Research in Developmental Disabilities

21 (2000) 393– 407

Functional analysis of aberrant behavior

maintained by automatic reinforcement:
assessments of specific sensory reinforcers夞
Meeta R. Patel*,1, James E. Carr2, Christine Kim3,
Adel Robles, Dixie Eastridge
University of Nevada, Reno, NV 89557-0062, USA

Received 16 June 1999; accepted 6 July 1999

Abstract

The purpose of this study was to develop a systematic functional assessment package
for aberrant behaviors maintained by nonsocial (automatic) reinforcement. The assess-
ment package included four components: (1) functional analysis, (2) antecedent assess-
ment of specific automatic reinforcement sources, (3) stimulus preference assessment, and
(4) treatment evaluation. Functional analysis data indicated automatic reinforcement
functions of the stereotypy exhibited by a 10-year-old male and the self-injury (SIB)
exhibited by a 30-year-old male. Antecedent assessments of sensory classes indicated that
auditory stimulation and tactile stimulation were associated with stereotypy and SIB,
respectively. A multiple-stimulus-without-replacement procedure was conducted with
each participant to identify the most- and least-preferred stimuli within the identified
sensory classes. In an attempt to validate the assessment package for each participant, a
DRO procedure was implemented using a reversal design with a multielement component.
DRO procedures using stimuli within the targeted sensory classes were successful in
eliminating the aberrant behaviors of both participants. The results are discussed in the

夞 This study is based on a master’s thesis submitted by the first author to the Department of
Psychology at University of Nevada. We thank MaryAnn Demchak, Patrick Ghezzi and John Rapp
for their helpful comments on an earlier version of the manuscript.
1
Meeta R. Patel is now at the Marcus Institute, 1605 Chantilly Dr., Atlanta, GA.
2
James E. Carr is now at Western Michigan University.
3
Christine Kim is now at West Virginia University.
* Corresponding author. Tel.: ⫹1-404-727-9462; fax: ⫹1-404-727-9550.
E-mail address: Meeta.Patelat@Marcus.org (M.R. Patel).

0891-4222/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 8 9 1 - 4 2 2 2 ( 0 0 ) 0 0 0 5 1 - 2
394 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

context of improving the methodology for assessing and treating automatically reinforced
behaviors. © 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction

An abundance of research on the treatment of aberrant behavior maintained by

social reinforcement exists in the applied literature. However, relatively few
studies have examined the analysis and treatment of behavior maintained by
automatic reinforcement (in the absence of the social environment). One of the
earliest studies to investigate aberrant behavior maintained by automatic rein-
forcement examined “sensory extinction” (Rincover, 1978). Rincover posited
that by masking the sensory consequences postulated to maintain a particular
behavior, the behavior would extinguish. The author demonstrated this interven-
tion with an individual who frequently spun plates on tabletops, presumably to
obtain auditory stimulation. Rincover masked the auditory stimulation produced
by the plate spinning and the behavior decreased.
Favell, McGimsey, and Schell (1982) hypothesized that the self-injurious
behaviors (SIB) exhibited by six individuals were maintained by automatic
reinforcement. The authors provided the participants with noncontingent tangible
stimuli that presumably substituted for the sensory stimulation produced by those
behaviors. Pica and hand mouthing decreased when individuals were given
rubber toys and eye poking decreased after alternate visual stimulation was
presented. The authors interpreted the successful treatment effects as evidence
that these behaviors were maintained in the absence of the social environment.
The aforementioned studies are particularly important to the treatment liter-
ature, as they were the first to address the specific putative reinforcers for
aberrant behavior maintained by automatic reinforcement. However, one limita-
tion of these studies is a lack of functional assessment to determine behavioral
function prior to intervention. Without such data it is difficult to predict or even
expect successful treatment effects (e.g., Rincover, 1978). In addition, successful
treatments might be indicative of reinforcer competition instead of substitution;
thus, making it difficult to infer behavioral function from treatment effects.
Although a few recent studies have not conducted functional assessments
prior to successful treatment implementation (Luiselli, 1994; Sigafoos & Pennell,
1995), most researchers studying automatic reinforcement have utilized the
functional analysis methodology. Slifer, Iwata and Dorsey (1984) concluded, via
a functional analysis, that a case of eye-gouging behavior was maintained in the
absence of social consequences. No further analyses of the sensory consequences
were conducted; it was assumed that the behavior was maintained by visual
stimulation. The behavior was then successfully treated using a response inter-
ruption procedure. A more recent study found that eye-poking behavior was
maintained by the consequence of the response (Kennedy & Souza, 1995). The
authors hypothesized that visual stimulation maintained eye poking and pre-
sented alternate sources of stimulation to the participant. Two sources of stim-
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 395

