OPERANT BEHAVIOR Summary

A TRANSLATIONAL ANALYSIS OF ABA AND ABC RENEWAL OF OPERANT BEHAVIOR

CLARE J. LIDDON UNIVERSITY OF NORTH FLORIDA

MICHAEL E. KELLEY THE SCOTT CENTER FOR AUTISM TREATMENT, SCHOOL OF BEHAVIOR ANALYSIS, FLORIDA INSTITUTE OF

TECHNOLOGY

CATALINA N. REY KENNEDY KRIEGER INSTITUTE AND JOHNS HOPKINS UNIVERSITY SCHOOL OF MEDICINE

ASHLEY P. LIGGETT MUNROE MEYER INSTITUTE, UNIVERSITY OF NEBRASKA MEDICAL CENTER

AURELIA RIBEIRO THE SCOTT CENTER FOR AUTISM TREATMENT, SCHOOL OF BEHAVIOR ANALYSIS, FLORIDA INSTITUTE OF

TECHNOLOGY

Following reduction, behavior may recur when organisms are exposed to new or different envi- ronments. Such recurrences, called treatment relapse, are characterized by an increase of a previ- ously reduced behavior. Renewal is a type of treatment relapse resulting from changes in stimulus contexts despite the ongoing maintenance of treatment. Renewal types include varying arrangements, such as ABA and ABC renewal, where each letter represents a different context. In the present series of translational analyses, we evaluated both ABA and ABC renewal and assessed whether these two renewal arrangements may be present in a single participant’s behav- ior. Results produced evidence of both ABA and ABC renewal, and both types of renewal were detected in an individual participant’s behavior. Key words: ABA renewal, ABC renewal, context, problem behavior, treatment relapse

Behavioral interventions are effective in the assessment and treatment of problem behavior (Foxx, 1996) and the acquisition and mainte- nance of adaptive behavior (e.g., Peters-Schef- fer, Didden, Korzilius, & Sturmey, 2011). Though initially effective, treatments often encounter challenges to their maintenance. For

example, treatments may be challenged by errors of omission or commission (St. Peter Pipkin, Vollmer, & Sloman, 2010), inadvertent application of extinction, schedule thinning (e.g., Hanley, Iwata, & Thompson, 2001; Volkert, Lerman, Call, & Trosclair-Lasserre, 2009), or various schedule or response manipu- lations (Horner & Day, 1991). Behavior ana- lysts seek to implement treatments that are robust against treatment challenges and persist despite manipulations that might compromise their efficacy. Some intervention challenges may produce

various types of treatment relapse, such as resurgence, reinstatement, spontaneous recov- ery, and renewal (Bouton, 2002). Each type of

This research was conducted in partial fulfillment of the dissertation requirement of the doctor of philosophy degree by the first author. Correspondence concerning this article should be

addressed to Clare J. Liddon, Department of Exceptional, Deaf, and Interpreter Education, Building 57, Suite 3500, University of North Florida, 1 UNF Drive, Jacksonville, FL 32224. Email [email protected] doi: 10.1002/jaba.496

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2018, 51, 819–830 NUMBER 4 (FALL)

© 2018 Society for the Experimental Analysis of Behavior

819

treatment relapse is characterized by a specific procedural arrangement and has specific impli- cations for responding. For example, renewal is defined as the recurrence of previously extin- guished behavior as a function of a change in the stimulus context. There are several different types of procedures for detecting renewal, but all variations generally include a three-phase arrangement. In operant ABA renewal, Context A involves reinforcement contingent on an operant response, followed by operant extinc- tion in Context B, and continued operant extinction upon the return to Context A. The renewal effect is observed when responding recurs upon return to Context A. In operant ABC renewal, Context A involves reinforce- ment contingent on an operant response, fol- lowed by operant extinction in Context B, and continued operant extinction upon the expo- sure to Context C. The renewal effect is observed when responding recurs upon expo- sure to Context C. Both ABA and ABC renewal effects have

been demonstrated in respondent and operant learning of nonhuman animals across areas such as fear conditioning (Bouton & Bolles, 1979; Bouton & King, 1983), lever pressing (Bouton, Todd, Vurbic, & Winterbauer, 2011; Naka- jima, Tanaka, Urushihara, & Imada, 2000), and drug seeking (Bossert, Liu, Lu, & Shaham, 2004; Crombag & Shaham, 2002; Hamlin, Clemens, & McNally, 2008; Hamlin, Newby, & McNally, 2007; Kearns & Weiss, 2007). Demonstration of the renewal effect in humans is relatively less common, and arguably deficient, given the potential for renewal to explain common treatment failures, explain how responses maintain, and provide a frame- work for strengthening interventions. For example, consider an individual who engages in problem behavior in one environment (e.g., home or school; Context A). Following successful treatment in a clinic (Context B), the client is discharged into the original environ- ment (Context A). Results of basic research

