further evaluation of leisure items in the attention condition of functional analyses

FURTHER EVALUATION OF LEISURE ITEMS IN THE ATTENTIONCONDITION OF FUNCTIONAL ANALYSES

EILEEN M. ROSCOE, ABBEY CARREAU, JACKIE MACDONALD, AND SACHA T. PENCE

NEW ENGLAND CENTER FOR CHILDREN

NORTHEASTERN UNIVERSITY

Research suggests that including leisure items in the attention condition of a functional analysismay produce engagement that masks sensitivity to attention. In this study, 4 individuals’ initialfunctional analyses indicated that behavior was maintained by nonsocial variables (n 5 3) or byattention (n 5 1). A preference assessment was used to identify items for subsequent functionalanalyses. Four conditions were compared, attention with and without leisure items and controlwith and without leisure items. Following this, either high- or low-preference items wereincluded in the attention condition. Problem behavior was more probable during the attentioncondition when no leisure items or low-preference items were included, and lower levels ofproblem behavior were observed during the attention condition when high-preference leisureitems were included. These findings suggest how preferred items may hinder detection ofbehavioral function.

DESCRIPTORS: assessment, autism, functional analysis, problem behavior

_______________________________________________________________________________

The functional analysis model proposed byIwata, Dorsey, Slifer, Bauman, and Richman(1982/1994) has had a substantial impact onthe assessment and treatment of problembehavior. By determining the contingenciesthat maintain problem behavior, functionalanalyses allow the development of effectivefunction-based treatments (Iwata et al., 1994).Since the advent of this procedure, several directand systematic replications across populations,settings, and topographies have been reported(Hanley, Iwata, & McCord, 2003). In thisgeneral model, test conditions that contain anestablishing operation (Michael, 1982), a dis-criminative stimulus, and a source of reinforce-ment are alternated with a control condition(that removes the establishing operation andconsequences present in the test conditions) toidentify the maintaining variables for problembehavior. Data are typically interpreted throughvisual inspection, and differentiated outcomesare determined based on levels of problem

behavior observed during each test conditionrelative to the control condition. For example,differentially higher levels of behavior duringthe attention condition would suggest mainte-nance by social-positive reinforcement. Whendifferentiated outcomes are obtained, the rele-vant consequence can be altered as treatment forthe problem behavior.

Although functional analyses have beenreported to frequently produce differentiatedoutcomes (Derby et al., 1992; Iwata et al.,1994; Vollmer, Marcus, Ringdahl, & Roane,1995), unclear outcomes (e.g., variable and lowlevels of behavior across conditions) have beenobserved, reducing the likelihood of identifyinga maintaining variable. To enhance the likeli-hood of obtaining informative results, a numberof procedural refinements have been developed.One such adaptation involves manipulatingspecific aspects of antecedent events that mayaffect functional analysis outcomes (Berg et al.,2000; O’Reilly et al., 2006; Smith, Iwata, Goh,& Shore, 1995).

For example, O’Reilly (1999) observedhigher levels of problem behavior during anattention condition following a period in whichaccess to attention was restricted than when

Address correspondence to Eileen M. Roscoe, Assis-tant Director of Research, New England Center forChildren, 33 Turnpike Rd., Southborough, Massachu-setts 01772 (e-mail: [email protected]).

doi: 10.1901/jaba.2008.41-351

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2008, 41, 351–364 NUMBER 3 (FALL 2008)

351

attention was continuously presented prior to asession. Smith et al. (1995) examined the effectsof manipulating antecedent events associatedwith the demand condition, including tasknovelty, session duration, and rate of taskpresentation. Results showed that presentationof novel demands, increases in session length,and increases in the rate of task presentationresulted in increases in escape-maintainedbehavior for some of the participants.

