Abstract
In this study, we extended the literature on negatively-reinforced mands by teaching multiple novel, socially appropriate alternative mands to two young children with autism spectrum disorder (ASD). First, we replaced existing mands (e.g., pushing away) with two novel, socially appropriate, negatively-reinforced mands. Next, we examined responding under immediate- and delayed-reinforcement conditions to assess resurgence to existing mands and to determine whether the function of the communicative behavior influences the consistency with which different mands are emitted. Finally, we examined generalization to different social partners. Our data suggest that both children acquired the new mands and used them to avoid nonpreferred items. Resurgence to existing mands during delayed-reinforcement conditions was documented for one child, and the sequence in which mands were emitted within a response class was not influenced by the function of the communicative behavior. Generalization data indicate that both children emitted the new mands and one of the two children alternated between the two mands with a social partner who was not involved in the training. We discuss the importance of teaching multiple negatively-reinforced alternative mands to children with autism in applied settings.
Two core characteristics of children with autism spectrum disorder (ASD) are deficits in social and verbal communication (American Psychiatric Association, 2013). These deficits have a negative impact on children’s quality of life by interfering with friendships with peers, by restricting access to leisure activities and community participation, and by increasing social isolation (Knott, Dunlop, & Mackay, 2006; Ozonoff, Goodlin-Jones, & Solomon, 2005). Young children with ASD who do not communicate with language may use such nonvocal behaviors as gestures, body movements, or facial expressions to access reinforcers in their environment (Drasgow & Halle, 1995). That is, these nonvocal behaviors serve communicative functions (e.g., Keen, Sigafoos, & Woodyatt, 2001; Lerman et al., 2005).
One communicative function of nonvocal behaviors is manding to escape or avoid an aversive stimulus (Yi, Christian, Vittimberga, & Lowenkron, 2006). An aversive stimulus is a stimulus for which removal or termination functions as a reinforcer (Skinner, 1953). For example, a child with ASD may learn that turning his head away, a nonvocal behavior, in the presence of a nonpreferred toy offered by a social partner results in its removal, thus producing reinforcement (i.e., termination of the nonpreferred stimulus). Turning one’s head away, however, may not be a socially appropriate behavior to escape a nonpreferred or aversive stimulus and may not be immediately reinforced by social partners across most situations encountered throughout the day.
One example in which a nonvocal behavior may not be immediately reinforced and, therefore, result in escape or avoidance of a nonpreferred or aversive stimulus is related to those situations when social partners are unfamiliar with the child and with his or her behavioral repertoire. Thus, they are unable to identify the function of the nonvocal behavior emitted by the child and do not respond to his or her communicative attempts (Matthews-Somerville & Cress, 2005; Meadan, Halle, & Kelly, 2012). When a social partner does not respond immediately to a child’s communicative attempt, the child may persist with the same nonvocal behavior (i.e., extinction burst; Lerman, Iwata, & Wallace, 1999), may engage in inappropriate behavior that was reinforced by social partners in the past (i.e., resurgence; Lieving, Hagopian, Long, & O’Connor, 2004), or may eventually refrain from further interaction because of a history of past communication failures, thereby continuing to be exposed to the aversive situation (i.e., learned helplessness or extinction; Guess, Benson, & Siegel-Causey, 1985; Seligman, 1975).
Teaching children with ASD to emit socially appropriate and effective negatively-reinforced mands may be a potential avenue for promoting personal independence by allowing children to exert control over their environment, while reducing or preventing resurgence to problem behavior (Borrero & Borrero, 2008; Fritz, Iwata, Hammond, & Bloom, 2013; Groskreutz, Groskreutz, Bloom, & Slocum, 2014; Reeve, & Carr, 2000; Yi et al., 2006). A mand is a verbal operant for which the response is under the control of the motivating operation in effect for a specific reinforcer and reinforcement is mediated by a social partner (Skinner, 1957). For example, a child may say “Turn down the TV, please” to escape a noisy environment or may sign “No, thank you” to avoid a nonpreferred food item.
One strategy that has been used to teach novel, socially acceptable and effective mands to young children with ASD is mand training. Mand training consists of first examining the various topographies of communicative behavior existing in a child’s behavioral repertoire and identifying the function of the child’s behavior (Drasgow, Sigafoos, Halle, & Martin, 2009; Lerman et al., 2005). Next, environmental variables that function as aversive or nonpreferred stimuli are identified through assessment. A novel, socially appropriate and effective mand is then selected as the target behavior to be taught during training. Teaching consists of reinforcing the novel mand while withholding access to the reinforcer (i.e., extinction) for the emission of the inappropriate existing mand (Sigafoos, Arthur-Kelly, & Butterfield, 2006), a strategy referred to as differential reinforcement.
The existing literature on mand training for children with ASD has advanced our understanding of the importance of expanding a child’s behavioral repertoire by teaching novel, socially appropriate, and effective mands to promote increased personal independence in applied settings. However, at least three aspects of mand training warrant further investigation. One aspect pertains to the manding function targeted during training. Although the manding function of escaping or avoiding nonpreferred or aversive stimuli is one of the first communicative functions to emerge in children (Carpenter, Mastergeorge, & Coggins, 1983), a limited number of investigators have addressed the topic of teaching negatively-reinforced mands to escape to children with ASD (Dyer & Kohland, 1991; Frost & Bondy, 2002; Groskreutz et al., 2014; Yi et al., 2006). The participants in the studies above were children with autism between the ages of 4 and 9 years. Most of these children used pictures or signs to communicate two- or three-word phrases with the exception of two children who communicated vocally using three- or four-word utterances. All children engaged in problem behavior including pinching, whining, crying, or biting to escape or avoid nonpreferred items or activities.
A second aspect that warrants further investigation pertains to the number of mands taught during training. The investigators in the above studies, focusing on negatively-reinforced mands, taught only one novel, socially appropriate alternative mand. Teaching only one novel mand may not enable a child to become a resilient communicator when confronted with complex and unpredictable situations encountered in his or her natural environment. For example, a child who learned to sign “stop” emits this behavior in the presence of a peer who is playing loud music in his or her physical proximity. If the peer does not understand sign language, thus failing to provide access to the reinforcer (terminating the music), the child may resort to previously reinforced inappropriate behaviors (e.g., pushing away, hitting) to escape the aversive situation, in what is termed resurgence. Resurgence refers to the emission of a previously reinforced and extinguished response when another response is placed on extinction (Epstein, 1985).
