Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by deficits in social communication and interaction and restricted and/or repetitive patterns of behavior, interests, or activities (American Psychiatric Association, 2013). ASD may present in many ways, ranging from individuals who display mild social impairments to individuals who are severely impacted with significant intellectual disabilities. For the latter group, the presence of challenging behavior is common and can limit opportunities to access the community (Hill et al., 2014). As such, practitioners have an obligation to develop expertise in the use of reinforcement-based behavioral interventions (Scheuermann, Webber, Boutat, & Goodwin, 2003).

Differential reinforcement of other behavior (DRO) is an extensively used reinforcement-based intervention that is effective for reducing problem behavior (Jessel & Ingvarsson, 2016; Lennox et al., 1988). When a DRO procedure is implemented, a reinforcer is provided contingent upon the absence of a specified response within a specific period of time (Homer & Peterson, 1980). In addition to their well-documented efficacy, DRO reinforcement systems are useful in that they can be implemented in a flexible manner and tailored to an individual’s needs. DRO reinforcement systems can be implemented in many different ways, including: with fixed or variable time intervals (Della Rosa et al., 2015; Nuernberger et al., 2013; Rapp et al., 2017; Rozenblat et al., 2009), on fixed or momentary schedules (Bergstrom et al., 2011), and in signaled and unsignaled preparations (Hammond et al., 2011; Jessel & Ingvarsson, 2016; Watts et al., 2013).

Of the DRO variations mentioned above, there have been relatively few studies investigating the use of signals to clarify reinforcement contingencies. In fact, in their review of 45 articles that employed DRO systems, Weston et al. (2018) found that 61% of the reviewed studies did not use any signal to indicate the contingency associated with the DRO schedule. When signals were integrated, they were often used as a cue to alert the individual to the length of time left to contact reinforcement, presumably clarifying the expectations of the contingency. Signals used in DROs have taken many forms and have included turning away from, prompting, or praising the behavior (Harris & Wolchik, 1979), gestures (Repp et al., 1983), and tones/gestures/music (Kahng et al., 2001; Sisson et al., 1988). While several types of signals have been used with DROs, the effects on target behavior have been largely inconsistent (Hammond et al., 2011). Further, few studies have explicitly compared signaled and unsignaled DRO systems to determine the relative effects signals may have on dependent variables.

Comparisons of Signaled and Unsignaled DRO Systems

While the use of signaled delays to reinforcement is fairly new in applied research, there have been some applications in the basic literature. For example, Schall and Branch conducted a series of studies evaluating the use of signaled delays in conjunction with variable-interval reinforcement schedules on key pecking in pigeons (Schall & Branch, 1988; Schall & Branch, 1990; Schall, Schuh, & Branch, 1992). Interestingly, there tended to be increases in sustained responding (i.e., key pecking) during signaled reinforcement delays relative to unsignaled delays. In addition, intrasubject variability in responding was common across the studies, raising questions about the mechanism of action for the signals. More recently, researchers have applied the use of signals to differential reinforcement procedures in applied settings, particularly in conjunction with DRO reinforcement systems. Hammond et al. (2011) examined the use of signaled and unsignaled fixed momentary DRO contingencies on problem behavior (aggression, self-injury) exhibited by four children with developmental disabilities. Three conditions were used to evaluate the effects of DRO reinforcement systems on problem behavior using an ABAC reversal design (i.e., baseline, unsignaled DRO, baseline, signaled DRO). During baseline conditions, the participant was given 90-s pre-session access to preferred tangible items. The items were then removed at the start of the session. The occurrence of problem behavior resulted in access to the tangible items (for 10-s). In the unsignaled DRO condition, reinforcement was provided for the absence of target behavior over a 300-s (5-min) interval. In this condition, no signal was present during the DRO interval. In the signaled DRO condition, reinforcement was also provided for the absence of target behavior over a 300-s (5-min) interval. However, in the signaled DRO condition, the experimenter used a signal (i.e., holding the reinforcer above the experimenter’s head) for the final 3 s of the DRO interval. The results from the investigation indicated that all participants responded favorably to the DRO reinforcement contingencies relative to baseline. More robust reductions were found in the unsignaled DRO condition for two participants, while the other two had more pronounced reductions in the signaled DRO condition. These findings suggest that, in some cases, signals may aid in behavioral discrimination in the presence of varied reinforcement contingencies within DRO systems.

