Transferring Control of the Mand to the Motivating ... · Transferring Control of the Mand to the...

The Analysis of Verbal Behavior

89

2007, 23, 89–102

The importance of developing an effectivemand repertoire for children with autism andother developmental disabilities has been welldocumented in current literature concerninglanguage training (Michael, 1988; Sundberg,1993; Shafer, 1994; Charlop-Christy, Carpen-ter, LeBlance, & Kellet, 2002; Sundberg &Michael, 2001). Mand training offers a num-ber of benefits including reductions in inap-propriate behavior (Carr & Durand, 1985;Shafer, 1994; Charlop-Christy, et al., 2002) andincreased effectiveness of language training forother verbal operants (Carroll & Hesse, 1987;Arntzen & Almås, 2002). Perhaps most sig-nificantly, however, a strong mand repertoireallows persons with developmental disabilitiesto effectively control their social environmentby increasing access to unconditioned and con-ditioned reinforcers and by increasing the valueof interacting with other members of the ver-bal community.

In his book Verbal Behavior, Skinner definedthe mand as “. . . a verbal operant in which theresponse is reinforced by a characteristic con-sequence and is therefore under the functional

control of relevant conditions of deprivationor aversive stimulation” (1957, pp. 35–36).Keller and Schoenfeld (1950) later termedthese conditions as establishing operations, andas Skinner’s (1957) definition of the mand sug-gests, they function as a controlling variablefor the mand (p. 35). The control exerted bythe establishing operation over the mand dis-tinguishes it from other primary verbal oper-ants in which control is determined by an an-tecedent discriminative stimulus (Michael,1988). The concept of the establishing opera-tion was later refined by Michael (1993, 2000).He defined it as any stimulus, condition, orevent that (a) momentarily alters the value ofsome stimulus as a reinforcer and (b) evokesall responses that have produced that reinforcerin the past (Michael, 1993). The term estab-lishing operation was later replaced with mo-tivating operation (MO) to more accuratelydescribe the bidirectional effects of establish-ing and abolishing the value of reinforcement,and the resultant evoking or abating of the rel-evant behavior (Laraway, Snycerski, Michael,& Poling 2003).

As with other behavioral variables, such asdiscriminative stimuli and consequences, theMO may be manipulated to alter the frequencyof some behavior and should therefore be con-sidered an independent variable to be studiedin terms of its control over behavior (Sundberg,2005). Analysis of the MO as an independentvariable is necessary in order to determine itsexact effect on the occurrence or nonoccurrence

Transferring Control of the Mand to the Motivating Operationin Children with Autism

Emily J. Sweeney-Kerwin, Vincent J. Carbone, Leigh O’Brien,Gina Zecchin, and Marietta N. Janecky, Carbone Clinic

Few studies have made use of B. F. Skinner’s (1957) behavioral analysis of language and precise tax-onomy of verbal behavior when describing the controlling variables for the mand relation. Consequently,the motivating operation (MO) has not typically been identified as an independent variable and the natureof a spontaneous mand has been imprecisely described. The purpose of this study was to develop proce-dures to bring the mand response under the control of the relevant MO and therefore free it from themultiple controls that are more easily identified by practitioner’s who rely on Skinner’s analysis and tax-onomy. Using a rolling time delay and prompt fade procedure both participants’ mand repertoires weresuccessfully transferred to the relevant MO and a listener and described within the context of a behavioralanalysis of language.

Key words: verbal behavior, motivating operation, mand, autism.

We would like to thank the parents of the partici-pants in this study for their support and coopera-tion. Portions of this paper were presented at theannual meeting of the New York State Associationfor Behavior Analysis, Saratoga Springs, NY, No-vember 2005. Address correspondence to: Emily J.Sweeney-Kerwin, Carbone Clinic, Valley Cottage,NY 10989; e-mail: emilyjanesweeney@ gmail.com.

90 EMILY J. SWEENEY-KERWIN, et al.

of behavior and its application to the develop-ment of effective procedures to bring aboutchange in behavior. This point is of particularimportance when considering the mand reper-toire of children with autism. Many childrenwith autism develop mand repertoires that aremultiply controlled, and they fail to exhibitspontaneous mands (Charlop, Schreibman, &Thibodeau, 1985; Charlop-Christy, et al., 2002;Sundberg, 2005). Multiply-controlled mandsdevelop when both the MO and some otherstimulus is present during acquisition and sub-sequently both stimuli in combination exertcontrol over the response. These stimuli mayinclude an echoic/mimetic stimulus presentedfor the purpose of modeling the correct topog-raphy of the mand, or the presence of an itemthat serves as reinforcement for the mand.Multiply-controlled mands may therefore bepartially controlled by an echoic or nonverbalstimulus and the MO, or a combination of allthree variables. Although useful, multiply-con-trolled mands reduce the effectiveness of aperson’s repertoire to control the environmentunder all conditions, specifically conditions inwhich the supplementary verbal or nonverbalstimuli are not present. In the absence of thesupplementary stimuli, a multiply-controlledrepertoire is ineffective in producing reinforce-ment regardless of the MO’s value altering ef-fect. Take for example a child who learns toeffectively mand for a cookie under the stimu-lus control of both the cookie’s presence andthe MO. In conditions in which the cookie ispresent and some level of deprivation has in-creased the value of the cookie as a reinforcer,the child may mand frequently by saying“cookie.” However, when placed in conditionsin which the cookie is not present but depriva-tion has again increased the value of a cookieas a reinforcer, the child may not reliably mandfor it. In both instances described above, theeffectiveness of the cookie serving as a rein-forcer did not change, however, the frequencyof behaviors that have produced that reinforcerin the past (manding) decreased. This changeshould therefore be attributed, not to a changein the motivation for the reinforcer, but ratherto the controlling influence of the response bythe discriminative nonverbal stimulus. Manychildren with autism experience the difficul-ties described above and therefore have mandrepertoires that do not include spontaneousmands or mands that occur solely under the

control of the MO and a listener (Sundberg,2005). Language training programs focusingon the development of the mand repertoire inchildren with autism should include proceduresdesigned to increase spontaneous mand re-sponses and to transfer stimulus control ofmultiply-controlled mands to the MO.