suggest that problem behavior may re-emerge post discharge due to the history of reinforce- ment for the behavior in that context, despite the maintenance of the treatment. Conversely, examples of beneficial renewal may also be identified in the everyday environment. During toilet training, a child learns to void appropri- ately at his primary caregiver’s home (Context A) following reinforcement for appropriate voids. He may then visit another caregiver for an extended period of time (Context B), where appropriate voids are not reinforced, and are thus extinguished. Results of basic research sug- gest that appropriate behavior may re-emerge upon the child’s returning home due to the his- tory of reinforcement for the behavior in that context, despite the absence of previously pre- sent reinforcement contingencies. Kelley, Liddon, Ribeiro, Greif, and Podle-

snik (2015) were the first to examine ABA renewal with pigeons and humans within a translational preparation. In Experiment 1, six pigeons were exposed to alternating keylight colors that signaled the contexts. In the first exposure to Context A (RFT), keypecks resulted in a hopper presentation and access to grain and occurred at steady rates. In Context B (EXT), keypecks did not produce grain and extinguished. Upon return to Context A, key- pecks re-emerged despite the continuation of EXT. In Experiment 2, two children diagnosed with autism participated in a similar prepara- tion. Contexts were signaled using colored stimuli (e.g., therapist t-shirts, poster board, and task materials) and a particular therapist for each context. Reponses included engage- ment in mastered, preacademic task completion (e.g., letter tracing and picture sorting). Results of Experiment 2 were similar to those of Exper- iment 1 and demonstrated generality across species to tasks of social significance. Further- more, these results were obtained with a popu- lation that is likely to engage in problem behavior, and thus provide promise for clinical application.

CLARE J. LIDDON et al.820

Though promising, these results are limited to ABA renewal. Additional translations of renewal in humans still need to be extended to other types, such as ABC renewal. Further- more, it has yet to be demonstrated whether a single human’s behavior is subject to multiple types of renewal. The implications for under- standing treatment relapse are greater if renewal is demonstrated as a robust phenomenon across multiple procedural variations (e.g., both ABA and ABC renewal) within a single participant. Therefore, the primary purpose of the present studies was to expand upon the operant renewal literature through a translational analy- sis of ABA and ABC renewal. We aimed first to evaluate the ABA and ABC renewal effect in humans in a translational preparation, possibly providing further support for these effects in humans. Second, we aimed to examine whether both ABA and ABC renewal could be demon- strated within individual participants’ behavior.

METHOD

Participants, Setting and Materials Participants included six individuals diag-

nosed with autism who received services from a university-based, early intervention program. At the time of the study, Bella, Liam, Vaughn, Der- mot, and Cole were 3 years old. Preston was 4 years old. The level of communication skills varied across participants. Bella communicated using complete vocal sentences and frequently engaged in reciprocal conversation with others. Preston communicated using three- to five-word vocalizations. Liam communicated using ges- tures and some one- to two-word vocalizations. Cole, Dermot, and Vaughn were all nonvocal speakers, but communicated through a picture exchange system using single picture icons only. Cole and Vaughn participated in the ABA

arrangement, Bella and Dermot participated in the ABC arrangement, and Liam and Preston participated in the ABC/ABA arrangement. Table 1 depicts the task, reinforcer, and

renewal exposure for each participant. All ses- sions were conducted in a treatment room or classroom of a university-based, early interven- tion clinic. Session rooms included a table, two chairs, a video camera and tripod, and the materials necessary for each condition (e.g., reinforcers, task materials, condition sig- nals, etc.). In the classroom setting, partitions were used to separate the session context from the rest of the classroom. The therapist con- ducting the session was present.

Response Measurement and Interobserver Agreement During all reinforcement and extinction con-

ditions, therapists collected count measures on free operant task completion (i.e., correct plus incorrect, or total responses), using laptop com- puters during each 5-min session. Therapists defined task completion and any other relevant target behaviors independently for each partici- pant, depending on the mastered task chosen. Some examples included sorting two- dimensional stimuli or tracing letters and num- bers. Dermot’s mastered task was depositing blocks into a large bin (i.e., a fine-motor task that did not require discrimination between stimuli), and thus, he did not have the oppor- tunity to engage in an incorrect response. A complete list of tasks by participant is available upon request from the first author. A second, independent observer collected data

during a minimum of 25% of sessions across all conditions, for each participant. Each interval was 10 s in duration, and we calculated interob- server agreement (IOA) by comparing observer records using the exact interval method. Inter- vals in which both observers recorded exactly the same number of responses were counted as agreements. Therapists calculated IOA by divid- ing the number of agreements by the number of agreements plus disagreements for a session and multiplying by 100. IOA was calculated for Bella during 35% of sessions. The mean IOA

