Increasing Exercise Intensity: Teaching High-Intensity Interval Training to Individuals with Developmental Disabilities Using a Lottery Reinforcement System

Abstract

Rates of overweight and obesity are above 70% in typically developing adults in the United States, with higher rates observed in individuals diagnosed with developmental disability (DD). Lottery reinforcement systems have been validated as effective exercise interventions for individuals with DD. Although high-intensity interval training (HIIT) has demonstrated health benefits, it has not been studied using individuals within this population. The purpose of this study was to implement a lottery reinforcement system to systematically increase heart rate (HR) during 30-min HIIT sessions with 3 adults with DD. Results demonstrated increases in HR from below to within the prescribed range in all 3 participants. For 1 participant, weight decreased by 10.8 pounds during the 9-week program. Implications include that lottery systems increase exercise intensity with adults with DD, that HR during exercise can be reliably controlled using a lottery system, and that similar programs may result in health benefits.

Social Significance

Overweight and obesity is a socially significant problem in the United States, which negatively impacts both quality of life and longevity. According to the Centers for Disease Control and Prevention (CDC, 2019), 71.6% of adults over the age of 20 in the United States were overweight or obese. In 2014, national health care costs associated with obesity were $149 billion (Kim & Basu, 2016), with the annual obesity-related costs specifically for individuals with disabilities accounting for $44 billion of the estimated cost (CDC, 2019). Individuals with developmental disability (DD) are at an even greater risk (CDC, 2019; Melville et al., 2008), contributing to a higher prevalence of heart problems, cardiovascular and respiratory disease, diabetes, cancer (World Health Organization [WHO], 2010), and reduced life expectancy (Patja, Iivanainen, Vesala, Oksanen, & Ruoppila, 2000). Although biological factors including higher rates of leptin resistance (Magge, O’Neill, Shults, Stallings, & Stettler, 2008) and hypothyroidism (Hardy et al., 2004) require consideration, a reduced rate of physical activity (Cervantes & Porretta, 2010) represents both a functional cause of overweight and obesity and an intervention target (Hausman & Kahng, 2015). Given the implications to health, quality of life, and the economic impact associated with overweight and obesity, it is important to examine treatments that promote physical fitness in at-risk populations, including individuals with DD.

Motivating Factors

Understanding the motivating factors involved in the choice between exercise and more sedentary behavior is critical to the treatment of overweight and obesity. Three key principles will be outlined in this context: motivating operations, the matching law, and temporal discounting. First, motivating operations are environmental events that occur prior to the onset of a target behavior (Michael, 1982). These stimuli alter both the reinforcing value of a stimulus and the likelihood that a behavior will be evoked to access it (Michael, 1993). Skipping breakfast likely increases the value of food as a reinforcer, thus increasing the likelihood of food-seeking behavior (Tapper, 2005). Second, the matching law describes the phenomenon in which response rate matches the rate of reinforcer delivery when an individual is presented with concurrent behavioral options (Herrnstein, 1961); it has been conceptualized by behavioral scientists as the mechanism of choice. According to the matching law, lean schedules of reinforcement for healthy nutrition and exercise behaviors would be selected less frequently than unhealthy but immediately reinforcing alternatives, such as streaming movies from the couch and eating calorically dense foods (Critchfield & Kollins, 2001). When evaluated in the context of fitness, specific attention should be paid to the small cumulative effects exercise and nutrition exhibit on physiological markers such as weight or body composition. Third, temporal discounting (TD) is defined as the decrease in reinforcer value as the time between the target behavior and its reinforcer is increased. Individuals with steeper discounting rates may be more likely to exhibit behaviors such as accessing immediately available sweets and television rather than those resulting in delayed physical benefit. TD was positively correlated with body mass index (BMI) and poor nutrition and negatively correlated with exercise frequency (Barlow, Reeves, McKee, Galea, & Stuckler, 2016; Sweeney & Culcea, 2017). Conversely, lower discounting rates were associated with positive outcomes for those involved in fitness interventions (Barlow et al., 2016).

