JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1994) 279 331-344 NUMBER2 (summER 1994)

TREATMENT OF SELF-INJURY AND HAND MOUTHING FOLLOWING

INCONCLUSIVE FUNCTIONAL ANALYSES
TIMOTHY R. VOLLMER, BETHANY A. MARcus, AND LINDA LEBLANC
LOUISIANA STATE UNIVERSITY

Numerous studies have used functional analyses to prescribe interventions for severe behavior
disorders. The majority of these studies have focused on behavior that is dearly maintained by
socially mediated reinforcement, such as contingent access to attention, tangibles, or escape from
instructional demands. However, a significant proportion of functional analyses do not yield con-
clusive results. We examined interventions for 3 children with severe disabilities following incon-
dusive functional analyses. First, preferred stimuli were identified for each child via a stimulus
preference assessment. Second, a functional analysis was conducted for 2 of the 3 children. High
rates of aberrant behavior were seen even (if not especially) in no-interaction sessions. The 3rd child
was observed for several consecutive no-interaction sessions; behavior persisted in this condition.
Third, interventions based on environmental enrichment were analyzed in an analogue setting. For
all of the children, environmental enrichment decreased aberrant behavior if preferred stimuli were
used in the procedure. To obtain further reductions in aberrant behavior, explicit reinforcement of
toy play was required for 2 children, and a brief (5-s) time-out was necessary with 1 child. Finally,
effects of treatment carried over to the school or home environment following teacher or family
training. Results are discussed in the context of basic reinforcement principles and future directions
for research.
DESCRIPTORS: environmental enrichment, functional analysis, stimulus preference, self-in-
jurious behavior, family training

The use of pretreatment functional analyses has yses have proven to be successful, the generality of
increased the likelihood of successful treatment for the current literature may be limited because most
severe behavior disorders displayed by individuals studies have focused on behavior that is clearly
with developmental disabilities (Mace, Lalli, & Pin- maintained by socially mediated reinforcement (e.g.,
ter-Lalli, 1991). When a functional analysis is con- attention, tangible reinforcement, or escape from
ducted before intervention, the experimenter or ser- instructions). In an epidemiological analysis, Iwata
vice provider may be able to develop an effective et al. (1994) reported that approximately 5% of
treatment, because variables that maintain a be- their functional analyses of self-injurious behavior
havior can be identified and rearranged to reduce were incondusive, and more than 25% did not
the likelihood of problem behavior while reinforc- show a sensitivity to socially mediated reinforce-
ing appropriate behavior (Iwata, Dorsey, Slifer, ment; Derby et al. (1992) reported that 34% of
Bauman, & Richman, 1982). For example, the their clients displayed aberrant behavior that ap-
intervention known as functional communication peared to be maintained by automatic sensory con-
training (FCT) involves replacing an undesirable sequences. Despite these data, disproportionately
behavior with an appropriate communicative re- few studies have focused on the treatment of be-
sponse that serves the same function (Carr & Du- havior disorders following undifferentiated results
rand, 1985). of functional analyses.
Although interventions based on functional anal- The focus on interventions for behavior main-
tained by socially mediated reinforcement may be
because an experimenter or clinician can control
We thank John Northup for helpful comments on an social contingencies. For example, if a behavior is
earlier version of this manuscript. maintained by attention, the environment can be
Reprints may be obtained from Timothy R. Vollmer,
Department of Psychology, Louisiana State University, Baton arranged so that attention is withheld or delivered
Rouge, Louisiana 70803. at the experimenter's discretion (e.g., Vollmer, Iwa-
331
332 TIMOTHY R. VOLLMER et al.
ta, Zarcone, Smith, & Mazaleski, 1993). Similarly, shift toward appropriate behavior. Even if biolog-
if behavior is maintained by escape, access to escape ical mechanisms, such as dopamine depletion (Bau-
can be withheld or delivered at the experimenter's meister, Todd, & Sevin, in press), are implicated
discretion (e.g., Iwata, Pace, Kalsher, Cowdery, & as maintaining factors, behavioral interventions may
Cataldo, 1990). However, if a behavior persists depend on the potency of the reinforcers used to
independent of the social environment or has been override the maintaining conditions. Second, "sen-
undifferentiated in a functional analysis, an exper- sory extinction" procedures may allow appropriate
imenter or dinician is less able to identify, control, behavior to compete with aberrant behavior because
or manipulate the relevant variables that maintain some of the stimulus products of aberrant behavior
the behavior. may be attenuated (Lovaas, Newsom, & Hickman,
Functional analyses may yield undifferentiated 1987; Rincover, 1978). Third, punishment of ab-
results for several reasons. One possibility is that errant behavior may shift response allocation to-
the behavior is maintained by automatic positive ward appropriate behavior by increasing the relative
or negative reinforcement that occurs independent value of reinforcement for the appropriate alter-
of social variables (see Vollmer, in press, for a recent native (Deluty, 1976).
review). For example, automatic positive reinforce- In this study we examined the role of stimulus
ment may take the form of endogenous opiate preference in interventions for self-injury and hand
production (Sandman, Barron, & Colman, 1990); mouthing following inconclusive functional anal-
automatic negative reinforcement may result from yses. In addition to assessing interventions for be-
pain-attenuating behavior such as self-scratching havior problems following undifferentiated results
(Carr & McDowell, 1980) and ear hitting (Cataldo of functional analyses, this study may advance the
& Harris, 1982). A second possibility is that be- current literature in the following ways: (a) Stim-
havior is multiply controlled (Smith, Iwata, Voll- ulus preference assessments are used in conjunction
mer, & Zarcone, 1993). A third possibility is that with environmental enrichment and reinforcement-
discrimination problems may occur due to rapidly based procedures; (b) the individual and combined
alternating conditions or subtle extraneous variables effects of treatment packages are assessed; (c) a
(Vollmer, Iwata, Duncan, & Lerman, 1993). Fi- method is described for systematically adding com-
nally, experimenters may fail to indude all relevant ponents to treatment packages to improve overall
variables in the analysis, such as biological factors treatment efficacy; (d) the results have implications
or idiosyncratic environmental stimuli. In this worst for early intervention (the participants were in pre-
case, the current state of our assessment method- school); and (e) the generality of the procedures
ology may not allow identification of all the vari- was assessed through systematic family training.
ables that maintain self-injury for given clients.
Behavior analysts are frequently asked to identify GENERAL METHOD
effective and acceptable interventions in the face of
undifferentiated or incondusive assessment data. At Participants and Setting
least three general strategies could be examined to Three children who attended a noncategorical
increase the likelihood of successful behavioral in- preschool participated in the study. The participants
tervention. First, the quality of the reinforcers in were the first 3 children referred to our program
an enriched environment (Homer, 1980) or in a for the analysis and treatment of self-injury whose
differential reinforcement procedure could be max- aberrant behavior showed no obvious differential
imized by using methods for ascertaining stimulus sensitivity to socially mediated reinforcement (see
preference (e.g., Fisher, Piazza, Bowman, Hago- description of functional analysis). The children's
pian, & Owens, 1992). The higher the quality of parents provided informed consent to participate
alternative reinforcers, the more responding should in the study.
 SELF-INJURY AND MOUTHING 333

