REINFORCER SUBSTITUTABILITY: IMPLICATIONS FOR THE
ASSESSMENT AND TREATMENT OF STEREOTYPIC SELF-INJURY
BRIDGET A. SHORE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I wish to thank those individuals who have helped to make these experiments
possible. I thank Dr. Brian Iwata, my committee chair and advisor, for his professional
support and advice throughout this project. I also thank the other committee members, Dr.
Marc Branch, Dr. Timothy Hackenberg, Dr. Jeffrey Farrar, and Dr. Steven Smith, for their
kind assistance. Additional thanks are extended to Han Goh, Iser Deleon, Richard Smith,
and Sung Woo Kahn, who acted as therapists for the experiments. I also thank my parents
for their emotional and financial support. Finally, I thank my friend and companion,
Joshua, who never doubted for a moment that I could do this.
TABLE OF CONTENTS
LIST OF FIGURES.......................................................................... iv
Behavior Disorders Maintained by Automatic Reinforcement...................... 2
Behavioral Interventions for Stereotypic Self-Injury....................... ...... 4
The Matching Law and Reinforcer Substitutability......................... ........ 7
Reinforcer Substitutability and Applied Behavior Analysis............................. 10
GENERAL METHOD........................................................................... 13
Subject Descriptions....................................................................... 13
Functional Analyses...................................................................... 14
Stimulus Preference Assessments...................................................... 14
Response Measurement and Reliability................................................ 15
STUDY ONE.................................................... ............................. 18
Experiment One......................................................................... 18
Experiment Two.......................................................................... 22
Experiment Three......................................................................... 32
STUDY TWO................................................................................ 43
Experiment One......................................................................... 45
Experiment Two............................................................................ 48
Experiment Three....................................................................... 54
GENERAL DISCUSSION................................................................ 61
BIOGRAPHICAL SKETCH.................................................................. 74
LIST OF FIGURES
1 Percentage of intervals of hand mouthing and toy playing during alone and
toy play conditions for Caryl, Merry, and Randy................................. 21
2 Percentage of intervals of hand mouthing/arm rubbing during the alone
baseline and across DRO conditions for Caryl, Merry, and Randy.
Numbers above each condition reflect DRO interval length (top #)
and toy access time (bottom #)................................................... 26
3 Mean percentage of intervals of hand mouthing/arm rubbing during alone
baseline and across DRO conditions for Caryl, Merry, and Randy.
Numbers below each DRO condition reflect DRO interval length
(top #) and toy access time (bottom #).......................................... 30
4 Percentage of intervals of hand mouthing/arm rubbing and toy playing
during the alone baseline and across string length conditions for Caryl,
Merry, and Randy. Numbers above each condition reflect length of
the string attached to toy (top #) and proportion of string length while
the subject was seated in an upright position (bottom #) ........................ 36
5 Mean percentage of intervals of hand mouthing/arm rubbing and toy
playing during each string length condition for Caryl, Merry, and
Randy. Numbers at the bottom of each graph reflect length of the string
attached to toy (top #) and proportion of string length while the subject
was seated in an upright position (bottom #) ..................................... 40
6 Percentage of intervals of hand mouthing and eating during food and alone
conditions for Chuck and Matt ..................................................... 47
7 Percentage of intervals of hand mouthing and eating during each of the food
conditions for Chuck and Matt. Numbers above each graph reflect the
number of cookie pieces (Chuck) and the milliliters of pudding (Matt) ........ 51
8 Mean percentage of intervals of hand mouthing and eating across food
conditions for Chuck and Matt. Numbers below each graph reflect
the number of cookie pieces (Chuck) and milliliters of pudding (Matt) ........ 53
9 Percentage of intervals of hand mouthing and number of blocks placed
in bucket for Chuck across baseline and reinforcement condition................ 58
10 Percentage of intervals of hand mouthing and eating during the alone and
DRO conditions for M att ........................................................... 60
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirement for the Degree of Doctor of Philosophy
REINFORCER SUBSTITUTABILITY: IMPLICATIONS FOR THE ASSESSMENT
AND TREATMENT OF STEREOTYPIC SELF-INJURY
Bridget A. Shore
Chair: Brian A. Iwata
Major Department: Psychology
The concept of reinforcer substitutability proposes that in any given situation, a
continuum of possible interactions among reinforcers can exist. At one end of this
continuum, reinforcers are functionally similar and therefore "substitutable," with one
reinforcer being readily traded for another. At the other end of the continuum, reinforcers
are consumed jointly and are therefore "complementary," with consumption of one
reinforcer resulting in an increase in consumption of another reinforcer. Study 1 examined
the interaction between substitutable reinforcers with three developmentally disabled
individuals who engaged in self-injurious behavior maintained by automatic reinforcement
Two individuals who engaged in hand mouthing, and a third who engaged in arm rubbing,
participated. Results of three experiments showed that (a) substitution effects occurred
when toys and hand mouthing/arm rubbing were concurrently available, with subjects
showing a strong preference for toy manipulation; (b) attempts to reduce hand
mouthing/arm rubbing using the preferred toys as reinforcers in DRO contingencies were
unsuccessful for all three subjects; and (c) all three subjects' preferences for hand
mouthing/arm rubbing or toy play systematically changed when reinforcer "cost" was
varied. Study 2 examined the interaction between complementary reinforcers. Two
developmentally disabled men who engaged in hand mouthing participated. Results of
three experiments showed that (a) both subjects' hand mouthing increased when food was
consumed; (b) varying the amount of food had no effect on hand mouthing for either
subject; and (c) food reinforcers were ineffective in either decreasing hand mouthing (one
subject) or increasing appropriate behavior (the other subject). The results of the two
studies illustrate the importance of examining interactions among concurrently available
reinforcers when conducting reinforcer assessments.
It has long been recognized that behavior is affected in numerous and sometimes
complex ways by stimuli in the environment. For example, the probability of responding
is affected not only by events that precede and follow a particular behavior, but also by the
presence of alternative (concurrent) response options. In applied settings, where stimuli
are often not amenable to direct control, the analysis of environment-behavior interactions
is particularly troublesome when behavior produces its own consequences. The source of
reinforcement for such behavior is often unclear and, even if the stimulation can be
identified, it is usually difficult or impossible to separate the behavior from its product.
Therefore, most reinforcement-based interventions aimed at reducing automatically-
reinforced aberrant behavior have necessarily involved choices between concurrently
A considerable amount of basic research has examined the interaction between
concurrently available reinforcers. For example, the matching law (Herrnstein, 1961,
1970) was an initial attempt to account for the relativistic nature of reinforcement: The effect
of a given reinforcer is dependent upon the availability of other reinforcers within a
particular situation. Hemnstein's matching law was quite accurate in predicting outcomes
when concurrent reinforcers were qualitatively similar (and usually identical), but less so
when choices were between qualitatively different reinforcers (Baum, 1974). Subsequent
research has expanded our understanding of the nature of reinforcement by incorporating
economic principles, specifically that of reinforcer substitutability, into behavioral analyses
when reinforcers are not qualitatively identical (Rachlin, Green, Kagel, & Battalio, 1976).
Given the findings of behavioral economic research, an examination of the parameters
affecting the relationship between automatically-reinforced behavior and other
environmental stimuli seems long overdue. The purpose of the two studies that follow is to
examine both substitutable and complementary relationships between stereotypic self-injury
and environmental stimuli. In Study One, three experiments were conducted. Experiment
One was a simple demonstration of the substitutability of toy play for hand mouthing (two
subjects) and arm rubbing (one subject), Experiment Two examined the efficacy of DRO
using the toys as reinforcers, and Experiment Three examined the effect of altering
"response effort" for the preferred reinforcer (toy play) on the amount of hand
mouthing/arm rubbing and toy play. Study Two examined complementary relationships
between hand mouthing and food consumption. Experiment One demonstrated a
complementary relationship between hand mouthing and food consumption with two
subjects, Experiment Two examined the effect of amount of food on time spent hand
mouthing, and Experiment Three examined the efficacy of food as a reinforcer in a DRO
contingency (one subject) and on a free operant task (one subject).
Behavior Disorders Maintained by Automatic Reinforcement
Although considerable research has shown that many behavior problems are
maintained by social reinforcement such as attention from caregivers (Day, Rea, Schussler,
Larsen, & Johnson, 1988), or escape from task instructions (Iwata, Pace, Kalsher,
Cowdery, & Cataldo, 1990), it has also often been found that some behavior problems
persist in the absence of social consequences (Iwata et al., 1994). Persons with severe
developmental disabilities have frequently been found to engage in behaviors that appear to
be maintained by directly (automatically) produced sensory consequences. Examples of
such behavior include repetitive and rhythmic "stereotypic" movements such as body
rocking, object manipulation, and complex hand and finger movements (Repp & Karsh,
1990), and some cases of self-injury such as hand mouthing (Rast & Jack, 1992),
trichotillomania (Rothbaum, 1992), aerophagia (Barrett, McGonigle, Ackles, & Bruhart,
1987), and pica (Danford & Huber, 1982).
Functional analyses designed to identify variables affecting the occurrence of
behavior disorders (Carr & Durand, 1985; Iwata, Dorsey,Slifer, Bauman, & Richman,
1982; Mace & Lalli, 1991; Wacker, et al., 1990) have been demonstrated as useful tools
for identifying social functions for problem behavior and in the subsequent development of
effective treatments. For example, successful interventions based on results of functional
analyses have been demonstrated for self-injury maintained by escape from instructions
(Iwata, Pace, Kalsher, Cowdery, & Cataldo, 1990; Steege, Wacker, Berg, Cigrand, &
Cooper, 1989), and by contingent attention (Day et al., 1988; Mazaleski, Iwata, Vollmer,
Zarcone, & Smith, 1993). In addition, there is evidence that identification of the
maintaining variables improves the efficacy of treatment (Iwata, Pace, Kalsher, Cowdery,
& Cataldo, 1990; Repp & Karsh, 1990). The utility and generality of functional analyses
have also been supported by numerous studies across response topographies, subject
populations, and settings (e.g., see Iwata, Vollmer, & Zarcone, 1990; and Mace, Lalli, &
Pinter-Lalli, 1991, for reviews).
As with the assessment of socially-mediated behavior, identification of sources of
control for nonsocially-mediated behavior could lead to more effective treatment
interventions. But unlike socially-mediated behavior, for which the maintaining variables
can be identified among observable events, many automatically-reinforced behaviors have
their sources of control from unobservable events that are difficult to identify and
manipulate. Typically, researchers and therapists would assume automatic reinforcement
as the maintaining variable based on a functional analysis when exhaustive assessment
manipulations produced undifferentiated results, or if the behavior persisted in an alone
condition (Shore & Iwata, in press). Therefore, automatic reinforcement has been assumed
to be the maintaining variable based on a process of elimination of possible social
functions, rather than on identification of a specific source of control. For example, Iwata,
et al. (1994) recently presented results of functional analyses of SIB obtained for 152
subjects over an 11-year period. Interestingly, automatic reinforcement was identified as
the maintaining variable (through a process of elimination) for 25.7% (39) of the subjects.
Behavioral Interventions for Stereotypic Self-Iniury
Based on the assumption that some cases of self-injury or stereotypy are maintained
by their sensory consequences, one treatment strategy known as sensory extinction
(Rincover, 1978) has been to attenuate or eliminate the sensory consequences of the
behavior. For example, Rincover and Devany (1982) used a padded helmet for one boy
who banged his head, covered the floor and walls with padding for another boy who head
banged, and placed rubber gloves on a girl who scratched her face. Results showed
decreases in self-injury for all three subjects. Although this type of intervention has often
been shown effective in reducing stereotypic and self-injurious behavior, the underlying
process involved may not be extinction (Mazaleski, Iwata, Rodgers, Vollmer, & Zarcone,
1994). For example, does a helmet merely alter the sensory consequences of head hitting,
making it less reinforcing (extinction), or does it increase stimulation to the hand of an
individual who strikes the helmet (punishment)? Thus, it is unclear if the actual source of
stimulation has been identified and masked or if punishment is the behavioral process
involved in behavioral reduction.