ulation were compared: music (auditory) and a video game (visual). Only the
presentation of the visual stimulus had a reductive effect on the behavior,
supporting the hypothesis that the behavior was maintained by visual stimulation.
Behavior maintained in the absence of social contingencies may require an
extended series of assessments to determine the specific sensory stimuli that
maintain or are at least correlated with the target behavior. To date, few studies
on automatic reinforcement have gone beyond the traditional analog analysis.
Goh et al. (1995) conducted a functional analysis, which indicated that partici-
pants’ hand mouthing was maintained by automatic reinforcement. The authors
then extended the analysis in an effort to determine the specific stimuli main-
taining hand mouthing. Resulting data indicated that hand mouthing was primar-
ily maintained by hand stimulation (as opposed to oral stimulation). Although the
authors did not conduct a treatment evaluation in order to verify their assessment,
their method has proven useful in the development of more specific analyses of
behavior associated with automatic reinforcement. Another recent study exam-
ined sensory stimuli that may compete with stereotypy and SIB (Sprague,
Holland, & Thomas, 1997). Their extended analysis consisted of potential com-
peting sensory consequences. The authors demonstrated that alternative sensory
stimulation was a more effective consequence when compared to social praise,
supporting their automatic reinforcement hypothesis.
Piazza, Hanley, and Fisher (1996) combined an experimental functional anal-
ysis with stimulus preference assessments to treat cigarette pica. First, the authors
conducted a multielement assessment, which demonstrated that pica occurred
more often with cigarette butts containing nicotine than with those that did not.
Next, they conducted a stimulus preference assessment, which showed that
tobacco was highly preferred compared with the other components of a cigarette.
Third, the authors conducted a functional analysis to demonstrate that pica was
maintained independent of the social environment. Cigarette consumption was
reduced to near-zero levels when the hypothesized response-reinforcer relation-
ship was interrupted using a noncontingent reinforcement procedure combined
with response blocking. Finally, a stimulus control procedure was implemented
to improve effectiveness of the intervention in unsupervised environments.
Piazza et al. (1998) concluded that the pica of three individuals was either
maintained by automatic reinforcement alone or both automatic and socially
mediated reinforcement (i.e., attention or access to tangible items). A stimulus
preference assessment was conducted to determine if stimuli producing oral
stimulation (matched stimuli) were preferred over other types of stimuli (un-
matched stimuli). Matched stimuli were generally preferred when compared to
unmatched stimuli and treatments based on the presumed function of pica (oral
stimulation) were more effective than treatments unrelated to the hypothesized
function. A further analysis was also conducted to determine the exact properties
of oral stimulation that were reinforcing. A firmness analysis provided additional
support that the pica of two individuals was maintained by oral stimulation. This
analysis confirmed that the texture of various food items was an important aspect
of preference as well as treatment effectiveness.
396 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

The purpose of the current investigation was to extend the above studies,
especially those by Piazza et al. (1996, 1998), by conducting functional analyses
and subsequent assessments to determine the specific sensory qualities correlated
with the aberrant behavior. After functional analyses were conducted, we imple-
mented antecedent assessment (similar to that reported by Goh et al., 1995) in
which sensory stimuli related to the target behaviors were presented or removed.
Stimuli identified in these assessments were then presented to the participants in
a stimulus preference assessment (similar to Piazza et al.). Subsequently, differ-
ential reinforcement interventions were then used to verify the predictions of the
assessment package.

