Professional Documents
Culture Documents
By
A DISSERTATION
Advisory Committee:
advanced degree at the University of Southern Maine, I agree that the library shall make
it freely available for inspection. I further agree that permission for “fair use” copying or
publication for scholarly purposes be granted by the Librarian. It is understood than any
copying or publication of this dissertation for financial gain shall not be allowed without
my written permission.
Signature:
Date:
ii.
Much of the behavior analytic research conducted in the past fifty years has
included studies that held motivation constant as a confound. However, there are
parameters of environmental context which change the probability of response rate within
the context of static rates of reinforcement. Keller and Shoenfeld (1950) and later
Michael (1982, 1993) described a variety of environmental operations that change the
learning history, whereas conditioned motivating operations do. There are three types of
Motivating Operations. These operations induce behavioral processes that change the
function of a previously neutral stimulus condition. The purpose of the current study is to
observe the process and corresponding behavior altering effects of the Surrogate and
ACKNOWLEDGEMENTS
I would like to personally thank Dr. Mark W. Steege for his patience, knowledge
and generous gift of time during the completion of this dissertation. His dedication to his
students and desire to bring the technology of behavior analysis to systems of public
education is genuine and laudable. Dr. Steege is an exceptional advocate for children
with disabilities and has provided a wonderful model of a professional who hold both
precision and compassion in high regard when working with people who are disabled.
Moreover I would also like to thank Dr. Rachel Brown-Chidsey for her exceptional
consistent guidance throughout the my graduate work at USM. Additionally, the author
would like to thank Dr. Harold Longenecker for his feedback on the technical aspect
papers and for the entertaining and thought provoking discussions of the ethics of
professional psychology over the past two years. Dr. F.C. Mace was an integral part of
the design process of these experiments. Without his knowledge of behavior and
experimental methodology this dissertation would not have been completed with the
same level of methodological soundness and precision. Dr. Mace has provided guidance
that has helped me understand behavior analysis at a level that would have been very
difficult without his teaching, and I thank him graciously for his contribution to my
professional growth. I would also like to thank the Assistant Director of Margaret
supportive context within which to conduct this research. Without such a collegial
TABLE OF CONTENTS
ACKNOWLEDGMENTS……………………………………………………………….iii
LIST OF TABLES……………………………………………………………………….vi
LIST OF FIGURES……………………………………………………………………..vii
Chapter
1. LITERATURE REVIEW……………………………………………………..…1
Contributions of J. Michael……………………………………………………….8
2. EXPERIMENT 1……………………………………………………………..…18
Method…………………………………………………………………………..18
Subjects………………………………………………………………….18
Procedure………………………………………………………………..22
Baseline Phase…………………………………………………...22
Conditioning Phase………………………………………………24
Test Phase………………………………………………………..25
Results……………………………………………………………………………26
Discussion………………..………………………………………………………29
Future Research………………………………………………………………….31
3. EXPERIMENT 2………………………………………………………………34
Method…………………………………………………………………………34
Subjects………………………………………………………………...34
Procedure………………………………………………………………36
Baseline Phase………………………………………………….36
Training Phase………………………………………………….36
Results………………………………………………………………………….40
Discussion……………………………………………………………………....42
Future Research………………………………………………………………....45
4. GENERAL DISCUSSION………......................................................................47
REFERENCES……………………………………………………………........53
LISTS OF TABLES
Table 3.3 Summary of the Descriptive from the Baseline and Training Phases of
Experiment 2…………………………………………………………….37
vii.
LIST OF FIGURES
Experiment 2...........................................................................................42
Conditioning CEO’s 1.
Chapter 1
Literature Review
“I am induced by many reasons to suspect that [the phenomena of nature] may all
depend on certain forces by which the particles of bodies, by some cause hitherto
unknown, are mutually impelled towards one another, and cohere in regular figures, or
are repelled and recede from one another.” – Sir. Isaac Newton (from Kileen, 1992)
events that contact an organism, and the organism’s response which contacts the outside
Consequence or S-R-C) has been offered as a model for understanding the reciprocal
interaction of the individual and the environment. This model describes how antecedent
stimuli occasion behavior and how consequences strengthen that contingency. Michael
(1982) expanded the discussion of the three-termed contingency by including the notion
of motivation and proposed a set of operations that change the motivational properties of
environmental events.
Skinner 1938, 1953; Smith & Iwata, 1997). Moreover, for the greater proportion of the
past 50 years, the impact of antecedent events other than discriminative stimuli on
behavior has largely been ignored (Smith, & Iwata 1997). Skinner's work emphasized
the selection of behavior by its consequences, and the field of behavior analysis has
closely followed this paradigm. The process of discrimination changes the behavioral
Conditioning CEO’s 2.
or S), which then comes to occasion behavior (i.e., is associated with a change in
response probability). This is not the only operant process that results in other types of
antecedent events that motivate behavior (i.e. that alter the probability of response).
Some of these events are unconditioned and some are conditioned, meaning that they
reinforcement. The pattern of the interactions between the individual and the
environment determine what class of antecedent the previously neutral event becomes.
Conceptually, there are three main lines of reasoning that define the structural
aspects of motivation from the perspective of behavior analysis. The three main accounts
of motivation arose from the works of Skinner (1938, 1953), Kantor (1977), and Michael
placed on Michael’s contribution as his work forms the conceptual base of the thesis of
neutral stimulus so it can later alter the probability of responses (1938, 1953). Therefore,
neutral event is paired temporally with the availability of reinforcement. When responses
are reinforced, the previously neutral event becomes a behaviorally functional unit that
signals the availability of reinforcement. Such an antecedent event that signals the
reinforcement. An SD, therefore, sets the occasion for behavior (it predicts a differential
probability that it will occur), but does not motivate behavior. In general parlance, an SD
tells an organism that something is available, but does not make the organism “want” that
Skinner also recognized that other events can change the probability of a
described. Skinner observed that more frequent responding would occur when an
organism was deprived of the type of reinforcement that was used in the experimental
level of that type of reinforcement just prior to exposure to the contingency (satiation).
