self-injurious behavior







A three-part controlled case study is presented in which severe and longstanding self-injurious behavior exhibited by a 9-year-old-boy was treated successfully with differential reinforcement of other behavior. In Phase 1, an experimental analysis demonstrated that the boy’s scratching was not maintained by environmental contingencies; instead, it appeared that the self-injurious behavior was a stereotypic (automatically reinforced) response. In Phase 2, the effects ofan escalating differential- reinforcement-of-other-behavior schedule mediated through token reinforcement (pennies) were evaluated in a reversal design. Results showed that differential-reinforcement-of-other-behavior eliminated self-injurious behavior very quickly and for periods of time as long as 30 min. A noteworthy side effect observed during Phase 2 was the occurrence of crying behavior following the nondelivery of reinforcement. In Phase 3, the token program was gradually extended in 30-min increments throughout the day. Additionally, results of a brief multielement manipulation showed that the effects oftoken reinforcement were superior to those ofa more easily administered differential reinforcement of other behavior based on social reinforcement, which differed little from baseline. DESCRIPTORS: differential reinforcement, DRO, functional analysis, self-injurious behavior,

side effects, stereotypic behavior

Differential reinforcement in its various forms (differential reinforcement ofother behavior [DRO], differential reinforcement of alternative or incom- patible behavior [DRA/DRI], and differential re- inforcement of low rates of behavior [DRL]) is one of the most basic, widely known, and commonly used techniques to suppress undesirable behavior. Although appropriate implementation requires the creation of a deprivation state-limited access to the reinforcer-as well as extinction, differential reinforcement does not involve extended interrup- tion of ongoing activities (e.g., time-out), contin- gent removal of positive reinforcers (e.g., response cost), or presentation of aversive stimuli (e.g., pun-

This research was supported by Grant HD 16052 from the National Institute of Child Health and Human Devel- opment. We thank Michael Kalsher, Kay McCosh, and Jay Quinn for assisting with various aspects of the research.

Reprints may be obtained from Brian Iwata, Psychology Department, University of Florida, Gainesville, Florida 32611.

ishment). These characteristics of differential rein- forcement make it the least intrusive of all behavioral interventions and probably account for its wide- spread popularity. Yet in spite of its many positive features, it is not dear that differential reinforcement is a highly effective intervention, particularly as a treatment for severe and longstanding behavior dis- orders. One such disorder is self-injurious behavior (SIB),

a chronic and intractable problem that takes mul- tiple forms and poses serious health risks. Reviews of the literature (e.g., Favell, Azrin, et al., 1982; Romanczyk, 1986) are generally consistent in con- duding that differential reinforcement alone is not very effective compared to most other behavioral treatments for SIB. In fact, out of the hundreds of research artides and case studies published, only a handful reported positive results when differential reinforcement was the only form of intervention (Allen & Harris, 1966; Favell, McGimsey, &Jones,


1990,231,497-506 NUMBER4 (wiNm 1990)


1978; Favell, McGimsey, & Schell, 1982; Frankel, Moss, Schofield, & Simmons, 1976; Steege, Wack- er, Berg, Cigrand, & Cooper, 1989; Weihar & Harman, 1975).

In light of the paucity of research supporting the use of differential reinforcement as a treatment for SIB, additional investigations are needed to doc- ument its effectiveness and to establish its legitimacy on empirical rather than philosophical grounds. During the course of treating an unusual and severe case of SIB, we had the opportunity to conduct such an investigation. The results of our behavioral assessment, and subsequent evaluation of DRO as a sole means of treating SIB, are reported in the present study.


Subject and Settings Jerry was a 9-year-old boy whose SIB consisted

of self-excoriation through either scratching or rub- bing. The problem dated back at least 6 years. Although he tested within the low-normal range of intelligence, he had never attended school due to the severity of his SIB. For the 2 years prior to his participation in this study, Jerry spent almost all of his time in hospitals, where he was examined and treated by a variety of specialists, induding dermatologists, neurologists, pediatricians, psychi- atrists, and psychologists. During his most recent hospitalization, the use of punishment (brief ap- plication of a cold pack to his skin contingent on scratching) had reduced SIB to approximately 30% of observed intervals. Neither this nor any of the other interventions tried in the past were judged to be “clinically adequate.” At the time of his entry into the study, he wore bandage wraps on his head, neck, arms, legs, and feet, and he had open but unbandaged lesions on his face, shoulders, hands, chest, back, buttocks, and penis. Jerry underwent additional medical tests throughout his participa- tion in this study to determine if the scratching might be the result of an allergic or dermatologic condition. All of these tests were negative.

