Teaching trainees to implement functional communication training with multiple schedules: An evaluation of training effects and durability

Abstract

We evaluated the effects of behavioral skills training on improving participant implementation of functional communication training with multiple schedules when working with a confederate. Behavioral skills training produced mastery-level responding for all six participants who required training, providing the first empirically supported training for this functional communication training approach. Next, we assessed durability during training challenges with (a) procedural changes to the original protocol, (b) a novel confederate with different discriminative stimuli and reinforcers, and (c) relapsed confederate destructive behavior. Training effects degraded at least once for all participants and in 62% of training challenges, although continuing to expose the participant to the challenging situations or providing post-session booster training resolved the degradation in most cases. We discuss these findings in relation to their clinical implications and directions for future research.

Approximately 94% of caregivers of children with autism spectrum disorder report at least one form of challenging behavior (Jang et al., 2011), with 68% reporting severe destructive behavior like aggression, self-injurious behavior, or property destruction (Kanne & Mazurek, 2011). Thus, it is likely that staff and trainees¹ will encounter severe destructive behavior when delivering applied-behavior-analytic services. For example, Ralston and Brown (2023) surveyed 142 Registered Behavior Technicians and found that 83% reported working with clients who displayed severe destructive behavior, and 75% sustained an injury when managing such behavior. In another study, Colombo et al. (2021) surveyed 125 Board-Certified Behavior Analysts (BCBAs) who worked with clients with severe destructive behavior and found that nearly half reported being assigned their first severe-behavior case without any initial or ongoing training.

Clearly, there is a need for greater access to empirically supported trainings for managing severe destructive behavior, especially for staff and trainees learning how to implement effective treatment for such behavior. Functional communication training with multiple schedules (mult FCT) is one effective treatment for severe destructive behavior (Greer et al., 2016) that staff and trainees can be taught to implement. Mult FCT is a stimulus-control extension of Carr and Durand’s (1985) seminal FCT paradigm in which implementers teach clients when reinforcement is and is not available for their functional communication responses (FCRs) using treatment signals. Most often, mult FCT involves alternating between a reinforcement component signaled by a stimulus correlated with reinforcement (S^D; e.g., a green card) and an extinction component signaled by a stimulus correlated with extinction (S^Δ; e.g., a red card). This approach has been successful in (a) treating severe destructive behavior while reducing rates of FCRs and reinforcement (Greer et al., 2016); (b) mitigating treatment relapse like resurgence (Fisher et al., 2020; Fuhrman et al., 2016); and (c) facilitating generalization to novel therapists, settings, and caregivers (Fisher et al., 2015; Greer et al., 2019; Neely et al., 2018).

To date, there is no empirically supported training for mult FCT outside of employment or practicum experiences at specialized severe-behavior clinics (Mitteer et al., 2024). Thus, a goal of the present study was to develop a behavioral skills training (BST; e.g., Parsons et al., 2012) package consisting of written instructions, modeling of target skills, and role play with performance feedback to teach trainees to implement mult FCT to mastery levels. Trainee mastery refers to performance criteria in which the trainee implements all intervention components with a certain level of accuracy. Often, BCBAs continue training until the trainee demonstrates 100% accuracy with the behavioral program during role play. Additionally, it is recommended that trainees then experience in-vivo, or on-the-job, training by supervisors to ensure that competency shown in role play extends to working with clients (Parsons et al., 2012).

Despite one’s best effort to prepare trainees for unique experiences they will encounter during BST, there will undoubtedly be situations in which a BCBA must empower a trainee to deliver services in a way that differs from initial training. Take, for example, a recent paper by Berdeaux et al. (2022) in which researchers taught five teachers to implement function-based treatments for escape-maintained challenging behavior displayed by confederates. Although the teachers’ procedural fidelity was high during routine implementation of the treatments, adding distractor confederates to the classroom disrupted implementation accuracy for at least one function-based treatment for all teachers. Within clinical settings, supervisees and trainees may need to (a) deliver services to less-familiar clients when staffing is limited, (b) make modest protocol adjustments following an updated treatment plan, or (c) respond immediately to unexpected client behavior. Although it is likely that trainees will encounter situations involving severe destructive behavior that differ from their training, a review of the BST literature by Smith et al. (2024) found that only two studies evaluated durability of training effects when participants learned to implement behavior-reduction procedures (Giles et al., 2018; Graudins et al., 2012). Graudins et al. (2012) assessed generalization of oral-care steps for children with challenging behavior to new X-ray procedures, and Giles et al. (2018) assessed confederate-to-child generalization of response-interruption-and-redirection procedures. In both Graudins et al. and Giles et al., participant procedural fidelity remained high during the generalization tests.

Implementing mult FCT will undoubtedly require that the trainee adjust their behavior to new situations that differ from the training context. In our clinic, we ask staff and trainees to work with multiple clients across the week and to be prepared for protocol changes; these contextual changes may disrupt implementer performance (Berdeaux et al., 2022; Mitteer et al., 2022; Williams et al., 2023). Even if protocols and clients remain the same, a myriad of treatment variables may lead to temporary increases in destructive behavior that trainees must manage in the moment (Briggs et al., 2018; Mitteer et al., 2022; Muething et al., 2020, 2021). An increase in client destructive behavior is likely to disrupt implementer performance (Addison & Lerman, 2009; Carr et al., 1991; Mitteer et al., 2018; Williams et al., 2023), further eroding treatment effects. Ultimately, it may be helpful for BCBAs to evaluate the durability of training effects under novel situations prior to trainees working directly with clients.

Durability refers to the persistence of treatment or training effects during changing environmental conditions, including under traditional stimulus-generalization and maintenance tests but also during more robust treatment or training challenges (Greer & Shahan, 2019; Mitteer, in press; Muething et al., 2022). In their discussion of durability, Nevin and Wacker (2013) noted that:

For maintenance to occur, the effects of treatment must persist when changes occur in both antecedent and consequent stimuli. The people, tasks, prompts, and consequences surrounding behavior change or vary over time. For long-term maintenance to be achieved, adaptive behavior must persist in the face of these challenges. (p. 123)

Researchers have examined the durability of the effects of mult FCT on destructive behavior as well as adaptive behavior (e.g., FCRs, cooperation) under treatment challenges like reinforcement schedule thinning (Briggs et al., 2018; Mitteer et al., 2022; Muething et al., 2022) and changes in implementers or settings (Mitteer et al., 2022; Muething et al., 2021). As we developed our training materials for mult FCT, we wanted to extend this analysis of durability to our training evaluation. The faculty in our severe behavior program came to a consensus on three training challenges to evaluate within our study, which are described in more detail within the method section. We involved confederates who role played as clients with destructive behavior to program equivalent opportunities for trainees to exhibit mult-FCT skills across sessions and participants and, importantly, it avoided a worsening of actual client behavior should the participants implement mult FCT incorrectly. Across two experiments, we evaluated the efficacy of our training on mastery of mult-FCT skills and durability during training challenges. We viewed this experiment with confederates as an initial evaluation of training efficacy and durability. If training in such an arrangement is demonstrated to be successful, then future research can examine the generality of training effects with confederates who implement mult-FCT to clients with severe destructive behavior who may respond differently than confederates.

