contingencies promote delay tolerance · 5/14/2015 · tolerance for delays to reinforcement (here...

CONTINGENCIES PROMOTE DELAY TOLERANCE

MAHSHID GHAEMMAGHAMI, GREGORY P. HANLEY, AND JOSHUA JESSELWESTERN NEW ENGLAND UNIVERSITY

The effectiveness of functional communication training as treatment for problem behaviordepends on the extent to which treatment can be extended to typical environments that includeunavoidable and unpredictable reinforcement delays. Time-based progressive delay (TBPD)often results in the loss of acquired communication responses and the resurgence of problembehavior, whereas contingency-based progressive delay (CBPD) appears to be effective forincreasing tolerance for delayed reinforcement. No direct comparison of TBPD and CBPD has,however, been conducted. We used single-subject designs to compare the relative efficacy ofTBPD and CBPD. Four individuals who engaged in problem behavior (e.g., aggression, vocaland motor disruptions, self-injury) participated. Results were consistent across all participants,and showed lower rates of problem behavior and collateral responses during CBPD than duringTBPD. The generality of CBPD treatment effects, including optimal rates of communicationand compliance with demands, was demonstrated across a small but heterogeneous group ofparticipants, reinforcement contingencies, and contexts.Key words: contingency-based delay, delayed reinforcement, functional communication

training, generality, schedule thinning, severe problem behavior

Functional communication training (FCT;e.g., Carr & Durand, 1985), a form offunction-based differential reinforcement, hasbeen shown to reduce problem behavior byteaching the individual an appropriate alterna-tive behavior that serves the same function asproblem behavior. In fact, FCT combined withextinction has been shown to be an efficacioustreatment for a variety of problem behaviorsthat differ both functionally and topographi-cally (Kurtz, Boelter, Jarmolowicz, Chin, &Hagopian, 2011; Tiger, Hanley, & Bruzek,2008). Problems arise, however, because care-givers cannot always reinforce requests immedi-ately, and these periods of nonreinforcementfor appropriate communication can lead to the

resurgence of problem behavior (Hanley,Iwata, & Thompson, 2001).To increase the generality of effects, thinning

the schedule of reinforcement for the functionalcommunication response (FCR) is often listedas an essential component of FCT when treat-ment is extended to the typical environment(e.g., Durand & Moskowitz, 2015; Kurtzet al., 2011). Various procedures for increasingtolerance for delays to reinforcement (heredefined as near-zero levels of problem behaviorand manding during extensive nonreinforce-ment periods and the resumption of mandingwhen appropriate) have been evaluated (seeHagopian, Boelter, & Jarmolowicz, 2011, for areview). One common procedure involves pro-graming gradually increasing delays betweenthe FCR and the delivery of the reinforcer,often indicated with a brief signal such as“wait” (Vollmer, Borrero, Lalli, & Daniel,1999). This procedure has been referred to as adelay schedule (Hagopian et al., 2011). Delayschedules have an intuitive appeal because thearrangement best emulates the typical situationsexperienced in the natural environment(i.e., when parents cannot provide requested

This study was conducted in partial fulfillment of therequirements for a PhD in Behavior Analysis fromWestern New England University by the first author. Wethank William H. Ahearn, Chata Dickson, and RachelThompson for their feedback on earlier versions of thismanuscript.

Address correspondence to Gregory P. Hanley, WesternNew England University, 1215 Wilbraham Road, Spring-field, Massachusetts 01119 (e-mail: [email protected]).

doi: 10.1002/jaba.333

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2016, 49, 548–575 NUMBER 3 (FALL)

548

items or interactions, they tell the child to waitand then provide that which was requestedwhen it is possible to do so). This procedure,however, frequently results in the loss of thenewly acquired FCR and a resurgence of prob-lem behavior, usually within the first 16 s ofdelay (Fisher, Thompson, Hagopian, Bow-man, & Krug, 2000; Hagopian et al., 2011;Hanley et al., 2001). Delayed reinforcementcould also elicit negative emotional responses aswell as evoke an excessive amount of manding(Fisher et al., 2000) before the resurgence ofproblem behavior, and these collateral responsesmay be as disruptive as the original problembehavior.The apparent obstacle to achieving general

effects of FCT may be partly due to theextinction-like periods created by the longdelays that could result in the resurgence of thepreviously reinforced problem behavior(Lieving & Lattal, 2003) and agitated or emo-tional responding (Lerman & Iwata, 1996).Lieving and Lattal (2003) showed that as sche-dules of intermittent reinforcement arethinned, longer periods of nonreinforcementare created that are functionally equivalent toconventional extinction and can lead to theresurgence of the previously reinforcedresponse. Volkert, Lerman, Call, and Trosclair-Lasserre (2009) demonstrated that problembehavior may resurge when the newly acquiredFCR is placed on an intermittent schedule ofreinforcement during generalization attemptswith FCT. In fact, deterioration of FCT treat-ment effects during implementation in moretypical environments has often been reported(Fisher et al., 2000; Hagopian, Fisher, Sullivan,Acquisto, & LeBlanc, 1998; Hagopian et al.,2011; Hanley et al., 2001; Rooker, Jessel,Kurtz, & Hagopian, 2013). For instance,Hagopian et al. (1998) found that when delaysto reinforcement and demand fading wereintroduced, the efficacy of FCT with extinctionwas maintained in less than one half of theapplications (i.e., clinically acceptable outcomes

were not achieved). In most of their cases, theaddition of punishment was necessary to attaina 90% reduction in problem behavior. Wackeret al. (2011) also showed that long periods ofFCT treatment (an average of 14 months) wererequired before treatment effects would persistduring 5-min periods of extinction, and evenlonger periods of treatment were required whena 15-min extinction period was used. In addi-tion, although problem behavior was reducedduring repeated extinction exposures, afternearly 2 years of treatment, problem behaviorwas not eliminated for half the children.The negative side effects observed with

delayed schedules may also be attributed to acontingency-weakening effect that occurs underthis arrangement (Hanley et al., 2001). Positivecontingency strength may be defined as theprobability of obtaining reinforcement given aresponse being greater than the probabilitygiven no response (Hammond, 1980). Luc-zynski and Hanley (2014) found that deliveringreinforcement after a delay resulted in a contin-gency strength of −1 (the weakest possible con-tingency) because no reinforcers were everdelivered in close temporal proximity to thecommunication response. Thus, the delayschedule created a context that was probablyaversive to the participating children, in thatthey preferred a context with no reinforcementat all to one in which delayed reinforcementwas programmed.To mitigate the contingency-weakening

effects associated with delay, multiple schedules(e.g., Fisher, Kuhn, & Thompson, 1998; Han-ley et al., 2001), chained schedules (e.g., Fisheret al., 1993; Lalli, Casey, & Kates, 1995), or acombination of the two (e.g., Falcomata, Mue-thing, Gainey, Hoffman, & Fragale, 2013)have been adopted and have successfully main-tained zero or near-zero rates of problem behav-ior during long periods of nonreinforcement.Multiple schedules involve a time-based alter-nation between reinforcement and extinctioncomponents, both of which are correlated with

549DELAY TOLERANCE

distinct stimuli (e.g., colored cards). Chainedschedules incorporate a response requirement,either a specific number of demands or a spe-cific duration of time engaged in work activity,to be completed, after which the first instanceof FCR results in reinforcement. Neither multi-ple nor chained schedules, however, preciselyemulate the unplanned and therefore unpre-dictable delays that are often experienced inhomes and schools.Chained and multiple schedules require par-

ents and teachers to plan periods of nonreinfor-cement or demand time, during which anindividual’s FCRs are ignored. After these peri-ods are over (either when a time criterion hasbeen met or through completion of requireddemands), an interval of reinforcement thensets in, and parents are advised to reinforcerequests immediately. Delays in the typicalenvironment, however, do not emulate thisarrangement, and are often sudden, unex-pected, and unplanned. Individuals can requesta variety of items at a given time, and in thesecases, caregivers may not know whether thereinforcer is available until the specific requesthas been made, making it difficult to plan forimmediate reinforcement. One must also beable to tolerate periods in which their reinfor-cing activities are suddenly interrupted andtheir requests are not granted in the absence ofclear stimuli that signal the unavailability ofreinforcement and even under stimulus condi-tions that would normally signal immediatereinforcement (e.g., a toy is available but thebattery runs out). In such cases, the only natu-rally occurring stimuli may be brief verbalresponses of “wait,” “not right now,” or “in aminute” to the request. The individual is, then,expected to wait for the request to be grantedwithout engaging in repeated manding, prob-lem behavior, and negative emotionalresponses. In addition, the individual will oftenbe required to comply with an adult’s requestsor acquiesce to someone else’s preferences dur-ing the delay. At other times, the individual

might need to scan the environment and findalternative activities while he or she waits forthe preferred items and others’ attention. Delayschedules are structurally ideal for teachingbehavioral expectations in these situations;however, strategies for mitigating the extensivenegative side effects associated with their appli-cation have not yet been articulated.One change to typical delay procedures that

may reduce the commonly reported negativeside effects is the addition of probabilisticimmediate reinforcement of the communica-tion response. The addition of immediate rein-forcement of some FCRs would increase theFCR-reinforcer positive contingency strength.This change may also increase the ecologicalvalidity of this procedure because requests inthe typical environment are also immediatelygranted sometimes. Another change that mayimprove the effectiveness of delay schedulesinvolves the addition of a response requirementduring the delay. In other words, the negativeside effects may be mitigated by changing froma time-based delay to a contingency-based delayin which a chain of responses after the FCRwill result in the delivery of reinforcement. Acontingency-based delay increases the FCRreinforcer positive contingency strength bybuilding a chain of responses that ultimatelycontacts reinforcement, thereby minimizing thecreation of long delays that emulate conven-tional extinction.These procedural changes to the delay sched-

ule were described in a study by Hanley, Jin,Vanselow, and Hanratty (2014) in whichprobabilistic immediate reinforcement andcontingency-based delay were used to treatproblem behaviors maintained by positive rein-forcement for one child and a synthesis of posi-tive and negative reinforcement for otherchildren. During contingency-based delay, fol-lowing the cue to wait, the children were ini-tially required to engage in a tolerance response(e.g., saying “okay”); progressively more diffi-cult response chains were then prompted before