Self-Control Interventions

Behavior analysis offers an effective teaching approach to self-control, defined as the delay of access to more immediately available but lower magnitude reinforcers for later access to larger magnitude reinforcers. Schweitzer and Sulzer-Azaroff (1988) demonstrated that preferences for delayed reinforcers could be shaped in typically developing preschool participants who exhibited high discounting rates at baseline. Dixon, Rehfeldt, and Randich (2003b) used a progressive delay procedure to shape preference for larger, delayed reinforcers over smaller, more immediate reinforcers in three participants with moderate to profound intellectual disabilities. Dixon, Horner, and Guercio (2003a) showed that response requirements and choice interventions could shift preference from small-magnitude reinforcers delivered immediately to delayed, larger magnitude reinforcers in a participant with traumatic brain injury.

Promoting Exercise in Participants With DD

Behavior-based interventions have a long history of promoting physical activity in participants with DD. Allen and Iwata (1980) increased exercise participation during physical education classes in 10 participants with DD by presenting a preferred exercise activity following completion of a nonpreferred exercise activity. Bennett, Eisenman, French, Henderson and Shultz (1989) provided token reinforcement for exercise participation to three individuals with Down syndrome. Participants increased exercise frequency and exhibited improvements in resting heart rate (HR) following intervention. Todd and Reid (2006) utilized a treatment package consisting of self-monitoring combined with verbal and edible reinforcement to increase distance walked, jogged, and snowshoed during 30-min sessions in three participants with DD. Physical activity levels continued to increase as the authors faded edible reinforcers. Krentz, Miltenberger, and Valbuena (2016) implemented token reinforcement to increase walking distance in four individuals with DD.

Lottery Reinforcement Systems

Lottery reinforcement systems provide advantages over fixed conditioned reinforcement procedures (e.g., token reinforcement) in their inherent use of variable reinforcement schedules for backup reinforcer delivery. Variable reinforcement schedules have demonstrated greater resistance to extinction than fixed reinforcement schedules and offer more promising long-term outcomes (Ferster & Skinner, 1957). Lottery reinforcement systems have been effective in promoting physical fitness in various populations (Epstein, Wing, Thompson, & Griffin, 1980; Li, Curiel, Ragotzy, & Poling, 2018; Thyer, Irvine, & Santa, 1984) and have demonstrated social validity (Li et al., 2018). In addition to behavioral changes, some lottery reinforcement systems have resulted in weight loss. Volpp et al. (2008) demonstrated that participants enrolled in a lottery system lost an average of over 13 pounds over a 16-week period. Petry, Martin, Cooney, and Kranzler (2000) first reported the use of a modified lottery system in which participants earned opportunities to draw a potential prize from a “prize bowl” (Petry et al., 2004) on a continuous reinforcement schedule. Backup reinforcers were delivered on a variable schedule according to the lottery slip drawn from the bowl. Washington, Banna, and Gibson (2014) were the first to apply this procedure to increase physical activity. The aforementioned studies demonstrate that behavior-analytic interventions, including lottery reinforcement systems, can improve exercise behavior across populations.

Exercise Intensity

De Luca and Holborn (1985, 1990, 1992) demonstrated the efficacy of fixed-interval, fixed-ratio, and variable-ratio reinforcement schedules on exercise duration and pedaling rate during stationary bike exercise with typically developing obese and nonobese children. Changing-criterion designs (Hartmann & Hall, 1976; Kazdin, 1982) were used to safely and systematically shape exercise behavior to greater duration and rate. Although rate is an objective measure of exercise speed (De Luca & Holborn, 1992), it is not the most valid measure of exercise intensity. Intensity of running or pedaling at a rate of 5 miles per hour (mph) for 5 min would vary based on an individual’s age, weight, current patterns of physical activity, and so forth. Exercise intensity is a key factor in maximizing the physiological benefits of exercise (Mayo Clinic Staff, 2019). Although other measures of intensity may provide additional benefit (Roy, 2015; Weltman et al., 1989), availability and practicality can be prohibitive. Measures based on HR represent a more valid target for overall health impact than previous studies’ measures (De Luca & Holborn, 1990, 1992; Washington et al., 2014) and a more generalizable solution for applied settings given the accessibility of heart rate monitors (HRMs). Moreover, Eckard, Kuwabara, and Van Camp (2019) demonstrated that HR can be modulated through operant contingencies.