Another variable that may affect differentia-tion in functional analysis outcomes is theinclusion of leisure items during the attentioncondition. Although leisure items are typicallyincluded during the attention condition todecrease the likelihood of observing automati-cally reinforced problem behavior, the inclusionof preferred items may mask detection ofattention as the maintaining reinforcer. A studythat raised this possibility was conducted byVollmer et al. (1995), who found clearfunctional analysis outcomes for 17 of 20individuals referred for assessment and treat-ment of severe behavior problems. For the 17individuals for whom a clear outcome wasobtained, attention was never identified as amaintaining variable. The authors noted thatone explanation for this finding was that theyincluded preferred stimuli (based on the resultsof a preference assessment) during the attentioncondition, and that these items may havecompeted with attention as a source of positivereinforcement. It is possible that attentionwould have served as a reinforcer for some ofthese participants if preferred items were notavailable concurrently.

McCord and Neef (2005) reviewed articlespublished during a 10-year period (from 1994to 2003) in the Journal of Applied BehaviorAnalysis and determined how many used leisureitems during the attention condition of func-tional analyses. The authors found that 87 of165 studies (53%) included leisure items duringthe attention test condition, and of the 87studies that included leisure items, only five

(6%) reported use of a stimulus preferenceassessment to aid in item selection. Thus,although leisure items are often included inthe attention condition of a functionalanalysis, the items included are usually ofunknown preference. The authors suggestedthat although the inclusion of items duringthe attention condition minimizes the occur-rence of behavior maintained by automaticreinforcement, it is possible that the availabilityof leisure items may sometimes compete withattention-maintained behavior, masking thispotential function of problem behavior. Inaddition, the authors noted that because fewstudies have reported using preference assess-ments for item selection, the items includedduring the attention condition may have beennonpreferred, increasing the likelihood of false-positive identification of attention as a rein-forcer (i.e., behavior observed in this conditionmay have been maintained by automaticreinforcement).

Recent research more directly suggests thatthe inclusion of preferred leisure items maymask detection of attention as a reinforcer.Ringdahl, Winborn, Andelman, and Kitsukawa(2002) evaluated the effects of providingcontinuous access to preferred items duringthe attention condition. A modified functionalanalysis was conducted that included threeconditions: a control condition and two variantsof the attention condition. During the controlcondition, the therapist delivered attentionindependent of responding, preferred leisureitems were available continuously, and instancesof the target behavior were ignored. During theattention-only condition, the therapist delivered30 s of attention following instances of thetarget behavior, and no leisure items wereavailable. During the attention-enriched condi-tion, the therapist delivered 30 s of attentionfollowing each instance of the target behavior,and preferred leisure items were availablecontinuously. Results indicated higher levels ofproblem behavior during the attention-only

352 EILEEN M. ROSCOE et al.

condition than in the attention-enriched con-dition, suggesting that the availability ofpreferred leisure items in the functional analysismay have abolished or competed with attentionas a reinforcer, resulting in false-negativeoutcomes. However, it is unclear how theinclusion of preferred items would have affectedoutcomes for participants whose behavior wasmaintained by automatic reinforcement. Also,the authors did not report how preferred leisureitems were identified, and it is unclear whetherbehavior may have been sensitive to automaticreinforcement because an alone condition wasnot included.

Although research has shown that theinclusion of high-preference (HP) items in theattention condition may result in false-negativeoutcomes for attention-maintained problembehavior, it is also possible that the omissionof items during the attention condition mayresult in false-positive identification of attentionas a maintaining variable for behavior main-tained by automatic reinforcement. In addition,it is unclear how the inclusion of items ofvarying preference value may affect outcomes.The purpose of the current study was to replicateand extend findings by Ringdahl et al. (2002).First, items used during the attention conditionof the functional analysis were selected based onthe results of a duration-based preferenceassessment that involved simultaneous measure-ment of item engagement and problem behaviorsimilar to that used by Piazza, Fisher, Hanley,Hilker, and Derby (1996). Second, we evaluatedwhether the inclusion of HP items during theattention condition affected outcomes for indi-viduals with behavior maintained by attention orautomatic reinforcement based on the outcomesof previous functional analyses. Third, weassessed whether including leisure items withdifferent preference values (i.e., HP vs. low-preference [LP] items) affected functional anal-ysis outcomes for individuals with problembehavior maintained by attention or by auto-matic reinforcement.

METHOD

Participants and Setting

The participants were 2 boys and 2 girls whohad been diagnosed with an autism spectrumdisorder; they had been referred by serviceproviders as exhibiting problem behavior thatinterfered with their participation in education-al activities. All participants attended a residen-tial school for individuals with developmentaldisabilities.