Ron was a 3-year-old boy who had not received tions, observational sessions lasted 10 min (see
a formal diagnosis for his disabilities. He did not Family Training).
have any conventional language skills and had not
yet developed self-care skills such as toileting or Response Measurement and Reliability
dressing. He was ambulatory and appeared to enjoy Target behaviors were SIB, hand mouthing, and
being around other people. Ron was referred for appropriate behavior. SIB consisted of head bang-
intervention due to self-injurious behavior (SIB) in ing (Korey)-contact of the head against a solid
the form of head hitting, ear poking, and biting object; head/face hitting (Korey, Ron)-forceful
and sucking on his fingers. Korey was a 3-year-old contact of a hand or arm with any part of the head
boy with multiple disabilities; like Ron, he did not or face; ear picking (Ron)-scraping of the fin-
display conventional language or self-care skills. He gernails against the skin on or in the ear; hand
also had significant visual and auditory impair- biting (Korey)-contact of the teeth with the hand.
ments. Korey was nonambulatory and was unable Hand mouthing (Rhonda) was defined as any con-
to use his wheelchair independently. Some of his tact of the hand or fingers with the mouth that
fingers were not separated, which limited his grasp- broke the plane of the lips. For Korey and Ron,
ing ability. Although Korey was frequently unre- appropriate behavior was defined as contact with
sponsive to social interaction, he would occasionally toys; for Rhonda, appropriate behavior was defined
smile, laugh, or make brief eye contact. Korey was as contact with toys, food, or therapist (eye contact,
referred for head banging, head hitting, and hand touching). Target behaviors were recorded by ob-
biting. At the time of referral, his behavior was not servers using hand-held computers (Assistant, Model
forceful enough to incur significant tissue damage, AST 102) or laptop PCs. Observation sessions last-
but the behavior was reportedly increasing in in- ed 10 min, and data were converted into percentage
tensity as he got older. Rhonda was a 4-year-old of intervals using a 10-s partial-interval recording
girl who engaged in many "autistic-like" behaviors procedure.
such as hand twirling and rocking. She also dis- To assess interobserver agreement, a second ob-
played no conventional communication. She was server concurrently but independently collected data
ambulatory but walked with an unusual and un- for each participant during at least 20% of the
steady gait. She had mild visual impairments that sessions in each condition of each phase (functional
were corrected through the use of glasses. Rhonda analysis, analogue treatment analysis, and family
was referred for treatment because she engaged in training). Each 10-s interval was scored as either
chronic hand mouthing that resulted in the appli- an agreement or a disagreement on the occurrence
cation of an arm brace throughout the school day or nonoccurrence of a behavior. The number of
before intervention. intervals scored as an agreement was divided by
Analogue sessions took place in a classroom at the total number of intervals in each session. Mean
the preschool, but each child was observed without agreement scores for SIB were 94.3% and 98.3%
other children in the room. The classroom was for Korey and Ron, respectively. Mean agreement
equipped with toys and a variety of academic ma- for mouthing was 96.0% for Rhonda. Mean agree-
terials that were put away during sessions. The area ment scores for appropriate behavior were 93.3%,
of the room used for the experimental sessions was 84.9%, and 85.2% for Korey, Ron, and Rhonda,
approximately 7 m2 with a carpeted floor; one wall respectively. The lowest agreement score during any
was lined with large windows. Observers sat at a condition for any of the 3 participants was 81.0%
teacher's desk or at a work table during sessions. (appropriate behavior for Rhonda during the an-
Family training sessions took place at the partici- alogue treatment analysis). Because there was no
pants' homes, and dassroom observations took place opportunity for appropriate behavior in the no-
in each child's regular classroom. With rare excep- interaction baselines (see Procedure below), no
334 TIMOTHY R. VOLLMER et al.