821ABA AND ABC RENEWAL

for Bella’s correct responding was 88.6% (range, 50%-100%) and 94.4% for incorrect respond- ing (range, 66.7%-100%). IOA was calculated for Cole during 26% of sessions. The mean IOA was 95% (range, 75%-100%) for Cole’s correct responding and 96% (range, 80%-100%) for incorrect responding. IOA was calculated for Dermot during 26% of sessions of ABC renewal. The mean IOA was 90.3% (range, 71.4%-100%). IOA was calculated for Liam during 26% of sessions of ABC renewal and 30% of sessions for ABA renewal. The mean IOA for Liam’s correct responding was 99.5% during ABC renewal (range, 95.5%-100%) and 94% during ABA renewal (range, 33.3%- 100%). The mean IOA for Liam’s incorrect responding was 100% during ABC renewal and 93.3% during ABA renewal (range, 33.3%- 100%). For Preston, IOA was calculated during 25% of sessions of ABC renewal and 29% of sessions for ABA renewal. The mean IOA for Preston’s correct responding was 96% during ABC renewal (range, 83.3%-100%) and 94% during ABA renewal (range, 75%-100%). The mean IOA for Preston’s incorrect responding was 100% during both ABC and ABA renewal. For Vaughn, IOA was calculated during 29% of sessions. The mean IOA for Vaughn’s correct responding was 91% (range, 80%-100%) and 100% for incorrect responding.

Experimental Design For ABA renewal, participants were exposed

to the stimuli in an ABA design

(i.e., Context A, Context B, Context A) and to the contingencies in an ABB design (reinforce- ment, extinction, extinction). For ABC renewal, participants were exposed to the stim- uli in an ABC design (i.e., Context A, Context B, Context C) and to the contingen- cies in an ABB design (reinforcement, extinc- tion, extinction). Both ABA and ABC renewal are defined by these particular experimental arrangements in previous research (see Bouton, 2002; Kelley et al., 2015).

General Procedures We signaled each context with the presence

of the same color of the following stimuli in the room: shirt worn by the therapist, task materials, and a poster board (e.g., Kelley et al., 2015). We correlated each color with a particu- lar context (e.g., all green stimuli correlated with Context A). In an effort to keep context changes consistent within and across partici- pants, the same therapist conducted sessions across all phases of the study (e.g., the first author served as the only therapist for Dermot) and was the only person present in the room other than the participant. Therapists placed the poster board where it was likely to be attended to by the participant (i.e., on the table as a place mat or hung on the wall in front of the participant).

Pre-Experimental Procedures Mastered task probe assessment. We conducted

an assessment to determine appropriate

Table 1 Participant Exposure, Materials, and Preassessment Reinforcement Schedule

Participant Task Reinforcer Renewal Exposure Preassessment Reinforcement Schedule

Bella Writing DVD player ABC VR 2 Cole Matching Items mandarin oranges ABC FR1 Dermot Depositing Blocks dried cranberries ABA FR2 Liam Matching Items marshmallows ABCABA VR2 Preston Sorting Pictures peanuts ABCABA VR2 Vaughn Depositing Blocks shape sorter ABA FR1

CLARE J. LIDDON et al.822

mastered tasks for the renewal conditions. This consisted of an experimenter arranging free- operant response opportunities (e.g., matching, sorting, letter tracing) and telling the partici- pant, “You can do as many as you want.” Upon task completion, praise was provided on an intermittent schedule (e.g., variable-ratio 2 [VR-2] schedule; see Table 1 for individual schedule values). Sessions were 2 min in dura- tion. Schedule of praise delivery was deter- mined based on the skill level of the participant (e.g., earlier learners had praise delivered on a denser schedule and advanced learners on a lea- ner schedule). Probes were conducted until par- ticipants demonstrated 90% independent, correct responses across three sessions. Data from these assessments are available upon request from the first author.

Experimental Procedures Participants were exposed to the contingen-

cies in ABA renewal, ABC renewal, or both. All sessions lasted 5 min. We exposed participants to each condition with their specified arrange- ment for five sessions or until stable responding was observed. Stable responding was deter- mined through visual analysis of the data. To minimize the likelihood that confounding vari- ables (e.g., spontaneous recovery) might influ- ence the participants’ rate of responding at the beginning of a phase, all condition changes were made within a day. For all participants, during Reinforcement (A), preferred stimuli identified through a paired-stimulus preference assessment (see Fisher et al., 1992) were deliv- ered contingent on a correct response. ABA Renewal: Reinforcement (A). We pre-