Watts et al. (2013) also evaluated the effectiveness of signaled and unsignaled DRO procedures using lower preference and higher preference (HP) toys. In this investigation, the authors evaluated a “rule” and “no rule” DRO contingency in a multielement design. The “rule” DRO was a vocal contingency (a signaled DRO) delivered to the learner, stating that he/she could earn reinforcement for not engaging with the HP toy. The “no rule” DRO was conducted the same as the “rule” condition, except a contingency (signal) was not given. This signal was provided at the onset of the DRO interval as opposed to seconds before the end of the interval as described in the Hammond et al. (2011) study. Results of the study indicated near-zero levels of HP toy play across all participants during the signaled DRO/“rule” contingency. Further, the “no rule” DRO contingency effectively reduced HP toy play to near-zero levels for two of the four participants. These results indicate that the use of signals may enhance the effectiveness of DRO procedures.

While previous studies have suggested that signaled DRO systems may be useful for decreasing problem behavior, the literature on the topic remains relatively undeveloped. Only a few evaluative studies exist and the findings from those investigations are mixed. In addition, the vast majority of the existing literature on the topic has focused on the treatment of target behavior in young children. In a review of the DRO literature, Weston et al. (2018) noted that 88% of study participants were 18 or younger and 71% of participants were under 13. Further, of the existing literature, very few studies have evaluated the use of technology to integrate signals into DRO systems. These gaps in the literature formed the basis for the current investigation. The purpose of the current study was to compare the relative effectiveness of signaled versus unsignaled DRO contingencies to reduce automatically reinforced stereotypical behavior exhibited by three adolescents and adults with ASD.

Methodology

Participants and Setting

Three individuals who attended a university-based program for children and adults with ASD participated in the current investigation. All participants had a diagnosis of ASD and were chosen for inclusion in this study for the treatment of motor or vocal stereotypy that disrupted their academic and vocational programming. Matt and Tim were 28 and 18 years old, respectively, and both engaged in vocal stereotypy (e.g., non-contextual vocalizations, repetitive utterances, humming); Antonio was a 17-year-old male who engaged in motor stereotypy (e.g., repetitive face tapping, head movements).

Sessions for Matt and Tim were conducted in a 4.9 m by 5.8 m conference room. Sessions for Antonio were conducted in a 6.4 m by 10.5 m classroom. Rooms were equipped with tables, chairs, and materials required to run the sessions (e.g., timer, data sheet, tablet with countdown application, and student-specific reinforcers). One to three sessions were conducted per day and sessions were typically conducted two to three days per week, dependent upon participant availability.

Functional Analysis

A functional analysis (FA) procedure was used to determine the function of the participants’ stereotypical behaviors (Iwata, Dorsey, Slifer, Bauman, & Richman, 1982, 1994; Querim et al., 2013). Specifically, an automatic reinforcement screening procedure with multiple ignore conditions run consecutively was implemented (Querim et al., 2013). Ignore conditions were 5-min in length with the participant and an instructor in the room. The instructor provided no consequences for the occurrence of stereotypy. Data were collected on vocal (Matt and Tim) and motor (Antonio) stereotypy.

The dependent variable measured during the FAs was the percentage of trials with vocal or motor stereotypy. Partial-interval data (20-s intervals) were collected for the presence or absence of stereotypical behavior. The 20-s interval was selected based on the average interresponse time between instances of stereotypy derived from daily data collection in the participants’ classroom. A second trained observer independently recorded data from video recordings of functional analysis screening sessions for 33.3% of sessions for Matt, 33.3% of sessions for Tim, and 33.3% of sessions for Antonio. Interobserver agreement was calculated using the scored-interval method. Mean agreement for the presence or absence of either vocal or motor stereotypy was 100.0% for Matt, Tim, and Antonio during the functional analysis screening.

Signaled v. Unsignaled DRO Analysis

Baseline Stereotypy data collected during the functional analysis screening served as the baseline for all three participants in the DRO analysis.