Sessions were conducted in a hospital inpatient unit in areas equipped with one-way observation

windows. Phases 1 and 2 were conducted privately in treatment rooms, whereas Phase 3 was conducted in a group activity area where 2 to 6 other clients were present at any given time.

Response Measurement and Reliability Jerry’s SIB was defined as scratching or rubbing

one body part against another, or rubbing a body part against a stationary object (e.g., furniture). During Phases 1 and 2, in which sessions lasted a maximum of 30 min, data were recorded during continuous 10-s intervals on a hand-held computer (Panasonic Model RL-H1800). Session length in Phase 3 sometimes exceeded 5 hr and rendered the use of 10-s intervals unfeasible. Therefore, data during Phase 3 were collected during continuous 5-min intervals. Although inaccurate with respect to the actual frequency of Jerry’s SIB, the 5-min intervals provided an extremely conservative mea- sure in that any response during the interval resulted in a positive score. All data were converted to per- centage of intervals during which one or more SIBs occurred. During Phases 1 and 2 of the study, crying behavior (defined as any audible sobbing, whining, or complaining) also was recorded during continuous 10-s intervals on the computer.

Reliability was assessed by having a second ob- server simultaneously but independently record data with the primary observer. Percentage agreement scores (overall, occurrence, and nonoccurrence) were calculated based on interval-by-interval compari- sons of observers’ records. Reliability was assessed during 65% of sessions in Phases 1 and 2, and overall agreement for SIB and crying averaged 96% and 92%, respectively. Reliability was assessed dur- ing 82% of the sessions (22% of all intervals) in Phase 3, and overall agreement for SIB averaged 91%.


Procedures Prior to treatment, an attempt was made to

identify the functional properties of Jerry’s SIB. A series of conditions based on the research of Iwata,



Dorsey, Slifer, Bauman, and Richman (1982) was presented in a multielement format. Briefly, these were (a) attention-toys were available for Jerry’s use, and the experimenter ignored Jerry except to deliver reprimands and/or statements of concern contingent on SIB; (b) demand-the experimenter presented learning trials to Jerry with a brief time- out contingent on SIB; (c) alone-Jerry was ob- served while alone in a room without access to materials; and (d) play-the experimenter provid- ed opportunities for toy play and attention contin- gent on the absence of SIB. Based on informal observations conducted at the time of Jerry’s ad- mission, a fifth condition, alone with toys, was added to his assessment in order to separate the potential influences of social versus nonsocial stim- ulation, both of which were present in the play condition. During alone with toys, Jerry had access to a video game and several other play materials.

All sessions were scheduled to last 15 min, but one was terminated after 2 min because Jerry’s scratching focused on an open wound that was judged to be extensive by his physician.

Results and Discussion Figure 1, during the initial baseline, shows data

obtained across the five assessment conditions. Cry- ing (lower panel) was not observed during any of the sessions. The results for SIB (upper panel) were quite dear, in that scratching was almost continuous when Jerry was left completely alone; by contrast, no SIB occurred during the other conditions. These results suggest that (a) neither contingent attention (attention condition) nor contingent escape (de- mand condition) were reinforcers for Jerry’s SIB; (b) numerous forms of stimulation (e.g., social, nonsocial, or even academic tasks) served as effec- tive distractors for SIB by providing opportunities for competing behavior; and (c) SIB occurred only when these sources of stimulation were unavailable. By default, then, Jerry’s SIB was considered a form of self-stimulation, which usually implies some un- identified source of positive reinforcement associ- ated with the behavior itself. It was equally possible, however, that the scratching was negatively rein- forced through temporary alleviation of some un-

known, physiologically irritating condition. Auto- matic reinforcement or stereotypic SIB are therefore more accurate functional descriptions of his SIB (see Iwata, Vollmer, & Zarcone, 1990, for further darification). Two approaches for treating stereotypic SIB have