sented participants with a free-operant, mas- tered (i.e., previously learned and identified through probe sessions) task in all contexts. In the reinforcement context, the therapist pre- sented the participant with the task while giv- ing the instruction, “You can do as many as you want.” Following a correct response, a

reinforcer was provided on a fixed-ratio 1 (FR- 1) schedule of reinforcement. Reinforcers con- sisted of 15-s access to leisure items or a single, bite-sized portion of an edible (e.g., one third of a mandarin orange slice for Cole). To main- tain the free-operant nature of the tasks, mate- rials remained present when the reinforcer was presented. All sessions were conducted in Con- text A. Table 1 shows a summary of all mate- rials and context-correlated stimuli used for each participant for this and subsequent conditions. Extinction (B). Procedures in this condition

were identical to those in Reinforcement (A) with the exception that no consequences were provided for task completion. All sessions were conducted in Context B. Extinction (A). Procedures in this condition

were identical to those in Extinction (B). All sessions were conducted in Context A. ABC Renewal: Reinforcement (A). Procedures

in this condition were identical to those in the Reinforcement (A) condition of ABA renewal. Extinction (B). Procedures in this condition

were identical to those in the Extinction (B) condition of ABA renewal. Extinction (C). Procedures in this condition

were identical to those in the Extinction (A) condition of ABA renewal. However, all sessions were conducted in Context C.

RESULTS

All renewal data are expressed as total responses per minute (rpm) across each condi- tion for each participant. Both correct and incorrect responses comprise total responses. However, incorrect responses rarely occurred. Across all participants and both renewal types, a renewal effect was determined by a relative increase in responding during the first session of the final condition (i.e., Extinction [A] or [C]), when compared to the final session of Extinction (B). This is consistent with renewal effects identified in previous research

823ABA AND ABC RENEWAL

(e.g., Berry, Sweeny, & Odum, 2014; Bouton et al., 2011; Kelley et al., 2015). Cole’s ABA renewal results are depicted in

Figure 1 (top panel). In Reinforcement (A) of ABA renewal, Cole demonstrated high and var- iable rates of total responding (M = 2.1 rpm, range = 0-2.8). Cole’s rate of total responding decreased (M = 0.6 rpm, range = 0-2.6) in Extinction (B) of ABA renewal. In the return to Extinction (A), Cole’s rate of total respond- ing increased, then returned to near zero levels (M = 0.05 rpm, range = 0-0.2). An ABA renewal effect is evident in the increase in responding for Cole during the first session of Extinction (A; session 16), compared to the final session of Extinction (B; session 15). Vaughn’s ABA renewal results are depicted

in Figure 1 (bottom panel). In Reinforcement (A) of ABA renewal, Vaughn demonstrated

moderate, stable rates of total responding (M = 1.6 rpm, range = 0.4-2.2). In Extinction (B) of ABA renewal, Vaughn’s rate of total responding increased, then decreased (M = 5.5 rpm, range = 0 -13.6). In Extinction (A), Vaughn’s rate of total responding increased, then returned to near-zero levels (M = 2.8 rpm, range = 0-8). An ABA renewal effect is evident in the increase in responding for Vaughn dur- ing the first session of Extinction (A; session 30), compared to the final session of Extinction (B; session 29). Bella’s ABC renewal results are depicted in

Figure 2 (top panel). In Reinforcement (A), Bella demonstrated moderate rates of total responding (M = 1.7 rpm, range = 1.4-2.2). In Extinction (B) of ABC renewal, Bella’s rate of total responding decreased to zero levels (M = 0.6 rpm, range = 0-1.6). In Extinction (C) of ABC renewal, Bella’s rate of responding initially increased, then returned to zero levels (M = 0.7 rpm, range = 0-1.8). An ABC renewal effect is evident in the increase in responding for Bella during the first session of Extinction (C; session 14), compared to the final session of Extinction (B; session 13). Dermot’s ABC renewal results are depicted

in Figure 2 (bottom panel). In Reinforcement (A) of ABC renewal, Dermot demonstrated high and variable rates of total responding (M = 2.5 rpm, range = 1.6-3.6). In Extinction (B) of ABC renewal, Dermot’s rate of total responding decreased (M = 1.5 rpm, range = 0-4. In Extinction (C), Dermot’s rate of total responding increased, then returned to near zero levels (M = 4 rpm, range = 0-1.6). No increase in responding was observed during the first session of Extinction (C; session 23), when compared to the final session of Extinction (B; session 22). Therefore, no ABC renewal effect was found according to our definition. Liam’s ABC and ABA renewal results are

depicted in Figure 3. Data in the upper panel reflect exposure to ABC renewal, and data in the lower panel reflect data from ABA renewal.