2. Method

2.1. Participants and setting

Two individuals participated. Austin was a 10-year-old male who was diag-
nosed with autism. He attended a special education school and lived with his
parents. He was referred by the school district for assessment and treatment of
rapid tongue movements. According to his teacher, this behavior distracted other
children in the class and made it difficult to teach Austin functional skills. Darion
was a 30-year-old male diagnosed with severe mental retardation and showed
signs of fetal alcohol syndrome. He was referred by his residential facility
because his self-injury produced scars and calluses on his forehead.
All sessions for Austin were conducted at his school in a setting (3 m x 3 m)
that was used as a conference room. The room contained a long table and several
chairs. A therapist, the participant, and another individual who videotaped the
sessions were present during all sessions. The first phase of the functional
analysis for Darion was conducted in an analog room unfamiliar to him. As head
hitting was eliminated in the analog setting, sessions were moved to the natural
setting. These sessions were conducted in the workroom (5 m x 3 m) at Darion’s
day program. The room was furnished with a long table, with several chairs, a
chalkboard, and some filing cabinets.

2.2. Data collection: dependent variable

The dependent variable for Austin was rapid tongue movements and for
Darion it was head hitting. Rapid tongue movement was operationally defined as
the continuous (more than three repetitions per incident) movement of the tongue
in an up-and-down, visible fashion. These rapid tongue movements also pro-
duced an audible product (i.e., when Austin engaged in the target behavior it
produced noise). Head hitting was operationally defined as any hand-to-head
contact toward the right side of his forehead that was repeated more than three
times. This operational definition was chosen to eliminate other responses such
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 397

as head scratching. All sessions were 10 min in duration and were conducted 3
to 4 times per day in the afternoon. Most sessions were videotaped and later
scored; however, some sessions included online data collection with a data
collector present in the session room. All sessions were scored using a continuous
partial-interval recording system. The target behavior was recorded if the indi-
vidual engaged in the behavior at any time during a given 10-s interval. Data
were then converted to a percentage of occurrences of the target behavior for a
given session.
Interobserver agreement (IOA) was calculated for 63% (functional analysis),
29% (antecedent assessment), and 39% (treatment evaluation) of sessions for
Austin. IOA was calculated for 25% (functional analysis), 65% (antecedent
assessment), and 27% (treatment evaluation) of sessions for Darion. IOA was
calculated by dividing the number of agreements by the number of agreements
plus disagreements and multiplying by 100%. Mean IOA scores for Austin and
Darion, respectively, for each phase are as follows: (1) functional analysis -
97.6% (range, 92–100) and 97.4% (range, 92–100); (2) antecedent assessment -
98.0% (range, 92–100) and 96.0% (range, 92–98); (3) stimulus preference
assessment - 100% for Austin (IOA was not calculated for Darion); and (4)
treatment evaluation - 98.5% (range, 97–100) and 95.3% (range, 85–100).

2.3. Data collection: independent variable

Independent variable (IV) integrity was evaluated in the functional analysis

phase and during the treatment evaluation. During the no-interaction condition of
the functional analysis, IV integrity was determined by recording any interactions
between the participant and the investigator. IV integrity in the attention condi-
tion was analyzed by recording the number of times attention was provided
contingent on the target behavior. During the demand condition IV integrity was
determined by recording the number of times a participant was allowed to escape
a given task contingent on the target behavior. IV integrity was evaluated during
the control condition by recording whether or not a therapist provided attention
on a fixed-time (FT) 30-s schedule. IV integrity was evaluated in the treatment
evaluation by observing the presentation of a particular stimulus on the prede-
termined schedule (in the absence of the target behavior).
IOA for IV integrity measures was obtained and calculated using the same
procedures described above for the dependent variables. IOA was calculated for
all of the sessions in which IV integrity data were calculated.
IV integrity data were calculated for 50% (functional analysis) and 44%
(treatment evaluation) of Austin’s sessions. IV integrity data were calculated for
86% (functional analysis) and 21% (treatment evaluation) of Darion’s sessions.
Mean IV integrity for Austin and Darion, respectively, for each phase were as
follows: (1) functional analysis - 100% and 99.6% (range, 93–100), and (2)
treatment evaluation -100% for each participant. Mean independent variable IOA
scores for Austin and Darion, respectively, for each phase were as follows: (1)
398 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

functional analysis - 98% (range, 93–100) and 100%, and (2) treatment evalua-
tion - 100% and 98.1% (range, 87–100).