Conditioning CEO’s 4.
Both satiation and deprivation are operations that change the probability of a response, by
altering the temporary value of the reinforcement. Keller and Schoenfeld (1950)
the strength of a response is called a “drive.” The drive was an important concept in the
development of the motivating operation which was discussed both by Skinner (1938,
1953) and Keller and Shoenfeld (1950). Conceptually, “the drive” is not a measurable
events. Because this definition is ambiguous and because of it is not directly measurable,
the “drive” has limited utility as an experimental tool. Within their conceptualization of
motivation, Skinner, as well as Keller and Schoenfeld, used the term “the drive” to
observable behavior. Both Skinner and Keller and Schoenfeld described a number of
(e.g., hunger drive, sex drive, etc.) These variables were not defined functionally, but
they were defined by characteristics that are correlated with physiologically relevant
topics. Although correlated with physiological variables, the drive is not a physiological
operation (e.g., deprivation, satiation) and process (e.g., increases and decreases in
(Skinner, 1953). These events are defined by three characteristics: a) they originate in the
class of behavior is altered. Again, although Skinner fully recognized the impact of
emotion on responding, he was careful not to explain the impact of emotion on human
Behavior, chapter 10 for a complete description), but was careful to describe such
that emotion and motivation likely overlap, but that science needs to use caution not to
describe the impact of emotion on human behavior using psychic or physiological terms.
aversive stimuli by their structural rather than functional qualities (Smith & Iwata, 1997).
Although he recognized that these events changed the probability of responding, the
functional properties of the relationships were not stated. Skinner described the process
of discrimination and elicitation functionally, but left behavior analysis with a structurally
contributions were mainly structural, and included little primary data to support his
views.
phenomenology of this construct differs little from the definition of a stimulus delineated
by Skinner. However, Kantor fundamentally defined this concept differently because his
description overtly stated the reciprocal nature of the relation of stimulus and response in
behavior. According to Kantor, the term “stimulus function” appears to indicate a static
nature of stimuli. This did not appear to be Skinner’s intent, however. In The Behavior
of Organisms (1938) Skinner stated that an event does not function as a stimulus until it
exerts control over behavior. By definition the process of stimulus control requires the
molecular perspective to describe the composition of behavior (Skinner, 1938, pp. 8-19).
His reliance on the reflex as the primary “fact” that is the analytical unit of behavior
behavior. Kantor defined setting components as “conditions in and around the organism
and its stimuli,” such as hygiene, or habituation. Moreover, he described the interaction
of the individual and the environment as “a mutually corresponsive process that could not
Conditioning CEO’s 7.
be dissected into individual behavioral and stimulus functions,” a totality which requires
defining behavior as a “whole” event. As reported by Smith and Iwata (1997), this is a
description of behavior that few behavioral analysts would disagree with in principle, but
that many might disagree with empirically. Such an analysis requires the reliable
quantification of molar classes of behavior, a feat not yet realized. In his analysis of
TEAB (1970), Kantor described his interbehavioral structures in great detail, having used
book.
functionalism), Kantor’s structural theory yielded constructs that are mentalistic and
phenomena that requires highly precise electronic technology beyond the direct
interaction of the organism and the environment. “Setting events” as described by Kantor
are (in part) determined by this interbehavioral function, and are correspondingly difficult
behavioral technologies that were not rooted in the science of behavior. Wahler and Fox
behavior analysts. Specifically, they contend that correlational methods for comparing
the molar effect of interbehavioral fields on behavior is adequate for the applied science
Conditioning CEO’s 8.
of human behavior. Certainly there is great room for discussing the need for behavior
(Baum, 2000), but there are too many methodological questions that need to be answered
psychology address some of the criticisms of behavior analysis, they are difficult to
incorporate into the behavior analytic consensus because they are outside of the prevue of
The science of behavior seeks to observe the “natural lines of fracture” that define
antecedent conditions that are global and based in structurally determined constructs.
analysis of behavior that is rooted in observing the molar effects of a set of conditions on
behavior analysis.
Contributions of J. Michael
Keller and Schoenfeld (1950) first coined the term “Establishing Operation”
and defined an establishing operation very broadly as having, “two general effects on
Conditioning CEO’s 9.
object or event as reinforcement, and the other is to evoke behavior that has been
followed by that object or event” (Michael, 1982, pp. 149). The former effect of the
establishing operation is termed the “value-altering effect,” whereas the latter effect is
termed the “behavior altering effect.” Michael proposed a structural account of these
behavior altering effects are either evocative (increase behavior) or abative (decrease
behavior).
framework has recently evolved to describe the conceptual framework of the establishing
describe the current consensus among behavior analysts regarding this concept.
that was rather confusing. The term “establishing” was used to describe the effect of
result, the term “establishing” was associated with conditions that evoked and/or
the term “establishing operations” because the same term was used to describe the same
effect on another event (e.g., establishment), but have opposite effects on behavior (either
Conditioning CEO’s 10.
abative or evocative). Laraway, Synersky, Poling, & Michael (2003) developed more
specific definitions for the operations and processes previously described in the literature.
called the Motivating Operation (Laraway, Synersky, Poling, & Michael 2003). Under
the umbrella “Motivating Operations,” are both establishing and abolishing operations,
both of which refer to value-altering effects the operation has on a stimulus class. An
“abolishing” effect on that event. Inversely, an operation that increases the value of an
event.