Jake was an 8-year-old boy who exhibitedhand biting. His primary modes of communi-cation were through the use of a pictureexchange communication system (PECS) anda speech-generating device. Beth was a 12-year-old girl who exhibited motor stereotypy. Shecommunicated using one-word vocal utterancesor complete sentences with a speech-generatingdevice. Amy was a 9-year-old girl who exhibiteda finger tapping stereotypy. She communicatedusing one- to two-word vocal utterances andPECS. Carl was a 12-year-old boy whoexhibited shirt twirling. He communicatedusing one- to two-word utterances and PECS.All of the participants followed one- to two-stepinstructions. Carl and Beth exhibited indepen-dent self-help skills (toileting, grooming, dress-ing, feeding).

All sessions were conducted in a room (1.5 mby 3 m) and equipped with a wide-angle videocamera, a microphone, videorecording equip-ment, materials necessary to conduct theexperimental conditions, and an appropriatelysized table with one or two chairs.

Response Measurement andInterobserver Agreement

Self-Biting (Jake) was defined as putting oneor more of his fingers into his mouth and bitingdown such that his upper and lower teeth madecontact with his fingers. Motor stereotypy (Beth)was defined as repetitive movement of any or allbody parts including rocking or swaying oftorso, head, feet or body; jumping; handflapping or posturing; finger rubbing; fanning

FUNCTIONAL ANALYSIS 353

or spreading of fingers; putting fingers in orover ears, mouth, or corner of eyes. Tapping(Amy) was defined as any instance of her handsor an object in her hands coming into contactwith another surface two or more times within a2-s period. Shirt twirling (Carl) was defined asholding onto his shirt with two or more fingerswhile moving his wrist or arm in a back-and-forth motion.

Previously trained graduate students orundergraduate students served as observerswho collected data by hand using data sheets.Data on hand biting were recorded using afrequency measure during all assessments. Dataon motor stereotypy, tapping, and shirt twirlingwere recorded using 10-s momentary timesampling during all assessments. Data on itemengagement (collected during the initial prefer-ence assessment) were recorded using 10-smomentary time sampling. For every 10 s ofsession time, an observation of 2 s occurredduring which the occurrence or nonoccurrenceof the behavior was recorded. Momentary time-sampling data were converted to percentageoccurrence by dividing the number of intervalsin which an occurrence was scored by thenumber of 10-s intervals in the session.

Interobserver agreement was assessed byhaving a second observer independently recordbehavior. Agreement for frequency data wascalculated by dividing each session into consec-utive 10-s intervals. Intervals in which the samenumber was scored by both observers wereassigned a value of 1. Intervals in which oneobserver scored 0 target responses and the otherscored anything other than 0 were assigned avalue of 0. For intervals in which differentnumbers were scored, the smaller number ofresponses scored was divided by the largernumber. The mean was calculated for the valuesgenerated for each interval, and this numberwas converted to a percentage. Interobserveragreement for momentary time-sampling datawas calculated by dividing the number ofagreements by the total number of agreements

plus disagreements and converting this ratio to apercentage. During the functional analysis,agreement was measured during 37%, 56%,21%, and 46% of functional analysis sessions,and mean agreement was 97% (range, 93% to100%), 93% (range, 85% to 100%), 95%(range, 92% to 100%), and 98% (range, 95%to 100%) for Jake, Beth, Amy, and Carl,respectively. During the preference assessment,agreement was measured during 40% ofsessions, and mean agreement was 95% (range,73% to 100%) and 99% (range, 93% to 100%)for problem behavior and item engagement,respectively. During the leisure-item analysisand the preferred or nonpreferred leisure-itemanalysis, agreement was measured during 47%,34%, 31%, and 32% of sessions, and meanagreement was 96% (range, 84% to 100%),92% (range, 83% to 100%), 98% (range, 92%to 100%), and 98% (range, 97% to 100%) forJake, Beth, Amy, and Carl, respectively.

