JOURNAL OF APPLIED BEHAVIOR ANALYSIS

REDUCING ESCAPE BEHAVIOR AND INCREASING TASKCOMPLETION WITH FUNCTIONAL COMMUNICATION TRAINING,

EXTINCTION, AND RESPONSE CHAININGJOSEPH S. LALLI

THE UNIVERSITY OF PENNSYLVANIA

AND

SEAN CASEY AND KELLY KATESCHILDREN'S SEASHORE HOUSE

The effects of functional communication training, extinction, and response chaining on 3 sub-jects' escape-maintained aberrant behavior were evaluated using a multielement design. Func-tional communication training consisted of teaching subjects a verbal response that was func-tionally equivalent to their aberrant behavior. Subjects initially were allowed to escape from atask contingent on the trained verbal response. In subsequent treatment phases, escape wascontingent on the trained verbal response plus the completion of the specified number of stepsin the task (response chaining). The number of steps was increased until a subject completedthe task to obtain a break. Results showed that the treatment reduced rates of aberrant behaviorand that the chaining procedure was effective in decreasing the availability of escape.DESCRIPTORS: extinction, functional communication training, response chaining, escape

behavior, task completion

In the treatment of aberrant behavior, func-tional communication training (FCT) is a two-step process. First, the therapist identifies thetarget behavior's operant function. Second, thetherapist reinforces an alternative response withthe same consequence as that produced by theaberrant behavior. FCT thus weakens the ab-errant response-reinforcer relation by providingthe reinforcer for inappropriate behavior con-tingent on the alternative response (Carr,1988). Researchers have used FCT as part of atreatment package to reduce aberrant behaviormaintained by attention (Carr & Durand,1985; Durand & Carr, 1991; Lalli, Browder,Mace, & Brown, 1993), escape (Bird, Dores,Moniz, & Robinson, 1989; Carr & Durand,1985; Durand & Carr, 1991; Fisher et al.,1993; Wacker et al., 1990), or access to pre-

We thank the reviewers for their helpful comments onearlier drafts of this manuscript.

Reprints may be obtained from Joseph S. Lalli, Chil-dren's Seashore House, 3405 Civic Center Blvd., Phila-delphia, Pennsylvania 19104.

ferred items (Bird et al., 1989; Fisher et al.,1993; Wacker et al., 1990).One potential benefit of FCT is that an in-

dividual can regulate delivery of the reinforcerto a greater degree when compared to treat-ments that rely solely on the passage of time(e.g., response-independent reinforcement),time plus the absence of aberrant behavior (e.g.,differential reinforcement of other behavior;DRO), or a predetermined performance re-quirement (e.g., task completion). However,this benefit is compromised to the extent thatthe reinforcer is not available immediately (e.g.,a snack before a meal) or may not be in thebest interest of the individual (e.g., escape fromself-care routines). For example, in the treat-ment of escape-maintained behavior, therapistshave sometimes terminated an ongoing task im-mediately when the subject emitted the trainedescape response (Bird et al., 1989; Fisher et al.,1993; Wacker et al., 1990). Thus, althoughFCT may result in decreased rates of aberrantbehavior, it may also disrupt an ongoing activityand limit teaching opportunities. For FCT to

261

1995, 282 261-268 NUMBER3 (FALL 1995)

JOSEPH S. LALLI et al.

be a viable treatment for escape-maintained be-havior, therapists must use it in a way that canmaintain treatment effects while improving par-ticipation in the required activity.To address this concern, Bird et al. (1989)

and Fisher et al. (1993) gradually increased theresponse requirements before their subjectscould escape with the trained verbal response.Bird et al. increased the number of task stepsrequired for completion before a subject couldexchange a token for a 60-s break; the initialincrease occurred after 6 weeks of FCT training,and subsequent increases occurred at 3-week in-tervals. Fisher et al. gradually increased thenumber of instructional requests presented to asubject from 2 to 26 before signing produced a30-s break. Both studies successfully maintainedlow rates of aberrant behavior as performancerequirements increased. However, neither studyprovided detailed descriptions of the proce-dures.

