Abstract
This study evaluated the effects of continuous and discontinuous work-reinforcer schedule arrangements on skill acquisition for three students with autism. Participants were initially exposed to both schedules in an alternating schedules condition where they were taught different but equivalent skills for each schedule. In the discontinuous schedule condition, participants completed work in small increments to gain access to a reinforcer for short periods of time. In the continuous schedule condition, participants completed larger increments of work to gain longer access to a reinforcer. Results showed that two participants mastered the target responses with both schedules and the third participant only met mastery criterion with the continuous schedule. Preference for schedules varied across participants. Session duration was consistently shorter during the continuous work-reinforcer schedule, suggesting that continuous work-reinforcer schedules are more efficient. Participants engaged with the reinforcer less when provided longer access, suggesting that reinforcer access might be reduced with continuous schedules for further efficiency gains.
Applied behavior analysis (ABA) has a rich history of optimizing instruction for children with developmental disabilities by manipulating antecedent and consequence variables. Recently, researchers have studied how manipulations of the distribution of responses and access to reinforcers within a session affect behavior. Researchers have examined two types of work-reinforcer arrangements, or schedules. Continuous schedules involve providing access to large reinforcers after large response requirements. Discontinuous schedules involve providing access to small reinforcers after smaller response requirements. In these studies, researchers typically controlled for the total number of responses required across arrangements as well as the total amount of reinforcement during a session. For instance, Fienup, Ahlers, and Pace (2011) provided a 16-year-old female participant with a brain injury with two work-reinforcer schedules and a control condition. In the continuous schedule, the participant was required to complete six 20-problem math worksheets to receive her preferred activity for 18 consecutive minutes. In the discontinuous schedule, the participant had to complete one 20-problem worksheet to gain access to her preferred activity for a shorter amount of time (3 min) and this was repeated six times during a session. In the control condition, she was required to complete all six worksheets consecutively and then the session ended. Fienup et al. found that the participant preferred to complete sessions according to the continuous work-reinforcer schedule.
A number of follow-up studies have replicated and extended research on work-reinforcer arrangements while examining preference. Ward-Horner, Pittenger, Pace, and Fienup (2014) varied the magnitude (duration) of reinforcement associated with the continuous schedule, while keeping the reinforcer magnitude constant in the discontinuous schedule. The authors found that this participant preferred the continuous schedule when it had 80% and 100% of the total magnitude of the discontinuous schedule, though, when the magnitude was dropped to 60% and 20%, the participant preferred the discontinuous schedule. Overall, the research has found that participants with developmental disabilities aged 13 to 20 years prefer continuous work-reinforcer arrangements when the reinforcer is an activity (Bukala, Hu, Lee, Ward-Horner, & Fienup, 2015; DeLeon et al., 2014; Fienup et al., 2011; Ward-Horner et al., 2014) or edible (three of four participants, DeLeon et al., 2014). In other words, these participants preferred completing all work requirements consecutively followed by longer access to a preferred stimulus.
A few studies have investigated the effects of work-schedules on performance. Bukala and colleagues (2015) measured session duration, work/task duration, and transition duration. The authors found that sessions were completed faster with continuous schedules for two of three participants. The continuous schedules resulted in faster task completion for one participant and resulted in decreased transition time for the second participant. DeLeon et al. (2014) evaluated performance on free operant and discrete trials tasks, and participants earned tokens for correct responses. DeLeon et al. termed the continuous condition accumulated because participants exchanged an accumulated number of tokens after several responses were emitted and at the end of a session. They termed the discontinuous condition distributed because participants exchanged tokens after each correct response. The authors found that three of three participants had a higher response rate during free operant accumulated conditions, four of four participants demonstrated less time to complete discrete trials tasks in the accumulated condition when the reinforcer was an activity, and two of four participants completed accumulated condition discrete trials tasks in less time when the reinforcer was an edible.
