Abstract
We conducted five experiments to evaluate the predictive validity of a free-operant competing stimulus assessment (FOCSA). In Experiment 1, we showed that each participant’s repetitive behavior persisted without social consequences. In Experiment 2, we used the FOCSA to identify high-preference, low-stereotypy (HP-LS) items for 11 participants and high-preference, high-stereotypy (HP-HS) items for nine participants. To validate the results of the FOCSAs (Experiment 3), we used a three-component multiple schedule to evaluate the immediate and subsequent effects of an HP-LS stimulus, an HP-HS stimulus, or both (in separate test sequences) on each participant’s stereotypy. Results of Experiment 3 showed that the FOCSA correctly predicted the immediate effect of the HP-LS stimulus for 10 of 11 participants; however, the FOCSA predictions were less accurate for the HP-HS stimulus. Results of Experiment 4 showed that a differential reinforcement of other behavior procedure in which participants earned access to the HP-LS for omitting vocal stereotypy increased all five participants’ latency to engaging in stereotypy; however, clinically significant omission durations were only achieved for one participant. Experiment 5 showed that differential reinforcement of alternative behavior in which participants earned access to the HP-LS stimulus contingent upon correct responses during discrete-trial training reduced targeted and nontargeted stereotypy and increased correct academic responding for all four participants. The potential utility of the FOCSA is discussed.
Rapp and Vollmer (2005) defined motor stereotypy as a repetitive, invariant noninjurious behavior that persists in the absence of social consequences, and Lanovaz and Sladeczek (2012) expanded that definition to explicitly include response forms that are categorized as vocal stereotypy. In a recent review of the literature, Chebli, Martin, and Lanovaz (2016) reported that 88% of individuals with autism spectrum disorder (ASD) engaged in at least one form of stereotypy. Common treatments for stereotypy include differential reinforcement procedures and noncontingent reinforcement (NCR; Rapp & Lanovaz, 2016). Although these treatments have been shown to reduce stereotypy and other automatically reinforced behavior, there is no one agreed upon method for which to identify reinforcers to use in these treatments (LeBlanc, Patel, & Carr, 2000). In fact, there are several options for identifying stimuli to use in the treatment of stereotypy (e.g., Ahearn, Clark, DeBar, & Florentino, 2005; Pace, Ivancic, Edwards, Iwata, & Page, 1985; Roane, Vollmer, Ringdahl, & Marcus, 1998), but the lack of a consensus regarding the best way to identify these items may decrease the efficiency and consistency of treatment implementation in applied practice.
Although NCR in the form of continuous access to preferred items is a widely cited intervention for decreasing various forms of stereotypy (see Rapp & Lanovaz, 2016, for a recent review), several studies have shown that noncontingent access to preferred stimulation can increase some individuals’ engagement in automatically reinforced behavior (Cook, Rapp, Gomes, Frazer, & Lindblad, 2014; Friman, 2000; Lanovaz, Robertson, Soerono, & Watkins, 2013; Petrongolo, DuBard, & Luiselli, 2015; Rapp, 2004, 2005; Van Camp et al., 2000). For example, Cook et al. (2014) found that music increased body rocking and hand flapping for an adolescent with ASD, Van Camp et al. (2000) demonstrated that certain stimuli (vibrating toys and social interaction) were associated with increased levels of stereotypy for two children with disabilities, and Rapp (2004) found that noncontingent access to music consistently increased object twirling for a young boy with ASD.
Relatedly, Rapp et al. (2013) found that noncontingent music not only decreased targeted vocal stereotypy for several children with ASD but also increased engagement in one or more forms of nontargeted motor stereotypy (e.g., hand flapping, body rocking) for about half of these individuals. Rapp et al. (2013) argued that some preferred items may function as an establishing operation, a subtype of motivating operation (MO; Laraway, Snycerski, Michael, & Poling, 2003; Laraway, Snycerski, Olson, Becker, & Poling, 2014), that simultaneously evokes a specific form of stereotypy and increases the value of the stimulation generated by engagement in that form of stereotypy. Taken together, the aforementioned studies suggest that NCR with preferred items could (a) decrease one or more forms of stereotypy, (b) decrease one form of stereotypy while increasing another, or (c) simply increase one or more forms of stereotypy.
Eikeseth and Grung (2017) reviewed studies wherein NCR or environmental enrichment reportedly increased individuals’ engagement in automatically reinforced behavior. In part, Eikeseth and Grung suggested that results of studies that utilized contingent removal might point to operant processes to account for this environment–behavior relation. Eikeseth and Grung also suggested that an explanation involving MOs has not yet been empirically tested. Although Eikeseth and Grung summarized possible explanations for this phenomenon, they neither recommended a methodology for studying such relations nor suggested a possible approach for incorporating such effects into an intervention.
As a step toward the latter, Frewing, Rapp, and Pastrana (2015) described a free-operant competing stimulus assessment (FOCSA) that involved the use of conditional durations to identify items that either increased or decreased a participant’s engagement in stereotypy. Based on the time participants spent manipulating each item and their engagement in stereotypy while manipulating each item, Frewing et al. categorized items as either high or low preference and either high or low stereotypy for each participant. Although Frewing et al. conducted separate treatment validations using standard multielement designs with each type of stimulus, our focus is on results for high-preference stimuli. To that end, the authors found that the FOCSA categorized a high-preference, low-stereotypy (HP-LS) stimulus for all four participants and, subsequently, demonstrated that the HP-LS decreased each participant’s stereotypy. Likewise, the FOCSA categorized a high-preference, high-stereotypy (HP-HS) stimulus for all four participants; however, the HP-HS increased stereotypy for only one participant. It is possible that the treatment validation sessions were too brief or that the methodology was not sufficiently sensitive for detecting small behavior increases. To address the latter possibility, we used a method that was developed to detect small changes in stereotypy. Although the relative merits of identifying and then providing preferred stimuli to decrease stereotypy are already evident in the literature (e.g., see Rapp & Lanovaz, 2016), the utility of identifying stimuli that temporarily increase stereotypy has yet to be explored.
The primary purpose of this series of experiments was to further evaluate the FOCSA as an assessment tool for reliably predicting the effects of preferred stimuli for either increasing or decreasing stereotypy. Because previous research has shown that decreases in one form of stereotypy may increase or decrease other forms of stereotypy (e.g., Rapp, 2005; Rapp et al., 2013; Rapp, Vollmer, St Peter, Dozier, & Cotnoir, 2004), the secondary purpose was to track changes in one or more nontargeted forms of stereotypy for each participant across experiments.
Thirteen children with ASD or a related disability participated in three to five experiments. In Experiment 1, we verified that each participant’s repetitive behavior persisted in the absence of social consequences. In Experiment 2, we conducted an FOCSA to identify HP-LS and HP-HS stimuli for each participant. In Experiment 3, we conducted a brief treatment evaluation to evaluate the predictive validity of the FOCSA. Specifically, the treatment evaluation in Experiment 3 examined both immediate and subsequent effects of the HP-LS, HP-HS, or both for each participant. In Experiment 4, we implemented a differential reinforcement of other behavior (DRO) procedure embedded in discrete-trial teaching (DTT) sessions using the HP-LS stimulus as the reinforcer for omitting the targeted stereotypy. In Experiment 5, we evaluated the efficacy of a differential reinforcement of alternative behavior (DRA) procedure for decreasing stereotypy, in which the participant earned access to the HP-LS stimulus for correct responding on tasks during DTT sessions.
Experiment 1: Verifying a Nonsocial Function
Querim et al. (2013) found that consecutive alone or no-interaction (NI) sessions could be used to verify the persistence of repetitive behavior in the absence of social consequences. This method has been used in several recent studies involving evaluations of different behavioral interventions for stereotypy (Carroll & Kodak, 2014; Enloe & Rapp, 2014; Frewing et al., 2015; Lanovaz, Rapp, Long, Richling, & Carroll, 2014; Rapp, Cook, McHugh, & Mann, 2017). The purpose of this experiment was to verify that each participant’s repetitive behavior persisted without social consequences.