Although the procedures implemented to evaluate resurgence may vary, several conditions are usually necessary when assessing resurgence. The first condition consists of providing immediate reinforcement contingent on the occurrence of a specific response (e.g., pushing away) until a pre-established criterion is met. The second condition consists of immediately reinforcing a different response (e.g., signing “No”) until criterion is met, while reinforcement of the first response is delayed or the response is placed on extinction. The third condition consists of delaying reinforcement for both responses. Resurgence is documented when the first response occurs under delayed-reinforcement conditions at levels that are higher or similar to those documented in the first condition. Resurgence has been demonstrated both in basic studies with non-humans (e.g., DaSilva, Maxwell, & Lattal, 2008; Doughty, DaSilva, & Lattal, 2007; Epstein, 1985) and in applied studies in relation to problem behavior (e.g., Lieving et al., 2004; St Peter Pipkin, Vollmer, & Sloman, 2010; Volkert, Lerman, Call, & Trosclair-Lasserre, 2009).
One potential strategy that may reduce resurgence of existing inappropriate mands and problem behavior in those situations when a child’s communicative attempt is not immediately reinforced by social partners in the natural environment consists of teaching multiple alternative mands within a response class (Drasgow, Martin, Chezan, Wolfe, & Halle, 2016). A response class consists of two or more topographically different behaviors that produce the same effect on the environment and are maintained by the same reinforcer (Catania, 1998; Johnston & Pennypacker, 1993). For example, a child with ASD may escape a nonpreferred activity either by signing “Break, please” or by screaming. Signing “Break, please” and screaming are two topographically different behaviors that produce the same effect on the environment, termination of a nonpreferred activity, thus forming a response class.
One critical feature of teaching multiple alternative mands is response variability. Response variability refers to the extent to which a member of a response class differs from another member of the same response class (Neuringer, 2002). For example, a child with ASD mands to escape from an aversive situation by saying “I do not like that,” “Not now, please,” or “No.” These three mands represent varied responding because they are topographically different from one another but produce the same reinforcer (termination of an aversive stimulus), thus they are members of the same response class. A child who has multiple alternative mands that are under stimulus control in his or her repertoire is more likely to engage in varied responding, thus delaying resurgence of problem behavior when one mand within a response class does not produce immediate reinforcement. For example, a child whose responding is varied may engage in several socially appropriate responses (e.g., touching one’s arm, saying “Talk,” or tapping one’s shoulder) to obtain an adult’s attention when the adult does not immediately respond to the child’s first communicative attempt, rather than engaging in problem behavior or persisting with the same response that may not contact reinforcement.
Hoffman and Falcomata (2014) examined resurgence of socially appropriate, positively-reinforced mands and problem behavior in three children with autism. The authors first taught two sequential novel mands (i.e., card exchange, and microswitch press) to mastery. Then, they implemented a continuous schedule of reinforcement for each mand to establish a reinforcement history with the newly acquired mands, and finally placed each novel mand on extinction. The authors then tested for resurgence of the first mand by placing all responses on extinction. Results indicated that resurgence of first mand occurred during 89% of sessions across participants. In addition, resurgence of socially appropriate mands occurred prior to resurgence of problem behavior. Thus, teaching multiple alternative mands within a response class may be a viable option to prevent or delay resurgence of problem behavior.
A third aspect that warrants further investigation is the generalization of newly acquired negatively-reinforced mands to other items or across contexts and social partners. Generalization of negatively-reinforced mands has been examined by a limited number of researchers (Groskreutz et al., 2014; Yi et al., 2006). The researchers in these studies used a generalization-promoting strategy (i.e., multiple exemplars) to facilitate manding in the presence of untrained nonpreferred items or across settings. Results indicated that generalization to untrained nonpreferred items occurred for all children after a negatively-reinforced mand was taught to mastery in the presence of at least two nonpreferred items. Although generalization across settings was reported in the Groskreutz et al. study, children manded only when the nonpreferred items used in the novel setting were identical to the items used during training (the common stimulus may have been the mediator). For example, children manded when a song was played via an MP3, but not when the song was sung live by a group of people. Instead of assessing generalization across items or settings, we chose to examine generalization across adult social partners.
Therefore, the purpose of our study was to extend the existing literature on teaching negatively-reinforced mands to children with ASD in applied settings. First, we examined the effectiveness of mand training in replacing existing mands in children with ASD by teaching two new, socially appropriate, negatively-reinforced alternative mands. Second, after the acquisition of each mand, we assessed their manding under immediate- and delayed-reinforcement conditions to test for resurgence to existing mands and to examine the structure of the response class. Third, we taught children to alternate between the two new mands to assess how engaging in varied responding influenced resurgence of previously reinforced mands. Finally, we assessed generalization of the two new mands to social partners without including any additional generalization-promoting strategies.
Method
Participants
Two children with a diagnosis of ASD who attended a self-contained elementary classroom for children with autism and developmental disabilities in a local suburban school participated in the study. The children received their diagnosis of ASD based on a psychological evaluation conducted by a licensed psychologist. Isaac was a 4-year-old male and Brenda was a 3-year-old female. Both children functioned within the severe range both intellectually and adaptively based on formal evaluation data on the Bayley Scales of Infant Development, Second Edition (Bayley, 1993) and the Vineland Adaptive Behavior Scales, Second Edition (Vineland-II, Sparrow, Cicchetti, & Balla, 2005). No standardized measures assessing the communication level of children were available at the time the study was implemented. The teacher nominated these children for the study because they (a) were nonvocal; (b) did not use sign language, picture systems, or other socially appropriate conventional modalities serving a negatively-reinforced manding function; and (c) used problem behaviors to reject nonpreferred items and activities offered to them by their primary teacher or other staff members. Problem behavior included crying, pushing others, and walking away. Both children had a limited communicative repertoire consisting of approximately two gestures (e.g., reaching or leading) and two or three signs (“Cookie” or “More”) serving a positive-reinforcement function.
Setting
The study was conducted at a local elementary school in the southeast. We conducted one training session per day 4 to 5 days a week in the elementary classroom during snack time. Six students, one special education teacher, and two paraprofessionals were present in the classroom during each session. Each training session occurred in the morning during the normal classroom routine at a table at the back of the classroom where snack typically occurred. Food was generally available only during snack time and teachers followed their normal schedule when the trainer was not present in the classroom or did not run the intervention procedures.
A small table and four chairs, located in the center of the room, comprised the snack area. Each session was conducted in a one-on-one format with the trainer and no other peers. The other students, the special education teacher, and the paraprofessionals continued to follow their normal schedule of activities during our training sessions. We used a digital camera and tripod positioned behind the trainer to record each training session.