agreement scores for that target response were cal- PHASE 1: FUNCTIONAL ANALYSIS
culated during that condition. Procedure
During family training, interobserver agreement The functional analyses were based on the pro-
measures were taken for correct antecedent (Korey's cedures described by Iwata et al. (1982), and in-
family), correct consequence to inappropriate be- duded an assessment of sensitivity to tangible pos-
havior (Rhonda's family), and correct consequence itive reinforcement. The stimuli identified during
for appropriate behavior (Korey's and Rhonda's the stimulus preference assessment were used as
family). Interobserver agreement scores for those preferred stimuli in relevant conditions. Sessions
categories averaged 8 1.3% (range, 46.7% to 100%), were conducted one to four times per day, 4 days
99.2% (range, 90% to 100%), and 92.1% (range, per week, depending on the participants' schedules.
80% to 100%), respectively. Korey and Rhonda were exposed to the follow-
ing conditions: attention (preferred stimuli were
Stimulus Preference Assessments available noncontingently, attention was delivered
A stimulus preference assessment was conducted only contingent on SIB), tangible positive rein-
prior to the functional analyses using the procedures forcement (preferred stimuli were delivered contin-
described by Fisher et al. (1992). The child was gent on SIB), instructions (academic instructions or
presented with randomized pairs of stimuli and was tasks were presented once every 30 s using a three-
asked to select the one he or she wanted. When prompt instructional sequence; instructions were
the child touched one of the two stimuli, the re- terminated contingent on SIB), no interaction (par-
sponse was scored as a "selection." The experi- ticipant was observed in an austere environment;
menter then allowed the child to contact the stim- no programmed consequence for SIB), and play
ulus for approximately 30 s. For example, if the (noncontingent access to preferred stimuli, fixed-
stimulus was a toy, the child could play with it. time 30-s attention, and no programmed conse-
At least 10 stimuli were assessed for each child, quence for SIB). These conditions were presented
and each item was paired with all other items at during 10-min sessions in a multielement format.
least once. The stimuli used in the assessment were Rhonda's arm brace was removed for all sessions.
either suggested by the child's teacher or parent or Before his intervention analysis, Ron was ex-
were items that the children routinely contacted at posed only to the no-interaction condition. We
the school, and included food, drinks, auditory toys wanted 1 participant to have no explicit social re-
(e.g., music), visual toys (e.g., lights), tactile toys inforcement history in the analogue setting. A no-
(e.g., Koosh® ball), and traditional toys (e.g., dolls, interaction baseline is not sufficient evidence to sug-
cars), among others. Randomized pairs were pre- gest that his behavior was insensitive to social con-
sented until dear preferences were identified, as tingencies, but only that the behavior would persist
defined by the highest percentage of selection per in the absence of programmed contingencies. Be-
item (for each child, at least two stimuli were se- tween sessions, Ron was given attention only fol-
lected on over 80% of the trials). Preferred stimuli lowing periods of several seconds without SIB, to
identified for Ron were toy rubber cirdes, blocks, reduce the possibility of accidental reinforcement.
and magnets. For Korey, preferred stimuli included Previous research has shown that SIB that is main-
a sound-making toy (e.g., See-n-Say®) and a music tained by social consequences will extinguish under
box. For Rhonda, preferred stimuli induded sound- similar circumstances with repeated no-interaction
making toys, cookies, and a Koosh ® ball. A second sessions (Vollmer, Iwata, Duncan, & Lerman,
stimulus preference assessment was conducted for 1993). Also, prior research has shown that the
Korey at a later time (see description of treatment noncontingent attention delivered between sessions
procedures). In the second assessment, Korey showed should not result in an accidental reinforcement
a preference for sound-making toys, a mirror, and effect (Vollmer, Iwata, Zarcone, Smith, & Maza-
a toy bus. leski, 1993).
 SELF-INJURY AND MOUTHING 335

Results 1 0N
Figure 1 shows the results of the analysis for
each participant. For Korey, the level of SIB was
undifferentiated across all conditions. The lowest
levels of SIB were seen in conditions in which pre-
ferred stimuli were available (attention and play).
The highest rates of SIB were seen in the conditions z
in which preferred stimuli were not available (de- I
mand, tangible, and no interaction). Korey was
observed for 11 consecutive no-interaction sessions, 0
and the SIB persisted (with the exception of one ai
session). For Rhonda, the target behavior occurred 100-
at a high level across all assessment conditions and C)
-i
during four consecutive no-interaction sessions. For a No interact
Ron, with the exception of Sessions 10 and 1, the LU Tangible
target behavior persisted at a level above 25% of Attention
z Demand
the intervals despite the absence of social conse- 50-
A

Play
cn o
quences during all of his assessment sessions. The
insensitivity of Ron's SIB to social contingencies LL
was further supported in his treatment analysis (see 0
Phase 2). LUJ
CD 0
PHASE 2: ANALOGUE TREATMENT ANALYSIS z
Procedure LUJ
100 -
0~
Baseline and treatment sessions generally lasted LU
10 min and were conducted one to four times per
day, 4 days per week. The baselines were taken
from the no-interaction observations reported in
Phase 1 (for Korey and Rhonda) and from the last 50 -
half of Ron's assessment observations in Phase 1.
In the descriptions below, the term enriched en-
vironment will be used to describe the environment
in comparison to the no-interaction condition. It is
acknowledged that the term should generally refer 0 10 20 30
to an environment enriched with preferred stimuli
only. SESSIONS
Ron. The treatment for Ron was analyzed with
1. Functional analyses for Korey (upper panel)
an ABCBCB reversal design. The A condition was andFigure Rhonda (center panel) and repeated no-interaction ses-
baseline, in which Ron was observed in no inter- sions for Ron (lower panel). Target behaviors are expressed
action (no toys, no therapist present) with no pro- as percentage of 10-s intervals observed.
grammed consequence for SIB. An observer was
present in the room. The B condition was an en- assessment) (EE -). The second C condition also
riched environment with preferred stimuli (EE+). was an enriched environment with nonpreferred
The C condition was an enriched environment with stimuli, but Ron was exposed to a single session
nonpreferred stimuli (those selected in less than of each of the conditions described for Korey's and
20% of the opportunities during the preference Rhonda's functional analyses (Phase 1). The final
336 TIMOTHY R. VOLLMER et al.