Procedure

Functional analysis. An experimental analysisbased on procedures described by Iwata et al.(1982/1994) was conducted. Three or fourconditions were arranged in a multielementdesign. For Jake, no-interaction, attention,demand, and control conditions were conduct-ed. For Beth, Amy, and Carl, alone, attention,and demand sessions were conducted with a 2:1ratio of alone to attention or demand sessions.Because it seemed likely that their stereotypywas maintained by automatic reinforcement,the play condition was omitted, and anextended series of alone sessions (similar to thatused by Vollmer et al., 1995) was conducted toverify that stereotypy persisted in the absence ofsocial consequences. However, periodic probesof attention and demand test conditions wereconducted. If behavior persisted in socialconditions and was low in the alone condition,further analyses would have been conducted todetermine whether the behavior was sociallymaintained. However, this pattern was notobserved for Beth, Amy, and Carl.


During the alone/no-interaction condition,the participant was alone in the room or at least1 m away from observers, and no materials wereavailable. No attention or other programmedconsequences were arranged.

During the attention condition, all partici-pants (except Jake) were provided with contin-uous access to preferred leisure items. At thestart of each session, the therapist stated that heor she ‘‘had work to do.’’ Contingent on theoccurrence of the target response, the therapistdelivered brief verbal and physical attention(e.g., saying ‘‘Don’t do that, you’re going tohurt yourself’’ or ‘‘Don’t do that, that doesn’tlook nice’’ paired with a light touch to theshoulder).

During the control condition, the participantwas provided with continuous access to pre-ferred leisure items. In addition, the therapistpresented brief vocal and physical attention on afixed-time 15-s schedule (or more frequently ifinitiated by the participant). No programmedconsequences were delivered for the targetbehavior.

During the demand condition, instructionswere presented continuously, using a three-stepprompting hierarchy (i.e., vocal, gestural, andphysical prompts). Contingent on compliance,the participant received brief praise (e.g., ‘‘nicework’’). Contingent on the target response, thetherapist said ‘‘you don’t have to,’’ delivered noinstruction or task materials, and turned awayfrom the participant for 15 s.

Preference assessment. Prior to conducting theleisure-item and preference analyses, a prefer-ence assessment was conducted for each partic-ipant to identify HP and LP leisure items forinclusion in the attention and control condi-tions. A duration-based preference assessmentsimilar to that described by Piazza et al. (1996)was used. Each item was singly presented, andthe participant had continuous access to theitem during four 3-min sessions (Beth) orduring two 5-min sessions (all other partici-pants). In addition to problem behavior, item

engagement was measured during the prefer-ence assessment. Item engagement, defined ascontact between the participant’s hands and theitem, was measured using 10-s momentary timesampling. Items associated with the highestlevels of item engagement and the lowest levelsof problem behavior were identified as HPitems, whereas items associated with the lowestlevels of item engagement and the highest levelsof problem behavior were identified as LPitems. However, in some cases, items that wereassociated with higher levels of problembehavior were identified as HP as long as theywere associated with at least 80% engagement,and items that were associated with lower levelsof problem behavior were identified as LP aslong as they were associated with no more than50% item engagement.

Leisure-item analysis. During this assessment,four conditions (two types of attention condi-tions and two types of control conditions) werealternated in a random order, and a multiele-ment design was used to demonstrate experi-mental control.

During the attention/HP items condition,procedures were identical to those describedabove for the attention condition. That is,participants were provided with continuousaccess to HP leisure items based on the resultsof the preference assessment. At the start of eachsession, the therapist stated that he or she ‘‘hadwork to do.’’ Contingent on the occurrence ofthe target response, the therapist delivered briefverbal and physical attention as describedpreviously.

During the attention/no-items condition,procedures also were identical to those de-scribed above; however, no leisure items werepresent. This condition was designed to testwhether the absence of items during theattention condition resulted in different out-comes than when items were present during theattention condition.

During the control/HP items condition,procedures were identical to those described


above for the control condition. That is, theparticipant was provided with continuous accessto the same HP items present during theattention/HP items condition. In addition, thetherapist presented brief vocal and physicalattention on a fixed-time 15-s schedule (ormore frequently if initiated by the participant).No programmed consequences were deliveredfor the target behavior.