Target Behaviors and Recording System
We adapted the recording and intervention procedures implemented in Drasgow et al. (2016) to conduct the present study. We assessed four target behaviors. First, we recorded existing mands: The child used pushing or looking away to avoid nonpreferred food items. Pushing away was defined as using an arm or hand to deflect a social partner’s hand or object when the social partner was offering or presenting a hand-held food item. Looking away was defined as turning head away from an offered hand-held food item. Second, we recorded “Card” if the child picked up a rejection card and extended his or her arm with the card facing toward the social partner. The rejection card consisted of a laminated square card (4 in. × 4 in.) marked with an X in the middle. Third, we recorded “No” if the child turned his or her head from side-to-side 2 or more times when offered a hand-held food item. Fourth, we recorded “Alternation” if the child switched from one new mand (e.g., “Card”) as a first response to the second new mand (e.g., “No”) as a second response when the first response was not immediately reinforced. Finally, we recorded any occurrences of problem behavior that consisted of crying, pushing others, or walking away.
We used a response-per-opportunity recording method. An opportunity consisted of the trainer or social partner presenting a nonpreferred food item (i.e., a trial) that required a response from the child. For each trial, we recorded the prompted or independent occurrence of the four target behaviors and documented the occurrence or non-occurrence of problem behavior. We further classified these target behaviors as the first form or as a delayed form based on the order in which they occurred during a trial. The first form was defined as the initial response emitted by the child when an adult partner (i.e., the trainer or the teacher) presented or offered a food item. Delayed forms were defined as any additional responses emitted by the child during a trial when the adult did not immediately remove the food item contingent on the child’s first response. Delayed forms consisted of existing mands, newly taught alternative, socially appropriate mands, or other responses (e.g., screaming, vocalizations) not observed during the initial assessment. We recorded the first form and the sequence of all additional forms during delayed-reinforcement trials when (a) a form met an operational definition and (b) there was at least a 1-s pause between identical forms.
Experimental Design and Trainers
We used an adapted reversal design (Betz, Higbee, Kelly, Sellers, & Pollard, 2011; Drasgow et al., 2016) to assess the effectiveness of mand training for (a) teaching new, negatively-reinforced mands; (b) teaching an alternation strategy to engage in varied responding; (c) producing generalization to social partners; and (d) assessing resurgence to existing mands and the consistency with which different mands appeared in sequence. We assessed during both immediate- and delayed-reinforcement trials. An immediate-reinforcement trial consisted of an adult removing a nonpreferred food immediately contingent on the child’s initial response. A delayed-reinforcement trial consisted of an adult inserting a 6-s delay between the emission of the child’s initial response and accessing the reinforcer (i.e., removal of the nonpreferred food).
Our adapted reversal design (i.e., ABACADA) consisted of the same sequence of baseline and intervention conditions included in a reversal design with one exception. Each intervention condition targeted the acquisition of either a novel mand (i.e., B and C) or of alternating mands (i.e., D) within the same response class. Our baseline condition was consistent with a conventional reversal design and we used the same intervention procedures to teach each new mand and alternation. Although our design did not enable us to demonstrate experimental control of our intervention effect through replication of the same behavior, it did allow us (a) to document acquisition of the two new mands and their alternation when and only when intervention was introduced, (b) to evaluate resurgence of existing mands, (c) to record the sequence with which mands were emitted within the response class, and (d) to evaluate generalization of the new mands and alternation with social partners different than the trainer.
The trainer for this study was a doctoral student in the Department of Special Education. He was a certified behavior analyst with more than 15 years of experience providing behavior support services to individuals with disabilities in applied settings. The teacher had a master’s degree in special education and 6 years of teaching experience working with children with autism in a preschool setting. She also was a doctoral student in the Department of Special Education at the time of the study.
Procedures
Preference assessment
We conducted a paired-stimulus preference assessment (Fisher et al., 1992) and ranked the food items for each child as high preferred and low preferred. Results indicated that the food items receiving the highest rankings (highly preferred) consisted of (a) gold fish, crackers, chips, Kool-Aid, cookies, cheerios, and skittles for Isaac and (b) gold fish, candy, and juice for Brenda. Findings also indicated that the food items receiving the lowest rankings (lower preferred) consisted of (a) broccoli, cauliflower, and water for Isaac and (b) carrots, apples, cheese, and water for Brenda.
Identification of existing mands
Prior to baseline, we identified existing positively and negatively-reinforced mands. These procedures consisted of indirect methods (i.e., interviews with the teacher and parents) and direct methods (i.e., A-B-C recording, direct observation). We observed each child for approximately 3 hr in the school setting over several days. Results revealed that both children emitted pushing away and looking away as their primary negatively-reinforced mands for avoiding or escaping aversive stimuli. Data also indicated that both children emitted positively-reinforced mands consisting of either reaching or signing “Cookie” or “More” to obtain access to preferred food items. When presented with a nonpreferred item, both children occasionally engaged in various topographies of problem behavior including crying, pushing others, and walking away.
Assessment of existing mands
We established a routine for both children. The purpose of the routine was to systematically assess the hypothesized function of existing mands, identified previously, and to teach new, socially appropriate alternative mands. Prior to the establishment of the routine, the adults (i.e., trainer and teacher) received behavioral skills training (Nigro-Bruzzi & Sturmey, 2010) consisting of instruction, modeling, rehearsal, and immediate feedback on how to implement the protocol. The training ended when the trainer and the teacher implemented with fidelity at least 80% of the steps for three consecutive training sessions for each condition included in the intervention. The routine consisted of the trainer first arranging the preferred and the nonpreferred food in proximity to where the child would sit, but out of his or her sight and reach. Next, the trainer gathered the materials (e.g., bowls, place mat, additional food, spoons, and napkins) necessary to conduct a session by placing them on a table in front of where the child would sit. Then, the trainer brought the child to the table and had him or her sit.