The positive effects of continuous, or accumulated, schedules have been observed with preference (Bukala et al., 2015; DeLeon et al., 2014; Fienup et al., 2011; Ward-Horner et al., 2014) and performance (Bukala et al., 2015; DeLeon et al., 2014). The research to date has focused on academic responding to tasks that are best characterized as maintenance tasks or tasks that were previously mastered. These are tasks that require little to no teaching for the participant to respond correctly to trials. With continuous/accumulated schedules, there is a greater delay between behavior and access to reinforcers, which according to basic principles, may be appropriate for behaviors in maintenance that have thinner reinforcement schedules (Mayer, Sulzer-Azaroff, & Wallace, 2012; Skinner, 1968). Skill acquisition, however, requires denser schedules of reinforcement, and delays between responding and reinforcement may impair learning. Thus, discontinuous/distributed schedules may be favorable for skill acquisition.
The purpose of the current study was to replicate and extend previous research on the effects of work-reinforcer schedules by examining academic behaviors in skill acquisition, as well as preference of work-reinforcer schedules when provided a choice. In addition, the study examined the effects of the respective arrangements on session duration to evaluate efficiency and reinforcer engagement to evaluate whether participants engaged with brief and long durations of reinforcement similarly.
Method
Participants and Setting
Three male students, aged 16 to 17 years, participated in this study. All participants attended a private behavioral school for children with autism and received one-to-one instruction throughout the school day. At the time of the study, 13 students were enrolled at the school and either 4 or 5 students were assigned to one of three classrooms. Chris and Lance were in the same classroom, and Alex was assigned to another classroom. All participants had a diagnosis of autism, could attend to instructor directions, and had a history of receiving discrete trial instruction. Alex was a nonverbal 17-year-old who used Proloquo2Go on an iPad® to communicate. Chris was a 17-year-old who communicated verbally using short phrases. Lance was 16 years old, had limited verbal communication skills, and used Proloquo2Go on an iPad® to communicate. Alex and Lance gestured to indicate preferred activities whereas Chris vocally requested preferred activities.
Three instructional staff, including the first author, implemented the instructional sessions with the participants. Staff were responsible for daily implementation of behavioral and academic programming. The instructors were female, had an average age of 27 years (range = 26-28), held at least a bachelor’s degree, and had worked at the school for 10 to 44 months (M = 23). The first author was enrolled in a master’s level graduate program in ABA during this study. One of the other instructors held a master’s degree and was enrolled in an advanced certificate program in ABA.
All sessions took place at school. Alex’s first teaching session took place in his classroom, and his remaining sessions were conducted in various rooms of the school depending on the instructions he was learning, including the exercise room, the kitchen, and the gym. The first teaching sessions for Chris took place in his classroom and all remaining sessions were conducted in the school’s gym, due to his preferred activity (riding a scooter). The gym was a large room that stored the school’s sports equipment and contained three cafeteria-style tables. Lance’s sessions were held in his classroom, which was an open-bay room divided into five work sections, one for each student assigned to the room.
Materials
Materials for all participants included color cards for a color preference assessment, choice cards for a stimulus preference assessment, and schedule cards for the alternating schedules and free choice phases (phases described below). Color cards were 5.08 by 2.54 cm pieces of construction paper in different colors. Choice cards were 2.54 by 1.27 cm and represented preferred activities during the preference assessment either by picture (Alex) or in text (Chris and Lance). Schedule cards were 20.32 by 27.94 cm and displayed the work-reinforcer schedules by illustrating the order of work and preferred activities. The continuous schedule was represented with one white rectangle divided into 25 boxes (Alex and Lance) or 50 boxes (Chris) and one large picture of the preferred activity. The discontinuous schedule was represented by five adjacent white columns divided into 5 boxes (Alex and Lance) or 10 boxes (Chris) followed by a smaller picture of the preferred activity. Additional materials included a binder to store data sheets and teaching materials, a pencil, a dry-erase marker, two timers, and a Flip MinoHD Video Camera® and tripod to record sessions.