High-Intensity Interval Training

High-intensity interval training (HIIT) is an exercise procedure with strong empirical support (Laursen & Jenkins, 2002). During HIIT, participants alternate between intervals of extreme aerobic activity and recovery periods in which HR is lowered. For example, an individual may walk for 3 min at a rate that elicits relatively low HR, then sprint at a rate that elicits an increase in HR for 1 min. Each interval is repeated. Ramos, Dalleck, Tjonna, Beetham, and Coombes (2015) reported that target HRs during high-intensity intervals ranged between 60% and 90% with an average of 74.4% of maximum HR (HRMax). HIIT is a time-efficient exercise procedure that has been shown to induce the metabolic benefits associated with prolonged endurance training in a much shorter time frame (Gibala & McGee, 2008). Vigorous physical activity produces health benefits equivalent to moderate levels of physical activity in half the time (Viana et al., 2019). HIIT is associated with reductions in total body fat and insulin (Trapp, Chisholm, Freund, & Boutcher, 2008) and improved cardiovascular and metabolic outcomes relative to steady-state aerobic exercise. Critical for generalized outcomes, HIIT has been shown to have higher compliance rates (Shiraev & Barclay, 2012). Interventions have been evaluated in both healthy and at-risk populations, but no research has examined the use of HIIT in participants with DD.

The purpose of this study was to assess the efficacy of a modified lottery system similar to that used by Washington et al. (2014) to increase HR during HIIT in three participants with DD. Specifically, experimenters hypothesized that implementing a lottery reinforcement system contingent on reaching target HRs that were systematically increased using a changing-criterion design (De Luca & Holborn, 1992; Hartmann & Hall, 1976; Kazdin, 1982) would yield increases in exercise intensity during treadmill exercise.

Method

Participants and Setting

Three participants residing in an independent supported-living (ISL) facility for high-functioning adults with DD participated in this study. As part of admissions criteria, residents were required to secure employment or a volunteer position for 20 hr per week, medicate with minimal oversight, and communicate needs effectively. Participants were all able to make a grocery list that could be checked by ISL staff prior to shopping. They were able to plan and order food at a restaurant and could discern, as well as select, basic healthy options from less healthy ones. For example, each could select light mayonnaise as opposed to regular or a healthier salad as opposed to a burger. Inclusion criteria for this study included a BMI in the overweight or obese range as defined by the CDC (2019) and the ability to modulate HR when instructed following training. All participants received medical clearance before participation. Exercise took place in the fitness center located in the ISL facility. This area included two identical treadmills, three stationary bikes, a dumbbell rack, and a Nautilus machine. Participant demographics, including age, height, weight, BMI, diagnoses, and terminal HR targets, are presented in Table 1.

Table 1.

Participant Demographics

Participant	Age	Height (feet/in.)	Weight (pounds)	BMI	Diagnoses	Terminal HR
Alex	32	5'2"	221.4	40.5	Down syndrome	139 (75% of HRMax)
Laura	40	5'5"	180	30	Intellectual disability	144 (80% of HRMax)
Carla	25	5'4"	165	28.3	Intellectual disability/ cerebral palsy	156 (80% of HRMax)

Previous Token Economy

All participants were involved in a token economy targeting physical health behavior for 2 years prior to program implementation. Individual goals to increase water consumption, frequency of exercise, and so forth were targeted, and tokens were provided each day contingent on meeting these goals. Exercise intensity was not targeted. Tokens were then exchanged for low-calorie snacks, one-on-one time with staff, or gift cards to a healthy restaurant. Exchanges took place weekly during a house meeting.