GENERAL METHOD

Recruitment

The experimenters offered study participation to graduate students enrolled in behavior-analytic coursework at an affiliated university and newly hired Registered Behavior Technicians employed by the outpatient clinic, which served as the setting for the study. Because FCT is the most widely used treatment for socially reinforced severe destructive behavior (Tiger et al., 2008), researchers expected that several potential participants would have experience with FCT or other similar differential-reinforcement interventions. Thus, experience with FCT was not an exclusion criterion but experience with mult FCT was. We did not have to exclude any participants due to previous mult-FCT experience, and we enrolled all interested trainees until we recruited four participants for Experiment 1 and three for Experiment 2. All participants completed a consent process approved by the respective institutional review board prior to beginning the experiment.

All participants except Tom in Experiment 1 were enrolled in a graduate course on single-case experimental designs during the study. Students could earn up to 500 maximum points in the graduate course, and they received 10 points of extra credit for participation irrespective of their performance or duration of participation. Intermittently during their participation, the experimenters reminded these participants that they could withdraw from the study at any point without losing their extra-credit points and reminded Tom that he could withdraw from the study at any point and continue to receive training in mult FCT related to his clinical duties. No participant withdrew voluntarily or involuntarily.

Settings and materials

Experimenters conducted sessions within 3-m by 3-m therapy rooms in an intensive-outpatient clinic. Therapy rooms contained padding on the floors and walls, one-way observation mirrors from which data collectors scored participant behavior, and audio-visual equipment for the data collector to hear the participant and confederate behavior. Clinic staff (e.g., BCBAs, Registered Behavior Technicians, practicum students) served as the confederates in the experiment. A BCBA or BCBA-Doctoral, each of whom had several years of experience training new staff in mult FCT, served as trainers during BST and delivered the training according to a specific protocol approved by the faculty in the severe behavior program.

All sessions included (a) a 10-cm by 15-cm card with blue and yellow sides to serve as discriminative stimuli, (b) a timer to assist in tracking component durations, and (c) a laminated mult-FCT protocol individualized to the experimental condition. Prior to each session, an experimenter provided the participant with a list displaying the order of programmed mult-FCT components (e.g., reinforcement, extinction, extinction, reinforcement, extinction, reinforcement). Except during the stimulus challenge that programmed an attention function for confederate behavior, the experimenter provided the participant with a bag of miniature cookies to serve as reinforcers for confederate behavior. Confederates also held small fidget toys (e.g., fidget spinner, bubble-pop ball) in their hands to simulate toy play that a client may emit while in therapy sessions.

Confederate scripts

Similar to Mitteer et al. (2018) and Williams et al. (2023), the experimenters created a series of audio scripts to maximize the consistency of confederate responding during and across sessions. Copies of the audio scripts and their written counterparts are available for download in the accompanying data repository (https://zenodo.org/doi/10.5281/zenodo.8360234). Each script involved an experimenter stating a confederate response (e.g., “FCR”) at a predetermined time (e.g., 30-s mark). Experimenters generated multiple audio recordings and assigned them to a given session. Confederates wore a Bluetooth earbud connected to a smartphone playing the audio scripts such that the prompts were only audible to the confederates. In all sessions except challenge probes, reinforcement and extinction components included six instances of destructive behavior and six FCRs dispersed across multiple 30-s components (see Response Measurement section for operational definitions). The scripts programmed the same total number of destructive responses and FCRs during reinforcement and extinction components but the timing of those responses and the number of each response per reinforcement or extinction presentation differed. No more than two responses of each type (e.g., FCRs, destructive behavior) occurred in a given reinforcement or extinction component. For example, in one session, the confederate engaged in two instances of destructive behavior and two FCRs during the first extinction component, two instances of destructive behavior and no FCRs during the second extinction component, and no destructive behavior and two FCRs during the third extinction component. There was a similar, but not identical, distribution of FCRs and destructive behavior during that session’s reinforcement components. Experimenters arranged this script format to decrease the predictability of confederate behavior and approximate the fluctuating rates of client behavior one may encounter in a typical session. When not engaged in destructive behavior or FCRs, confederates walked around the room and played with fidget items they held in their hands.

To maximize session time and efficiency, experimenters arranged edible items (e.g., cookies), and, later, brief attention (e.g., high fives) during the stimulus challenge to serve as the functional reinforcers maintaining the confederate’s destructive behavior as opposed to duration-based reinforcement (e.g., 20-s access to toys or a break from instructions). Confederates briefly paused their scripted behavior to simulate consumption. Because the study occurred during the COVID-19 pandemic, confederates wore medical face masks and placed edibles to their mouth before depositing them into their pocket to emulate edible consumption.

Response measurement

The primary dependent measure was the participant’s accuracy of mult-FCT implementation. Table 1 displays a list of mult-FCT component skills that trained data collectors used to record participant accuracy. Experimenters designed this checklist based on (a) the scoring sheet used in a previous study of caregiver mult-FCT implementation (Campos et al., 2020), (b) descriptions of mult-FCT features in the literature (Greer et al., 2016), and (c) consensus by the three BCBA-Ds and four BCBAs in the clinical program, all of whom have published regularly on mult FCT and agreed on inclusion of each skill. Correct steps consisted of skills the participant implemented accurately according to the checklist (e.g., delivered the programmed reinforcer within 3 s of a confederate’s FCR during the reinforcement component). Incorrect steps consisted of steps for which the participant deviated from the checklist (e.g., delivered the programmed reinforcer during the extinction component). Because the clinical program strives for perfect procedural fidelity during mult FCT, data collectors scored conservatively and marked a step as “correct” only if the participant implemented the skill correctly across all reinforcement or extinction components in a session. Not applicable was scored for steps that, due to timing and participant behavior, did not have the opportunity to occur during a particular session. Data collectors only scored “not applicable” twice during the study due to the session ending before completion of the confederate’s script. Data collectors calculated the percentage accuracy of mult-FCT by dividing the number of correct steps by the number of correct and incorrect steps, then they multiplied this quotient by 100. Because global measures of procedural fidelity could mask important error patterns (Cook et al., 2015), experimenters also graphed and monitored participant performance for each component skill.