Various materials were used for each participant’s skill acquisition programming. These included a portable DVD player, a television, a microwave, a sleeve of plastic cups, and Clorox® wipes for Alex’s receptive instruction objective; Post-it® notes to provide textual prompts for Chris’s spelling objective; and printed words posted on 10.16 by 12.7 cm index cards for Lance’s receptive word identification objective. Last, a scooter, helmet, iPod Touch® with headphones, and iPod Nano® with headphones served as the preferred activities for the participants.
Dependent Variables and Data Collection
The first dependent variable was the percentage of correct responses. A trial was scored as correct when the participant responded correctly and independently to the discriminative stimulus. For all participants, responses were scored as prompted if the instructor provided an additional antecedent stimulus that evoked the desired response. Responses were scored as incorrect if the participant did not respond within 3 s of the presented discriminative stimulus or if he provided a response different from the target response. Instructors collected data on a data sheet by circling whether a trial was correct, prompted, or incorrect. The number of correct trials was divided by the total number of trials and then multiplied by 100.
The second dependent variable was session duration. This variable was defined as the number of minutes from the beginning to the end of a session. The instructor collected session duration data by starting a timer at the onset of presenting the schedule and stopping the timer at the end of the final reinforcer consumption period.
The third dependent variable was the percentage of intervals with reinforcer engagement. Data were collected using 30 s whole interval recording. If the participant was engaged with his reinforcer for the entire 30 s interval, the instructor recorded that engagement occurred during that interval. The number of intervals with engagement was divided by 10, which was the total number of intervals, and then multiplied by 100.
The fourth dependent variable was work-reinforcer schedule preference. Data were recorded on the participant’s selection of the discontinuous schedule or continuous schedule during the free choice phase.
Procedure
The researchers identified skill acquisition programs that were part of each participant’s individualized education plan. Alex’s program was following one-step directions, during which he responded to a specific direction. Chris’s program was spelling, during which he responded to the direction, “Spell ______,” by spelling the word aloud. Lance’s program was receptive identification of written words. In response to the direction, “Point to _____,” Lance pointed to the corresponding word in a field of three words.
The study involved four phases. The first phase included two preference assessments to determine each participant’s lowest preferred color and highest preferred activity. The second phase consisted of a baseline in which a number of potential targets for all three participants were probed to determine unknown targets. The third and fourth phases involved the teaching of the skill acquisition programs. The third phase involved alternating the presentation of schedules, in which the participant completed both work-reinforcer schedules in a randomized order each day. The fourth phase involved a free choice phase, in which each participant chose which work-reinforcer schedule he wanted to complete.
Preference assessment
Prior to teaching, the researchers conducted a multiple stimulus without replacement preference assessment (MSWO; DeLeon & Iwata, 1996) to identify two non-preferred colors for each participant. The purpose of identifying two non-preferred colors was to use the colors to make the work-reinforcer schedules more salient with color, but not use colors that would affect preference. For each participant, the preference assessment was conducted three times during three separate sessions (separated by at least 1 hr). During each session, the researcher presented a participant with five to seven small pieces of colored construction paper and instructed him to choose a color. This continued until all the colors were selected, and after the three sessions, the researchers determined the average rank order for each color and the highest ranking colors were determined to be the least preferred colors.
A second MSWO preference assessment was conducted to determine a preferred stimulus from an array of activities and edibles for each participant. This MSWO was also conducted on three separate occasions. During the assessment for all participants, the instructor presented stimuli (pictures for Alex, written words for Chris and Lance) that represented activities or edibles and provided the instruction to choose a stimulus. All participants received 1 min access to the selected activity or one piece of the chosen edible. This continued until the participant selected all the activities or edibles. After the three sessions for each participant, the researchers determined the average rank order for each stimulus and identified the most preferred stimulus.