Unsignaled DRO. Prior to the start of each unsignaled DRO condition, a brief multiple stimulus without replacement (MSWO) preference assessment was conducted to identify a high preference item for the contingency (DeLeon & Iwata, 1996). A variety of tangible and edible items were offered in a five-item array at the start of each session. The same tangible/edible was consistently selected by all participants (i.e., candy for Matt, iPad for Tim and Antonio). At the start of the unsignaled DRO condition, a therapist briefly explained the reinforcement contingency to each participant (e.g., “If you sit with quiet hands, you can earn _____”). No other cues were present during this condition to indicate the contingency required to earn reinforcement. Reinforcement was provided along with vocal praise (e.g., “Great sitting. You earned ___,” or similar variation) for the absence of stereotypy for the entire 20-s interval. If the participant engaged in stereotypy, he did not contact reinforcement and the DRO interval expired prior to initiating another trial. No vocal feedback was provided to indicate that the participant did not earn reinforcement.

Signaled DRO Prior to the start of each signaled DRO condition, a brief MSWO preference assessment was conducted to identify a high preference item for the contingency (DeLeon & Iwata, 1996). The procedures for the signaled DRO condition were identical to those of the unsignaled DRO condition with the exception of the use of a countdown timer application on a tablet (Children’s Countdown Timer ©, Fehners Software, LLP) to visually show the passage of time until the end of the interval. The tablet was placed directly in front of the participant with the application open. Following the presentation of the vocal contingency to earn reinforcement, the therapist would start the timer by pressing the “Start” icon on the screen of the tablet. The application included both visual and auditory components. As the timer counted down, a picture of the reinforcer to be earned was gradually revealed. When the interval expired, the full image of the reinforcer was displayed, and a foghorn sound played from the tablet. Reinforcement (presentation of the chosen item contingent upon the absence of stereotypy) was paired with vocal feedback (e.g., “Great sitting. You earned ___,” or a similar variation). If the participant engaged in stereotypy, he did not contact reinforcement and the DRO interval expired prior to initiating another trial. No vocal feedback was provided to indicate that the participant did not earn reinforcement.

Response Measurement and Reliability

The dependent variable measured was the percentage of trials with vocal or motor stereotypy. Each session consisted of ten DRO trials and 20-s partial-interval data were collected on stereotypy. Data were then converted to the percentage of trials with motor or vocal stereotypy. Data were not collected for stereotypy during the reinforcement interval for any of the participants, as doing so would place additional requirements on busy therapists. A second trained observer independently recorded data during the session or from video recordings of sessions for 42.7% of sessions for Matt, 33.3% of sessions for Tim, and 30.0% of sessions for Antonio. Interobserver agreement was calculated using the scored-interval method. Mean agreement for the presence or absence of either vocal or motor stereotypy during the DRO comparison was 98.3% for Matt (range: 90.0% to 100%), 100% for Tim, and 91.3% (range: 70.0–100%) for Antonio.

Design The signaled and unsignaled DRO conditions were compared using an alternating treatments design for all three participants.

Signaled DRO Schedule Thinning (Antonio only)

After the completion of the initial DRO analysis, a schedule thinning procedure was implemented for Antonio to gradually increase the delay to reinforcement.

Baseline. During baseline conditions, Antonio was seated at a table. The instructor was seated next to him throughout the session. Ten 20-s trials were conducted during each baseline condition. No instructions were provided, and stereotypical behavior contacted no differential consequence during baseline.

Signaled DRO Thinning. In this condition, the effective signaled DRO interval from the initial analysis was used (20 s). The reinforcement schedule was then thinned following, a) two consecutive sessions with no motor stereotypy or b) three sessions with motor stereotypy in 20% of intervals or fewer. Following the aforementioned criteria for thinning the DRO interval, the initial 20-s interval was thinned to 30-s and then 40-s. At this point, a reversal to baseline was conducted to exhibit experimental control of the signal over Antonio’s stereotypy. Following the reversal, the DRO interval was thinned to 60-s, 80-s, 100-s, and then finally 120-s following the same criteria for thinning. The length of the reinforcement period was also gradually increased to match the increasing DRO interval. The terminal reinforcement period for Antonio during schedule thinning was 45-s. Interobserver agreement was calculated using the scored-interval method for 34.5% of sessions. Mean agreement was 95.0% (range: 80.0% to 100%) during the signaled DRO thinning analysis.