TABLE 1.

Measured trainee skills during functional communication training with multiple schedules.

Reinforcement-Component Skills	Shortened Term
The therapist initiates the discriminative stimulus for reinforcement (SD) within 3 s of programmed timing in the absence of destructive behavior.	SD Initiation
The therapist presents the SD for approximately 30 s (i.e., 27–33 s) and the SD is clearly visible for 90% of this duration.	SD Visibility
The therapist will only deliver reinforcement (Sr+) within 3 s of an independent functional communication response (FCR) only if the individual does not already have access to the reinforcer.	Sr+ Delivery
The therapist delivers only one reinforcer per correct FCR.	Sr+ Magnitude
The therapist withholds reinforcement and attention following destructive behavior for at least 3 s (i.e., 3-s changeover delay; COD).	Extinction + COD
Extinction-Component Skills	Shortened Term
The therapist initiates the stimulus delta for extinction (SΔ) within 3 s of when programmed.	SΔ Initiation
The therapist presents SΔ for approximately 30 s. Stimulus change is evident and the card is visible 90% of this duration.	SΔ Visibility
The therapist does not provide reinforcement (edible) or attention when the confederate engages in the correct FCR.	Extinction
The therapist ignores all destructive behavior and FCRs and does not reintroduce SD for 3–5 s following the end of destructive behavior.	SΔ to SD COD

Note: Data collector scored “Yes,” “No,” or “Not Applicable” for each of these skills during every session.

Data collectors scored the frequency of confederate behavior via video recordings of the sessions. Confederate destructive behavior included property destruction (e.g., tearing materials), self-injurious behavior (e.g., headbanging), and aggression (e.g., lightly hitting the participant’s arm). Specific topographies differed between the primary confederate and the stimulus-challenge confederate (described in more detail below). Aggression emulated the intensity used during clinical role plays that do not leave lasting tissue damage or other adverse effects. Confederate FCRs consisted of vocal requests (“Cookies” or “High Five,” depending on the condition).

Experimental design

Experimenters followed a sequential design involving (a) baseline with a behavioral protocol, (b) BST if baseline was insufficient at producing trainee mastery, and then (c) a series of three training challenges. For participants who did not display mastery during repeated baseline sessions, experimenters staggered implementation of BST within a nonconcurrent multiple-baseline-across-participants design. With the exception of BST skill-out training (described below in Experiment 1), the mastery criteria for each condition were two consecutive sessions with 100% accuracy. Aside from the teaching portions of BST, all sessions were 3 min.

EXPERIMENT 1: FIXED ORDER OF TRAINING CHALLENGES WITH IMMEDIATE BOOSTER TRAINING

Experimenters assessed training durability following mastery of mult-FCT implementation with three training challenges arranged in presumed difficulty from least to most challenging according to the consensus of three BCBA-Ds in our clinical program. It was posited that working with the same confederate client but with a different mult-FCT schedule would be the least challenging training extension, followed by a change in confederate client and function with the same mult-FCT schedule, and that an increased rate of confederate destructive behavior and FCRs would be most challenging to participant accuracy because this type of training challenge frequently disrupted FCT treatment implementation by caregivers in Mitteer et al. (2018). Additionally, we anticipated that most clinicians would immediately correct staff or trainee errors during supervised sessions with clients; thus, experimenters provided booster training consisting of performance feedback or role play following sessions with participant errors after BST.

Participants

According to the logic of a prospective consecutive controlled case series design (Hagopian, 2020), we enrolled all the participants who responded to our recruitment request and opted to present their data regardless of training outcomes or the need for BST following baseline. Tom was a 23-year-old non-Hispanic White male enrolled in his first year as a master’s student in the affiliated applied behavior analysis (ABA) program and a newly hired technician in the experimenters’ clinic. Lucy, a 27-year-old non-Hispanic White female, was in her second year of the ABA program and employed as a classroom assistant at an unaffiliated behavior-analytic school. Yara was a 22-year-old non-Hispanic Asian female in her first year of the ABA master’s program. Yara was not employed or completing practicum within the ABA field at the time of the study. Jolie, a 26-year-old non-Hispanic White female, was a doctoral student in the clinical psychology Psy.D. program at the affiliated university and was employed as a classroom assistant in an unaffiliated ABA school.

Procedural fidelity and interobserver agreement

Prior to conducting any sessions with participants, confederates role played with an experimenter who simulated participant behavior until the confederate demonstrated three consecutive sessions with 100% procedural fidelity. Before each session, confederates reviewed the experimental protocol prior to entering the room. As noted in the response-measurement section, data collectors recorded the frequency of confederate destructive behavior and FCRs per session. This is because the frequency of confederate behavior could affect the number of opportunities to score participant skills (e.g., fewer destructive responses could result in fewer opportunities to implement extinction). Given that confederates continued with their programmed scripts irrespective of participant behavior (see information on scripts below), examining the consistency of confederate behavior within and across participants is likely the most relevant measure of procedural fidelity for this experiment. Table 2 summarizes these confederate data.

TABLE 2.

Confederate destructive behavior and functional communication responses.

Experiment and Participant	Baseline		Behavioral Skills Training (BST)		Post-BSTa		Schedule Challenge		Stimulus Challenge		Treatment-Adherence Challenge
	DB	FCR	DB	FCR	DB	FCR	DB	FCR	DB	FCR	DB	FCR
Experiment 1
Tom	13.50 (12–15)	12.50 (11–14)	16 (15–18)	14 (12–18)	12.50 (12–13)	12.50 (12–13)	16 (16–16)	14.50 (14–15)	12 (12–13)	12 (12–13)	32 (28–38)	24 (20–31)
Lucy	14 (12–16)	13 (12–14)	14 (12–17)	16 (12–20)	13.50 (13–14)	12 (13–14)	14 (12–17)	14.50 (14–16)	13 (12–13)	12 (12–13)	30 (29–31)	22.50 (22–25)
Yara	14 (13–16)	13 (12–13)	13.50 (10–16)	12.50 (7–18)	12.50 (12–13)	12.50 (12–14)	15 (12–17)	14 (12–17)	12.50 (12–13)	12 (11–13)	28 (27–29)	20.50 (19–22)
Jolie	12 (11–20)	13 (11–17)	N/A	N/A	13.50 (12–15)	13 (13–13)	14 (12–15)	18 (15–23)	13 (12–15)	14 (13–16)	29 (29–30)	21 (20–22)
Experiment 2
Nicole	12 (12–13)	12 (11–12)	13.50 (13–14)	13.50 (13–14)	13 (12–14)	12 (12–12)	15 (15–15)	14 (14–14)	13 (12–15)	12 (12–12)	31 (30–32)	20.50 (20–21)
Caseyb	-	-	16 (15-17)	17 (17)	16 (15–18)	13 (12–14)	16 (15–18)	17 (16–19)	13 (12–15)	12.50 (12–13)	30 (22–38)	18 (14–22)
Sophia	12 (11–15)	13 (11–15)	15 (12–19)	11 (10–12)	13 (11–14)	12 (10–13)	15 (14–16)	16 (15–18)	12.50 (12–13)	12 (12–12)	34 (29–40)	20 (16–22)

Note: Means and ranges of confederate destructive behavior (DB) and functional communication responses (FCRs) reflected as a count during 3-min sessions.