A number of studies have sought to over-come the problem of spontaneity by using fixedtime delay procedures to eliminate teacher ini-tiated vocal prompts. Halle, Marshall, andSpradlin (1979) used a time delay procedureto increase vocalizations of severely retardedchildren living in an institutional setting. Thestudy was conducted during meal times and theresponse targeted for increase was meal re-quest. During the baseline conditions, partici-pants were called by a staff member to walk toa food counter and select one of several foodtrays placed on the counter. Requests for mealswere not required during the baseline condi-tion. During the experimental conditions, how-ever, after the participants had been called bya staff member and approached the foodcounter, the delivery of food trays was delayedfor 15 s or until a request for the meal occurred.If a request did not occur during the 15 s delay,a model of the appropriate meal request wasprovided and following the imitation of theresponse, the food tray was delivered. The re-searchers were able to effectively increase thevocalizations of three of the six participants inthe study by simply implementing the timedelay. Two of the six participants initially re-quired a modeled prompt but were later able toask for the items without the model, and oneparticipant learned, after intensive training, torequest the meal tray without prompting. Theresearchers concluded that the time delay wasan effective procedure for increasing vocaliza-tions if the targeted response was previouslylearned. A subsequent study by Halle, Baer, andSpradlin (1981) replicated these results in adifferent setting, with a shorter time delay, andacross different motivational categories.

Another commonly used time-delay proce-dure involved gradually increasing the lengthof the time delay as a method of fading promptsand transferring the stimulus control of the re-sponse to MO control. Charlop, Schreibman,and Thibodeau (1985) used a gradual time de-lay procedure to teach children with autism torequest items without an echoic prompt. Theresearchers gradually increased the time be-

91TRANSFERRING CONTROL OF THE MAND

tween the presentation of an item and the echoicprompt, and were able to successfully transferthe stimulus control of each child’s vocaliza-tions from the teacher’s echoic prompt to thepresence of the item. It was therefore concludedthat a time delay procedure may be effectivein transferring the stimulus control of mandsinitially prompted by some echoic stimulus tomands that are controlled by the item’s pres-ence. Time delay procedures have also beenused to effectively transfer the stimulus con-trol of other types of vocalizations, includingverbalizations of affection (Charlop & Walsh,1986), social pleasantries (Matson, Sevin,Fridley, & Love, 1990; Matson, Sevin, Box, &Francis, 1993), task appropriate speech duringplay (Ingenmey & Van Houten, 1991), andsocial greetings (Charlop & Trasowech, 1991;Matson, et al., 1993).

Most of the studies cited have failed to de-fine their terms according to Skinner’s classi-fication of verbal operants and the variablesthat control each. This set of circumstances hasled to imprecise descriptions of the controllingvariables. For example, the failure to identifythe role of the MO as an independent variablehas led to terminological confusion about howto define a spontaneous mand. The authors ofmost of these studies have treated “requests”that occurred without vocal prompts in the pres-ence of the item desired to be spontaneous.These types of “spontaneous” responses mayactually be part tact and part mand and maynot occur in the future without the control ex-erted by the nonverbal stimulus. On the topicof spontaneous mands, Sundberg (2005) states,“. . . to be optimally useful a mand should oc-cur in the absence of the object or conditionthat is reinforcement for the mand; it shouldoccur primarily under the control of the MO”(p. 14). Sundberg and Michael (2001) indicatedthat there are substantial benefits to be gainedfrom discussing language training in terms ofthe specific controlling variables that accountfor each of the verbal operants and identifyingthe operants according to the terms B. F. Skin-ner suggested in his book Verbal Behavior(1957). Notwithstanding these advantages,only a few studies have described methods thathave brought the mand response primarily un-der the control of the MO and a listener andtherefore appropriately defined the meaning ofspontaneous responding. One such study wasconducted by Carr and Kologinksy (1983) in

which the researchers asserted that effectivesign repertoires should not be brought underthe control of stimuli such as objects or verbalquestions but instead should be under the con-trol of more broadly defined stimuli such asthe presence of an appropriate audience. In ac-cordance with this assumption, six childrenwith autism were taught to use sign languageto request reinforcers using imitative prompt-ing, fading, and differential reinforcement inthe absence of any verbal or nonverbal stimuli.A similar study by Carr and Durand (1985)taught developmentally disabled children tovocally request the attention of adults as analternative to maladaptive behavior. These re-quests were taught during demand conditionsunder which the researchers had establishedthat assistance or praise would serve as a rein-forcer. An echoic stimulus was first used toprompt the appropriate response and was thensystematically faded. Although both of thesestudies were able to effectively bring the mandrepertoire solely under the control of the MOand a listener, neither study identified the re-sponses according to Skinner’s taxonomy ofverbal behavior or identified the MO as an in-dependent variable.