2.3.1. Phase 1: functional analysis

To determine the maintaining variables of the target behaviors, a functional
analysis was conducted before treatment. An extended analysis within a multi-
element design was used for both participants. The test conditions were similar
to the ones described by Iwata, Dorsey, Slifer, Bauman and Richman (1994/
1982) and included attention, escape/avoidance, no interaction, and control
conditions.

2.3.2. Phase 2: antecedent assessment of sensory stimuli

The purpose of the antecedent assessment was to determine the nature of the
sensory stimuli that were associated with the target behaviors. A multielement
design was used for both participants. Austin’s antecedent assessment consisted
of providing him access to different forms of stimulation. Prior to this phase we
audio recorded the sound produced when he engaged in rapid tongue movements.
Subsequently, during the auditory condition, this recording was played back to
Austin (noncontingently). The purpose of the auditory condition was to deter-
mine if the behavior was maintained by the auditory product of the target
response. A no-interaction condition was included to establish a “baseline” for
comparison. We assessed the auditory condition initially because it appeared to
be the most likely variable that was associated with the target behavior. More-
over, the extended auditory conditions were presented to eliminate any possible
novelty effect. Austin’s antecedent assessment concluded with two vibratory
conditions and moisturizer conditions. In the vibratory #1 condition noncontin-
gent access to a vibrating toothbrush was provided. In vibratory #2 condition
noncontingent access to a vibrating candy holder was provided. The vibratory
conditions were included in order to determine if the behavior was maintained by
the vibration produced from the tongue movements. Finally, the moisturizer
condition was included in order to determine if these tongue movements were
maintained by automatic negative reinforcement, in the form of relief from dry
lips.
In Darion’s antecedent assessment, stimulation was either presented or elim-
inated. There were four conditions: forehead stimulation, reduced visual stimu-
lation, forehead attenuation, and the no-interaction condition. Again, the no
interaction condition was imbedded with the other test conditions for compari-
son. The forehead stimulation condition consisted of noncontingent access to a
manual head massage on a FT 15s schedule (that was determined by calculating
the mean interresponse time (IRT) in the no- interaction conditions of the initial
functional analysis). The purpose of the forehead stimulation condition was to
determine if head hitting was maintained by head stimulation (tactile stimula-
tion). The reduced visual stimulation condition consisted of blocking visual
stimulation by having Darion wear tinted goggles. The purpose of this condition
was to determine if the behavior was maintained by visual stimulation in the form
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 399

of hand movements past his visual field. In the forehead attenuation condition
Darion wore a ski hat that covered his forehead. The purpose of this condition
was to determine if the target behavior was maintained by hand stimulation.

2.3.3. Phase 3: stimulus preference assessment

We conducted 5 preassessment sessions in which the participants had free
access to the stimuli for 30 min. Stimuli for both participants were chosen
according to the results of Phase 2. All stimuli in Austin’s preference assessment
consisted of auditory toys and all stimuli in Darion’s assessment consisted of
items that could provide head stimulation. Preference was assessed using a
multiple-stimulus without replacement (MSWO) procedure (DeLeon & Iwata,
1996). The therapist presented 5 items to the participants in a linear array. The
participants were permitted to choose one stimulus item from the array. After a
particular stimulus was chosen, they had 10s access to the item, after which time
the trials resumed. This procedure continued until all items were chosen, or until
no choice was made. This procedure was repeated 3 times. Preference was
determined as the percentage of times an item was selected. The most- and
least-preferred stimuli were used during the treatment evaluation phase. The
second stimulus preference assessment was conducted in a paired-stimulus for-
mat. The above two stimuli were presented to each participant 6 times. After
choosing a particular stimulus, the participants had 10s of access to the item.