During the control/no-items condition,procedures were also identical to those de-scribed above; however, no leisure items werepresented. This condition was designed to test ifnoncontingent attention alone (without leisureitems as in the standard control condition)resulted in similar outcomes to the attention/no-items condition. Low levels of problembehavior during this condition would provideadditional support that problem behavior wasmaintained by social-positive reinforcementbecause noncontingent attention served as anabolishing operation, whereas high levels ofbehavior during this condition would provideadditional support that participants’ problembehavior was not maintained by social-positivereinforcement because noncontingent attentiondid not serve as an abolishing operation.

Preferred-item analysis. Three of the 4participants were included in the preferred-itemanalysis. During this assessment, three condi-tions (two types of attention and one type ofcontrol condition) were alternated in a randomorder, and a multielement design was used todemonstrate experimental control.

During the attention/HP items condition,procedures were identical to those describedabove for the attention/HP items condition inthe leisure-item analysis. During the attention/LP items condition, procedures were alsoidentical to those described above; however,LP items rather than HP items were availablecontinuously. This condition was designed totest whether the inclusion of LP items resultedin higher levels of problem behavior during theattention condition for the participant previ-

ously identified as having attention-maintainedproblem behavior and in lower levels ofproblem behavior during the attention conditionfor participants previously identified as havingautomatically reinforced problem behavior.During the control/HP items condition, proce-dures were identical to those described above.

RESULTS

Results for the initial functional analysis forall participants are depicted in Figure 1. ForJake, hand biting occurred at differentiallyhigher levels during the attention and demandconditions, indicating that his behavior wasmaintained by social-positive reinforcement inthe form of attention and by social-negativereinforcement in the form of escape. Motorstereotypy for Beth, tapping for Amy, and shirttwirling for Carl occurred at differentiallyhigher levels in the alone condition than indemand and attention conditions, suggestingthat these stereotypies were maintained byautomatic reinforcement. However, shirt twirl-ing for Carl was low and variable during the lastfew alone sessions. Following the initial func-tional analysis, an extended series of alonesessions was conducted, which showed that shirttwirling persisted in the absence of socialconsequences.

The results of the preference assessment forall participants are depicted in Figure 2. Stimuliare ordered based on their associated levels ofitem engagement (highest to lowest). In general,items associated with the highest levels of itemengagement and the lowest levels of problembehavior were selected for inclusion in thesubsequent functional analyses. For Jake, toytrains were identified as the HP item and astuffed dog was identified as the LP item.Because he did not engage in problem behaviorduring the preference assessment, only itemengagement was used for determining itemselection. A toy train was selected over jacks andwatersnake because it resulted in similarly highlevels of item engagement and was associated


Figure 1. Frequency of hand biting and percentage occurrence of motor stereotypy, tapping, and shirt twirlingduring functional analyses for Jake, Beth, Amy, and Carl, respectively.


Figure 2. Percentage occurrence of problem behavior and item engagement during the preference assessment forJake, Beth, Amy, and Carl.


with more appropriate item engagement. ForBeth, a bumble ball was identified as the HPitem and a book was identified as the LP item.For Amy and Carl, two HP items (electronicbook and a vibrating toy for Amy, Slinky andsquirmy wormy for Carl) and two LP items(magnet toys and a stuffed kitten for Amy,musical keyboard and top for Carl) wereidentified for use during the subsequentfunctional analyses.

The results of the leisure-item analysis for allparticipants are depicted in Figure 3. For Jake,hand biting occurred at higher levels during theattention/no-items condition than in the atten-tion/HP items, control/HP items, and control/no-items conditions. In the absence of otherdata, the lower levels of problem behaviorobserved during the attention/HP items condi-tion might be interpreted as suggesting thatbiting was not maintained by attention, anoutcome that is inconsistent with results fromhis initial functional analysis (in which leisureitems were not included in the attentioncondition). Higher levels of biting observedduring the attention/no-items condition suggestthat biting was maintained by attention, anoutcome that is consistent with results from hisprevious functional analysis. In addition, lowerlevels of biting were observed during thecontrol/no-items condition, suggesting thatnoncontingent attention alone suppressed bit-ing, providing further support that biting wasmaintained by attention.