^aFor Jolie, post-BST data reflect continued baseline sessions following mastery.

^bCamera issues during Casey’s baseline precluded data collectors from accurately scoring confederate behavior from the videos.

Trained data collectors, all of whom were deemed reliable across more extensive clinical data-collection systems, reviewed operational definitions prior to each research appointment. A second, independent observer recorded data on at least one-third of sessions across experimental phases. Experimenters compared each row of the procedural-fidelity checklist across observers and marked an agreement when both observers scored “Yes,” “No,” or “N/A.” Then, they computed exact agreement by dividing the number of agreements by the total number of rows on the procedural-fidelity checklist and multiplied that quotient by 100. Interobserver agreement for accuracy of mult-FCT implementation was 100% for Lucy, Yara, and Jolie, and it averaged 99% (range: 88%–100%) for Tom. Experimenters also computed total-agreement coefficients for session counts of confederate destructive behavior and confederate FCRs by dividing the smaller number of the respective confederate behavior by the larger number of the confederate behavior and multiplying that quotient by 100 to yield a percentage. Interobserver-agreement coefficients for confederate destructive behavior averaged 96% (range: 92%–100%) for Tom, 93% (range: 74%–100%) for Lucy, 95% (range: 84%–100%) for Yara, and 98% (range: 91%-100%) for Jolie. Interobserver-agreement coefficients for confederate FCRs averaged 97% (range: 82%–100%) for Tom, 93% (range: 81%–100%) for Lucy, 96% (range: 92%–100%) for Yara, and 96% (range: 92%–100%) for Jolie.

Procedures

Baseline

Baseline was designed to simulate situations in which BCBAs provide a new trainee with a behavioral protocol to review and then ask them to implement that program with their client. Prior to each session, an experimenter guided the participant into the therapy room and provided the participant with a mult-FCT protocol. This protocol, adapted from the research team’s clinical documents, provided a step-by-step description of a mult 30/30 (i.e., 30 s reinforcement, 30 s extinction) session. The experimenter also provided a programmed order in which the participant should present the mult-FCT components for that session (e.g., reinforcement, extinction, extinction, reinforcement, extinction, reinforcement). See the data repository (https://zenodo.org/doi/10.5281/zenodo.8360234) for these materials. The experimenter permitted the participant as much time as they needed to review the protocol prior to each session and allowed the participant to make notes on the protocol document but did not answer questions or provide additional instructions. After the participant told the experimenter they were ready to begin the session, the confederate entered the room and followed the audio script irrespective of participant behavior.² The experimenter provided no additional guidance or feedback during the session. Experimenters conducted a minimum of three baseline sessions for all participants and extended the baseline phase across participants according to multiple-baseline logic.

BST

If participants did not master mult-FCT implementation during baseline with the written protocol, a BCBA implemented BST to conduct more thorough teaching of mult-FCT skills. Behavioral skills training consisted of instructions, modeling, and role play with feedback. To make BST more practical for experimental and clinical use, the experimenters automated the instruction and modeling portion (Weston et al., 2020) within a PowerPoint presentation. The scripted PowerPoint (a) explained the conceptual background of mult FCT, (b) described and modeled each component skill, and (c) included comprehension questions that required accurate responses before progressing to the next section of the PowerPoint.

Because this training was intended to be used for staff and trainees outside of the experimental context or our clinic, the instructions and modeling portion of BST incorporated teaching strategies such as multiple-exemplar training and training loosely (Stokes & Baer, 1977) within the PowerPoint. The written instructions and video models described how to implement mult FCT across three common social functions of severe destructive behavior (i.e., attention, escape, and tangible), FCR topographies (e.g., exchanging a communication card, vocal mand, and manual sign), and discriminative stimuli (e.g., red and green cards on a lanyard, blue and yellow wristbands, black and white papers). Please see the data repository (https://zenodo.org/doi/10.5281/zenodo.8360234) for training materials. The PowerPoint was not intended to provide explicit guidance on the features of the training challenges that the participants would experience later in the experiment.

During rehearsal and feedback, participants implemented mult FCT with the trainer in 3-min sessions akin to baseline and post-BST sessions with the identical discriminative stimuli and edible function of confederate destructive behavior. The trainer provided corrective feedback immediately and behavior-specific praise intermittently. After participant responding met mastery during role play, experimenters terminated BST and initiated the post-BST condition.

BST skill-out teaching (Yara Only)

Because Yara did not master mult-FCT implementation during typical BST, the experimenters conducted an error analysis to detect the type of errors she made despite rehearsal and feedback. The most common error Yara made was incorrect implementation of a changeover delay (COD; Herrnstein, 1961), which involved Yara switching from the extinction component to the reinforcement component despite the confederate engaging in FCRs or destructive behavior. The correct response in these situations would be to implement a COD by waiting for 3 s without FCRs or destructive behavior before switching to the reinforcement component to avoid adventitiously reinforcing these problematic responses.

The BCBA conducted skill-out teaching for the COD in a two-step sequence. Each session consisted of six trials, with mastery defined as correct COD implementation for all six trials in a session. Each trial consisted of a 10-s extinction component and 5-s reinforcement component to allow for rapid practice of the COD implementation. During the first step of skill-out teaching, the confederates engaged in destructive behavior at the end of the extinction period in half the trials and FCRs at the end of the extinction period in the other half, randomly dispersed across the six-trial session. After Yara mastered the COD in this format, she proceeded to the second step of skill-out teaching, in which half of the trials were identical to the first step and the other half contained no confederate destructive behavior or FCRs. This arrangement emulated the typical sessions Yara experienced outside of skill-out teaching in which not every programmed switch from extinction to reinforcement required COD implementation. After mastering the COD in this final step, Yara returned to the 3-min BST sessions.