Design. The signaled DRO schedule thinning was evaluated using a reversal design to compare baseline to intervention.

Results

Functional Analysis

The results from the automatic reinforcement screening for all three participants are summarized in the top, middle, and bottom panels of Fig. 1 (Matt, Tim, and Antonio, respectively). under the condition indicated as “Baseline” High levels of stereotypical behavior were observed for all participants, suggesting that stereotypy was maintained by automatic reinforcement for all three.

Fig. 1
figure 1

Automatic reinforcement screening functional analysis results for Matt, Tim, and Antonio.

Full size image

Signaled v. Unsignaled DRO Analysis

Results from the signaled v. unsignaled DRO analysis for all three participants are summarized in Fig. 1.

Fig. 2
figure 2

. Results of the DRO comparisons across baseline and intervention phases for Matt, Tim, and Antonio

Full size image

Matt The results for Matt are depicted in the top panel of Fig. 1. During baseline, vocal stereotypy occurred in 100.0% of 20-s intervals. In the unsignaled DRO condition, vocal stereotypy decreased to an average of 38.0% of intervals, which was a 62.0% reduction from baseline. In the signaled DRO condition, vocal stereotypy decreased to an average of 32.0% of intervals, which was a 68.0% reduction from baseline.

Tim The results for Tim are depicted in the middle panel of Fig. 1. During baseline, vocal stereotypy occurred in 100.0% of intervals. In the unsignaled DRO condition, vocal stereotypy occurred in 96.7% of intervals, which was a 3.3% reduction from baseline. In the signaled DRO condition, vocal stereotypy decreased to an average of 36.7% of intervals, which was a 63.3% reduction from baseline.

Antonio The results for Antonio are depicted in the bottom panel of Fig. 1. During baseline, vocal stereotypy occurred in 100.0% of intervals. In the unsignaled DRO condition, vocal stereotypy decreased to an average of 78.0% of intervals, which was a 22.0.% reduction from baseline. In the signaled DRO condition, vocal stereotypy decreased to an average of 25.0% of intervals, which was a 75.0% reduction from baseline.

Signaled DRO Thinning (Antonio only)

The results for Antonio’s signaled DRO thinning analysis are summarized in Fig. 2 and Table 1. In the initial baseline, Antonio engaged in motor stereotypy in an average of 90.0% of 20-s intervals. Motor stereotypy decreased to 0.0% of 20-s intervals at the 20-s signaled DRO interval. Levels of stereotypy remained low when the DRO intervals were increased to 30-s and 40-s (10.0% of intervals with stereotypy). During the reversal to baseline, motor stereotypy increased to 100.0% of intervals. Upon returning to treatment, motor stereotypy remained low at the 60-s, 80-s, 100-s, and 120-s signaled DRO intervals (Fig.3) (6.7%, 16.7%, 16.0%, and 14.0% of intervals, respectively). Overall, the signaled DRO resulted in an 85.5% reduction from baseline and remained low at the 120-s interval.

Fig. 3
figure 3

Results of schedule thinning during signaled DRO condition for Antonio

Full size image
Table 1 Schedule thinning percent reductions of stereotypy from baseline: Antonio
Full size table

Discussion

The purpose of the current investigation was to evaluate the relative effectiveness of signaled and unsignaled DRO reinforcement systems for decreasing stereotypical behavior in three adolescents and adults with ASD. The results of the comparative DRO analysis show that the unsignaled DRO was effective for only one of the three participants (Matt), while the signaled DRO was effective for reducing stereotypical behavior for all three participants. The present findings suggest that a relatively simple procedural modification (the presence of a visual countdown application) enhanced the effectiveness of the DRO procedure. In addition, it was possible to gradually increase the DRO interval length from 20-s to 120-s for one of the participants (Antonio). These findings add to a relatively limited literature on the use of signaled delays and schedule thinning while using DRO reinforcement systems (Hammond et al., 2011; Weston et al., 2018). These results are clinically significant in that participants of this study all engaged in stereotypical behavior 100% of 20-s intervals during the FA screening procedure.