Experimental Design

A nonconcurrent multiple-baseline across-participants design (Watson & Workman, 1981) was combined with a changing-criterion design (Hartmann & Hall, 1976; Kazdin, 1982) to safely and systematically shape target HR during HIIT sessions. HRMax was identified by subtracting age from 220, as recommended by the Mayo Clinic (2019). Terminal targets were identified at 75% of HRMax (Alex) and 80% of HRMax (Carla and Laura). For Carla and Laura, three total phases were implemented to increase HR from baseline to the terminal target and are presented in Table 2. Criteria changes for Carla and Laura averaged 14 beats per minute (BPM) and 15 BPM, respectively. These numbers were selected by dividing the difference between the average BPM during baseline and the terminal target by three treatment phases. For Alex, six total phases were implemented to increase HR from baseline levels to the terminal target, including one reversal. All target HRs are presented in Table 2. Phase 5 consisted of a return to a lower criterion of HR in which Alex was provided a target HR near baseline BPM, further demonstrating that target behavior was controlled by programmed contingencies. Phases for all participants varied in length to increase experimental control but required at least three consecutive data points at or above criteria before moving to the next phase. Additional experimental control was increased in Alex’s design by varying the percentage that HR increased between conditions (Kazdin, 1982).

Table 2.

Represents target heart rate for each phase change for each participant.

Target HR
Participant	Baseline	Phase 1	Phase 2	Phase 3	Phase 4	Phase 5	Phase 6
Alex	NA	110	115	125	130	110	137
Carla	NA	130	145	160
Laura	NA	115	130	145

Materials

HR was measured using a Garmin Premium Heart Rate Monitor Soft Strap. The HRM was attached just below the center of the participant’s sternum using elastic straps that wrapped securely around the chest. The HRM displayed HR on a screen on the TRUE Fitness PS100T treadmill where it was clearly visible to both the participant and the first author. The same treadmill was used for the duration of this study. An iPhone 4s was used to record video of HR displayed on the treadmill screen during high-intensity intervals. An Ozeri Precision digital bath scale was used for all weight measurements, which took place on the wood floor outside the fitness room following each exercise session.

Preparation

The back of the HRM was moistened by holding it under running water, as recommended by the HRM’s instructions. Participants were assisted in strapping the HRM under their shirts with skin contact just below the sternum and the label facing outward. The participant was instructed to stand with feet straddling the running belt, then turn on the treadmill. HR was observed for 30 s to ensure stability and accuracy. If HR was highly variable (changes by more than 10 BPM total) during this observation period, the HRM was remoistened and reapplied. This process was repeated until HR remained stable (within 10 BPM for the entire 30-s duration).

Training Procedures

Prior to study implementation, the first author conducted a self-report preference assessment with each participant that documented chosen stimuli among possible rewards. Stimuli included social activities—for example, watching a movie with staff—and edible items. A behavioral skills training (BST) procedure was used to demonstrate the effects of exercise speed on HR. During the didactic instruction phase, HR was described to each participant. Each participant was shown where to find HR displayed on the treadmill screen. The first author stood with feet straddling the running belt with HR displayed on the treadmill screen. The participant stood near the treadmill screen so that the HR display was visible. Each participant then read the first author’s resting HR from the HR display. The first author stated that HR increases when speed increases and that HR decreases when speed decreases. The participant was probed for comprehension (“If I run faster, will my heart rate go up or down?”). Corrective feedback was provided if the participant responded incorrectly and repeated until the participant responded correctly to three consecutive trials for both increased and decreased HR. During the modeling phase, the first author turned on the treadmill and engaged in a 1-min walk at 1 mph, a 1-min sprint at 10 mph, and a 1-min cooldown at 1 mph.