MAHSHID GHAEMMAGHAMI et al.550

reinforcers would be delivered. In this way, theexperimenters were able to extend the delay topractical levels that included completion of age-appropriate demands and engagement withappropriate leisure items during the delay with-out the resurgence of problem behavior.Given that Hanley et al. (2014) implemen-

ted multiple changes to the way in whichdelays are traditionally scheduled, the extent towhich each of the changes is necessary for thesuccess of this treatment remains unclear. Forexample, the addition of probabilistic reinforce-ment to increase the contingency strength ofFCRs may be sufficient to produce the sameresults with time-based delay. Also, the merepresence of and redirection to an alternativeactivity may be sufficient to maintain zerolevels of problem behavior during the delay.For instance, Fisher et al. (2000) showed thatin one case the addition of an alternative workactivity, without a contingency, was enough toreduce positively reinforced problem behaviorand collateral responses during nonreinforce-ment intervals. These authors, however, didnot report on the rate of excessive manding orcompliance with demands during these inter-vals. The extent to which the mere presence ofan alternative activity during delays to rein-forcement, without a response contingency, willbe sufficient to eliminate severe problem behav-ior without the emergence of other collateralresponses remains to be investigated.The main purpose of this study was to evalu-

ate the direct effects of a response contingencyduring delayed reinforcement. Althoughcontingency-based delay has been used as themain treatment (Hanley et al., 2014) or a com-ponent of treatment (e.g., Carr & Carlson,1993), the effects of a response contingencyalone have not been evaluated. We thereforeconducted a comparative analysis of time-based(TBCD) versus contingency-based (CBPD)delay tolerance training. To isolate the effectsof a response contingency alone, we includedboth probabilistic reinforcement and alternative

activities in both time-based and contingency-based delay conditions. The second purpose ofthis study was to evaluate the direct effects ofcontingency-based delay on collateral responses(e.g., excessive manding, negative emotionalresponding) and compliance with adult instruc-tions, so multiple measures were collectedacross all participants. The third purpose of thisstudy was to assess the generality of delay toler-ance training. In addition to the systematic rep-lication of the comparison across a wide rangeof participant characteristics and different rein-forcement contingencies, we also evaluated theextent to which behavior changes that occurredas a function of experience with either delayprocedure would generalize to a second contextin which problem behavior during the delaywould be reinforced (i.e., a context that emu-lates typical environments with no extinctionduring delays).

GENERAL METHOD

Participants and SettingsFour individuals, ranging in age from

21 months to 30 years, who had been referredto our university-based outpatient services, par-ticipated in this study. Nico was a 23-month-old typically developing boy who reportedlyhad difficulty waiting for preferred items andactivities. Nico’s parents reported that he wouldoften repeat his requests multiple times, say“no” when told to wait, and would sometimeshave a tantrum that included crying and flop-ping if his requests were not granted. Nicocould follow multistep vocal instructions, had atypically developing imitation repertoire andfine and gross motor skills, and communicatedusing gestures, single words, and partiallyframed sentences. He had an age-appropriateplay repertoire including imaginative play. Nicoattended a center-based day-care.Will was a 30-year-old man with a diagnosis

of pervasive developmental disorder, comorbidintellectual disability, attention deficit

551DELAY TOLERANCE

hyperactivity disorder, and episodic mooddisorder. He had a long history of severeself-injurious behavior (SIB), which consistedprimarily of hand-to-head hitting that often ledto open wounds on his forehead, as well asfinger biting. Will reportedly engaged in SIBthroughout the day at his rehabilitation center,his group home, and during transport to andfrom the center. Staff reported that they oftengave him food and drinks to calm him down.He was nonvocal and had no formal communi-cation system. He could follow some simplegestural prompts, had no echoic or motor imi-tation repertoire, and had limited gross motorand fine motor skills. He could walk independ-ently and feed himself but was not toilettrained and had no independent play or leisureskills. He attended the day habilitation center5 days per week and spent the majority of histime eating, taking walks, or sleeping.Jack was a 21-month-old boy with a diagno-

sis of an autism spectrum disorder (ASD) whoengaged in severe problem behavior multipletimes each day. His problem behavior includedmotor disruptions and aggression toward peersand adults (usually toward his mother) whenpreferred activities were interrupted and whenan item he requested was not available. Theseepisodes usually led to the family leaving socialsettings or providing Jack with the requesteditem or other preferred items as well as physicaland verbal attention (e.g., hugs, squeezes, repri-mands). He could follow simple vocal instruc-tions, had good fine motor skills, could imitatesimple motor responses and partially echo two-word phrases, and communicated using ges-tures and word approximations. He had noindependent play or leisure skills. He receivedearly intervention services at home thatincluded early intensive behavioral intervention,speech language services, and occupational ther-apy. He also attended a home-based day-care.Alex was a 6-year old boy with a diagnosis of

ASD who engaged in daily episodes of severeproblem behavior that included vocal and

physical disruptions and aggression towardadults and peers. Alex reportedly became highlyemotional and aggressive at home, school, andother community outings when his preferredactivities were interrupted or not available andwhen asked to comply with demands. Theseepisodes usually led to adults complying withAlex’s requests and providing access to pre-ferred items to calm him down as well as physi-cal and verbal attention (e.g., hugs, squeezes,reprimands). He could follow multistep vocalinstructions, had a typically developing imita-tion repertoire, fine and gross motor skills, andspoke in full sentences. He had a developmen-tally appropriate play repertoire including imag-inative play. He had some difficultypronouncing certain sounds and was receivingspeech services. He attended a public school inwhich he spent the majority of his time in aresource room that included the support ofparaprofessionals and had an individualizededucational plan.All sessions for Nico, Jack, and Alex were

conducted in small treatment rooms (4 m by3 m) at a university setting equipped with aone-way observation panel, audio-video equip-ment, child-sized tables, two chairs, and aca-demic and play materials as needed. All sessionsfor Will were conducted in an open area in theday rehabilitation center that containedcafeteria-style tables and adult-sized chairs. Ses-sions were conducted 2 to 4 days per week,two to eight times each day. Sessions lasted3 to 5 min throughout the functional analysesand mand analysis. Sessions lasted 3 to 5 min(Nico and Jack) or five evocative trials (Willand Alex) during FCT. An evocative trialinvolved the presentation of the evocative situa-tion and was as long as necessary to allow thetarget FCR to occur and for any scheduleddelays and reinforcement periods to be pre-sented. Sessions lasted for five evocative trialsthroughout the comparative analyses unless thetime-based session termination criteria weremet (20-min session for Nico and 10 min of


crying for Jack). A minimum of three evocativetrials was required for a session to be includedin the analyses.

Data Collection and Interobserver AgreementTrained observers recorded data using pencil

and paper during the functional and preferenceanalyses for Will. Otherwise, trained observerscollected data via computers that provided asecond-by-second account of participants’responses and relevant contextual features. Spe-cific response definitions and data collectionand conversion details are provided in the rele-vant sections.Interobserver agreement was assessed by hav-

ing a second observer collect data on all targetbehaviors simultaneously but independentlyduring at least 20% (range, 20% to 60%) ofthe sessions in each condition for each partici-pant. Each session’s data were divided into 10-sintervals and compared on an interval-by-interval basis. Agreement percentages were cal-culated by dividing the smaller number ofresponses or duration (in seconds) in eachinterval by the larger number, averaging thefractions, and converting the result to a per-centage. Interobserver agreement averaged 95%(range, 76% to 100%) for Nico, 96% (range,80% to 100%) for Will, 93% (range, 76% to100%) for Jack, and 92% (range, 74% to100%) for Alex.

Preference AssessmentWe conducted a preference assessment that

involved presenting an array of items simulta-neously, similar to a multiple-stimulus-without-replacement preference assessment(e.g., DeLeon & Iwata, 1996). For Jack, Alex,and Nico, highly preferred and neutral itemsand activities nominated by caregivers duringan open-ended interview along with other age-appropriate toys and academic activities (10 to20 items) were arbitrarily arranged on twosemicircle-shaped tables before sessions. The

items included varied for each child butremained the same throughout all assessmentand treatment analyses for each child. Duringthe assessment, the analyst directed the childto the tables and reviewed the items that wereavailable by touching and naming each activ-ity. She then allowed the child to walk aroundthe table and manipulate the items brieflybefore prompting the child to choose three tofive preferred items to bring into the sessionroom. The analyst then selected one to threeitems that the child had not chosen duringthe previous two selection opportunities andused those as the neutral items for the alterna-tive activity or demands. Access to the tableswas typically provided after two to foursessions.

PART 1: FUNCTIONAL ASSESSMENT

Data Collection and Response DefinitionsTarget problem behavior for Jack and Alex

included aggression (defined as hitting, biting,kicking, hair pulling, head butting, and push-ing) and disruptions (defined as both physicaldisruptions such as throwing, ripping, swiping,and pushing items, banging items together, andvocal disruptions such as a high-pitchedscream). Will’s problem behavior was self-injurious behavior in the form of hand-to-headhits and knuckle biting. Target behavior forNico included problem behavior (i.e., aggres-sion), minor problem behavior (defined as cry-ing, whining, throwing, ripping, and swiping),gestures (defined as reaching and pointing), sin-gle words, or framed mands. Counts of partici-pants’ problem behavior were collected andconverted to a rate for all analyses.