Establishing operations can establish the effect of both punishment and reinforcement.
Operations that establish a stimulus as a punisher also have an abative effect on behavior,
may increase the value of a sweater as a reinforcer and subsequently evoke sweater
there are establishing operations for both punishment and reinforcement and abolishing
operations for both punishment and reinforcement. The linguistic framework delineated
by Laraway et al (2003) will be used as a basis for further discussion in this paper.
Conditioning CEO’s 11.
not require a learning history. Deprivation and satiation of specific classes of reinforcers
are operations that change the probability of responding within the context of consistent
behavior related to these consequences. For example, an animal that is deprived of food
.is more likely to engage in classes of behavior that had previously resulted in access to
food. Again this change in responding occurs regardless of the differential availability of
change the probability that a previously established SD will evoke behavior, if it signals
the availability of the same class of reinforcer altered by the motivating operation.
All UMO’s are related to primary reinforcement. Michael defined nine broad
categories of UMO’s (Michael, 2005). The primary reinforcers food, water, sleep,
activity, and oxygen are altered through the environmental operation of deprivation or
but may be invoked by other operations as well (e.g., pheromones, hormones etc.) The
UMO’s related to being too warm or too cold have separate physiological determinates,
The last UMO is related to an increase in painful stimuli. Traditionally behavior analysis
context within which pain attenuating behavior can be reinforced, and thus are
Food Deprivation/Satiation
Water Deprivation/Satiation
Oxygen Deprivation/Satiation
Activity Deprivation/Satiation
Sleep Deprivation/Satiation
Sex Complex
categories of Conditioned Motivating Operations (CMO’s) that evolve within the lifespan
motivating operation. For example, watching television while waiting for dinner to cook
correlates the television (neutral stimulus) with a state of food deprivation. This temporal
contiguity could impart the motivational properties of food deprivation on the presence of
the television. Television could then evoke behavior motivated by food deprivation,
functioning as a Surrogate CMO. There has been little basic research on the Surrogate
Conditioning CEO’s 14.
CMO. Calvin, Bicknell and Sperling (1953) correlated a striped box with varying levels
of deprivation in rats. One rat was deprived of food for 22 hours and the other for 1 hour.
After the training period the rats were allowed to feed after 11.5 hours of deprivation in
the striped box and those who experienced greater deprivation ate more. It must be noted
that attempts to replicate this research have not yielded very consistent findings (Cravens
and responses that occur within that neutral stimulus both eliminate the warning stimulus
condition (Michael, 1993; 2000). Michael termed these operations “Threat or Promise”
would function as negative reinforcement. The individual, having escaped the warning
stimuli, would have also avoided the worsening condition. The offset of the warning
the response(s) associated with escaping the warning stimulus. Another set of
located that indicates a study that has analyzed a Reflexive CMO for an improving
condition, but there have been a multitude of research studies that have analyzed
because a stimulus functions as a Reflexive CMO only if responses that eliminate the
stimulus are systematically correlated with removal of the worsening condition during the
temporally paired with another stimulus condition (S1) that precedes a worsening or
worsening conditions could occur. The behavioral process develops via an unconditioned
second stimulus (S2) is paired with the first stimulus (S1), where S1 becomes available
punishment. The second stimulus should acquire the ability to evoke or abate behavior
through second order conditioning of the UMO (either improving or worsening). There
has been some applied research on the Transitive CMO, but basic research that has been
conducted on nonhuman subjects has used operations that invoke a behavioral process
that appears to resemble a chained schedule. However, currently there has not been a
study that observes the manner by which the first stimulus in this operation functions
Conditioning CEO’s 16.
differentially than a chained schedule (Michael, 1993). The process of T-CMO’s has
conditioned motivating operations) evoke or abate behavior depend on the operations that
invoke them. Because these conditions (the operations themselves) originate in the
processes, so the general behavioral relation can be observed. Once established, this
general behavioral relation can be used to inform applied research, and subsequent
technology (Mace, 1996). As of yet, these general behavioral relations have not been
They are the aggregate effect of stimulus conditions on behavior. As such, a technology
designed to change socially valid behavior should comprise interventions that are molar
in nature, but the molecular underpinnings that control the dynamic nature of behavior
will determine more specifically exactly what comprises such behavioral relations.
In sum, behavior analysis will likely evolve into a science that uses its
technology is not available in a form that is accurate enough to record molar behavior
change within the repertoires of individuals, nor efficient enough to be used in applied
settings. The development of such technology will require the full understanding of
stimulus functions, and stimulus function changes. Much of the research in behavior
analysis has focused on the process of discrimination as the major determiner of stimulus
The purpose of this current study is to arrange functional relations in the manner
delineated by Michael (2000), and see if such operations result in the same processes he
stimulus function change will expand the science of behavior to inform methodologies
that can be used to differentiate between how various stimulus conditions function in real
following study was conducted to examine the how the Reflexive and Surrogate
Conditioned Motivating Operations develop and effect behavior, and how those
operations can be used to motivate socially valid behavior. It was hypothesized that
positive reinforcement can be arranged in such a fashion that a neutral stimulus can be
reinforcement.
Conditioning CEO’s 18.