In the present study, we extended this line ofresearch by evaluating the effects of responsechaining following the implementation of FCT.We taught subjects an alternative escape re-sponse and then added a response chaining pro-cedure to increase their participation in thetask. We combined these treatment procedureswith extinction throughout the study. Our ob-jective was to first teach a subject to escape task-related activities with a socially appropriate re-sponse, and then to teach a subject to toleratea delay (i.e., an increase in the performance re-quirement) prior to obtaining a break from thetask.

METHOD

Subjects and SettingThree individuals admitted to a hospital in-

patient unit specializing in the treatment of se-vere problem behavior participated. All were en-rolled in the unit's special education programthroughout their admission. The teacher re-ported that each subject engaged in high rates

of aberrant behavior during instructional activ-ities. Subjects' admissions were sequential.

Joe was 10 years old with moderate mentalretardation and was admitted for treatment ofhis self-injurious behavior. Joe interacted withothers using utterances and responded to hisname and some words. He was ambulatory, fol-lowed a few one-step directions, and could feedhimself but he required physical prompting tocomplete his other self-care activities. Jen was15 years old with moderate mental retardationand autism, and she was admitted for treatmentof her self-injury. She interacted with others us-ing one-word utterances and responded to fa-miliar names, objects, and activities. Jen wasambulatory, followed one-step directions, andindependently completed self-care activities.Kim was 13 years old with moderate mentalretardation and autism, and was admitted fortreatment of her aggressive behavior. She inter-acted with others using utterances and gestures(pointing), and responded to her name and oth-er familiar names, objects, and activities. Kimwas ambulatory, followed one-step directions,and demonstrated adequate fine and gross mo-tor responses for participation in self-care rou-tines.

All sessions were conducted individually inone of two therapy rooms (3 m by 3 m), andobservers recorded data from behind a one-waymirror. A therapist and subject were presentduring sessions. Two 1 5-min sessions were con-ducted daily 5 days per week, with a minimumof 120 min between sessions.

Dependent Variables and Data CollectionSelf-injury was defined as forceful contact of

one's head or mouth against an object Joe) orclosure of upper and lower teeth on one's handden). Aggression was defined as forceful hittingor throwing objects at others (Kim). Appropriateverbalization was defined as a subject indepen-dently giving the therapist an index card with"BREAK" Joe), saying "no" Jen), or movingone's head horizontally to indicate "no" in re-sponse to a therapist's request (Kim). In the

262

TREA TMENT OF ESCAPE BEHAVIOR

FCT plus extinction with response chainingphases, appropriate verbalizations were definedas the subject independently requesting a breakafter completing the task requirement (i.e., ona fixed-ratio, FR, schedule). Compliance wasdefined as independently initiating a task within10 s of the therapist's request. Data were alsocollected on the therapist's instructions, praise,and withdrawal of instructional materials tomonitor procedural fidelity.

Observers used a computerized event-record-ing procedure for all topographies (Repp, Har-man, Felce, VanAcker, & Karsh, 1989). Weconverted subjects' appropriate verbalizationsand compliance to percentage of requests re-sponded to independently with the respectiveresponses. A second observer independently col-lected data during an average of 30% of thesessions, equally distributed across all phasesand subjects. Interobserver agreement was de-termined using the "reliable" program (Repp etal., 1989). Occurrence agreement was scoredwhen two observers recorded the onset of a tar-get behavior within 2 s of each other. Occur-rence agreement averaged 88% (range, 83% to97%) across topographies, phases of the study,and subjects. Procedural fidelity data showedthat the therapist correctly carried out the pro-cedures on an average of 93% of the opportu-nities across all subjects.

Experimental Designs and SequenceThe teacher reported that the subjects en-

gaged in aberrant behavior exclusively duringinstructional activities. Therefore, we initiallyassessed the subjects' aberrant behavior via afunctional analysis to verify the hypothesized es-cape function. Conditions were presented dur-ing 15-min sessions in a multielement design.The second phase of the study began with an

evaluation of FCT plus extinction using a mul-tielement design in which FCT plus extinctionconditions alternated with baseline conditions.We paired each condition with a different ther-apist and setting to help subjects discriminatethe conditions in effect. Subsequent treatment

phases (FCT plus extinction with responsechaining) consisted of gradually increasing theresponse requirement prior to obtaining escapefrom the task.

