Abstract
The authors evaluated the effects of matched and unmatched stimuli on immediate and subsequent engagement in targeted vocal stereotypy (Experiment 1) and untargeted motor stereotypy (Experiment 2). Results of Experiment 1 showed that (a) matched stimulation decreased immediate engagement in vocal stereotypy for 8 of 11 participants and increased subsequent engagement in vocal stereotypy for only 1 of the 8 participants and (b) unmatched stimulation decreased immediate engagement in vocal stereotypy for only 1 of 10 participants and did not increase subsequent engagement in vocal stereotypy for this participant. Results of Experiment 2 showed that for 8 of 14 participants, untargeted stereotypy increased when the matched or unmatched stimulus was present, after it was removed, or both. The authors briefly discuss the potential clinical implications of using matched stimulation to decrease vocal stereotypy and limitations of the findings.
Keywords
A recent review by Lanovaz and Sladeczek (2012) concluded that a number of antecedent and consequent interventions were effective for decreasing vocal stereotypy displayed by children and adolescents with autism spectrum disorder. Among the most empirically supported antecedent interventions for vocal stereotypy is noncontingent reinforcement (NCR) with matched stimulation. Piazza, Adelinis, Hanley, Goh, and Delia (2000) distinguished between matched and unmatched preferred stimulation based on the overt structure of the problem behavior: the former contains stimulation that is putatively “matched” or similar to stimulation produced by the problem behavior, whereas the latter does not. To date, only a few studies have evaluated the effects of unmatched preferred stimulation on vocal stereotypy (Lanovaz & Sladeczek, 2012). In addition, few studies have examined the (a) subsequent effects of interventions on vocal stereotypy or (b) collateral effects (i.e., changes in untargeted problem or appropriate behavior) of interventions designed to treat vocal stereotypy (Lanovaz & Sladeczek, 2012).
Lanovaz, Rapp, and Fletcher (2010) outlined the use of a three-component multiple-schedule procedure for assessing automatically reinforced behavior immediately before an intervention is implemented (first component), during implementation of an intervention (second component), and after an intervention is removed (third component). Results of the assessment may assist practitioners in determining whether the intervention decreases immediate engagement in vocal stereotypy without producing a subsequent increase in vocal stereotypy. The methodology involves alternating between a baseline sequence, which comprises three consecutive 10-min components containing no social consequences, and a test sequence, which is identical to the baseline sequence except that the second component contains a behavioral intervention. Interventions that decrease immediate engagement in vocal stereotypy without increasing subsequent engagement in vocal stereotypy may be useful in applied settings, particularly if vocal stereotypy remains low for a period of time after the intervention is discontinued (e.g., Lanovaz, Fletcher, & Rapp, 2009; Rapp, 2007). By contrast, interventions that decrease immediate engagement in vocal stereotypy but also increase subsequent engagement in vocal stereotypy may have very limited applied utility (Lanovaz et al., 2009).
The following experiments evaluated the extent to which (a) matched or unmatched items decreased immediate engagement in vocal stereotypy, (b) decreases in vocal stereotypy that are produced with matched or unmatched stimuli were followed by subsequent increases in vocal stereotypy, and (c) decreases in targeted vocal stereotypy coincided with increases in untargeted forms of motor stereotypy. The procedures described in this study were intended to serve as pretreatment evaluation that potentially predicted what effects matched or unmatched items would exert on vocal and motor stereotypy when implemented on a broader basis in the participant’s environment. If a matched stimulus decreased an individual’s immediate engagement in vocal stereotypy without increasing either his or her subsequent engagement in vocal stereotypy or his or her immediate or subsequent engagement in an untargeted stereotypy, the intervention could presumably be extended to the individual’s typical environment without untoward side effects. By contrast, if a matched stimulus decreased an individual’s vocal stereotypy, but increased his or her engagement in either subsequent vocal stereotypy or an untargeted (collateral) stereotypy, an alternative intervention for his or her vocal stereotypy would be considered.
General Method
Participants and Response Definitions
Twenty-two children who were diagnosed with autism and who displayed vocal stereotypy participated in one or both of the experiments. One participant was also diagnosed with Down’s syndrome. Participants were 6 to 17 years of age. For each participant, we defined vocal stereotypy as the emission of acontextual audible sounds or words for 2 s or longer. We defined object manipulation as the participant touching the item with any part of his or her hands or feet. For each target event, we utilized a 2-s offset criterion. For example, observers scored engagement with a given item at the very moment it was first touched but did not stop scoring engagement with that item until contact with the item was terminated for 2 s. Sessions took place in a specific room in each participant’s school. The room was equipped with a table, two chairs, and a video camera.