Results of the leisure-item analysis weresimilar for Beth, Amy, and Carl. For each ofthese participants, the target behavior occurredat higher levels during the attention/no-itemsand the control/no-items conditions thanduring the attention/HP items and the con-trol/HP items conditions. Lower levels ofproblem behavior observed during the atten-tion/HP items condition suggest that theirbehavior was not maintained by attention, anoutcome that is consistent with results fromtheir initial functional analyses. In the absence

of other data, elevated levels of problembehavior observed during the attention/no-items condition may be interpreted as suggest-ing behavioral maintenance by attention, anoutcome that is not consistent with results fromparticipants’ initial functional analyses. Inaddition, the high levels of problem behaviorobserved in the control/no-items conditionsuggested that noncontingent attention alonedid not suppress this behavior. This providesfurther support that these participants’ targetbehavior was not maintained by attention.

The results of the preferred-item analysis forJake, Carl, and Amy are depicted in Figure 4.For Jake, biting occurred at higher levels duringthe attention/LP items condition than duringthe attention/HP items and control/HP itemsconditions. In the absence of other data, lowerlevels of biting during the attention/HP itemscondition might be interpreted as suggestingthat his behavior was not maintained byattention, an outcome that is inconsistent withresults from his initial functional analysis (inwhich toys were not included in the attentioncondition). Elevated levels of biting observedduring the attention/LP items condition suggestthat biting was maintained by attention, anoutcome that is consistent with results from hisinitial functional analysis. These findings indi-cate that including HP items in the attentioncondition may mask detection of attention-maintained behavior, whereas the inclusion ofLP items did not have such an effect.

Results of the preferred-item analysis weresimilar for Amy and Carl. For each of theseparticipants, the target behavior occurred athigher levels during the attention/LP itemscondition than during the attention/HP itemsand control/HP items conditions. Lower levelsof problem behavior during the attention/HPitems condition suggested that target behaviorwas not maintained by attention, an outcomethat is consistent with results from their initialfunctional analyses. In the absence of otherdata, elevated levels of problem behavior


Figure 3. Frequency of hand biting and percentage occurrence of motor stereotypy, tapping, and shirt twirlingduring the leisure-item analysis for Jake, Beth, Amy, and Carl, respectively.


Figure 4. Frequency of hand biting and percentage occurrence of tapping and shirt twirling during the preferred-item analysis for Jake, Amy, and Carl, respectively.


observed during the attention/LP items condi-tion might be interpreted as suggesting behav-ioral maintenance by attention, an outcome thatis not consistent with results from the initialfunctional analyses. These findings indicate thatincluding LP items in the attention conditionmay not eliminate the establishing operation forautomatically reinforced behavior (i.e., depriva-tion from sensory stimulation persists in thepresence of LP items). By contrast, including HPitems in the attention condition may haveresulted in a finding consistent with theoutcomes from the initial functional analysis byterminating the establishing operation for auto-matically reinforced behavior. This also suggestsa method for treating problem behavior.

DISCUSSION

The present study found that the inclusion ofleisure items during the attention condition ofthe functional analysis in some cases facilitatedand in other cases impeded identification of thefunction of problem behavior. For participantswhose problem behavior was maintained byautomatic reinforcement, the inclusion ofleisure items in the attention condition facili-tated clear detection of the behavior’s main-taining variable. However, for the participantwhose problem behavior was maintained byattention, the inclusion of HP leisure itemshindered detection of the behavior’s maintain-ing variable.

These findings were consistent with thosereported by Ringdahl et al. (2002) for theparticipant whose problem behavior was main-tained by attention. That is, noncontingentaccess to leisure items during the attentioncondition resulted in low levels of behaviorduring attention (a false-negative outcome).The present results also extended the findings ofRingdahl et al. in a number of ways. First, aninitial functional analysis that included an alonecondition was conducted prior to the leisure-item analysis to identify whether behavior wasmaintained by attention or automatic reinforce-

ment. Second, a duration-based leisure-itemassessment was conducted to identify items thatwere highly preferred (defined as those resultingin high engagement and low problem behavior).Third, we included 3 participants whosebehavior was maintained by automatic rein-forcement to determine how the inclusion ofleisure items during the attention conditionaffected automatically reinforced behavior.