ProcedureOpen-ended functional assessment interview

and interactive observation. An open-endedinterview, as described by Hanley (2012), wasconducted with the participants’ caregivers pri-marily to discover potential reinforcers that

553DELAY TOLERANCE

might influence the individual’s problembehavior and contexts in which problem behav-ior was most likely. The interview lasted 45 to60 min and was followed by a 20-min informalobservation of the participant interacting withparents (Nico, Jack, and Alex) or staff (Will) inwhich play preferences, language skills, topogra-phies of problem behavior, fine and motorskills, and other unique characteristics describedby caregivers during the interview were directlyobserved to individualize and prepare foranalyses.Mand analysis. The open-ended interview

with Nico’s parents revealed that concerns cen-tered exclusively on situations when an item oractivity or their attention was not immediatelyavailable and Nico was asked to wait, duringwhich time he would mostly engage in exces-sive manding and minor problem behaviorsuch as whining, crying, and throwing items.Given that these behaviors often followed one-word or framed mands for preferred items andthe parent reported that Nico seldom engagedin any severe problem behavior, a mand analy-sis (Hernandez, Hanley, Ingvarsson, & Tiger,2007) was determined to be more suitable foridentifying the predominant response form thatfunctioned as a mand for tangible items. Theanalysis involved rapidly alternating betweentwo conditions. The test consisted of differen-tial reinforcement of target responses (DRA),whereas the control consisted of continuousnoncontingent reinforcement (NCR). DuringNCR, the preferred toys, DVDs, and activitieswere made available freely and continuously.During DRA, the preferred items were placedon a table but access was blocked by the ana-lyst. Access to items was provided for 30 s con-tingent on any target response.Functional analyses. Following open-ended

interviews, functional analyses were designedfor Will’s, Jack’s, and Alex’s problem behavior.The analyses involved rapidly alternatingbetween test and control sessions by(a) presenting the reported evocative situation

(e.g., presenting writing tasks, taking away toysor tablet, removing attention) in test sessionsand allowing 30-s access to the reported conse-quences immediately after problem behavior,and (b) withholding the same evocative situa-tions in control sessions by presenting the puta-tive reinforcers continuously. Events that werenot suspected of maintaining problem behavior(e.g., escape from demands for Jack, analyst’sattention and escape from demands for Will)were freely available in both the test and con-trol conditions, ensuring that the only differ-ence between test and control conditions wasthe programmed reinforcement contingency.Will. Staff reported that whenever Will

appeared agitated or started to engage in self-injury, they gave him snack items. Based onthe results of this interview and the brief obser-vation, an analysis of a social-positive reinforce-ment contingency was conducted using thetypically available food items (e.g., raisins,crackers, peanuts, cheese, cookies). Two tothree of these snack items were visible butslightly out of reach in both test and controlconditions. During the control condition, verysmall bites of each snack were placed on aplate, and the plate was presented to Willapproximately every 10 s independent of hisbehavior. Following his selection, the plate wasremoved. By contrast, the plate with the snackitems was presented during the test sessionsonly after instances of head hitting or fingerbiting; each instance resulted in the plate beingpresented and a snack bite obtained.Jack. Based on the interview results and

observation with Jack’s mother, a synthesizedcontingency of attention and tangible itemswas analyzed in one context conducted by hismother (Context 1) and in another by the ana-lyst (Context 2). Two analyses were conductedto create two baselines from which the directand general effects of the delay procedurescould be evaluated. Preferred items (e.g., a hairbrush, broom and dust pan, DVDs) identifiedby Jack’s mother during the interview and


some additional age-appropriate toys and activ-ities were placed in the preference assessment.Given the mother’s report that to calm Jackdown she would often attend to him and pro-vide access to the preferred items, a synthesizedattention and tangible reinforcement contin-gency was tested in the analysis. During thecontrol condition, Jack had continuous andnoncontingent access to his mother’s (or theanalyst’s) attention (e.g., sitting in her lap, pre-tend cooking with her) and access to preferreditems. During the test condition, the adult pre-tended to be busy with one of the items andalso blocked access to all other preferred items.Contingent on any instance of problem behav-ior, the adult immediately attended to Jack(e.g., comforted and played with him) and gavehim access to the preferred items for 30 s.Alex. During the interview, Alex’s mother

reported that dinosaurs were his favorite topic.He often engaged in imaginative play and con-structed elaborate dinosaur theme sets that“had to remain untouched” in the familyhome. He would demand that his parents andyounger brother play along with the very spe-cific roles he would assign to them. Any move-ment of these items by others, interruption ofplay, or failure to assume the assigned roleresulted in severe tantrums that includedaggression and could last up to 30 min. Alex’sspecific requests extended to other activities inthe home and school. For example, he oftendemanded that his peers play by his rules inthe gym, and he insisted on doing academictasks in a specific manner regardless of the tea-cher’s instructions. Most interruptions or redir-ections of preferred activities resulted in severetantrums. Alex’s mother reported that whenthese tantrums occurred, she helped Alex calmdown by removing her demands and encourag-ing him to take a breath and tell her what hewants, which then resulted in the resumptionof his preferred activity and compliance withhis requests. The results of the open-endedinterview suggested that problem behavior was

evoked when adults stopped complying withhis requests (see Bowman, Fisher, Thomp-son, & Piazza, 1997, for a similar functionalrelation) and interrupted his preferred activitiesto place demands to engage in other tasks.Given that problem behavior often resulted inthe simultaneous delivery of attention, removalof demands, and adult compliance with mands,a synthesized contingency of positive and nega-tive reinforcement was arranged in two analysesconducted by the same analyst but in two dif-ferent contexts. Context 1 contained materialsselected from the preference assessment thatdid not include any dinosaur-related items.Instead, items in the preference assessmentincluded other activities reported as highly pre-ferred by Alex’s mother (e.g., drawing activity,Legos, tablet), other age-appropriate toys andactivities, and demand materials. Context2 included only Alex’s most preferred activity,which was dinosaur figure sets along withdinosaur-themed books and stickers. Duringthe control condition, Alex was given uninter-rupted access to his preferred activity and theanalyst complied with all of his reasonablerequests (i.e., those that could be granted in thesession room safely) and presented no demands.During the test condition, the analyst inter-rupted play, denied his requests, and presenteda demand (e.g., the therapist deviated from theplay as instructed by Alex and told him to dosomething else). Three-step prompting wasused to ensure compliance with demands. Con-tingent on any instance of problem behavior,the analyst removed demands, allowed Alex toresume his activity in his preferred manner,and honored his requests for 30 s.

ResultsNico. Although all of Nico’s target responses,

including aggression, would have producedreinforcement during the DRA condition ofthe mand analysis, only minor problem behav-ior (e.g., whining and throwing), gestures, and

555DELAY TOLERANCE

single-word and partially framed mands wereemitted. Single-word mands, however, emergedas Nico’s predominant response (Figure 1).The rate of single-word responses was consist-ently higher in the DRA sessions, and Nicoengaged exclusively in single-word mands dur-ing the last test–control dyad. The results sug-gested that his predominant response forpreferred tangible items and adult attention wasa single-word mand.Will. Problem behavior was observed exclu-

sively in test sessions in which Will’s problembehavior resulted in snack items (Figure 1).The result of the functional assessment processshowed that problem behavior was maintainedby access to food.Jack. Problem behavior was observed exclu-

sively in test sessions in which Jack’s problem

behavior yielded access to preferred items andadult attention, irrespective of whether hismother or the analyst implemented the contin-gency (Figure 1). The result of the functionalassessment process suggested that problembehavior was maintained by a combination ofsocial-positive reinforcers. Although the exactrole of each reinforcer included in the synthe-sized contingency was not isolated and theextent to which main effects, interactions, orboth were maintaining problem behavior werenot determined, the analysis did emulate thetypical conditions Jack experienced and identi-fied a context that demonstrated control overhis problem behavior. The inclusion of all pos-sible contingencies of reinforcement resulted ina reliable baseline from which to evaluate theeffects of FCT and a highly challenging context

Figure 1. Results of the mand form analysis for Nico and the interview-informed synthesized contingency analysesfor Will, Jack, and Alex.


to evaluate reinforcement delay (see Ghaemma-ghami, Hanley, Jin, & Vanselow, 2016; Han-ley et al., 2014; and Jessel, Hanley, &Ghaemmaghami, 2016, for more detailed dis-cussions of the interview-informed synthesizedcontingency analysis).Alex. Problem behavior was observed exclu-

sively in the test conditions when Alex’s prob-lem behavior terminated adult instruction andallowed him access to preferred activities, adultattention, and having his requests granted(Figure 1). The result of the functional assess-ment process suggested that problem behaviorwas sensitive to a combination of social-positiveand negative reinforcers. Isolating the suspectedcontingencies of reinforcement was also notdesirable or possible in Alex’s case, becausemost events that evoked problem behaviorinvolved a simultaneous provision of both neg-ative and positive reinforcement. For example,removing interruptions of activity meant thathe simultaneously escaped the adult instructionand resumed uninterrupted access to his pre-ferred activity. That is, both positive and nega-tive reinforcement operated in tandem. Similarto Jack, however, the inclusion of all possiblereinforcers in the contingency also provided uswith a challenging and reliable baseline fromwhich to evaluate communication and toler-ance skills.

PART 2: FUNCTIONALCOMMUNICATION TRAINING

Data Collection and Response DefinitionsIn addition to problem behavior and

responses defined above, the followingresponses were also measured during this phase.Nico’s initial FCR was a single-word mand(e.g., “music,” “dance”) identified via the mandanalysis. His target FCR was a fully framedmand (e.g., “I want [item] please,” “More[activity] please”). Will’s target FCR consistedof handing a food icon to the analyst. Jack’starget FCR consisted of a hand gesture to his

chest or a vocal response of “my way.” GivenJack’s limited vocal imitation repertoire, a novelhand gesture was added to supplement thevocal response and allow immediate prompting.Alex’s initial FCR was “my way, please.” Histarget FCR consisted of saying “excuse me” andthen waiting for acknowledgment before saying“May I have my way please?” FCRs were con-sidered prompted if the analyst prompted anypart of the FCR within 10 s of the participantemitting the response. Only independent FCRsare reported. Counts of participants’ communi-cation responses and problem behavior werecollected and converted to a rate for allanalyses.