Each interval was completed twice during the training session. Before the conclusion of each interval, the first author asked the participant to verbally identify current HR and whether it was increasing or decreasing, and provided any necessary corrective feedback. Before the modeling phase concluded, the participant was probed for comprehension a second time (“If I run faster, will my heart rate go up or down?”). The participant was required to demonstrate three consecutive correct answers for both increasing and decreasing HR before moving to the rehearsal phase. During the rehearsal phase, the HRM was strapped to the participant, who stood with feet straddling the running belt, and whose HR was displayed on the screen. The first author stood nearby to the left side of the participant. Participants read their resting HR aloud and turned on the treadmill and were instructed to begin walking. A 1-min self-paced walk was followed by the instruction to “raise your heart rate.” The participant was then instructed to “lower your heart rate.” Additional instructions to run faster, decrease speed, and so forth were not provided. The participant was required to correctly respond three times to each instruction before moving to the treatment phase. Responses were considered accurate when speed was (a) increased upon instruction to increase HR and (b) lowered upon instruction to lower HR. Each participant required only three probes per instruction within one training session to meet competency.

Session Overview and Data Collection Methodology

A script was read aloud to the participant prior to beginning each session. This document outlined the overall structure of each exercise session and provided a verbal instruction of target HR to the participant, if applicable. Scripts are included in Appendices 1 and 2. Participants exercised on the treadmill for a total of 30 min during each session. Baseline consisted of three 10-min blocks of self-paced treadmill exercise. A blank Post-it note was secured adjacent to the HR display on the treadmill screen. HR was recorded during the final 2 min of each block using 15-s momentary time sampling procedures, as outlined in the data collection section. During HIIT, each session consisted of three 10-min blocks separated into three distinct intervals: a 6-min self-paced interval, a 2-min warm-up interval to raise HR to criterion, and a 2-min high-intensity interval. HR was recorded only during the 2-min high-intensity interval using a 15-s momentary time sampling procedure. The 2-min high-intensity interval began the moment target HR was reached. If target HR was reached before the end of the 2-min warm-up period, the high-intensity interval began early. In this scenario, remaining session time was added to the end of the 10-min training block as self-paced exercise. If target HR was not reached during the 2-min warm-up interval, data were collected from the 8-min to the 10-min mark of the block using the same 15-s momentary time sampling procedure. HR data were averaged and plotted for each high-intensity interval.

During baseline and HIIT sessions, verbal feedback (“Nice job exercising!”) was provided every 30 s regarding current HR and time remaining in the 2-min interval. A performance chart was secured on the wall directly in front of the participant. This display included the date and three blank spots for each 10-min block completed during each session. Immediately following each 10-min block, the first author verbally informed the participant if he or she met his or her goal and placed a token (star sticker) under the corresponding block. If the participant did not walk for the entire block, the first author drew an “x” through the corresponding block number displayed on the performance chart. These tokens were used to bridge the gap between the completion of each training block and the lottery drawing as discussed in the following section.

Lottery Procedures

One token was provided contingent on either (a) completion of the entire 10-min block regardless of HR in the baseline condition or (b) reaching and maintaining target HR for the entire duration of the 2-min high-intensity interval. This arrangement allotted up to three tokens per session (three 10-min intervals), which were exchanged for a lottery slip at a one-to-one ratio immediately following the session. Lottery slips were drawn from a large cardboard box containing slips of paper printed with names of preferred items or activities in a manner similar to that of Washington et al. (2014); 25% could be exchanged for a food or drink item, 25% could be exchanged for 10 min of one-on-one time with the first author, and 50% read “No Reward.” Lottery winnings were created based on the preference assessment conducted before study implementation and included 100-calorie snack items, sugar-free Gatorade, protein shakes, and 10 min of one-on-one time, with the lottery slips for edibles exchanged immediately. Interaction with the first author took place either immediately following exchange or was scheduled for a time later the same day.