Method
Participants, target behaviors, and settings
Each participant was diagnosed with ASD except for Alex who was diagnosed with a speech delay. Table 1 lists each participant’s age and response definitions for targeted and nontargeted stereotypy. Sessions for Steven and James took place in 3 m × 3 m rooms within their public elementary schools. Sessions for Casey took place in a 3 m × 4 m room within a university clinic. Sessions for Jeremy, Max, Aiden, Bentley, Caden, and Gavin were conducted in the same 2.9 m × 2.3 m therapy rooms within the treatment center, wherein they received behavior-analytic services. Sessions for Mark, Alex, Nick, and Sam were conducted in a 1.5 m × 4.5 m or a 3.6 m × 3.6 m therapy room in a university clinic. The 1.5 m × 4.5 m room contained only a file cabinet and table. The 3.6 m × 3.6 m room was devoid of all materials and it contained a one-way mirror on one wall.
Participants’ Age and Response Definitions for Targeted and Nontargeted Stereotypy.
Procedures
Each participant was exposed to a minimum of three consecutive NI sessions, wherein a therapist was present, no social consequences were provided for participant’s behavior, and no items were available to the participant. Sessions were 10 min in duration for Mark and were 5 min in duration for the other participants. Table 2 (second column) shows the number of sessions that were conducted with each participant. For each participant, therapists conducted three to five sessions daily, 3 to 4 days per with week.
Results for the FOCSA and Means of Targeted and Nontargeted Stereotypy During the No-Interaction Sessions.
Note. Number in parentheses in the second column denotes number of NI sessions in Experiment 1. FOCSA = free-operant competing stimulus assessment; NI = no interaction; HP-LS = high-preference, low-stereotypy; HP-HS = high-preference, high-stereotypy; ND = not determined.
Denotes mean for most probable nontargeted stereotypy.
Denotes highest preference item from the FOCSA.
Denotes less than 30% engagement during FOCSA.
Data collection and interobserver agreement (IOA)
For James, Steven, and Casey, observers collected data on repetitive behavior with continuous duration recording (CDR) from digital recordings of the sessions using laptop computers equipped with a data collection program that recorded the duration of each response. We converted the duration of repetitive behavior into a percentage of time measure by dividing the number of seconds engaged in the behavior by the total number of seconds in the session and multiplying by 100%. For the other 10 participants, observers collected in vivo data using momentary time sampling (MTS) 1 with 10-s intervals. Some additional post hoc scoring was completed from video recordings of the sessions when a secondary observer was not available at the session time. Each MTS observation period was signaled by an auditory and textual stimulus from an iPhone application. Several studies have shown that 10-s MTS detects a wide range of changes in duration events such as stereotypy (e.g., Devine, Rapp, Testa, Henrickson, & Schnerch, 2011; Meany-Daboul, Roscoe, Bourret, & Ahearn, 2007; Rapp, Colby-Dirksen, Michalski, Carroll, & Lindenberg, 2008) and may be relatively easy for practitioners to use (Kolt & Rapp, 2014). We converted data for each response into a percentage of 10-s intervals measure by dividing the number of scored intervals by the total number of intervals in the session and multiplying by 100%.
Board-certified behavior analysts and undergraduate and graduate students with training in applied behavior analysis served as primary and secondary observers. We assessed IOA for at least 29% of sessions for each participant. To calculate IOA scores for data collected using CDR, we compared data collected by the primary observer with data collected by the secondary observer (from recorded sessions) using the block-by-block method within 10-s bins (Mudford, Martin, Hui, & Taylor, 2009). For data collected using MTS, IOA was calculated on an interval-by-interval basis by dividing agreements by agreements plus disagreements, and multiplying by 100% (e.g., Mudford et al., 2009). The mean IOA score for each participant’s targeted and nontargeted stereotypy was 86% or higher.
Results and Discussion
Results for each participant indicated that one or more forms of their repetitive behavior persisted without social consequences across NI sessions. Table 2 (second column) shows the means for each participant’s stereotypy (targeted and nontargeted) from the consecutive NI sessions. In most cases, the response form that we designated as targeted was the highest probability behavior from this assessment. Based on the persistence of one or more response forms across sessions, all 13 participants were eligible for inclusion in Experiment 2.
Experiment 2: FOCSA
Currently, there is no consensus among researchers about how to best identify stimuli to be used in the treatment of stereotypy and other automatically reinforced behavior (Frewing et al., 2015). One option is to conduct a competing stimulus assessment to identify one or more items that compete with stereotypy; however, there is no one agreed upon assessment for which to do so (Lanovaz, Rapp, & Fletcher, 2010). In one of the first studies to use a competing stimulus assessment, Piazza, Fisher, Hanley, Hilker, and Derby (1996) employed a single-stimulus presentation method to identify stimuli associated with different levels of automatically reinforced self-injurious behavior. Piazza et al. (1996) then used items associated with different levels of the target behavior in a DRO preparation. Subsequent studies have used the single-stimulus approach to identify items that competed with automatically reinforced behavior (e.g., Ahearn et al., 2005; Piazza, Adelinis, Hanley, Goh, & Delia, 2000; Piazza et al., 1998); however, a multiple-stimulus approach could be more efficient for practitioners (Frewing et al., 2015).
Frewing et al. (2015) proposed using an FOCSA comprised of three or more 10-min sessions to identify a participant’s stimulus preference and the conditional percentages of stereotypy associated with each stimulus. An FOCSA is an adaptation of the free-operant stimulus preference assessment (Roane et al., 1998), but it might yield more predictive information. To date, only Frewing et al. (2015) have evaluated the predictive validity of the FOCSA, thus more research is needed to validate this tool.
The FOCSA described by Frewing et al. (2015) prescribes calculating conditional measures of target behavior given specific item engagement. By calculating these conditional percentages of stereotypy, researchers can determine both the individual’s relative preference for each item and the level of stereotypy associated with each item. Two of the classifications yielded from the FOCSA are the HP-LS classification and the HP-HS classification. An item identified as HP-LS results in lower percentages of stereotypy while the participant is engaging with the item when compared with the background percentage of stereotypy (Frewing et al., 2015). By contrast, an item identified as HP-HS results in higher percentages of stereotypy while the participant is engaging with the item when compared with the background percentage of stereotypy (Frewing et al., 2015).
To date, no study has attempted to utilize evocative stimuli as a treatment component for reducing stereotypy. Thus, it is not yet known how an HP-HS item may be used in the treatment of stereotypy. A first step toward initiating this line of inquiry is to determine whether an HP-HS can be reliably identified with an assessment tool. The purpose of Experiment 2 was to identify an HP-LS stimulus and an HP-HS stimulus (or a low-preference, high-stereotypy [LP-HS] stimulus if an HP-HS was not identified) for each participant.
Method
Participants, target behavior, setting, and data collection
Each individual who participated in Experiment 1 also participated in this experiment. Likewise, we collected data in the same settings on the same target behaviors using the same response definitions described in Experiment 1. In addition, for James, Steven, and Casey, we defined object manipulation as touching an item with any part of the participant’s body. For James and Steven, we defined music and movie interaction as the presence of auditory stimulation coming from either the CD or DVD player. For Bentley, Caden, Gavin, Max, Aiden, and Jeremy, we defined object manipulation as making contact with any part of his body, excluding mouthing (defined as making contact with an object using his open mouth). For Mark, Alex, Nick, and Sam, we defined object manipulation as contact between an item and any part of the participant’s body for nonauditory items (e.g., plastic figurine) and contact between an item and any part of the participant’s body and/or eye gaze directed toward an item that produced auditory stimulation (e.g., an iPad™). Researchers defined object manipulation with slight variations across participants due to unique participant characteristics and previously observed behavioral patterns.