There are several drawbacks in the use of sensory extinction. Assuming that the
procedure does represent a form of extinction, identification of the behavior's maintaining
reinforcers is required. However, many behaviors produce multiple modes of stimulation
(e.g.,visual, tactile, auditory). For example, hand mouthing produces tactile and
gustatorial stimulation, and object twirling produces tactile, auditory, and visual
stimulation, any or all of which could be the maintaining reinforcer. Another potential
problem with sensory extinction is that the procedure may be impractical (e.g., eliminating
visual stimulation) unless used on a response-contingent basis or intermittently. Finally,
extinction may produce an initial burst in responding and/or other emotional behavior that
may preclude its use (Luiselli, 1981).
Given that it may be difficult or impossible to use sensory extinction in some cases,
many researchers have focused on reinforcement-based alternatives. Examples include
continuous access to alternative reinforcers such as toys (Berkson & Mason, 1965;
Davenport & Berkson, 1963; Favell, McGimsey, & Schell, 1982), differential
reinforcement of alternative behavior or DRA (Lockwood & Bourland, 1982; Mulick,
Hoyt, Rojahn, & Schroeder, 1978), and differential reinforcement of other behavior or
DRO (Cowdery, Iwata, & Pace, 1990; Luiselli & Krause, 1981, Repp, Deitz, & Deitz,
Many studies have found that the availability of certain toys decreased the
occurrence of stereotypic behavior. For example, Favell et al. (1982) provided alternative
sensory activities to six profoundly retarded subjects who displayed self-injury. In all
cases, when these activities were available, eye poking, hand mouthing, and pica were
substantially reduced. Selection of the items was based on their similarity to the presumed
sensory consequences produced by the target behavior (e.g., brightly colored beads and
mirrors for the subjects who eye poked, large rubber balls and popcorn for the subjects
who engaged in pica). Thus, it was shown that, in some cases, environmental stimuli
could substitute for the automatically-produced reinforcers.
Another strategy for increasing the effectiveness of interventions that provide access
to items that substitute for aberrant behavior has been to use differential reinforcement of
alternative behavior (Favell et al., 1982; Lockwood & Bourland, 1982; Mulick et al.,
1978). For example, Favell et al. used social reinforcement to increase more appropriate
toy play (e.g., toy holding versus toy mouthing) and further decrease hand mouthing for
three subjects, and used popcorn contingent on appropriate toy usage to decrease toy
chewing and pica for two other subjects. Lockwood and Bourland (1982) showed that
although toys affixed to the wheelchair of a developmentally disabled subject reduced SIB
by approximately 50%, the addition of differential reinforcement (praise and brief physical
contact) for sustained toy use and the absence of finger biting was necessary to achieve
clinically acceptable effects. Similarly, Mulick et al. (1978) decreased finger picking and
nail biting by differentially reinforcing independent toy play with social and edible
Another approach, differential reinforcement of other behavior (DRO), involves the
delivery of reinforcement at the end of a predefined interval provided that the aberrant
behavior has not occurred during that time. Cowdery et al. (1990), for example, showed
that access to video games contingent on the nonoccurrence of self-scratching was an
effective intervention to eliminate self-injury. Repp et al. (1976) reduced the hairtwirling,
handbiting, and thumbsucking behavior of three developmentally delayed children in a
classroom setting by delivering praise and edible reinforcement in a DRO contingency.
Other studies have also shown DRO as an effective intervention for the treatment of
stereotypic behavior ( Luiselli & Krause, 1981; Repp, Deitz, & Speir, 1974).
Interestingly, interventions that do not include an extinction component (e.g.,
access to substitutable items as described above, or many cases of DRA or DRO)
necessarily produce conditions of choice between concurrently available reinforcers
because of the continued availability of the inappropriate behavior (and its resulting
consequences) and the arbitrary reinforcer delivered in the intervention. The parameters
that affect choice between concurrently available reinforcers have been studied extensively
by basic researchers and their findings have important implications for the assessment and
treatment of automatically-reinforced problem behavior.
The Matching Law and Reinforcer Substitutabilitv
Herrstein's matching law (1961, 1970) made explicit the relativistic nature of
reinforcement. The matching law predicted that the probability of a given response is
influenced not only by the reinforcers contingent upon it, but also by other reinforcers
contingent on other responses within the particular situation. The matching law stated that
the ratio of responses emitted for two reinforcers would be equal to the ratio of the
reinforcement obtained for those responses. A generalized version of the matching law
was proposed by Baum (1974), with additional parameters (sensitivity and bias) added to
account for cases in which deviations from matching occurred. A considerable amount of
basic research has examined the accuracy of the matching law in predicting relative
allocation of responding (see de Villiers, 1977, for a review). Most of this research,
however, has focused on reinforcers that differ in their frequency, amount, delay, or
probability, all of which are quantitative dimensions. For example, choice has typically
been examined in procedures with pigeons (or some other nonhuman animal) choosing
between identical food reinforcers that differ along one or more of these dimensions. Much
less research, however, has been conducted to examine the interactions between
qualitatively different reinforcers (Green & Freed, 1994). In its ideal form, the matching
law's assumes that reinforcers are perfectly substitutable, yet evidence from basic research
shows that under and overmatching occur frequently (Baum, 1979).
The concept of reinforcer substitutability is based on economic theory and was
proposed as a possible extension to the generalized matching law (see Green & Freed,
1994, for a review). This theory describes a continuum of interactions between
concurrently available reinforcers. At one end of this continuum are complementary
reinforcers for which increased consumption of one alternative would result in an increase
in consumption of its complement. For example, increased consumption of salty food
could result in increased consumption of water. At the other extreme of the continuum are
substitutable reinforcers, for which an increase in consumption of one alternative would
result in a decrease in consumption of its substitute. For example, a pencil may be readily
traded for a pen when writing. In the middle of the continuum, reinforcers are
independent, with consumption of one having no effect on consumption of another.
Substitutes and complements, however, are not fixed points along a continuum. For
example, although one may readily trade a pencil for a pen when jotting down a quick
note, that same person may seek out a pencil instead of a pen when doing a crossword
puzzle. Therefore, substitutability and complementarity are not static properties of single
reinforcers, but descriptions of the relationship between reinforcers in a given context.
A considerable amount of research over the past decade has been conducted to
examine the substitutability of qualitatively different reinforcers. That is, results of these
studies show that substitution effects can occur independent of initial preference.
Increasing the price of a preferred reinforcer through changes in rate, magnitude, or delay
to reinforcement has been shown to eliminate preference for that reinforcer over a
substitutable one (Rachlin et al., 1976). For example, in a basic demonstration of
substitutability (Kagel et al., 1975), rats responded for either root beer or Tom Collins mix
on concurrent fixed ratio (FR) schedules of reinforcement. Each reinforcer was associated
with a different response lever and FR requirement. The rats lived in the experimental
chamber and were limited to a fixed number of lever presses in a given 24-hr period. In
economic terms, the pairs of reinforcers represented different "commodities", the FR
requirement represented the "price" of each good, and the total number of lever presses
allotted represented the rats' "income". By altering the schedules for each reinforcer, the
"price" for each reinforcer could be varied. The results showed that both rats had a strong
preference for root beer when the "price" (FR requirement) and "income" (total number of
lever presses allotted) were equal. This relationship was then altered by making "income-
compensated price changes." This was accomplished by reducing the FR schedule
requirement (price) for Tom Collins by half and doubling the FR requirement (price) for
root beer, while adjusting the allotted number of lever presses (income) to obtain the same
combination of root beer and Tom Collins as was obtained in the first condition. Results of
this manipulation showed much more consumption of the now "cheaper" Tom Collins than
the now more "expensive" root beer. Thus, altering the price of the concurrently available
reinforcers eliminated the preference that had been shown when prices were equal. Studies
such as this show that the context in which a putative reinforcer is used can alter its
In a more recent case study on reinforcer substitutability in human subjects, Tustin
(1994) examined relative preference for reinforcers shown by subjects with developmental
disabilities under constant and several different fixed-ratio schedule requirements. For the
first subject, two series of schedules were presented singly. In the first series, complex
sensory stimuli were presented in a series of FR schedules (e.g., FR 1, FR 2, FR 5, FR
10, FR 20). The second series was identical to the first series of FR schedules, but with
attention as the reinforcer. Results showed that, as the FR schedule requirements
increased, the subject increased responding more for the complex sensory reinforcer than
for the attention reinforcer. This study demonstrated changes in relative preference as FR
schedule requirements increased. In the second case study, another subject was exposed to
two series of concurrent fixed-ratio schedules. In the first series, the subject responded for
a choice between visual reinforcers consistently delivered on an FR 5 schedule, and
auditory stimuli delivered in several different FR schedules (the same series described in
the first study). Results showed an initially greater number of reinforcers were earned for
the auditory stimuli than for the visual stimuli when response requirements for visual
stimuli were lower than for auditory stimuli. This preference, however, was eliminated
when the schedule requirement for auditory stimuli became greater than that for the constant
requirement for visual stimuli. Thus, a substitutable relationship was demonstrated
between the visual and auditory stimuli. The third case study showed changes in
preference as a function of schedule requirements. The subject was exposed to one series
of concurrent schedules, with choices between constant color stimuli and complex sensory
stimuli, both delivered on the same FR schedule. The results showed that in the FR 1
schedule requirement, the constant stimuli were preferred over the complex stimuli. This
preference, however, switched as schedule requirements were increased, with increased
responding for the complex stimuli and decreased responding for the constant stimuli.
In light of these and other similar basic research studies demonstrating the
contextual nature of reinforcement, further analysis of the variables affecting stimulus
preference in applied settings seems warranted.
Reinforcer Substitutability and Applied Behavior Analysis
Numerous applied studies have been conducted in recent years assessing stimulus
preference. These studies have demonstrated the effectiveness of assessing preference for
stimuli prior to using them as reinforcers in teaching adaptive skills (Pace, Ivancic,
Edwards, Iwata, & Page, 1985) or in reducing problem behavior (Steege et al., 1989).
Basic research studies examining choice between qualitatively different reinforcers,
however, suggest that stimulus preference changes with changing context. The variables
affecting choice in applied settings and the extent to which assessing reinforcing efficacy in
one context is predictive of reinforcer efficacy in another context are clearly in need of
further investigation. Indeed, Green and Freed (1994), in their review of studies of
substitutability, emphasized that consideration of substitutability is necessary for an
adequate understanding of the interactions among reinforcers and their influence on
behavior. The study of reinforcer interactions, they suggested, is particularly relevant to
applied behavior analysts. In spite of the importance of such research, however, the
concept has received little attention in applied behavioral research.
Although not interpreted in terms of reinforcer substitutability, many applied studies
have shown substitution effects between environmental stimuli and stereotypic or self-
injurious behavior. For example, Favell et al. (1982) effectively decreased
developmentally disabled subjects' hand mouthing, pica, and eye poking by providing
access to toys, popcorn, and visual toys, respectively. Hence, by matching the presumed
source of sensory stimulation produced by the self-injurious behavior to that produced by
toy play, they were able to replace the more severe form of self-stimulation with a more
benign topography. Favell et al. suggested that these stimuli were "substitutes" for the
inappropriate behavior. Other studies have also shown substitutability between toys and
stereotypic or self-injurious behavior (Bailey & Meyerson, 1970; Rincover, 1978).
Davenport and Berkson (1963), however, showed that this reciprocal relationship varied
according to which toys were used. Therefore, this intervention may be effective only to
the extent that alternative sensory activities are preferred over the self-stimulatory behavior.
Although these simple demonstrations of substitution effects have been reported, the
parameters affecting these relationships have not been explored in any systematic manner.