2.3.4. Phase 4: treatment evaluation

A treatment evaluation was conducted using a reversal design with a multi-
element component. Both participants were exposed to baseline condition until
their data were stable. A differential reinforcement condition was then imple-
mented using the most- and least-preferred stimuli from the preference assess-
ments in a multielement fashion. Once data were stable, baseline conditions were
again implemented, followed by a final treatment condition. The baseline ses-
sions were conducted in the no-interaction context.
During the treatment phase, the two stimuli were alternated in a differential
reinforcement of zero rates of responding (DRO) schedule that was determined
by using the mean IRT established during baseline (Repp, Deitz, & Deitz, 1976).
Austin’s mean IRT during baseline was 20 s (with 10 s access to the stimulus),
and Darion’s mean IRT during baseline was 15 s (with 5 s access to the stimulus).
Access to the stimulus was contingent upon the nonoccurrence of the target
behavior during the specified interval.

3. Results

The results of the functional analysis for each participant are shown in Fig. 1.
These data represent an undifferentiated functional analysis with three final
consecutive no interaction sessions. Mean percentage occurrence for each test
400 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

Fig. 1. Percentage occurrence of target behaviors during functional analysis conditions for Austin (top
panel) and Darion (bottom panel).

condition was as follows: no interaction, 49.7% (SD ⫽ 5.2), attention, 47.7%

(SD ⫽ 6.4), escape, 41.7% (SD ⫽ 14.0), and control, 38.7% (SD ⫽ 7.6). Because
Darion’s behavior was eliminated during the analog sessions, we then conducted
assessments in his work area of the day program he attended. Head hitting
persisted throughout the no- interaction condition and was at low levels in all the
other conditions. Mean percentage occurrence for each test condition was as
follows: no interaction, 72.6% (SD ⫽ 19.5), attention, 3.7% (SD ⫽ 1.2), escape,
12.8% (SD ⫽ 9.7), and control, 9.75% (SD ⫽ 8.0).
The results of the antecedent assessment for each participant are shown in Fig.
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 401

Fig. 2. Percentage occurrence of target behaviors during antecedent assessment conditions for Austin
(top panel) and Darion (bottom panel).

2. For Austin, responding during the auditory condition was variable, but reached
stability by session nine. The target behavior decreased by about 50% when
compared to functional analysis data with noncontingent access to auditory
stimulation (M ⫽ 31.0%, SD ⫽ 13.5%). The no interaction condition produced
variable and generally high levels of responding throughout the antecedent
assessment (M ⫽ 61.3%, SD ⫽ 6.2%). The level of responding during the
no-interaction condition of the antecedent assessment is consistent with the data
presented in the functional analysis. Data for the vibratory #1 condition initially
402 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

were at low levels with a subsequent increasing trend (M ⫽ 37.0, SD ⫽ 17.1%).