In studies designed to increase a spontane-ous mand repertoire, only a few experimentershave explicitly defined their terms accordingto Skinner’s analysis of verbal behavior andhave systematically manipulated the MO inorder to examine its effect over the mand rep-ertoire. One such study was conducted by Halland Sundberg (1987) in which the researcherswere able to teach persons with developmen-tal disabilities to mand under the control of theMO by arranging chains of responses that,when completed, resulted in the delivery ofsome terminal reinforcer. Each response in thechain was trained until the participant was ableto complete the chain independently. Thechains were then interrupted by removing anitem necessary for the completion of the chain.The absence of that item subsequently in-creased the value of that item as a reinforcerand increased the frequency of behaviors thathad produced that item in the past. The itemwas therefore established as a conditional con-ditioned reinforcer based on its relation to thecompletion of the response change and ulti-mately gaining access to some terminal rein-forcer. The researchers trained the manual signresponse for the missing item using an imita-


tive prompting and fading procedure. Uponcompletion of training, participants beganmanding under the control of the MO for themissing item under conditions in which thatitem would serve as a reinforcer. Hall andSundberg (1987) therefore demonstrated thatthe control for the mand could be transferredto relevant MO or motivational variables. Simi-lar results were obtained by Sigafoos, Doss,and Reichle (1989) using graphic symbols asthe response form.

The purpose of this study was to replicatethe findings of previous research related tomand training. The study was also designedspecifically to extend the technology of mandtraining by testing a modified time-delay pro-cedure to free the mand from the discrimina-tive control exerted by the presence of the de-sired item. In addition, this study was designedto extend the findings of prior studies the treat-ment of children with autism and to developprocedures that could be easily implementedby teachers and instructors in educational set-tings. It was hypothesized that the addition ofthe terminological refinements brought aboutby including Skinner’s (1957) classification oflanguage and Michael’s (1988) identificationof the MO as the primary controlling variablefor the mand would bring analytic and proce-dural clarity to the task.

METHOD

Participants and Setting

There were two participants in this study.Both participants were enrolled for five, 3-hoursessions per week at a private educational set-ting offering one-on-one intensive teaching inthe form of discrete trial training interspersedwith teaching in the natural environment fa-cilitated through play based activities. Bothchildren were selected as participants for thisstudy due to their lack of a mand repertoireprimarily under motivational (MO) control.

Martin was a 3-year-old male diagnosed withautism in the moderate range of disabilities. Hedemonstrated an echoic repertoire and had ac-quired over 100 multiply-controlled vocalmands. His articulation was poor but was im-proving through differential reinforcement ofbetter word approximations. His tact andintraverbal repertoires were limited but devel-oping. When denied access to a reinforcer or

transitioned from a highly preferred to less pre-ferred activity, Martin would sometimes engagein problem behavior in the form of crying,whining, or flopping.

Jeff was a 7-year-old boy diagnosed with au-tism in the moderate to severe range. Jeff hada poor echoic repertoire and therefore manualsign language was chosen as his verbal re-sponse form. He had begun to vocally approxi-mate some words, but due to poor articulationthe use of manual sign was still required. Histact and intraverbal repertoires were weak.When denied access to a reinforcer, Jeff wouldengage in problem behavior in the form of hit-ting, yelling, or pinching.

Response Definitions

The dependent variable in this study was thefrequency of MO-controlled mands. MO-con-trolled mands were defined as any mand thatoccurred at least 15 s after the desired item hadbeen displayed and consumed and any mandthat occurred without the desired item everhaving been displayed to the participantthroughout the session. For example, if 15 s ormore elapsed since a desired item had last beendisplayed, this mand was recorded as MO con-trolled. If however, another mand occurred inless than 15 s since the last display of a desireditem, this mand was recorded as multiply con-trolled. This time interval was chosen because15 s was the approximate time required for eachparticipant to receive and consume the rein-forcer from the time of its display. Any responseprior to the 15 s elapsing was recorded as mul-tiply controlled because the reinforcer wouldlikely still be visible to the participant.

Recording Procedure and Calculation ofInter-observer Agreement

The participants’ instructors served as thedata recorders throughout the experiment. Theprimary data recorder was seated next to thechild, either at a table or on the floor, with adata sheet on a clipboard. A tally mark was re-corded in the appropriate column on the datasheet for responses that were judged to be con-trolled by the MO, or multiply controlled ac-cording to the response definition above. Datawere also recorded on the times that the re-sponse was prompted by the display of the tar-geted item. The frequencies of MO-controlled


and multiply-controlled mands were talliedafter each session and plotted on the graphsshown in Figures 1 and 2.

Additional instructors were trained to recordobservations of the dependent variables inde-pendently for the purposes of inter-observeragreement (IOA). For purposes of calculatingIOA the data recording of the primary recorderwas compared to that of the secondary recorder.An agreement occurred when both recordersrecorded the observed response as evoked bythe same controlling variable, either multiplycontrolled or MO-controlled. A disagreementoccurred when one observer recorded the oc-currence of a behavior that the other observerdid not record, or the observers did not recordthe same controlling variable for a particularresponse. The IOA was calculated by dividingagreements by agreements plus disagreementsand multiplying by 100. Reliability observa-tions were conducted for about 15% of ses-sions across all conditions. The IOA was 100%for all sessions.

Design

A multiple baseline design across behaviorswas used to verify the effectiveness of the in-dependent variables (Baer, Wolf, & Risely,1968).

Pre-baseline

For several weeks prior to the baseline con-dition every mand response was recorded dur-ing the daily 3-hour sessions to determine mandresponses that were MO and multiply con-trolled. Several mand responses related to fooditems occurred for both participants at highrates under the multiple control of the item’spresence and the MO. Mand responses for theseitems were occurring dozens of times per dayfor many weeks prior to the experimental con-ditions. None of the mand responses occurredsolely under the control of the MO and a lis-tener during this period. For Martin, four itemsfrequently manded for under the control of theitem’s presence and the MO were identified andselected as targets for inclusion in this study.Targeted items included fries, lollipop, pretzels,and chips. Two items frequently manded forunder the control of the item’s presence andthe MO were identified and selected as targetsfor inclusion in this study for Jeff. The targeted

items selected were bacon and biscuits.