Other applied research has shown that some behaviors tend to directly covary, with
engagement in one behavior resulting in engagement in another behavior. For example,
Knight and McKenzie (1974) examined the effects of time out from bedtime stories on the
thumbsucking behavior of three small children. All three girls were reported to suck their
thumbs most often at bed time, when being read to, and when holding a favorite blanket.
Thus, thumbsucking can be viewed as a complementary reinforcer to hearing bed time
stories and/or holding onto a blanket. Following a baseline in which the mothers read bed
time stories without contingencies for thumbsucking, the mother's stopped reading
whenever the child started to thumbsuck and only resumed when the child did not thumb
suck. Although the authors interpreted this as time out, another interpretation is possible.
The time out contingency may have merely reduced consumption of one reinforcer
(thumbsucking) by restricting access to its complement (bed time stories). Other studies
suggesting that self-stimulatory behavior may have a complementary relationship with other
reinforcers can be found in research examining situational effects on rates of stereotypy.
For example, Kaufman and Levitt (1967) found that body rocking displayed by
developmentally delayed subjects increased steadily prior to lunch and at times when
institutional staff changed shifts. For these subjects, stereotypic behavior may have had a
complementary relationship to eating meals and/or staff attention. Complementary
relationships such as these, and the parameters affecting those relationships, clearly warrant
In light of basic research examining the interaction between concurrently available
qualitatively different reinforcers, the interaction between automatically-reinforced problem
behavior and other environmental stimuli should be explored. Study One examined
substitutable relationships between stereotypic self-injury and preferred play items, and
Study Two examined the complementary relationship between hand mouthing and food
consumption. The experiments for both studies explored the contextual nature of these
types of interactions.
Five individuals with developmental disabilities participated in the two studies. All
lived in a public residential facility, and all had been diagnosed with severe/profound
mental retardation. The subjects were referred to a specialized program for the assessment
and treatment of their SIB. Three subjects, Caryl, Merry, and Randy, participated in Study
One. Two subjects, Chuck and Matt, participated in Study Two.
Caryl was a 30-year-old woman whose SIB consisted of hand mouthing that
resulted in tissue damage. She was non-ambulatory, displayed no expressive language,
and did not appear to respond to directions from caregivers. Merry was a 34-year-old
woman whose SIB also consisted of hand mouthing that resulted in substantial tissue
damage. She was non-ambulatory, displayed no expressive language, and did not appear
to respond to directions from caregivers. Randy was a 33-year-old man whose SIB
consisted of high rates of arm rubbing against a chair, table, or other stationary object His
arm rubbing produced only mild skin abrasions, but it interfered considerably with other
activities. Randy walked with an unsteady gate and did not display any expressive
language, but he did respond to a few simple requests. Chuck was a 45-year-old man
whose SIB consisted of head hitting and hand biting. Following assessment and
successful treatment for these behaviors, staff reported that Chuck also engaged in hand
mouthing when eating certain foods. Chuck was ambulatory and did not display any
expressive language, but did respond to simple requests. Matt was a 30-year-old male
whose SIB consisted of hand mouthing that resulted in tissue damage. He was non-
ambulatory, had no expressive language, and did not appear to respond to directions from
All of the experiments for both studies were conducted at a day program for the
assessment and treatment of SIB, located on the grounds of the subjects' residence.
Sessions were conducted by graduate students, with the location of the sessions always the
same within each condition. Therapy rooms contained chairs, tables, and other
furnishings, as well as materials that varied according to the conditions of the experiments.
Sessions lasted for 15 min unless otherwise noted. Between one and three sessions were
conducted each day, and sessions were typically conducted four or five days per week.
Functional analysis assessments as described by Iwata et al. (1982) were conducted
for all five subjects. Results for Caryl, Merry, and Matt showed that hand mouthing
occurred across all conditions with the highest percentage of intervals in the alone
condition, suggesting that SIB was automatically reinforced. Randy's functional analysis
showed that his arm rubbing occurred at high rates across all conditions, also suggesting
his SIB was automatically reinforced. Although no hand mouthing occurred during
Chuck's functional analysis for SIB, it was observed that when snacks were delivered, he
engaged in high rates of hand mouthing.
Stimulus Preference Assessments
Probe sessions were conducted for Caryl, Merry, and Randy to determine toy
preferences. Sessions were 10 min in length and consisted of continuous access to a toy.
Toys were selected for each subject based on their ability to compete with hand mouthing
(e.g., the toy that resulted in the lowest rate of hand mouthing/arm rubbing). Two large
plastic rings were selected for Caryl, a small plastic tube for Merry, and a vibrator for
Probe sessions were conducted for Chuck and Matt to determine food preferences.
Sessions were 15 min in length. For Chuck, different food items were placed on a paper
plate in front of him at the start of the session. Matt could not feed himself; therefore,
different food items were delivered on a spoon placed at his lips approximately every 10 s.
For both subjects, observers recorded occurrences of eating and hand mouthing, and the
food that was associated with the highest amount of hand mouthing was selected for
inclusion in the study. Kit Kat cookies was selected for Chuck and pudding was selected
Response Measurement and Reliability
The dependent variables for the experiments were operationally defined as follows:
hand mouthing (Caryl, Merry, Chuck, and Matt)--insertion of the hand or fingers past the
plane of the upper and lower lips, or protrusion of the tongue out of the mouth onto the
hand or fingers; arm rubbing (Randy)--scraping the arm against the surface of a stationary
object; toy play (Caryl, Merry, and Randy)--holding a toy in hand; eatin (Chuck and
Matt)--food entering the mouth or visible chewing motions of the mouth; blocks in bucket
(Chuck)--blocks putting a block into a bucket. Therapist behavior was also scored and
defined as follows: toy delivery (Caryl, Merry, and Randy)-- therapist hands the toy to the
subject; feeding (Chuck and Matt)-- therapist delivers the food item to the subject; prompts
(Chuck)-- therapist places a block in the bucket.
Subject and therapist behaviors were recorded on a hand-held computer (Assistant,
Model AST 102) during continuous 10-s intervals. Because the duration of hand
mouthing, toy play, and eating varied considerably, a partial-interval scoring procedure
was used, in which observers marked the occurrence of the behavior if it was observed at
all during a 10-s interval. Data were converted to the percentage of intervals during which
responding occurred. Arm raising was scored as responses per minute by dividing the
number of arm raises by the number of minutes in the session. Blocks in bucket, prompts,
and feeding were scored as number per session.
A second observer simultaneously but independently recorded data during at least
19% of the sessions in each condition (range across subjects, 19.4% to 48%).
Interobserver agreement scores were calculated by first dividing session time into
consecutive 10-s intervals. The smaller number of responses was divided by the larger
number of responses recorded during each interval, and those values were averaged across
the session. Interobserver agreement data were collected for SIB, toy play, and eating.
Table One shows the percentage of sessions with interobserver reliability and mean
percentage agreement scores obtained for all subjects during each of the experimental
Percentage of Sessions with Interobserver Reliability and Mean Percentage Agreement
Scores for Self-Injury, Toy Play, and Eating, during each Experiment of Study One and
Percentage of Sessions
SIB Toy Plav/Eat
SIB Toy Play/Eat
SIB Toy Play/Eat
Mean Percentage Agreement
For three of the subjects previously described, Caryl, Merry, and Randy, toys were
found during probe sessions that appeared to reduce drastically the occurrence of
stereotypic self-injury; that is, were substitutable for hand mouthing. The purpose of
Study One was to explore the relationship between toy play and hand mouthing or arm
rubbing. Three experiments were conducted: (a) a basic demonstration of the relationship
between toy play and hand mouthing; (b) an examination of the effectiveness of the toys as
reinforcers in DRO procedures to reduce hand mouthing; and (c) a parametric analysis of
the effects of response effort on the occurrence of hand mouthing.
In Experiment One, the relationship between toy play and hand mouthing was
examined when both reinforcers were continuously available. The conditions were
presented to each subject in a reversal (ABAB) experimental design.
Alone. The subject was alone in the room (with the exception of the observer),
with no toys or other materials available. No interactions occurred between the observer
and subjects. This condition was identical to the alone condition of the functional analysis.
Toys available. These sessions were identical to baseline sessions, with the
exception that toys were given to the subjects at the start of each session. For Caryl, two
plastic rings were placed on a tray attached to her wheelchair. For Merry, the plastic tube
was placed in her hand at the start of the session. For Randy, a vibrator was given to him at
the start of each session.
Results and Discussion
Figure 1 shows the percentage of intervals containing hand mouthing and toy
playing for all three subjects across conditions. Caryl's data are shown in the top panel of
Figure 1. During the alone condition, Caryl engaged in high levels of hand mouthing.
When toys were provided, however, high levels of toy play were observed, while hand
mouthing decreased to 0% for all six sessions. When the alone condition was reinstated,
hand mouthing returning to its previous high level. When toys were reintroduced, high
levels of toy play were observed, while hand mouthing again decreased to 0%.
Results for Merry are shown in the middle panel of Figure 1. During the alone
condition, Merry's hand mouthing was variable. When toys were introduced, hand
mouthing was virtually eliminated and was replaced with high levels of toy play. A return
to Baseline showed variable amounts of hand mouthing. When toys were reintroduced,
high levels of toy play were observed, while hand mouthing decreased.
Results for Randy are shown in the bottom panel of Figure 1. He engaged in a
considerable amount of arm rubbing during the initial alone condition. When the vibrator
was introduced, however, high levels of to play were observed, while arm rubbing was
observed only once during the four sessions. A return to the alone condition showed arm
rubbing again at high levels. When the vibrator was reintroduced, high levels of toy play
were again observed, while no occurrences of arm rubbing were observed during any of
the four sessions.
Thus, results obtained for each subject indicated that when toys were continuously
available, hand mouthing/arm rubbing were almost totally eliminated. These findings
Figure 1. Percentage of intervals of hand mouthing and toy playing during
alone and toy play conditions for Caryl, Merry, and Randy.
--- Hand Mouthing
S-- Toy Play
Lm -- --I I ____________-l- -.L GBk n- i -
suggest that the stimulation obtained from toy play was preferred over that obtained from
the hand mouthing/arm rubbing, and replicated the results of previous studies showing
substitutable relationships between toy play and self-stimulatory behavior (Berkson &
Mason, 1965; Davenport & Berkson, 1963; Favell et al., 1982; Goh et al., in press).
Results of basic research examining reinforcer substitutability, however, suggest that
substitution effects are context dependent: Parameters such as rate, magnitude, or delay to
reinforcement have been shown to totally eliminate preferences. Therefore, it seemed
reasonable to explore further the parameters affecting the substitutable relationship
demonstrated in Experiment One.
Cowdery et al. (1990) suggested that although providing alternative play activities
may be effective in reducing or eliminating self-injury or stereotypic behavior, it may not be
viable or practical for several reasons. They stated that these activities are often
unavailable, may require one-to-one supervision to ensure continued play, and can lose
their reinforcing functions due to satiation. In addition, it would be difficult to train other
skills if toy play had to be continuously available. Cowdery et al. evaluated the effects of
DRO in reducing the stereotypic scratching of a young boy, and suggested that, if effective,
DRO could be used during times when alternative activities were not available. Using
pennies that could be exchanged for preferred play items as reinforcers, results showed that
the DRO successfully eliminated the boy's self-scratching.
Most research examining the efficacy of DRO/DRA procedures to reduce
automatically-reinforced aberrant behavior, such as in the Cowdery et al. (1990) example,
have used arbitrary reinforcers without eliminating the reinforcement directly produced by
the aberrant behavior (i.e., extinction) (see Vollmer & Iwata, 1992, for a recent review).
Thus, the efficacy of differential reinforcement may depend on how well the stimulus used
as the arbitrary reinforcer competed with the maintaining reinforcer. One method to
increase the likelihood that a reinforcer will compete with that obtained from the aberrant
behavior would be to conduct a stimulus preference assessment. For example, Steege et al.