Next, the trainer placed a snack-size portion (e.g., three gold fish or two pieces of candy) of preferred food on a napkin on the table out of the child’s reach, but in sight and waited for the child to respond. Each child responded by reaching or signing “Cookie” or “More.” Contingent on the child’s response, the trainer provided access to the food and the child consumed it. On randomly selected trials, the trainer placed the food in front of the child within reach where the child could independently access it, thus providing noncontingent access to a preferred food. The purpose of the noncontingent access was to determine whether the child would emit one of the mands identified previously or would access the food independently. On these trials, reaching or signing “Cookie” or “More” did not occur. These mand forms occurred only when access to a preferred food was restricted (i.e., the food was in sight but out of reach) and did not occur when access to a preferred food was readily available (i.e., food placed in front of the child within reach). Therefore, we concluded that reaching or signing “Cookie” or “More” served a manding function of requesting access to preferred food; thus, they were positively-reinforced mands. The trainer randomly selected which highly preferred food to use during a trial, and reinforced all three responses (i.e., reaching, and signing “Cookie” or “More”) emitted during this condition by providing access to the preferred food.
We conducted the same procedures with nonpreferred food with one exception: The trainer either presented a nonpreferred food to the child by holding it within 8 inches of the child’s chest or placed it on the table in the child’s sight but did not directly offer it to the child. Both children emitted looking and pushing away when a nonpreferred food was presented by the trainer. Looking and pushing away failed to occur when the trainer placed a nonpreferred food on the table within the child’s sight, but did not directly offer it to the child. Thus, we determined that looking and pushing away served a manding function of avoiding nonpreferred food, thus they were negatively-reinforced mands. The trainer reinforced the two responses (i.e., pushing and looking away) emitted during this condition by removing the nonpreferred food. The procedure was repeated for 8 to 10 min and conducted for 4 weeks prior to the child entering baseline, thus allowing the trainer multiple opportunities for assessing existing mands and establishing stimulus control over the mands prior to baseline. Neither of the children engaged in problem behavior during these trials.
Baseline
Baseline procedures were consistent with the procedure used during the assessment of existing mands. The trainer randomly selected the preferred and nonpreferred food items to be used during a session from the food items receiving the highest and lowest rankings during the preference assessment. Baseline sessions consisted of approximately 10 preferred-item trials (opportunities) to request food and approximately five nonpreferred-item trials (opportunities) to reject food during snack time in the classroom. Occasionally, we would conduct more or fewer trials depending on the child’s interest and motivation on that particular day. On two nonpreferred-item trials, immediate reinforcement was delivered for the first form emitted and on three nonpreferred-item trials, reinforcement was delayed for 6 s to allow the child time to emit more than one form.
The order of the immediate- and delayed-reinforcement nonpreferred-item trials was unpredictable throughout the 15-trial session. The first trial of every session was a preferred-item trial; otherwise, we randomly embedded the immediate- and delayed-reinforcement nonpreferred-item trials within the preferred-item trials to reduce the likelihood of creating an aversive situation for the children during the assessment. The preferred-item trials were all immediate-reinforcement trials. During baseline, the trainer reinforced all negatively and positively-reinforced mands emitted by the child. The exact number of preferred- and nonpreferred-item trials during immediate- and delayed-reinforcement conditions varied because they were mediated by the child (e.g., a preferred item could have been rejected by the child). The purpose of this procedure was to ensure that the trainer provided consequences corresponding to the motivating operation in effect at that time.
The protocol for conducting an immediate-reinforcement preferred-item trial consisted of the adult (i.e., trainer, teacher) first placing a small portion of preferred food in front of the child, but out of his or her reach. Second, the adult waited 3 to 5 s for a request (i.e., either with an existing mand or one of the two identified signs). If the child did not emit a mand within 3 to 5 s, the adult “straightened or arranged” the food (to direct the child’s attention toward the food) and gave the child an additional 5 s to request the food. If the child did not respond to this second opportunity, the adult removed the food and assumed that it was not sufficiently reinforcing at that moment. After 3 to 5 s, the adult then presented the next trial with a different highly preferred food. If the child requested the food within 3 to 5 s, the adult immediately provided the item contingent on the child’s request. If the child rejected the food either with an existing mand or with one of the two pre-established, functionally equivalent, alternative responses, the adult immediately removed the food contingent on the child’s response (i.e., negative reinforcement).
The protocol for conducting an immediate-reinforcement nonpreferred-item trial consisted of the same protocol as the preferred-item trials except that the adult offered a nonpreferred food item to the child. The protocol for conducting a delayed-reinforcement nonpreferred-item trial consisted of the same protocol as the immediate-reinforcement nonpreferred-item trials except that when the child emitted the first response, the adult looked at the child and continued to present the item by holding it within 8 inches of the child’s chest instead of immediately providing reinforcement (termination of the aversive stimulus by removing the unpreferred food). The purpose of the 6-s delay was to simulate a naturalistic communication breakdown (i.e., a delay in reinforcement due to a social partner failing to understand or acknowledge the communicative signaling of another) that might occur when a parent does not respond immediately to a child’s initial response. We selected the 6-s delay (a) to provide the child sufficient time to emit more than one mand and yet (b) to lessen the likelihood of escalation to problem behavior by keeping the delay brief (Dixon & Cummings, 2001). If the child emitted additional responses within the 6 s, the adult continued to glance at the child, recording each mand form, and after 6 s removed the food item, thus providing the reinforcer (termination of aversive stimulus). If the child did not respond during the 6-s delay, the adult re-presented the item by saying “Come on, try a little bit.” If the child did not emit a second mand to reject the item (i.e., persisted with the first mand), the adult continued to look at the child, recorded each occurrence of the repeated mand (i.e., defined as a 1-s delay between mands) and after 6 s removed the food, and then presented the next trial. If the child requested the item, the adult immediately provided the food contingent on the child’s response and recorded the emitted form as a request. If the child did not consume the food item within 6 s, the adult removed the food, and presented the next trial. The adult ended a session 30 to 40 s after the final trial.
Intervention
Mand training “Card” (B)
Phase B of the intervention began when the adult implemented mand training to replace existing mands with the use of a “Card.” The trainer conducted all intervention sessions and all trials were immediate-reinforcement trials (i.e., the trainer provided the reinforcer immediately contingent on the child’s first response). During Phase B, the trainer reinforced all prompted and independent responses consisting of the use of “Card” for nonpreferred-item trials, and the emission of either of the two positively-reinforced mands (i.e., “Cookie,” or “More”) for preferred-item trials and placed all existing mands and problem behavior on extinction. Similar to baseline, each “Card” trial began when the trainer placed a preferred or a nonpreferred food in the child’s sight, but out of reach. The trainer then waited 3 to 5 s for the child to respond. If the child initiated a request with one of the two identified mands (i.e., “Cookie,” or “More”), the trainer immediately responded by providing the child with a small portion of whatever food was available at the time. If the child failed to respond within 3 to 5 s, the trainer “straightened or arranged” the food item and gave the child an additional 3 to 5 s to respond. If the child requested the item, the trainer immediately responded by providing the child with the food item; if the child did not respond during this second opportunity, the trainer removed the item and then presented the next trial with a different food item after 3 to 5 s elapsed. If the child rejected the item with an existing mand, the trainer immediately physically prompted (e.g., guided the child’s hands) the child to pick up the card and extend his or her arm with the card in the direction of the trainer, who then removed the item contingent on the response of extending the card. If the child rejected the item with one of the two pre-established, functionally equivalent, alternative responses, the trainer immediately responded by removing the item contingent on the occurrence of the child’s response and refrained from providing verbal praise or physical touch. We systematically faded our prompts across sessions as the children acquired the new mand. We did this by decreasing the amount of physical assistance the child needed to emit the new response from full physical to partial physical to shadowing the child’s hands as he or she emitted the new mand (Wolery, Ault, & Doyle, 1992).