Functional AnalysisWe assessed the subjects' aberrant behavior

under demand and play conditions, based onprocedures described by Iwata, Dorsey, Slifer,Bauman, and Richman (1982/1994). We se-lected tasks from the subjects' educational plansand developed 16-step task analyses for usethroughout the study. During the demand con-dition, the therapist presented an instruction tothe subject once every 30 s (approximately 16per session), provided descriptive praise for cor-rect responses, and provided a 30-s time-outfrom the task (escape) contingent on the oc-currence of aberrant behavior. During the playcondition, the therapist provided the subject ac-cess to preferred items, delivered neutral com-ments every 30 s, and did not respond to ab-errant behavior.

Treatment ConditionsBaseline. Baseline conditions during the FCT

plus extinction evaluation were the same asthose of the functional analysis demand condi-tion (described above) with one exception. Dur-ing the baseline condition, the therapist provid-ed descriptive praise contingent on the trainedverbal response ("no") and restated the instruc-tion. That is, trained verbal responses wereplaced on extinction during these conditions.FCTplus extinction. The therapist presented

a request to the subject as in the functionalanalysis demand condition. However, the ther-apist also said, "If you do not want to worknow, say 'no' [or the other verbal responses]."The therapist allowed the subject 10 s to emitthe trained verbal response independently. Con-tingent on noncompliance, the therapist provid-ed either physical prompts Joe) or modeledprompts den and Kim) for the appropriate ver-bal response. If aberrant behavior occurred dur-ing a trial, the therapist waited 10 s after it

263

JOSEPH S. LALLI et al.

ceased, then prompted the verbal response. Thetherapist provided descriptive praise and a 30-stime-out from the task contingent on the verbalresponse (independent or prompted). In thismanner, the subject escaped each request con-tingent on the verbal response.FCT plus extinction with response chaining.

The objective of these conditions was to grad-ually increase the response requirement beforethe subject could request a break from task de-mands. Thus, the FR designation refers to thenumber of steps in the task that the subject wasrequired to complete before asking for a breakwas reinforced (i.e., FR 1 to FR 16). Escapewas contingent on completion of the requirednumber of steps in the task and emission of thetrained verbal response. The number of requestsper session depended on the FR schedule in ef-fect. For example, in the FR 1 phase, the ther-apist presented 16 requests, with each requestcorresponding to one step in the task analysis,with a break allowable following the completionof each step. In this manner, the subject com-pleted the entire task by the end of the session.In subsequent treatment phases, the therapistpresented progressively fewer requests based onthe FR schedule in effect (e.g., FR 2, eight re-quests; FR 4, four requests), and the subjectreceived correspondingly fewer breaks. Thisprocedure continued until the therapist pre-sented one request per session (i.e., FR 16),which required the subject to complete the en-tire task before requesting a break.

During the response chaining phase, thetherapist provided a request to the subject, stat-ed the criterion for earning a break, and pointedto the instructional materials related to thespecified criterion. Contingent on the verbal re-sponse, the therapist said, "Good saying no, butyou have to do [task step] then you can ask fora break." Contingent on noncompliance, thetherapist physically guided the subject to emitthe task-related response. The therapist praisedcompliance, provided the level of assistance nec-essary (i.e., least-to-most prompt hierarchy)during the task, and did not respond to aber-

rant behavior that occurred during the task. Ifthe subject did not ask for a break within 10 sof completing the criterion, the therapist said"Good, you finished. Do you want morework?" If aberrant behavior occurred after thesubject completed the criterion, the therapistwaited 10 s after it ceased, then repeated theabove statement. The therapist immediatelyprovided a 30-s break contingent on indepen-dent or prompted target verbal responses. Weinitially established the criterion for a break ascompletion of one step of the task (FR 1) andsubsequently doubled the criterion for the nextphase after three consecutive sessions withoutaberrant behavior.