We assessed IOA for each dependent variable for at least 33% of sessions for each participant using either the block-by-block method within 10-s bins for CDR data or the interval-by-interval method for MTS data (e.g., Mudford et al., 2009). The mean IOA score was 85% or higher for each dependent measure for each participant.
Competing stimulus assessment
We conducted the FOCSA sessions in the manner described by Frewing et al. (2015). Frewing et al. correctly identified an HP-LS item for all four participants with three 10-min FOCSA sessions; however, the same number and duration of sessions correctly identified an HP-HS item for one of four participants. We conducted more than three sessions with some participants in an attempt to identify items that would fit the two stimulus categories. Specifically, therapists conducted four 10-min assessments with James, Caden, Gavin, and Bentley; three 10-min assessments for Steven, Max, Aiden, Alex, and Jeremy; six 10-min assessments with Casey; and one 10-min assessment with Bentley. During the initial assessment, Mark and Sam simultaneously manipulated the iPad™ and at least one other item and Nick engaged in almost exclusive manipulation with the iPad™. Therefore, therapists conducted an additional series of three 10-min sessions with Mark, Sam, and Nick during which access to the iPad™ was not provided.
Assessment sessions for each participant were typically separated by at least 2 hr with a maximum of two assessments per day. During each session, participants were given access to the same four to 10 stimuli. We included stimuli that would provide several forms of sensory stimulation including tactile stimulation (e.g., pinscreen), visual stimulation (e.g., light spinner, iPad™), and auditory stimulation (e.g., music, iPad™). For all participants, each item was placed in an array on the floor, equidistant to other items (approximately 0.3 m apart) and 1 m from the participant. Prior to each session, researchers implemented forced exposure to each item for approximately 10 s. At the start of the session, the experimenter stated “You can play with whatever you want.” To access music or a movie, Steven, James, or Casey had to move into an area designated by tape on the floor. When the participant entered the area, the experimenter started the CD or DVD player. When the participant left the designated area, the experimenter stopped the CD or DVD player. For Bentley, Caden, Alex, and Gavin, access to any item was available by approaching it. If they attempted to leave the general area with the item, the experimenter took the item from them, placed it back in the initial location, and stated “this stays here” (there were not any physical markers to demarcate the boundaries). This was done to minimize simultaneous item engagement, which affected the conditional engagement calculations. Because a prior study demonstrating that the iPad™ (an empirically identified preferred stimulus) decreased Bentley’s vocal stereotypy (Rapp et al., 2017), therapists conducted his sessions without the iPad™.
During each assessment session, observers collected data on object manipulation and stereotypy. In keeping with the manner in which results from a free-operant stimulus preference assessment are typically depicted (e.g., Roane et al., 1998), we combined all the sessions for a given participant, and then calculated conditional and unconditional probabilities. For data collected with CDR, we calculated unconditional percentages and conditional percentages of stereotypy based on the procedures employed by Frewing et al. (2015) and Lanovaz, Rapp, and Ferguson (2013). Specifically, we calculated unconditional percentages for stereotypy by dividing the number of seconds engaged in stereotypy by the total number of seconds and multiplying by 100%. We determined conditional percentages by dividing the number of seconds the participant engaged in stereotypy while manipulating the object by the total number of seconds the participant manipulated the object and then multiplying by 100%. For data collected with 10-s MTS, we calculated the unconditional percentage of stereotypy by dividing the number of intervals engaged in stereotypy by the total number of intervals and then multiplying by 100%. As with CDR, we determined conditional percentages of stereotypy by dividing the number of intervals the participant engaged in stereotypy while manipulating the object by the total number of intervals the participant manipulated the object and multiplying by 100%. Conditional percentages were calculated separately for each item that the participant manipulated during the assessment. If the conditional percentage of stereotypy was higher than the unconditional percentage of stereotypy in both the FOCSA and NI sessions, we hypothesized that the object would increase stereotypy when presented alone. By contrast, if the conditional percentage was lower than the unconditional percentage of the FOCSA and NI sessions, we hypothesized that the item would decrease stereotypy if presented alone.
The purpose of conducting the FOCSA was to identify stimuli that increased stereotypy, decreased stereotypy, or both for each participant. We categorized an item as an HP-LS stimulus based on the item (a) producing lower than average (unconditional) levels of stereotypy during the Competing Stimulus Assessment ([CSA] i.e., engagement in stereotypy was lower while the participant manipulated that object than the overall level of stereotypy throughout the CSA) and (b) being ranked in the top 50% of items presented during the CSA as indicated by percentage of item manipulation. (Frewing et al., 2015, p. 39)
By contrast, we categorized an item as an HP-HS stimulus if the item “(a) produced higher than average (unconditional) levels of stereotypy during the CSA and (b) was ranked in the top 50% of items presented during the CSA” (Frewing et al., 2015, p. 39).
As previously noted, the criteria used by Frewing et al. (2015) to categorize HP-HS stimuli correctly predicted that said item increased stereotypy for only one of four participants. Given the relatively weak predictive validity for this stimulus category, we modified the categorization such that an HP-HS stimulus had to increase conditional engagement in stereotypy relative to both the unconditional level of stereotypy in the FOCSA and the mean level from the NI sessions from Experiment 1. Because most practitioners would likely conduct consecutive NI sessions to determine whether repetitive behavior persisted without social consequences, this comparison would not require extra resources.
Results and Discussion
Figure 1 shows the results of the FOCSA for Alex; these results are representative of the FOCSA outcomes for all participants. During Alex’s FOCSA series, the percentage of 10-s intervals with iPad™ and Octonauts™ engagement (black bars) fell within the top 50% of items provided and, thus, both were labeled as high-preference items. The iPad™ was associated with a lower conditional percentage of stereotypy (white bar) when compared with both unconditional percentages of stereotypy during the FOCSA (light gray bar) and the mean from the NI sessions (dark gray bar). The Octonauts™ was associated with a higher conditional percentage of stereotypy (white bar) when compared with both unconditional percentages of stereotypy (i.e., FOCSA and NI; light and dark gray bars, respectively). Alex’s overall results suggest that the iPad™ was an HP-LS stimulus and that the Octonauts™ was an HP-HS stimulus.
Percentage of 10-s intervals with item engagement (primary y axis), conditional percentage of stereotypy given item engagement, and background percentage of stereotypy (secondary y axis) during the FOCSA and NI sessions for Alex.
Table 2 summarizes the FOCSA results for all 13 participants. The left side of the table provides the mean percentage of time (Steven and James only) or 10-s intervals with targeted and nontargeted stereotypy during the NI sessions; this value is equivalent to the unconditional percentage of stereotypy in NI sessions. The right side of the table lists the HP-LS and HP-HS items that were identified with the FOCSA for each participant. The FOCSA identified an HP-LS stimulus for all 13 participants and an HP-HS stimulus for nine participants. The failure to identify an HP-HS for Max, Jeremy, Casey, and Gavin appeared to be a function of their near exclusive allocation to the HP-LS stimulus. Sam, Nick, and Mark also exhibited the same problem with the HP-LS item (i.e., an iPad™). Subsequently, therapists conducted three additional sessions, wherein the iPad™ was not available and identified an HP-HS stimulus for these three participants. The purpose of Experiment 3 was to validate the categories provided by the FOCSA by evaluating changes in stereotypy when these stimuli were provided separately.
Experiment 3: Immediate and Subsequent Effects of HP-HS and HP-LS Stimuli
Lanovaz et al. (2010) outlined a methodology for studying changes in MOs that may be produced during and after the implementation of behavioral interventions for stereotypy and other automatically reinforced behavior. Briefly, this methodology involves an NI control sequence and one or more test (treatment) sequences that are conducted in an alternating fashion across separate days. Each sequence is comprised of three contiguous 5- or 10-min components. During all three components of the NI sequence, a therapist withholds social consequences. During test sequences, a therapist implements the intervention during only the second component; the first and third components are conducted in the same manner as the components in the NI control sequence. This methodology provides a succinct process by which to evaluate immediate and subsequent effects of treatment on stereotypy (Enloe & Rapp, 2014). One such treatment that has been evaluated using this process is NCR with matched stimulation (Lanovaz, Sladeczek, & Rapp, 2012).