Chapter 2
EXPERIMENT 1
Method
Subjects
children with multiple disabilities in a city of 35,000 residents in Southern Maine. One
participant completed each of the experiments. Participant 1 was a 17 year old male with
Anxiety Disorder (GAD). He had the visual acuity, language skills, and fine motor
All experimental sessions were conducted in a room (3 x 3.5 meters) with a 0.87
by 1.33 meter window and 0.67 by 1.75 meter table and two chairs. The room was
largely quiet and free of distraction, and was separated from the classrooms and other
high-activity areas of the school. 60% of the sessions were videotaped for purposes of
inter-observer agreement. All procedures for both experiments were reviewed and
with one 3 cm hole cut 3 cm from the top, with the center of the hole 31.5 cm from the
sides of the board, and two other holes 9 cm in diameter in line with one another
horizontally, 24 cm inches from the top of the board, and 15 cm from each edge of the
Conditioning CEO’s 19.
apparatus (one 15 cm from the left and one 15 cm from the right). Under each of the two
larger holes are slots, each 9 cm in width, and 1 cm in height. The small hole was
covered by white fabric, and contained 2 speakers which emitted tones. One speaker ran
at 6 volts and emitted a 400 Hz tone at 75 decibels. The other speaker ran at 12 volts and
emits a 4200 Hz tone at 75 decibels. The tone of the speakers was suppressed by 4 oz. of
cotton felt within the speaker box. Each of the larger holes were covered by a light
and lights. The left box (from the participant’s perspective) contained no lights, and the
right box contained 4-2 watt colored bulbs (red, green, yellow and blue) and one 10 watt
white bulb. The outside of the apparatus was painted light gray (see figure 1 for a
Figure 1. Schematic of the apparatus used for experiment 1 and experiment 2 from the
Buzzer Speaker
Light 1 Light 2
Plastic disks were used as a media for the dependent measure. Each plastic disk
was a flat cylinder 3.5 cm wide, 1.5 mm thick and that weighed 1.5 grams.
The dependent measure for Experiment 1 was the number of insertions of plastic
disks into the experimental apparatus. The insertion response was defined as the “passing
(number of disks inserted into the apparatus). Insertion rate was calculated by dividing
the number of disks inserted by the duration of time spent interacting with the apparatus
to obtain rate. Inter-observer agreement data were collected by counting the number of
disks inserted on 72% of the total sessions. These data were collected on at least one
function of the aggregate total of behavior, and therefore the smaller number of responses
of observer A or B in the entire session was divided by the larger of the two(either A or
B), and multiplied by 100. The average inter-observer agreement for this experiment was
100%.
reinforcement was delivered within 2 seconds of the target 30th correct response.
Through baseline and experimental sessions, the amount of time since last eating was
held between 120 and 135 minutes. The second independent measure was 10 minutes
sessions recorded on videotape. These data comprised the length of session and the rate
of reinforcement during the session. These data were collected on 14% of the
components and averaged 99% for session length, and 100% for reinforcer deliveries.
Experiment 1
Inter-observer 100%
agreement
Reinforcer Deliveries 100%
Session Length 99%
Procedure
reinforcement was counterbalanced across two sessions per day at 10:00 am and 2:15 pm
until stable responding was observed. The participant was told the rule which described
the schedule of reinforcement in the component to minimize the likelihood that rule-
governance would alter the probability of his responding. The experimenter said before
each session “every time you put 30 disks into the slot, you will receive Chex-Mix© as a
reward.” During the sessions, the participant placed disks through the slot in the right
side of the apparatus. During each session, either the red or white light was activated.
Each session of the experiment comprised one 10 minute component. During baseline,
both FR-30 schedules were preceded with 120-135 minutes of deprivation of access to
food. The baseline condition comprised 14 sessions that were randomized across
Randomization was achieved by assigning each of the conditions with a set of numbers
Conditioning CEO’s 24.
(i.e. 0-4, 5-9), and then using a random number list (digits 0-9) to chose the next
component. See table 2.2 for the randomized schedule of the sessions.
Conditioning Phase. During the conditioning phase, the white component was preceded
by a period of deprivation of food, while the red component was preceded by a satiation
by free access to Chex-Mix© period just prior to the start of the experimental session.
The conditioning sessions comprised 14 sessions randomized across the red and white
conditions, as well as morning and afternoon conditions. Although the quantity of time
for these sessions were variable, they lasted approximately 10 minutes. The experimenter
said “you may eat as much as you want, please let me know when you do not want any
more” to initiate these sessions. The quantity of Chex-Mix© consumed varied across
Test Phase. The persistence of the effects of satiation was observed through a
comparison between the red and white lights across baseline, conditioning, and test
conditions. The test condition comprised a multiple FR-30 schedule and was
implemented in one session. 10-1 minute components were presented to the participant
with a 30 second dark-key interval between each component. The conditions (red light
vs. white light) were presented in the following randomized pattern {W, R, W, R, R, W,
component (0-4, 5-9), and then drawing a number from a computer generated single digit
Conditioning CEO’s 26.
(0-9) random number generator. Data collected within this session are displayed in figure
3.
Results
Baseline responding was consistent across red and white conditions. Mean
baseline responding in the red condition was 49.1 disks per minute and 50.1 disks per
The mean deprivation interval in the AM sessions was 120 minutes since last food
and the mean deprivation interval was 135 minutes since last food for the PM sessions.
sessions and during 43% of the PM sessions. The average quantity of Chex Mix©
consumed during each session of the satiation condition was 128.6 grams. See table 2.2
for a complete descriptive analysis of the individual components by session, and table 2.3
Table 2.3.