Baseline probe and target response identification
During baseline phases, the researchers presented an instruction to the participant, with no prompts or reinforcement, and recorded whether he responded correctly or incorrectly. For a target to be selected for use in the study, the participant had to demonstrate 0% accuracy across three separate sessions. Baseline was repeated before each alternating schedules and free choice phase.
Once target behaviors with 0% accuracy were identified, the instructor randomly assigned targets to the continuous and discontinuous work-reinforcer schedules. Table 1 displays the target responses that were assigned to each participant for the respective conditions. Based on Alex’s history of slower skill acquisition relative to the other two participants, only one target response was assigned to each schedule in a given phase, whereas Chris and Lance were taught three target responses per schedule and phase. For all participants during the free choice phase, the same target responses were taught regardless of the participant’s work-reinforcer schedule choice.
Target Behaviors.
Note. Conditions 1 and 2 were alternating schedules phases and Choice represents the free choice condition.
Alternating schedules phase(s)
The effects of work-reinforcer schedules were evaluated in an adapted alternating treatments design (Sindelar, Rosenberg, & Wilson, 1985). During each session, instructors exposed the participants to both work-reinforcer schedules in a randomized order, generated by www.randomizer.org. Only one session occurred in a day, and sessions occurred up to 3 days per week.
The continuous and discontinuous schedules both involved earning the same total number of check marks and 5 total minutes of access to preferred activities during a given session. The contingencies for earning a check mark were identical across work-reinforcer schedules. The difference was how the responses and reinforcers were distributed in the session: The continuous schedule required a participant to earn the total amount of check marks to gain access to the full duration of a preferred activity, whereas the discontinuous schedule required a participant to earn a portion of the total amount of check marks to gain access to a shorter duration of a preferred activity. With the discontinuous schedule, this sequence was repeated until all check marks were earned and the total duration of preferred activity was consumed.
The work-reinforcer schedules included boxes representing the number of check marks needed to access reinforcement. In accordance with the teaching philosophy of the school, target responses were initially prompted to promote errorless learning. As trials were repeatedly presented, prompts were faded (i.e., reduced in magnitude), which allowed the opportunity for independent responding. To reduce the likelihood of prompt dependency and to promote independent responding, correct responses were differentially reinforced. If a participant responded correctly and independently, he received two check marks, whereas, a prompted response resulted in one check mark. The instructor delivered verbal praise contingent on prompted and independent responses as well as appropriate attending behavior. If the participant responded incorrectly, the instructor provided feedback by indicating the correct answer and no check mark was provided. To avoid repeated errors, incorrect trials were followed by a prompted trial.
During the continuous work-reinforcer schedule, the instructor presented the visual stimulus card depicting the continuous schedule to the participant and explained it to the participant. When the participant earned the predetermined number of check marks, he gained access to a long duration of preferred activity. Alex received 5 min on his iPod® contingent on earning 25 check marks (FR25/5-min), Chris received access to 5 min of scooter contingent on earning 50 check marks (FR50/5-min), and Lance received access to 5 min on his iPod Touch® contingent on earning 25 check marks (FR25/5-min).
During the discontinuous work-reinforcer schedule, the instructor presented the visual stimulus card depicting the discontinuous schedule to the participant and explained it to the participant. When the participant earned one fifth of the total check marks, he gained access to his preferred activity for one fifth of the total time. Alex received 1 min on his iPod® contingent on earning 5 check marks (FR5/1-min), Chris received 1 min of scooter contingent on earning 10 check marks (FR10/1-min), and Lance received 1 min on his iPod Touch® contingent on earning 5 check marks (FR5/1-min). After 1 min access to a preferred activity, the instructor directed the participant back to his instructional area and the sequence repeated until the participant completed five sets and had five 1-min opportunities to engage with his preferred activity.