When data indicated that the child met the acquisition criterion of using the “Card” independently on 100% of the trials for three consecutive snack sessions including using the “Card” on the first trial of each session, we returned to baseline conditions. The conditions programmed during the second baseline were the same as those in the original baseline. During delayed-reinforcement trials, we alternated social partners (i.e., trainer and teacher) to assess the child’s generalized use of the “Card” with a different adult than the one who delivered training. The protocol for conducting delayed-reinforcement nonpreferred-item trials during the second baseline (after acquiring the use of the “Card”) was consistent with delayed-reinforcement trial procedures conducted during the initial baseline (Phase A) except that we expected the use of the “Card” in addition to existing mands as a potential response option.
Mand training “no” (C)
The instructional procedures for Phase C of the intervention were the same as those for Phase B, except if the child rejected a food item within 3 to 5 s by using an existing mand or the “Card,” the trainer immediately physically prompted the child to sign “No” by gently physically turning his or her head to the left, then right, and back to the left 2 or more times. During this phase, the trainer reinforced all prompted and independent responses of “No” and placed existing mands, the “Card,” and problem behavior on extinction. After meeting the criterion for acquiring “No,” the protocol for conducting delayed-reinforcement nonpreferred-item trials during the third baseline was the same as that for the “Card” except that the child could emit existing mands, the “Card,” or “No” to reject food.
Alternation (D)
We implemented Phase D to teach the two children to engage in varied responding by alternating between the two newly acquired alternative replacement mands; when one mand did not produce immediate reinforcement, switching to the other was reinforced. During this phase, the trainer reinforced the second mand form (e.g., “No”) that differed from the first (e.g., “Card”) and placed all existing mands and problem behavior on extinction. Phase D consisted of 10 preferred-item trials (opportunities) to request food and approximately five nonpreferred-item trials (opportunities) to reject food: two immediate-reinforcement trials and three delayed-reinforcement trials. The immediate-reinforcement trials consisted of the trainer removing food contingent on the child’s use of either the “Card” or “No” as the first response. Each delayed-reinforcement trial began with a 3 to 5 s delay following the child’s use of either the “Card” or “No” to reject a food item. Neither of the children used existing mands as their first response during this phase. The adult waited 3 to 5 s to assess whether the child would emit the second functionally equivalent trained response during the delay to reject the food. If the child used a second mand that differed from the first (i.e., “Card” or “No”) within 3 to 5 s, then the adult removed the food. If the child used the same mand on both occasions, the adult prompted the “other” mand. Specifically, if the child used the “Card” as a first mand and as a second mand within 3 to 5 s, then the adult physically prompted “No” and immediately removed the food. If the child failed to respond with the second trained mand within 3 to 5 s of emitting the first, the adult physically prompted the mand not used as the first form. If the child requested the food item with either an existing mand or one of the two identified forms (i.e., “Cookie” or “More”), then the adult provided the food contingent on the child’s request. When data indicated that the child alternated responding for 90% of more of the delayed-reinforcement trials for three consecutive sessions, we returned to baseline conditions. The protocol for conducting delayed-reinforcement trials during the fourth baseline, after Phase D, was the same as that employed for the prior three baseline phases.
Generalization probes
We conducted immediate- and delayed-reinforcement generalization probes during each of the four baselines. The purpose of generalization probing was to determine the mands that the children would emit in the presence of a familiar social partner (i.e., a teacher who was not associated with training) before intervention, after the acquisition of each mand, and after they mastered alternation between the two newly acquired mands. We selected the teacher to conduct generalization probes because (a) she consented to participate in the study, (b) she did not conduct acquisition training sessions with children, and (c) she would be a social partner in the children’s everyday settings. The protocol for conducting immediate- and delayed-reinforcement trials was the same as the protocol implemented during baseline sessions.
Interobserver Agreement (IOA)
The trainer videotaped all sessions and coded children’s responses after each session. The trainer served as the primary recorder during immediate- and delayed-reinforcement trials conducted by him; the teacher served as the primary recorder during all trials conducted by her during baseline sessions. Three special education graduate students served as reliability observers. The trainer provided instruction to the reliability observers prior to their collecting IOA data. Instruction was complete when the trainer and a reliability observer reached 80% or better agreement on each coding category for three consecutive days. The primary recorder and the reliability observer independently watched the videotapes and coded the occurrence and sequence of each target behavior for each trial. We collected IOA on at least 30% of the sessions across all baseline and intervention phases for each child. An agreement was scored if the primary recorder and the reliability observer recorded (a) the same mand on immediate-reinforcement trials and (b) the same occurrence and sequence (i.e., point-by-point) of mands on delayed-reinforcement trials. We calculated the percentage agreement scores by dividing the total number of agreements by the total number of agreements plus disagreements and multiplying the quotient by 100.
For Isaac, we calculated IOA for 36% (942 of 2,600) of the immediate-reinforcement preferred trials, for 36% (324 of 896) of the immediate-reinforcement nonpreferred trials, and for 41% (20 of 49) of the delayed-reinforcement nonpreferred trials. Agreement on immediate-reinforcement (i.e., first form) was for 99% (934 of 942) for preferred trials and 100% (324 of 324) for nonpreferred trials. Agreement on delayed-reinforcement (i.e., multiple forms) nonpreferred trials was 100% (46 of 46). For Brenda, we calculated IOA for 33% (583 of 1,800) of the immediate-reinforcement preferred trials, for 34% (129 of 381) of the immediate-reinforcement nonpreferred trials, and for 37% (16 of 43) of the delayed-reinforcement nonpreferred trials. Agreement on immediate-reinforcement trials was 99% (580 of 583) for preferred trials and 100% (129 of 129) for nonpreferred trials. Agreement on delayed-reinforcement trials was 100% (16 of 16).