Disk insertions per minute were the dependent measure that was graphed on
Microsoft Excel for visual interpretation. The suppressive effect of satiation was
calculated as a function of baseline rates of responding. The differences in the red and
white conditions were calculated using the same formula through baseline and
conditioning phases. The formula used to calculate the effect of satiation on responding
is r(W) – r(R) / r(W) where r(W) is the response rate in the presence of the white stimulus
light, and r(R) is the rate of response in the presence of the red light. As aforementioned,
the deprivation condition was correlated with the white light and the satiation condition
is evidence of a Surrogate CMO. Within the baseline phase, responding in the red
the conditioning phase, responding after the satiation procedure suppressed responding
during the red condition 12.4% lower than the deprivation condition {(W-R)/W}. Figure
3 displays this relationship. Within the test condition the rate of responding in the red
Figure 3 graphically displays this relation. This may indicate evidence of the Surrogate
CMO, as a 2.58% difference between the red and white components in the test condition
comprises 20.8% of the effect of satiation when there was direct temporal contiguity
between the satiation procedure and the associated stimulus (red light). See figures 4 and
Figure 4. Differences between red and white components during baseline, conditioning
condition
Conditioning CEO’s 29.
Discussion
1938, 1953). However, evidence of the persistence of the effect of satiation is scant at
operation may persist even when the satiation operation is removed. In experiment 1 this
Conditioning CEO’s 30.
was demonstrated by first pairing the red light with a satiation condition multiple times,
and then presenting the red light within a deprivation condition. Compared to the control
condition (i.e., white light), rates of responding were lower during the experimental and
Although the magnitude of the difference between the red and white conditions is
less than what is expected in applied research, they do support evidence of a Surrogate
Limitations of Experiment 1
satiation on responding, and the reduced discriminative control that was conditioned
during the satiation procedure. There are numerous operant processes occurring, and it
reduction in responding that was observed could have weakened the control of the
discriminative stimulus that was established during baseline. One feature of behavior
that determines the contingency between the stimulus and the response is the frequency
of reinforcement. If this is the case, the conditioning phase of this experiment may have
simply weakened the contingency between the response and the reinforcing stimulus,
because there were fewer contacts between the responses and the reinforcing
consequence (Catania, 1994). Because the participant responded less in the conditioning
phase and therefore received lowered rates of reinforcement, the contingency was likely
Conditioning CEO’s 31.
weakened. Moreover, because of the value altering effect of the satiation procedure in
the conditioning phase, the occasioning effect of the SD may have been diminished.
2. Because the participant was required to respond laterally across the experimental
apparatus, it is possible that handedness may have played a role in respond rate.
Future Research
that could parametrically differentiate between the effect of an Abative Surrogate CMO,
baseline phases on a fixed ratio schedule would run in a randomized fashion each
correlated with a different salient stimulus condition, and preceded by a period of food
deprivation. During the conditioning phase, one component would remain identical to
baseline, and like the experiment conducted in this study, one component would be
preceded by a satiation procedure while the third component would comprise a fixed
interval schedule which matches the frequency of reinforcement of the previous satiation
component. For example, if the last FR-30 satiation procedure resulted in the participant
be FI-30 sec. After the period of conditioning, the test phase would be very similar to the
test phase in this experiment with the exception of three components being compared
Conditioning CEO’s 33.
conducted in one session with the total amount of time comprising the test session equal
to the same amount of time that comprised each session of the baseline and conditioning
phases (e.g., the participant would contact the stimulus conditions for the same amount of
time during the test as s/he did during each session of the baseline and conditioning
phases). It must be acknowledged that this would require a great deal of time on the part
of the participant, and especially in the case of people with developmental disabilities the
ethics of the cost-benefit analysis associated with such an inquiry should be thoroughly
Chapter 3
EXPERIMENT 2
Method
Subjects
Participant 2 was a 13 year old male with a diagnosis of Autistic Disorder. His
current level of intellectual functioning is in the range of mild mental retardation (IQ =
65). He had the visual acuity, language skills, and fine motor control necessary to
The same experimental setting and apparatus used for experiment 1 was used in
experiment 2.
The dependent measure for experiment 2 was the rate (per min) of the insertion of
plastic disks into the experimental apparatus. The insertion response was defined as
removing a disk from a container in front of the apparatus and placing it through a slot in
count. Global quanta of responses were the unit of behavior analyzed. A second
independent observer collected inter-observer data on 41% of the total sessions. Some of
the inter-observer agreement data were collected via a video tape and some of the data
were collected through examination of the quantity of disks inserted after the session.
Conditioning CEO’s 35.
average inter-observer agreement of the sessions analyzed for experiment 2 was 98.23%.
These data were collected on at least one baseline and one experimental session. Inter-
observer agreement was calculated as a function of the aggregate total of behavior, and
therefore the smaller number of responses of observer A or B in the entire session was
divided by the larger of the two (either A or B), and multiplied by 100. This method was
used because global quanta of data were compared, not the direct contiguity of molecular
Integrity data were collected on the integrity measures of reinforcer deliveries and
length of session on 18% of the sessions. These data were collected via a videotape by a
delivery, and a correspondence of 98.7% on session length. Again global quanta of data
were analyzed where the total length of sessions in seconds was compared and the
Experiment 2
Inter-observer 98.2%
agreement
Reinforcer Deliveries 100%
Session Length 98.7%
schedules of reinforcement, and the temporal contiguity of colored lights and a tone. The
sample cup used to deliver reinforcement had a 4 fluid ounce capacity, and was filled
with 3 kernels of popcorn. Microwave popcorn (ACT© 94% fat-free butter) was used as
schedule of reinforcement. After baseline was established, the second set of independent
measures was presented in the form of colored lights and a tone. These events occurred
just prior to the presentation of the schedules of reinforcement (see procedure for details).