(1990) successfully treated subjects displaying stereotypy and SIB with DRO procedures
by first conducting a stimulus preference assessment and then using the most preferred
stimuli as reinforcers in a DRO contingency to reduce SIB. The use of DRO without
extinction as a treatment intervention exemplifies one case in which interactions between
concurrently available reinforcers may require careful consideration. In the current study,
Experiment One showed that continuous access to toys virtually eliminated hand mouthing
for two subjects and arm rubbing for another subject; the reinforcement obtained from the
toys was preferred over reinforcement obtained from hand mouthing or arm rubbing.
Thus, the results suggest that these items might be effective in a DRO contingency to
reduce hand mouthing/arm rubbing. Thus, the purpose of Experiment Two was to
examine the efficacy of DRO using the preferred toys as reinforcers.
A parametric analysis of varying DRO schedules was conducted for all three
subjects. Each DRO schedule had two manipulable components: the DRO interval length
and the duration of access to the reinforcer. These two parameters were manipulated in
various combinations in an attempt to find an effective DRO schedule for reducing hand
mouthing/arm rubbing. The conditions are described below.
Alone. This condition was identical to the previous alone condition in Experiment
One and served as an initial Baseline.
DRO. The therapist delivered the preferred toy item according a resetting DRO
schedule. If the subject did not engage in hand mouthing/arm rubbing during an interval,
the toy was delivered at the end of the interval for a prespecified amount of time. If the
subject engaged in hand mouthing/arm rubbing at any time during an interval, the DRO
timer was reset. The DRO interval length and toy access time for the DRO schedules varied
across conditions. The therapist kept track of session time, DRO interval time, and toy
access time with two stop watches. Session time was stopped during toy access time to
keep the total amount of time the subject was without the toy constant. This control
procedure was included to insure that changes in responding from baseline to treatment
were not merely a function of increased access to the toys. Therefore, although total
session time varied according to how many times a subject met the DRO reinforcement
criterion, the amount of time in session (without the toy) remained constant at 10 min.
When the subject did not engage in hand mouthing/arm rubbing for the predefined DRO
interval, the therapist stopped the session time watch, handed the subject the toy and started
the toy access stopwatch. At the end of the access interval, the toy was removed, and the
next DRO interval began. For each subject, toy access time was 15 s, 30 s, and 60 s,
according to the predetermined schedule. The initial DRO interval was slightly shorter than
the mean interresponse time obtained during the Alone condition, and subsequent DRO
interval lengths were arbitrarily varied. Observers recorded the occurrence of hand
mouthing/arm rubbing and toy play, and the number of times the toy was given to the
subject, as described previously.
Results and Discussion
Figure 2 shows the percentage of intervals containing hand mouthing or arm
rubbing across sessions and conditions for all three subjects. The top panel for Figure 2
shows the results for Caryl. During the alone condition, Caryl engaged in high and
variable amounts of hand mouthing. In the next five conditions, the DRO interval length
varied (20 s, 30 s, 40 s, 60 s, and 10 s), while toy access time remained constant at 15 s.
Figure 2. Percentage of intervals of hand mouthing/arm rubbing during
the alone baseline and across DRO conditions for Caryl, Merry,
and Randy. Numbers above each condition reflect DRO
interval length (top #) and toy access time (bottom #).
'-" Reinforcement Interval
1ls 45& 10s I1.L 3ps
15s 15s 15s 30s 30s
1 I I I t
No reductions in hand mouthing occurred during any of the conditions, although the
number of times that Caryl met the criterion for reinforcement varied according to the DRO
schedule. The mean numbers of times that Caryl met the criterion for reinforcement for
each condition were: 1.8 for DRO 60 s, 3.4 for DRO 40 s, 8.8 for DRO 30 s, 15.1 for
DRO 20 s, and 12 for DRO 10 s. During the next four DRO schedules (DRO 60 s, 45 s,
30 s, and 10 s), toy access time was increased to 30 s. Again, although there were no
reductions in the amount of hand mouthing across conditions, the number of times that
Caryl met the criterion for reinforcement varied according to DRO schedule: 0.8 for DRO
60 s, 3 for DRO 45 s, 2.9 for DRO 30 s, and 15.5 for DRO 10 s. The last attempt to
reduce hand mouthing was a DRO 5 s schedule, during which 60 s access to the toys was
provided. This condition appeared to increase, rather than decrease hand mouthing, and
mean number of times she met criterion for reinforcement was 18.3.
The middle panel of Figure 2 shows Merry's results. During the alone condition,
Merry engaged in high and variable amounts of hand mouthing. In the next four
conditions, the DRO interval length varied (45 s, 30 s, 20 s, and 10 s), while the toy access
time remained at 15 s. No reductions in hand mouthing occurred during any of the
conditions, although the number of times Merry met the criterion for reinforcement varied
according to DRO schedule. The mean number of times Merry met the criterion for
reinforcement were: 0.2 for DRO 45 s, 1.3 for DRO 30 s, 4.9 for DRO 20 s, and 11.6 for
DRO 10 s. During the next four conditions (DRO 60 s, 45 s, 30 s, and 10 s), the toy
access time was increased to 30 s. Again, there was no reduction in hand mouthing in any
of these conditions, and number of times Merry met criterion for reinforcement varied
according to DRO schedule: 0.4 for DRO 60 s, 3.1 for DRO 45 s, 2.2 for DRO 30 s, and
8.5 for DRO 10 s. The last attempt to reduce hand mouthing was a DRO 5 s schedule,
during which 60 s access to the toys was provided. This condition appeared to increase,
rather than decrease hand mouthing, and mean number of times she met criterion for
reinforcement was 21.8.
The bottom panel of Figure 2 shows Randy's results. During the alone condition,
Randy engaged in high levels of arm rubbing. In the next three conditions, the DRO
interval length varied(10 s, 30 s, and 45 s), while vibrator access time remained at 15 s.
There were no reductions in arm rubbing during any of these conditions, although the
number of times Randy met criterion for reinforcement varied according to the DRO
schedule. The mean numbers of times Randy met criterion for reinforcement were: 15 for
DRO 10 s, 2.6 for DRO 30 s, and 0 for DRO 45 s. During the next three conditions (DRO
10 s, 30 s, and 45 s), vibrator access time was increased to 30 s. Again, there was no
reduction in arm rubbing in any of these conditions, and number of times met criterion for
reinforcement again varied according to DRO schedule: 21.8 for DRO 10 s, 1.8 for DRO
30 s, and 1 for DRO 45 s. The last attempt to reduce arm rubbing was a DRO 5 s
schedule, during which 60 s access to the vibrator was provided. This condition appeared
to increase, rather than decrease arm rubbing, and mean number of times he met the DRO
Figure 3 shows the mean percentage of intervals containing hand mouthing or arm
rubbing across conditions for all three subjects. The top graph shows the results for Caryl,
the middle graph shows results for Merry, and the bottom graph shows results for Randy.
For all three subjects, there were no clinically significant reductions in hand mouthing or
arm rubbing from the alone condition during any of the DRO schedule manipulations.
There were also no consistent parametric differences across schedules, except that all three
subjects engaged in the most hand mouthing/arm rubbing during the DRO 5 s/60 s
schedule. These results seem unusual because the most hand mouthing/arm rubbing for all
three subjects occurred in a condition most closely resembling continuous access to the toy
(DRO 5 s, with 60 s access to the toys). One plausible explanation for the increase is the
Figure 3. Mean percentage of intervals of hand mouthing/arm rubbing
during alone baseline and across DRO conditions for Caryl,
Merry, and Randy. Numbers below each DRO condition
reflect DRO interval length (top #) and toy access time
30s 40s 60s
fl lb"'~ ~~L I C1 IIIII II ~ ~~ Lb~~I SCICI I I III .11 ~~~ L IICI ~ IJ~I~ ~ Li
10s 45s 10s
15s T5s 30s
45s' 60s' 5s
3s 3s 0 DY
I Yiv W JSl r Iff l
45s 10s 30s 45s 5s
15s 30s 30s 30s 60s
DRO Interval/Reinforcement Interval
n Ierners r1rrr neUra r5Nerr rrrrrr a verrrrnf r rn, vA vrr
U .- .. ...
'" "' ....... .. n ff fmmffJ "" I
nature of the measurement method combined with the very short DRO interval length.
Because the DRO interval length (5 s) was shorter than the observation interval length (10
s), the subject could emit a response at the beginning of an observation interval (resulting in
a scored interval), but still meet the DRO criterion within that same scored interval. In
addition, short interresponse times were being reinforced (5 s) with very long
reinforcement times (60 s).
Even though a clear preference for toys was shown for all three subjects when toys
were continuously available during Experiment One, none of the DRO schedules showed
the toys as an effective reinforcer to reduce hand mouthing/arm rubbing in Experiment Two
in spite of numerous manipulations in the DRO schedule. Given the results of basic
research showing changing preference under differing contexts, it seems reasonable to
assume that preference for toys vs. hand mouthing/arm rubbing might also prove to be
context-dependent. Basic researchers have shown that changes in reinforcer dimensions
such as rate, magnitude, schedule, or delay to reinforcement, can effectively eliminate
preferences (Kagel et al., 1975; Rachlin et al., 1976). In behavioral economic terms, these
parameters represent "price requirements." Similarly, DRO schedules might be viewed as
another sort of "price requirement." The price requirement in DRO, however, involves not
responding to obtain the alternative reinforcer. Under these conditions, the subject's
responding was not affected by any of the DRO manipulations. Therefore, perhaps a better
way to examine changes in preference would be to alter some other parameter of "price"
that involves responding to obtain the alternative reinforcer. Experiment Three, therefore,
further examined parameters affecting the preference for toys over hand mouthing or arm
Most behavioral economic research examining the effects of reinforcement on
behavior has defined price as the number of responses per reinforcer under fixed-ratio (FR)
schedules (e.g., Hursh, Raslear, Shurtleff, Bauman, & Simmons, 1988). Some
investigators, however, have suggested that a more fundamental definition of price would
also include the amount of work expended per reinforcer (Hursh, 1980). Thus, unit price
would be defined as a cost-benefit ratio that would specify the amount of work expended
per unit of reinforcer. Hursh et al. (1988) tested this concept of unit price with rats
pressing levers for food pellets in a closed economy (the entire daily ration of food was
earned during the experimental procedure). Four dimensions of price were varied: (a) the
response requirement for food delivery (FR schedule); (b) the number of pellets delivered
per FR completion; (c) the effort required to make a response (lever force); and (d) the
probability that completion of a response requirement would result in food delivery. The
results confirmed that effort expended per unit of food value earned was the underlying
dimension of price determining consumption of food in a closed economy.
In a more recent study examining the reinforcer and response dimensions that
influence choices made by students with emotional disorders, Neef, Shade, and Miller
(1994) examined how reinforcer rate, quality, delay, and response effort combined to affect
time allocation across alternative math tasks on concurrent variable-interval schedules of
reinforcement. Neef et al. assessed the effects of these reinforcer and response dimensions
by counterbalancing the competing dimensions across six conditions, permitting
examination of the effects of each dimension on time allocation. Results showed that time
allocated to the different problem sets was differentially affected by the reinforcer and/or
response dimensions. These studies suggest that one other parameter affecting reinforcer
efficacy is response effort.
The purpose of Experiment 3 was to examine the effects of systematically altering
response effort in a parametric analysis of choice between the toy items and hand
mouthing/arm rubbing. Effort to obtain one reinforcer (hand mouthing or arm rubbing)
would be difficult to manipulate and therefore remained constant, but effort to obtain the
other reinforcer (toys) could be (and was) manipulated. This was accomplished by
attaching the toys with string to a table in front of of each subject; response effort was then
manipulated by varying the distance between the subjects and the toys.