Results

Heart Rate

Average HRs during 2-min high-intensity intervals are included for all three participants in Fig. 1. Each data point represents one full 10-min block, but HR was recorded only during the 2-min high-intensity intervals. Data for all participants show a consistent grouping pattern around each criterion, demonstrating experimental control. Figure 2 represents the average HR for all participants during baseline and in their final criterion during intervention. Alex raised his average HR during treadmill exercise from 97 BPM during baseline to an average of 134 BPM during the final intervention criterion, representing a 38% increase. Laura raised her average HR during treadmill exercise from 100 BPM during the baseline condition to 152 BPM in the final intervention criterion, representing a 52% increase. Carla raised her average HR during treadmill exercise from 118 BPM during the baseline condition to 162 BPM during the final intervention criterion, representing a 37% increase. Table 3 represents range, mean, and standard deviation for each participant during each criterion. Means for each criterion were calculated by adding the average HR per block and dividing by the total number of blocks per criterion.

Fig. 1.

Average heart rate during high-intensity interval training for each participant. Vertical lines depict changes in target heart rate. Horizontal lines depict the target heart rate for each criterion

Fig. 2.

Average heart rate at baseline and during the final condition for each participant

Table 3.

Range, mean, and standard deviation for each participant during each criterion.

Participant	Phase	Criteria	Range	Mean	Standard Deviation
Alex	Baseline		90–105	96.8	4.78
	1	110	104.5–116.5	111.4	4.01
	2	115	84–124	113.6	16.86
	3	125	130–134	132.2	2.04
	4	130	108.5–139	128.1	12.02
	5	110	115–117	116.3	0.95
	6	137	108–144	133.8	13.41
Laura	Baseline		90–105	99.5	5.11
	1	115	127–148	135.7	7.1
	2	130	113–146	135.1	9.97
	3	144	150–155	152.4	1.59
Carla	Baseline		107–131	117.5	7.45
	1	130	136–154	146	6.03
	2	145	147–159	153.6	4.43
	3	156	126–174	162	11.05

Secondary Outcome

The results for Alex’s weight are included in Fig. 3. Each data point represents a separate session. Phase changes are included and correspond with HR criteria. During the duration of his 9-week participation, Alex lost a total of 10.8 pounds, representing about 1.5 pounds lost per week; 4.8 pounds were lost during baseline. He lost an additional 0.5 pounds during the first criterion. He lost an additional 2.2 pounds following the first session in the second criterion but gained 1 pound following the next session. He lost 1 pound following the first session in the third criterion and then gained 3 pounds following the first session of the fourth criterion. By the end of this condition, Alex weighed 213.4 pounds, a loss of almost four pounds within this criterion. When weighed following the one session in the fifth criterion, he had lost an additional 2 pounds. Another pound was lost when he was weighed following Session 1 of the sixth criterion. His weight then stabilized at just over 210 pounds for the final three sessions, representing a total weight loss of 10.8 pounds over the 9 weeks.

Fig. 3.

Weight following sessions for Alex

Discussion

This study builds upon published research regarding exercise programs for individuals with DD by introducing an effective treatment protocol that can be easily and affordably implemented in applied settings. Higher obesity rates and associated secondary health problems are observed in individuals with DD relative to their nondisabled peers (CDC, 2019). Behavior-based interventions promoting exercise in participants with DD have been well established (e.g., Allen & Iwata 1980). HIIT at target average HRs of 74.4% of HRMax (Ramos et al., 2015) has been shown to improve physiological improvements in shorter time frames (Gibala & McGee, 2008) and results in higher adherence rates compared to steady-state cardio (Shiraev & Barclay, 2012). Evidence demonstrates that exercise intensity (De Luca & Holborn, 1985, 1990, 1992) and HR (Eckard et al., 2019) can be impacted using differing schedules of reinforcement, but HIIT had not been targeted by structured reinforcement systems.