Although preferred stimuli could be provided either contingently or noncontingently to reduce stereotypy, we opted to assess the effects of the HP-LS item and HP-HS item when each was provided continuously and noncontingently because prior studies have shown that such access can produce subsequent changes in stereotypy (e.g., Rapp, 2006, 2007; Rapp et al., 2017; Rapp et al., 2013). If the FOCSA results are validated, we would expect the data path for the second component of the HP-LS sequence to be differentiated below the data path for the NI sequence and the converse for the data path of the HP-HS sequence. That is, the data path for the second component of the HP-HS sequence should be differentiated above the data path for the NI sequence. Furthermore, we hypothesized that some items may produce subsequent decreases in stereotypy once the item is removed. If an item produces a subsequent decrease in stereotypy, then the data path for the third component of that test sequence will be differentiated below the data path of the NI sequence in the third component (Lanovaz et al., 2010).
Method
Participants, settings, and target behavior
Each individual who participated in Experiments 1 and 2 also participated in this experiment. Observers collected data on the same target behaviors in the same settings using the same response definitions described in Experiments 1 and 2. We assessed IOA for each participant’s stereotypy for at least 33% of sessions using the same methods described for Experiments 1 and 2. The mean IOA score for each dependent measure for each participant was 88% or higher.
Experimental design and procedures
We evaluated the effects of preferred stimuli on stereotypy based on the procedures described by Lanovaz et al. (2010). Specifically, we plotted the data paths for first, second, and third components of each sequence in multielement designs. If the data paths for the second components of the NI sequence and a test sequence were differentiated after we conducted three sessions with each sequence, we stopped the analysis (e.g., Rapp et al., 2013). If the NI sequence and a test sequence were not differentiated after three sessions with each sequence, we continued until we completed a total of five sessions with each sequence, and then we discontinued the analysis. Because the purpose of this experiment was to validate the FOCSA results by evaluating the immediate effects of preferred stimuli on stereotypy, we only evaluated the data in third components if the second components of the test and NI sequences were differentiated. For Nick, Mark, Sam, Alex, James, and Steven, each sequence contained three consecutive 10-min components (30 min total per session). For Max, Aiden, Jeremy, Casey, Caden, Gavin, and Bentley, each sequence contained three consecutive 5-min components (15 min total per session). The difference in component duration was due to participant scheduling and availability.
NI sequence
In this sequence, each component was the same. During each component, the participants did not have access to preferred stimuli. An experimenter was present, but there were no programmed consequences for stereotypy. This sequence served as a control and was designed to assess natural fluctuations in levels of stereotypy against which the effects of one or more test sequences (described below) could be assessed. This sequence was conducted with all 13 participants.
Test sequences
In this sequence, the first and third components were identical to the NI sequence. During the second component, participants were given continuous and noncontingent access to either an HP-HS (LP-HS for Nick) stimulus or an HP-LS stimulus. Participants were assigned to the HP-LS sequence, the HP-HS sequence, or both sequences based on his or her program goals and availability. Because most of the participants in this study were receiving clinical services, we asked each participant’s team leader for his or her input regarding the intervention that would ultimately be used for stereotypy. In most cases, team leaders were most interested in using the HP-LS stimulus to compete directly with stereotypy during leisure periods. When time permitted for a given participant, we endeavored to test both the HP-LS stimulus and the HP-HS stimulus, particularly when the HP-HS was most preferred. James and Steven participated in only the HP-HS test sequence. Max, Jeremy, Gavin, and Casey participated in only the HP-LS test sequence. Aiden, Mark, Nick, Sam, Alex, Bentley, and Caden participated in both the HP-HS test and HP-LS test sequences.
Data analysis
We evaluated the data using a between-sequence analysis (Lanovaz et al., 2010). In this analysis, we compared the level of stereotypy in first and second components of the NI sequence with the level of stereotypy in the first and second components of one or more test sequences, respectively. The purpose of the first component is to identify potential confounding variables. In the absence of a confounding variable, we expected the first components to be undifferentiated. The second components of each sequence (hereafter denoted the “immediate” effects) assessed the changes in stereotypy in the presence of either the HP-LS stimulus or the HP-HS stimulus. We considered the results conclusive if the data path from a test sequence was differentiated (after three to five sessions with this sequence) from the data path of the NI sequence in the second components. As previously noted, we evaluated the third components of the sequences only when the second components were differentiated. If the data paths in the second components of the NI and test sequences were not differentiated (this applied only to Steven and Jeremy) after five sessions with each sequence, the analysis was discontinued and we concluded that the test stimulus did not alter stereotypy. Thus, the immediate effects of the HP-LS and HP-HS stimuli were determined by visual analysis of both trends and mean levels of the test and control data paths (Bartlett, Rapp, & Henrickson, 2011).
Results and Discussion
Figure 2 shows the results for targeted stereotypy for James, Steven, and Caden; results for their nontargeted stereotypy are summarized in Table 3. Figure 2 (upper three panels) shows the percentage of time James engaged in hand gestures during the first, second, and third components of the NI and HP-HS test sequences. Results for the first components show that his stereotypy was undifferentiated in the NI (M = 19.7%) and HP-HS (M = 19.2%) sequences. During the second components, his stereotypy was differentiated and higher in the HP-HS sequence (M = 46.2%) than in the NI sequence (M = 19.9%). In the third components, his stereotypy was again undifferentiated in the HP-HS (M = 19.1%) and NI (M = 9.9%) sequences. Results show that the HP-HS stimulus increased James’s engagement in hand gestures.
Percentage of time James engaged in hand gestures across the first, second, and third components of the NI and HP-HS test sequence (upper three panels); percentage of time Steven engaged in vocal stereotypy across the first, second, and third components of the NI and HP-HS test sequences for the between-sequence analysis (center three panels); percentage of time Caden engaged in vocal stereotypy across the first, second, and third components of the NI, HP-HS, and HP-LS test sequences (lower three panels).
Immediate and Subsequent Effects of HP-LS and HP-HS Stimuli on Stereotypy.
Note. HP-LS = high-preference, low-stereotypy; HP-HS = high-preference, high-stereotypy; T = targeted stereotypy; N = nontargeted stereotypy; − denotes a differentiated decrease; + denotes differentiated increase, = denotes undifferentiated; LP-HS = low-preference, high-stereotypy.
Denotes highest preference from the free-operant competing stimulus assessment.
Denotes multiple nontargeted stereotypy.
Denotes use of an LP-HS stimulus.
Figure 2 also shows the percentage of time Steven engaged in vocal stereotypy during the first, second, and third components of the NI and HP-HS test sequences (three center panels). Results for the first components show that his stereotypy was undifferentiated in the NI (M = 60.7%) and HP-HS (M = 65.4%) sequences. During the second components, his stereotypy was again undifferentiated in the HP-HS (M = 70.9%) and NI (M = 64.4%) sequences. In the third components, his stereotypy was, likewise, undifferentiated in the HP-HS (M = 54.6%) and NI (M = 66.1%) sequences. Results were inconclusive for the effect of the HP-HS stimulus. In part, visual analysis of the data paths was complicated by the decreasing trends in Steven’s stereotypy across sessions in all three components in both sequences.
Results for the first components for Caden (Figure 2, lower three panels) show that his stereotypy was undifferentiated in the NI (M = 46.7%), HP-HS (M = 45.3%), and HP-LS (M = 32.3%) sequences. During the second components, his stereotypy became differentiated such that it was lower in the HP-LS sequence (M = 13.3%) and higher in the HP-HS sequence (M = 63.3%) relative to the NI sequence (M = 35.3%). In the third components, his stereotypy was variable and undifferentiated across the HP-HS (M = 55.3%), HP-LS (M = 44.3%), and NI (M = 34.3%) sequences. Results show that the HP-LS stimulus decreased and the HP-HS increased Caden’s immediate engagement in vocal stereotypy.