A reversal design with replications across subjects was used in the parametric
demonstration. Following an alone baseline, response effort was systematically altered
beginning with the easiest (least effortful) condition. Once preference switched from toy
play to hand mouthing or arm rubbing, conditions on each side of this "switch point" were
replicated. The conditions are described below.
Alone. This condition was identical to the previously described alone sessions in
Experiments One and Two.
String length. All three subjects participating in this experiment (Caryl, Merry, and
Randy) engaged in toy play while sitting in an upright position. Therefore, the "least
effortful" condition was designed to require no change in the subjects' body position while
interacting with the toys. Prior to beginning the study, the experimenter measured the
distance between the edge of the table/lap tray, where the string would be attached to the
table surface, and the subject's mouth while the subject was seated in an upright position.
This was considered the least effortful position and was arbitrarily designated as 1.0
(proportion of string length to that obtained for the upright measurement). This distance
was 20 in, 17.5 in, and 67 in, for Caryl, Merry, and Randy, respectively. Thus, the most
effortful condition for all three subjects was one in which the toy was tied at its anchoring
point on the table/lap tray (0 in for all subjects, corresponding to 0 proportion of upright
position). The proportion of the upright string length was then varied for each subject in an
attempt to determine a switch point (e.g., the distance at which preference for the toy was
eliminated), and conditions on each side of the switch were replicated. For example, the
order of conditions for Caryl was 20 in (1.0), 0 in (0), 10 in (0.5), 15 in (0.75), 12.5 in
(0.62, switch point), 20 in (1.0), 10 in (0.5), and 15 in (0.75).
Results and Discussion
Figure 4 shows the percentage of intervals containing hand mouthing or arm
rubbing and toy play across conditions for all three subjects. The top panel shows the
results for Caryl. During the alone condition, Caryl engaged in high levels of hand
mouthing. The next condition was the "least effortful" condition (20 in or 1.0). During
this condition, Caryl's hand mouthing was virtually eliminated, and she engaged in high
levels of toy play. The next condition was most effortful (0 in or 0). During this
condition, hand mouthing increased (M=56.9%) and toy play was completely absent for all
seven sessions. When the 20 in (1.0) condition was reinstated, hand mouthing was
initially higher than in the previous 20 in (1.0) condition, but decreased across sessions,
while toy play again increased. In the next condition, 10 in (0.5), string length was
reduced to one-half its original length. During the first two sessions of this condition, toy
play was higher than hand mouthing, but then switched, with a steady increase in hand
mouthing and a steady decrease in toy play. The next condition was 15 in (0.75 of string
length). This condition was selected because it was halfway between 20 in (1.0) condition
in which toy play was preferred and 10 in (0.5) condition in which hand mouthing was
preferred. During this condition, hand mouthing decreased and toy play increased.
Because the 15 in (0.75) condition showed a preference for toy play, and the 10 in (0.5)
condition showed a preference for hand mouthing, the next condition was midway between
these two conditions at 12.5 in (0.625 of string length). During this condition, most
Figure 4. Percentage of intervals of hand mouthing/arm rubbing and toy
playing during the alone baseline and across string length
conditions for Caryl, Merry, and Randy. Numbers above each
condition reflect length of the string attached to toy (top #) and
proportion of string length while the subject was seated in an
upright position (bottom #).
-- Hand Mouth/Arm Rub
-- Toy Play
Alone 1.0 0
67 16.75 33.5 50.25
Alonel.0 .25 .5 .75
58.6 50.25 54.4
.875 .75 .825 String Length (inches)RANDY
JfProportion of Upright
| String Length
20 40 60 80
sessions showed hand mouthing lower than toy play, but both responses showed
considerable variability and some crossover. A return to the 20 in (1.0) condition showed
decreased hand mouthing, and high and variable amounts of toy play. A reintroduction of
the 10 in (0.5) condition showed toy play initially higher (as in the first 10 in condition),
but hand mouthing increased to high levels and toy play decreased to low levels. The final
condition replicated the 15 in (0.75) condition and showed hand mouthing low, and toy
The middle panel of Figure 4 shows the results for Merry. During the alone
condition, Merry engaged in variable amounts of hand mouthing. During the next
condition, 17.5 in (1.0), Merry's hand mouthing decreased, and toy play was high. In the
0 in (0) condition that followed, Merry's hand mouthing increased and toy play was not
observed in any of the sessions. When the 17.5 in (1.0) condition was reinstated, hand
mouthing decreased, while toy play increased to high levels for all but two sessions. In the
next condition, 8.75 in (0.5), string length was half of its original length. During all
sessions in this condition, hand mouthing was higher than toy play. The next condition,
13 in (0.75), was halfway between the 17.5 in (1.0) and 8.75 in (0.5) conditions. During
this condition, hand mouthing remained high, and toy play remained low. Because the 8.75
in (0.5) and 13 in (0.75) conditions still showed hand mouthing higher than toy play, the
next condition, 15.75 in (0.9), was selected midway between the 17.5 in (1.0) and 13 in
(0.75) conditions. During this condition, seven out of nine sessions showed hand
mouthing completely suppressed, and toy play was high. Because 15.75 in (0.9) condition
showed preference for the toy, the next condition, 14.5 in (0.825), was selected midway
between the 13 in (0.75) and 15.75 in (0.9) conditions. During this condition, a complete
suppression of hand mouthing occurred in all six sessions and toy play was high.
Therefore, the next condition was 13.75 in (0.785), midway between the 13 in (0.75) and
14.5 in (0.825). During this condition, hand mouthing was suppressed in six out of nine
sessions, but higher in the remaining three sessions. Toy play showed an inverse pattern,
with six of the nine sessions showing high levels of toy play and the remaining three
sessions with low levels of toy play. A return to the 15.75 in (0.9) condition showed hand
mouthing at low levels, except in one session, and toy play high, except for that same
session. A reintroduction of the 13 in (0.75) condition showed hand mouthing again
higher in all but one session, and toy play low in all but that same one session. The final
condition replicated the 15.75 in (0.9) condition with hand mouthing low, and toy playing
The bottom panel of Figure 4 shows the results for Randy. During the initial alone
condition, Randy engaged in high levels of arm rubbing. The 67 in (1.0) condition that
followed showed an almost complete suppression of arm rubbing, and high levels of toy
play. The next condition, 16.75 in (0.25), showed arm rubbing at high levels, and toy
play at lower levels. Because preference in the 16.75 in (0.25) condition was for arm
rubbing, the next condition was 33.5 in (0.5). During this condition, switching of
preferences occurred across sessions, with arm rubbing variable but higher in most
sessions than toy playing. The next condition was 50.25 in (0.75). Although the first few
sessions showed no clear preference for either response, subsequent sessions showed an
increase in arm rubbing, and a decrease in toy play. The next condition, 58.6 in (0.875),
showed a clear preference for toy play over arm rubbing. The 50.25 in (0.75) condition
was then replicated, and preference again switched (arm rubbing increased and toy play
decreased). The final condition, 54.4 in (0.825), showed decreased arm rubbing after the
first four sessions, and increased toy play.
Figure 5 shows the mean percentage of intervals containing hand mouthing/arm
rubbing and toy play for all subjects during each string length manipulation. The top panel
shows the results for Caryl. At 20 in (1.0), a clear preference for toy play over hand
mouthing is shown. As string length was shortened, however, mean percentages of hand
Figure 5. Mean percentage of intervals of hand mouthing/arm rubbing
and toy playing during each string length condition for Caryl,
Merry, and Randy. Numbers at the bottom of each graph reflect
length of the string attached to toy (top #) and proportion of string
length while the subject was seated in an upright position
Q Hand Mouthing/Arm Rubbing
M Toy Play
15in 12.5in 10in
(.75) (.625) (.5)
67in 58.6in 54.4in 50.25in 33.5in 16.75in
(1.0) (.875) (.825) (.75) (.5) (.25)
STRING LENGTH (PROPORTION OF UPRIGHT POSITION)
mouthing increased and mean percentages of intervals toy play steadily decreased until
hand mouthing became the preferred response at 10 in (0.5 of the string length). The data
showed a clear and orderly inverse relationship between hand mouthing and toy play. The
middle panel shows results for Merry. At string lengths of 17.5 in (1.0), 15.75 in (0.9),
and 14.5 in (0.825), high mean percentage of intervals for toy play, and very low mean
percentages for hand mouthing were shown. At 13.75 in (0.785 of string length), toy play
decreased noticeably, and hand mouthing increased somewhat. At 13.5 in (0.75),
preference for toy play was eliminated. Randy's results are shown in the bottom panel of
Figure 5. His data show a decreasing trend for toy play and an increasing trend for arm
rubbing as string length decreased, with preference for toy play eliminated at 50.25 in
(0.75 of the string length).
These results show that altering response effort, as measured by distance of an
object to an individual in an upright position, reduced preference for toys over hand
mouthing or arm rubbing, and further demonstrate the contextual nature of reinforcer
efficacy. Although clear preference for toys was observed in Experiment One, those
preferences were reduced in Experiment Three merely by increasing the effort to obtain the
preferred reinforcer, while no contingency was placed on hand mouthing. Experiment
Two showed that no DRO schedule was effective in reducing hand mouthing/arm rubbing,
which also suggests that preference was altered. Thus, these three experiments provide
evidence that stimulus preference assessments may need to take variables such as
reinforcement schedule and response effort into account when attempting to predict
stimulus reinforcement efficacy.
Substitutable reinforcers represent one end of the continuum of possible interactions
among reinforcers. Substitutable reinforcers serve similar functions, with one reinforcer
replacing another. At the other end of the continuum are complementary reinforcers.
These stimuli are not functionally similar and are considered complements because, by
definition, they are consumed jointly. Thus, an increase in consumption of one reinforcer
would result in an increase in consumption of its complement. The implication for applied
behavior analysts is that complementary reinforcer interactions may occur between
concurrently available reinforcers that affect treatment efficacy. Study Two examined
complementary relationships between reinforcers and the effect of these interactions on the
assessment and treatment of hand mouthing.
Many applied studies have examined the effectiveness of providing competing or
substitutable reinforcers as interventions to reduce inappropriate behavior. Much less
applied research, however, has assessed interactions among complementary reinforcers.
Green and Freed (1994) discussed the implications of reinforcer substitutability for
treatment of inappropriate behavior in applied settings. According to the matching law,
there are two obvious reinforcement-based strategies for eliminating inappropriate
behavior. First, one could increase the rate of noncontingent reinforcement (NCR). In
Study One, for example, noncontingent access to toys almost completely eliminated the
stereotypic self-injury for all three subjects. A second strategy would be to increase the rate
of reinforcement for a concurrently available response option, as exemplified in studies on
differential reinforcement for alternative behavior (DRA) contingencies (Favell et al., 1982;
Lockwood & Bourland 1982; Mulick et al.,1978). Results of these DRA studies show that
when alternative stimuli do not entirely substitute for the aberrant behavior, additional
reinforcement for the alternative may be an effective treatment. These two treatment
strategies, however, may be effective only to the degree that the reinforcers are
substitutable. When reinforcers are complementary, reinforcement for an alternative
response may result in an increase in the behavior targeted for reduction. Therefore, an
understanding of complementary relationships and the parameters affecting those
interactions may be necessary when selecting potential reinforcers for use in behavior
acquisition or reduction procedures.
Few studies have examined complementary interactions between reinforcing
environmental stimuli and automatically-reinforced behavior. One possible example is
found in a study examining the effects of altering the physical environment on self-
stimulatory behavior (Duker & Rasing, 1989). The study showed that redesigning the
classroom environment by covering the walls, shelves, and replacing colorful curtains with
unicolored ones resulted in three developmentally disabled males displaying much less self-
stimulatory behavior and more on-task behavior during training sessions held in the room.
Thus, the visual stimulation available in the less barren classroom could have been
complementary to self-stimulation, with increasing amount of visual stimuli in the
environment resulting in increased self-stimulation. By decreasing the visual stimulation,
the complementary self-stimulation was also decreased. Thus, some self-stimulatory
behavior may be evoked or potentiated by various environmental stimuli; if so, these
stimuli may have adverse effects on behavior when used as reinforcers.