The vibrator #2 session represents behavior at levels similar to vibrator #1. The
moisturizer condition produced high levels of responding relative to the no
interaction condition, but with a slight increasing trend (M ⫽ 73.0%, SD ⫽
3.5%).
For Darion, responding in the forehead stimulation condition was at near-zero
levels throughout the analysis (M ⫽ 1.6%, SD ⫽ 2.3%). Responding in the
reduced visual stimulation condition was low and stable (M ⫽ 23.6%, SD ⫽
8.8%). The forehead attenuation condition data were more variable, but still
relatively low (M ⫽ 28.7%, SD ⫽ 10.4%). Finally, the no interaction condition
produced high levels of responding throughout the analysis (M ⫽ 66.8%, SD ⫽
16.9%), similar to that which is evident in the functional analysis.
Selection percentages for Austin’s first stimulus preference assessment are as
follows: musical bear (60%), drum (40%), musical mickey mouse toy (40%),
music box (25%), and the musical elmo toy (20%). Based upon these results, the
most and least preferred items were used in treatment evaluation. After session
22 (on Fig. 3), a second preference assessment was conducted in an effort to
explain the counterintuitive treatment effects assessed during DRO conditions for
the musical bear and the musical elmo toy. The second preference assessment
compared the musical elmo toy with the bear in a paired-stimulus assessment.
The musical elmo toy was chosen 100% of the time. Selection percentages for
Darion’s first stimulus preference assessment are as follows: thermomassager
(75%), body massager (40%), facial massager (40%), head pack (25%), and a
manual massage (15%). A second preference assessment was conducted after
session 16, as with Austin. The thermomassager and manual massage were
compared in a paired-stimulus assessment, resulting in each being chosen 50% of
the time.
Treatment evaluation data for both participants are shown in Fig. 3. For
Austin, baseline data had little variability and produced an upward trend (M ⫽
53.9%, SD ⫽ 7.9%). During the DRO assessment, the schedule that included the
bear produced lower levels of responding initially, with a steep upward trend
toward the end of the phase (M ⫽ 17.0%, SD ⫽ 15.9%). The DRO schedule that
included the musical elmo toy produced near-zero levels of responding through-
out the phase (3.3%, SD ⫽ 1.6%). Removal of the DRO contingencies produced
responding similar to baseline (M ⫽ 47.2%, SD ⫽ 9.8%). The final DRO
condition was implemented solely with the musical elmo toy and produced
responding similar to the initial implementation (M ⫽ 2.3%, SD ⫽ 0.6%).
Darion’s baseline data were high and stable (M ⫽ 70.7%, SD ⫽ 18.1). The
subsequent DRO phase included sessions with the thermomassager and the
manual massage in a multielement fashion. With the exception of a relatively
high data point during session 17, responding was at low levels during the
thermomassager condition (13.3%, SD ⫽ 12.9%). However, responding was
consistently much lower in the manual massage condition throughout the phase
(M ⫽ 3.7%, SD ⫽ 3.3%). Removal of the DRO contingencies produced re-
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 403

Fig. 3. Percentage occurrence of target behaviors during baseline and DRO conditions for Austin (top
panel) and Darion (bottom panel).

sponding similar to baseline (M ⫽ 80.7%, SD ⫽ 12.1%). The final DRO

condition was implemented solely with the manual massage and produced
responding similar to the initial implementation (M ⫽ 2.0%, SD ⫽ 3.5%).

4. Discussion

The functional analyses confirmed that both participants’ target behaviors

were maintained independent of the social environment. Our interpretations of
the antecedent assessment data were that Austin’s rapid tongue movements were
404 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

substantially reduced when the auditory product of this behavior was presented
noncontingently. Likewise, Darion’s head hitting was reduced when noncontin-
gent forehead tactile stimulation was provided. Stimulus preference assessments
indicated the most and least preferred stimuli within the identified sensory class.
Finally, treatment evaluations that were based on specific hypotheses of associ-
ated variables proved successful; however, the utility of the stimulus preference
assessment data were questionable during these evaluations.
The complexity of Austin’s antecedent assessment requires additional discus-
sion. In the auditory condition, Austin was provided with an almost-identical
representation of the auditory product of his behavior. The reduction in the target
behavior during the auditory condition of the antecedent assessment suggests that
auditory stimulation was contributing to the maintenance of Austin’s rapid
tongue movements. However, because complete reduction was not observed, it
was assumed that other stimuli may also be associated with the target behavior
(e.g., tactile stimulation produced by movement of the tongue within the mouth).
The most likely alternative is vibratory stimulation, which is indicated by the
initial reduction observed during the vibratory stimulus conditions. However, an
alternative explanation for that initial reduction is a novelty effect, which was
followed by habituation, as evident by the increasing trend across sessions. In
addition, during the initial vibratory stimulus conditions, Austin engaged with the
stimulus orally; however, during subsequent sessions he frequently placed the
vibrating toothbrush in his hands. This change in stimulus engagement argues
against a reinforcer substitution hypothesis as the engagement was topographi-
cally dissimilar to the target behavior. However, it is possible that the quality of
the vibration was not functionally equivalent to the maintaining reinforcer. Thus,
we cannot eliminate the possibility that vibratory stimuli were functional rein-
forcers of Austin’s behavior.
A reduction in Darion’s head hitting was observed across all test conditions in
the antecedent assessment, with a nearly complete reduction in the forehead
stimulation condition. In the forehead attenuation condition, Darion wore a ski
hat to cover his forehead, yet it was difficult to keep his forehead covered
throughout the sessions. He frequently moved the hat up and engaged in the
target behavior; thus, it was impossible to completely attenuate the stimulation he
received while engaging in head hitting. Reduction in the target behavior during
this condition might be explained as a sensory extinction effect.
Although the stimulus preference assessments for each participant produced
gradients that led to the selection of least- and most-preferred stimuli, the
gradients were only moderately significant. This finding may have been a result
of using a low number of stimuli. In addition, the novelty of the stimuli may have
suppressed consistent preference between items, as exposure to each stimulus
was brief.
During the initial sessions of each participant’s treatment evaluation, both
reductions were seen with each of the stimuli. With Austin, the musical elmo toy,
although the least-preferred stimulus, consistently functioned as an effective
stimulus within the DRO schedule. These results do not support the predictions
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 405