The alternating schedules phase was completed when the participant met mastery criterion for targets associated with one of the work-reinforcer schedules. Mastery criterion was set at 90% correct responding across three consecutive sessions, which was the standard criterion of the school. After meeting mastery criterion, baseline and the alternating schedules phases were repeated with Chris and Alex.
Generalization probes
Once a set of targets met mastery criterion during the alternating schedules phase, the instructor conducted a generalization probe for all target responses. The one-step directions that Alex learned were assessed in naturally occurring situations (e.g., responding to the direction to open the microwave while he was getting ready for lunch). For Chris, a generalization probe consisted of writing words he had learned to spell aloud. For Lance, the probe consisted of identifying words written on a piece of paper or typed on a computer screen that had been presented on flashcards. For all participants, each trial was marked correct, prompted, or incorrect. A percentage of correct trials was calculated for these generalization probes by dividing the number of correct trials by the total number of trials (range = 3-5 trials per target) and then multiplying by 100.
Free choice phase
During this phase, new target responses were identified (see Table 1) and the instructor allowed the participant to select the work-reinforcer schedule (either discontinuous or continuous) in a concurrent operants arrangement (e.g., DeLeon et al., 2014). The researcher presented both visual stimulus cards and instructed the participant to choose a schedule. The participant indicated which work-reinforcer schedule he wanted to complete by pointing to the respective schedule card or by saying the color associated with a schedule. The positions of the schedule cards were randomized during each session to control for a position bias. The free choice phase continued for each participant until he demonstrated a preference for one work-reinforcer schedule, which was determined by selecting the same schedule for four consecutive sessions.
Interobserver Agreement (IOA) and Treatment Integrity
For the purpose of calculating IOA, an independent observer viewed sessions in vivo or via video recording. For trial accuracy, an agreement occurred when both the observer and the instructor recorded the same response (correct or incorrect) on a trial. IOA was calculated by dividing the number of trials of agreements by the total number of trials and multiplying by 100%. IOA data for trial accuracy were conducted for 38% of sessions during the alternating schedules phase and mean agreement was 98% (range = 93%-100%). For session length, the observer also timed each session, and agreement was calculated by dividing the lower duration by the higher duration and multiplying by 100%. IOA for session length was conducted for 18% of sessions during the alternating schedule phase and the mean agreement was 99% (range = 98%-100%). For reinforcer engagement time, an agreement occurred when both the observer and the instructor recorded engagement or no-engagement during a particular interval. Agreement was calculated the same as with trial accuracy. IOA for the reinforcer engagement was conducted for 27% of sessions during the alternating schedule phase and mean agreement was 97% (range = 90%-100%). During 32% of free choice sessions, the researchers collected IOA data on the agreement on whichever schedule (continuous or discontinuous) the learner selected. Agreement was 100% for all participants.
Treatment integrity data sheets were created for both the alternating schedule and free choice phases. These data sheets listed all procedural steps, including setting up the teaching environment, the error correction procedures, schedule explanation and presentation, recording of session duration, delivering the proper number of checks for each response, and delivering the reinforcement associated with each schedule. During the alternating schedules phases, an independent observer collected treatment integrity data for 38% of sessions and mean integrity was 100% (range = 94%-100%). During the free choice condition, an independent observer collected treatment integrity data for 32% of sessions and integrity 100%.
Results
Alex, Lance, and Chris demonstrated 0% accuracy across three sessions for all target behaviors listed in Table 1, respectively. Baseline probes were conducted prior to alternating schedule and free choice phases. For Alex and Chris, baseline probes were conducted prior to the initial and second alternating schedules phases. These data are not displayed in graphs.