Baseline

During baseline, targeted items were re-moved from the participants’ view and the oc-currence of MO-controlled mands for each itemwas recorded during 3 hour teaching sessionseach day between 9:00 a.m. and 12:00 p.m. Noother edibles were made available to either par-ticipant during baseline and all experimentalsessions so as not to compete with the value ofthe targeted items. During the baseline andexperimental conditions, teaching sessions con-sisted of discrete trial training and natural en-vironment teaching. During discrete trial train-ing access to preferred videos, action figures,and other toys were used to reinforce correctresponding. During natural environment teach-ing, both participants participated in variouspreferred activities such as games, arts andcrafts, puzzles, videos, and outdoor activities.Opportunities to mand for these preferred itemsand activities during both discrete trial train-ing and natural environment teaching werecontinuously contrived and encouraged. Par-ents of both participants were asked to with-hold access to the targeted items immediatelyprior to each experimental session; however,access to all other food items was not controlledand each participant was typically fed break-fast prior to each teaching session. Multiply-controlled mands did not occur during thisphase because the removal of all targeted itemsprevented any multiply-controlled mands fromoccurring.

Rolling Time Delay

During the first experimental session for eachparticipant, one targeted item was displayed atthe beginning of the session, while the otheritems were maintained in baseline. If the par-ticipant manded for the item by saying thename, a small bite-size portion or equivalentof the item (a piece of pretzel, a lick of a lolli-pop, a piece of biscuit, etc.) was delivered im-mediately and the response was recorded as amultiply-controlled mand. A 2-min time delaythen occurred, during which the targeted itemremained out of view of the participant. If theparticipant manded for the item within this timeinterval, but at least 15 s after the display ofthe item, the response was recorded as MO con-


trolled and the mand was reinforced by thedelivery of a bite-size portion of the item. Thetime delay interval timer was reset and the next2-min time delay interval began. If, at the endof any two minute interval, an MO-controlledresponse had not occurred, the item was dis-played again as a prompt for the mand. If aresponse occurred it was reinforced with thedelivery of a bite-size portion of the item andthe next rolling 2-min time delay began.

Beginning with the second session and allsubsequent sessions in this phase, during thefirst 30 min of the session, each participant wasgiven the opportunity to mand for the targetitem without it being displayed. This compo-nent of the treatment was implemented to probefor stimulus-control transfer across sessionswhich also crossed days. This procedureguarded against providing visual prompts forthe response when they may not have beennecessary and therefore reducing dependencyon the visual stimulus. Stimulus-control trans-fer from a multiply-controlled mand to an MO-controlled mand was said to have occurred ifthe participant manded at least once solely un-der the control of MO and a listener during this30-min period at the start of each session. Ifthe participant manded for the item during thisdaily probe the mand was reinforced and re-corded as MO controlled. In addition, the itemwas never displayed for the remainder of the3-hour session and therefore only MO-con-trolled mands could occur during these ses-sions. If a mand did not occur during the probethe item was displayed at the end of the 30-min period and the rolling time delay andprompt fade procedures described above wereimplemented for the remainder of the session.

When MO-controlled manding began occur-ring during the treatment condition at levelssubstantially greater than during baseline, asecond item was then subjected to the inde-pendent variable. Criterion for introduction ofa new item to the treatment conditions was notestablished, since mand rates are under the con-trol of ever changing motivational operations.The procedures described above continueduntil all items underwent treatment.

Generalization and Response Maintenance

Generalization across instructors and re-sponse maintenance over time was assessed forone of the two participants (Martin). Jeff left

the program to return to his home outside ofthe United States before the generalization andresponse maintenance data could be assessed.To determine generalization across instructors,Martin’s sessions were conducted by four dif-ferent instructors. These instructors conductedMartin’s sessions using the procedures imple-mented during the baseline phase. Again, allitems were removed from view and the occur-rences of MO-controlled mands were recorded.The absence of the target items from the par-ticipants’ view prevented the opportunity formultiply-controlled mands to occur. The roll-ing time delay procedure was never imple-mented during the generalization and responsemaintenance session. The frequency of MO-controlled mands was recorded by each instruc-tor. Data were collected over a 3-month periodduring which time the targeted items werenever displayed. Consequently, maintenance ofthe MO-controlled mands over time was alsoassessed.

RESULTS

Baseline and treatment for all of Martin andJeff’s targeted responses are displayed in Fig-ure 1 and Figure 2 respectively. In addition,Figure 1 displays Martin’s generalization andmaintenance data. In each figure, two datapoints are displayed along the data path. Bothclosed and open circles indicate the number ofMO-controlled mands that occurred during thesession. However, a closed circle indicates thatan MO-controlled mand did not occur duringthe 30-min probe session and therefore all ofthe MO-controlled mands for that 3-hour ses-sion occurred after having seen the item at leastonce at the end of the 30-min probe. Opencircles indicate that an MO-controlled manddid occur during the 30-min probe period andtherefore each MO-controlled mand during theremainder of the session occurred without hav-ing seen the item since at least the end of thelast session during the previous day. The opendata points represent MO-controlled mands thatoccurred during a session when the targeteditem was never displayed.

As shown in Figure 1, MO-controlled mandsby Martin for the first item, fry, occurred dur-ing the first experimental session. The fre-quency of MO-controlled mands for fry rangedfrom 0 to 78 per treatment session. That is, from0 to 78 times per MO-controlled mand, the


mand was prompted from 0 to 78 times by dis-playing the required item. (These numbers donot appear on the figure.) This is in compari-son to a baseline level of zero occurrencesthroughout the phase. The mand for fry wasprompted an average of about 10 times per ses-sion. However, MO-controlled manding for fryoccurred without any prompts throughout thesession beginning with session 12 and only 1session in more than 50 subsequent sessionsrequired any prompt at all. This response hadbeen freed from the control of the nonverbalstimulus and was now occurring solely underthe control of the MO and the listener. Mandsfor fry occurred without prompting for 80% ofall the experimental sessions.