Food is a commonly used reinforcer in the treatment of aberrant behavior and in the
acquisition of adaptive behavior. For example, in a 20-year review of reinforcement-based
behavior reduction procedures, O'Brien and Repp (1990) reported that 60% of the DRO
studies and 20% of the DRI studies used food as the programmed consequence. In the
current study, a complementary relationship between hand mouthing and eating existed.
That is, for two of the subjects previously described, Chuck and Matt, food was found
during probe sessions to increase the occurrence of hand mouthing. For Chuck, Kit Kat
candy bars had been found to be highly preferred in a stimulus preference assessment and
to evoke the most hand mouthing during probe sessions. For Matt, pudding evoked the
most hand mouthing during probe sessions and was therefore selected as the reinforcer.
The purpose of Study Two was to explore the relationship between consuming these food
items and hand mouthing. Three experiments were conducted: (a) a basic demonstration of
the relationship between food and hand mouthing, (b) a parametric analyses of the effects
of amount of food on hand mouthing, and (c) an examination of the effectiveness of food
as a reinforcer in acquisition and DRO procedures.
In Experiment One, the relationship between eating and hand mouthing, when both
reinforcers were concurrently available, was examined. The conditions were presented to
each subject in a reversal design (ABAB or BABA) and are described below.
Alone. During the alone sessions, neither food nor social interaction was available.
This condition was identical to the alone sessions for Study One.
Food available. During this condition, food was presented to the subjects in the
following manner. For Chuck, Kit Kat candy bars were evenly divided into four parts,
with each piece counted as one cookie. At the beginning of each session, Chuck was
seated at a table, and a plate with 6 pieces of the Kit Kat was placed in front of him. After
placing the cookies in front of Chuck, the therapist left the room and session time began.
For Matt, the therapist measured 100 milliliters of pudding into a cup prior to each session.
Because Matt could not feed himself, Matt's sessions began with the therapist feeding him
at a slow, steady pace, and session time began when feeding started. When the pudding
was gone, Matt's therapist left the room. Observers recorded hand mouthing and eating for
Results and Discussion
The results for Experiment One are shown in Figure 6. The top panel of Figure 6
shows the results for Chuck. In the first condition, when cookies were available, Chuck
engaged in high levels of hand mouthing, and moderate amounts of eating. In the
subsequent alone condition, Chuck did not engage in hand mouthing for any of the five
sessions. When cookies were reintroduced, Chuck again engaged in high levels of hand
Figure 6. Percentage of intervals of hand mouthing and eating during
food and alone conditions for Chuck and Matt.
i ^ ^ -_ p ^
m m -In --- I
mouthing, and moderate amounts of eating. The alone condition was then reinstated, and
no hand mouthing was observed in any of the five sessions.
The bottom panel of Figure 6 shows the results for Matt. In the initial alone
condition, Matt engaged in low to moderate amounts of hand mouthing. When pudding
was available in the next condition, however, hand mouthing increased (the amount of
eating was low and stable). The alone condition was then reinstated, with hand mouthing
relatively low. When the pudding condition was reinstated, hand mouthing again increased
(time spent eating was again low and stable).
These results showed that, for both subjects, hand mouthing increased when food
was consumed. When food was not available, Chuck did not engage in hand mouthing at
all, and Matt's hand mouthing was reduced. Thus, consumption of food increased the
reinforcing efficacy of hand mouthing and may have served as an establishing operation
(Michael, 1982) for hand mouthing. Basic experimental studies examining complementary
relationships have shown that varying the amount of one reinforcer can affect the rate of
consumption of an alternative reinforcer. For example, Allison and Mack (1978) showed
(with rats lever pressing for food pellets) that eating and drinking were substitutable when
eating was suppressed, but complementary when drinking was suppressed. Thus, amount
of food consumed may be an important variable to consider when using food as a
reinforcer. Therefore, the purpose of Experiment Two was to determine if changes in the
amount of food available would alter the time spent hand mouthing.
A parametric analysis of varying amounts of food was conducted for both subjects.
For Chuck, several reversals occurred in the order of 12-6-12-3-12-1 cookies. For Matt,
milliliters of pudding were varied in the order of 100-50-200-25-12.5-6.25-3. Sessions
were conducted just as they had been in Experiment One, and observers scored the
occurrence of eating and hand mouthing as previously described.
Results and Discussion
The results of Experiment Two are shown in Figures 7 and 8. Figure 7 shows the
percentage of intervals containing hand mouthing and eating across sessions for both
subjects. The top panel of Figure 7 shows the results for Chuck. During the initial 12
cookies condition, Chuck engaged variable amounts of hand mouthing (M=25.8%) and
eating (M= 12.5%). Hand mouthing continued to be variable across the remaining
conditions, with mean percentages of intervals at 21.3% (6 cookies), 46.5% (12 cookies),
29.8% (3 cookies), 50.9% (12 cookies), and 30.1% (1 cookie). Although the amount of
hand mouthing did not change systematically with the number of cookies, predictably the
amount of eating did, with mean percentages of intervals eating increasing as the number of
The bottom panel of Figure 7 shows the results for Matt. During the initial pudding
condition (100 ml) Matt showed variable amounts of hand mouthing (M= 42.2%) and
stable amounts of eating (M=20%). Similar amounts of hand mouthing were shown across
the remaining conditions with mean percentages of intervals of hand mouthing at 37.6%
(50 ml), 30.2% (200 ml), 48.5% (25 ml), 28.5% (12.5 ml), 50.8% (6.25 ml), and 45.8%
(3 ml). Time spent eating predictably increased as amount of pudding increased.
Figure 8 shows the mean percentages of intervals spent hand mouthing or eating
across conditions. The top panel shows results for Chuck, and the bottom panel shows
results for Matt. Both subjects showed increasing mean percentages of intervals eating as
the amount of food increased, but neither showed any systematic change in the mean
percentages of intervals hand mouthing as a function of the amount of food consumed.
Figure 7. Percentage of intervals of hand mouthing and eating during
each of the food conditions for Chuck and Matt. Numbers
above each graph reflect the number of cookie pieces (Chuck)
and the milliliters of pudding (Matt).
# of Cookies
12 6 12 3 12 1
MI of Pudding MATT
S100 50 200 25 12.5 6.25 3
a 1 1 1 T
Figure 8. Mean percentage of intervals of hand mouthing and eating
across food conditions for Chuck and Matt. Numbers below
each graph reflect the number of cookie pieces (Chuck) and
milliliters of pudding (Matt).
[ Hand mouthing
MI of Pudding
These results suggest that the preferred food items established hand mouthing as
reinforcing regardless of the amount of food consumed.
Food is frequently used as a reinforcer in both behavioral acquisition programs and
in procedures used to reduce inappropriate behavior. In fact, for these two subjects,
stimulus preference assessments showed food to be highly preferred stimuli. The
demonstration in Experiment One illustrated, however, that for these two subjects,
consumption of preferred food increased the reinforcing efficacy of inappropriate behavior
(hand mouthing). The second experiment suggested that food would result in increased
hand mouthing across a wide range of food amounts. For Chuck, hand mouthing did not
appear to be a problem until food was introduced. Thus, in the absence of food, there was
no hand mouthing in need of treatment. Because, however, food seemed to be a highly
preferred stimulus for Chuck, it was used as a reinforcer in some of his training programs
and may have had the unintended effect of disrupting his performance. Therefore, an
examination of the effects of food as reinforcement in an acquisition-type context would be
of clinical interest for Chuck. Matt, however, engaged in hand mouthing in the absence of
consuming food. Thus, examining the use of food in a DRO contingency to reduce hand
mouthing would be of interest for Matt. The purpose of Experiment Three, therefore, was
to examine the efficacy of the edible reinforcers in acquisition training during a free operant
task (Chuck) and during a DRO contingency to reduce hand mouthing (Matt).
Acquisition procedures (Chuck)
All sessions were 10 min in length, and conditions were alternated in a reversal
design (ABA). The response of interest was placing blocks in a bucket. Chuck was seated
in a chair facing a table, on which were located a bucket and a number of small blocks.
The experimenter began the session by modeling the response (i.e., placing one block into
the bucket). This modeling was repeated on an FT (fixed t ime) 1 min schedule throughout
the session for a total of 10 modeling prompts. No further instructions or prompts were
provided. Observers scored the number of blocks placed in the bucket by Chuck, number
of modeled responses by the experimenter, and the occurrence of hand mouthing by
Baseline. The experimenter modeled the response on the FT 1 min schedule and
did not provide any further instruction, prompts, or reinforcement.
Reinforcement. This condition was identical to baseline, except that the
experimenter delivered one piece of Kit Kat cookie contingent on each occurrence of
placing a block in the bucket.
DRO procedure (Matt)
Alone. Baseline consisted of the alone condition previously described. Observers
recorded the occurrence of hand mouthing, and the experimenter measured the
interreponse times (IRTs) with a stopwatch. Mean IRT across the baseline sessions was
then determined by adding all durations of IRTs and dividing by the number of IRTs.
DQR. The DRO interval was set slightly greater than the mean IRT for hand
mouthing as determined from baseline observations. The experimenter fed one level spoon
of pudding to Matt contingent on the absence of hand mouthing according to a resetting
DRO 30 s schedule. If Matt did not engage in hand mouthing during a 30 s interval, the
food was delivered at the end of the interval. If the Matt engaged in hand mouthing at any
time during an interval, the DRO timer was reset. Observers recorded the occurrence of
hand mouthing, eating, and the number of times the pudding was delivered.
Results and Discussion
Figure 9 shows the results for Chuck. During Baseline, he displayed one instance
of hand mouthing during one session; no hand mouthing occurred in any of the remaining
nine sessions observed. He did, however, put blocks in the bucket, although the number
of blocks he put in the bucket was variable across sessions (M= 24.5). When cookies
were delivered contingent on putting blocks in the bucket, however, the response decreased
(M=10), and the amount of hand mouthing increased (M=40%). A return to baseline
showed a variable increase in the number of blocks in the bucket (M=40), and no hand
mouthing during any of the eight sessions.
Figure 10 shows the results for Matt. In the alone condition, Matt showed variable
amounts of hand mouthing (M= 30%). When pudding was delivered contingent on the
absence of hand mouthing for 30 s, a slight increase was observed in hand mouthing (M=
40.9%), and time spent eating was indicative of the number of times he met the
contingency (M= 33 reinforcers per session).
These results showed that both the acquisition procedure for Chuck and the DRO
procedure for Matt were ineffective when a highly preferred food was used as a reinforcer.
For Chuck, the Kit Kat candy presented contingent on putting blocks in the bucket reduced
rather than increased responding. Eating the cookies increased hand mouthing; hand
mouthing then interrupted responding. For Matt, the pudding increased the efficacy of
hand mouthing and therefore did not serve as a reinforcer for the absence of hand
mouthing. These results suggest that therapists should examine the interaction between
stimuli selected for use as reinforcers in therapeutic programming and the effect those
stimuli may have on undesirable behavior.
Figure 9. Percentage of intervals of hand mouthing and number of
blocks placed in bucket for Chuck across baseline and
--- Hand mouthing
-o- Blocks in Bucket
Baseline Reinforcement Baseline
0 10 20 30
Figure 10. Percentage of intervals of hand mouthing and eating during the
alone and DRO conditions for Matt.