Data Collection and Interobserver Agreement (IOA)
We videotaped and later scored each session for the duration of vocal stereotypy and object manipulation (during preference assessments only) using laptop computers. We converted duration events to a percentage of time measure by dividing the number of seconds engaged in the target event by the total number of seconds in a session and multiplying by 100%. For each participant, a secondary observer scored 20% or more of the sessions. We calculated IOA scores using the block-by-block (with 10-s bins) method (Mudford, Taylor, & Martin, 2009). For Experiment 1, the mean IOA scores across participants ranged from 74% to 95% for vocal stereotypy and 88% to 99% for object manipulation (all items combined). For Experiment 2, the mean IOA scores across participants for motor stereotypy ranged from 90% to 98% (IOA data for individual participants available from first author).
Design and Procedures
We evaluated each participant’s preference for leisure items during three 10-min free-operant stimulus preference assessments (FOSPAs) that we conducted on three separate days (Roane, Vollmer, Ringdahl, & Marcus, 1998). We opted to use this type of preference assessment because two prior studies (Lanovaz et al., 2009; Rapp, 2007) had used stimuli identified with this method to decrease vocal stereotypy. We provided each participant with continuous access to an array of 8 to 11 items; we used the same items in each session for a given participant. We included items in the array based on teacher and parent reports of the participant’s relative preference for various items. In addition, the items were intended to provide a broad range of visual, auditory, or tactile stimulation. Some items provided multiple types of stimulation (e.g., the DVD player generated auditory and visual stimulation). The stimulus array that we used with each participant contained 2 or 3 matched items (i.e., each primarily produced auditory stimulation).
After obtaining informed consent, we assigned individuals (with minor exceptions) to participate in either the unmatched (the most preferred only) sequence or the matched sequence in alternating waves of two or three participants. For example, we assigned the first three participants to the matched sequence, the next three to the unmatched sequence, the following three to the matched sequence, and so on. We used this approach to assign participants to the conditions because (a) we could only accommodate sessions for up to three participants in a given day and (b) we did not know a priori how many individuals would participate in the experiments. For each participant, the unmatched stimulus was also the most preferred item, which we defined as the item to which the participant allocated the most time across all three assessment sessions combined. Initially, our plan was to provide the most preferred item in this condition, regardless of whether it was or was not matched to the overt produced of vocal stereotypy. However, because each participant’s most preferred item was not matched to the product of his or her vocal stereotypy, we opted to simply refer to the item as an “unmatched” stimulus. Examples of unmatched stimuli included a vibrating massager, a blinking disco ball, a puzzle, a Rubik’s cube®, markers and paper, an action figure, a Slinky®, and a Teddy bear. By comparison, we defined the matched stimulus for each participant as an item that (a) matched the overt stimulation produced by vocal stereotypy (i.e., it produced auditory stimulation when manipulated appropriately) and (b) was in the top half of the hierarchy in terms of total time allocation across the three preference assessment sessions. Examples of matched items included music (from either a radio or a CD player), a remote-controlled car with sirens, toy laptop, a talking robot toy, and a drum. We provided only one stimulus to the participant during the unmatched and the matched conditions. For the preference assessment, we defined manipulation of a radio, a CD player, or a DVD player as the participant’s buttocks making contact with the seat of a designated chair. Prior to initiating the preference assessment, we trained participants to sit in this chair to turn “on” the respective item; leaving this chair turned the item “off” and sitting in other chairs did not produce stimulation from the item.
Experimental Design and Data Analysis
Using the methodology described by Lanovaz et al. (2010), we evaluated each participant’s vocal stereotypy during a no-interaction (NI) sequence and during one or two preferred stimulus (PS) sequences. We exposed each participant to the NI sequence, and we exposed most participants to either a PS sequence with a matched stimulus or a PS sequence with an unmatched stimulus. Each sequence comprised three consecutive 10-min components. We conducted only one sequence per day. We used a combination of either a two-component or a three-component multiple schedule and reversal design to evaluate the immediate effects of the matched stimulus or unmatched stimulus, which was provided during the PS sequence, on vocal stereotypy. We exposed three individuals to both a PS sequence with a matched stimulus and a PS sequence with an unmatched stimulus (two received the matched stimulus first). We conducted 3 to 10 sessions with each sequence for each participant. For most participants, the number of sessions with each sequence varied as a function of the stability of vocal stereotypy within the second components of the NI and PS sequences. For four participants, the stability in the first component determined the number of sessions conducted with each sequence because their behavior was evaluated with a two-component multiple schedule.