Baseline responding for the second item,lollipop, remained at zero spontaneous mandsper session during the implementation of treat-ment for the first item. However, immediatelyupon subjecting the response to the experimen-tal variables, the frequency of the mand forlollipop increased. The frequency of MO-con-trolled mands for lollipop ranged from 0 to 25during this phase compared to zero duringbaseline. The mand for lollipop was promptedan average of about 38 times per session (notshown on the figure). MO-controlled mandsfor lollipop required more frequent display ofthe item as compared to the response for fry.After the first treatment session, MO-controlledmanding for lollipop occurred without prompt-ing for only 10% of all additional experimen-tal sessions. While the increase in frequencyof MO-controlled mands was functionally re-lated to the implementation of the treatment, itwas not as robust as compared to the effectsupon the other responses. For example, duringonly four of the experimental phases did MO-controlled mands occur without any prompt-ing throughout the session.

Baseline responding for the third item, pret-zel, remained at zero MO-controlled mands persession during the implementation of treatmentfor the first two items. For pretzel, the fre-quency of MO-controlled mands ranged from1 to 24 per session as compared to zero duringall sessions in the baseline condition. Pretzelwas only prompted an average of about twotimes per session. MO-controlled manding forpretzel required only one session with the dis-play of the item to evoke unprompted mandingfor all the subsequent experimental conditions.In other words, after the first session MO-con-

trolled manding for pretzel occurred withoutprompting for all 21 of the experimental ses-sions that followed for this item.

Baseline responding for the fourth item, chip,remained at zero MO-controlled mands persession during the implementation of treatmentfor all other items. During the experimentalphase the frequency of MO-controlled mandsfor chip ranged from 1 to 20 per session. Chipwas prompted an average of about five timesper session. Chip began occurring without anyprompting after only one treatment session anda prompt was never required to maintain theresponding during the 10 subsequent treatmentsession.

Generalization and maintenance data forMartin are displayed in Figure 1. Probes forgeneralization across instructors were con-ducted by four different experimenters over thecourse of a three month period. The data in thisphase represent the frequency of MO-con-trolled mands aggregated across all four in-structors but presented by response topogra-phy. During this period MO-controlled mandsoccurred from 0 to 11 per times session for alltargeted items. While the effects were not aslarge as in the treatment condition when thesame instructor acted as the listener each day,the responses did occur more frequently thanbaseline across instructors and over time.

Figure 2 presents the frequency of MO-con-trolled mands during baseline and treatmentconditions for Jeff. MO-controlled manding byJeff for the first item, bacon, occurred duringthe first experimental session. The frequencyof MO-controlled mands for bacon rangedfrom 4 to 429 per treatment session as com-pared to zero occurrences of the mand duringall sessions of the baseline condition. The mandfor bacon was prompted an average of abouttwo times per session. After only four treat-ment sessions the mand for bacon occurredwithout prompting in all but one of the subse-quent 71 treatment sessions. This meant thatMO-controlled manding for bacon occurredwithout prompting for 96% of the additionalexperimental sessions. When a second itemunderwent treatment, MO-controlled mandingfor bacon was maintained for 55 sessions.

Before the second response, biscuit, was sub-jected to the experimental condition, Jeffmanded under the control of the MO for it onlyonce. The frequency of MO-controlled mandsfor biscuit ranged from 0 to 60 per session as


compared to zero during all baseline sessions.Biscuit was prompted an average of about 28times per session. MO-controlled manding forbiscuit occurred without prompting for 52%

of the additional experimental sessions.

DISCUSSION

Figure 1. Frequency of MO-controlled mands per sessions during baseline (BL), treatment, and generali-zation and maintenance conditions for all targeted items for Martin.


The results of the current study met the re-quirements of the multiple baseline designacross behaviors leading to the conclusion thatthe rolling time delay and prompt fading pro-cedures were effective in establishing an MO-controlled manding repertoire for both learn-ers in this study. As shown in Figures 1 and 2,baseline levels of MO-controlled manding forboth participants remained stable at zero mandsper session for all items. Following the imple-mentation of the rolling time delay procedureMO-controlled mands began occurring for bothlearners either within the first session or afteronly one experimental session, and increasedat various frequencies for all targeted items.These results suggest that by systematicallyfading the presence of the items, we were able

to transfer the stimulus control of the partici-pants’ manding repertoire from the multiplecontrol exerted by a discriminative stimulusand an MO to the control exerted by the MOand a listener. The participants’ motivation forthe targeted items was established during thepre-baseline condition of this study duringwhich multiply-controlled mands in the pres-ence of the targeted items occurred at high fre-quencies. When the baseline condition wasimplemented, however, and the targeted itemswere removed from the participants’ view, allmands for these items ceased. Manding onlyoccurred again after the items were reintro-duced during the experimental condition andstimulus-control transfer from multiply-con-trolled mands to mands controlled solely by

Figure 2. Frequency of MO-controlled mands per sessions during baseline (BL) and treatment conditionsfor Jeff.


the MO and a listener had occurred. These re-sults provide at least indirect evidence that thelearner’s motivation for the targeted items ex-isted throughout the baseline condition. Ulti-mately, during the rolling-prompt delay con-ditions the additional control exerted by thenonverbal stimulus (item) had an additive ef-fect and therefore the response was evoked onlywhen both controlling variables were present.It appears that the multiply-controlled mandmay have actually been part tact, and the trans-fer of stimulus-control procedures may havefreed it from the variables that control the tact,such as the presence of a nonverbal stimulus.