Post-BST

Post-BST³ procedures were identical to the baseline in that the trainer was no longer present during the session. The purpose of this condition was twofold. First, post-BST allowed an assessment of participant accuracy in the absence of immediate BCBA feedback. Second, conducting post-BST prior to each training challenge allowed for an examination of the deleterious effects of the challenge condition relative to the standard mult FCT 30/30 context with the initial confederate, potentially reducing sequence effects that could occur with back-to-back training challenges. For all participants, experimenters conducted at least one post-BST session to serve as a maintenance session between training challenges. Additionally, for any participant whose accuracy was less than 80% during baseline, we conducted at least two post-BST sessions immediately following BST to ensure that the training effects maintained before the first training challenge.

Booster training (Not graphically displayed)

As is recommended following trainee mastery of behavior-analytic skills (Lerman et al., 2015), experimenters monitored the maintenance of trainee performance and corrected errors that occurred. Following a session with less than 100% accuracy during post-BST and training challenges, the trainer met with the participant to review the specific errors that they made and provided corrective feedback along with opportunities for modeling or role play. Each booster training lasted 1–2 min. Experimenters did not provide any feedback to participants following sessions with 100% accuracy.

Training challenges

The training challenges were derived via consensus of the three BCBA-Ds in the experimenters’ severe behavior program as ones most relevant to trainees in their clinic. Regarding mult FCT, it is unlikely that (a) participants would incorporate a 30/30 schedule each time they implement mult FCT, (b) all clients would have the same discriminative stimuli or behavioral functions, and (c) that destructive behavior and incorrect FCRs would always remain at low rates during extinction periods. Prior to the first and second challenges, the experimenters noted that the protocol differed and recommended that the participant review materials carefully before initiating the session. This emulated standard clinical practice in the experimenters’ clinic in which trainees and employees are informed of programmed changes to protocols prior to conducting sessions. The experimenters provided no forewarning of upcoming changes to confederate behavior prior to the third challenge, as this was designed to emulate unexpected client behavior during a routine session (e.g., spontaneous recovery; Kimball et al., 2023; Wathen & Podlesnik, 2018). In lieu of any empirical data suggesting a clear sequence for these training challenges, the experimenters arranged them according to presumed difficulty.

Schedule challenge

The first challenge reduced the duration of the extinction component, which resulted in rapid component switches and potentially frequent implementation of the COD should destructive behavior occur within the shortened extinction components. The reinforcement component remained 30 s, but the extinction component was only 2 s (i.e., a 30/2 schedule). The printed protocol and order of mult-FCT components given to the participant reflected this change. The goal of the schedule challenge was to determine if mastery with the mult 30/30 schedule would predict accurate implementation of a dense mult 30/2 schedule, the latter of which often serves as an initial mult-FCT schedule in our clinic. The sessions remained 3 min but involved approximately twice as many schedule components due to the mult 30/2 schedule.

Stimulus challenge

The second challenge altered the colors of the discriminative stimuli and the topography and function of confederate FCRs and introduced a novel confederate. Experimenters reversed the S^D and S^Δ colors from the initial confederate’s protocol. This occurs somewhat regularly in the experimenters’ affiliated clinic when using similar colors (e.g., blue, yellow) across clients. Although multiple clients may have identical coloring pairings for their discriminative stimuli BCBAs ask each client and their caregivers to choose the colors associated with reinforcement and extinction. Although aimed at bolstering client input into treatments, this could also result in therapists making occasional errors when switching between clients, such as if the color blue is associated with reinforcement for their morning client but also associated with extinction for their afternoon client. Additionally, experimenters programmed an attention function for destructive behavior and a new vocal FCR (“High Five”). The printed protocol given to the participant reflected these procedural differences. The goal was to determine whether BST would prepare the participant for working with multiple clients with different protocols and behavioral functions.

Treatment-adherence challenge

The third challenge was modeled after the treatment-adherence challenges in Mitteer et al. (2018) and Williams et al. (2023). The session was like post-BST, but the confederate emitted higher rates of destructive behavior and FCRs during the extinction component. This emulated situations in which participants encounter relapse of child behavior despite correct treatment implementation (e.g., renewal, spontaneous recovery; Kimball et al., 2023; Wathen & Podlesnik, 2018). Prior to this study, experimenters determined how many responses a confederate could reasonably emit within a 30-s mult-FCT component. Within 30 s, the experimenters completed approximately 35 responses (e.g., FCRs, destructive behavior) reliably. As such, the treatment-adherence challenge contained approximately 35 scripted responses per mult-FCT component, with a mix of destructive behavior and incorrect FCRs during the extinction component and a higher proportion of destructive behavior to FCRs during the reinforcement component (i.e., simulating poor FCR discrimination or renewal of destructive behavior; e.g., Mitteer et al., 2022). Because the mult-FCT procedures were identical to baseline and post-BST from the participant’s perspective (i.e., only confederate behavior differed), the materials given to the participant were identical to those in baseline and post-BST.

Results

Figure 1 depicts the percentage of accurate implementation of the mult-FCT protocol. As seen in the first phase of Figure 1, only one participant (Jolie) mastered all relevant skills for mult-FCT implementation during baseline with a written protocol. Mean accuracy for Tom, Lucy, and Yara was 55%, 77%, and 26%, respectively. Table 3 displays an overview of participant errors during each experimental condition, with each black checkmark indicating that a participant in Experiment 1 made an error on that skill. As can be seen, three of the four participants erred on each skill aside from Sr+ magnitude (i.e., delivering only one edible per FCR) and S^Δ visibility (i.e., displaying the S^Δ consistently during the extinction component).

FIGURE 1.

Percentage accuracy of mult-FCT implementation (Experiment 1). This figure displays each participant’s accuracy when implementing mult FCT during Experiment 1. The dotted line for Jolie indicates the transition from baseline to the challenge sequence without an intermediate BST phase.

TABLE 3.

Error analysis across experimental conditions and mult-FCT skills.