Skill Acquisition
Figure 1 displays percentage of correct responding to the respective work-reinforcer schedules during the alternating schedules phase. During the first phase, Alex (top panel) met mastery criterion (responding correctly for 90% of trials for three consecutive sessions) for the target response taught with the discontinuous schedule (FR5/1-min) in five sessions. The target response associated with the continuous schedule (FR25/5-min) did not meet mastery criterion, but reached a level of 79%. When generalization was probed, Alex responded correctly for 100% of the trials for the target taught with the discontinuous schedule and he responded correctly for 50% of the trials for the target taught with the continuous schedule. During the second alternating schedules condition, Alex met mastery criterion for the target taught with the continuous schedule in six sessions. The target associated with the discontinuous schedule did not meet mastery criterion, but reached a level of 100%. When generalization was probed, Alex responded correctly for 80% of the trials for the target taught with continuous schedule, and he responded correctly for 100% of the trials for the target taught with the discontinuous schedule.
Percentage correct responding during the alternating schedules phases.
During the first phase, Chris (middle panel) met mastery criterion for the target responses taught with the continuous schedule (FR50/5-min) in six sessions. The target responses associated with the discontinuous schedule (FR10/1-min) did not meet mastery criterion but reached a level of 87%. When generalization was probed, Chris responded correctly for 93% of the trials for the targets taught with continuous schedule, and he responded correctly for 90% of the trials for the targets taught with the discontinuous schedule. During the second alternating schedules condition, Chris met mastery criterion for the targets taught with the discontinuous schedule in six sessions. The targets associated with the continuous schedule did not meet mastery criterion but reached a level of 86%. When generalization was probed, Chris responded correctly for 100% of the trials for the target responses taught with discontinuous schedule, and he responded correctly for 44% of the trials for the target responses taught with the continuous schedule.
Lance (bottom panel) experienced one alternating schedules phase. Lance met mastery criterion for the target responses taught with the continuous schedule (FR25/5-min) in 12 sessions. The targets associated with the discontinuous schedule (FR5/1-min) did not meet mastery criterion but reached a level of 53%. When generalization was probed, Lance responded correctly to 80% of the trials for the targets taught with both the continuous schedule and the discontinuous schedule.
Overall, the effects of schedules on sessions to skill acquisition were mixed. There was no consistent advantage of either schedule on Alex and Chris’ skill acquisition. Lance’s performance was higher with the continuous schedule, and generalization outcomes were identical.
Session Duration
Figure 2 displays session duration for the respective work-reinforcer schedules during the alternating schedules phase. During the first alternating schedules phase, Alex’s (top panel) mean session length was 10.5 min (range = 9.7-13.0) for the discontinuous schedule and 9.2 min (range = 8.7-10.7) for the continuous schedule. During the second phase, Alex’s mean session length was 10.7 min (range = 10.1-11.9) for the discontinuous schedules and 8.1 min (range = 7.2-9.2) for the continuous schedule. Across both phases, the continuous schedule sessions were completed in 12.0% and 23.9% fewer minutes than discontinuous schedule sessions. During the first phase, Chris’s (middle panel) mean session length was 13.4 min (range = 11.2-20.5) for the discontinuous schedule and 10.4 min (range = 8.4-14.4) for the continuous schedule. During the second phase, Chris’s mean session length was 11.8 min (range = 11.4-12.6) for the discontinuous schedule and 9.6 min (range = 8.8-10.6) for the continuous schedule. Across both phases, the continuous schedule sessions were completed in 22.2% and 18.8% fewer minutes than discontinuous schedule sessions. Lance’s (bottom panel) mean session length was 9.9 min (range = 9-11.5) for the discontinuous schedule and 8.7 min (range = 7.5-10.6) for the continuous schedule. Lance completed continuous schedule sessions in 11.5% fewer minutes than discontinuous schedule sessions. Overall, the continuous schedule consistently produced shorter session durations (11.5% less-23.9% less). Alex and Chris’s session durations were clearly discriminable. For Lance, there was not only overlap but also a 1-min mean difference and 13.6-min overall difference.
Session duration during the alternating schedules phases.