been described in the literature. One-sensory ex- tinction-involves attenuation of response-pro- duced stimulation (Dorsey, Iwata, Reid, & Davis, 1982; Rincover & Devaney, 1982). This interven- tion was impossible to implement procedurally due to the extent and nature of Jerry’s SIB. His body was covered with open lesions, some of which were quite large (e.g., one extended across the entire back of his neck from his hairline to his shoulders). The number and severity of these injuries precluded the possibility of “attenuating” stimulation due to scratching and rubbing at all potential locations. Even if it were possible to do so, it is unlikely that the procedure would have suppressed Jerry’s SIB, because he was frequendy observed to rub two heavily bandaged-wrapped parts of his body against each other (e.g., his arm against the back of his neck). A second approach, the development of alter-

native play behaviors that provide stimulation sim- ilar to that derived from the SIB (Favell et al., 1982), apparently was neither necessary nor suffi- dent for Jerry because the assessment data already suggested that almost any alternative behavior (not just those providing similar types of stimulation) might successfully compete with Jerry’s SIB. Thus, the mere availability of competing activities (e.g., toys) would be an effective intervention in a con- trolled demonstration of short duration (as it was during assessment), but continuous access to ma- terials that would maintain their reinforcing value indefinitely could not be guaranteed. Jerry’s pre- vious caretakers reported that they had tried this approach in the past with some success, but that it was impossible to ensure his continued partici- pation in an activity without continuous one-on- one supervision. Thus, the criterion for successful treatment was elimination of SIB at times when he would be left unattended with relatively little to do.




1001 BASELINE BL 110


C) -JUJ w –












w H.

20 40


Cn w

z20 –


z w -J z

10 ° C) w 0n

, 0


SESSIONS Figure 1. Session length and percentage 10-s intervals of SIB (upper panel), and percentage 10-s intervals of crying

(lower panel) across experimental conditions during Phase 1 (assessment) and Phase 2 (initial treatment).

As an alternative to the above strategies, we evaluated the effects of differential reinforcement of other behavior (DRO) as treatment for Jerry’s SIB. If successfal, the DRO could be used to bridge time gaps between activities and in situations in

which activities were unavailable or had lost their inherent reinforcing value. Because Jerry had good verbal skills, responded well to a wide range of potential reinforcers, and could describe the passage of time, we designed a treatment program based



on an escalating DRO schedule, which was me- diated with token reinforcement.



Procedures Treatment effects were elevated in a reversal de-

sign. Al sessions were conducted during the alone condition, while Jerry sat in a therapy room un- attended and without access to any play materials. At the beginning of each session, the experimenter told Jerry that she “had to leave the room for a while,” and asked him not to scratch while she was gone. During baseline sessions, the experimenter then left the room, returned at the end of the session, and brought Jerry to his next scheduled activity. Data obtained during the alone condition of Phase 1 served as the initial baseline.

At the beginning of the first treatment session, the experimenter gave the same instruction that she had during baseline. She also noted that she would give Jerry a penny upon returning if he did not scratch. The experimenter then left the room and watched him through the observation window. She returned at the end of the interval, briefly examined Jerry to “see if he had scratched,” and praised him and gave him a penny if no SIB had occurred. If SIB had occurred, the experimenter indicated in a nonpunitive manner that Jerry had scratched while she was gone (it was possible to identify the area because it had been observed directly), regretted that he had not earned the penny, and asked him to try again. This sequence was repeated two more times, resulting in three back-to-back DRO inter- vals per session. At the completion of each session, Jerry was allowed to exchange earned pennies for access to TV, snacks, video games, and other play materials. The DRO interval was set initially at 2 min (the longest amount of time we had previously seen Jerry refrain from scratching while left alone) and was expanded to 4 min by the end of the first treatment condition by adding 1 min to one or more intervals, resulting in session lengths ranging

from 6 min (three 2-min intervals) to 12 min (three 4-min intervals).

At the beginning of sessions during the reversal condition, the experimenter explained that she did not have any pennies but asked Jerry to refrain from scratching nonetheless. Because Jerry’s pre- vious scratching had done extensive damage and because session length had not been held constant, the duration of sessions during this condition was limited to 2, 6, and 8 min. When treatment was reinstated, session length

was gradually increased from 6 to 18 min, generally in 1-min increments. During this time, the number of pennies Jerry could earn increased to five per session (one penny for each of five successfully com- pleted DRO intervals), where it remained constant. Also, to increase the likelihood that Jerry would refrain from SIB during all intervals, he was given a bonus nickel if he earned all five pennies during a session. When session length reached 18 min (consisting of three 4-min intervals and two 3-min intervals), a probe was conducted to determine the necessity of continuing to increase interval length in 1-min increments and of delivering pennies at the end of each interval. Session 61 consisted of a single 15-min DRO interval in which Jerry could earn 104 (five pennies plus the bonus nickel). Dur- ing subsequent sessions, the interval was increased to 25 and then to 30 min.