In addition, the effects of using leisure itemsof varying preference was evaluated to identify apotential solution to the confounding effectproduced by including leisure items during theattention condition of the functional analysis.Results of such analyses may indicate whetherlower preference items can serve as an abolish-ing operation for the response product ofbehavior maintained by automatic reinforce-ment while not competing with attention-maintained problem behavior. Ringdahl et al.(2002) suggested that the use of leisure items ofdifferent preference (i.e., not HP items) maylead to a better procedure for obtaining clearoutcomes from functional analyses. Resultsfrom the preferred-item analysis showed thatlevels of behavior during the attention/LP itemscondition were as high as those obtained in theattention/no-items condition. This suggested aconsistent outcome for Jake, whose behaviorwas maintained by attention, but an inconsistentoutcome for Carl and Amy, whose behavior wasmaintained by automatic reinforcement. More-over, problem behavior occurred at low levelsduring the attention/HP items condition (sug-gesting an inconsistent outcome for Jake and aconsistent outcome for Carl and Amy). Thus, assuggested by McCord and Neef (2005), it wasdifficult to identify items that would simulta-neously control for automatically reinforcedbehavior while not impeding detection of anattention function.

The present findings have a number ofimplications for whether or not to include HPor LP leisure items (or no items) in functionalanalyses. For example, including HP leisure


items in the attention and control conditionsmay yield a false-negative outcome for atten-tion-maintained problem behavior (i.e., lowlevels of behavior during the attention condi-tion when attention is the maintaining vari-able). However, including HP items duringattention and control conditions could yieldcorrect identification of automatically rein-forced behavior (i.e., low levels of behaviorduring the attention condition when attentionis not the maintaining variable). By contrast, ifHP items are included in only the controlcondition (and LP items or no items are usedduring the attention condition), attention-maintained problem behavior would be identi-fied correctly, avoiding a false-negative outcomefor attention. However, including LP or noitems during attention may lead to false-positiveidentification of an attention function forbehavior that is maintained by automaticreinforcement (i.e., elevated levels of behaviorwould be observed in both the attention andalone conditions).

Results of the present study demonstrate theutility of including a control/no-items condi-tion (i.e., noncontingent attention). Whenproblem behavior was maintained by attention,low levels of problem behavior were observed inthe control/no-items condition, whereas whenproblem behavior was maintained by automaticreinforcement, high levels were observed in thiscondition. Based on this information, onepossibility would be to remove items from theattention condition when conducting a func-tional analysis. If problem behavior occurs athigh levels during the attention and aloneconditions, an extended alone condition couldbe conducted (Vollmer et al., 1995) with thepossible addition of a control/no-items condi-tion. If problem behavior decreases acrosssuccessive sessions during the extended alonecondition and is low in the control/no-itemscondition, this would provide additional sup-port that this behavior was maintained byattention. By contrast, if problem behavior is

maintained during the extended alone condi-tion and is high during the control/no-itemscondition, this would suggest that the problembehavior was, most likely, not maintained byattention but by automatic reinforcement,because extinction did not occur during theextended alone condition and because noncon-tingent attention during the control conditiondid not act as an abolishing operation.

Although not the focus of the current study,the present findings extend research on non-contingent reinforcer delivery of HP stimuli fortreatment of attention-maintained problembehavior. Noncontingent presentation of HPstimuli has been shown to reduce problembehavior maintained by attention when com-bined with extinction (Hanley, Piazza, &Fisher, 1997) and when extinction was not ineffect (Fischer, Iwata, & Mazaleski, 1997). Thefindings obtained for Jake during the leisure-item and preferred-item analyses replicate thosefound by Fischer et al. That is, lower levels ofproblem behavior were observed during theattention/HP items condition than during theattention/no-items condition, demonstratingthat the continuous presentation of HP itemsmay reduce problem behavior even when itcontinues to be reinforced by attention. Resultsobtained for Jake during the preferred-itemanalysis also replicate findings by Fisher,O’Connor, Kurtz, DeLeon, and Gotjen(2000), who showed that noncontingent pre-sentation of stimuli reduced attention-main-tained behavior in the absence of extinctionwhen HP stimuli (but not LP stimuli) wereused. Thus, the present findings provideadditional support for the use of noncontingentpresentation of HP stimuli for treating atten-tion-maintained problem behavior.