Component and Skill		Experimental Condition
		Baseline	Behavioral Skills Training (BST)	Post-BST	Schedule Challenge	Stimulus Challenge	Treatment-Adherence Challenge
Mult-FCT Skills	Reinforcement Component
	SD Initiation	✔✔✔✔✔✔		✔	✔	✔	✔
	SD Visibility	✔✔✔✔✔✔	✔		✔		✔
	Sr+ Delivery	✔✔✔✔✔	✔✔✔	✔	✔✔		✔
	Sr+ Magnitude	✔✔					✔
	Extinction + COD	✔✔✔✔			✔✔		✔✔
	Extinction Component
	SΔ Initiation	✔✔✔✔✔✔	✔✔	✔	✔✔	✔	✔
	SΔ Visibility	✔✔✔✔✔		✔
	Extinction	✔✔✔✔✔	✔✔	✔	✔✔	✔✔✔✔	✔
	SΔ to SD COD	✔✔✔✔✔✔	✔✔✔	✔	✔✔	✔	✔

Note: Checkmarks indicate the number of participants who displayed an error for the mult-FCT skill during the specified experimental condition. Black and gray checkmarks denote errors for participants in Experiment 1 and 2, respectively. Empty cells indicate that no participants made that particular error in the condition during either experiment. S^D = discriminative stimulus for reinforcement, S^r+ = reinforcement, S^Δ = stimulus delta for extinction, COD = changeover delay.

Tom, Lucy, and Yara participated in BST. Tom and Lucy each completed the role-play component within four sessions, but Yara required substantial performance feedback and supplemental training techniques to complete BST. Following seven sessions of BST, the BCBA trainer determined that the most consistent error Yara made was implementing a COD following destructive behavior or incorrect FCRs at the end of the extinction component. The BCBA conducted skill-out training for this error across two steps, first in a six-trial format in which Yara practiced using a COD on each trial and next during six-trial sessions in which half of the trials required the implementation of the COD and the other half did not (similar to what is required in the target sessions). Yara required an additional session of the skill-out training after her performance deteriorated in the BST role plays. In total, Yara required 14 sessions of BST and three skill-out sessions to reach mastery.

In general, mastery of mult-FCT implementation maintained during post-BST sessions. All participants displayed 100% accuracy during the maintenance probes that occurred between training challenges. However, recall that experimenters opted to conduct a series of post-BST sessions immediately following training for those with baseline accuracy below 80%. For Tom and Yara, who met these criteria, we observed slight decrements in the second consecutive post-BST session that followed BST, which warranted booster training.

Every participant’s performance degraded at least once during the sequence of durability tests (schedule, stimulus, treatment-adherence). Across the four participants, there were 12 training challenges (i.e., three per participant), and we observed reductions in mult-FCT accuracy during 8 of those 12 (67%) challenges. The schedule challenge disrupted accuracy for three of four participants. Interestingly, none of the error forms (see Table 3) for these participants were ones the experimenters expected during this challenge. The stimulus challenge disrupted participant accuracy in three of four cases. All three participants who erred failed to implement extinction appropriately and Jolie displayed difficulty in initiating the reinforcement and extinction components correctly. During the treatment-adherence challenge, half of the participants (Lucy and Yara) displayed durable training outcomes with 100% accuracy. The other two participants (Tom and Jolie) did not implement extinction accurately, with Jolie delivering reinforcement during the extinction component and Tom inadvertently reinforcing problematic confederate behavior during the reinforcement component by omitting a COD. Booster trainings eradicated these errors.

EXPERIMENT 2: RANDOMIZED ORDER OF TRAINING CHALLENGES WITH DELAYED BOOSTER TRAINING

The results of Experiment 1 provided preliminary support for the efficacy of our mult-FCT training. However, one participant (Yara) required extensive teaching, and another (Jolie) mastered mult-FCT without needing to experience BST. Therefore, recruiting additional participants to evaluate the effects of BST on acquisition of mult-FCT implementation would allow us to better assess the generality of teaching effects. Additionally, two aspects of Experiment 1 posed interpretation difficulties for the training-challenge data. Although we explicitly programmed training challenges in Experiment 1 according to presumed difficulty, the third challenge (treatment-adherence) disrupted performance for fewer participants than the earlier challenges (schedule, stimulus). Thus, it may have been that sequencing of the challenges moderated performance across the challenges. Further, providing immediate booster training following a session with errors made it difficult to determine whether participant performance would have improved with continued exposure to the training challenge. Thus, in Experiment 2, we (a) recruited three new participants to further assess teaching effects, (b) randomized the sequence of training challenges to understand effects of challenge order versus challenge type, and (c) delayed booster training following initial errors to assess whether participant performance would increase with repeated practice during the training challenges.

Participants

Nicole was a 25-year-old non-Hispanic Black female enrolled in her first year of the school psychology Psy.D. program at the affiliated university and was completing her practicum at a local public school. Casey was a 23-year-old non-Hispanic White female who was enrolled in her first year of the school psychology Psy.D. program at the affiliated university. Casey was employed as a Registered Behavior Technician at an unaffiliated in-home company and completing her practicum at a local public school. Sophia was a 23-year-old non-Hispanic White female enrolled in her first year of the affiliated ABA program and employed as a Registered Behavior Technician at an unaffiliated in-home company.

Procedural fidelity and interobserver agreement

Experimenters used the same procedures as in Experiment 1 to assess procedural fidelity and interobserver agreement. Table 2 summarizes the confederate data relevant to assessing procedural fidelity. Please note that due to camera problems during Casey’s baseline sessions, data collectors could not review her videos and accurately measure confederate behavior for post-hoc analysis. However, a second live data collector did record data on participant behavior during Casey’s baseline sessions. Interobserver agreement for accuracy of mult-FCT implementation was 100% for Nicole and averaged 99% (range: 88%–100%) for Sophia and 86% (range: 55%–100%) for Casey. Interobserver-agreement coefficients for confederate destructive behavior averaged 98% (range: 86%–100%) for Nicole, 89% (range; 86%–94%) for Casey, and 96% (range: 86%–100%) for Sophia. Interobserver-agreement coefficients for confederate FCRs averaged 98% (range: 92%–100%) for Nicole, 98% (range: 92%–100%) for Casey, and 99% (range: 95%–100%) for Sophia.

Procedures

Procedures for Experiment 2 were identical to Experiment 1 except experimenters (a) programmed post-BST sessions immediately after BST for all participants and not just those with baseline performance below 80%, (b) did not implement post-session booster training immediately following errors during post-BST or training-challenge sessions, and (c) randomized the order of the training challenges for each participant. Initial post-BST sessions and training challenges occurred for at least two sessions. If mastery occurred during those two sessions, the condition ended. If mastery did not occur, experimenters conducted a third session. If the third session of post-BST or the training challenge did not result in 100% accuracy, the experimenters delivered booster training as described in Experiment 1. If the participant displayed 100% accuracy during the third session, the experimenters conducted a fourth session to determine if the participant would reach mastery criteria with repeated exposure to the condition. If performance decreased below 100% again, the experimenters delivered booster training at this time. Unlike in Experiment 1, the order of training challenges differed across participants to understand the relevance of order effects on performance during training challenges.