--S- Hand mouthing
DRO 30" (Pudding)
0 5 10 15 20 25
Results of the two studies illustrate that relationships among concurrently available
reinforcers can affect behavior in numerous and complex ways. Specifically, research
examining reinforcer substitutability and complementarity may provide a guide for applied
behavior analysts who seek to identify potential reinforcers for both acquisition training and
behavioral reduction. Understanding and identifying the parameters that alter these
reinforcer relationships may be the next step in developing a technology for reinforcer
Study One examined substitutability between two automatically-reinforced
behaviors-- toy play and either hand mouthing or arm rubbing. For all three subjects, time
spent hand mouthing or arm rubbing and time spent toy playing were inversely related
when both responses were freely available (Experiment One). These preferences for the
toys, however, were shown to be context-dependent, in that preference was readily
eliminated when response effort was increased (Experiment Three). In addition, even
though toys were highly preferred in the free access condition (Experiment One), the toys
were not effective reinforcers in DRO contingencies to reduce hand mouthing or arm
rubbing (Experiment Two). These findings are consistent with results of basic research on
variables affecting choice (Green & Freed, 1994).
One explanation for the results of Study One involves reinforcer access time. Hand
mouthing/arm rubbing was continuously available in all conditions of all three experiments.
Therefore, presentation of an alternative stimulus would have to compete with continuous
access to hand mouthing. Although toys were preferred when both stimuli were
continuously available, the DRO contingencies and string length contingencies showed that
when toy play was less than continuously available, it no longer competed with hand
mouthing or arm rubbing.
Another possible related explanation for the results involves temporal separation
between the two reinforcers. Temporal separation has been shown to reduce the degree of
substitutability between otherwise identical reinforcers. Hursh and Bauman (1987), for
example, compared consumption of one reinforcer as a function of its relative price when
an alternative identical reinforcer was present in three conditions of temporal separation:
concurrent schedules, multiple schedules, and across conditions of the same experiment.
(Green and Freed (1994) suggested this was similar to comparing prices of items on the
same shelf, in different stores, or over months of shopping, respectively.) Results showed
that the greater the temporal separation between identical reinforcers, the less substitutable
they became. The DRO conditions in Experiment Two and the shorter string length
conditions of Experiment Three resulted in temporal separations between the self-
stimulation produced by toy play relative to hand mouthing (which was always immediately
available). In the DRO contingencies (Experiment Two), the subjects had to not engage in
hand mouthing, which presumably produced similar stimulation to that produced by toy
play. In the string length conditions (Experiment Three), decreased string length created a
temporal separation by forcing the subjects to engage in additional behavior to play with the
toys (e.g., bending over to place the toy against the face or in the mouth).
Results of Study Two also have important implications for the assessment and
treatment of automatically-reinforced aberrant behavior. The results of these experiments
suggest that reinforcer assessments that do not take contextual variables into account, such
as relationships between concurrently available reinforcers, may result in identification of
putative reinforcers that do not serve as reinforcers, and that may in fact be contraindicated.
Experiment One demonstrated that some stimuli that would be identified as preferred in a
standard stimulus preference assessment (e.g., Pace et al., 1990) may increase the
reinforcing efficacy of inappropriate behavior. When food was available, both subjects'
time spent hand mouthing increased over that observed when food was not consumed.
Altering the amount of food, however, did not have any effect on the amount of hand
mouthing. At least for these two subjects, any amount of food appeared to increase the
reinforcing efficacy of hand mouthing.
Given that food is often used as a reinforcer to teach new skills or in DRO/DRA
contingencies to decrease inappropriate behavior, the results of Experiment Three have
important implications for applied behavior analysts. When treating automatically-
reinforced behavior problems, the importance of examining the collateral effects of
reinforcer consumption on the occurrence of other behavior may be particularly relevant.
Data from Experiment Three (Figure 9) showed that food was ineffective as a reinforcer for
Chuck's block-in-bucket responding. In fact, responding was suppressed when food was
delivered contingent on performance. For Matt (Figure 10), the DRO contingency resulted
in a slight increase in hand mouthing. These results are not surprising when one considers
that, for both subjects, any amount of the preferred food item increased hand mouthing.
Hand mouthing, therefore, had a disruptive effect on acquisition. Further research should
be conducted to examine when, or if, a complementary reinforcer that establishes a problem
behavior as reinforcing, can serve as a reinforcer for other behavior.
There are several limitations to these experiments worth noting. First, sequence
effects may have occurred in the parametric studies. In the DRO experiment of Study One
(Figure 2), for example, all three subjects were exposed to first 15 s, and then 30 s
reinforcement intervals. Although the DRO intervals varied between subjects, it is not
known if the sequence of conditions affected the results. Similar problems are found in
Experiment Three of Study One (Figure 4). The sequence of conditions may have either
facilitated or hampered performance in adjacent conditions. There is, in fact, some
indication that preceding conditions affected subsequent conditions, in that initial
performance in some conditions showed a switch after the first couple of sessions. In
Experiment 2 of Study Two (Figure 7) for Chuck, there was an attempt to control for
sequence effects by returning to the 12 cookie condition every other condition. The order
of conditions was different for both subjects, and similar results were obtained, suggesting
that the sequence of conditions was not a confounding factor.
Another potential limitation is the experimental design used in Experiment Three of
Study Two (Figure 10). For Matt, an AB demonstration was conducted and further
manipulations were not pursued because no reinforcement effect was observed.
Experiment Three of Study Two consists of two case studies; therefore, the results should
be considered preliminary. Further research is needed to examine these variables under
more rigorous experimental conditions.
Stimulus preference assessments have been receiving considerable attention in
recent years (e.g., Fisher, Piazza, Bowman, Hagopian, & Langdon, 1994; Pace, et al.,
1985) and have facilitated acquisition of adaptive behavior (e.g., Pace et al., 1985) as well
as reduction in aberrant behavior (Steege et al., 1989). The results of this study, however,
suggest that preference in one context may not be predictive of preference in another
context. This may be particularly true when automatically produced reinforcement is
involved. Further research is needed to develop a technology for reinforcer assessment that
takes contextual variables into account.
The results of these experiments on substitutable and complementary reinforcers
shed new light on previous research examining the efficacy of various interventions to
reduce stereotyped aberrant behavior (i.e., behavior not maintained by social
reinforcement). Directions for future research using concepts borrowed from behavioral
economic theory are numerous. For example, one common method of treatment for
automatically-reinforced stereotypic behavior has been to manipulate stimuli in the
environment that appear to be correlated with a decrease in the occurrence of the aberrant
behavior. For example, stereotypic behavior has been shown in some cases to occur more
frequently in barren environments (Berkson & Mason, 1965; Homer, 1980; Warren &
Burs, 1970), and in other cases to occur more often when external stimulation is high
(Adams, Tallon, & Stangle, 1980; Duker & Rasing, 1989). These studies suggest that
stereotypic behavior may be established as reinforcing under conditions of either low or
high sensory stimulation. For example, Duker and Rasing (1989) covered the walls and
windows of a classroom to reduce the visual stimulation that appeared to be complementary
to their subjects' stereotypic behavior. The current studies suggest that researchers and
clinicians should look to relationships between reinforcers as possible establishing
operations for other automatically-produced reinforcers.
When stereotypic behavior occurs most frequently in relatively barren
environments, the appropriate intervention would consist of enriching the environment
This "enriched environment" approach to treatment, however, can be interpreted in terms of
substitutability of reinforcers: Provide alternative sources of stimulation that substitute for
the inappropriate self-stimulation. For example, Berkson and Mason (1965) found that
subjects engaged in more stereotyped movements when alone than in a condition in which
the experimenter handed the subjects toys and provided attention. This approach to
treatment has the advantage of being relatively simple for caregivers to implement (Boe,
1977; Vollmer, 1994). When, however, alternative stimuli do not entirely substitute for the
inappropriate behavior, combining access to the stimulus with other preferred stimuli may
be effective. For example, Lockwood and Bourland (1982) used praise and brief physical
contact contingent on sustained toy usage to further reduce SIB below a condition of toys
only. The use of multiple reinforcers that substitute for aberrant behavior when one
reinforcer is insufficient provides a good example of how the concept of substitutability
may facilitate the development of more effective reinforcement-based interventions for
automatically reinforced aberrant behavior.
Enriched environments, however, are not always effective in reducing aberrant
behavior. For example, Adams, et al. (1980), found lower levels of stereotypic behavior
under quiet and "easy listening" music conditions than in a television-on condition. They
also found that the addition of toys to the environment did not result in decreased self-
stimulatory behavior. Thus, for some individuals, increased environmental stimulation
may establish stereotypic behavior as reinforcing, and treatment would entail less
stimulation. These studies also may be interpreted in terms of reinforcer substitutability.
For some individuals, other sources of environmental stimulation may be complementary
reinforcers to the automatically-reinforced behavior, with an increase in environmental
stimulation resulting in an increase in the stereotypic behavior.
Given that extinction of automatically-reinforced behavior is often difficult or
impossible to achieve, reinforcement-based treatments almost always involve concurrent
schedules of reinforcement. Therefore, applied researchers should examine the plethora of
basic research examining the parameters affecting preference between concurrently
available reinforcers and assess those relationships in the treatment of automatically-
reinforced problem behavior. In addition, identifying the parameters that affect preferences
and/or alter the reinforcing efficacy of aberrant behavior may be facilitated by consideration
of the relationships between concurrently available reinforcers. There is much to be learned
about the assessment and treatment of nonsocially-mediated behavior problems, and the
concept of reinforcer substitutability may facilitate that understanding.
Adams, G.L., Tallon, R.J., & Stangle, J.M. (1980). Environmental influences on
self-stimulatory behavior. American Journal of Mental Deficiency, 85, 171-175.
Allison, J., & Mack, R. (1982). Polydipsia and autoshaping: Drinking and
leverpressing as substitutes for eating. Animal Learning & Behavior, 1Q, 465-475.
Bailey, J., & Meyerson, L. (1970). Effect of vibratory stimulation on a retardate's
self-injurious behavior. Psychological aspects of disability, 1, 133-137.
Barrett, R.P., McGonigle, J.J., Ackles, P.K., & Brukhart, J.E. (1987). Behavioral
treatment of chronic aerophagia. American Journal of Mental Deficiency, 91, 620-625.
Baum, W.M.(1974). On two types of deviation from the matching law: Bias and
undermatching. Journal of the Experimental Analysis of Behavior, 22, 231-242.
Baum, W.M. (1979). Matching, undermatching, and overmatching in studies of
choice. Journal of the Experimental Analysis of Behavior, 32, 269-281.
Berkson, G., & Mason, W.A. (1965). Stereotyped movements of mental detectives:
4. The effects of toys and the character of the acts. American Journal of Mental Deficiency,
Boe, R.B. (1977). Economical procedures for the reduction of aggression in a
residential setting. Mental Retardation, 15, 25-28.
Carr, E.G., & Durrand, (1985). Reducing behavior problems through functional
communication training. Journal of Applied Behavior Analysis, 18, 111-126.
Cowdery, G.E., Iwata, B.A., & Pace, G.M. (1990). Effects and side effects of DRO
as treatment for self-injurious behavior. Journal of Applied Behavior Analysis, 23, 497-
Danford, D.E., & Huber, A.M. (1982). Pica among mentally retarded adults.
American Journal of Mental Deficiency, 87, 141-146.
Davenport, R.K., & Berkson, G. (1963). Stereotyped movements in mental
detectives: Effects of novel objects. American Journal of Mental Deficiency, 67, 879-882.
Day, R.M., Rea, J.A., Schussler, N.G., Larsen, S.E., & Johnson, W.L. (1988). A
functionally based approach to the treatment of self-injurious behavior. Behavior
Modification, 12, 565-589.
de Villiers, P. (1977). Choice in concurrent schedules and a quantitative formulation of
the law of effect. In W. K. Honig & J.E.R. Staddon (Eds.) Handbook of operant
behavior (pp. 233-287). Englewood Cliffs, NJ: Prentice-Hall.
Duker, P.C., & Rasing, E. (1989). Effects of redesigning the physical environment on
self-stimulation and on-task behavior in three autistic-type developmentally disabled
individuals. Journal of Autism and Developmental Disorders, 19, 449-461.