Figure 3 shows the results for Sam’s targeted stereotypy (three upper panels), nontargeted stereotypy (three center panels), and item engagement (lower panel). Sam’s targeted stereotypy was undifferentiated in the NI (M = 30%), HP-HS (M = 44.4%), and HP-LS (M = 31.7%) during the first components. During the second components, his targeted stereotypy in the HP-LS sequence (M = 7.2%) became differentiated below the NI sequence (M = 43.3%); his stereotypy in the HP-HS sequence (M = 36.1%) was undifferentiated from the NI sequence. In the third components, Sam’s targeted stereotypy in the HP-LS sequence (M = 15.6%) was differentiated below the NI sequence (M = 28.9%) and the HP-HS sequence (M = 33.3%) was undifferentiated from the NI sequence. Results show that the HP-LS stimulus decreased Sam’s immediate and subsequent engagement in targeted stereotypy. Results also show that the HP-HS stimulus did not increase Sam’s targeted stereotypy.
Percentage of 10-s intervals Sam engaged in vocal stereotypy (targeted; top panel) and motor stereotypy (nontargeted; middle panel); percentage of 10-s intervals Sam engaged with HP-LS and HP-HS items during the second components of the HP-LS and HP-HS sequences (bottom panel).
Figure 3 (three center panels) shows that Sam’s nontargeted stereotypy was undifferentiated in the NI (M = 22.8%), HP-HS (M = 46.7%), and HP-LS (M = 18.3%) during the first components. During the second components, his nontargeted stereotypy in the HP-LS sequence (M = 6.1%) was differentiated below the NI sequence (M = 51.7%) and his nontargeted stereotypy in the HP-HS sequence (M = 50.6%) was undifferentiated from the NI sequence. In the third components, Sam’s nontargeted stereotypy was undifferentiated across the HP-LS (M = 27.8%), the NI (M = 13.3%), and the HP-HS (M = 23.9%) sequences. Results show that the HP-LS stimulus decreased Sam’s immediate engagement in nontargeted stereotypy and did not alter his subsequent engagement in stereotypy. As with Sam’s targeted stereotypy, the HP-HS stimulus did not increase his nontargeted stereotypy.
The lower panel of Figure 3 shows the percentage of intervals Sam engaged with the HP-LS item and the HP-HS item during the second components of the HP-LS and HP-HS test sequences, respectively. Results show relatively high percentages of item engagement for both HP-LS and HP-HS stimuli across sessions. Specifically, item engagement was slightly higher during the HP-HS sequence, which is consistent with the FOCSA predictions of Sam’s relative levels of item preference. This outcome is noteworthy because it shows that providing Sam his most preferred item (i.e., the HP-HS) failed to decrease his stereotypy. Instead, the combination of time allocation and conditional engagement in stereotypy correctly predicted that the HP-LS would decrease his stereotypy
Table 3 provides a summary of outcomes for the HP-LS stimuli and the HP-HS stimuli across Components 2 (immediate effects) and 3 (subsequent effects) and the mean percentage of engagement with each stimulus. The HP-LS decreased immediate engagement in the target stereotypy for 10 of 11 participants and did not change the other participant’s target stereotypy. In addition, the HP-LS decreased immediate engagement in nontargeted stereotypy for six of 10 participants and had no effect on the remaining participants’ nontargeted stereotypy. The HP-LS decreased subsequent engagement in targeted stereotypy for one participant and had no effects on the other participants’ subsequent engagement in the targeted stereotypy. The HP-LS did not produce subsequent changes (either decreases or increases) for any participant’s nontargeted stereotypy. By contrast, the HP-HS increased immediate engagement in the target stereotypy for five of nine participants and had no effect on the remaining participants’ immediate target stereotypy. The HP-HS increased one of nine participant’s immediate nontargeted stereotypy and had no effect on the remaining participants’ nontargeted stereotypy. Notably, the HP-HS did not alter any participant’s subsequent engagement in either targeted or nontargeted stereotypy. Across participants, all stimuli identified by the FOCSA as high preference were associated with high levels item engagement during this experiment. Likewise, the only low-preference stimulus that was evaluated (for Nick) produced low levels of item engagement, as predicted by the FOCSA. In addition, for eight of nine participants, the highest ranked item by the FOCSA was associated with the highest mean engagement during the test sequences.
Taken together, results of Experiment 3 illustrate two broad strengths of the FOCSA. First, the FOCSA accurately predicted the immediate effects of HP-LS stimuli for nearly every participant. That is, the FOCSA consistently identified HP-LS items that decreased immediate engagement in targeted stereotypy for 10 of 11 participants without increasing subsequent stereotypy for any participant. In addition, the same HP-LS items decreased immediate engagement in nontargeted stereotypy for six of 10 participants without increasing subsequent engagement in nontargeted stereotypy for any participant. This outcome is noteworthy because interventions for stereotypy are often associated with unintended increases in other automatically reinforced behavior (e.g., Rapp et al., 2013). Second, the FOCSA accurately predicted that high-preference items would produce the highest level of item engagement with the top-ranked item (based only on time allocation) for eight of nine participants during the second components of the test sequences. Put differently, participants typically manipulated the most preferred item from the FOCSA, whether it was an HP-LS stimulus or an HP-HS stimulus, for the highest percentage of time across sessions. Because the results from Experiment 3 validated the predictions of the FOCSA for HP-LS stimuli, Experiments 4 and 5 evaluated two differential reinforcement procedures for decreasing stereotypy with HP-LS stimuli in an instructional context.
Experiment 4: Treating Stereotypy Using DRO With the HP-LS Stimulus
Cariveau, Kodak, and Campbell (2016) found that participants engaged in more stereotypy during DTT with longer intertrial intervals (ITIs) when compared with DTT with shorter ITIs. One potential solution to address this problem may be to implement a DRO procedure during work sessions, in which a practitioner provides access to a reinforcer contingent upon the omission of the target behavior for a predetermined duration (Vollmer & Iwata, 1992). Rapp et al. (2017) described a trial-based DRO in which an empirically identified, functionally matched stimulus was delivered contingent upon omitting stereotypy for a predetermined duration. Rapp et al. thinned the DRO interval up to 10 min for one participant; however, they did not implement the DRO procedure during DTT for any of their participants.
The purpose of Experiment 4 was to examine the efficacy of a DRO procedure in which the HP-LS stimulus that was validated in Experiment 3 was delivered to the participant contingent upon meeting the DRO criterion. We originally hypothesized that to produce meaningful treatment effects in a DRO preparation, that an HP-HS stimulus should produce immediate increases and subsequent decreases in stereotypy engagement. Given that we did not observe this outcome for any participant in Experiment 2, we did not evaluate the effects of an HP-HS stimulus for any participant in this experiment. In addition, we predicted that HP-LS items that decreased both immediate and subsequent stereotypy (i.e., items that were functionally matched to stereotypy; Rapp, 2007) would lead to greater treatment gains in a DRO preparation than HP-LS items that only decreased immediate stereotypy. In relation to the current study, Experiment 3 showed that Sam’s HP-LS produced immediate and subsequent decreases in his targeted stereotypy. Thus, we predicted that Sam would have the most success with treatment in DRO.
Method
Participants, setting, and response definitions
Mark, Nick, Aiden, Alex, and Sam participated in Experiment 4. All settings and stereotypy response definitions were the same as in Experiment 3.
Data collection and reliability
Researchers collected data on the latency to vocal stereotypy during baseline and DRO trials for all participants. Trials ranged from a few seconds to several minutes in duration (depending on the participant’s engagement in stereotypy). A secondary observer collected data for at least 33% of sessions across participants. Researchers calculated IOA score using a time-window analysis method, where an agreement was scored if the latency recorded by the second observer was within 2 s of the latency recorded by the first observer (Mudford et al., 2009; Rapp et al., 2017). For each session in which latency IOA was collected, researchers divided the number of agreements by the number of agreements plus disagreements and multiplied by 100%. Mean IOA scores for each participant’s targeted stereotypy were 97% or higher.
