect.
• The most convincing results occur when the
change in pattern is immediate and large.
21
4 types of change
• Change in average level:
– change in average baseline rate and average treatment rate
• Immediate change in level
– Compare the last point in one phase with the first point in the
following phase
• Change in trend:
– Compare the slope of the trend in baseline with the shape of
the trend in treatment
• Latency of change.
– Compare the latency of change in baseline with latency of
change in treatment.
22
23
2- immediate change in level
24
Change in Trend
25
Latency in change
26
The problem with single subject design:
Need to find a way to show causation
27
THE ABAB REVERSAL DESIGN
• The majority of small N research studies use
some form of the ABAB design
• Consists of four phases:
–
–
–
–
A baseline phase ( A)
Followed by treatment ( B)
Then a return to baseline ( A)
Finally a repetition of the treatment phase ( B).
• Why end with a treatment?
28
Effective
29
Not Effective
30
Limitations of the ABAB Design
• The implemented treatment has corrected a
problem behavior, and when the treatment is
removed, the correction continues.
• A second problem with an ABAB design
concerns the ethical question of withdrawing
a successful treatment.
31
Variations on the ABAB Design
32
B not working, introduce C
33
B not working add C
B= Graduated exposure
C= Reinforcement
34
MULTIPLE- BASELINE DESIGNS
• Uses multiple baselines and multiple interventions
(hence the name!)
–
–
–
–
Across stimuli
Across environments
Across individuals
Across behaviors
• Eliminates the need for a return to baseline and
therefore
• Is particularly well suited for evaluating treatments
with long- lasting or permanent effects.
35
Person1
Person2
2 different students
36
Yelling
Crying
2 different behaviors
37
School
Home
2 different situations.
38
Weaknesses of the
Multiple- Baseline Design
• Risk that a treatment applied to one behavior
may generalize and produce changes in the
second behavior.
• E.g., Treating stuttering may help treating
aggressive behavior
• Or: getting a great “down” for your dog results in
loss of the “sit”
• Solution: chart and monitor behavior change in
both
• This could be good OR bad change
39
Weaknesses of the
Multiple- Baseline Design
• One behavior may show a large and immediate change,
but the second behavior may show only a minor or
gradual change when the treatment is introduced.
• Floor and ceiling effects
• Treatments may have different levels of effect on different
behaviors
• Can convert data to proportions to see if it is a
measurement issue
• The same problem can occur with research involving
different participants with similar behavior problems.
40
Dismantling design
• Dismantling design
– Also called a component- analysis design
– Consists of a series of phases in which each phase
adds or subtracts one component of a complex
treatment
– Allows one to determine how each component
contributes to the overall treatment effectiveness.
41
Example
42
The Changing- Criterion Design
• The criterion level is changed from one phase
to the next.
• Add to, take away or change criterion
– E.g., out of seat behavior:
• Phase 1: 5 out of seats/day
• Phase 2: 3 out of seats/day
• Phase 3: 1 out of seat/day
43
Smoking Treatment
44
The Alternating- Treatments Design
• Also called a discrete-trials design,
• Two ( or more) treatment conditions
– Randomly alternated from one observation to the
next.
– No set schedule; subject doesn’t know which is
coming when
• E.g., scent enrichment in zoo animals
45
Example 1- Alternate weeks
46
Example 2- 9 cases for each method
47
GENERAL STRENGTHS OF
SMALL N DESIGNS
• Conducted with only one participant or a very
small (N~5) group.
• Tends to be much more flexible than a
traditional group study.
• Single- subject designs require continuous
assessment.
48
General Weaknesses Of
Single- Subject Designs
• Participant’s behavior may be affected not only by the
treatment conditions but also by the assessment
procedures.
– Sometimes measuring behavior changes it
– Draws attention to the behavior; subject alters behavior
• Another concern for single- subject designs can be
absence of statistical controls.
– Can address this with additional statistical measures
designed for behavior analysis
– Typically included in EAB research; now seeing more and
more in ABA research
49
In Summary
• Functional analysis is determining the function of the
behavior: What purpose is the behavior serving for the
person?
• Behaviorists tend to use small N or single subject designs
focusing on individual behavior change
• Behaviorists use different forms of the experimental
method but CAN show cause and effect!
• The behavioral approach is not “instead of” the traditional
experimental approach, but provides an alternative level of
analysis.