B condition was an enriched environment with pre- therapist assisting Rhonda in completing the re-
ferred stimuli, but Ron was exposed to conditions sponse. Because Rhonda's mother and teacher had
described for Korey's and Rhonda's functional expressed concern about her lack of social inter-
analyses in Phase 1 (a tangible-item condition was action, social contacts such as eye contact and touch-
not conducted here because by definition he had ing were followed by 5 s of verbal praise and arm
free access to preferred stimuli in the enriched en- touching. The E condition was identical to the D
vironment). condition except that mouthing resulted in 5 s of
Korey. The treatment for Korey was analyzed hands-down time-out, in which the therapist
using an ABCAC reversal design. The A condition blocked the response and held Rhonda's hands at
was baseline, in which Korey was observed in no her side. Blocked responses were scored as an in-
interaction (no toys, no therapist interaction) seated stance of mouthing, and time-out intervals were
next to a large table with no programmed conse- not calculated as part of the 10-s interval.
quence for SIB. The B condition was an enriched
environment with preferred stimuli (placed within Results and Discussion
arm's reach next to the wheelchair). The only dif- Ron. Figure 2 shows the results of Ron's treat-
ference between the environmental enrichment and ment analysis. His SIB persisted (M = 43.1%;
the play conditions during the functional analysis range, 5% to 73.3%) in the absence of programmed
was that toys were always kept within Korey's reach reinforcement contingencies (baseline). When pro-
and there was no attention delivered. This was vided with preferred stimuli (EE+), the percentage
accomplished by seating Korey in front of a larger of intervals with SIB decreased markedly (M =
table and replacing dropped toys immediately. The 2.8%; range, 0% to 23.3%); most of his time was
C condition was an enriched environment combined spent holding or playing with the preferred stimuli
with reinforcement for toy contact; a therapist ac- (M = 93%; range, 55% to 100%). When the
tivated a toy when Korey touched it. During the preferred stimuli were replaced with nonpreferred
final C condition, the stimuli used in the enriched stimuli (EE-), SIB increased (M = 49.6%; range,
environment were based on a second preference 1.7% to 88.3%) and appropriate behavior de-
assessment conducted because treatment gains be- creased (M = 14.2%; range, 0% to 56.7%). The
gan to dedine. At the end of the final C condition, effects of EE+ were replicated when a reversal to
Korey's sound-making toys were activated by a that condition was conducted (M for SIB = 0.2%;
microswitch, eliminating the need for a therapist M for appropriate = 99.4%). Finally, exposure to
to complete the response. conditions identical to the fimctional analyses for
Rhonda. Rhonda's arm brace was removed dur- Korey and Rhonda resulted in high rates of SIB
ing all sessions. Treatment for Rhonda was analyzed only when toys were unavailable (M when un-
using an ABCDEBE reversal design. The A baseline available = 52.1%; M when available = 0.2%).
condition was identical to that of Ron and Korey. These results demonstrate the value of a preinter-
The B condition was an enriched environment with vention stimulus preference assessment; it was not
nonpreferred stimuli (EE-). The C condition was the availability of stimulation per se that resulted
an enriched environment with preferred stimuli in a decrease in SIB; preferred stimuli were required.
(EE+). As with Korey, EE sessions differed slightly In addition, there was an inverse relation between
from play sessions. During EE, preferred stimuli appropriate behavior and SIB, suggesting that re-
were explicitly kept within an approximately 2-m inforcement derived from the preferred stimuli may
radius (Rhonda almost always elected to stay seated have overridden the variables that maintained SIB.
on the carpet during observational sessions). The Finally, the limitations of Ron's pretreatment as-
D condition was positive reinforcement combined sessment were addressed by demonstrating that so-
with EE+: Hand-mouthing responses were blocked, cial contingencies did not influence his SIB.
and stimulus interactions were reinforced by the Following the analogue intervention, recom-
 SELF-INJURY AND MOUTHING 337
Test conditions

Baseline EE + EE- EE+ EE- EE

1 00'

50

C)
Baseline EE+ EE+/Sr+ BL EE+/Sr+ EE+/Sr+ (new assessment)
1 00.

w
H-

z
LL
0
Cw) 50

z
Lu]
0w 0.
LLJ

1 00'

50*

0.