Procedures and design
Therapists conducted 10-min sessions with a trial-based changing criterion design as described by Rapp et al. (2017). Therapists delivered the HP-LS stimulus contingent upon the participant meeting the DRO criterion. Participants received the HP-LS stimulus for 20 s, 30 s, or 1 min depending on the current DRO criterion. For criteria less than 30 s, the participant received the HP-LS for 20 s; for criteria greater than 30 s but less than 1 min, participants received the HP-LS for 30 s; and for criteria more than 1 min, participants received the HP-LS for 1 min.
NI baseline
Trials in this phase were conducted to a similar manner as the NI conditions that were described in Experiments 1 and 3. Observers collected data on the time between successive instances of each participant’s targeted stereotypy.
DTT baseline
Trials in this phase were the same as in the NI baseline phase except for the inclusion of DTT instruction. Before beginning a trial, the researcher stated, “It’s time to work” and started the timer on the table. The time counted down from 10 min. During the trial, the researcher delivered gross-motor imitation instructions with a prompting strategy that included most to least prompts and a progressive time delay. If a participant emitted a correct response (i.e., prompted at the prescribed prompt level or independent), the researcher delivered enthusiastic behavior-specific praise and either an edible (e.g., half a skittle, half a gummy) or high-quality attention (e.g., tickles, picking the child up and spinning them) if the participant did not accept edibles (Mark). If a participant emitted an incorrect response (i.e., any response other than the target response or no response), the researcher conducted an error-correction procedure that included hand-over-hand prompting. If a participant engaged in targeted or nontargeted stereotypy, the researcher continued with instructions and prompting as prescribed by the DTT protocol and did not provide any differential consequences.
DRO
Trials in this phase were conducted as described in the DTT baseline phase. The first DRO criterion was set at half of the last 10 vocal stereotypy interresponse times (IRTs) in the DTT baseline phase (excluding a 300-s outlier trial for Nick). Before beginning a trial, the researcher stated, “You can earn the [HP-LS item] for working quietly.” The researcher then started the timer, which counted down from the current criterion (Rapp et al., 2017). Therapists delivered instructions as described in the DTT baseline sessions above. If a participant engaged in targeted stereotypy before the DRO criterion had elapsed, the therapist ended the trial, reset the timer, restated the contingency, and immediately began a new trial. If a participant refrained from engaging in targeted stereotypy throughout the DRO criterion, the researcher delivered the HP-LS for the duration described above. No differential consequences were provided for nontargeted stereotypy. Successive increases were systematically varied, in that some were large steps and others were more moderate steps to strengthen the experimental control with the changing criterion designs (Kazdin, 2011). Researchers attempted to thin each participant’s DRO schedule to at least 300 s for an increased likelihood of acceptability in clinical settings (Rapp & Lanovaz, 2016). Initially, criteria to increase the DRO subphase was set at 10 consecutive successful trials for Mark, Nick, and Aiden. Likewise, the criteria to decrease a DRO subphase was set at 10 consecutive failed trials. Using the 10-trial criteria, Aiden did not advance beyond the 20-s subphase. After Mark and Nick failed to progress beyond the 30-s subphase, therapists modified the criteria to four of five successful and unsuccessful trials to either increase or decrease the subphase, respectively. Sam began this experiment after Mark and Nick completed this experiment. Given the problems we observed for Mark and Nick, the therapist used only the modified criteria with Sam.
Results and Discussion
Mark’s results are depicted in Figure 4. Mark typically displayed his targeted stereotypy with short IRTs in the NI baseline and DTT baseline phases. Mark’s mean IRT in the last 10 trials of the DTT baseline phase was 6 s; thus, his first criterion was set at 3 s. Mark met the first criterion in 10 trials. Mark exhibited some success followed by difficulty moving past 28 s. At this point, researchers modified criterion to increase and decrease subphases as outlined above. After therapists implemented the modified criterion (lower panel, right side), Mark completed the 28-s subphase in 12 trials. Again, Mark had some initial success with the new criterion but later was unable to move past 28 s.
Latency to targeted vocal stereotypy in seconds across Trials 1 through 146 (upper panel), Trials 147 through 403 (middle panel), and Trials 404 through 630 (lower panel) during NI BL, DTT BL, and DRO phases for Mark.
At this point, Mark had participated in a total of 627 trials of the DRO procedure for 700 total minutes across 24 days. The clinical acceptability of the procedure seemed questionable as a criterion of only 24 s had been obtained across this time. That is, the DRO 24-s procedure allowed therapists to deliver approximately two discrete-trial instructions before providing access to the HP-LS stimulus. The results for Nick, Aiden, and Alex were similar to those for Mark. Specifically, none was able to progress beyond DRO 48 s.
Sam’s results are shown in Figure 5. Sam participated in only a DTT baseline phase; his mean DTT baseline IRT was 16 s. Because the results of Experiment 3 showed that the HP-LS exerted abative effects for subsequent engagement in stereotypy, therapists set Sam’s first criterion slightly above his mean baseline IRT at 20 s. Sam met this criterion in 11 trials and the next criterion was set at 30 s. Sam initially experienced some success in increasing his latency to vocal stereotypy but this was followed by required decreases in criteria.
Latency to targeted vocal stereotypy in seconds across Trials 1 through 98 (upper panel), and Trials 99 through 177 (lower panel) during DTT BL and DRO phases including the 10-s HFED for Sam.
After three additional decreases and proportionally smaller increases in DRO subphases, therapists noted that Sam was engaging in vocal stereotypy when he ingested the edible he had earned for correct responding during DTT. The vocal stereotypy was low magnitude and sounded like a continued emission of the sound “mmm” with a closed mouth. As this sound appeared to be evoked by consuming the edible and it did not interfere with instruction, therapists placed a 10-s hold on the contingency for stereotypy after edible delivery. In other words, if Sam engaged in the “mmm” behavior, therapists did not reset the DRO interval if said behavior occurred within 10 s of Sam receiving an edible.
After therapists implemented this modification for Sam, they set the criterion at 25 s. Sam met this criterion in five trials and the next criterion was set at 40 s, which Sam met in four trials. Following several additional successful increases, Sam met a 300-s criterion in seven trials. Following the modification, Sam met a clinically acceptable DRO duration (5 min) in a total of 29 trials provided over 80 min total across 3 days. After Sam completed the 300-s subphase, therapists returned to the DTT baseline phase to test the durability of Sam’s behavior change. Sam’s highest IRT during this phase was 851 s, but this uniquely long IRT was immediately followed by several short and variable IRTs that were consistent with the original DTT baseline phase. Therapists reintroduced the DRO with the HP-LS at 300 s and Sam mastered this criterion in four trials. In a deliberate attempt to test the upper range of the DRO procedure, therapist continued to increase the criterion as Sam was successful. Sam exhibited success at 460 s, which was then followed by shorter latencies to vocal stereotypy. At this point, Sam had participated in 177 trials across 48 sessions (480 min) and 17 days. As a whole, the results suggested that the upper range of the DRO interval for Sam was approximately 300 s, which was the initially stated treatment goal.
The results of the Experiment 3 correctly predicted that Sam would be the most successful with the DRO procedure, in that he was the only participant who showed immediate and subsequent decreases in stereotypy after NCR with the HP-LS stimulus. Across the five participants, the DRO procedures could only be thinned to a practical variation for Sam. As an alternative to DRO, Experiment 5 evaluated the extent to which the HP-LS could be used in a DRA procedure to directly increase correct pre-academic and academic responding and indirectly decrease stereotypy.