For participants with whom we used a three-component multiple schedule, we first conducted a between-sequence visual analysis of the data paths from the first and second components. If the data paths for the first components (preintervention) were undifferentiated and the data paths for the second components of the two sequences were differentiated, both of which were determined by a consensus decision of the first author and a junior author, we also inspected the data paths for the third components of the two sequences. We defined a stable data path as one that had less than 20% variability between three or more successive data points without a decrease or increase trending. Based on definitions provided by Barlow, Nock, and Hersen (2009) and Bartlett, Rapp, and Henrickson (2011), we defined differentiation as when (a) the two data paths were stable with at least 10% separation, (b) the two data paths were trending in opposite directions, or (c) one data path was stable and the other data path was trending away from it or was consistently variable but still separated from the other. If the data paths for the third components were differentiated, we stopped the analysis. If the data paths for the third components were not differentiated, we then conducted a within-sequence analysis wherein we separately compared the data for the first and third components for the NI sequence and PS sequence, respectively (Lanovaz et al., 2010). For participants with whom we used a two-component schedule, we first conducted a between-sequence visual analysis of the data paths from the first components of the two sequences. If the data paths from the first components were differentiated, we also conducted a visual analysis of the data paths for the second components. We could not conduct a within-sequence analysis for individuals with whom we used a two-component multiple schedule.
Experiment 1: Immediate and Subsequent Effects of Matched and Unmatched Stimuli on Targeted Vocal Stereotypy
The purpose of this experiment was to extend the results from prior studies (e.g., Ahearn, Clark, DeBar, & Florentino, 2005; Lanovaz et al., 2009; Rapp, 2007) by evaluating the extent to which matched or unmatched (e.g., Piazza et al., 2000) items decreased each participant’s immediate and subsequent engagement in vocal stereotypy.
Method
Participants and Procedures
Twenty-one individuals participated in this experiment. We provided 11 participants with a matched stimulus and 10 participants with an unmatched stimulus. We exposed each participant to two sequences; each sequence comprised either two consecutive 10-min components or three consecutive 10-min components. We typically exposed participants to each sequence on an alternating basis; however, there were several occasions (e.g., when toys were not available or a prior session was not completely processed) where we conducted the same sequence on consecutive days.
NI sequence
During this sequence, a trainer was present in the room with the participant, but she did not provide social consequences for any of the participant’s behavior. Contingent on requests to use the bathroom, the trainer paused the session, escorted the participant to the bathroom, and immediately resumed the session upon return to the room. The NI components were signaled by the absence of toys or other material in the room. The purpose of this sequence was to (a) demonstrate the persistence of vocal stereotypy in the absence of social consequences (Iwata & Dozier, 2008) and (b) serve as a baseline with which the effects of the matched or unmatched stimulus could be compared.
PS sequence with matched or unmatched stimulus
When we used a three-component sequence, the first and third components were identical to the components in the NI sequence, and the matched or unmatched stimulus was provided during the second component. When we used a two-component sequence, the matched or unmatched stimulus was provided during the first component, and the second component was a NI condition. The purpose of the PS sequence was to determine whether the matched or unmatched stimulus (a) decreased vocal stereotypy when the item was continuously available and (b) increased vocal stereotypy after the stimulus was removed.
Data Depiction and Analysis
We used the steps for a between-sequence and within-sequence analysis described by Lanovaz et al. (2010). Results of the within-sequence analyses are graphically presented in the same manner as described in a recent study by Lanovaz, Sladeczek, and Rapp (2011) to compare levels of vocal stereotypy in the first and third components of each sequence.
Results and Discussion
Figures 1 and 2 contain representative results (other data sets are available from the first author on request). For each participant, data paths for the first component were undifferentiated and, therefore, were not depicted. Figure 1 shows the percentage of time Alan (left column) and Barry (right column) engaged in vocal stereotypy during the second (first panel) and third (second panel) components of the NI sequence and PS sequence with a matched stimulus. Results from the second components show that Alan’s vocal stereotypy was lower during the PS sequence than during the NI sequence; however, the data paths for the third components were undifferentiated. Results of the within-sequence analysis show that Alan’s vocal stereotypy was higher in the third component of the NI sequence than in the first component of the NI sequence for only one of five sessions (third panel); a similar pattern was produced in the PS sequence (fourth panel). Together, the results from the between- and within-sequence analyses suggest that the matched stimulus produced an immediate reduction in Alan’s vocal stereotypy and did not increase his subsequent engagement in vocal stereotypy. Barry’s vocal stereotypy was lower during the second component of the PS sequence than during the second component of the NI sequence; however, the NI and PS data paths were undifferentiated during the third components of the two sequences. Results of the within-sequence analysis show that Barry’s vocal stereotypy was higher in the third component of the NI sequence than in the first component of the NI sequence for four of five sessions (third panel). By comparison, his vocal stereotypy was higher in the third component of the PS sequence than in the first component of the PS sequence for only one of five sessions in the PS sequence. As with Alan, the results from the between- and within-sequence analyses show that the matched stimulus decreased immediate engagement in vocal stereotypy for Barry and did not increase his subsequent engagement in vocal stereotypy.