ProcedureWhen reinforcers were identified for prob-

lem behavior or predominant mand forms, weattempted to replace problem behavior andsimple mand forms with more socially accepta-ble and developmentally appropriate mandforms via FCT plus extinction. The effects ofFCT plus extinction were demonstrated in aconcurrent operants AB design for Nico andWill and a concurrent operants within a multi-ple baseline design across contexts for Jack andAlex. The test sessions of the mand or func-tional analysis served as the baselines for allFCT evaluations.During FCT, access to reinforcers was pro-

vided for approximately 1 min before each ses-sion; the session started by the removal of allreinforcers and the presentation of an evocativesituation for each participant (e.g., the analystpaused the DVD player and turned away fromJack, or the analyst placed a bite of food on aplate visible to Will but out of his reach). A tar-get FCR was selected and reinforced on a fixed-ratio (FR) 1 schedule in which each instance ofthe FCR resulted in 30 s of reinforcement. Allproblem behavior was placed on extinction.A small number of presession training trials(up to five) were conducted before introduction

557DELAY TOLERANCE

of FCT. These trials included a brief instruc-tional statement, modeling of the FCR, role-play of emitting the FCR, and accessingreinforcement and praise or correction of theFCR. During sessions, a most-to-least prompt-ing hierarchy was used to teach the target FCRuntil 80% of FCRs were independent, afterwhich prompts were faded to a vocal promptevery 60 to 90 s as needed. For Alex, whenproblem behavior was eliminated in both con-texts and initial FCRs were emitted independ-ently for two consecutive sessions, the analystincreased the complexity of the responserequired via prompting and differentialreinforcement.

ResultsFigure 2 depicts the results of FCT. There

was a reduction in initial FCRs and variablerates of the target FCR observed with Nico,but after a period of variability, the target FCRwas emitted exclusively and at an optimal rate.FCT resulted in an immediate elimination ofproblem behavior for Will and the acquisitionof the target FCR. Despite some variability inproblem behavior, FCT resulted in an eventualelimination of problem behavior and acquisi-tion of the FCR with Jack in both contexts.FCT resulted in an immediate reduction ofproblem behavior for Alex and the acquisitionof the initial FCR in both contexts.

Figure 2. Results of the functional communication training plus extinction (FCT + EXT) for Nico, Will, Jack,and Alex.


Independent and target FCRs eventuallyoccurred at a higher rate than simple FCRs andto the exclusion of problem behavior in bothcontexts. By the end of FCT, all participantsemitted the target FCR at an efficient rate,maximizing reinforcement to near-continuousaccess, in the absence of problem behavior(Will, Jack, and Alex) or initial FCRs (Nico).

PART 3: COMPARATIVE ANALYSISOF TOLERANCE TRAINING

Data Collection and Response DefinitionsIn addition to response rates defined above

(e.g., target problem behavior and FCRs), thefollowing responses and rates were also meas-ured during the comparative analysis. The tol-erance response (TR) for Nico, Jack, and Alexconsisted of saying “okay” in an appropriatetone and volume within 5 s in response to thedelay cue. TRs were considered prompted ifthe analyst prompted any aspect of the responsewithin 10 s of the participant emitting theresponse. Only independent TRs are reported.For all participants, an optimal rate of FCRwas calculated by dividing the number of evoc-ative trials presented by the total duration ofthat session. For Nico, an optimal rate of TRwas also calculated for each session by dividingthe total number of opportunities (i.e., the totalnumber of delay trials) by the total duration ofthat session. The optimal rates are depicted ineach figure as a dotted data path.Nico’s collateral responses were in the form

of excessive mands (defined as any additionalrequests during the delay that were differentthan the target FCR that initiated the delayinterval). Will’s collateral responses includedmotor disruptions (defined as throwing, swiping,and pushing items) and grabbing others. Jack’scollateral responses were in the form of negativeemotional responding (defined as crying, pout-ing, and saying “no”). Alex’s collateralresponses included attempts to control duringthe delay (defined as negotiating to change the

qualitative features of the task or the amount,vocally refusing or completing the task in amanner different than what the analyst indi-cated, and making additional requests).Alternative activity engagement was defined as

actively manipulating, responding to(e.g., dancing to music), or orienting towardsmaterials (e.g., neutral toys, beads) as instructedby the analyst without problem behavior or col-lateral responses for Nico, Will, and Jack.Engagement was recorded using a 3-s onset–offset delay. For Alex, counts were collected oneach verbal and gestural instruction issued bythe adult and his compliance with each instruc-tion. Compliance was defined as orientingtowards the materials within 5 s of the instruc-tion and completing the task correctly withoutany collateral responses or problem behaviorand without any need for a physical promptfrom the adult.Counts of participants’ communication and

tolerance responses, problem behavior, and dis-crete collateral responses (i.e., excessive man-ding, grabbing and motor disruptions, andattempts to control) were collected and con-verted to a rate for all analyses. Duration datawere collected on other collateral responses(e.g., crying), scheduled and experienced delays(the interval of time between the delivery ofthe delay cue, e.g., “wait,” and the delivery ofreinforcement), and engagement in the alterna-tive activity during delays. The percentage ofsession engaged in negative emotional respond-ing was calculated by dividing the duration ofnegative emotional responding by the sessionduration. The percentage of delay intervalengaged in alternative activity was calculated bydividing the duration of alternative activityengagement by the delay duration.

General ProcedureAfter 1-min access to all reinforcers, every ses-

sion started with the removal of all reinforcersand the presentation of the participant-specific

559DELAY TOLERANCE

evocative situation (e.g., taking away toys ortablet, removing attention, presenting writingtasks). Sessions were as long as necessary toallow five presentations of the evocative situa-tions and all the scheduled delays and rein-forcement periods, henceforth referred to astrials. This resulted in session durations thatranged from 2.5 to 40 min depending on thedelay. During all delay conditions, FCRs werereinforced immediately on two of five ran-domly selected trials. On the remaining threetrials, the FCR resulted in one of several briefverbal delay signals (e.g., “wait,” “not yet,” “ina minute”), that were rotated within partici-pants, and reinforcement was provided aftereither the scheduled amount of time (TBPD)or the scheduled response requirement(CBPD). The specific response contingenciesfor each participant during CBPD are sum-marized in Table 1 and explained in detailunder specific procedures. The contingencyand prompting procedures chosen in CBPDwere dictated by the parental goals and expec-tations during the delay.A geometric progression starting at 1 s

(i.e., 1, 2, 4, 8, …) was used to reach the

terminal delay in an efficient manner and per-haps to allow differences in the procedures to berevealed more readily than with a less rapid pro-gression. This geometric progression was used asa guide for increasing the scheduled delay. Thegeometric progression was used until the targetterminal delay was reached, at which point thescheduled delay was capped at that level. Thetarget terminal delay for each participant wasguided by caregiver and setting requirements.The participant-specific criteria to increasedelays are described in the specific proceduresbelow. The experienced duration of each delayin a CBPD session was determined based on thespeed with which the participant completed theresponse requirement and refrained from engag-ing in problem behavior or collateral responses.The experienced duration of each delay in aTBPD session, however, did not always pre-cisely match this programmed delay due to(a) being yoked to the experienced duration ofthe delay in the CBPD session (Nico only),(b) slight variations in the time required to resetthe reinforcing materials (Will and Context 1 ofJack and Alex), or (c) termination of the delaydue to problem behavior (Context 2 of Jack and

Table 1Participant-Specific Prompts, Response Contingency, and Consequences During the Delays in the Comparative Analysis

Time-baseddelay

Contingency-based delay

Participant Prompts Prompts ContingencyConsequences for

PB and CR

Nico None Gestural and physicalevery 30 to 45 s

DRA-DRO: TR and continuous engagement in alternativeactivity without PB or CR (say “okay” then play in areawith low-preference toys)

Restart delay interval

Will Single vocal Three-step every 15 s DRA: Cumulative number of compliance responses (placeall the provided beads on a string)

Block PB

Single vocal DRO: Cumulative amount of time without PB or CR(engage in any or no activity)

Block PB, pause,and hold up timer

Jack Single vocal Single vocal DRA-DRO: TR and cumulative amount of time withoutPB or CR (say “okay” and engage in any or no activity)

Pause timer

Alex Three-step asneeded

Three-step as needed DRA-DRO: TR and cumulative number of complianceswithout PB or CR (say “okay” and independently complywith a mixture of adult-directed academic and toy-baseddemands)

Continue demands

Note. PB = problem behavior, CR = collateral responses, DRA = DRA-based contingency delay, DRO = DRO-basedcontingency delay, TR = tolerance response, Three-step = vocal, model, physical prompting.


Alex). Reinforcement intervals were increasedfrom 30 s to 120 s (30 s for delays of 0 to 32 s,60 s for delays of 64 to 128 s, and 120 s fordelays of 256 to 600 s).No-delay baseline. These sessions were identi-

cal to the final sessions of FCT. Reinforcementwas withheld until the target communicationresponse was emitted. In all trials, the FCR wasimmediately reinforced and neither problembehavior nor collateral responses resulted inprogrammed consequences.Time-based progressive delay (TBPD). On the

three delay trials, the FCR resulted in a delaysignal and either no additional prompts (Nico),a single prompt (Will and Jack), or multipleprompts (Alex) to engage in the alternativeactivity or comply with demands. Although thealternative activity or instructional materialswere present and freely available during thesesessions, there was no requirement for the par-ticipant to engage these materials or independ-ently comply with demands (i.e., the delayended based on time alone). At the end of thescheduled delay, the reinforcers were deliveredwith a verbal statement (e.g., “Now you canhave —,” “Here you go”). Problem and collat-eral behavior resulted in no programmed conse-quences throughout the session.Tolerance response (TR) training. TR training

was conducted before the start of CBPD forNico, Jack, and Alex. Training sessions of10 trials, 60% of which were delay trials, wereused to teach a specific TR (“okay”) to theadult’s delay cues. A minimum of two sessionswith 80% independent FCRs and TRs wereconducted before the start of CBPD. The train-ing sessions started with a brief instructionalstatement, modeling of the FCR followed bythe delay cue and the TR, role-play of emittingboth the FCR and the TR to access reinforce-ment, and ended with praise or any necessarycorrections. A most-to-least prompting proce-dure was used during each trial.Contingency-based progressive delay (CBPD).