Treatment Fidelity
The first author collected procedural fidelity data for 20% of the sessions for each baseline and intervention phase for each child with the exception of the last baseline phase for Brenda. During the last baseline phase for Brenda, school officials asked us to conduct the remainder of the study in the cafeteria. However, we could not videotape these sessions because other students for whom we did not have permission to videotape were present in the cafeteria at the time of the study.
The first author watched the videotapes and used a checklist to collect procedural fidelity data. The checklist consisted of a task analysis of the steps required to conduct preferred and nonpreferred, immediate- and delayed-reinforcement trials. She scored each step indicated on the checklist as completed or not completed. We calculated procedural fidelity by dividing the number of steps completed by the total number of steps completed plus not completed and multiplying the quotient by 100. Procedural fidelity was 100% for all preferred and nonpreferred immediate- and delayed-reinforcement trials for both participants during each baseline and intervention session.
Results
Isaac
Figure 1 displays the percentage of Isaac’s independent first form responses during Phases A (i.e., all four baselines), B (i.e., “Card”), and C (i.e., “No”), and the percentage of independent alternation in Phase D. Data prior to the final three sessions that met the criterion are presented in seven-session blocks (“Card” phase) and four-session blocks (“No” and “Alternation” phases). To interpret Isaac’s data and to determine the effectiveness of our intervention, the pattern of responding during Phases B, C, and D should be analyzed in relation to the purpose of each phase. Specifically, our purpose in Phases B and C was to introduce in Isaac’s communicative repertoire two new functionally equivalent, socially appropriate, negatively-reinforced mands (i.e., “Card” and “No”). Therefore, the pattern of responding should assume an upward trend and an increase in the level of newly taught mands (i.e., an acquisition curve) that stabilizes at 100% for the final three consecutive sessions. Our purpose in Phase D was to increase the likelihood of varied responding by teaching Isaac to spontaneously alternate between the two newly acquired mands. Thus, the expected pattern would be less stable compared with Phases B and C and would reflect more variability with the exception of the last three sessions, when the pattern stabilized at 100%, suggesting that Isaac acquired the independent alternation strategy.
Isaac’s percent of first form responses for each mand type under immediate (open) and delayed (closed) reinforcement trials and independent alternation from first to second form.
Figure 2 presents the cumulative number of communication forms on delayed-reinforcement trials during the four baseline phases. During the first baseline, Isaac never emitted (as first or additional forms) either of the two yet-to-be-taught negatively-reinforced mands to avoid nonpreferred food during immediate- and delayed-reinforcement trials with either the trainer or the teacher. Isaac required 97 sessions and a total of 524 trials to reach the acquisition criterion for “Card” (i.e., using the “Card” independently on 100% of the trials for three consecutive snack sessions and using the “Card” on the first trial of each session), 47 sessions and a total of 248 trials for “No,” and 15 sessions and a total of 46 trials for “Alternation.”
Cumulative distribution of communication forms across the four baselines during delayed-reinforcement trials.
During the second baseline, Isaac used the “Card,” the newly acquired mand, as a first form on 100% of the immediate- and delayed-reinforcement trials with the trainer. Generalization data revealed that he used the “Card” as a first form on 100% of the immediate- and delayed-reinforcement trials when the teacher conducted the trials. Delayed-reinforcement data revealed that Isaac used the “Card” as a second or third response during the 6-s delay. During the third baseline, Isaac emitted “No” as a first form on 100% of the immediate- and delayed-reinforcement trials in the two sessions in which the trainer conducted the trials and in the two sessions with the teacher (Generalization data). Delayed-reinforcement data revealed that Isaac emitted “No” as a second or third form during the 6-s delay, but he did not use the “Card.” During the fourth baseline, Isaac emitted the two new mands (i.e., “Card” and “No”) as a first form during immediate- and delayed-reinforcement trials with the trainer and the teacher. He alternated on 100% of the trials with the trainer in one session and failed to alternate when the teacher conducted the trials. Delayed-reinforcement data revealed that Isaac emitted only one mand (i.e., “No”) as a second response and he never used the “Card” as an additional response during the 6-s delay. Isaac did not engage in any occurrences of problem behavior throughout the study. He did not show any signs of food selectivity after mand acquisition.
Brenda
Figure 3 displays Brenda’s independent first form occurrences during Phases A (i.e., all four baselines), B (i.e., “Card”), and C (i.e., “No”), and the percentage of independent alternation in Phase D. Data in sessions prior to mastery are presented in four-session blocks (“No” and “Alternation” phases). Brenda’s data should be interpreted using the same guidelines described for Isaac. Figure 4 presents the cumulative number of communication forms on delayed-reinforcement trials during the four baseline phases. During the first baseline, Brenda did not emit the two new negatively-reinforced mands to avoid nonpreferred food during immediate- and delayed-reinforcement trials with either the trainer or the teacher. Delayed-reinforcement data revealed that she never used either of the two new mands as additional forms to avoid nonpreferred food. Brenda required nine sessions and a total of 135 trials to reach the acquisition criterion for “Card,” 28 sessions and a total of 416 trials for “No,” and 13 sessions and an additional 10 booster sessions for a total of 613 trials for “Alternation.”
Brenda’s percent of first form responses for each mand type under immediate (open) and delayed (closed) reinforcement trials and independent alternation from first to second form.
Cumulative distribution of communication forms across the four baselines during delayed-reinforcement trials.
During the second baseline, Brenda emitted the newly acquired mand (i.e., “Card”) as a first form to avoid nonpreferred food on 100% of the immediate- and delayed-reinforcement trials with the trainer. Generalization data revealed that Brenda used the “Card” as a first form on 100% of the immediate- and delayed-reinforcement trials when the teacher conducted the trials. Delayed-reinforcement data indicated that Brenda used the “Card” as a second, third, or fourth response during the 6-s delay. During the third baseline, Brenda emitted “No” as a first form on 67% and 75% of the immediate-reinforcement trials and on 100% and 0% of the delayed-reinforcement trials in the two sessions with the trainer. She also used existing mands on 100% of the immediate-reinforcement trials with the trainer in one session. Generalization data indicated that Brenda used “No” as a first form on 33% and 43% of the immediate-reinforcement trials and on 0% of the delayed-reinforcement trials in the two sessions when the teacher conducted the trials. She also used existing mands as a first form on 100% of the delayed-reinforcement trials with the teacher. Delayed-reinforcement data revealed that Brenda emitted “No” as a second, third, or fourth response during the 6-s delay. During the fourth baseline, Brenda emitted the two new mands (i.e., “Card” and “No”) as first responses during immediate- and delayed-reinforcement trials with the trainer and the teacher. She alternated on 33% of the trials during both sessions with the trainer and in one session with the teacher and she alternated on 66% of the trials in one session with the teacher. Delayed-reinforcement data revealed that Brenda used both forms as a second, third, or fourth response during the 6-s delay. Brenda engaged in no occurrences of problem behavior throughout the baseline and intervention phases of the study. She did not show any signs of food selectivity after mand acquisition.