The percentage of time Alan (left column, upper two panels) and Barry (right column, upper two panels) engaged in vocal stereotypy during the NI sequence and PS sequence with a matched stimulus across sessions during the second (first panel) and third (second panel) components, and the percentage of time Alan (left column, lower two panels) and Barry (right column, lower two panels) engaged in vocal stereotypy during the first and third components of the NI sequence and the PS sequence with a matched stimulus.
The percentage of time Hal (left column, upper two panels), Devon, (right column, upper two panels) and Marty (right column, lower two panels) engaged in vocal stereotypy during the NI sequence and PS sequence with a matched (Hal) or unmatched (Devon and Marty) stimulus across sessions during the second and third components, and the percentage of time Hal (left column, lower two panels) engaged in vocal stereotypy during the first and third components of the NI sequence and the PS sequence with a matched stimulus.
Figure 2 shows the percentage of time Hal (matched stimulus; left column), Devon (unmatched stimulus; right column, upper two panels), and Marty (unmatched stimulus; right column, lower two panels) engaged in vocal stereotypy during the second and third components of the NI sequence and PS sequence. Results for the second components (first panel) show that Hal’s vocal stereotypy was lower during the PS sequence than during NI sequence; however, his vocal stereotypy was high and undifferentiated during the third components (second panel) of the two sequences. Results of the within-sequence analysis show that Hal’s vocal stereotypy was higher in the third component than in the first component for four of eight sessions with the NI sequence (third panel) and five of eight sessions with the PS sequence (fourth panel). Results from the between- and within-sequence analyses show that a matched stimulus produced an immediate decrease in Hal’s vocal stereotypy and did not produce an increase in his subsequent engagement in vocal stereotypy. Results for Devon’s vocal stereotypy show that the data paths for the second (first panel) and third (second panel) components of the NI and PS sequences were undifferentiated. As such, we did not conduct a within-sequence analysis of the subsequent effects. Thus, the results indicate that the unmatched stimulus did not decrease Devon’s immediate engagement in vocal stereotypy. Marty’s vocal stereotypy was initially lower in the second (right column, third panel) and third (right column, fourth panel) components of the PS sequence than in the respective components of the NI sequence. However, because the data path for the second component of the NI sequence was decreasing across Sessions 1 to 8, we extended the analysis until his responding ultimately became stable and undifferentiated in the second component for both sequences. Thus, the results indicate that the unmatched stimulus did not consistently decrease Marty’s immediate engagement in vocal stereotypy. The results for the unmatched stimulus with Devon and Marty are consistent with the majority of the participants.
Results showed for all 21 participants indicated that (a) matched stimuli decreased immediate engagement in vocal stereotypy for 8 of 11 participants and increased subsequent engagement in vocal stereotypy for 1 of 8 participants and (b) an unmatched stimulus decreased immediate engagement in vocal stereotypy for 1 of 10 participants, and did not increase subsequent vocal stereotypy for that participant. As a whole, results from this experiment suggest that practitioners who use NCR to decrease vocal stereotypy should provide preferred items that produce auditory stimulation.
Experiment 2: Immediate and Subsequent Effects of Matched or Unmatched Stimuli on Untargeted Motor Stereotypy
Although there is considerable empirical support for the use of NCR to treat automatically reinforced behavior, a handful of studies have shown that preferred stimulation may actually increase some individuals’ engagement in automatically reinforced behavior (McAdam, DiCesare, Murphy, & Marshall, 2004; Piazza, Fisher, Hanley, Hilker, & Derby, 1996; Rapp, 2004, 2005; Van Camp et al., 2000). The purpose of this experiment was to determine whether access to preferred stimulation (matched or unmatched), which was intended to decrease vocal stereotypy, increased immediate engagement in untargeted stereotypy, subsequent engagement in untargeted stereotypy, or a combination of both. Information about the effects of interventions that are designed to treat vocal stereotypy on engagement in untargeted stereotypy is potentially important for at least two reasons. First, on a practical level, undesirable increases in collateral behavior (e.g., motor stereotypy) may influence a practitioner’s decision to use preferred stimuli, contingently or noncontingently, to decrease vocal stereotypy. Second, on a conceptual level, the extent to which decreases in vocal stereotypy coincide with increases in motor stereotypy may contribute to our understanding of the processes governing behavior–behavior relations (e.g., Rapp, 2008).