Results and Discussion Figure 1 (upper panel) shows data on Jerry’s

SIB, as well as on session length, during the two DRO conditions and brief baseline reversal. During the first treatment condition, SIB immediately de- creased to zero from a baseline (alone condition) average of 78% and remained generally low as session length was extended from 6 to 12 min. Jerry’s scratching quickly returned to near-baseline levels during the brief reversal and decreased again when treatment was reinstated. Across subsequent treatment sessions, there were periodic occurrences of SIB, but the behavior gradually decreased to zero as session length and the duration of DRO intervals were extended. Jerry exhibited no SIB during the 15-min DRO interval in Session 61. A



few sessions later, we conduded Phase 2 at the 30- min interval length, which far exceeded the amount of time we considered reasonable for parents or caretakers to leave Jerry unattended.

The lower panel of Figure 1 presents data on Jerry’s crying and related disruptive behaviors. By comparing the upper and lower panels, it can be seen that crying was not a correlate of SIB, because crying was not observed during either baseline con- dition. Crying or disruption also was not observed during the first three sessions of the initial DRO condition (Sessions 16-18). During the next session (19), scratching first occurred during the DRO condition (upper panel). When told at the end of the DRO interval that he would not receive a penny, Jerry became visibly upset and began crying (lower panel). In subsequent sessions, crying was observed only when SIB occurred during a treat- ment condition. These results indicate that crying was a function of nondelivery of reinforcement- only when Jerry scratched and failed to earn a penny did he exhibit any negative “emotional” behaviors.

The other behaviors that Jerry exhibited during treatment sessions were quite varied, and we did not take any systematic data on their occurrence. Some of the more frequent behaviors induded sing- ing, humming, walking around the therapy room, rocking slowly while in a kneeling position with his hands folded between his knees, and sitting quietly either watching the therapy room door or with his eyes dosed.


Procedures By the end of Phase 2, Jerry’s SIB was under

very good control during treatment sessions but remained generally high throughout the day. Much of his time was spent in a group activity area, where 2 to 6 clients and 1 to 4 staff participated in a variety of leisure activities, instructional sessions, or free time. Baseline observations were begun, using the 5-min interval procedure described previously, and lasted the entire time that Jerry was present in

the area. After seven sessions, the token reinforce- ment procedure was initiated in the activity area, beginning with a single 30-min DRO interval, which was measured with a timer. If, when the timer went off, no SIB had occurred during the previous half hour, a staff member praised Jerry for not scratching and gave him a penny. The occurrence of SIB during an interval resulted in resetting the timer. During subsequent sessions, the amount of time in treatment was increased (and, consequently, the amount of time in baseline was decreased) by introducing additional 30-min DRO intervals in an alternating manner during the morn- ing and afternoon. When the program was in effect for an hour each morning and afternoon (Session 16), the penny contingency was supplemented by giving Jerry a nickel at the end of each morning and afternoon session if no SIB had occurred. By Session 25, the program was in effect during the entire time Jerry was present in the activity area. The money he earned could be exchanged imme- diately for access to the same reinforcers that were available in Phase 2 of the study, or it could be saved for special outings (e.g., to a restaurant).

During Phase 3, we also attempted to determine whether token reinforcement was a necessary com- ponent of the DRO program. Such procedures are generally complicated to implement and maintain; ifJerry’s SIB could be managed at a low level with social reinforcement alone, transfer of the program to his home environment might be achieved more easily. During Sessions 10 through 14, at desig- nated times when neither baseline nor the token reinforcement program was in effect, social rein- forcement (e.g., praise, hugs, etc.) was provided for the absence of SIB on a DRO 5-min schedule.

In summary, Phase 3 involved the presentation of one, two, or three conditions daily for varying amounts of time in a multielement design.

Results and Discussion Figure 2 shows the percentage of 5-min intervals

during which Jerry exhibited SIB, as well as the number of hours per day spent in each of the three c