Procedure
Baseline Phase. Reinforcement was arranged in a two component multiple schedule that
occurred twice per day (i.e., 10 am and 2:30 pm). Each session comprised four-2-minute
components that were counterbalanced and randomized (i.e., the conditions were
alternated during sessions and counterbalanced across sessions) to control for an order
effect. The sessions were limited to four-2 minute components because the participant
was observed to become highly distracted after a brief period of time. The first
component of the multiple schedule was correlated with a white light and a FR-10
schedule of reinforcement. The second component was correlated with a blue light and a
FR-1 schedule of reinforcement. The participant was told the ratio of responses to
Specifically, the participant was told “You may put these colored disks into the slot to get
Conditioning CEO’s 37.
popcorn as a reward. Every time you put 1(or 10) disk(s) into the slot you will receive
popcorn.”
Training Phase. During the training phase, a two component multiple-chained schedule
conditions. An FR-5 schedule that was correlated with a red light and a tone was
presented to the participant. Upon completion of the FR-5 schedule (inserting 5 disks
into the slot), the red light, and tone shut off and a blue light correlated with a FR-1
schedule of reinforcement appeared. The blue light and FR-1 schedule of popcorn
reinforcement remained until the end of the component. This situation was alternated (in
a counterbalanced fashion) with a component where a red light was correlated with a FR-
5 schedule (but no tone), and upon completion of the FR-5 schedule, the red light shut off
and a white light appeared. The white light was correlated with a FR-10 schedule of
reinforcement, and remained on until the end of the component. Training occurred twice
per day both in the morning and in the afternoon (i.e., at 10:00 am and at 2:15 pm).
Table 3.3 Descriptive data from the Baseline and Conditioning phases of experiment 2
fr1 Blue 6
7 PM fr10 White 18
fr1 Blue 5.75
8 AM fr1 Blue 6.25
fr10 White 16
9 AM fr10 White 17
fr1 Blue 6.5
10 AM fr1 Blue 7.25
fr10 White 17.5
11 PM fr10 White 18.5
fr1 Blue 7
12 AM fr10 White 16.5
fr1 Blue 6.75
13 PM fr1 Blue 6.25
fr10 White 15.75
14 PM fr10 White 16
fr1 Blue 6.5
15 AM fr10 White 24.75
fr1 Blue 6.5
16 AM fr10 White 13.5
fr1 Blue 7.25
17 AM fr10 White 16.75
fr1 Blue 7.25
18 PM fr1 Blue 7.75
fr10 White 20.25
19 AM fr10 White 17
fr1 Blue 8.75
schedules of reinforcement. In these phases the participant was able to place disks in
either the right or the left slot in the apparatus. During this phase the participant had
access to red, white and blue disks at the same time. During baseline and conditioning
phases the participant had access to only one color of disk at a time.
Within the first test phase, during the first component, on the right side of the
apparatus, an FR-5 schedule in the presence of a red light and a tone lead to a white light
correlated with a FR-10 schedule of reinforcement. On the left side of the apparatus
within the first component, the insertion of one disk stopped the tone and lead to a white
In the second component of the first test phase, the right slot was correlated with a
yellow light and a tone on an FR-5 schedule which lead to a white light and a FR-10
schedule of reinforcement. In this component, the insertion of one disk in the left slot did
not terminate the tone. This arrangement tested if the termination of the tone has
In the first component of the second test phase, the right slot was correlated with a
red light and a FR-5 schedule which lead to a blue light and a FR-1 schedule of
reinforcement. The insertion of one disk into the left slot produced the tone.
In the second component of the second test phase, a yellow light was correlated
with FR-5 schedule and lead to a blue light and a FR-1 schedule of reinforcement.
However, the insertion of one disk into the left slot did not produce the tone. This
Conditioning CEO’s 41.
Across both test phase 1 and test phase 2, the terminal link was 2 minutes in
Results
Figure 6 displays data collected under the baseline and conditioning phases of
experiment 2. Baseline indicates a differential effect of the FR-1 and FR-10 schedules of
reinforcement on the rate of disk insertion. As expected, mean response rate was higher
for the FR-10 component than for the FR-1 component (Ferster & Skinner, 1957).
Mean response rate for the FR-1 condition in baseline was 6.38 disk insertions per
minute and 16.54 disk insertions per minute for the FR-10 condition. During the
conditioning phase mean response rate for the FR-1 schedule was 7.53 disk insertions per
minute and 19.22 disk insertions per minute for the FR-10 condition. Overall response
rates were elevated slightly during the conditioning phase possibly due to slightly
increased ratio of response to reinforcer delivery (e.g. a FR-5 initial link was added prior
to both the FR-1 and FR-10 schedules of reinforcement in the conditioning phase). This
operation thinned the overall reinforcement schedule for each component, and may be
Differential responding was observed across test phase one and two. Specifically,
the participant inserted disks in the left slot of the apparatus at a rate of 27 insertions per
minute during test phase 1, and at a rate of 0.25 insertions per minute in test phase two.
Likewise, differential responding between test phases was observed on the right side of
Conditioning CEO’s 42.
the apparatus as the participant inserted 17 disks per minute in test phase one, and 33
Within test phase one, responding was similar across the left and right stimulus
conditions. The participant inserted an average of 27 disk insertions per minute into the
left slot, and an average of 17 insertions per minute into the right slot.