10 20 30 40 50 60

SESSIONS
Figure 2. Upper panel: Percentage of intervals with SIB and appropriate behavior for Ron during baseline (no interaction),
enriched environment (preferred stimuli) (EE+), enriched environment (nonpreferred stimuli) (EE-), and enriched envi-
ronment (EE+). Finally, functional analyses were conducted in the context of EE- versus EE+. Center panel: Percentage
of intervals with SIB and appropriate behavior for Korey during baseline (no interaction), enriched environment (preferred
stimuli) (EE+), EE+ and positive reinforcement, baseline, EE+ and positive reinforcement, and EE+ and positive
reinforcement following an updated stimulus preference assessment. Lower panel: Percentage of intervals with hand mouthing
and appropriate behavior for Rhonda during baseline, enriched environment (nonpreferred stimuli) (EE-), enriched en-
vironment (preferred stimuli) (EE+), EE+ and positive reinforcement, EE+ and positive reinforcement plus time-out,
EE-, and EE+ and positive reinforcement plus time-out.
338 TIMOTHY R. VOLLMER et al.
mendations were made to enrich Ron's classroom low level and appropriate behavior stayed at a rel-
environment during free time (i.e., make preferred atively high level.
toys available). Ron's teacher was taught to conduct Subsequent to our treatment analysis, Korey's
daily abbreviated stimulus preference assessments, teacher conducted abbreviated preference assess-
and she then made the preferred toys available to ments and provided continuous access to preferred
Ron on a nearly continuous basis (except at times stimuli during free time at school. The EE classroom
when toys would interfere with learning activities). was equipped with switch-activated toys as well as
Follow-up data were collected after 4.5 months in other toys (such as Korey's mirror and bus). Two
Ron's classroom during six separate 10- to 15-min 10-min follow-up observations (not depicted in
observations (not depicted in Figure 2). Follow-up Figure 2) were conducted in Korey's new classroom
observations during regular ongoing classroom ac- 5 months after completion of our analogue treat-
tivities showed that SIB remained at low levels ment analysis. Follow-up observations were un-
when preferred toys were available (M = 2.2% of announced and occurred during regular classroom
intervals). Follow-up data in the classroom also activities. He displayed no SIB and showed high
showed that SIB remained at high levels when rates of appropriate behavior (M = 75% of inter-
preferred stimuli were unavailable (M = 57.0%). vals). One observation was conducted during switch-
Before observations with stimuli unavailable, pre- toy play and no teacher interaction, and a second
ferred toys were removed from the room by an observation was conducted during other toy play
experimenter. and teacher interaction. Additional generaliza-
Korey. Figure 2 shows that Korey's SIB also tion and follow-up data for Korey are reported in
decreased relative to baseline (baseline M = 48.8%; Phase 3.
range, 0% to 80%) as a function of an environment Korey's results further demonstrate the utility of
enriched with preferred stimuli (EE+) (M = a preintervention stimulus preference assessment.
13.9%; range, 6.7% to 21.7%); however, there Not only did access to preferred stimuli reduce
was a slight increasing trend in SIB correlated with occurrences of SIB, but those reductions were ap-
a sharp decreasing trend in appropriate behavior parently a function of a current preference for the
(M for appropriate = 57.1%; range, 36.7% to stimuli. Thus, stimulus preferences may need to be
85%). It was noted anecdotally that Korey was updated on a frequent basis (Mason, McGee, Farm-
unable to manipulate the preferred toys effectively er-Dougan, & Risley, 1989). The results also il-
due to his extensive motor and physical disabilities. lustrate that, if a child is unable to manipulate his
When the therapist began activating the toy con- or her preferred stimuli in an enriched environment,
tingent on toy contact (EE+/Sr+), the percentage explicit reinforcement or assistance from a therapist
of intervals with appropriate behavior increased (M may be required.
= 82.6%; range, 70% to 97%) with a concurrent Rhonda. Figure 2 shows that an enriched en-
decrease in SIB (M = 3.8%; range, 1.7% to 8.3%). vironment had only a moderate effect on Rhonda's
Following a reversal to baseline (M = 46%; range, hand mouthing (baseline M = 87.2%; EE- M
16.7% to 73.3%), the EE+/Sr+ condition was = 52.9%). The enriched environment containing
reimplemented; however, treatment effects began preferred stimuli (EE + ) reduced mouthing slightly
to deteriorate (SIB M = 18%; range 0% to 33.3%; more than the enriched environment with nonpre-
appropriate M = 69.7%; range, 25% to 88.3%). ferred stimuli (EE-) (M = 47.5%; range, 28.3%
After conducting a second stimulus preference as- to 65%). EE+ also resulted in an increase in ap-
sessment, Korey's SIB was again reduced (M = propriate behavior when compared with EE - (M
3.8%; range, 0% to 13.3%) with a concurrent = 42.5% vs. 21.4%). Despite the reduction,
increase in appropriate behavior (M = 88%; range, mouthing still occurred at high levels. When pos-
75% to 100%). Finally, when Korey was able to itive reinforcement was implemented (EE +/Sr+),
activate his sound-making toys, SIB stayed at a appropriate behavior was further increased (M =
 SELF-INJURY AND MOUTHING 339