RESULTSFunctional Analysis

Response patterns were similar for the 3 sub-jects; that is, aberrant behavior occurred exclu-sively in the demand condition (Figure 1).These findings suggested that aberrant behav-iors were maintained by escape.

FCTplus ExtinctionResults presented in Figure 2 show lower

rates of aberrant behavior during the FCT plusextinction condition compared with the base-line condition for each subject. Joe's data showa gradual decrease in rates of self-injury, whicheventually reached zero. By contrast, Jen's andKim's data show more rapid reduction in ratesof aberrant behavior. During the FCT plus ex-tinction condition, Joe's independent verbaliza-tions gradually increased from 0% to 69%, Jen'sincreased from 38% to 81%, and Kim's in-creased from 25% to 88% of the therapist's re-quests per session. Joe's independent verbaliza-tions averaged 3% during the baseline condi-tion (not graphically presented), and Jen andKim never emitted the verbal response duringbaseline conditions. The subjects' task-relatedcompliance in the FCT plus extinction condi-tions was zero, because escape was contingenton the trained verbal response. Task-related

264

TREATMENT OF ESCAPE BEHAVIOR

2.8-

2.4'

2.0-

1.6-

1.2-

0.8'

wI-0

0

050.4'

0.0'

Functional Analysis

Demand

Play

0 O2

2.8'

2.4'

w 2.0'I--

z 16'

= 1.2

X 0.8'0.4

0.0'2

zs

z

C,

a,

0.9-

0.6-

0.3-

0.0

1 2

SESSIONSFigure 1. Responses per minute across asse

ditions.

compliance in the baseline condition averaged3% for Joe, 13% for Jen, and 17% for Kim(not graphically presented).

FCT Plus Extinction withResponse ChainingWe observed low rates of aberrant behavior

for each subject during the initial FCT plus ex-tinction with response chaining condition (FR

rJOE i 1, Figure 2). The subjects' rates of aberrant be-havior stayed low throughout the remaining

. phases when performance requirements for ob-tamning a break were increased. Interestingly, we

3 observed temporary increases in rates of aber-Y rant behavior in each chaining phase for Jen

and Kim and for Joe in the first and thirdchaining phases. However, the subjects' rates ofaberrant behavior quickly decreased and re-mained stable throughout the phases.

In the response chaining phases, subjects hadthe opportunity to independently request abreak following completion of the FR schedule.

JEN For example, in the FR 1 phase, subjects had16 opportunities (i.e., following each completed

=v request), considered as trials, to request a break3 independently. The number of trials decreased

proportionately across subsequent FR phases.Therefore, data on appropriate verbalizations inthe response chaining phases reflect the per-centage of trials in which the subjects indepen-dently requested a break following completionof the FR schedule. Subjects always respondedto the therapist's statement "Do you want morework?" by saying "no." However, these were

KIM considered to be prompted responses and arenot graphically presented. Joe's independent

° verbalizations showed a gradual increase across3 response chaining phases before eventually av-

eraging 100% in the FR 16 phase. By contrast,Jen's and Kim's independent verbalizations

ssment conl showed a more rapid increase. Jen's and Kim'sindependent verbalizations averaged 100% bythe fifth session of the FR 2 phase and the thirdsession of the FR 4 phase, respectively.

During the response chaining phases, we ob-served an increase in subjects' independent

265

1

0 -0I

0

0 -0

JOSEPH S. LALLI et al.

r - - FCT + EXT (WITH RESPONSEI (FRI(FR2)2) (FR4)

2.8 I 100

-W-BL(SIB)

2.4

2.0 4

941 5 1 1 t 601.2

I4 0p4010.8 1 2 1

0.4- FCT + EXT I

FCTI (TR) )0.0 (SIB)