of the stimulus preference assessment, but again the limited exposure to the
stimuli may have skewed the preference assessment. A second preference as-
sessment confirmed that the musical elmo toy was the most-preferred stimulus
when compared to the bear at that assessment time. Toward the end of treatment
phase 1, the percent occurrence of the target behavior during the DRO schedule
with the bear was similar to baseline, indicating that the bear no longer func-
tioned as an effective stimulus within the intervention. It could be that preference
changed with time, or that we were unable to accurately measure preference from
the beginning.
Darion’s treatment evaluation was similar to Austin’s in that initial reduction
was achieved when using both stimuli that provided head stimulation. Although
the thermomassager was the most preferred stimulus, reduction was not as robust
as with the manual massage. While there was no significant increasing trend
when implementing the thermomassager, the manual massage did produce better
suppressive effects. In order to determine whether preference changed across
time, a second stimulus preference assessment was conducted. There was no
difference in preference between both stimuli. It could be that both stimuli were
equally reinforcing, but a more viable alternative may be that preference was not
adequately assessed for the manual massage. The other items in the array were
represented in actual form; however, the manual massage was represented as an
open hand.
The mechanism responsible for treatment effects are uncertain since a dis-
crepancy exists between the first and second preference assessment. If the
mechanism responsible for treatment effects was simply reinforcement effects,
then it can be argued that the second preference assessment is more valid (in
Austin’s case). If the mechanism involves competition, then it is not possible to
determine which preference assessment was more valid. Another possible mech-
anism responsible for treatment effects may be reinforcer substitution. The
antecedent assessment may have determined possible substitutable sensory rein-
forcers associated with the target behavior. This explanation is less viable for
Austin as the auditory conditions only produced a reduction, rather than a
complete elimination of the target behavior. Reinforcer competition and substi-
tution are viable mechanisms for our treatment effect, but more research is
necessary in this area.
The limitations of the study generally concern the antecedent assessment and
stimulus preference assessment phases. During the antecedent assessment, we
were unable to manipulate all possible maintaining variables (i.e., sensory mech-
anisms) associated with the target behavior (cf. Piazza et al., 1996, 1998). This
finding is typical with behaviors maintained in the absence of social contingen-
cies for which the correlated stimuli cannot be directly related (as with pica).
However, noncontingent presentation of the auditory product did result in con-
sistently lowest levels of the target behavior. In addition, the stimulus preference
assessment did not prove useful in either case as greater reductions were ob-
served during the lower-preference stimulus conditions. As mentioned earlier,
406 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407