While this intervention did not totally eliminate stereotypical behavior and automatically maintained stereotypy is historically challenging to treat, the target behaviors were reduced to levels that were much more manageable. The reductions observed during the DRO comparison resulted in more learning opportunities across participants, as they were then available to attend and respond to instructional discriminative stimuli (SDs). Additionally, participants were able to engage in stereotypy during the reinforcement intervals. This represents an advantage of the present DRO procedure as permitting access to an automatically maintained, non-destructive behavior both could serve to reinforce abstinence from stereotypy during the DRO interval as well as control for the potential unintended side effects of a procedure aimed to eliminate such behavior completely such as more intense forms of problem behavior (e.g., aggression, destruction, self-injury).

In addition to adding to the sparse literature regarding the effects of signaled delays, the findings from the current investigation also highlight the potential utility of signaled delays for fading the delay to reinforcement while maintaining intervention effects (Antonio). Previous research has suggested that clinicians often experience treatment failures and reemergence of problem behavior when introducing delayed access to reinforcement for responses that were previously reinforced immediately. While the use of signals in DRO systems has been shown to help individuals tolerate delays to reinforcement (Kelley et al., 2011), much more research is necessary to evaluate the effectiveness of signals for increasing delays to reinforcement within DRO reinforcement systems.

The findings from the current investigation are of particular importance from an ethical standpoint for practitioners. For two of the three participants, the use of a traditional, unsignaled DRO procedure (even at very short intervals) was not effective for decreasing stereotypical behavior. The addition of a signaled delay had clinically significant effects for these two participants. This is particularly notable as these two individuals otherwise may have been considered “non-responders” to reinforcement-based procedures as they are typically used in applied settings (DRO systems implemented without signals). Assuming these two participants had been in a traditional placement (e.g., a school or adult program), next steps may have included more intrusive intervention strategies (i.e., punishment-based interventions) to address their stereotypical behavior. An assumption such as this poses significant ethical issues for practitioners and could explain increases in the use of other punishment-based procedures to address stereotypy in applied settings (Campbell, 2003). It is important to note a multitude of reinforcement-based procedures are available and should be attempted prior to considering punishment-based procedures.

While all participants in the current investigation responded favorably to the signaled DRO, the added complexity entailed (i.e., navigating the app, taking photographs of preferred items, carrying around an iPad/tablet) could lead to errors and/or be stigmatizing when used in community settings. The current protocol could potentially be used as a screening procedure to determine whether the use of signaled DROs is necessary while transitioning to unsignaled DRO systems that may be easier to implement and less intrusive in community settings. The protocol was brief and user-friendly, which may make it an ideal assessment for that purpose.

While the results of the current study clearly suggest that there may be benefits to using signaled DRO reinforcement systems, there remain several areas that require further investigation. For instance, the signaled delay in the current study was a timer application on an electronic tablet. While participants were not allowed to directly interact with the tablet during the analyses, prior reinforcement history with tablets may have contributed to the treatment effects. Future studies could investigate the utility of signaled delays to reinforcement with non-electronic devices (e.g., an analogue visual timer) or smaller devices (e.g., smart phone/watch). Given the small sample size in the current investigation, researchers should replicate these findings with more children and adults with ASD.

Additionally, future research could evaluate the feasibility of continued fading of the current intervention. For example, Antonio’s terminal DRO schedule was 120-s with a 45-s reinforcement interval. Logical next steps could include both continued fading to a longer DRO schedule as well as the insertion of task demands during what can be conceptualized as the S-delta component of a multiple schedule (e.g., Hagopian, Bruzek, Bowman, & Jennet, 2007). In doing so, stimulus control of stereotypy maintained by automatic reinforcement might be further strengthened while academic and vocational SDs are presented.

Finally, the present study addresses a gap in the literature related to research in best practices for adults with ASD. A signaled DRO procedure may benefit adults with ASD who engage in automatically reinforced problem behavior. Further, this procedure may reduce situationally inappropriate and/or disruptive behaviors in community-based settings (e.g., hand flapping and/or loud vocalizations in an office) and allow for increased opportunities for integration. Future studies may investigate the ability of adults with ASD to self-manage the signaled DRO contingency and thus promote further independence for a population of individuals with already limited access to resources and funding for quality instructional and behavioral programming.