REFERENCES

Berg, W. K., Peck, S., Wacker, D. P., Harding, J.,McComas, J., Richman, D., et al. (2000). The effectsof presession exposure to attention on the results ofassessments of attention as a reinforcer. Journal ofApplied Behavior Analysis, 33, 463–477.


Derby, K. M., Wacker, D. P., Sasso, G., Steege, M.,Northup, J., Cigrand, K., et al. (1992). Brief func-tional assessment techniques to evaluate aberrantbehavior in an outpatient setting: A summary of 79cases. Journal of Applied Behavior Analysis, 25,713–721.

Fischer, S. M., Iwata, B. A., & Mazaleski, J. L. (1997).Noncontingent delivery of arbitrary reinforcers astreatment for self-injurious behavior. Journal ofApplied Behavior Analysis, 30, 239–249.

Fisher, W. W., O’Connor, J. T., Kurtz, P. F., DeLeon, I.G., & Gotjen, D. L. (2000). The effects ofnoncontingent delivery of higher and lower qualityreinforcers on destructive behavior. Journal of AppliedBehavior Analysis, 33, 79–83.

Hanley, G. P., Iwata, B. A., & McCord, B. E. (2003).Functional analysis of problem behavior: A review.Journal of Applied Behavior Analysis, 36, 147–185.

Hanley, G. P., Piazza, C. C., & Fisher, W. W. (1997).Noncontingent presentation of attention and alterna-tive stimuli in the treatment of attention-maintaineddestructive behavior. Journal of Applied BehaviorAnalysis, 30, 229–237.

Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E.,& Richman, G. S. (1994). Toward a functionalanalysis of self-injury. Journal of Applied BehaviorAnalysis, 27, 197–209. (Reprinted from Analysis andIntervention in Developmental Disabilities, 2, 3–20,1982)

Iwata, B. A., Pace, G. M., Dorsey, M. F., Zarcone, J. R.,Vollmer, T. R., Smith, R. G., et al. (1994). Thefunctions of self-injurious behavior: An experimental-epidemiological analysis. Journal of Applied BehaviorAnalysis, 27, 215–240.

McCord, B. E., & Neef, N. A. (2005). Leisure items ascontrols in the attention condition of functionalanalyses. Journal of Applied Behavior Analysis, 38,417–426.

Michael, J. (1982). Distinguishing between discrimi-native and motivational functions of stimuli. Journalof the Experimental Analysis of Behavior, 37, 149–155.

O’Reilly, M. F. (1999). Effects of presession attentionon the frequency of attention-maintained be-havior. Journal of Applied Behavior Analysis, 32,371–374.

O’Reilly, M. F., Sigafoos, J., Edrisinha, C., Lancioni, G.,Cannella, H., Choi, H., et al. (2006). A preliminaryexamination of the evocative effects of the establishingoperation. Journal of Applied Behavior Analysis, 39,239–242.

Piazza, C. C., Fisher, W. W., Hanley, G. P., Hilker, K., &Derby, K. M. (1996). A preliminary procedure forpredicting the positive and negative effects ofreinforcement-based procedures. Journal of AppliedBehavior Analysis, 29, 137–152.

Ringdahl, J. E., Winborn, L. C., Andelman, M. S., &Kitsukawa, K. (2002). The effects of noncontingentlyavailable alternative stimuli on functional analysisoutcomes. Journal of Applied Behavior Analysis, 35,407–410.

Smith, R. G., Iwata, B. A., Goh, H., & Shore, B. A.(1995). Analysis of establishing operations for self-injury maintained by escape. Journal of AppliedBehavior Analysis, 28, 515–535.

Vollmer, T. R., Marcus, B. A., Ringdahl, J. E., & Roane,H. S. (1995). Progressing from brief assessments toextended experimental analyses in the evaluation ofaberrant behavior. Journal of Applied Behavior Anal-ysis, 28, 561–576.

Received April 8, 2006Final acceptance October 18, 2006Action Editor, Iser DeLeon


further evaluation of leisure items in the attention condition of functional analyses

Documents