One deviation from Experiment 1 and the rest of participants in Experiment 2 is that we limited Sophia’s training challenges to a maximum of 10 sessions per condition due to her restricted availability for participation. Experimenters concluded her initial training challenge prior to mastery to assess her performance in the other two training challenges.

Results

Figure 2 displays the accuracy of mult-FCT implementation for participants in Experiment 2. None of the three participants displayed mastery-level performance during baseline with the protocol alone, with mean performances of 66%, 50%, and 2% for Nicole, Casey, and Sophia, respectively. The gray checkmarks in Table 3 reflect the errors made by participants in Experiment 2, with the most common baseline errors consisting of accurate S^D initiation and visibility. Sophia had the lowest baseline of any participant across Experiments 1 and 2, implementing most skills incorrectly across her sessions.

FIGURE 2.

Percentage accuracy of mult-FCT implementation (Experiment 2). This figure displays each participant’s accuracy when implementing mult FCT during Experiment 2.

Nicole, Casey, and Sophia all mastered mult-FCT implementation within three sessions during the BST role play. This performance continued at mastery during post-BST for Casey and Sophia and at high levels for Nicole. However, Nicole made errors related to S^D and S^Δ initiation (i.e., starting the reinforcement and extinction components on time) during later post-BST sessions, which warranted booster training.

Like in Experiment 1, the challenge sequence consisting of the schedule, stimulus, and treatment-adherence challenges disrupted performance at least once for all three participants. Across the three participants, there were nine training challenges (i.e., three per participant) and mult-FCT accuracy degraded in 5 of 9 challenges. As seen in Table 3, error types varied for Experiment 2 participants during the training challenges but added to the generality of some error types observed in Experiment 1 (e.g., accurate extinction implementation being difficult during the stimulus challenge).

Recall that participants experienced a fixed order of training challenges (schedule, stimulus, treatment adherence) during Experiment 1, whereas those in Experiment 2 experienced a randomized order of training challenges. Figure 3 displays an across-experiment analysis of disruption across the challenges. The top panel displays the percentage of training challenges that produced disruption (i.e., participant fidelity errors) across consecutive training challenges. During Experiment 1, the first two challenges presented to the participants disrupted performance in an equivalent number of cases, whereas the third challenge did so in only half of cases. The results during Experiment 2 did not follow an orderly pattern, as the first challenge (irrespective of challenge type) disrupted every participant’s performance, the second challenge did not disrupt any participant’s performance, and the third challenge resulted in fidelity errors in a majority of cases. Ultimately, one would need six permutations of the challenges with multiple data sets per group to definitively answer the role of sequence effects; however, these results provide preliminary support for a diminished role of sequence effects in the outcomes of Experiment 1.

FIGURE 3.

Percentage of challenges with errors across challenge number and type. This figure displays the percentage of training challenges that resulted in at least one participant error. The higher the level of the data path, the more participants made an error during that particular challenge number or type. Participants in Experiment 1 experienced the same sequence of training challenges (schedule, stimulus-generalization, treatment-adherence). Experimenters randomized the order of training challenges for participants in Experiment 2.

The bottom panel of Figure 3 displays the same data but according to the type of training challenge. Similar to Experiment 1, participant accuracy during Experiment 2 degraded most often during the schedule challenge (2 of 3 participants) and stimulus challenge (2 of 3 participants) and less so during the treatment-adherence challenge (1 of 3 participants). Taken together, we observed performance degradation in 71% of schedule and stimulus challenges and 43% of treatment-adherence challenges across the experiments. Although the prevalence of response disruption is important, it is equally imperative to understand the magnitude of disruptions when they occur (Mitteer et al., 2022). Figure 4 shows the lowest level of participant accuracy when a training challenge disrupted their performance, with the schedule challenge producing the lowest mean performance during response disruption (76%), followed by the treatment-adherence challenge (78%), and then the stimulus challenge (84%). Although these decrements were not robust, readers may consider the error analysis (Table 3) in which the disrupted treatment accuracy (e.g., implementing extinction incorrectly) may have countertherapeutic effects on client outcomes.

FIGURE 4.

Magnitude of disruption in training challenges. This figure displays the lowest percentage of mult-FCT accuracy during training challenges that produced response disruptions. Gray bars display the mean across participants and white data points represent individual participant data.

A final difference in Experiment 2 is that experimenters did not deliver post-session booster training immediately following a training-challenge session with low performance. Participant accuracy improved with continued exposure in 4 of 5 training challenges that had disrupted performance initially. This corresponds to findings by Berdeaux et al. (2022) showing that accuracy may improve with repeated practice to challenging situations. However, although delaying feedback may be acceptable during simulated trainings, supervisors should still consider delivering immediate correction of staff or trainee errors during client sessions.

GENERAL DISCUSSION

The purpose of this study was twofold. First, we sought to develop a training procedure for mult FCT, which is a highly effective version of FCT (e.g., Greer et al., 2016) but most often used by doctoral-level providers in intensive-outpatient and inpatient programs (Mitteer et al., 2024). In line with the larger BST literature (Lerman et al., 2015; Smith et al., 2024), our training produced mastery-level responding for six participants who did not fully acquire mult-FCT skills during baseline, though additional modifications to BST (e.g., practicing specific skills in pull-out sessions) were needed for one participant prior to concluding BST. Although empirically supported trainings can sometimes be kept behind paywalls or select training opportunities, we placed all of the English, Chinese, and Spanish training materials into a repository for anyone to use (https://zenodo.org/doi/10.5281/zenodo.8360234). Doing so is one example of behavior analysis moving toward “open science,” which involves transparent and accessible knowledge that is developed and shared within the scientific community (Vicente-Saez & Martinez-Fuentes, 2018; see Tincani et al., 2024, for a recent discussion of open-science practices in behavior analysis). We hope other groups will find the training to be helpful in preparing staff and trainees to implement mult FCT with high fidelity.

A systematic review of the BST literature (Smith et al., 2024) found that few studies examined BST’s durability across more than one scenario (e.g., generalization across settings) and only two studies (Giles et al., 2018; Graudins et al., 2012) focused on trainings related to the treatment of challenging behavior. Thus, the second aim of the study was to evaluate the effects of our training across several challenges. We found that the durability of training effects was tenuous, with each participant’s accuracy falling below mastery at least once during the training challenges regardless of the order in which participants experienced them. Recall that we arranged our training challenges in Experiment 1 according to presumed difficulty, with us hypothesizing that the treatment-adherence challenge would be the most difficult based on similar treatment-adherence challenges disrupting implementation in previous studies (Mitteer et al., 2018; Williams et al., 2023). Counter to this assumption, the schedule and stimulus challenges disrupted participant accuracy most often, although the treatment-adherence challenge resulted in pervasive errors for Sophia.