4 812 16 20 24 2MULTIELEMENT --- FCT + EXT (WITH RESPONSE CHAINING -----I c

3. ALUATION (Ri) (FR2) (FR4) (FRO) (FR16) 1

3.0

80

2.5 I cUj >~IFigure 2.0RM LU~~~~

40IIL1.0 0

~~~~~~~~~~~~~~~~~Nz20 UZ

0.0 Jj0E0.5

12 18 24 30 36 54 a2.0 -MULTIELEMENT r -- FCT + EXT (WITH RESPONSE CHAINING) A-----

=fncioalcomuictinr ;(FR1) (FR (FR4) (FR8) (FR16)EVALUATION100

1.6-

1.2

~~~~~~~~~~~60

0.840

0Q4- iKIM

0.0

1 4 8 1 2 1 6 20 24 28 32 36

SESSIONS

Figure 2. Responses per minute of aberrant behavior and the trained verbal response during the multielement

evaluation and FCT plus extinction with response chaining. ALT-R = appropriate verbalization; EXT = extinction;

FCT = functional communication training; FR through FR 16 = task-related criterion for break across FCT plusextinction with response chaining phases.

266

TREATMENT OF ESCAPE BEHAVIOR

compliance with requests compared with base-line. During baseline, compliance was low andvariable. Joe's compliance averaged 20%, 46%,81%, 83%, and 100% across the responsechaining phases (i.e., FR 1 to FR 16). Jen'scompliance averaged 48%, 63%, 77%, 85%,and 100% across the response chaining phases.Kim's compliance averaged 32%, 58%, 70%,80%, and 80% during response chaining con-ditions.

DISCUSSIONResults of this study extend previous work on

the effectiveness of FCT plus extinction (Carr& Durand, 1985; Durand & Carr, 1991). Dur-ing the FCT plus extinction conditions (withand without response chaining), we observed aninverse relationship between subjects' appropri-ate verbal responses and their aberrant behavior(i.e., increased verbal responses and decreasedaberrant behavior). Low rates of aberrant be-havior were maintained throughout the re-sponse chaining phases when the requirementsfor obtaining a break were increased. In addi-tion, we observed gradual increases in compli-ance with requests throughout the study. Thesefindings suggest that combining FCT plus ex-tinction with response chaining in this mannermay be an effective way to maintain low ratesof aberrant behavior while increasing subjects'participation in the task.

Fisher et al. (1993) recently reported thatFCT plus extinction failed to reduce a subject'srate of escape-maintained aberrant behavior andspeculated that treatment effects may be ham-pered by the temporal proximity between theappropriate and aberrant responses. That is, theappropriate and aberrant behaviors may haveformed a response chain. We addressed thisproblem by requiring the absence of aberrantbehavior for 10 s before providing a break con-tingent on the trained verbal response. A secondconcern reported by Fisher et al. was the relativereinforcement histories for the appropriate andaberrant responses. That is, the appropriate re-

sponse probably has a shorter reinforcementhistory than the aberrant behavior. To addressthis problem, we started functional communi-cation training immediately upon Jen's andKim's admission to the hospital program. Weinitially taught Jen and Kim the appropriateverbal response by presenting them with non-preferred items and telling them to say "no" ifthey did not want the item. We started func-tional communication training for Joe duringthe first FCT plus extinction phase. Interesting-ly, a comparison of the FCT plus extinctiondata showed that decreases in aberrant behaviorand increases in independent verbalizationswere obtained more quickly with Jen and Kimthan with Joe. We suggest that differences inresponding among the subjects might be a func-tion of their individual histories with functionalcommunication training.A major focus of the present study was an

evaluation of a response-chaining component asa means of increasing compliance with task de-mands as well as reducing the availability of es-cape. Although FCT plus extinction reducedrates of aberrant behavior initially, subjects didnot participate in the task because escape wascontingent only on the trained verbal response.This highlights a potential limitation of FCT-allowing an individual to escape from a requiredinstructional activity (even if contingent on asocially appropriate response). To address thisproblem, we started treatment sessions with in-structional requests and provided a break fromthe task contingent on compliance with one re-quest (i.e., FR 1) and the trained verbal re-sponse. We doubled the criterion for a breakacross phases, with the criterion in the finalphase requiring the subject to complete the en-tire task before earning a break. In this manner,subjects eventually completed the task in itsnatural progression. FCT plus extinction withresponse chaining gradually increased compli-ance with requests and maintained low rates ofaberrant behavior.