Method
Participants and response definitions
Fourteen individuals participated in this experiment. We identified 10 participants from Experiment 1 who exhibited one or more forms of untargeted motor stereotypy. In addition, we included 4 participants who exhibited untargeted motor stereotypy from a study by Lanovaz, Rapp, and Ferguson (in press), which decreased vocal stereotypy using noncontingent access to music. To be categorized as motor stereotypy, the behavior in question had to be emitted for a mean of 10% or more of at least one component in either the baseline sequence or a test sequence. Table 1 provides response definitions for each participant’s motor stereotypy.
Response Definitions for Each Participant’s Untargeted Motor Stereotypy in Experiment 2.
Data collection
We scored videotaped sessions from Experiment 1 for engagement in untargeted forms of motor stereotypy using a duration measure. We converted data for each participant’s engagement in motor stereotypy to a percentage of time measure by dividing the number of seconds engaged in the target behavior by the total number of seconds in a session and then multiplying by 100%.
Data depiction and analysis
We opted to use only the between-sequence analysis (Lanovaz et al., 2010), because it detects larger changes in behavior. As such, increases in untargeted motor stereotypy that are detected at this level of analysis would likely influence a practitioner’s decision to implement an intervention for targeted vocal stereotypy.
Results and Discussion
The data paths for the first components of the two sequences were undifferentiated for each participant; therefore, we did not depict the data in the figures (data available from first author). Figures 3 and 4 contain representative results (other data sets are available from the first author on request). Figure 3 shows the percentage of time Mason (left column) engaged in jumping during the second (first panel) and third (second panel) components and hand flapping during the second (third panel) and third (fourth panel) components of the NI sequence and PS sequence. Results for the second and third components show that Mason’s jumping was typically higher during the PS sequence than during the NI sequence, suggesting that the matched stimulus, which decreased his vocal stereotypy, produced an immediate and subsequent increase in Mason’s jumping. Results for the second components also show that Mason’s hand flapping was marginally higher in the PS sequence than during the NI sequence; however, the data paths for the third components were undifferentiated.
The percentage of time Mason engaged in jumping (left column, upper two panels) and hand flapping (left column, lower two panels), Eric engaged in hand mouthing (right column, upper two panels), and Hal engaged in ear covering (right column, lower two panels) during the NI sequence and PS sequence sessions in the second and third components.
The percentage of time Jared engage in body rocking (left column) and Oliver engaged in hand flapping (right column) during the NI sequence and PS sequence sessions during the second (first panel) and third (second panel) components.
Figure 3 also shows the percentage of time Eric engaged in hand mouthing (right column; upper two panels) and Hal engaged in ear covering (right column, lower two panels) during the second and third components of the NI sequence and PS sequence. Results for both the second and third components show that Eric’s hand mouthing was higher in the PS sequence than during the NI sequence. Thus, the matched stimulus, which decreased his vocal stereotypy, increased Eric’s immediate and subsequent engagement in hand mouthing. Results for the second components show that Hal’s ear covering was higher in the PS sequence than during the NI sequence; however, the data path for the PS sequence decreased across sessions. Data paths for the third components were undifferentiated. Results for Hal suggest that the PS increased his immediate engagement in ear covering but did not increase his subsequent engagement in ear covering.
Figure 4 shows the percentage of time Jared engaged in body rocking (left column) and Oliver engaged in hand flapping (right column) during the second (first panel) and third (second panel) components of the NI sequence and PS sequence. Results for the second components show that Jared’s body rocking was higher during the PS sequence than during the NI sequence. Results for the third components show that Jared’s body rocking was marginally lower during the PS sequence than during the NI sequence, suggesting that the matched stimulus, which did not produce a meaningful reduction in his vocal stereotypy, produced an immediate increase in Jared’s body rocking, and a subsequent decrease in Jared’s body rocking. Results for the second components show that Oliver’s hand flapping was undifferentiated across the PS and NI sequences. During the third components, Oliver’s hand flapping was variable but higher during the PS sequence than during the NI sequence. Results for Oliver demonstrate that the matched stimulus, which did not appreciably decrease his vocal stereotypy, did not alter his immediate engagement in hand flapping but increased his subsequent engagement in hand flapping.