66.6%; range, 58.1% to 85%) with concurrent visit. One to four observation blocks were dispersed
decreases in mouthing (M = 29.3%; range, 11.7% throughout the hour. Although reactivity to ob-
to 51.9%). However, mouthing still occurred at servers was expected, reactive effects were held con-
fairly high levels and showed a gradual increasing stant across baseline and treatment as much as
trend. Finally, when response blocking (one com- possible.
ponent of the EE+/Sr+) was replaced with a 5-s Rhonda. Rhonda's mother and older sister (12
time-out, mouthing decreased (M = 11.1%; range, years old) participated in the training. Rhonda's
0% to 48.3%) with concurrent increases in appro- mother identified the most problematic times of
priate behavior (M = 73.9%; range, 41.7% to day as before and after dinner when Rhonda's
96.7%). These effects were replicated following a mother was cooking or her sister was cleaning.
reversal to EE-. Follow-up data and additional Thus, the hour before or after dinner was targeted
evidence of experimental control are reported for for implementation of the procedure.
Rhonda in Phase 3. Following family and teacher During baseline, the experimenters placed pre-
training, her arm brace was permanently removed ferred stimuli in the room, but otherwise did not
at home and at school. alter the typical family environment. Following a
Rhonda's results point to some limitations of series of baseline observations alternating between
interventions that rely on competition between re- the mother and sister in the family room with
inforcers delivered by the environment and un- Rhonda, training was conducted. Rhonda's mother
known variables maintaining aberrant behavior. was trained before her sister, and the procedure was
Specifically, preferred stimuli apparently were not the same as that used in the terminal condition of
potent enough to compete with the variables that Phase 2 (EE + /Sr+ /time-out). First, the experi-
maintained Rhonda's hand mouthing. Despite a menter demonstrated an abbreviated version of the
modest reduction in mouthing during the enriched stimulus preference assessment, which allowed
environment condition, it was not until a time-out Rhonda to select preferred stimuli for that evening.
contingency was implemented that satisfactory Second, training consisted of a verbal description
changes in behavior were observed. It is unknown of our findings in the analogue setting. Although
whether the hands-down time-out was effective be- the family was generally aware of the procedures
cause of contingent aversive stimulation (punish- we used, baseline data showed that the family was
ment) or a contingent loss of opportunity to obtain not conducting the procedure at home. Third, the
sensory reinforcement (time-out). As with Ron and procedure was modeled by an experimenter for 10
Korey, Rhonda's behavior showed a strong inverse min with accompanying verbal descriptions and
relation between appropriate and inappropriate re- explanations. Fourth, about 1 min of role playing
sponding. Thus, mild punishment did not result was conducted between the trainee and the exper-
in a generalized suppression of behavior. imenter. Fifth, the mother and sister conducted the
procedure for 10 min with immediate feedback
from the experimenter (e.g., "She just put her hand
PHASE 3: FAMILY TRAINING in her mouth, be sure to hold her hands down
Procedure now"; "Good, you turned on the toy for her when
Rhonda, Korey, and their families participated she touched it"). Sixth, a 10-min session was con-
in the family-training component at their homes; ducted by the mother and sister with the experi-
Ron and his family were unable to participate. menter providing feedback after the completion of
Family training visits were usually conducted once the session (e.g., "There was one instance when
per week for approximately 1 hr per family for 7 you forgot to put her hands down"). The family
weeks (not including follow-up). During baseline was asked to conduct the procedure every night,
and posttraining observations, 5-min blocks of time and posttraining observations commenced the fol-
were randomly selected for observation during the lowing week.
340 TIMOTHY R. VOLLMER et al.

The dependent measures for Rhonda and her grammed, Korey's parents occasionally prompted
mother or sister were percentage of intervals with the older siblings to play with him. The observers
hand mouthing, percentage of intervals with ap- never prompted play or interaction.
propriate behavior, percentage of correct responses The dependent measures for Korey and his
to hand mouthing (hands down for 5 s + 1 s, brothers were percentage of intervals with SIB, per-
calculated as a percentage of opportunity given the centage of intervals with appropriate behavior, per-
initiation of hand mouthing), and percentage of centage of intervals with correct antecedent (toys
correct responses to appropriate behavior (toy play within arm's reach), and percentage of correct re-
contingent on toy touch or social interaction con- sponses to appropriate behavior (toy activation or
tingent on eye contact or touching, calculated as a play contingent on toy touching, calculated as a
percentage of opportunities given the initiation of percentage of the opportunities given an initiation
appropriate behavior). Interobserver agreement of appropriate behavior). Interobserver agreement
measures were reported in the General Method measures were reported in the General Method
section above. section above.
Korey. Korey's 2 brothers (7 and 9 years old)
participated in family training. Preliminary obser- Results
vations showed that Korey's mother interacted with Training was conducted for Rhonda's mother,
him in a manner consistent with our procedure (she Rhonda's sister, and Korey's brothers sequentially
had discovered that toy play and other stimulation in a multiple baseline design. Figure 3 shows the
resulted in low rates of SIB; also, she allowed him results of training for each dyad or triad. In each
to select stimuli by pointing from an array). We case, dear reductions in the target behavior and
were unable to observe Korey interacting with his concurrent increases in appropriate behavior were
father, but his mother reported that their interac- correlated with the completion of family training.
tions were consistent with our program. An inter- These results replicate those seen in the analogue
view with Korey's mother indicated that she had treatment analysis.
time to play with him and take him on walks Figure 3 also shows the results of 1- and 5-month
through the neighborhood after school, but that follow-up observations with Rhonda's family. Ko-
before and after dinner were problematic times be- rey's family was not available for follow-up obser-
cause his brothers did not interact well with him. vations. Both families reported that the procedures
We recommended that the family purchase switch- were conducted virtually every evening that the
activated toys to use during this time. Although family was home.
she agreed that switch-activated toys would be ben- Treatment integrity measures showed improve-
eficial, Korey's mother expressed an interest in en- ment for both families from baseline to posttrain-
couraging his brothers to play with him at least 1 ing. Rhonda's mother never provided the correct
hr per day before or after dinner. consequence for mouthing in baseline, but that
Following a series of baseline observations with measure increased to a mean of 76.6% of oppor-
both brothers in the room (Korey's preferred stim- tunities following training (range, 10% to 100%).
uli were available but were not placed within his Her correct consequences for appropriate behavior
reach), training was conducted. Training followed increased from a mean of 55% during baseline
the same sequence used with Rhonda's family; (range, 42.9% to 66.7%) to 89.5% following train-
however, both brothers were trained as a pair, with ing (range, 69.2% to 100%). During baseline,
either child delivering the antecedents (toys within Rhonda's sister rarely provided the correct conse-
arm's reach) and consequences (activation or ma- quence for mouthing (M = 0.8%; range, 0 to
nipulation of toys contingent on touching). Post- 3.3%), but this measure increased to a mean of
training observations commenced the week after 87.4% following training (range, 69.2% to 100%).
the training session. Although it was not pro- During baseline, Rhonda's sister responded cor-
 SELF-INJURY AND MOUTHING 341
Follow up
mo . 5 mo