During test phase two differential responding was observed across the concurrent
components as well. The participant inserted an average of 33 disks per minute into the
right slot, but only 0.25 disks per minute into the left slot. See Figures 7 and 8 for
Figure 8. Interactions of response rate within and across test phases of Experiment 2
Discussion
within test phases one and two. These results indicate that there may be evidence of a
Reflexive CMO using the operations in this study, but it is not compelling.
improving condition may make that neutral event become a motivationally relevant
and a response eliminates that stimulus and also eliminates the improving condition, the
probability that the previously neutral event will motivate behavior should decrease. The
most salient comparisons in the test data are both within test conditions and across test
Conditioning CEO’s 44.
conditions because each test condition supported the hypothesis of the tone to either
These relations are illustrated in Figures 9 and 10. Responding was significantly
lower in the left component of test phase two than in the left component of test phase 1
(see figure 8). This indicates that the stimulus condition (red light plus tone) that was
terminated upon a response and led to a improving condition, suppressed behavior. The
operations used in the study dramatically changed the probability of behavior across red
and across test conditions 1 and 2. There was a great deal more responding in the right
component of test phase 2 than test phase 1. Operationally, it would be expected that a
higher response rate would be allocated to the right component of test phase 1, as that
component was correlated with a thinner schedule of reinforcement (test phase 1 was
correlated with a FR-10 schedule of reinforcement, and test phase 2 was correlated with a
phase two may have been motivated not only by the FR-1 schedule of the right
component, but also because the right component was the only “good” choice. The
participant may have responded to the right component because historically responding to
the left component would have stopped the tone condition, which once eliminated-denied
Within each testing session responding was consistent (see Figure 8). However in
test phase one the participant showed very similar rates of responding during each of the
Conditioning CEO’s 45.
components. Alternatively, in test phase two, the participant displayed highly different
response rates across the left and right components. The stimuli correlated with the two
components in test phase one had similar evocative power after conditioning, so the
responding in test phase two indicates that the stimuli correlated with each of the two
Limitations of Experiment 2.
During baseline, one aberrant data point was noted where the participant
optimized his responding. Instead of reaching into the can of disks, retrieving one disk,
and placing it through the slot, he picked up multiple disks, and held them in one hand
close to the slot and feed them through the slot one by one (i.e., optimization). That
session is denoted by an arrow in Figure 5. Just prior to the commencement of the next
session, the participant was reminded of the correct procedure: placing one chip into the
slot at a time. During the session where the optimization occurred, no correction trial
was used to redirect responding, because such a correction could have functioned as
punishment. It should also be noted that overall response rates increased slightly over the
course of the baseline and conditioning phases. This may have been due to the
participant intermittently optimizing insertions over the course of the experiment (i.e., he
was observed to occasionally pick up multiple disks in his hand and insert them quickly).
Although he did not consistently engage in this response pattern, he did slightly optimize
his responding on occasion throughout the latter stage of baseline and conditioning.
Conditioning CEO’s 46.
predicted. He would frequently take the small cup of popcorn and dump it into his hand,
and then use his free hand to respond. When he earned the next cup of popcorn, he
occasionally dumped it on top of the first pile of popcorn. This may have interrupted the
2. Because the participant was required to respond laterally across the experimental
apparatus, it is possible that handedness may have played a role in respond rate.
The most major limitation of this study is the applicability of the experimental
demanded by this sort of inquiry requires the level of procedural accuracy only
These data indicate very tentative evidence that the operations used in experiment
2 produced a change in the behavioral function of a previously neutral event that was
consistent with the Reflexive CMO as conceptually delineated by Michael, (1982, 1993,
2000). However due to the limitations described above, the results should be interpreted
very cautiously.
Future Research:
condition, one response, and a consequence, would yield a more accurate embryonic
CMO.
reinforcement. After a stable baseline is observed, a neutral event (tone) is presented for
a period of time. If responding occurred during the tone, the next series of responses
would not be reinforced. If no responding occurred during the tone, reinforcement would
be available after the tone shut off on its own. The number of reinforcers missed if
responding occurred during the tone would be parametrically equivalent to the number of
reinforcers missed (on average) that comprises the duration of the tone. This would
would be evaluated using a cumulative record, where it could be observed whether the
CMO for a improving condition by suppressing behavior that eliminates the availability
of a improving condition.
Conditioning CEO’s 48.
Chapter 4
GENERAL DISCUSSION
The purpose of the two experiments in this dissertation was to document a change
the observation of change in probability (or rate) that is the most important determinant in
validating effect (Skinner, 1966). The magnitude of such effect is not necessarily an
important variable in assessing the validity of the principle, certainly not as important as
function change (Skinner, 1938; Catania, 1994; and Michael 1982). These processes
result in a previously neutral event becoming a stimulus that changes the probability of
behavior. There are both respondent and operant relations that can alter the behavioral
class of stimulus function: conditioned elicitors. Operant relations result in the remaining
four classes of stimulus function: Discriminative Stimuli (SD and SΔ), Conditioned
preparation of the Surrogate CMO as proposed by Michael (1993) can produce a process
may still question the validity of the results. An analysis of other satiation procedures
was conducted to see the level of suppression observed by other researchers. Nevin,
Tota, Torquato, and Shull (1990) observed the responding of three white carneau pigeons
Conditioning CEO’s 49.
which were previously deprived of food until they reached 80% of their free feeding
weight. During their inquiry the authors observed the pigeons responding to various
indicates that responding was suppressed in experiment one of that paper by an average
research.
The schedules of children in educational settings do not take into account the
neutral event can alter the future probability of responding within the context of that
types of operations that change the value of primary reinforcers. It is possible that the
consistent correlation of events that increase the value of a reinforcer and events that are
otherwise unrelated to the same reinforcer, may acquire the value and behavior altering
effects of the UMO. For example the long term contiguity of an individual subject (i.e.
math) with the end of school day may cause stimuli conditions related to math (i.e. math
work sheets) to acquire the value and behavior altering effects of fatigue. Although
research has been conducted on the immediate effects of fatigue on academic responding,
the longitudinal effects of correlating a neutral event with fatigue have not been analyzed.