modification in the preference assessment method might allow better predictive

validity of treatment effects.
The current study contributes to the literature in two ways. First, this is one of
the only studies to implement a systematic methodology to evaluate specific
sensory stimuli that are correlated with the target behavior. Second, this study
devised a comprehensive assessment package that was validated by a functional
treatment approach. Previous studies addressing automatic reinforcement and
function-based treatments focus on behaviors where the reinforcer can be con-
trolled (such as pica), but the current investigation addresses behaviors where the
reinforcer is controlled by the individual.
Behavior maintained by reinforcers that are not socially mediated are among
the most difficult to treat. However, by developing a systematic assessment
strategy, function-based treatments may be prescribed with greater confidence.
Although the results of this study are promising, more research is needed in the
area. Future research areas include modifications to stimulus preference methods
that allow multiple stimuli to be assessed in an equal manner over time. Simi-
larly, the use of lengthier preexposure sessions needs to be explored so that we
can account for stimulus novelty effects. In order to better understand the
mechanism responsible for treatment effects future researchers may want to
compare matched (stimuli within the identified sensory class) versus unmatched
stimuli (stimuli unrelated to the identified sensory class) in treatment. Systematic
assessment methods, such as those reported in the current study should be
addressed in a brief fashion so that behavior maintained in the absence of the
social environment can be analyzed in a time- and cost-prohibitive manner.
Finally, the treatments prescribed from specific functional analyses need to be
compared with treatments prescribed by stimulus preference assessments
(Ringdahl, Vollmer, Marcus, & Roane, 1997) in terms of magnitude, satiation
effects, and long-term maintenance of treatment gains.

References

DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple-stimulus presentation format for
assessing reinforcer preferences. Journal of Applied Behavior Analysis, 29, 519 –533.
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.
Goh, H., Iwata, B. A., Shore, B. A., DeLeon, I. G., Lerman, D. C., Ulrich, S. M., & Smith, R. G.
(1995). An analysis of the reinforcing properties of hand mouthing. Journal of Applied Behavior
Analysis, 28, 269 –283.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1994). Toward a
functional analysis of self-injury. Journal of Applied Behavior Analysis, 27, 197–209. (Reprinted
from Analysis and Intervention in Developmental Disabilities, 2, 3–20, 1982).
Kennedy, C. H., & Souza, G. (1995). Functional analysis and treatment of eye poking. Journal of
Applied Behavior Analysis, 28, 27–37.
Luiselli, J. K. (1994). Effects of noncontingent sensory reinforcement on stereotypic behaviors in a
child with posttraumatic neurological impairments. Journal of Behavior Therapy and Experi-
mental Psychiatry, 25, 325–330.
 M.R. Patel et al. / Research in Developmental Disabilities 21 (2000) 393– 407 407

Piazza, C. C., Hanley, G. P., & Fisher, W. W. (1996). Functional analysis and treatment of cigarette
pica. Journal of Applied Behavior Analysis, 29, 437– 450.
Piazza, C. C., Fisher, W. W., Hanley, G. P., LeBlanc, L. A., Worsdell, A. S., Lindauer, S. T., &
Keeney, K. M. (1998). Treatment of pica through multiple analyses of its reinforcing functions.
Journal of Applied Behavior Analysis, 31, 165–189.
Repp, A. C., Deitz, S. M., & Deitz, D. E. (1976). Reducing inappropriate behaviors in classrooms and
in individual sessions through DRO schedules of reinforcement. Mental Retardation, 14, 11–15.
Rincover, A. (1978). Sensory extinction: A procedure for eliminating self-stimulatory behavior in
developmentally disabled children. Journal of Abnormal Psychology, 6, 299 –310.
Ringdahl, J. E., Vollmer, T. R., Marcus, B. A., & Roane, H. S. (1997). An analogue evaluation of
environmental enrichment: The role of stimulus preference. Journal of Applied Behavior Anal-
ysis, 30, 203–216.
Sigafoos, J., & Pennell, D. (1995). Noncontingent application versus contingent removal of tactile
stimulation: Effects on self-injury in a young boy with multiple disabilities. Behaviour Change,
12, 139 –143.
Slifer, K. J., Iwata, B. A., & Dorsey, M. F. (1984). Reduction of eye gouging using a response
interruption procedure. Journal of Behavior Therapy and Experimental Psychiatry, 15, 369 –375.
Sprague, J., Holland, K., & Thomas, K. (1997). The effect of noncontingent sensory reinforcement,
contingent sensory reinforcement, and response interruption on stereotypical and self-injurious
behavior. Research in Developmental Disabilities, 18, 61–77.