In the present study, initial treatment effectsmay have been a function not only of FCT and

267

268 JOSEPH S. LALLI et al.

extinction but also of altering the establishingoperation for escape. That is, by initially reduc-ing the requirement for a break, the subjects'motivation to engage in escape behavior mayalso have decreased. Subsequently, the effective-ness of the response-chaining component maybe attributed either to the gradually increasingresponse requirement (similar to that describedby Zarcone, Iwata, Smith, Mazaleski, & Ler-man, 1994) or to the fact that the instructionalprocedures (therapist's instructions, task consis-tency, and materials) increased the predictabilityof the situation (Flannery & Horner, 1994; Lal-li, Casey, Goh, & Merlino, 1994).

N.,

REFERENCESBird, F., Dores, P. A., Moniz, D., & Robinson, J. (1989).

Reducing severe aggressive and self-injurious behav-iors with functional communication training. Amen-can Journal on Mental Retardation, 94, 37-48.

Carr, E. G. (1988). Functional equivalence as a mecha-nism of response generalization. In R Homer, R.Koegel, & G. Dunlap (Eds.), Generalization andmaintenance: Life-style changes in applied settings (pp.221-241). Baltimore: Paul H. Brookes.

Carr, E. G., & Durand, V. M. (1985). Reducing behaviorproblem through functional communication training.Journal ofApplied Behavior Analysis, 18, 111-126.

Durand, V. M., & Carr, E. G. (1991). Functional com-munication training to reduce challenging behavior:Maintenance and application to new settings. JournalofApplied Behavior Analysis, 24, 251-264.

Fisher, W., Piazza, C., Cataldo, M., Harrell, R., Jefferson,G., & Conner, R. (1993). Functional communica-

tion training with and without extinction and pun-ishment. Journal ofApplied Behavior Analysis, 26, 23-36.

Flannery, K. B., & Horner, R H. (1994). The relation-ship between predictability and problem behavior forstudents with severe disabilities. Journal ofBehavioralEducation, 4, 157-176.

Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K E.,& Richman, G. S. (1994). Toward a functional anal-ysis of self-injury. Journal ofApplied Behavior Analysis,27, 197-209. (Reprinted from Analysis and Interven-tion in Developmental Disabilities, 2, 3-20)

Lalli, J. S., Browder, D. M., Mace, F. C., & Brown, D.K (1993). Teacher use of descriptive analysis data toimplement interventions to decrease students' prob-lem behaviors. Journal of Applied Behavior Analysis,26, 227-238.

Lalli, J. S., Casey, S., Goh, H., & Merlino, J. (1994).Treatment of escape-maintained aberrant behaviorwith escape extinction and predictable routines. Jour-nal ofApplied Behavior Analysis, 27, 705-714.

Repp, A. C., Harman, M. L., Felce, D., VanAcker, R., &Karsh, K L. (1989). Conducting behavioral assess-ment on computer collected data. Behavioral Assess-ment, 2, 249-268.

Wacker, D. P., Steege, M. W., Northup, J., Sasso, G., Berg,W., Reimers, T., Cooper, L., Cigrand, K., & Donn,L. (1990). A component analysis of functional com-munication training across three topographies of se-vere behavior problems. Journal ofApplied BehaviorAnalysis, 23, 417-429.

Zarcone, J. R., Iwata, B. A., Smith, R. G., Mazaleski, J.L., & Lerman, D. C. (1994). Reemergence and ex-tinction of self-injurious escape behavior during stim-ulus (instructional) fading. Journal ofApplied BehaviorAnalysis, 27, 307-316.

Received August 15, 1994Initial editorial decision October 14, 1994Revisions received February 16, 1995; March 20, 1995Final acceptance April 20, 1995Action Editor, Brian Iwata