Figure 1. Total responses per minute, across all condi- tions, during ABA Renewal for Cole (top panel) and Vaughn (bottom panel).

CLARE J. LIDDON et al.824

In Reinforcement (A) of ABC renewal, Liam demonstrated high and variable rates of total responding (M = 9.1 rpm, range = 0.6-17). In Extinction (B) of ABA renewal, Liam’s rate of total responding decreased (M = 1.4 rpm, range = 0-7.4). In Extinction (C), Liam’s rate of total responding increased, then reduced to zero levels (M = 0.8 rpm, range = 0-2.2). In Reinforcement (A) of ABA renewal, Liam dem- onstrated high and variable rates of total responding (M = 12.8 rpm, range = 10-18.2). In Extinction (B) of ABA renewal, Liam’s rate of total responding decreased (M = 4.3 rpm, range = 0-15.2). In Extinction (A), Liam’s rate of total responding increased, then returned to near zero levels (M = 3.3 rpm, range = 0-12.2). An analysis of Liam’s data reveals an ABC and an ABA renewal effect. This is evi- dent in the increase in responding during the

first session of the final conditions (i.e., Extinction [C] session 30 and Extinction [A] session 34), compared to the final sessions of Extinction (B) for both arrangements (i.e., session 29 for ABC Renewal and session 33 for ABA renewal). Preston’s ABC and ABA renewal results are

depicted in Figure 4. In the upper panel are data from ABC renewal, and the lower panel the data from ABA renewal. In Reinforcement (A) of ABC renewal, Preston demonstrated moderate rates of total responding (M = 1.5 rpm, range = 0.2-2.8). In Extinction (B) of ABA renewal, Preston’s rate of total responding was initially variable, then decreased (M = 1.8 rpm, range = 0-5.4). In the return to Extinc- tion (C), Preston’s rate of total responding remained at near zero levels (M = 0.1 rpm, range = 0-0.2). In Reinforcement (A) of ABA

Figure 2. Total responses per minute, across all condi- tions, during ABC Renewal for Bella (top panel) and Der- mot (bottom panel).

Figure 3. Total responses per minute, across all condi- tions for Liam, during ABC Renewal (top panel) and ABA Renewal (bottom panel).

825ABA AND ABC RENEWAL

renewal, Preston demonstrated high and vari- able rates of total responding (M = 4 rpm, range = 2-7.4). In Extinction (B) of ABA renewal, Preston’s rate of total responding decreased (M = 0.5 rpm, range = 0-2.8). In

Extinction (A), Preston’s rate of total respond- ing increased, then returned to near zero levels (M = 0.4 rpm, range = 0-1.6). An analysis of Preston’s data reveals an ABA renewal effect. This is evident in the increase in responding during the first session of Extinction (A; session 13), compared to the final session of Extinction (B; session 12). Given that no increase in responding was observed during the first session of Extinction (C; session 28) when compared to Extinction (B; session 27), no ABC renewal effect was found according to our definition. Table 2 depicts the renewal effects for each

participant, expressed in an increase in responses per minute from the last session in Extinction (B) to the first session in Extinction (A) or (C). Overall, larger rate increases were observed during ABA renewal than ABC renewal. Proportional changes were generally greater during ABA Renewal than ABC Renewal (see Supporting Information).

DISCUSSION

We observed an ABA renewal effect in all of the four participants’ behavior exposed to that arrangement. We also observed an ABC renewal effect in two of the four participants’ behavior exposed to that arrangement. We observed both an ABA and an ABC renewal effect in one out of two participants’ behavior exposed to both arrangements. Overall, these results replicate previous human studies on ABA renewal (e.g., Kelley et al., 2015). Fur- thermore, these results extend previous findings of operant ABC renewal (e.g., Zironi, Burat- tini, Aicardi, & Janak 2006). Finally, this is the first demonstration of within-subject renewal effects in humans across renewal types. To date, only one study has demonstrated

the presence of both ABA and ABC renewal within subjects. Berry et al. (2014) evaluated the effects of rich and lean reinforcement rates on ABA and ABC operant renewal in pigeons. In Experiments 1 and 2, pigeons were exposed

Figure 4. Total responses per minute, across all condi- tions for Preston, during ABC Renewal (top panel) and ABA Renewal (bottom panel).

Table 2 Renewal Effects

Participant ABA Effect ABC Effect

Bella -- .8 Cole 0.2 -- Dermot -- 0 Liam 12 1.6 Preston 1.6 0 Vaughn 8 --

Note. Numbers expressed as an increase in responses per minute (rpm) from the last data point in Extinction (B) to the first data point in Extinction (A) or (C).