As a whole, results indicate that immediate engagement in motor stereotypy, subsequent engagement in motor stereotypy, or both increased for 8 of 14 participants when a PS was presented noncontingently to treat vocal stereotypy. Specifically, 5 participants (including Mason, Eric, and Hal; Figure 3) exhibited a decrease in vocal stereotypy and an increase in motor stereotypy, 5 participants exhibited a decrease in vocal stereotypy and no change in motor stereotypy, 3 participants (including Jared and Oliver; Figure 4) exhibited no change in vocal stereotypy and an increase in motor stereotypy, and 1 participant exhibited no changes in vocal or motor stereotypy. Results from this experiment suggest that practitioners who treat vocal stereotypy with matched stimulation should also evaluate the immediate and subsequent changes in untargeted forms of motor stereotypy before adopting the intervention for broader application.
General Discussion
Results of Experiment 1 showed that (a) a matched stimulus produced an immediate decrease in vocal stereotypy for 8 of 11 participants and a subsequent increase in vocal stereotypy for only 1 of 8 participants and (b) an unmatched stimulus decreased vocal stereotypy for 1 of 10 participants and did not produce a subsequent increase for that participant. Results of Experiment 2 showed that for 8 of 14 participants, noncontingent access to either a matched or unmatched item increased immediate engagement in an untargeted stereotypy, subsequent engagement in an untargeted form of stereotypy, or both. Results from Experiments 1 and 2, combined with those from the Lanovaz et al. (in press) study, show that a matched stimulus decreased vocal stereotypy without increasing subsequent vocal stereotypy or immediate or subsequent motor stereotypy for just 4 of 15 participants. Thus, even though there is empirical support for treating vocal stereotypy with matched stimulation (Lanovaz & Sladeczek, 2012), practitioners should be aware of the potential for collateral behavior changes.
For some participants, results from Experiment 1 are consistent with results from studies showing that items producing auditory stimulation may decrease immediate and subsequent engagement in vocal stereotypy (Lanovaz & Argumedes, 2009; Lanovaz et al., 2009; Lanovaz et al., 2011; Rapp, 2007). By contrast, results from Experiment 1 for immediate effects of unmatched stimuli on vocal stereotypy are not consistent with findings from the Ahearn et al. (2005) and Lanovaz et al. (2009) studies. It is not clear why our findings for unmatched (i.e., preferred only) stimuli were not consistent with those reported in the literature; however, it is possible that some procedural differences contributed to the differential outcomes. For example, Ahearn et al. evaluated their participant’s preference using an 8-min, single-item preference assessment, whereas the present study used three 10-min, free-operant sessions. It is possible that repeated exposure to the PS during the preference assessment session may have produced satiation for the stimulation generated by the preferred item. Although plausible, this account seems unlikely because the matched stimulus was also available for three 10-min sessions before the PS sequences were conducted. If satiation for the items occurred during those sessions, we would expect, at least to some extent, to also see satiation for the matched items; this did not appear to be the case. Likewise, although Lanovaz et al. used multiple free-operant sessions to identify preferred items, they evaluated unmatched stimuli during relative brief assessments. Had we conducted only two or three sessions with each sequence, our conclusions may have been different for some participants (e.g., see results for Marty). To this end, some studies have found that brief assessments conducted using multielement designs may give rise to a higher percentage of false positives (e.g., Bartlett et al., 2011; Kahng & Iwata, 1999). Thus, conclusions based on lengthier assessments may be more valid.
There are at least three ways to conceptualize the decrease in vocal stereotypy (Experiment 1 and the Lanovaz et al., in press, study) and the concomitant increase in motor stereotypy (Experiment 2) observed for five participants. First, it is possible that the matched item generated stimulation that was functionally similar to the product of vocal stereotypy; thus, vocal stereotypy decreased because the functional reinforcer was provided independent of vocal stereotypy. In this way, the PS directly altered engagement in vocal stereotypy. Thereafter, participants’ engagement in untargeted (i.e., lower probability) motor stereotypy increased as a result of the time made available following reduction of vocal stereotypy. This interpretation is consistent with previous research that has evaluated covariation of multiple forms of automatically reinforced behavior (e.g., Rapp, Vollmer, St. Peter, Dozier, & Cotnoir, 2004). Rapp et al. (2004) found that restricting the most probable form of stereotypy increased untargeted forms of stereotypy for some participants. Thus, decreasing engagement in one automatically reinforced behavior may increase engagement in another automatically reinforced behavior, which may or may not be functionally related to the decreased behavior.
It is also possible that vocal stereotypy decreased because the matched stimulus increased the value of engaging in previously low-probability forms of motor stereotypy. In so doing, the matched stimulus indirectly decreased engagement in vocal stereotypy by evoking other automatically reinforced behavior (i.e., motor stereotypy) and increasing the value of stimulation generated by engaging in that behavior, which ultimately displaced vocal stereotypy. Although Laraway, Snycerski, Michael, and Poling (2003) suggested that a single stimulus event may exert dynamic changes in motivation for multiple consequent events as reinforcers or punishers, this interpretation seems unlikely because (a) many of the behaviors that increased when the PS was present were not physically incompatible with vocal stereotypy and (b) the PS did not decrease vocal stereotypy for three participants who exhibited increases in untargeted stereotypy. That is, participants could, and often did, exhibit vocal stereotypy and other multiple forms of motor stereotypy simultaneously.