Discussion
Our first purpose in this study was to evaluate the effectiveness of mand training in replacing existing mands with two new, negatively-reinforced, socially appropriate alternative mands, and of teaching two children with ASD and language delays to engage in a varied pattern of responding by alternating between the two newly acquired mands. Our second purpose, after the acquisition of the new mands and alternation between them, was to assess children’s responding under immediate- and delayed-reinforcement conditions to test for resurgence to existing mands and to determine the consistency with which new mands appeared in sequence. Our third purpose was to determine the generalization of the two new mands to different social partners who were not involved in training without including any additional generalization-promoting strategies.
Baseline data suggest that, before introducing mand training, the two children used existing mands to avoid nonpreferred food during both immediate- and delayed-reinforcement nonpreferred-item trials. Acquisition data reveal that mand training was effective in replacing existing mands with two new, negatively-reinforced, alternative mands for both children. Data also suggest that for Brenda the alternation training was effective in producing varied responding when we ignored the first mand, whereas Isaac engaged in varied responding on only one occasion with the trainer. Generalization data indicate varied patterns of responding as suggested by the alternation of the two newly acquired mands in the presence of a different social partner. Isaac did not alternate between the new mands when the teacher conducted the trials and Brenda displayed moderate levels of alternation on generalization trials (i.e., 50%, range = 33%-66%). Delayed-reinforcement data show that both children used the new mands as additional responses when social partners did not provide access to reinforcer (termination of aversive stimulus) contingent on the child’s first response. However, various levels of resurgence to existing mands during delayed-reinforcement were documented for both children.
The findings of our study provide empirical evidence supporting the use of mand training as an effective intervention when teaching two alternative, socially acceptable mands that serve a negative-reinforcement function to young children with ASD and severe language delays. Data suggest that our mand training procedures consisting of prompting, differential reinforcement, and extinction resulted in acquisition of the two negatively-reinforced, alternative communication responses for both children. The clinical significance of these results is important for at least two reasons. First, it emphasizes the importance of using instructional strategies to promote varied responding when implementing communication interventions for young children with ASD and language delays. Varied responding that allows a child to alternate between different existing and newly acquired mands may prevent or delay resurgence to problem behavior by increasing the likelihood that the newly acquired mands may resurge prior to more severe inappropriate behavior (Hoffman & Falcomata, 2014). Second, it adds additional empirical evidence to the limited body of literature on the effectiveness of mand training to produce acquisition of negatively-reinforced mands to avoid or escape a nonpreferred stimulus (Groskreutz et al., 2014; Yi et al., 2006).
The findings of the study also indicate that our mand training procedures were effective in promoting varied patterns of responding as suggested by the alternation between the two negatively-reinforced mands for one of the two children. However, one child (i.e., Isaac) did not spontaneously alternate between the two newly acquired mands and instead resorted to existing mands (i.e., pushing away) when the first form did not result in immediate-reinforcement. One potential explanation for the lack of spontaneous alternation between the two mands may be related to behavioral rigidity. Individuals with autism are characterized by restrictive and repetitive behavior (American Psychiatric Association, 2013). For example, a child with autism may always reject a nonpreferred item by pushing the item away instead of using various response forms (e.g., “no,” “not now, thank you”). When only one form is reinforced by social partners in the environment, the child does not have to engage in varied responding to contact reinforcement. In our study, it is possible that teachers and caregivers may have reinforced existing mands used by Isaac to avoid nonpreferred food throughout the day while the study was ongoing, and thus, resulting in Isaac’s lack of alternation between the newly acquired mands.
Data on the response pattern during delayed-reinforcement conditions provide supporting evidence that different mands within a response class may not always occur in a stable hierarchy. Our results were consistent with findings of previous studies on positively-reinforced mands indicating that some children with ASD may not emit mands in a predictable temporal sequence when social partners do not provide immediate access to reinforcers (Drasgow et al., 2016). Specifically, after acquiring the new mands and the alternation strategy, children did not emit the existing or new mands in a predictable and consistent sequence across various opportunities. For example, Isaac emitted either “Card” or “No” as a first response during delayed-reinforcement trials; he also engaged in existing mands or “No” as a second response during delayed-reinforcement trials. The same pattern of responding was documented for Brenda. For example, Brenda emitted “Card,” “No,” or existing mands as a first response across different trials. She also emitted “Card,” “No,” existing mands, or new mand topographies as a second response across trials.
Furthermore, data on the response pattern during delayed-reinforcement conditions provide supporting evidence that the structure of the response-class hierarchy is not influenced by the function of the communicative behavior. In our study, we examined the sequence of responses under delayed conditions after the acquisition of two negatively-reinforced mands and the alternation between them. Our results reveal that the order in which the newly acquired, negatively-reinforced mands occurred under delayed conditions was not always stable and predictable. These findings are consistent with the results documented in previous studies with children with ASD that examined the structure of positively-reinforced response class members (e.g., Drasgow et al., 2016). These studies indicated that different positively-reinforced response-class members did not occur in a predictable temporal sequence during delayed-reinforcement conditions.
Our results suggest that extinction of a newly acquired negatively-reinforced mand may result in resurgence to existing mands that were reinforced by social partners in the past. These findings are closely aligned with previous results reported in the literature on resurgence during differential reinforcement procedures (e.g., Lieving et al., 2004; Wacker et al., 2013). In our study, resurgence of existing mands was documented for both children during delayed-reinforcement trials. However, the pattern and the frequency of responses were different for each child. For Isaac, resurgence consisted of initial forms of existing mands (i.e., pushing away, looking away) documented during our initial assessment. For Brenda, resurgence consisted of initial forms of existing mands (i.e., looking away, pushing away) and other responses (i.e., screaming and vocalizations) that were not observed during the initial assessment.