On the three delay trials, the FCR resulted in a

delay signal and either a single prompt (Jack,Will in DRO) or multiple prompts (Nico, Willin DRA, and Alex) to engage in the alternativeactivity or comply with demands. The partici-pant was required to emit the TR or eitherengage in additional specific responses orrefrain from engaging in problem behavior orcollateral responses to terminate the delay. Inother words, reinforcement was withheld untilthe participant completed the responserequirements.

Specific ProceduresNico. The relative efficacy of CBPD and

TBPD was evaluated with Nico in a multiele-ment design. TR training was conducted beforethe start of the comparative analysis. Through-out the comparative analysis, the two condi-tions were presented as a dyad in which thefirst condition to be presented in each dyadwas randomly selected. During both delay con-ditions, the highly preferred toys were placedon a table where Nico and the analyst sat, andthe neutral toys used as the alternative activitywere placed on foam mats. Green (TBPD) andred (CBPD) plastic sheets on the wall and thetable of the session room were correlated witheach condition. In addition, the positions ofthe table and foam mats were flipped duringTBPD and CBPD.During TBPD, FCRs were reinforced imme-

diately on two of five randomly selected trialsby providing access to requested toys and atten-tion. On the remaining three trials, the FCRresulted in a brief verbal delay signal(e.g., “wait”), and access to the highly preferredtoys and attention was withheld until thescheduled amount of time had elapsed. Prob-lem behavior resulted in no programmed conse-quences throughout the session. CBPD wassimilar to TBPD except that DRA-DRO wasused to terminate the delay. After the delay sig-nal, Nico was required to say “okay” and thenplay with the less preferred or neutral toys on

561DELAY TOLERANCE

the foam mats for a target amount of time.Access to the highly preferred toys and atten-tion was provided only after continuousengagement with the alternative activity (neu-tral toys) for the target amount of time. Thedelay was restarted if Nico stopped engaging inthe alternative activity or if he engaged in prob-lem behavior or collateral responses. Gesturaland physical prompts to engage in the alterna-tive activity were provided every 30 to 45 s ifhe was not doing so.Both conditions progressed through the

delay levels according to the following schedule:one session at each of the first three delays(approximately 1, 2, and 4 s), two sessions ateach of the next three delays (approximately8, 16, and 32 s), and one session at a delay of64 s, irrespective of problem behavior or collat-eral responses. The exact duration of each delaywithin a session in TBPD was yoked to theexperienced duration of the delay in the CBPDsession. The comparative analysis was con-cluded at the delay of 64 s, but CBPD wasused to increase the delay to 128 s, after whichthe treatment was extended to Nico’s fatherand then his mother at the terminal delayof 256 s.Will. We systematically replicated the com-

parison between CBPD and TBPD using aslightly modified contingency in an ABACdesign to allow a more independent evaluationof the presence of a response contingency dur-ing delay. For Will, the delay response contin-gency during CBPD was modified to requirecompletion of a cumulative, rather than a con-secutive, number of beading tasks (i.e., the con-tingency was not reset if Will stopped engagingin the activity). The terminal delay durationwas set at 180 s or the placement of roughly10 beads on the string.During both delay conditions, the edible

items were placed on a table where Will andthe analyst sat, and various toys and beadingmaterials were also placed on the table in frontof Will. The toys and beading materials were

freely available throughout all sessions. TBPDwas introduced first. During TBPD, FCRswere reinforced immediately on two of five ran-domly selected trials by providing access to abite of food. On the remaining three trials, theFCR resulted in a brief verbal delay signal(e.g., “in a minute”) and a single vocal prompt(“you can play or bead if you want”) to high-light the option of engaging with the neutralactivities (i.e., there was no requirement toengage with these items). Access to food waswithheld until the scheduled amount of timehad elapsed. Problem behavior was blocked butresulted in no other programmed consequencesthroughout the session.After a return to the no-delay baseline,

CBPD was introduced. During CBPD, two offive randomly selected trials included immedi-ate reinforcement of FCRs, and the remainingthree trials included a delay. In this condition,however, after the delay signal, Will wasprompted to engage in a beading task (“firstput the beads on”), and access to food was pro-vided after completion of a predeterminednumber of beads that corresponded to the tar-get delay. Vocal and model prompts to engagein the beading task were provided every 15 s ifhe was not doing so. Attempts at self-injurywere blocked but resulted in no other pro-grammed consequences. In both conditions,delays were increased after one session with noproblem behavior and collateral responses orafter two sessions if there were any instances ofthese behaviors. The first comparative analysiswas concluded at a delay of 180 s.Among other factors, the pace of the pro-

gression may have contributed to the persist-ence of Will’s problem behavior and collateralresponses during the first comparative analysis.A second analysis that used a multielementdesign was conducted to evaluate the effects oftwo versions of CBPD, DRA only and DROonly, simultaneously against TBPD with aslower programmed progression of the delay.The three conditions were presented in a


random and counterbalanced order. The condi-tions were signaled using color-correlatedstimuli (tablecloths, plates, and beads). Thetime-based condition was signaled with yellowstimuli and was identical to the previous TBPDcondition. The DRA condition was signaledwith red stimuli and was identical to the previ-ous CBPD condition. The DRO condition wassignaled with blue stimuli and was similar tothe CBPD condition described above, exceptthat the response contingency was further mod-ified to consist of the absence of problembehavior and collateral responses for a cumula-tive amount of time. A single vocal prompt(“you can play or bead if you want”) to high-light the option of engaging with the neutralactivities (i.e., there was no requirement toengage with these items) was provided after thedelay cue. No additional vocal prompts wereused to direct Will to play or bead, but in addi-tion to blocking self-injury, the analyst held upthe timer and paused if Will engaged in anyproblem behavior or collateral responding. Thiscomparison was started at a 64-s delay, whichwas the point at which problem behavior andcollateral responses emerged during the previ-ous analysis. When stable and desirable trendswere observed in one condition, the delay wasincreased to 90 s, 120 s, and finally 180 s.Jack and Alex. Although the multielement

designs provided a clear demonstration of therelative efficacy of each condition, there wassome apparent carryover across conditions(e.g., the TR generalized to the TBPD contextwith Nico). Therefore, to isolate the effects ofeach condition better, a multiple baselinedesign across participants was used with Jackand Alex to evaluate the direct effects of aresponse contingency during the delay, whilethe general effects were demonstrated in thesecondary context with each participant.During both CBPD and TBPD, highly pre-

ferred toys were placed on a table where thechildren and the adult sat, and the neutral toysused as the alternative activity were placed in

the corner of the room for Jack or the instruc-tional materials were placed on the table forAlex. The direct effects of TBPD and CBPDwere evaluated in Context 1, and general effectswere evaluated in Context 2 using terminaldelay probes (described below). TBPD wasintroduced first followed by the no-delay base-line, TR training, and finally CBPD. Duringboth TBPD and CBPD conditions in Context1, terminal delay baseline probes were con-ducted on every fifth session in Context2. Finally, CBPD was implemented in Con-text 2.Terminal delay probes (generality test). This

condition was arranged with Jack and Alex toevaluate the extent to which treatment effectswould generalize to a context in which prob-lem behavior during the delay was reinforced(i.e., no extinction during the delay). Context2 (the analyst context for Jack and the dino-saur context for Alex) was designated as thegeneralization context. The terminal delays ofapproximately 5 min for Jack and 10 min forAlex were used during these probes. All prob-lem behavior before the emission of FCR wasplaced on extinction. However, any instancesof problem behavior after the delivery of thedenial cue terminated the delay and resultedin the immediate delivery of the reinforcers.The alternative activity or demands wereavailable throughout this condition; therewas, however, no engagement or compliancerequirement. If no problem behavior wasemitted during the delay, the reinforcers weredelivered at the end of the scheduled terminaldelay. This condition served as a rigorous testof the generality of delay treatments, giventhat direct reinforcement of problem behaviorwas programmed.Jack. The general TBPD and CBPD proce-

dures described above were implemented withJack. The terminal delay was set at 256 s. Thedelay response contingency used during CBPDincluded emitting the tolerance response andengagement in any activity without engaging in

563DELAY TOLERANCE

problem behavior or collateral responses(i.e., both a DRA and a DRO contingency) fora cumulative amount of time.During both conditions, after the delay cue,

a single vocal prompt to engage with the alter-native activity (i.e., “play with [activity] if youwant”) was issued. No additional prompts wereused throughout the delay. During CBPD, Jackwas required to say “okay” in response to thedelay signal and access to the high-preferencetoys and attention was withheld until he metthe target cumulative amount of time notengaging in any problem behavior or collateralresponses. During both TBPD and CBPD,delay levels were increased after one session ifno problem behavior or collateral responsesoccurred, and after two to four sessions if anyof these responses occurred. The comparativeanalysis was concluded at the delay of 256 s,after which CBPD was also implemented inContext 2 and also with a new adult (Jack’sfather).Alex. The general TBPD and CBPD proce-

dures described above were implemented withAlex. The terminal delay was set at 600 s orroughly 50 age-appropriate demands. Giventhat problem behavior was at least partly main-tained by escape from demands, the delayresponse contingency used during CBPDincluded emitting the tolerance response andcompliance with a fixed number of adultinstructions without engaging in problembehavior or attempts to control (i.e., both aDRA and a DRO contingency).During both conditions, following the delay

cue (“not right now” or “wait”), demands(e.g., “write J,” “Color the bird blue”) were pre-sented and three-step prompting (vocal, model,full physical) was used to ensure compliancewith demands. During CBPD, after the delaysignal, Alex was required to say “okay” andcomply with a cumulative number of adultinstructions without engaging in any problembehavior or collateral responses. At the begin-ning, he was required to say “okay” and sit

facing the therapist or the demand materials.Starting at the 4-s delay, a demand was addedto this response chain. The number of demandswas then increased using a geometric progres-sion (1, 2, 4, …). During the delay, demandscontinued until Alex complied with the targetnumber of demands. In both conditions, delaylevels were increased after one session with noproblem behavior or collateral responses, orafter two sessions if there were any instances ofthese behaviors. The comparative analysis wasconcluded at the delay of 64 s or 16 demands.We then merged both contexts into one and

further extended the delay and demand require-ments to 32 demands. The response contin-gency during the delay was then changed fromcompliance with a cumulative number ofdemands to a consecutive number. To signalthis change in the contingency, tokens (check-marks) were introduced. Alex initially earned acheckmark for each demand he completed andlost all checkmarks earned in each section if heengaged in any problem behavior or collateralresponses. Finally, the demand requirementwas changed to a variable ratio of 50 demands.Checkmarks were earned for an average ofthree demands completed, and a total of16 checkmarks were required to earn 120 s ofreinforcement (i.e., an average of 24 demandsneeded to be completed in a row without anyproblem behavior or collateral responses to earnreinforcement time).