Rhonda w/mother

20

llq
LL

0
w
0O | Rhonda w/sister
0

Korey w/brotherz

10 20

SESSIONS
Figure 3. Effects of family training on aberrant and appropriate behavior for Rhonda with her mother (upper panel)
and sister (center panel) and for Korey with his brothers (lower panel). Target behaviors are expressed as percentage of 10-s
intervals observed.
342 TIMOTHY R. VOLLMER et al.
rectly to appropriate behavior during an average of duction in hand mouthing was clinically significant.
42.3% of the opportunities (range, 25% to 5 5.5%), Collectively, these results indicate that the appli-
which increased to a mean of 90.7% after training cation of a preintervention stimulus preference as-
(range, 65.2% to 100%). During baseline, Korey's sessment may help to override unknown maintain-
brothers never provided correct antecedents or cor- ing variables.
rect consequences for appropriate behavior. Follow- From a conceptual standpoint, the results of this
ing treatment, they provided correct antecedents study are relevant to an understanding of variables
during an average of 77.8% of the intervals (range, that maintain aberrant behavior. In all three cases,
60% to 96.7%) and correct consequences for ap- concurrent reductions in SIB or mouthing and in-
propriate behavior for a mean of 77.3% of the creases in appropriate behavior showed an inverse
opportunities (range, 1 1. 1% to 100%). relation. Neef, Mace, and Shade (1993) suggested
A booster training session was conducted for that functional analyses may examine not only the
Rhonda's mother after Session 8 and for Rhonda's type of reinforcement contingencies but also indi-
sister after Session 10. Booster sessions replicated vidual sensitivities to different reinforcement di-
the initial training, but with 5-min rather than 10- mensions. For Ron, the relative quality of his pre-
min modeling and feedback sessions. The sessions ferred stimuli appeared to compete effectively with
were conducted because of decreasing trends in the aberrant behavior. However, when the quality of
integrity of treatment implementation. The trends the stimuli was reduced (as defined by lower pref-
were reversed following the boosters. Korey's broth- erence in the assessment), behavior was allocated
ers did not receive further training after the initial almost exclusively to the unknown schedule main-
session, but the parental prompting was first noted taining aberrant behavior. Presumably some mod-
in Session 16; these prompts may have served the erately preferred stimulus would have resulted in
function of a booster session. response allocations somewhere between the two
extremes, as with Rhonda during her EE+ and
EE- conditions.
GENERAL DISCUSSION For Rhonda, the relative quality of the preferred
The results of this study show that, despite in- stimuli appeared to approximate the potency of the
condusive functional analyses, aberrant behavior reinforcers or other variables that were maintaining
can be treated using interventions based on envi- hand mouthing (see EE+ condition). When a time-
ronmental enrichment. Treatment packages were out contingency was introduced, responding shifted
highly individualized, due to idiosyncratic prefer- further to appropriate behavior. This finding is
ences for reinforcers across participants. With each compatible with basic research showing shifts in
participant, the introduction of preferred stimuli response allocation on concurrent schedules when
into the environment also resulted in increases in punishment is introduced (Deluty, 1976). Alter-
appropriate behavior. For Ron, the enriched en- natively, the time-out contingency may have been
vironment competed with unknown variables effective because of a reduced probability of rein-
maintaining SIB. For Korey, the enriched environ- forcement for aberrant behavior.
ment competed with unknown variables when al- Limitations of this study should be pointed out.
ternative responding was explicitly reinforced and First, although a functional analysis was conducted,
when stimulus preference was regularly reassessed. the specific maintaining variables for SIB were not
The enriched environment with preferred stimuli identified for each child; thus, behavioral reduction
competed with Rhonda's aberrant behavior only to based on overriding the maintaining variables rath-
a degree; responding was allocated about equally er than eliminating them may not result in long-
between aberrant and appropriate behavior. It was term reduction of behavior.
not until responding produced time-out that a re- A second limitation is the possibility that socially
 SELF-INJURY AND MOUTHING 343