CLARE J. LIDDON et al.826

to rich (variable interval [VI] 30s) and lean (VI 120 s) schedules of reinforcement followed by two subsequent phases of extinction while the context was manipulated: an ABA renewal context arrangement in Experiment 1 and an ABC renewal context arrangement in Experi- ment 2. Seven of the eight pigeons participated in both experiments and both renewal effects were detected across all subjects. The authors did note, however, that renewal was greater in the ABA arrangement, compared to the ABC arrangement. Berry et al. noted, though, a con- founding variable in their design that may have influenced this disparity: the order of renewal arrangements (i.e., ABA then ABC). The cur- rent study addresses this confound by introduc- ing ABC renewal before ABA renewal (for Liam and Preston), while producing the same result. Future researchers may consider combin- ing the order of renewal arrangements (i.e., ABA then ABC and ABC then ABA) in an effort to produce a robust, within-subject replication. Researchers have generally found the ABA

renewal effect to be stronger (i.e., producing more relapse in the form of higher response rates) than ABC renewal effects, likely due to a negative generalization gradient. Podlesnik and Miranda-Dukoski (2015) evaluated the negative generalization gradient hypothesis using specific keylight-alternation frequencies to signal reinforcement and extinction contin- gencies with pigeons in an operant chamber. In Phase 1, pigeons’ pecking a keylight that alternated between red and white every 0.5 s produced reinforcement. In Phase 2, the key- light color alternated every 0.13 s and pecking was placed on extinction. In Phase 3, the keylight color changed quasirandomly (i.e., alternating anywhere between 0.13 and 1.96 s), while extinction remained in place. An analysis of the results showed that keylight pecking increased as the alternation frequency became more dissimilar to the frequency in Phase 2 (i.e., extinction), representing a peak

shift similar to those observed in stimulus gen- eralization studies. The results obtained by Podlesnik and

Miranda-Dukoski (2015) supported the hypothesis that the processes responsible for stimulus generalization gradients may also be those that are observed in operant renewal. In the current study, differences between ABA and ABC renewal were not only observed within subjects, but also across subjects. For example, within subjects, ABA renewal was observed in both Liam and Preston’s behavior, while ABC renewal was observed at either a lower relative level (Liam) or not at all (Preston). Across subjects, ABA renewal was always observed (Cole and Vaughn) and ABC renewal was either not observed (Dermot) or produced a significantly smaller increase in responses per minute than ABA (Bella; see Table 2 for direct comparisons using increase in responses per minute). Thus, the current study’s results produced a replication of previ- ous ABA and ABC renewal studies, both within and across participants (Berry et al., 2014; Bouton et al., 2011). Furthermore, sup- port for the negative generalization gradient hypothesis highlights the importance of con- trolling for environmental factors that contrib- uted to the contexts in the current study. Though it is not possible to state with absolute certainty where the participants’ attention was primarily focused, many precautions were taken to ensure context changes were, at a minimum, salient and consistent. For example, the same therapist was used across all conditions for each participant, such that only the session room and the color of the stimuli within it changed; the camera and tripod remained in the rooms for all study sessions, regardless of whether it was recording. These consistencies, seemingly trivial, may have helped to mimic the salient changes that occur in the natural environment, such as going from your home to a clinic or from a special education to a general education classroom.

827ABA AND ABC RENEWAL

The results from the current study may have implications for clinical practice. Specifically, practitioners may consider programming for context changes throughout the progression of treatment. This may be considered during both the reduction of problem behavior and the acquisition of new skills. For example, when transferring a treatment for problem behavior to a familiar or novel environment (e.g., the learner’s home or new classroom), practitioners may consider ensuring they are present to aid caregivers in working through common initial increases in problem behavior. This may help to maintain caregiver treatment integrity in these contexts and reduce the need for future treatment. During the acquisition and mainte- nance of new skills, practitioners may consider reinforcing desired behavior in the terminal context. This may help to ensure the mainte- nance of skills in the presence of omission treatment errors (i.e., failing to provide a conse- quence when the schedule prescribes). Though these suggestions represent what is often occur- ring in practice, the results of the current study suggest that not adopting these practices may have detrimental effects when transferring interventions back to terminal contexts. Although these analogies are consistent with

basic research, results remain preliminary, and warrant further investigation. In the current study, some idiosyncratic participant features seemed to contribute to a variation in the mag- nitude of ABA and ABC renewal effects observed. For example, Preston, Liam, and Vaughn’s responding did not readily adapt to the contingencies in Context B, requiring extended exposure to extinction, particularly in the first renewal arrangement to which they were exposed. In his exposure to the second renewal arrangement, Preston’s behavior adapted to the contingencies in fewer sessions (consistent with a repeated-extinction effect; Sidman, 1960). This finding suggests a poten- tial problem with the use of extinction as a dis- rupter in renewal with humans. Participants