Experiment 5: Treating Stereotypy Using DRA With the HP-LS Stimulus
Caregivers and clinicians may prefer to utilize a DRA treatment for problem behavior as it involves the selection and teaching of an alternative, socially appropriate behavior. In a review of the literature, Lanovaz, Rapp, and Ferguson (2013) identified five studies that utilized some form of differential reinforcement to decrease stereotypic behavior and increase alternative socially appropriate behaviors. Lanovaz et al. noted that only one study in their review had targeted correct responding on academic tasks, and that study had used a DRO procedure. The lack of research on a DRA to decrease stereotypy and increase academic skills is surprising as a frequently noted reason to treat stereotypy is its detrimental impact on skill acquisition (Cunningham & Schreibman, 2008; Koegel, Firestone, Kramme, & Dunlap, 1974). As such, the current study evaluated the extent to which a DRA procedure in which participants earned the HP-LS stimulus contingent upon correct responses on acquisition tasks during DTT sessions decreased targeted and nontargeted stereotypy.
Method
Participants, setting, and response definitions
Alex, Nick, Mark, and Sam participated in this study. Alex participated in Experiment 5 before Experiment 4. Nick, Mark, and Sam participated in Experiment 4 first, and did so a few months before Alex entered the study. Results of Experiment 3 suggested that participants who exhibited an immediate and subsequent decrease from the HP-LS stimulus would have the most success in a DRO preparation. As such, we hypothesized that Alex would not reach clinically acceptable durations in a DRO preparation given his data pattern from Experiment 3. Thus, we chose to move Alex into Experiment 5 before Experiment 4. In part, this decision was due to our limited time with Alex as a participant. As secondary measures, observers collected data for (a) participants’ responding on gross-motor imitation tasks to ensure target acquisition and (b) the frequency of therapist demands, which were defined as any instance in which the therapist delivered a directive to the participant.
Data collection and reliability
Researchers collected data on both targeted and nontargeted stereotypy using 10-s MTS as described in Experiment 1. Researchers conducted secondary observations for stereotypy for at least 50% of sessions across participants. Mean IOA scores for targeted and nontargeted stereotypy were always above 90%. Secondary observations were conducted for the rate of demands for at least 29% of sessions across participants. Researchers calculated IOA scores for rate of demands using the block-by-block method within 10-s bins (Mudford et al., 2009). Mean IOA scores for rate of demands were 96% or higher.
Procedures and design
Researchers used a four-tiered nonconcurrent multiple baseline across participants design with an embedded ABACACD reversal in the first tier and an embedded ABAB reversal in the third tier to evaluate the effects of DRA with the HP-LS item. Therapists implemented 3-min sessions throughout Experiment 5 for Mark, Nick, and Sam. We chose to conduct 3-min sessions because this duration fit the natural stream of instruction for these participants. Alex participated in 5-min sessions. This experiment served as a treatment evaluation from which a broader intervention plan could be developed (Rapp et al., 2013). Because the focus of this experiment was to further validate the results of the FOCSA, we depict results only for each participant’s targeted and nontargeted stereotypy.
Baseline
Sessions in this phase were identical to the NI conditions described in prior experiments.
DRA with edibles condition (Alex only)
Prior to beginning this phase, therapists conducted a paired-choice stimulus preference assessment with eight edible items (Fisher et al., 1992). Prior to beginning sessions each day, therapists conducted a brief multiple-stimulus preference assessment (Carr, Nicolson, & Higbee, 2000) with the top three items from the initial paired-choice assessment; the first-selected item was used for sessions that day. During this phase, researchers administered gross-motor imitation teaching in the form of DTT (prompting and error-correction procedures were identical to those used in Experiment 4). Contingent upon correct responses for acquisition targets, the therapist delivered behavior-specific praise and one small edible (e.g., half of a skittle or half of an M&M). The effects of DRA with edibles were evaluated as a comparison with DRA with the HP-LS. The time required for Alex to accept and ingest each edible item was included in the total session time.
DRA with HP-LS
During this phase, researchers administered gross-motor imitation teaching in the form of DTT. Contingent upon correct responses for acquisition targets, the therapist delivered behavior-specific praise and 30 s of access to the HP-LS item (the iPad™ for all participants). For each participant, the time spent with access to the HP-LS was included in the total session time.
DRA with praise (Alex only)
In this phase, therapists administered gross-motor imitation teaching in the form of DTT. Contingent upon correct responses, the therapist delivered behavior-specific praise only. The effects of DRA with praise on stereotypy were compared with the effects of DRA with the HP-LS on stereotypy.
Results and Discussion
Figure 6 shows the results for each participant’s targeted stereotypy. Across three baseline sessions, Alex (top panel) engaged in variable and moderate percentages of vocal stereotypy (M = 33%). The first phase for Alex was DRA with edibles. During this phase, Alex displayed low percentages of vocal stereotypy (M = 7%). Therapists then reversed back to a baseline phase, in which Alex engaged in moderate percentages of vocal stereotypy (M = 24%). Therapists then implemented DRA with the HP-LS stimulus. During this phase, Alex engaged in low percentages of vocal stereotypy (M = 6%). Therapists then reversed back to a baseline phase in which Alex engaged in moderate percentages of vocal stereotypy (M = 41%). Therapists then reimplemented the DRA with HP-LS phase and, again, Alex engaged in low percentages of vocal stereotypy (M = 9%). Finally, therapists evaluated DRA with praise, and Alex engaged in moderate, increasing percentages of vocal stereotypy (M = 48%). Across five baseline sessions, Nick (second panel) engaged in moderate to high percentages of vocal stereotypy (M = 64%). When DRA with HP-LS was implemented, Nick engaged in variable, low to moderate percentages of vocal stereotypy (M = 28%). Across seven baseline sessions, Mark (third panel) engaged in moderate percentages of vocal stereotypy (M = 44%). When DRA with HP-LS was introduced, Mark engaged in low percentages of stereotypy (M = 9%). A return to baseline resulted in a quick increase to baseline levels of stereotypy (M = 60%). The reintroduction of DRA with HP-LS again produced a rapid reduction in Mark’s stereotypy (M = 2%). Across nine baseline sessions, Sam (bottom panel) engaged in moderate percentages of vocal stereotypy (M = 51%). When DRA with HP-LS was introduced, Sam’s stereotypy quickly decreased to low levels (M = 3%).
Percentage of 10-s intervals with targeted stereotypy across sessions for Alex (top panel), Nick (second panel), Mark (third panel), and Sam (bottom panel) during BL, DRA with edibles, DRA with HP-LS item, and DRA with praise phases.
Figure 7 shows that each participant’s nontargeted stereotypy decreased in the same manner as his targeted stereotypy during the DRA phases. The mean percentages of correct responses (prompted and independent) across DRA sessions for Alex, Nick, Mark, and Sam were 90%, 87%, 89%, and 99%, respectively. The mean response per minute of therapist’s demands across DRA sessions for Alex, Nick, Mark, and Sam were 1.4, 1.4, 1.3, and 3.0, respectively. For Alex, the rate of demand in the six DRA with edibles sessions was 1.3 rpm and the rate of demand in the nine DRA with HP-LS sessions was 1.4 rpm.
Percentage of 10-s intervals with nontargeted stereotypy across sessions for Alex (top panel), Nick (second panel), Mark (third panel), and Sam (bottom panel) during BL, DRA with edibles, DRA with HP-LS item, and DRA with praise phases.
Results of Experiment 5 show that providing contingent access to the HP-LS directly increased desirable behavior and indirectly decreased both targeted and nontargeted stereotypy for all four participants. These findings suggest that DRA with an HP-LS stimulus is a viable treatment option for individuals who show either only immediate decreases in stereotypy with the HP-LS stimulus or both immediate and subsequent decreases in stereotypy with the HP-LS stimulus (as shown in Experiment 3). Although each participant’s stereotypy decreased during DRA with the HP-LS relative to baseline, Sam’s decrease was the largest. The DRO procedure was also most effective for Sam as indicated by his omission of stereotypy for clinically acceptable durations of at least 5 min. Together, findings from Experiments 4 and 5 indicate that an HP-LS stimulus that produces immediate and subsequent decreases on stereotypy (i.e., one that is clearly functionally matched to stereotypy) might give rise to multiple treatment options. It is noteworthy that the DRA with edibles procedure also decreased Alex’s stereotypy; it is possible DRA with other arbitrary reinforcers could have decreased other participants’ stereotypy as well. Because we did not compare the effects of arbitrary reinforcers with the effects of the HP-LS stimuli for each participant, it is not clear whether the effects of the HP-LS are superior to the effects produced with other consequent events.