ResultsNico. The no-delay baseline showed that tar-

get FCRs occurred at an optimal rate, problembehavior was at zero, and no TRs occurred(Figure 3). After TR training and with theintroduction of progressive delay, target FCRsslowly decreased but remained near the optimalrate in both conditions. The TR was observedin both conditions; however, they occurred atan optimal rate in CBPD whereas excessiveamounts were emitted during TBPD. After a


few sessions of CBPD, Nico spent approxi-mately 80% of the delay engaging in the alter-native activity (third panel). Engagement in thealternative activity did not occur for several ses-sions in TBPD and then never exceeded 75%.Collateral responses and problem behavior werehighly variable but occurred almost exclusivelyin TBPD, despite the fact that the experienceddelays were similar across CBPD and TBPD.Overall, it appeared that CBPD was moreeffective than TBPD at increasing Nico’s toler-ance for delayed reinforcement. CBPD treat-ment effects maintained as the delay wasextended to an average of 5 min and treatmentwas implemented by Nico’s parents (data avail-able from the second author).

Will. During the no-delay baseline in Will’sinitial analysis (Figure 4), target FCRs occurredat an optimal rate, and problem behavior(i.e., SIB) and collateral responses were at zeroor near-zero levels. FCRs decreased but main-tained at an optimal rate, and no engagementin the alternative activity was observed duringTBPD. Problem behavior remained low ini-tially; however, as the delays increased, collat-eral responses such as grabbing others andswiping materials emerged and maintained andSIB resurged. The return to the no-delay base-line resulted in an immediate reduction of SIBand collateral responses, and FCRs persisted.The introduction of CBPD resulted in a grad-ual reduction of FCRs toward an optimal rate,high and variable engagement in the alternativeactivity, and zero levels of SIB and collateralresponses, but these latter behaviors resurged asthe demand requirements were increased. Dueprimarily to the resurgence of SIB and collateralresponses as the delays were increased, neitherCBPD nor TBPD was effective in developingtolerance for delayed reinforcement with Will.Will’s limited fine-motor repertoire and

independent play skills may have contributedto the moderate level of engagement in thealternative activity, which in turn may havecontributed to the resurgence of problembehavior. He also may have required a slowerprogression of the response requirement duringthe delay to allow him to acquire the beadingskills relevant to the alternative activity. Theseconsiderations informed the second analysis,the results of which are depicted in Figure 5.The optimal number of FCRs per reinforcer ineach condition was one; this was obtained dur-ing most DRA sessions. By contrast, Willemitted twice that many FCRs during TBPDand DRO. Alternative activity engagement wasexclusively observed during DRA. Both SIBand collateral responses occurred at higher ratesduring TBPD and DRO than during DRA.These patterns persisted as the delay increasedto 180 s. Overall, CBPD using a DRA

Figure 3. Results of the time-based versuscontingency-based comparative analysis for Nico.

565DELAY TOLERANCE

contingency was more effective for increasingWill’s tolerance for delayed reinforcement thana time-based or DRO-based contingency.Jack. During the no-delay baseline in both

contexts, FCRs occurred at an optimal rate,and problem behavior and collateral responseswere at zero (Figure 6). With TBPD in Con-text 1, there was an increase in the rate ofFCRs with a spike at the 16-s delay. Althoughsome engagement in the alternative activity was

observed as the delay was increased to 256 s,target FCRs were emitted at a higher than opti-mal rate during the majority of TBPD sessions.Problem behavior occurred at high and variablerates throughout TBPD, and collateralresponses (e.g., crying) increased as the delaywas increased. Problem behavior also remainedat strength in Context 2. Overall, TBPD didnot produce tolerance for delayed reinforce-ment. The return to the no-delay baseline in

Figure 4. Results of the time-based versus contingency-based comparative analysis for Will.


both contexts resulted in the elimination ofproblem behavior and collateral responses, andoptimal rates of FCRs.After TR training, the introduction of

CBPD in Context 1 resulted in near-optimalrates of FCRs and TRs, low but persistentengagement in the alternative activity, and con-tinued zero rates of problem behavior and col-lateral responses as the delay was increased to5 min. The DRA-DRO contingency duringdelay for Jack required him to emit the TR andthen refrain from engaging in any problembehavior or collateral responses for the requiredamount of time (i.e., there was no requirementto engage with the alternative activity). In addi-tion, the data from Context 2 provide evidenceof the generality of CBPD training. While Jackexperienced TBPD in Context 1, he

consistently experienced a shorter delay thanthat scheduled in Context 2 because he termi-nated the delay through problem behavior.Despite the presence of the same “inappropri-ate” contingency in Context 2, Jack toleratedthe scheduled delay when CBPD was pro-grammed in Context 1, even though the delaysin Context 2 could have been terminated atany point by engaging in problem behavior.The TR of “okay” as well as other appropriateplay responses generalized, and presumably as aresult, lower rates of problem behavior andlower rates of FCR were observed in this sec-ond context. There were, however, some resid-ual collateral responses; these were eliminatedafter implementation of CBPD in Context2. In summary, CBPD was an effective treat-ment for increasing Jack’s tolerance for delayed

Figure 5. Results of the comparison between time-based versus DRO-based versus DRA-based delay for Will. Thescheduled delay increased at each phase line.

567DELAY TOLERANCE

reinforcement. These treatment effects main-tained when treatment was extended andimplemented by Jack’s father (data availablefrom the second author).

Alex. During the no-delay baseline in bothcontexts, FCRs occurred at an optimal rate,and problem behavior and collateral responseswere at zero (Figure 7). With the introduction

Figure 6. Results of the time-based versus contingency-based comparative analysis for Jack.


of TBPD in Context 1, there was a gradualdecrease in the rate of FCRs corresponding tothe optimal rate, but compliance with demandsremained at zero throughout this condition,and there was an immediate increase in the rateof problem behavior with a spike at the delayof 16 s, at which point TBPD was terminated.Collateral responses also increased starting atthe delay of 4 s. Overall TBPD was not aneffective treatment for increasing tolerance fordelayed reinforcement with Alex. The return tothe no-delay baseline in both contexts resultedin the elimination of problem behavior, andFCRs persisted. After TR training, the intro-duction of CBPD in Context 1 resulted inoccurrences of the TR, a gradual reduction ofFCRs toward an optimal rate, high but variablelevels of compliance during delays, and near-zero rates of problem behavior throughout. Inaddition, the data from Context 2 provide evi-dence of the generality of CBPD training.While Alex experienced TBPD in Context1, he complied with no demands in Context2 and terminated the delay through problembehavior. By contrast, while Alex experiencedCBPD in Context 1 and despite the presenceof the same “inappropriate” contingency inContext 2, he emitted the TR and compliedwith almost half of the demands presented inContext 2 before he engaged in problem behav-ior to terminate the delay. When CBPD wasintroduced in Context 2, high and stable levelsof compliance were achieved, FCRs and TRspersisted, and problem behavior and collateralresponses occurred at zero or near-zero levels.By the end of treatment, Alex engaged inapproximately 50 demands and experienceddelays to reinforcement of approximately10 min with CBPD.

GENERAL DISCUSSION

Contingency-based delays were more effec-tive than time-based delays in developing parti-cipants’ tolerance for delayed reinforcement.

CBPD increased alternative activity engage-ment and compliance while it maintained zeroor near-zero rates of problem behavior and col-lateral responses and optimal rates of communi-cation. Our finding that TBPD is an ineffectivemethod for increasing the generality of FCTtreatment is consistent with previous research(Fisher et al., 2000; Hagopian et al., 1998;Hanley et al., 2001). In fact, despite the variousprocedural improvements to the manner inwhich TBPD is usually programmed, TBPDwas still found to be ineffective in our study.For example, although, the addition of probabi-listic reinforcement appeared to result in themaintenance of the communication responseduring TBPD, problem behavior resurged in allcases and within the first 16 s of delay for threeof four cases. Although the recovery of problembehavior during this delayed reinforcement pro-cedure is likely due to resurgence, as suggestedby Lieving and Lattal (2003) and Volkertet al. (2009), we did not arrange for the neces-sary controls to label this recovery as resurgencewith confidence instead of other extinction-related phenomena (Bruzek, Thompson, &Peters, 2009). Furthermore, the mere presenceof an alternative activity and prompts to engagein these activities or comply with demandswere not sufficient to mitigate the negative sideeffects of TBPD. Therefore, it appears that theresponse contingency during the delay is thenecessary component for the effectiveness ofprogressive delay-tolerance training. Our find-ing, that the presence of a contingency in addi-tion to the alternative activity during the delayis important for achieving delay tolerance, isconsistent with the findings from translationalresearch on self-control by Mischel, Ebbesen,and Raskoff Zeiss (1972), Dixon and Cum-mings (2001), and Dixon, Rehfeldt, and Ran-dich (2003). For example, Dixon andCummings have shown that requiring partici-pants to engage in an alternative response dur-ing delay aids in shifting preference from thesmaller immediate reinforcer to the larger

569DELAY TOLERANCE

Figure 7. Results of the time-based versus contingency-based comparative analysis for Alex.