Finally, it is possible that continuous access to stimulation (e.g., music) during the second component of the PS sequence disrupted or otherwise extinguished the product of engaging in vocal stereotypy. After the consequence for engaging in vocal stereotypy became unavailable, participants reallocated their responding to other forms of stereotypy, which generated stimulation that was not affected by the matched stimulus. Nevertheless, this account seems unlikely because participants could have avoided (i.e., not selected) the stimulus provided during the second component of the PS sequence during the three preference-assessment sessions. Regardless of the operant processes governing the participants’ behavior, results from this study suggest that researchers should evaluate immediate and subsequent changes in untargeted behavior when treating vocal stereotypy.
Two related outcomes from Experiment 2 warrant additional discussion. During the PS sequence, three participants exhibited both immediate and subsequent increases in motor stereotypy, whereas two participants exhibited only subsequent increases in motor stereotypy. The subsequent increases in motor stereotypy may be attributable to the residual value-altering effects exerted by an establishing operation (EO), which was provided in the second component. Once motor stereotypy was evoked by the matched stimulus, and the behavior contacted reinforcing stimulation, it may have gained “momentum” and become resistant to change even after the PS was removed (Ahearn, Clark, Gardenier, Chung, & Dube, 2003). Alternatively, the matched stimulus may have altered the value of engaging in immediate stereotypy, but each participant had already emitted motor stereotypy at ceiling levels in the absence of the matched stimulus. Put differently, even though the stimulus product of motor stereotypy may have become more valuable in the presence of the matched stimulus, the participants were physically unable to engage in higher levels of stereotypy. Thus, the value-altering effect of the EO in the second component may have extended the time for which the product of motor stereotypy was a reinforcing event, as opposed to increasing immediate engagement in motor stereotypy. Nevertheless, both accounts are speculative.
Some potential clinical implications of the findings should be briefly noted. First, response reallocation may be a problem when noncontingent access to matched or unmatched stimulation is used to treat vocal stereotypy. That is, NCR may decrease immediate engagement in vocal stereotypy; however, it may also increase immediate or subsequent engagement in other forms of stereotypy. Given this possibility, researchers and practitioners who treat automatically reinforced behavior should evaluate the immediate and subsequent changes in targeted and untargeted automatically reinforced behavior. Second, results indicate that selecting items to compete with vocal stereotypy based on preference alone may not be sufficient. Instead, the items should be selected based on both preference and similarity with the putative sensory product of vocal stereotypy. Third, as suggested by the results for Marty, brief assessments (i.e., those with less than three sessions per sequence) may not be sufficient for identifying items that consistently compete with vocal stereotypy over time. Thus, pretreatment evaluations should contain three or more sessions with each sequence (e.g., Bartlett et al., 2011).
The present study potentially contributes to the literature on assessment and treatment of automatically reinforced behavior in at least three ways. First, the results contribute to the growing literature on the use of two- or three-component methodologies for evaluating the effects of interventions on various forms of stereotypy and other automatically reinforced behavior (Chung & Cannella-Malone, 2010; Lang et al., 2009; Lang et al., 2010; Lanovaz et al., 2009; Lanovaz et al., 2011; Rapp, 2006, 2007; Schumacher & Rapp, 2011; Simmons, Smith, & Kliethermes, 2003). Second, the results of this study provide some additional support for the use of structurally matched stimulation for treating automatically reinforced behavior with NCR (e.g., Higbee, Chang, & Endicott, 2005; Patel, Carr, Kim, Robles, & Eastridge, 2000; Piazza et al., 2000; Rapp, 2006, 2007; Saylor, Sidener, Reeve, Fetherston, & Progar, 2012); however, the results are also consist with studies suggesting that noncontingent access to preferred items may produce undesirable side effects (e.g., Friman, 2000; Rapp, 2005; Van Camp et al., 2000).