whose responding persisted in the absence of consequences may have been doing so simply due to a lack of alternative sources of reinforce- ment. That is, because there were no other options for engagement during the sessions, participants may have allocated responding toward the target behavior despite the discon- tinuation of the reinforcement contingency. Translating basic findings to humans, partic-

ularly with responses of social importance such as problem behavior, often introduces chal- lenges that require multiple studies for estab- lishing generalized effects. For example, in the current studies, several important variables were necessarily different from previous studies. In basic research, the responses typically include key pecking or lever pressing. It is not possible for these types of responses to be incorrect; the animal either presses the lever/pecks the key- light or does not. In the clinical environment, practitioners target skills for acquisition that are not dichotomous in nature and have the possi- bility of being “incorrect.” Accuracy is a neces- sary component when looking at the translation to practice; practitioners would be unlikely to reinforce an incorrect response without the future plan to shape it into a cor- rect one. Therefore, translating procedures from an experimental to a clinical model is challenging in that the responses targeted are qualitatively different. Though we did not focus on accuracy in the current study, previous research demonstrates the importance of con- text on accurate responding (e.g., Godden & Baddeley, 1975). Therefore, future researchers may wish to evaluate the extent to which response parameters interact with the renewal arrangements. We should interpret the results of the cur-

rent study with caution due to several limita- tions. First, we used an FR-1 reinforcement schedule for each participant during the Rein- forcement (A) phase. The density of this sched- ule arranges a highly discriminable change from reinforcement to extinction. Thus, the

CLARE J. LIDDON et al.828

discriminability of the schedule may affect the magnitude of renewal effects. For example, dur- ing Extinction (A) with Cole, only a single response illustrates his renewal effect (0.2 rpm; see Figure 1, bottom panel, session 16), before he ceased responding completely for the remainder of the condition. This abrupt end in responding could have been a result of the highly-discriminable schedule change. Further- more, it is unlikely that caregivers are able to implement an FR-1 schedule of reinforcement when competing contingencies (e.g., caring for other children, getting house work done, teach- ing an academic lesson) are in place. Second, in a clinical environment, it is unlikely for an extinction schedule to be programmed for newly acquired responses. Third, visual analysis was used to determine extinction effects in Extinction (B), reducing the certainty that all participants’ response rates had completely extinguished. Finally, during the pre- experimental assessment, we did not assess the target responses in the absence of reinforce- ment. This may have contributed to the varia- tion in the participants’ extinction curves. Future research should consider using inter-

mittent schedules of reinforcement as opposed to continuous schedules. The use of intermit- tent schedules may provide less discriminability in the change from reinforcement to extinction, allowing a more in-depth analysis of the con- textual control exerted over the behavior. Addi- tionally, future studies should attempt to test the renewal effect in the presence of treatment errors that are likely to occur with caregivers, such as intermittent reinforcement schedules, and delays to reinforcer delivery. Future researchers might also consider providing a low to moderately preferred item or activity avail- able to provide access to alternative response options to more closely simulate conditions in the natural environment. Finally, future researchers should consider a pre-experimental assessment without overt reinforcement in combination with using a quantifiable

extinction criterion, rather than visual analysis prior to the final context change. This may mitigate the possibility of alternative forms of treatment relapse confounding data interpreta- tion. All of these proposed extensions may fur- ther the general understanding of the clinical implications of context-related treatment relapse.

REFERENCES

Berry, M. S., Sweeney, M. M., & Odum, A. L. (2014). Effects of baseline reinforcement rate on operant ABA and ABC renewal. Behavioural Processes, 108, 87-93. https://doi.org/10.1016/j.beproc.2014.09.009

Bossert, J. M., Liu, S. Y., Lu, L., & Shaham, Y. (2004). A role of ventral tegmental area glutamate in contex- tual cue-induced relapse to heroin seeking. The Jour- nal of Neuroscience, 24, 10726-10730. doi:https://doi. org/10.1523/JNEUROSCI.3207-04.2004

Bouton, M. E. (2002). Context, ambiguity, and unlearn- ing: Sources of relapse after behavioral extinction. Biological Psychiatry, 52, 976-986. https://doi.org/10. 1016/S0006-3223(02)01546-9

Bouton, M. E., & Bolles, R. C. (1979). Contextual con- trol of the extinction of conditioned fear. Learning and Motivation, 10, 445-466. https://doi.org/10. 1016/0023-9690(79)90057-2

Bouton, M. E., & King, D. A. (1983). Contextual control of the extinction of conditioned fear: Tests for the associative value of the context. Journal of Experimen- tal Psychology: Animal Behavior Processes, 9, 248-265. https://doi.org/10.1037/0097-7403.9.3.248