Previous studies have defined spontaneousmanding as mands that occur without a vocalprompt (Halle, et al., 1979; Halle, et al., 1981;Charlop, et al., 1985; Matson, et al., 1990;Matson, et al., 1993). The current study extendsthe previous research findings on this topic byusing both Skinner’s analysis of verbal behav-ior and the literature concerning the MO(Michael, 1993; Laraway, et al., 2003) to de-fine spontaneous manding as MO-controlledmands, or mands that occur solely under thecontrol of the MO and a listener. In addition,prior research has demonstrated that the con-trol of the manding repertoire can be transferredfrom an echoic prompt to a prompt by the non-verbal stimulus (Halle, et al., 1979; Halle, etal., 1981; Charlop, et al., 1985; Matson, et al.,1990; Matson, et al., 1993) but have failed todemonstrate transfer from the non-verbalstimulus to the MO. The current study adds tothe research findings that a manding repertoirecan be brought primarily under the control ofthe MO and supports Michael’s (1988) andSundberg’s (2005) contention that the MOserves as an independent variable in the devel-opment of language in the form of the mandresponse.

Further analysis of the results shows that onlya few prompts in the form of the visual displayof the targeted item were necessary to produceMO-controlled mands. Both learners beganmanding solely under the control of the MOafter no more than one experimental sessionfor all items. In addition, for the majority oftargeted responses, transfer from multiply-con-trolled mands to MO-controlled mands oc-curred within only a few experimental sessions.With the exception of the lollipop response,MO-controlled mands began occurring withinthe first experimental session for Martin for all

targeted items. In addition, once MO-controlledmands began occurring, stimulus-control trans-fer from a multiply-controlled mand to an MO-controlled mand occurred for the responses fry,pretzel, and chip within no more than 12 ex-perimental sessions. For pretzel and chip re-sponses, stimulus-control transfer occurredafter only one experimental session. For Jeff,MO-controlled mands also began occurringafter no more than one experimental session,and during the majority of experimental ses-sion a display of the targeted item was not re-quired, indicating that stimulus-control trans-fer from multiply-controlled mands to MO-controlled mands had occurred. These resultsfurther support the assumption that the moti-vation for the targeted items existed during thebaseline condition but the control exerted bythe MO alone was insufficient to evoke theresponse. Once implementation of the rolling-time delay and prompt-fade procedure had ef-fectively transferred control to the MO, MO-controlled mands began occurring readily forboth learners. The speed with which MO-con-trolled mands began occurring also speaks tothe efficiency of the procedure tested and fur-ther verifies the overall effectiveness of therolling-time delay procedure.

The variability in the frequency of MO-con-trolled mands as evidenced in Figures 1 and 2can be explained by considering the nature ofthe MO as defined by Michael (1993, 2000).The MO is a stimulus, condition, or event thatmomentarily alters the value of some stimulusas a reinforcer and evokes all responses thathave produced that reinforcer in the past. Thekey term in this definition when consideringthe variability in the data presented in Figures1 and 2 is “momentarily.” As Michael’s (1993)definition suggests, the MO’s establishing orabolishing effect may change the value of itemsas reinforcers and therefore change the fre-quency of any behavior that successfully pro-duced that reinforcer in the past. The MO forany particular item or activity may be fleetingand motivation for multiple items may existsimultaneously, producing competing values ofreinforcement. The differing rates of MO-con-trolled manding across sessions and itemsfound in this study would therefore be ex-pected, as the participants’ motivation for dif-ferent items shifted according to the values ofadditional and competing forms of reinforce-ment. During the course of each experimental


session, the participants were exposed to vari-ous putative reinforcers, such as games, toys,videos, and physical activities that may havecompeted with the value of the edible itemstargeted in this study and accounted for thesome of the variability in frequency of MO-controlled mands.

Furthermore, the introduction of additionalitems into the treatment condition may alsoaccount for the changing frequency of MO-controlled mands for previously introduceditems, and for the seemingly downward trendof MO-controlled mands over time. Newly in-troduced items may have competed with thevalue of previously introduced items and there-fore altered that value of these items as rein-forcers and decreased the MO-controlled re-sponses that had produced these reinforcers inthe past. As shown in Figure 1, the frequencyof MO-controlled mands for lollipop during thetreatment condition, though greater than dur-ing baseline, was low relative to the frequencyof MO-controlled mands for fry. These resultsmust be analyzed in terms of a number of vari-ables. First, lollipop was introduced after MO-controlled mands had begun occurring for fry,which may have produced a competing rein-forcer effect and therefore resulted in a lowerfrequency of MO-controlled mands for lolli-pop and a decrease in MO-controlled mandsfor fry. Once MO-controlled mands for lolli-pop began occurring, two additional items,pretzel and chip, were then subjected to theeffects of the independent variable. With theintroduction of each new item, MO-controlledmanding for all items previously in the treat-ment condition decreased. These data suggestthat the competing value of the introduction ofthe new items may have altered the value ofthe previous items as reinforcers and thereforedecreased the occurrences of the MO-con-trolled responses for those items. Other vari-ables outside the control of the experimenters(e.g., the amount of food each child consumedduring a meal prior to the experimental ses-sions) may also have altered the value of theedible reinforcers selected for this study.

An alternative explanation for the gradualreduction in the frequency of MO-controlledmands across treatment sessions is that the dep-rivation of the targeted items during thebaseline condition may have temporarily in-creased the value of these items as reinforcers.Consequently, when the items were initially

reintroduced during the treatment conditionhigh levels of MO-controlled mands occurred.As these items were delivered during the treat-ment sessions satiation may have occurred re-sulting in a decrease in the reinforcing valueof the items and the gradual decrease in MO-controlled mands.