delayed reinforcer while lower levels of problembehavior are maintained.The effects of CBPD were systematically

replicated across participants aged 21 monthsto 30 years old, with and without developmen-tal disabilities or autism, and across bothsocial-positive and social-negative reinforcementcontexts, replicating the results of Hanleyet al. (2014). In addition to the elimination ofproblem behavior during treatment, CBPDresulted in the acquisition of an appropriateresponse to a delay signal (i.e., the TR of“okay”) and a set of developmentally appropri-ate responses (e.g., compliance with academicdemands, functional play skills) that also gener-alized to a context in which extinction was notfully in place. In other words, with CBPD wewere able to shape a repertoire of “waiting” thatgeneralized to contexts with an inappropriatecontingency (i.e., availability of reinforcementfor problem behavior during delay). This shap-ing of a waiting repertoire was systematicallyreplicated across participants with varyingdegrees of baseline language, adaptive, and lei-sure skills. Desirable patterns of behavior dur-ing extended delays were produced for allparticipants without the need for positive pun-ishment (Fisher et al., 1993; Hagopian et al.,1998) or additional noncontingent or differen-tial reinforcement procedures (Hagopian, Con-trucci Kuhn, Long, & Rush, 2005; Rookeret al., 2013) during the delay.The relative speed with which these treat-

ment effects were obtained (2 to 8 hr distribu-ted across 4 to 24 weeks) suggests that CBPDmay be a desirable alternative to long-termFCT treatment necessary for persistence ofeffects during extinction (Wacker et al., 2011).Wacker et al. (2011) noted that the antecedentsand consequences surrounding the response inthe natural environment often vary from thespecific conditions used during treatment. Theysuggest that variables that enhance treatmentpersistence such as extensive experience withFCT (nearly 16 months) should be included in

treatment. Another variable that may play arole in the persistence of treatment effects isthe manner in which antecedents and conse-quences are arranged during treatment. Ourfindings suggest that alterations to the design oftreatment such as those included in CBPDmay be an efficient means of obtaining similarresistance in light of changing contexts.Although FCT, by the nature of its design,

exposes the individual to the natural maintain-ing contingencies of the response, exposing theindividual to sufficient exemplars of antecedentconditions may also be important to ensuregeneralized responding (Tiger et al., 2008).Rather than relying on the use of a single,tightly controlled context with a specific taskand clear discriminative stimuli for the delay,we used multiple exemplars of delay cues to sig-nal the onset of delay, a variable array of activ-ities and tasks based on child-selected itemsfrom the preference assessment, and adult-selected demand items that changed every twoto four sessions. Within 8 hr of treatment, forexample, Alex was able to engage in appropriatecommunication and tolerance responses in thepresence of a variety of evocative situations(e.g., interruption of drawing activity, removalof toys, denial of a request) and to toleratedelays of approximately 10 to 15 min andcomplete roughly 50 demands that involvedvarious academic and toy-based activities.The specific response requirements during

the delay were also closely matched to thebehavioral expectations regularly experiencedby the participants. The selection of the alter-native activities and the most appropriateprompting procedures, in addition to the spe-cific evocative contexts and reinforcers, wereguided by the results of the open-ended inter-views with caregivers. For example, Jack’smother reported that a common situationinvolved her preparing dinner and requiringJack to stay away from the stove and do “some-thing else,” with very little supervision or atten-tion. Given this context and Jack’s lack of an

571DELAY TOLERANCE

independent play repertoire, a DRO contin-gency seemed the most suitable. It allowed Jackto engage in a variety of other responses with-out any prompting from his mother and stillsatisfied the mother’s request that he stay awayfrom her cooking area for a few minutes. Theseconsiderations increased the ecological validityof treatment, which may further enhance themaintenance of treatment effects. We also useddelay cues that were commonly presented inthe natural environment (i.e., “wait,” “in aminute,” “not right now”) to increase the simi-larity of the training context and the contexttypically experienced by the participant.Finally, for two participants, treatment wassequentially implemented in a second context,and for all participants, treatment was extendedto the relevant context and training was pro-vided to the caregivers who would be responsi-ble for treatment maintenance.Although other procedures have been shown

to maintain low levels of problem behavior dur-ing planned delays of practical duration, theseeffective procedures also rely on strong contin-gencies. For example, Luczynski and Hanley(2014) showed that the strong positive contin-gencies within multiple schedules were respon-sible for their efficacy. Multiple schedules areoften used in the treatment of positively rein-forced problem behavior (Hagopian et al.,2011), whereas chained schedules are oftenused to treat escape-motivated problem behav-ior (e.g., Hagopian et al., 1998). Multiple sche-dules have been shown to retain zero or lowlevels of problem behavior and sufficient levelsof communication even when nonreinforce-ment periods are scheduled for up to 80% ofthe observation period (Betz, Fisher, Roane,Mintz, & Owen, 2013; Fisher et al., 1998;Hagopian et al., 2011; Hanley et al., 2001).Multiple schedules, however, have often beenprogrammed using somewhat artificial stimuli(but see Kuhn, Chirighin, & Zelenka, 2010)such as different-colored cards that must beprogrammed in the natural environment

(Hagopian et al., 2011; Hanley et al., 2001).In addition, obtaining stimulus control overthe occurrence of FCR can be difficult, some-times resulting in high rates of FCR during theextinction component and some recovery ofproblem behavior as the extinction componentis increased (see Hanley et al., 2001, for exam-ples). Given the current state of evidence, adirect comparison of multiple schedules andCBPD is warranted. In particular, it is impor-tant to evaluate the extent to which caregiversare able to maintain treatment integrity witheach procedure and whether they prefer oneover the other. The recipient’s preference forthese procedures should also be directly evalu-ated and considered.Chained schedules have historically been

referred to as demand fading and have beenused to treat negatively reinforced problembehavior (Hagopian et al., 2011; Lalli et al.,1995). Although supplemental strategies(including punishment) have often been neces-sary to achieve the desired outcomes withdemand fading (Hagopian et al., 1998), morerecent evaluations by Falcomata et al. (2013)and Falcomata, Roane, Muething, Stephenson,and Ing (2012) have been conducted in whichvarious elements of both multiple schedules(the discriminative stimuli) and chained sche-dules (the contingency-based alternation of thecomponent change) were used to treat problembehavior maintained by a synthesis of both pos-itive and negative reinforcement. The contin-gency arranged in the traditional chainedschedules is somewhat different than thearrangement used in CBPD. Chained schedulesused by Hagopian et al. (1998), Lalliet al. (1995), and Falcomata et al. (2013) canbe represented as an FR x FR 1 schedule, inwhich a certain number of demands are com-pleted, after which the communicationresponse is reinforced immediately. CBPD, bycontrast, can be represented as an FR 1 FRx schedule, in which the communicationresponse is followed by a chain of responses


that are progressively increased to accommodatethe length of delay necessary, includingunplanned delays that may naturally occur.This arrangement also allows the recursiveimplementation of treatment. This is impor-tant, given the continuous nature of interac-tions between individuals and their caregivers.For example, a child may request a break fromwork and be told to wait and finish his home-work first. After he is done, the child is pro-vided with a break with his toys. During thisbreak, however, the child may ask that hismother play along with him. At this point, themother can again repeat the treatment proce-dure and ask the child to wait and play alonewhile she finishes her cooking. When themother joins the play, however, the child maymake another request for a drink, which themother may immediately reinforce. In this way,CBPD can be practiced continuously because ithas a natural fit with common situations.The CBPD procedure, as described in this

study, is not without its limitations. Some par-ticipants’ performance, in particular whendemands were presented, required monitoringduring the delay. For example, Will requiredintermittent prompting to continue beading.Alex’s treatment included discrete presentationof demands and three-step prompting (instruc-tion, model, physical). The need for continu-ous monitoring may present a barrier toimplementation when caregivers are busy withother tasks or other individuals. One possibleextension of this research could involve evaluat-ing the use of product monitoring as the crite-rion for the contingent delivery ofreinforcement. Another strategy that couldimprove the practicality of CBPD involves theaddition of self-monitoring of performance(Connell, Carta, & Baer, 1993). Individualscould be taught to self-assess and to recruitreinforcement when a task is complete. Thisstrategy could reduce the amount of monitor-ing that caregivers must provide and increasean individual’s independent task engagement

during delays. Finally, the efficacy of a DRA-based contingency using momentary and spo-radic monitoring remains to be assessed.Some additional questions arise from the

manner in which CBPD was programmed inthis study. One question concerns the predicta-bility of the delay. Predictability can be definedin various ways. One aspect is related to the rel-ative proportion of delayed and immediatereinforcement for FCRs. A second aspect relatesto the extent to which the duration of eachdelay requirement is fixed or variable. Predicta-bility may also involve cues that inform theparticipant of the ensuing delay requirement(e.g., contingency-specifying statements or vis-ual cues such as token boards). Future researchshould examine the impact of predictable ver-sus unpredictable delay terminationrequirements.The main advantage of CBPD lies in its abil-

ity to create desirable patterns of behavior whileit emulates situations that involve unplanneddelays and in its ability to yield generalizablepatterns of behavior that appear to protect indi-viduals from mismanaged contingencies (seealso Luczynski & Hanley, 2013). Future inves-tigations into the procedural variations thatmay enhance the efficacy of this treatment, itsgenerality, and its social validity are stillwarranted.