One potential explanation for the differences in the relative frequency and the existing mands response pattern after the acquisition of the new replacement mands for the two children may be related to the history of reinforcement for existing mands. Previous research indicates that the length of the reinforcement history influences the probability of resurgence during extinction. Specifically, responses with a longer history of reinforcement are more likely to resurge compared with responses with a shorter history of reinforcement (Bruzek, Thompson, & Peters, 2009). In our study, certain topographies (e.g., screaming, vocalizations) may have been reinforced for a longer time compared with other topographies (e.g., pushing or looking away) and, thus, resurged first when existing mands were placed on extinction.
Anecdotal reports from teachers indicate that, over the last year since Brenda started attending the preschool program, she engaged in screaming and vocalizations when presented with a nonpreferred item during class routines. This behavior resulted in the teacher or other staff members removing the nonpreferred item to prevent escalation to more problematic behavior, thus facilitating a history of reinforcement for the above communicative responses in the presence of nonpreferred items. It is possible that the teacher served as a discriminative stimulus that increased the likelihood of topographies such as screaming and vocalizations when newly acquired mands were not reinforced immediately. The above topographies of behavior did not occur in the presence of the trainer who did not have a history of reinforcement with Brenda. In his presence, Brenda was more likely to emit existing mands that were reinforced during the initial phases of the study. Promoting response variability may be difficult when children have a long reinforcement history for emitting only one or two response topographies within a response class (Wolfe, Slocum, & Kunnavatana, 2014).
Our findings indicate that both children generalized the new negatively-reinforced mands in the presence of different social partners who were not involved in training without any additional generalization-promoting strategies. This finding is consistent with previous studies demonstrating that negatively-reinforced mands can generalize to untrained stimuli (Groskreutz et al., 2014; Yi et al., 2006). Moreover, it provides supporting evidence that for the two children with ASD included in this study, the newly acquired mands generalized as a result of mand training without additional programming. However, only one child (i.e., Brenda) engaged in varied responding by alternating between the two mands in the presence of a different social partner. These results suggest that, for some children, teaching more complex skills such as an alternation strategy may require additional programming to promote generalization to social partners not involved in training.
Two aspects of the study merit further discussion. The first aspect is related to the length of training necessary for the two children included in this study to acquire the new negatively-reinforced mands and the alternation strategy between them. For example, to reach the acquisition criterion, Isaac needed 524 trials for the first mand, 248 trials for the second mand, and 46 trials for the alternation strategy. Brenda needed 135 trials for the first mand, 416 trials for the second mand, and 613 trials for the alternation strategy. Previous research on mand acquisition for children with ASD and language delays suggests that when teaching a multiple mands or a more complex skill (e.g., two-way discrimination), children need extensive training ranging from hundreds to thousands of trials to acquire the skill (Carr, Binkoff, Kologinsky, & Eddy, 1978; Drasgow, Halle, Ostrosky, & Harbers, 1996). Furthermore, the communicative repertoire of children included in this study may have been a contributing factor to their slow acquisition rate of negatively-reinforced mands. For example, children in previous studies addressing the topic of teaching negatively-reinforced mands used pictures or signs to communicate two- or three-word phrases or communicated vocally using three- or four-word utterances. Children included in our study were nonvocal and had a limited repertoire consisting of approximately three or four signs (e.g., “Cookie,” “More”) serving a positive-reinforcement function.
The second aspect is related to children’s selection bias for participating in studies conducted in applied settings. Specifically, researchers may select for their studies only children for whom the intervention would be effective and, thus, allow them to demonstrate a strong functional relation between the target behavior and the intervention, and exclude children who would negatively influence the effectiveness of an intervention (Durand & Rost, 2005). The results of the current study demonstrate that for Isaac, the mand training to teach the acquisition of the two new mands was more challenging and required a longer time compared with the mand training for Brenda, and thus, not allowing the researchers to demonstrate an immediate effect of the intervention on the child’s level of performance. However, immediacy of effect, is just one of the six features of the single-case research design used to interpret the results of a study and it should be considered collectively with other features such as level, trend, variability, overlap, and consistency across similar phases (Kratochwill et al., 2010). Moreover, predicted delayed effects have to be considered in certain cases including the situation when participants require a large number of trials to acquire complex skills (Carr et al., 1978).
The results of the study should be interpreted with caution in the context of several limitations. One limitation concerns the external validity of the study. Only two children participated in this study, and therefore, the generalizability of the findings to other children with ASD and severe language delays is limited. In fact, these two participants responded somewhat differently to the intervention procedures. Future studies are needed to replicate these procedures with other children with ASD in different settings and with different communicative goals. Second, we did not collect social validity data to evaluate whether staff members perceived the replacement responses as meaningful and efficient for each child. Collecting social validity data may have allowed us to better program for maintenance and generalization of the new skills in each child’s natural setting. Practitioners are likely to abandon interventions perceived as irrelevant, ineffective, or ones that require too much effort and, thus, reduce the probability of response maintenance (Kennedy, 2002; Reid & Parsons, 2002). Third, potential threats to internal validity should be considered within the context of the experimental design when interpreting the findings of this study. For example, the fact that we taught the new mands and the alternation strategy sequentially may have compromised functional control because it did not allow us to replicate the findings for the same behavior, although it did allow us (a) to document acquisition and generalization of the two new mands and their alternation, (b) to evaluate resurgence of existing mands, and (c) to record the sequence with which mands were emitted within the response class. Fourth, our procedures for assessing resurgence were slightly different from the standard procedures implemented in previous studies. Specifically, we assessed resurgence under a 6-s delayed-reinforcement condition rather than placing all mands on extinction in a separate phase. We decided to implement this variation because of ethical concerns related to placing newly acquired mands on extinction. Finally, we did not conduct an experimental analysis to determine the function of problem behavior or existing mands present in the children’s behavioral repertoire prior to implementing the mand training.
In sum, the results of our study suggest that mand training is an effective intervention in replacing existing mands with two alternative, socially appropriate, negatively-reinforced mands. Both children acquired the two replacement mands, but only one child engaged in varied patterns of responding by spontaneously alternating between the two mands when the first form did not produce immediate-reinforcement. Resurgence to existing mands was also documented for both children under delayed-reinforcement conditions, and the pattern of responding under delayed-reinforcement conditions suggested the response-class members do not always occur in a predictable sequence. Moreover, both children used the newly acquired mands with different social partners, and one of the two children alternated the replacement mands to avoid nonpreferred food in the presence of a social partner not involved in training.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.