mediated sources of reinforcement were maintain- REFERENCES

ing the behavior on intermittent schedules that were Baumeister, A. A., Todd, M., & Sevin, J. A. (in press).
not identified or detected in our analysis. Although Efficacy and specificity of pharmacological therapies for
we were careful not to provide explicit consequences behavioral disorders in persons with mental retardation.
for the behavior, the intermittency of reinforcement Clinical Neuropharmacology.
Carr, E. G., & Durand, M. V. (1985). Reducing behavior
schedules is never certain in naturalistic contingen- problems through functional communication training.
cies. More extended no-interaction sessions may be Journal of Applied Behavior Analysis, 18, 111-126.
required in future investigations, but ethical con- Carr, E. G., & McDowell, J. J. (1980). Social control of
siderations (given the age, educational environ- self-injurious behavior of organic etiology. Behavior
Therapy, 11, 402-409.
ment, and undifferentiated analyses for the partic- Cataldo, M. F., & Harris, J. (1982). The biological basis
ipants in this study) precluded further observation for self-injury in the mentally retarded. Analysis and
in austere environments. Despite this limitation, Intervention in Developmental Disabilities, 2, 21-39.
our results are consistent with other studies showing Deluty, M. Z. (1976). Choice and the rate of punishment
in concurrent schedules. Journal of the Experimental
behavioral insensitivity to social reinforcement (e.g., Analysis of Behavior, 25, 75-80.
Mazaleski, Iwata, Rodgers, Vollmer, & Zarcone, Derby, K. M., Wacker, D. P., Sasso, G., Steege, M., Nor-
1994; Van Houten, 1993). thup,J., Cigrand, K., & Asmus,J. (1992). Brief func-
A third limitation is that, on a practical level, it tional assessment techniques to evaluate aberrant behav-
ior in an outpatient clinic: A summary of 79 cases.Journal
would have been difficult to conduct some of the of Applied Behavior Analysis, 25, 713-721.
procedures on a 24-hour basis. For the children Egel, A. L. (1981). Reinforcer variation: Implications for
involved in family training, it appeared that the motivating developmentally disabled children. Journal
of Applied Behavior Analysis, 14, 345-350.
procedures were implemented at home at appro- Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P.,
priate times, but future research may incorporate & Owens, J. C. (1992). A comparison of two ap-
observational data to isolate problem periods for proaches for identifying reinforcers for persons with severe
which labor-intensive interventions can be imple- and profound disabilities. Journal of Applied Behavior
Analysis, 25, 491-498.
mented most stringently. One potential problem Horner, R. D. (1980). The effects of an environmental
with enriched environment treatments is that con- 'enrichment" program on the behavior of institution-
tinuous exposure to preferred stimuli may result in alized profoundly retarded children. Journal of Applied
eventual satiation. However, satiation effects were Behavior Analysis, 13, 473-491.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E.,
not pronounced in this investigation. We interpret & Richman, G. S. (1982). Toward a functional analysis
the absence ofobvious satiation as primarily a func- of self-injury. Analysis and Intervention in Develop-
tion of updating stimulus preference, and teachers mental Disabilities, 2, 1-20.
Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R.,
and families allowing a wide range of choices. Vary- Vollmer, T. R., Smith, R. G., Rodgers, T. A., Lerman,
ing reinforcers to avoid satiation is consistent with D. C., Shore, B. A., Mazaleski, J. L., Goh, H., Cowdery,
previous accounts of maintaining reinforcer efficacy G. E., Kalsher, M. J., McCosh, K. C., & Willis, K. D.
(e.g., Egel, 1981). (1994). The functions of self-injurious behavior: An
experimental-epidemiological analysis. Journal of Ap-
Finally, more exhaustive functional analyses are plied Behavior Analysis, 27, 215-240.
required to identify specific maintaining variables Iwata, B. A., Pace, G. M., Kalsher, M. J., Cowdery, G. E.,
for undifferentiated SIB and stereotypy; such anal- & Cataldo, M. F. (1990). Experimental analysis and
yses may require greater interaction among biolog- extinction of self-injurious escape behavior. Journal of
Applied Behavior Analysis, 23, 11-27.
ical and behavioral scientists. Until these methods Lovaas, 0. I., Newsom, C. D., & Hickman, C. (1987).
are developed, environmental enrichment, differ- Self-stimulatory behavior and perceptual reinforcement.
ential reinforcement, and mild punishment appear Journal of Applied Behavior Analysis, 20, 45-68.
Mace, F. C., Lalli, J. S., & Pinter-Lalli, E. (1991). Func-
to be viable treatment options for behavior disor- tional analysis and treatment of aberrant behavior. Re-
ders that produce incondusive results from a func- search in Developmental Disabilities, 12, 155-180.
tional analysis. Mason, S. A., McGee, G. G., Farmer-Dougan, V., & Risley,
344 TIMOTHY R. VOLLMER et al.

T. R. (1989). A practical strategy for ongoing reinforcer Van Houten, R. (1993). The use of wrist weights to reduce
assessment. Journal of Applied Behavior Analysis, 22, self-injury maintained by sensory reinforcement.Journal
17 1-179. of Applied Behavior Analysis, 26, 197-203.
Mazaleski, J. L., Iwata, B. A., Rodgers, T. A., Vollmer, T. Vollmer, T. R. (in press). The concept of automatic re-
R., & Zarcone, J. R. (1994). Protective equipment as inforcement: Implications for behavioral research in de-
treatment for stereotypic hand mouthing: Sensory ex- velopmental disabilities. Research in Developmental
tinction or punishment effects? Journal of Applied Be- Disabilities.
havior Analysis, 27, 345-355. Vollmer, T. R., Iwata, B. A., Duncan, B. A., & Lerman, D.
Neef, N. A., Mace, F. C., & Shade, D. (1993). Impulsivity C. (1993). Extensions of multielement functional anal-
in students with serious emotional disturbance: The in- yses using reversal-type designs. Journal of Develop-
teractive effects of reinforcer rate, delay, and quality. mental and Physical Disabilities, 5, 311-325.
Journal of Applied Behavior Analysis, 26, 37-52. Vollmer, T. R., Iwata, B. A., Zarcone, J. R., Smith, R. G.,
Rincover, A. (1978). Sensory extinction: A procedure for & Mazaleski, J. L. (1993). The role of attention in the
eliminating self-stimulatory behavior in developmentally treatment of attention-maintained self-injurious behav-
disabled children. Journal of Abnormal Child Psychol- ior: Noncontingent reinforcement (NCR) and differential
ogy, 6, 299-3 10. reinforcement of other behavior (DRO). Journal of Ap-
Sandman, C. A., Barron, J. L., & Colman, H. (1990). An plied Behavior Analysis, 26, 9-26.
orally administered opiate blocker, naltrexone, attenuates
self-injurious behavior. American Journal on Mental
Retardation, 95, 93-102. Received October 13, 1993
Smith, R. G., Iwata, B. A., Vollmer, T. R., & Zarcone, J. Initial editorial decision December 27, 1993
R. (1993). Experimental analysis and treatment of mul- Revisions receivedJanuary 26, 1994; February 22, 1994
tiply controlled self-injury. Journal of Applied Behavior Final acceptance February 24, 1994
Analysis, 26, 183-196. Action Editor, F. Charles Mace