Favell, J.E., McGimsey, J.F., & Schell, R.M. (1982). Treatment of self-injury by
providing alternate sensory activities. Analysis and Intervention in Developmental
Disabilities, 2, 83-104.
Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., & Langdon, N. A.
(1994). Empirically derived consequences: A data-based method for prescribing treatments
for destructive behavior. Research in Developmental Disabilities, 15, 133-149.
Goh, H., Iwata, B.A., Shore, B.A., DeLeon, I.G., Lerman, D.C., Ulrich, S.M., &
Smith, R.G An analysis of the reinforcing properties of hand mouthing. Journal of
Applied Behavior Analysis, in press.
Green, L., & Freed, D.E. (1993). The substitutability of reinforcers. Journal of the
Experimental Analysis of Behavior, 60, 141-158.
Herrnstein, R.J. (1961). Relative and absolute strength of response as a function of
frequency of reinforcement. Journal of the Experimental Analysis of Behavior 4, 267-
Herrnstein, R.J. (1970). On the law of effect. Journal of the Experimental Analysis of
Behavior, 13, 243-266.
Homer, R.D. (1980). The effects of an environmental "enrichment" program on the
behavior of institutionalized profoundly retarded children. Journal of Applied Behavior
Analysis 13, 473-491.
Hursh, S.R. (1980). Economic concepts for the analysis of behavior. Journal of the
Experimental Analysis of Behavior 34, 219-238.
Hursh, S.R., & Bauman, R.A. (1987). The behavioral analysis of demand. In L.
Green & J.H. Kagel (Eds.), Advances in behavioral economics (Vol. 1, pp. 117-165).
Norwood, NJ: Ablex.
Hursh, S.R., Raslear, T.G., Shurtleff, D., Bauman, R., & Simmons, L. (1988). A
cost-benefit analysis of demand for food. Journal of the Experimental Analysis of
Behavior, 5, 419-440.
Iwata, B.A., Dorsey, M.F., Slifer, K.J., Bauman, K.E., & Richman, G.S. (1982).
Toward a functional analysis of self-injury. Analysis and Intervention in Developmental
Iwata, B.A., Pace, G.M., Dorsey, M. F., Zarcone, J.R., Vollmer, T.R., Smith,
R.G., Rodgers, T.A., Lerman, D.C., Shore, B.A., Mazaleski, J.L., Goh, H., Cowdery,
G.E., Kalsher, M.J., & Willis, K.D. (1994). The functions of self-injurious behavior:
An experimental-epidemiological analysis. Journal of Applied Behavior Analysis, 22,
Iwata, B.A., Pace, G.M., Kalsher, M.J., Cowdery, G.E., & Cataldo, M.F. (1990a).
Experimental analysis and extinction of self-injurious escape behavior. Journal of Applied
Behavior Analysis. 2, 11-27.
Iwata, B.A., Vollmer, T.R., & Zarcone, J.R. (1990b). The experimental (functional)
analysis of behavior disorders: Methodology, applications, and limitations. In A.C. Repp
& N.N. Singh (Eds.), Perspectives on the use of nonaversive and aversive interventions
for persons with developmental disabilities (pp. 301-330). Sycamore, IL: Sycamore
Kagel, J.H., Battalio, R.C., Rachlin, R., Green, L., Basmann, R.L., & Klemm,
W.R. (1975). Experimental studies of consumer demand behavior using laboratory
animals. Economic Inquiry, 13, 22-38.
Kaufman, M.E., & Levitt, H. (1965). A study of three stereotyped behaviors in
institutionalized mental detectives. American Journal of Mental Deficiency, 6, 467-473.
Knight, M.F., & McKenzie, H.S. (1974). Elimination of bedtime thumbsucking in
home settings through contingent reading. Journal of Applied Behavior Analysis, 2, 33-
Lockwood, K., & Bourland, G. (1982). Reduction of self-injurious behaviors by
reinforcement and toy use. Mental Retardation, 20, 169-173.
Luiselli, J.K. (1981). Behavioral treatment of self-stimulation: Review and
recommendations. Education and Treatment of Children, 4, 375-392.
Luiselli, J.K., & Krause, S. (1981). Reduction in stereotypic behavior through a
combination of DRO, cueing, and reinforcer isolation procedures. The Behavior Therapist.
Mace, F.C., & Lalli, J.S. (1991). Linking descriptive and experimental analyses in
the treatment of bizarre speech. Journal of Applied Behavior Analysis 24, 553-562.
Mace, F. C., Lalli, J. S., & Pinter Lalli, E. (1991). Functional analysis and treatment
of aberrant behavior. Research in Developmental Disabilities, 12, 155-180.
Mazaleski, J.L., Iwata, B.A., Rodgers, T.A., Vollmer, T.R., & Zarcone, J.R. (1994).
Protective equipment as treatment for stereotypic hand mouthing: Sensory extinction or
punishment effects? Journal of Applied Behavior Analysis, 22, 345-355.
Mazaleski, J.L., Iwata, B.A., Vollmer, T.R., Zarcone, J.R., & Smith, R.G. (1993).
Analysis of the reinforcement and extinction components of DRO contingencies with self-
injury. Journal of Applied Behavior Analysis, 26, 143-156.
Michael, J.L. (1982). Distinguishing between discriminative and motivational
functions of stimuli. Journal of the Experimental Analysis of Behavior, 37, 149-155.
Mulick, J.A., Hoyt, P., Rojahn, J., & Schroeder, S. (1978). Reduction of a
"nervous habit" in a profoundly retarded youth by increasing toy play. Journal of Behavior
Therapy and Experimental Psychiatry.., 381-385.
Neef, N.A., Shade, D., & Miller, M.S. (1994). Assessing influential dimensions of
reinforcers on choice in students with serious emotional disturbances. Journal of Applied
Behavior Analysis. 27, 575-584.
O'Brien, S., & Repp, A.C. (1990). Reinforcement-based reductive procedures: A
review of 20 years of their use with persons with severe or profound retardation. The
Journal of the Association for Persons with Severe Handicaps, 15, 148-159.
Pace, G.M., Ivancic, M.T., Edwards, G.L., Iwata, B.A., & Page, T.A. (1985).
Assessment of stimulus preference and reinforcer value with profoundly retarded
individuals. Journal of Applied Behavior Analysis, 18, 249-255.
Rachlin, H., Green, L., Kagel, J.H., & Battalio, R.C. (1976). Economic demand
theory and psychological studies of choice. In G.H. Bower (Ed.), The psychology of
learning and motivation (Vol. 10, pp. 129-154). New York: Academic Press.
Rast, J., & Jack, S. (1992). Mouthing. In E.A. Konarski, J.E. Favell (Eds.).,
Manual for the assessment and treatment of the behavior disorders of people with mental
retardation. (pp. 1-11). Morganton, NC: Western Carolina Center Foundation.
Repp, A.C., Deitz, S.M., & Deitz, D.E.D. (1976). Reducing inappropriate behaviors
in classrooms and in individual sessions through DRO schedules. Mental Retardation. 15,
Repp, A.C., Deitz, S.M., & Speir, N.C. (1974). Reducing stereotypic responding of
retarded persons by the differential reinforcement of other behaviors. American Journal of
Mental Deficiency, 79, 279-284.
Repp, A.C., & Karsh, K.G. (1990). A taxonomic approach to the nonaversive
treatment ofmaladaptive behavior of persons with developmental disabilities. In A.C.
Repp & N.N. Singh (Eds.), Perspectives on the use of nonaversive and aversive
interventions for persons with developmentally disabilities (pp. 331-347). Sycamore, IL:
Sycamore Publishing Co.
Rincover, A. (1978). Sensory extinction: A procedure for eliminating self-stimulatory
behavior in developmentally disabled children. Journal of Abnormal Child Psychology,6,
Rincover, A., & Devany, J. (1982). The application of sensory extinction procedures
to self-injury. Analysis and Intervention in Developmental Disabilities, 2. 67-81.
Rothbaum, B. 0. (1992). The behavioral treatment of trichotillomania. Behavioural
Psychotherapy, 20, 85-90.
Shore, B.A., & Iwata, B.A. (in press). Assessment and treatment of behavior
disorders maintained by nonsocial (automatic) reinforcement. In A.C. Repp & R.H.
Homer (Eds.), Functional analysis of problem behavior: From effective assessment to
effective support. Pacific Grove, CA: Brooks/Cole.
Steege, M.W., Wacker, D.P., Berg, W.K., Cigrand, K.K., & Cooper, L.J. (1989).
The use of behavioral assessment to prescribe and evaluate treatments for severely
handicapped children. Journal of Applied Behavior Analysis, 22, 23-33.
Steege, M.W., Wacker, D.P., Cigrand, K.C., Berg, W.K., Novak, C.G., Reimers,
T.M., Sasso, G.M., & DeRaad, A. (1990). Use of negative reinforcement in the treatment
of self-injurious behavior. Journal of Applied Behavior Analysis, 2, 459-468.
Tustin, R.D. (1994). Preference for reinforcers under varying schedule arrangements:
A behavioral economic analysis. Journal of Applied Behavior Analysis, 22, 597-606.
Vollmer, T.R. (1994). The concept of automatic reinforcement: Implications for
behavioral research in developmental disabilities. Research in Developmental Disabilities,
Vollmer, T. R., & Iwata, B. A. (1992). Differential reinforcement as treatment for
behavior disorders: Procedural and functional variations. Research in Developmental
Disabilities. 1, 393-417.
Wacker, D.P., Steege, M.W., Northrup, J., Sasso, G., Berg, W., Reimers, T.,
Cooper, L., Cigrand, K., & Donn, L. (1990). A component analysis of functional
communication training across three topographies of severe behavior problems. Journal of
Applied Behavior Analysis, 23, 417-430.
Warren, S.A., & Burs, N.R. (1970). Crib confinement as a factor in repetitive and
stereotyped behavior in retardates. Mental Retardation, 8, 25-28.
After receiving a Bachelor of Science degree in public relations at the University of
Florida in 1979, Bridget A. Shore worked for The Comer Drugstore in Gainesville as the
Volunteer Coordinator. There she was first taught to use behavioral procedures in
developing self-management programs for drug offenders and in teaching volunteers to
answer a 24-hour telephone crisis intervention hot-line. In 1982, she began working at
Sunland Center (now Tacachale) at Gainesville, an institution for persons with
developmental disabilities, where over a five-year period she advanced from training
specialist to human services senior supervisor. Through self-study, she learned more
about behavior analysis, and subsequently took the state certification exam to become an
HRS Certified Behavior Analyst in 1986. In 1987, Bridget became a member of the
Behavior Program Review Committee at Sunland and began developing and monitoring
behavior programs for persons with severe behavior problems. With this experience, she
obtained the position of Behavior Consultant with HRS/DD and worked for the next two
years providing assistance in the development of treatment programs for clients with
behavior problems throughout a sixteen-county district. In 1990, she left her job to attend
graduate school in applied behavior analysis at the University of Florida.
During Bridget's five years in graduate school, she has been a research assistant for
Dr. Brian Iwata at the Florida Center on Self-Injury, located at Tachachale. As a research
assistant, she has been involved in daily assessment and treatment sessions, data analysis,
staff training, administrative activities, supervision of research projects, and dissemination
of results through publication and conference presentations. Specific projects in which she
has had extensive involvement include staff training, generalization of treatment outcomes,
and the assessment and treatment of automatically-reinforced self-injury.
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
rian A. Iwata, Chair
Professor of Psycholog
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Marc N. trancif
Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
iothy D. H'k nberg
distant Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Michael J. Farr
Associate Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.
Assistant Professor of Special Education
This dissertation was submitted to the Graduate Faculty of the Department of
Psychology in the College of Liberal Arts and Sciences and to the Graduate School and
was accepted as partial fulfillment of the requirements for the degree of Doctor of
Dean, Graduate School
UNIVERSITY OF FLORIDA
3Il2l ii 85lll llI 91l0 1iii
3 1262 08553 9160