General Discussion
Experiment 1 showed that each participant’s repetitive behavior persisted in the absence of social consequences and, thus, met the functional definition of stereotypy (Lanovaz & Sladeczek, 2012; Rapp & Vollmer, 2005). Experiment 2 showed that the FOCSA identified an HP-LS item, an HP-HS item, or both for each participant. However, to identify the HP-HS stimulus, a response-restriction assessment was required for about half of the participants. Experiment 3 showed that results of the FOCSAs correctly predicted the immediate effects of the HP-LS and HP-HS for 10 of 11 participants and five of nine participants, respectively. Results from Experiment 3 also showed that the HP-LS decreased nontargeted stereotypy for six of 10 participants. In addition, Experiment 3 showed that noncontingent access to the HP-LS led to a subsequent decrease in targeted stereotypy for one participant (Sam). Experiment 4 showed that a trial-based DRO with the HP-LS stimulus increased latency to vocal stereotypy for five of five participants during DTT sessions relative to baseline; however, a clinically acceptable duration of 5 min was obtained for only one of five participants. Experiment 5 showed that DRA with contingent access to the HP-LS item for correct responding on pre-academic and academic tasks decreased both targeted and nontargeted stereotypy for four of four participants.
Superficial analysis of the results from Experiment 2 might suggest that identifying items that decrease stereotypy may only require a free-operant stimulus preference assessment (Roane et al., 1998). Nevertheless, a more nuanced analysis of the results suggests practitioners should conduct an FOCSA. Notably, the HP-LS item was most preferred for only seven of 13 participants (it was second-most preferred for Caden and Sam) and the HP-HS was most preferred for three of 13 participants (Aiden, Caden, and Sam). Subsequently, Experiment 3 showed that of the 20 high-preference items evaluated (11 HP-LS stimuli and nine HP-HS stimuli), only the HP-LS stimuli (10 of 11) decreased engagement in targeted stereotypy. Similarly, Frewing et al. (2015) found that even though the HP-HS stimulus was most preferred by three of four participants, only the HP-LS, which was most preferred by only one participant, decreased stereotypy for all four participants. Thus, choosing a treatment stimulus solely based on a client preference could result in a practitioner’s selection of an item that does not compete with a client’s stereotypy (Groskreutz, Groskreutz, & Higbee, 2011).
From a practical standpoint, it is important to note that just because noncontingent access to a preferred item (as in NCR) does not decrease stereotypy does not mean that said item cannot be used in another behavioral intervention to decrease stereotypy. For example, after Watkins and Rapp (2014) identified participants’ preferred items using the multiple stimulus without replacement assessment (DeLeon & Iwata, 1996), they found that noncontingent access to those stimuli did not decrease stereotypy during leisure periods for five individuals with ASD; however, they demonstrated that response-contingent removal of the same stimuli did decrease stereotypy to varying degrees for all five participants. In short, identifying a preferred, competing stimulus might not be an essential feature for treating stereotypy displayed by individuals with ASD. Nevertheless, results of this study suggest that using the FOCSA increases the probability that practitioners will identify an effective treatment for stereotypy.
There were some minor procedural differences across participants that warrant discussion. First, observers collected data using CDR for three participants, but used 10-s MTS for the other 10 participants. We suspect that practitioners will prefer to use 10-s MTS given the reduced effort for (a) collecting data on both stereotypy and object manipulation in vivo and (b) calculating conditional and unconditional percentages. Importantly, collecting data with 10-s MTS allowed researchers to code stereotypy and object manipulation at the same time. Second, in Experiment 3, therapists used 10-min components (30-min sequences) with four participants, but 5-min components (15-min sequences) with the other nine participants. The difference in component duration did not appear to affect our ability to validate the predictions made by the FOCSA; however, Rapp (2007) suggested that briefer components could decrease the probability of detecting subsequent changes in stereotypy.
Relatedly, one potential limitation to the FOCSA model is that it requires the participant to manipulate two or more items across the assessment sessions. Without distributed response allocation, the analysis is similar to a brief free-operant stimulus preference assessment for which the limitations of exclusive allocation to a single, highly preferred item have been previously noted (e.g., Rapp, Rojas, Colby-Dirksen, Swanson, & Marvin, 2010; Roane et al., 1998). We addressed this problem by (a) conducting additional sessions with all the items and (b) removing the highly preferred item and conducting additional sessions. Although participants who were exposed to these modifications manipulated more objects, results from Experiment 2 suggest that (b) was the more efficient modification.
One limitation of Experiment 5 is that we did not pause the session timer when the participant had access to the HP-LS. As such, it is possible that decreased stereotypy in the DRA with HP-LS phase was due to decreases in stereotypy when the participant was manipulating the HP-LS. Despite the potential limitation, we chose to collect data in this manner to capture the full effect of the treatment throughout the academic work session. Cook, Rapp, Burji, McHugh, and Nuta (2017) showed that access to an iPad™ increased multiple forms of stereotypy for a child with ASD. By keeping the HP-LS access time in the session, we ensured that stereotypy (both targeted and nontargeted) did not increase when the participant had access to the iPad™.
Results from Experiments 2 and 3 suggest that the best way to identify an HP-HS stimulus is not yet clear. The FOCSA accurately predicted the effects of the HP-LS stimulus for 10 of 11 participants, but accurately predicted the effects of the HP-HS stimulus for only five of nine participants. Frewing et al. (2015) recommended comparing the conditional percentages of stereotypy with the unconditional percentage of stereotypy in the FOCSA sessions. However, results of the current study showed that each participant’s unconditional percentages of stereotypy in the NI sessions (i.e., the mean percentage of stereotypy throughout the initial NI sessions) were always higher than those of the FOCSA sessions. As such, comparing the conditional percentage of stereotypy with the background percentage of stereotypy in the NI sessions may yield different conclusions. Future research should determine the most reliable comparisons for identifying high-stereotypy items.
We foresee at least two potential treatment options with HP-HS stimuli. First, noncontingent provision of HP-HS stimuli may be a viable alternative treatment for individuals for whom HP-LS stimuli cannot be identified. For example, when the HP-HS stimulus is used during academic instruction, it may be possible to decrease a child’s motivation to engage in stereotypy during instructional time by increasing their engagement in stereotypy with an HP-HS stimulus during signaled break periods. Second, several recent studies (e.g., Cook et al., 2014; O’Connor, Prieto, Hoffman, DeQuinzio, & Taylor, 2011) have evaluated procedures for producing stimulus control of stereotypy. Because the HP-HS stimulus, by definition, evokes stereotypy, including it in the condition wherein stereotypy is permitted, or otherwise encouraged, may enhance the effectiveness of those procedures. Studies such as those suggested here will help determine (a) how much of an increase in stereotypy must be produced with an HP-HS stimulus to contribute to an overall treatment effect and (b) the situations and participant characteristics for which HP-HS may be therapeutically indicated.
Footnotes
Acknowledgements
We thank Katie McHugh, Carla Burji, Raluca Nuta, Katelyn Prince, Carissa Balagot, Kayla Crouchman, Claire Jenkins, Sidrah Karim, Chelsea Watters-Wybrow, Jodi Coon, and Anna Kate Edgemon for their assistance with data collection.
Authors’ Note
Portions of this study were conducted in partial fulfillment of a master’s degree in behavior analysis by Kristen M. Brogan and Lisa A. Sennott.
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.
Supplementary Material
Supplementary material is available for this article online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