In a procedure similar to that of Washington et al. (2014), HR was shaped to levels within the prescribed range during intervention. Each of the participants exercised with HRs below the average intensity recommended by Ramos et al. (2015) at baseline. Over the 9-week procedure, Alex raised average HR during treadmill exercise from 51% (96.8 BPM) to 71% (133.8 BPM) of HRMax, representing a 38% increase. Laura raised average HR during treadmill exercise from 55% (99.5 BPM) to 85% (152.4 BPM) of HRMax, representing a 52% increase. Carla raised average HR during treadmill exercise from 60% (117.5 BPM) to 83% (162 BPM) of HRMax, representing a 37% increase. Alex met or exceeded target HR in 85% of training blocks, Laura met or exceeded target HR in 95% of training blocks, and Carla met or exceeded target HR in 92% of training blocks. The significant increases in HR during exercise and the tight adherence to the criteria set in each condition provide strong evidence of socially significant change and experimental control.

These data demonstrate that a lottery reinforcement system can shape target HR to criteria during HIIT in participants with DD. It is important to note that the individuals in this study were in an ISL facility for high-functioning adults with DD, as defined by the ISL facility’s admission criteria. Participants were able to increase speed and corresponding HR during the first training session. Although it is promising that individuals were able to demonstrate these skills with minimal instruction, tailoring the training approach to the participants is crucial. Replication of this protocol in populations with different skill sets, including typically developing children and in large-group randomized control trials, would further assess more widespread application. A single experimenter trained participants and implemented procedures throughout the study. The use of additional staff to implement this procedure would have improved its strength as an applied intervention. Procedural integrity measures should have been included to ensure that treatment was implemented consistently during sessions. Regardless, continuing with a competency-based BST model is recommended for the training of additional staff and future participants.

Estimates of target HRs have been well established (American Heart Association, 2015) but do not account for individual differences that could be relevant for identifying target HRs. Measuring resting HR prior to study implementation would allow for a more precise determination of target HR (WHO, 2010). Additionally, Jensen, Suadicani, Hein, and Gyntelberg (2013) confirmed that resting HR is a key variable used in assessing physical fitness and mortality risks. A 15-s momentary time sampling technique was used to minimize overestimates and underestimates inherent in interval sampling procedures. Although discontinuous measurements inherently introduce sampling errors, significant HR changes would unlikely be isolated within a 15-s interval. All three individuals exercised with baseline HRs below the threshold recommended by Ramos et al. (2015). Establishing an accurate baseline measure is particularly important in interventions targeting exercise behavior. During early baseline intervals, HR was relatively variable for all three participants. When replicating this procedure, it is equally important to ensure that baseline measures stabilize at a representative rate. If baseline data were above a true baseline exercise intensity, they would cause initial targets to be higher than what the individual is accustomed to and require greater response effort in initial treatment sessions. Increasing exercise intensity too quickly could superadd risks of muscle strains, bursitis, stress reactions, or fractures and reduce compliance even if injuries were not sustained. Two types of baseline measurements would have strengthened the experimental control in this study: (a) a baseline without visual feedback and (b) an instruction-only baseline. As HR had never been used as part of an intervention package with these participants, feedback provided by the HRM likely influenced their exercise speed, and thus HR. This reactivity effect lasted only a few sessions, allowing intervention to take place relatively quickly. An instruction-only baseline, in which the individual is instructed to raise HR to criteria without lottery incentives, would help isolate the functional properties of lottery systems from rule-following contingencies that may have impacted participants in this study.