Conditioning CEO’s 50.
educational settings. There are two inverse ways a R-CMO can be used. First, it can be
used to increase behavior that results in the avoidance of an aversive. This operations has
widely been used in schools with minimal success. Often the punishment operations used
in schools are not very aversive, and may inadvertently serve to reinforce behavior. Thus
during a warning stimuli results in a improving condition. This set of conditions is likely
to also be effective in an educational setting. For example academic behavior often does
not get completed when it is supposed to during the day at school. One use of the R-
CMO could be the correlation of an academic task, with a improving situation (i.e. free
time). If interfering behavior (i.e. talking loudly etc.), occurs within that period of time, a
transition to another academic task is invoked, and the initial academic task is removed.
If the initial academic task is removed, the free time condition would not be available,
however if no interfering behavior occurs, free time would be available. Thus the initial
academic task would serve as a “promise stimuli” that indicates the improving condition
programming for these participants, close observation during the session did result in new
information that could be used in programming. During the initial stages of experiment
questioning, the participant reported that he did not know what the schedule was, and was
using covert verbal behavior to try to figure out why he was completing the study. His
covert verbal behavior likely formed rules which altered his behavior independent of
environmental contingencies. Once told the contingency, (e.g. every time you put 30
chips into this slot you will receive Chex-mix), his behavior stabilized. This information
work, but to figure out why it is relevant. During experiment 2 the participant was
able to use vocal language, his responding especially in the baseline and conditioning
The operations used in both experiments were conducted in molar manner. That
is, they were designed to inform a behavioral process across two discriminated operants.
Specific to experiment 2, the resulting process should have been observable in the
differences in overall responding in the two test phases. Although a behavior change was
observed across and within the test phases, it would be a deductive assumption to cleanly
attribute those changes to the manipulation of the independent variables during the
conditioning stages.
Conditioning CEO’s 52.
In the early stages of behavior analysis, Skinner proposed the molecular analysis
of behavior based on the premise of the reflex. Within this paradigm the direct contiguity
of the interactions of the organism and the environment comprised a unit of behavior that
were used in his initial explorations. As behavior analysis has evolved, behavior has
been viewed as encompassing time, and not necessarily being composed only of the
immediate direct contact between the environment and the organism (Baum and Rachlin,
1969). Although the molecular interpretation of behavior is perhaps not the most
2000), the initial components of newly discovered units of behavior may most accurately
However, because the general behavioral relation that results from the Reflexive CMO is
yet unknown parametrically, a molecular analysis which first describes the process
resulting from the Reflexive-CMO may be most informative to the science of behavior
Humans as a species are a unique member of the animal kingdom. Our ability to
use language, and to use symbols to permanently record language has created a context
represent a significant deviation from other species. This deviation extends conceptually
to motivation, as humans are motivated by patterns of word in the form of rules. These
another person. These rules develop within each person, but also across people and
Conditioning CEO’s 53.
comprise part of the cultural context within which language and contingency develop.
Because of such a unique context of motivation, it would by untenable to study such a set
of variables in organisms that do not have the biological framework to support such
behavior. Thus, it is logical that a mature science of human behavior isolate and verify
behavior analysis will begin to view response rate as a dependent measure that is tenable
References
Baum, W., & Rachlin (1969). Choice as time allocation. Journal of the Experimental
Brown-Chidsey, R., Steege, M.W., & Mace, F.C. (2008) Best practice in evaluating the
efficacy of intervention using case study data. In A. Thomas, & J. Grimes (Eds.),
Carr, E.G., & Durrand, V.M. (1985) Reducing behavior problems through functional
Catania, A.C. (1994). Learning; 4th edition. Upper Saddle River, NJ: Prentice-Hall.
Calvin, J.S., Bicknell, E.A., & Sperling, D.S. (1953). Establishment of a conditioned
Cravens, R., & Renner K.E. (1970). Conditioned appetitive drive states; Empirical
D’Amato, M.R., Fazzaro, J., & Etkin, M. (1968). Anticipatory responding and
Ferster, C.B. & Skinner, B.F. (1957). Schedules of Reinforcement, Action, MA: Copely
Publishing Group.
Conditioning CEO’s 55.
Hayes, S. C., Barnes-Holmes, D., & Roche, B. (2001). Relational Frame Theory: A
267-272.
Johnston, J, & Pennypacker, H.S. (1980). Strategies and Tactics of Human Behavioral
Keller, F.S., & Shoenfeld, W.N., (1950). Principals of Psychology. New York: Appleton
Century-Crofts.
Killeen, P. (1992). The mechanics of the animate. Journal of the Experimental Analysis
Laraway, S., Snycerski, S., Michael, J., & Poling A. (2003). Motivating operations and
Mace, F. C., Gritter, A., Johnson, P., Malley, J., & Steege, M.W. (2006). Contingent
http://jackmichael.org/publications/presentations.index.html
Neef, N.A., Mace, F.C., Shea, M.C., Shade, D. (1992). Effects of reinforcement rate and
Nevin, J.A., Tota, M.E., Torquato, R.D., & Shull, R.L. (1990). Alternative reinforcement
Prebrazhenskaya, L.A., Ioffe, M.E., & Mats, V.N. (2005). The effects of the quality of
Skinner, B.F. (1938). The Behavior Of Organisms, New York: Appleton Century-
Crofts.
Skinner, B.F. (1953). Science and Human Behavior, New York: The MacMillan
Company
Conditioning CEO’s 57.
Skinner, B.F. (1966). What is the Experimental analysis of behavior. The Journal of the
Wahler, R. G., & Fox, J. J. (1981). Setting events in applied behavior analysis: towards a
14, 327-338.
Conditioning CEO’s 58.
Paul E. Johnson Jr. completed his B.A in Psychology, M.S. in School Psychology
enjoys research in the experimental analysis of behavior and providing clinical service to
people with developmental disabilities. He lives in Windham Maine with his wife Beth.