Mands that only occur under tightly circum-scribed conditions such as presence of only onelistener will be less beneficial to a speaker thana response repertoire that is controlled by manydifferent listeners. Stimulus generalizationacross listeners and environments is impera-tive to the development of an effective mandrepertoire. Generalization data for Martin arepresented in Figure 1. During the generaliza-tion and maintenance condition, assessment ofstimulus generalization across instructors wasconducted. Although the frequency of MO-controlled mands decreased during this condi-tion, MO-controlled mands did occur at higherfrequencies than during baseline sessions.Moreover, these generalized responses oc-curred without programming for generaliza-tion. Although no formal measure of generali-zation across settings was conducted, it shouldbe noted that MO-controlled mands occurredthroughout the educational environment. Mar-tin manded solely under the control of the MOand a listener in both instructional and playconditions, as well as in unfamiliar environ-ments such as the staff kitchen and conferenceroom. This suggests that in addition to gener-alization across instructors, generalizationacross settings may have also occurred. Un-fortunately, due to circumstances beyond thecontrol of the experimenters, the second par-ticipant, Jeff, left the study before his gener-alization and maintenance data could be gath-ered. It is important to note that incidentalprobes of stimulus generalization across in-structors occurred over a 2-week period dur-ing which Jeff’s primary instructor was onleave. During this time, his experimental ses-sions were conducted by various replacementinstructors and MO-controlled mands for thetargeted item, bacon, continued to occur dur-ing all sessions. In addition, Jeff manded solelyunder the control of the MO and a listenerthroughout the educational setting includinginstructional and play environments, as wellas outside the building during outdoor activi-ties.

Generalization across instructors and settings


was most likely facilitated by the ease of imple-mentation of the rolling-time delay and prompt-fade procedures which allowed for all experi-mental sessions to be conducted under natu-rally occurring conditions. All experimentalsessions were conducted for the duration of theparticipant’s teaching session and did not in-clude any isolation or separation of the partici-pant from the typical instructional environment,or any differentiation of routine. Generaliza-tion of MO-controlled mands across instruc-tors and environments further supports the useof the rolling-time delay and prompt-fade pro-cedures as effective interventions in the devel-opment of MO-controlled mand repertoires.

Although the results support the contentionthat generalization of MO-controlled mandsoccurred for Martin, the results indicated thatresponse generalization did not occur. Re-sponse generalization in this study would haveoccurred if the participants began mandingsolely under the control of the MO and a lis-tener for items that had not been subjected tothe independent variable. Clearly, from thebaseline data presented in Figures 1 and 2, nei-ther participant’s response repertoire was char-acterized by response generalization. It is con-ceivable that this effect would not be seen with-out the training with a greater number of ex-emplars.

With respect to the durability of the effect,the maintenance data collected for Martin overa 3-month period indicate that the MO-con-trolled responses continued to occur across alltrained responses for more than 45 sessionsafter the experimental condition was discon-tinued. As shown in Figure 1, MO-controlledmands decreased during the generalization andmaintenance condition relative to the treatmentcondition. All items targeted during this studywere edible items and therefore the simulta-neous change in MO-controlled mands for alltargeted items may suggest that during sessionsin which MO-controlled responses were low,the value of edible items in general may havebeen low. Again, maintenance data for Jeff werenot collected, however, it should be noted thatwhile biscuit was subjected to the independentvariable, MO-controlled mands for bacon con-tinued to occur without prompting and there-fore this condition was tantamount to a main-tenance probe condition.

The results of this study lend support to theconcept of the functional independence of the

verbal operants and that stimulus-control trans-fer procedures are effective and necessary totransfer control across operant classes. Skin-ner (1957) explained, “Classifications of re-sponses are useful only in separating the vari-ous types of controlling relations, and someresponses may show features of both mand andtact” (p.189). Both participants in this studyinitially emitted responses that had “featuresof both mand and tact.” Skinner contends thatwe should not expect that the response willspontaneously transfer to a more pure form ofmand, or tact for that matter, without contact-ing the conditions that account for the inde-pendent operant. In this study none of the re-sponses spontaneously took on the features ofthe mand without the arrangement of condi-tions that account for the separate mand rela-tion. Only after arrangement of the indepen-dent variable in the form of stimulus-controltransfer procedures was the response trans-ferred from a response that had features of boththe mand and the tact to one that was morepurely defined as a mand.

The results of this study have important im-plications for the use of selection-based re-sponding. Programs that require the speakerto mand by selecting a picture of the desireditem may produce a useful mand repertoire withchildren with autism. However, the very na-ture of the response form which requires thepresence of a stimulus to be selected that hasalmost identical appearance to the item re-quested may in fact preclude the developmentof an MO-controlled mand repertoire. In thisstudy, a non-vocal learner was taught a topog-raphy-based verbal response form, manual signlanguage, instead of a picture selection methodfor several reasons, but one consideration wasthe value of freeing the response from the pres-ence of the item in the form of a picture or icon.Consequently, when choosing a response formthat is either topography- or selection-based(Michael, 1985) for a non-vocal learner, theclinician may want to consider the findings ofthis study.

This study is limited by the fact that find-ings from only two participants are reported.Replication with additional learners with au-tism who have varying characteristics will beneeded to determine the benefits of these pro-cedures for a broad participant group. In addi-tion, the inclusion of only edible items limitsthe generality of the findings. MOs related to


other reinforcers may react differently whensubjected to the independent variables studiedin this report.