The variable reinforcement schedule created by lottery systems offers four main benefits to health promotion when compared to more continuous reinforcement systems. First, cost can be an important factor when evaluating the potential utility of an intervention in an applied setting. The use of a lottery system maximizes available reinforcers by providing them on an intermittent schedule. Second, variable reinforcement schedules are more resistant to extinction than continuous schedules of reinforcement (Ferster & Skinner, 1957). The long-term maintenance of exercise behavior is especially important given that ongoing programmed reinforcement may not be available. Third, the variable schedule created by the lottery system minimized additional calories consumed as a result of goal attainment. Edible items were rated as the most highly preferred by all participants during a preference assessment survey completed prior to baseline implementation and were thus used throughout this study. Although efforts were made to ensure that edible items available were low calorie, using an intermittent reinforcement schedule was an effective way to limit the consumption of additional calories. Further efforts should include parametric analyses of reinforcement-schedule density on exercise and nutrition behaviors. For example, in this study lottery slips were provided on a continuous schedule, and secondary reinforcers were provided at a rate of 50%. Both the schedule of lottery slip provision and exchange rates could be adjusted while evaluating the effects on target behavior. Fourth, although social validity was not directly assessed in this study, lottery reinforcement systems that promote exercise in participants with DD have been shown to have high levels of social validity (Li et al., 2018). Future research should include social validity questionnaires for HIIT procedures in participants with DD.

The experimental design would have been strengthened by including reversals for Laura and Carla and by varying criterion changes, similar to the design used with Alex. A previous token economy targeting nutrition and exercise frequency was implemented with all three participants prior to the initiation of this study. Each participant increased their exercise frequency during token implementation as measured by staff-validated instances of 30-min exercise sessions and pedometer use. It is possible that participants learned to use the treadmill during the implementation of the original token economy or required an increase in exercise frequency prior to targeting exercise intensity. However, no lottery system had been used previously, nor did the token economy target HIIT training or exercise intensity. Future research should study exercise intensity in isolation or counterbalanced with exercise frequency targets.

Carla and Laura completed the HIIT procedures prior to Alex. Although weight loss was observed in both participants, it was not tracked with appropriate experimental rigor. Thus, weight data were available only for Alex throughout the course of this study. Future research should prioritize physiological markers of fitness, including weight, measurements of resting HR, peak oxygen uptake, or blood pressure, in addition to behavior change to improve the external validity of similar programs.

Maintenance and generalization would be improved by programming a system to allow the participant to engage in HIIT independently rather than with a single experimenter. Free-operant behavior would provide a better indicator of the maintenance of this behavior once programmed reinforcement schedules are thinned and/or eliminated. Implementing a self-monitoring approach would increase the generalizability of this procedure to applied settings where staffing ratios limit intensive intervention. Independent transitions would be improved by timer signals between intervals without experimenter involvement as used in natural settings. Eckard et al. (2019) demonstrated that participants’ HR was impacted by the exercise activity in which they engaged. For example, participants demonstrated lower average HRs on stationary bikes and elliptical machines than during exergaming or basketball. Promoting exercise with appropriate intensity across a broader range of exercise procedures or during more highly preferred exercise methods may promote better generalization and long-term outcomes.

Despite these limitations, high accuracy and low variability by all participants within intervention conditions suggest strong experimental control. Results provide initial support for the operant control of HR (Eckard et al., 2019) in participants with DD using a modified lottery system. Moreover, this study was the first to demonstrate that HR, using both valid and accessible measures, can be impacted by operant reinforcement in participants with DD during HIIT.

Appendices

Baseline Script

“I would like you to use the treadmill for 10 minutes at whatever speed you choose. I will be watching and occasionally writing things down. If you use the treadmill for the entire 10-minute period, you will receive a token, which can be used to purchase one of the lottery tickets in our prize box. Please begin whenever you are ready.”

Treatment Script

“Today, your target heart rate is [target]. For the first 6 minutes, you can walk at whatever speed you choose. I will then ask you to raise your heart rate to [target]. You will have 2 minutes to get your heart rate to [target]. Once you reach your target, you must keep your heart rate at [target] for the next 2 minutes in order to earn a token. This token can be used to purchase one of the lottery tickets from our prize box. Please begin whenever you are ready.”

Compliance with Ethical Standards

Conflict of Interest

All authors declare no conflicts of interest.

Ethical Approval

Procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent

Informed consent was obtained from all individual participants included in the study.