Some specific findings from this investigation warrant discussion. First, the results provide additional support for the potential clinical utility of the procedures described by Lanovaz et al. (2010) for detecting changes in motivating operations after an intervention has been removed. Specifically, results for Hal (see Figure 2) demonstrate effects of an intervention that decreased immediate engagement in vocal stereotypy and does not increase subsequent engagement in vocal stereotypy. By comparison, results for Barry (see Figure 1) illustrate effects of an intervention that decreased immediate and subsequent engagement in vocal stereotypy. The latter outcome is indicative of a functionally matched intervention for vocal stereotypy (Lanovaz et al., 2010; Rapp, 2006, 2007), and the former is indicative of an intervention that did not produce an EO for subsequent engagement in vocal stereotypy. Although both outcomes are clinically desirable, the effects for Barry were optimum because his vocal stereotypy remained below baseline levels for a period of time after the matched stimulus was removed (Lanovaz et al., 2009). In addition, results for Eric (see Figure 3) illustrate effects of an intervention that decreased vocal stereotypy but also increased both immediate and subsequent engagement in an untargeted motor stereotypy. More specifically, Eric’s results show that increases in untargeted stereotypy can persist or even increase after the intervention is removed. Clinically, the net result of the NCR intervention may not be desirable for Eric.
Some potential limitations of this study should be discussed. First, we did not collect data on participants’ engagement in appropriate vocalizing. Given that no social consequences were provided for either appropriate or inappropriate behavior in either sequence (i.e., each component was conducted during a NI condition), changes in the participants’ appropriate verbal behavior were not expected during this assessment. Nevertheless, it may be beneficial for practitioners to include contingencies for increasing appropriate vocalizing (e.g., Colón, Ahearn, Clark, & Masalsky, 2012), correct responding on academic tasks (e.g., Chung & Cannella-Malone, 2010), or both when treating vocal stereotypy.
A second potential limitation is that we did not evaluate the effects of matched and unmatched stimuli with each participant. Thus, it is possible that an unmatched stimulus could have decreased vocal stereotypy for some of the participants whom a matched stimulus was effective. Nevertheless, the participants were assigned to the conditions using a semirandom process. In addition, for three individuals who participated in both conditions, a matched stimulus decreased vocal stereotypy for two participants, whereas an unmatched stimulus did not decrease vocal stereotypy for any of the three participants. As a related limitation, decreases in vocal stereotypy for some participants were either quite marginal or were sufficiently large but not necessarily clinically meaningful. As such, it may be necessary to supplement NCR with consequent interventions for vocal stereotypy, alternative appropriate behavior, or both.
A third potential limitation is that we did not conduct a complete functional analysis (Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/1994) to rule out the effects of social consequences on the participants’ vocal stereotypy. Despite the fact that we did not evaluate the sensitivity of each participant’s vocal stereotypy to social consequences (e.g., contingent attention), we conducted numerous 30-min sessions (the NI sequence) on alternating days; these sessions should be sufficient for demonstrating the persistence of each participant’s vocal stereotypy in the absence of social consequences.
Finally, we obtained mean IOA scores of less than 80% for some participants whose vocal stereotypy typically occurred between 40% and 60% of the time during sessions in Experiment 1. Further analysis of sessions from Experiment 1 indicated that (a) there was no pattern of systematic bias by observers (e.g., neither observer consistently scored vocal stereotypy for a higher percentage of the sessions in either sequence) and (b) the overall duration for which vocal stereotypy was scored in a given session was typically very similar for both observers. Thus, the lower IOA scores for some participants’ vocal stereotypy in Experiment 1 should not necessarily detract from the believability of the findings.
The results of this study provide some avenues for future research. First, it is possible that combining antecedent procedures, such as continuous access to matched stimuli, with consequent procedures, such as response interruption and redirection, could produce greater decreases in immediate engagement in vocal stereotypy than either type of procedure alone. Future research should explore this possibility. Second, future research should evaluate the extent to which arbitrarily providing structurally matched stimuli decreases specific forms of stereotypy. For example, Lanovaz et al. (2011) recently evaluated the effects of different parameters of auditory stimulation on the vocal stereotypy of two participants without conducting a formal, pretreatment preference assessment. Lanovaz et al. found that auditory stimulation decreased both participants’ immediate engagement in vocal stereotypy and did not increase either participant’s subsequent engagement in vocal stereotypy. If the structural or overt match of the stimulation is more important than the individual’s preference for the item, practitioners may be able to forgo formal preference assessments in lieu of choosing items that match the putative products of automatically reinforced behavior.
Footnotes
Authors’ Note
Portions of this study were conducted in partial fulfillment of a master’s degree in behavior analysis by Regina A. Carroll and Stephanie M. Sheridan.
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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by funds provided to John T. Rapp and Ethan S. Long by Autism Speaks Grant 2852.
Portions of the study were the project were supported by an experimentation grant from the Office des personnes handicapées du Québec provided to John T. Rapp and Marc J. Lanovaz. We thank the Centre de réadaptation de l’Ouest de Montréal for their collaboration with the study.