In terms of practical significance the resultsof this research demonstrate that the occurrenceof multiply-controlled mands may be an im-portant issue to assess within language-train-ing programs for children with autism. More-over, in some learners it appears that multiply-controlled mands will require explicit and pro-grammed stimulus-control transfer proceduresto free them from the control of other stimuli.Therefore, programs designed to teachmanding to children with autism may need toinclude procedures to systematically fade thestimulus control of reinforcing items from thepresence of the item and the MO to the moti-vation for the item or condition and a listeneralone.

REFERENCES

Arntzen, E., & Almås, I. K. (2002). Effects ofmand-tact versus tact-only training on theacquisition of tacts. Journal of Applied Be-havior Analysis, 35, 419–422.

Baer, D., Wolf, M., & Risley, T. (1968). Somecurrent dimensions of applied behavioralanalysis. Journal of Applied BehavioralAnalysis, 1, 91–97.

Carr, E. G., & Durand, V. M. (1985). Reducingbehavior problems through functional com-munication training. Journal of AppliedBehavioral Analysis, 18, 111–126.

Carr, E.G., & Kologinsky, E. (1983). Acquisi-tion of sign language by autistic childrenusing a time delay procedure. Journal ofApplied Behavioral Analysis, 16, 297–314.

Carroll, R. J., & Hesse, B. E. (1987). The ef-fects of alternating mand and tact trainingon the acquisition of tacts. The Analysis ofVerbal Behavior, 5, 55–65.

Charlop, M. H., Schreibman, L., & Thibodeau,M. G. (1985). Increasing spontaneous ver-bal responding in autistic children using atime delay procedure. Journal of AppliedBehavioral Analysis, 18, 155–166.

Charlop, M. H., & Trasowech, J. E. (1991). In-creasing autistic children’s daily spontane-ous speech. Journal of Applied BehavioralAnalysis, 24, 747–761.

Charlop, M. H., & Walsh, M. E. (1986). In-creasing autistic children’s spontaneousverbalizations of affection: An assessment

of time delay and peer modeling procedures.Journal of Applied Behavior Analysis, 19,307–314.

Charlop-Christy, M., Carpenter, M. L.,LeBlanc, L. A., & Kellet, K. (2002). Usingthe picture exchange communication sys-tem (PECS) with children with autism: As-sessment of PECS acquisition, speech, so-cial-communicative behavior, and problembehavior. Journal of Applied BehaviorAnalysis, 3, 213–232.

Hall, G. A., & Sundberg, M. L. (1987). Teach-ing mands by manipulating conditioned es-tablishing operations. The Analysis of Ver-bal Behavior, 5, 41–53.

Halle, J. W., Baer, D. M., & Spradline, J. E.(1981). Teacher’s generalized use of delayas a stimulus control procedure to increaselanguage use in handicapped children. Jour-nal of Applied Behavioral Analysis, 14, 389–409.

Halle, J. W., Marshall, A. M., & Spradlin, J. E.(1979). Time Delay: A technique to increaselanguage use and facilitate generalization inretarded children. Journal of Applied Be-havior Analysis, 12, 431–439.

Ingenmey, R., & Van Houten, R. (1991). Us-ing time delay to promote spontaneousspeech in an autistic child. Journal of Ap-plied Behavioral Analysis, 24, 591–596.

Keller, F. S., & Schoenfeld, W. N. (1950). Prin-ciples of psychology. New York: Appleton-Century Crofts.

Laraway, S., Snycerski, S., Michael, J., & Pol-ing, A. (2003). Motivating operations andterms to describe them: Some further refine-ments. Journal of Applied Behavior Analy-sis, 36, 407–414.

Matson, J. L., Sevin, J. A., Fridley, D., & Love,S. R. (1990). Increasing spontaneous lan-guage in three autistic children. Journal ofApplied Behavioral Analysis, 23, 227–233.

Matson, J. L., Sevin, J. A., Box, M. L., Francis,K. L., (1993). An evaluation of two meth-ods for increasing self-initiated verbaliza-tions in autistic children. Journal of AppliedBehavioral Analysis, 26, 389–398.

Michael, J. (1985). Two kinds of verbal be-havior plus a third. The Analysis of VerbalBehavior, 3, 1–4.

Michael, J. (1988). Establishing operations andthe mand. The Analysis of Verbal Behavior,6, 3–9.

Michael, J. (1993). Establishing operations.


The Behavior Analyst, 16, 191–206.Michael, J. (2000). Implication and refinements

of the establishing operation concept. Jour-nal of Applied Behavior Analysis, 33, 401–410.

Shafer, E. (1994). A review of interventions toteach a mand repertoire. The Analysis ofVerbal Behavior, 12, 53–66.

Sigafoos, J., Doss, S., & Reichle, J. (1989).Developing mand and tact repertoires inpersons with severe developmental disabili-ties using graphic symbols. Research inDevelopmental Disabilities, 10, 183–200.

Skinner, B. F. (1957). Verbal behavior.Englewood Cliffs, NJ: Prentice Hall.

Sundberg, M. L. (1993). The application ofestablishing operations. The Behavior Ana-lysts, 16, 211–214.

Sundberg, M. L. (2005). A behavioral analysisof motivation and its relation to mand train-ing. In L. W. Williams (Ed.). Developmentdisabilities: Etiology, assessment, interven-tion, and integration (pp. 1–22). Reno, NV:Context Press.

Sundberg, M. L., & Michael, J. (2001). Thevalue of Skinner’s analysis of verbal behav-ior for teaching children with autism. Be-havior Modification, 25, 698–724.

Transferring Control of the Mand to the Motivating ... · Transferring Control of the Mand to the...

Documents

Transcript of Transferring Control of the Mand to the Motivating ... · Transferring Control of the Mand to the...