REFERENCES

Betz, A. M., Fisher, W. W., Roane, H. S.,Mintz, J. C., & Owen, T. M. (2013). A componentanalysis of schedule thinning during functional com-munication training. Journal of Applied BehaviorAnalysis, 46, 219–241. doi: 10.1002/jaba.23

Bowman, L. G., Fisher, W. W., Thompson, R. H., &Piazza, C. C. (1997). On the relation of mands andthe function of destructive behavior. Journal ofApplied Behavior Analysis, 30, 251–265. doi:10.1901/jaba.1997.30-251

Bruzek, J. L., Thompson, R. H., & Peters, L. C. (2009).Resurgence of infant caregiving responses. Journal ofthe Experimental Analysis of Behavior, 92, 327–343.doi: 10.1901/jeab.2009-92-327

573DELAY TOLERANCE

http://dx.doi.org/10.1002/jaba.23

http://dx.doi.org/10.1901/jaba.1997.30-251

http://dx.doi.org/10.1901/jeab.2009-92-327

Carr, E. G., & Carlson, J. I. (1993). Reduction of severebehavior problems in the community using a multi-component treatment approach. Journal of AppliedBehavior Analysis, 26, 157–172. doi: 10.1901/jaba.1993.26-157

Carr, E. G., & Durand, V. M. (1985). Reducing behaviorproblems through functional communication train-ing. Journal of Applied Behavior Analysis, 18,111–126. doi: 10.1901/jaba.1985.18-111

Connell, M. C., Carta, J. J., & Baer, D. M. (1993). Pro-gramming generalization of in-class transition skills:Teaching preschoolers with developmental delays toself-assess and recruit contingent teacher praise. Jour-nal of Applied Behavior Analysis, 26, 345–352. doi:10.1901/jaba.1993.26-345

DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of amultiple-stimulus presentation format for assessingreinforcer preferences. Journal of Applied BehaviorAnalysis, 29, 519–533. doi: 10.1901/jaba.1996.29-519

Dixon, M. R., & Cummings, A. (2001). Self-control inchildren with autism: Response allocation duringdelays to reinforcement. Journal of Applied BehaviorAnalysis, 34, 491–495. doi: 10.1901/jaba.2001.34-491

Dixon, M. R., Rehfeldt, R. A., & Randich, L. (2003).Enhancing tolerance to delayed reinforcers: The roleof intervening activities. Journal of Applied BehaviorAnalysis, 36, 263–266. doi: 10.1901/jaba.2003.36-263

Durand, V. M., & Moskowitz, L. (2015). Functionalcommunication training: Thirty years of treatingchallenging behavior. Topics in Early Childhood Spe-cial Education, 35, 116–126. doi: 10.1177/0271121415569509

Falcomata, T. S., Muething, C. S., Gainey, S.,Hoffman, K., & Fragale, C. (2013). Further evalua-tions of functional communication training andchained schedules of reinforcement to treat multiplefunctions of challenging behavior. BehaviorModification, 37, 723–746. doi: 10.1177/0145445513500785

Falcomata, T. S., Roane, H. S., Muething, C. S.,Stephenson, K. M., & Ing, A. D. (2012). Functionalcommunication training and chained schedules ofreinforcement to treat challenging behavior main-tained by terminations of activity interruptions.Behavior Modification, 36, 630–649. doi: 10.1177/0145445511433821

Fisher, W. W., Kuhn, D. E., & Thompson, R. H.(1998). Establishing discriminative control ofresponding using functional and alternative reinfor-cers during functional communication training. Jour-nal of Applied Behavior Analysis, 31, 543–560. doi:10.1901/jaba.1998.31-543

Fisher, W., Piazza, C., Cataldo, M., Harrell, R.,Jefferson, G., & Conner, R. (1993). Functional com-munication training with and without extinction and

punishment. Journal of Applied Behavior Analysis, 26,23–36. doi: 10.1901/jaba.1993.26-23

Fisher, W. W., Thompson, R. H., Hagopian, L. P.,Bowman, L. G., & Krug, A. (2000). Facilitating tol-erance of delayed reinforcement during functionalcommunication training. Behavior Modification, 24,3–29. doi: 10.1177/0145445500241001

Ghaemmaghami, M., Hanley, G. P., Jin, S. C., &Vanselow, N. R. (2016). Affirming control by multi-ple reinforcers via progressive treatment analysis.Behavioral Interventions, 31, 70–86. doi: 10.1002/bin.1425

Hagopian, L. P., Boelter, E. W., & Jarmolowicz, D. P.(2011). Reinforcement schedule thinning followingfunctional communication training: Review andrecommendations. Behavior Analysis in Practice,4, 4–16.

Hagopian, L. P., Contrucci Kuhn, S. A., Long, E. S., &Rush, K. S. (2005). Schedule thinning followingcommunication training: Using competing stimuli toenhance tolerance to decrements in reinforcer density.Journal of Applied Behavior Analysis, 38, 177–193.doi: 10.1901/jaba.2005.43-04

Hagopian, L. P., Fisher, W. W., Sullivan, M. T.,Acquisto, J., & LeBlanc, L. A. (1998). Effectivenessof functional communication training with and with-out extinction and punishment: A summary of21 inpatient cases. Journal of Applied BehaviorAnalysis, 31, 211–235. doi: 10.1901/jaba.1998.31-211

Hammond, L. J. (1980). The effect of contingency uponthe appetitive conditioning of free-operant behavior.Journal of the Experimental Analysis of Behavior, 34,297–304. doi: 10.1901/jeab.1980.34-297

Hanley, G. P. (2012). Functional assessment of problembehavior: Dispelling myths, overcoming implementa-tion obstacles, and developing new lore. BehaviorAnalysis in Practice, 5, 54–72.

Hanley, G. P., Iwata, B. A., & Thompson, R. H. (2001).Reinforcement schedule thinning following treatmentwith functional communication training. Journal ofApplied Behavior Analysis, 34, 17–38. doi: 10.1901/jaba.2001.34-17

Hanley, G. P., Jin, C. S., Vanselow, N. R., &Hanratty, L. A. (2014). Producing meaningfulimprovements in problem behavior of children withautism via synthesized analyses and treatments. Jour-nal of Applied Behavior Analysis, 47, 16–36. doi:10.1002/jaba.106

Hernandez, E., Hanley, G. P., Ingvarsson, E. T., &Tiger, J. H. (2007). A preliminary evaluation of theemergence of novel mand forms. Journal of AppliedBehavior Analysis, 40, 137–156. doi:10.1901/jaba.2007.96-05

Jessel, J., Hanley, G. P., & Ghaemmaghami, M. (2016).Defining and improving the efficiency and control ofa functional analysis of a problem behavior. Journal ofApplied Behavior Analysis.












http://dx.doi.org/10.1177/0271121415569509

http://dx.doi.org/10.1177/0271121415569509

http://dx.doi.org/10.1177/0145445513500785

http://dx.doi.org/10.1177/0145445513500785

http://dx.doi.org/10.1177/0145445511433821

http://dx.doi.org/10.1177/0145445511433821



http://dx.doi.org/10.1177/0145445500241001

http://dx.doi.org/10.1002/bin.1425

http://dx.doi.org/10.1002/bin.1425




http://dx.doi.org/10.1901/jeab.1980.34-297






Kuhn, D. E., Chirighin, A. E., & Zelenka, K. (2010).Discriminated functional communication: A proce-dural extension of functional communicationtraining. Journal of Applied Behavior Analysis, 43,249–264. doi: 10.1901/jaba.2010.43-249

Kurtz, P. F., Boelter, E. W., Jarmolowicz, D. P.,Chin, M. D., & Hagopian, L. P. (2011). An analysisof functional communication training as an empiri-cally supported treatment for problem behavior dis-played by individuals with intellectual disabilities.Research in Developmental Disabilities, 32,2935–2942. doi: 10.1016/j.ridd.2011.05.009

Lalli, J. S., Casey, S., & Kates, K. (1995). Reducingescape behavior and increasing task completion withfunctional communication training, extinction, andresponse chaining. Journal of Applied BehaviorAnalysis, 28, 261–268. doi: 10.1901/jaba.1995.28-261

Lerman, D. C., & Iwata, B. A. (1996). Developing atechnology for the use of operant extinction in clini-cal settings: An examination of basic and appliedresearch. Journal of Applied Behavior Analysis, 29,345–382. doi: 10.1901/jaba.1996.29-345

Lieving, G. A., & Lattal, K. A. (2003). Recency, repeata-bility, and reinforcer retrenchment: An experimentalanalysis of resurgence. Journal of the ExperimentalAnalysis of Behavior, 80, 217–233. doi: 10.1901/jeab.2003.80-217

Luczynski, K. C., & Hanley, G. P. (2013). Prevention ofproblem behavior by teaching functional communica-tion and self-control skills to preschoolers. Journal ofApplied Behavior Analysis, 46, 355–368. doi:10.1002/jaba.44

Luczynski, K. C., & Hanley, G. P. (2014). How shouldperiods without social interaction be scheduled? Chil-dren’s preference for practical schedules of positive

reinforcement. Journal of Applied Behavior Analysis,47, 500–522. doi: 10.1002/jaba.140

Mischel, W., Ebbesen, E. B., & Raskoff Zeiss, A. (1972).Cognitive and attentional mechanisms in delay ofgratification. Journal of Personality and SocialPsychology, 21, 204–218. doi: 10.1037/h0032198

Rooker, G. W., Jessel, J., Kurtz, P. F., & Hagopian, L. P.(2013). Functional communication training with andwithout alternative reinforcement and punishment:An analysis of 58 applications. Journal of AppliedBehavior Analysis, 46, 708–722. doi: 10.1002/jaba.76

Tiger, J. H., Hanley, G. P., & Bruzek, J. (2008). Func-tional communication training: A review and practi-cal guide. Behavior Analysis in Practice, 1, 16–23.

Volkert, V. M., Lerman, D. C., Call, N. A., & Trosclair-Lasserre, N. (2009). An evaluation of resurgenceduring treatment with functional communicationtraining. Journal of Applied Behavior Analysis, 42,145–160. doi: 10.1901/jaba.2009.42-145

Vollmer, T. R., Borrero, J. C., Lalli, J. S., & Daniel, D.(1999). Evaluating self-control and impulsivity inchildren with severe behavior disorders. Journal ofApplied Behavior Analysis, 32, 451–466. doi:10.1901/jaba.1999.32-451

Wacker, D. P., Harding, J. W., Berg, W. K., Lee, J. F.,Schieltz, K. M., Padilla, Y. C., … Shahan, T. A.(2011). An evaluation of persistence of treatmenteffects during long-term treatment of destructivebehavior. Journal of the Experimental Analysis ofBehavior, 96, 261–282. doi: 10.1901/jeab.2011.96-261

Received May 14, 2015Final acceptance December 3, 2015Action Editor, John Borrero

575DELAY TOLERANCE


http://dx.doi.org/10.1016/j.ridd.2011.05.009








http://dx.doi.org/10.1037/h0032198






Copyright of Journal of Applied Behavior Analysis is the property of Wiley-Blackwell and itscontent may not be copied or emailed to multiple sites or posted to a listserv without thecopyright holder's express written permission. However, users may print, download, or emailarticles for individual use.

contingencies promote delay tolerance · 5/14/2015 · tolerance for delays to reinforcement (here...

Documents