An interesting outcome was that Jolie was the only participant who did not receive BST because she mastered mult-FCT skills in baseline; however, she was the only participant for whom every training challenge disrupted implementation accuracy. The remaining six participants performed at mastery levels during at least one of the training challenges, potentially suggesting that the background information, modeling, and other features of BST may be important for individuals to experience regardless of baseline performance. However, that is difficult to conclude from one data set. Nevertheless, our findings support the recommendation by Smith et al. (2024) that researchers ought to examine multiple forms of training durability. Indeed, sampling only one training challenge may have misrepresented the durability of our mult-FCT training’s effects.

Promisingly, performance often improved during training challenges with exposure alone or with short (e.g., 1–2 min) booster trainings consisting of performance feedback or role play. Trainers should be mindful that initially poor performance during a challenging situation may resolve itself through repeated exposure within a simulated training challenge without intervention by the trainer. One consideration for future research would be to better understand the conditions under which performance improves during training challenges. The findings in Experiment 2 correspond to Berdeaux et al. (2022) showing that accuracy will improve with repeated exposure to a training challenge. Yet, other studies (Mitteer et al., 2018; Williams et al., 2023) have shown the opposite. This may have to do with the training and skill sets of the implementers, as the participants in Experiment 2 and Berdeaux et al. were behavior-analytic trainees and teachers, respectively, whereas participants in Mitteer et al. (2018) and Williams et al. (2023) were caregivers. With that said, trainee errors should likely be addressed immediately if their errors have countertherapeutic effects on client outcomes. Brief performance feedback is relatively low-cost other than factoring in the supervisor’s time (Lerman et al., 2015). For the busy practitioner, this might mean conducting brief in-person or virtual observations when a trainee or supervisee implements mult FCT (a) for the first time with a client, (b) with a substantive protocol change, or (c) during escalated behavioral situations.

A reasonable next step for this training research would be to consider elements of multiple-exemplar training and training loosely (Stokes & Baer, 1977) within the role-play portion of BST (along with baseline and post-BST if conducting an experiment) to bolster durability to post-training challenges. This would expose trainees to various mult-FCT scenarios and might consolidate participant errors to the programmed training period when feedback is readily available and where the effects of trainee errors are isolated to simulations without deleterious effects to their clients. Another possibility would be to pair the new trainee with an expert or more experienced trainee during client sessions while gradually having the new trainee implement more of the mult-FCT components as they and the client meet certain criteria (i.e., zero client destructive behavior, zero trainee errors). As Allen and Warzak (2000) noted in their conceptualization of treatment nonadherence, excessive skill complexity may pose a barrier to accurate protocol implementation. A scaffolding approach to mult-FCT implementation may reduce this complexity during initial training and promote a more errorless approach to learning. Our clinic uses this approach during caregiver training in which the caregiver takes on more of the session responsibilities over time; this may be helpful for staff and trainees as well.

Several limitations worth considering may guide directions for future research. First, participant performance may have differed when working with clients who engage in severe destructive behavior. This study was translational in that it involved scripted confederates who did not cause actual harm to themselves or the participants. Presumably, fidelity errors may be more likely to occur when responding to non-simulated self-injury or aggression. Second, all the participants were graduate students enrolled in a training program, which may have led to improved performance relative to those without higher education because they had received training on topics like extinction prior to study participation. Third, we assessed the extension of training effects to relatively similar types of confederates, discriminative stimuli, and functions. It is unknown how participants would respond to confederates who engaged in very different types of destructive behavior (e.g., pica, elopement) or whose treatment involved dissimilar schedule-correlated stimuli (e.g., wristbands, posters). The edible and attention functions programmed during sessions and stimulus challenge were similar (e.g., both social-positives involving equivalent participant responding); a separate training may be needed for participants to accurately deliver the antecedent and consequence strategies of an FCT treatment for an escape function (e.g., chained FCT; Greer et al., 2016).

Fourth, it is unclear whether the mastery of skills during role play with confederates would generalize outside of the experimental context. Although we observed continued optimal performance from Tom who was employed in our clinic, experimenters were unable to monitor the other six participants following the study. Per our hospital’s regulations, individuals with client-facing roles (e.g., staff, practicum students) must complete hospital onboarding and other necessary trainings (e.g., crisis management, safe-patient handling) prior to working with a client, and this process usually lasts two months. This would have been impractical for the participants and the hospital for the purpose of an experiment. Nevertheless, having data on participant performance during clinical care would bolster the predictive validity of training effects observed during role play or protocol mastery. Additionally, enrolling participants other than graduate students would ensure that the effects of BST and booster trainings generalize to other potential implementers of mult FCT (e.g., caregivers, teachers).

Finally, it is unknown what effect the types and frequencies of participant errors would have on client performance during mult FCT. Although researchers have evaluated the effects of omission errors (i.e., failing to deliver reinforcers for appropriate behavior) and commission errors (i.e., reinforcing target behavior) with differential-reinforcement procedures like FCT, these were not done within the context of a multiple schedule. Because mult FCT’s effects depend on the congruence between the discriminative stimuli and the associated contingencies for FCRs and destructive behavior, such errors might disrupt client discrimination and lead to treatment relapse (e.g., Mitteer et al., 2022). In contrast to reinforcer omission and commission errors, which have been investigated more frequently in behavior-analytic literature (Kimball et al., 2023), the effects of errors like omitting a COD or obscuring the discriminative stimulus have yet to be evaluated in relation to client behavior.

A natural progression of the experiment in this study is to simulate common errors made by the participants within a mult-FCT intervention, perhaps in a translational manner that would avoid exacerbating treatment relapse of severe destructive behavior. For example, Mitteer et al. (2021) emulated caregiver behavior from Mitteer et al. (2018) with three children with autism who emitted arbitrary responses. Although emulation of caregiver errors produced robust increases in target behavior with the three children and often extinguished their alternative behavior, these deleterious effects were isolated to the translational arrangement. A similar approach could be used to reverse-translate the error patterns in the current study with children who experience a multiple schedule in relation to arbitrary responses. Determining if these errors do in fact degrade treatment efficacy would provide clear rationale for resolving errors for particular mult-FCT skills and potentially enhancing the existing BST approach.

DATA AVAILABILITY STATEMENT

All data and training materials are available via Zenodo (https://zenodo.org/doi/10.5281/zenodo.8360234).