Bouton, M. E., Todd, T. P., Vurbic, D., & Winterbauer, N. E. (2011). Renewal after the extinc- tion of free operant behavior. Learning and Motivation, 1, 57-67. https://doi.org/10.3758/ s13420-011-0018-6

Crombag, H. S., & Shaham, Y. (2002). Renewal of drug seeking by contextual cues after prolonged extinction in rats. Behavioral Neuroscience, 116, 169. doi: https://doi.org/10.1037/0735-7044.116.1.169

Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., Owens, J. C., & Slevin, I. (1992). A comparison of two approaches for identifying rein- forcers for persons with severe and profound disabil- ities. Journal of Applied Behavior Analysis, 25, 491- 498. doi: https://doi.org/10.1901/jaba.1992.25-491

Foxx, R. M. (1996). Twenty years of applied behavior analysis in treating the most severe problem behavior: Lessons learned. The Behavior Analyst, 19, 225.

Godden, D. R., & Baddeley, A. D. (1975). Context- dependent memory in two natural environments: On land and underwater. British Journal of Psychology, 66,

829ABA AND ABC RENEWAL

325-331. doi:https://doi.org/10.1111/j.2044-8295. 1975.tb01468.x

Hamlin, A. S., Clemens, K. J., & McNally, G. P. (2008). Renewal of extinguished cocaine-seeking. Neurosci- ence, 151, 659-670. https://doi.org/10.1016/j. neuroscience.2007.11.018

Hamlin, A. S., Newby, J., & McNally, G. P. (2007). The neural correlates and role of D1 dopamine receptors in renewal of extinguished alcohol-seeking. Neurosci- ence, 146, 525-536. https://doi.org/10.1016/j. neuroscience.2007.01.063

Hanley, G. P., Iwata, B. A., & Thompson, R. H. (2001). Reinforcement schedule thinning following treatment with functional communication training. Journal of Applied Behavior Analysis, 34, 17-38. doi:https://doi. org/10.1901/jaba.2001.34-17

Horner, R. H., & Day, H. M. (1991). The effects of response efficiency on functionally equivalent com- peting behaviors. Journal of Applied Behavior Analysis, 24, 719-732. doi:https://doi.org/10.1901/jaba.1991. 24-719

Kearns, D. N., & Weiss, S. J. (2007). Contextual renewal of cocaine seeking in rats and its attenuation by the conditioned effects of an alternative reinforcer. Drug and Alcohol Dependence, 90, 193-202. https://doi. org/10.1016/j.drugalcdep.2007.03.006

Kelley, M. E., Liddon, C. J., Ribeiro, A., Greif, A. E., & Podlesnik, C. A. (2015). Basic and translational eval- uation of renewal of operant responding. Journal of Applied Behavior Analysis, 48, 390-401. doi:https:// doi.org/10.1002/jaba.209

Nakajima, S., Tanaka, S., Urushihara, K., & Imada, H. (2000). Renewal of extinguished lever-press responses upon return to the training context. Learning and Motivation, 31, 416-431. https://doi.org/10.1006/ lmot.2000.1064

Peters-Scheffer, N., Didden, R., Korzilius, H., & Sturmey, P. (2011). A meta-analytic study on the

effectiveness of comprehensive ABA-based early inter- vention programs for children with autism spectrum disorders. Research in Autism Spectrum Disorders, 5, 60-69. https://doi.org/10.1016/j.rasd.2010.03.011

Podlesnik, C. A., & Miranda-Dukoski, L. (2015) Stimu- lus generalization and operant context renewal. Beha- vioural Processes, 119, 93-98. https://doi.org/10.1016/ j.beproc.2015.07.015

Sidman, M. (1960). Tactics of scientific research: Evaluating experimental data in psychology. New York, NY: Basic Books.

St. Peter Pipkin, C. S., Vollmer, T. R., & Sloman, K. N. (2010). Effects of treatment integrity failures during differential reinforcement of alternative behavior: A translational model. Journal of Applied Behavior Anal- ysis, 43, 47-70. doi: https://doi.org/10.1901/jaba. 2010.43-47

Volkert, V. M., Lerman, D. C., Call, N. A., & Trosclair- Lasserre, N. (2009). An evaluation of resurgence dur- ing treatment with functional communication train- ing. Journal of Applied Behavior Analysis, 42, 145- 160. doi: https://doi.org/10.1901/jaba.2009.42-145

Zironi, I., Burattini, C., Aicardi, G., & Janak, P. H. (2006). Context is a trigger for relapse to alcohol. Behavioural Brain Research, 167, 150–155. https:// doi.org/10.1016/j.bbr.2005.09.007

Received October 14, 2016 Final acceptance September 3, 2017 Action Editor, Claire St. Peter

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher’s website.

CLARE J. LIDDON et al.830