Reinforcer Engagement
Figure 3 displays the percentage of 30-s intervals during which the participants continually engaged with the preferred activity during reinforcer consumption periods. During the first alternating schedules phase, Alex’s (top panel) mean percentage engagement was 96% (range = 80%-100%) for the discontinuous schedules and 74% (range = 30%-90%) for the continuous schedule. During the second phase, Alex’s mean percentage engagement was 92% (range = 80%-100%) for the discontinuous schedule and 78% (range = 60%-100%) for the continuous schedule. During the first phase, Chris’s (middle panel) mean percentage engagement was 97% (range = 90%-100%) for the discontinuous schedule and 82% (range = 50%-100%) for the continuous schedule. During the second phase, Chris’s mean percentage engagement was 88% (range = 80%-100%) for the discontinuous schedule and 68% (range = 40%-90%) for the continuous schedule. In Sessions 2 and 10, Chris repeatedly asked for a different activity during the continuous schedule. Lance’s (bottom panel) mean percentage engagement was 92% (range = 60%-100%) for the discontinuous schedule and 88% (range = 10%-100%) for the continuous schedule. During Session 6 for both schedules, the Internet connection was intermittently interrupted when the learner was attempting to watch YouTube® videos. Overall, there was considerable overlap in data paths; however, the mean percentage engagement was consistently lower when participants earned access to longer durations of reinforcement (continuous schedule).
Percentage of engagement with reinforcers during the alternating schedules phases.
Schedule Preference
Figure 4 displays the cumulative number of times a participant selected a specific schedule during the free choice phase. Alex demonstrated a preference, which was defined as selecting the same schedule for 4 consecutive sessions, for the discontinuous schedule in 10 sessions and for 70% of these sessions. Additional sessions were not conducted because Alex mastered the target behaviors on the final session that is plotted. Chris demonstrated exclusive preference for the continuous schedule across 4 sessions and mastered the targets on that fourth session. Lance did not demonstrate a consistent preference during 11 sessions; however, he selected the discontinuous schedule during 6 of 11 free choice sessions.
Cumulative number of schedule choices during the free choice phase.
Discussion
The current study extended previous research on work-reinforcer arrangements in important ways. First, this study expands this area of research by examining how work-reinforcer schedules affect skill acquisition, which is an important addition because a goal of ABA is effective and efficient teaching of socially significant repertoires. As discussed in the introduction, basic learning principles regarding the immediacy of reinforcement needed to acquire new skills (Mayer et al., 2012; Skinner, 1968) suggest that discontinuous work-reinforcer schedules may produce better skill acquisition than continuous schedules because discontinuous schedules result in more frequent delivery of primary reinforcers. The results of the current study did not favor discontinuous schedules for skill acquisition. In terms of sessions to reach mastery criterion, Alex and Chris’s skill acquisition were not differentially affected by work-reinforcer schedules and Lance met criterion only with targets associated with the continuous schedule. In terms of session duration, all three participants completed continuous schedule session in an average of 1.1 to 3.0 min (M = 2.1) less than discontinuous schedule sessions, which was between 11.5% and 23.9% less time than discontinuous schedule sessions. Taken together, continuous schedules produced similar (Alex and Chris) or better (Lance) sessions to criterion and more efficient session lengths, thus favoring the use of continuous work-reinforcer schedules for skill acquisition of children with similar characteristics to the participants in this study. Although no studies to date have evaluated the effects of work-reinforcer schedules with skill acquisition tasks, the efficiency data were similar to those found in studies that evaluated previously mastered tasks (Bukala et al., 2015; DeLeon et al., 2014).
The discrepancy between learning principles and the outcomes of this study could be due to a few factors. First, instructors presented conditioned reinforcers (check marks) contingent on correct and prompted responses. This aspect of the study may help children tolerate delays to primary reinforcement (DeLeon et al., 2014). All the current participants had a long history with token economies. In addition, continuous schedules involve increased repetition of trials, which may promote acquisition and fluency. This factor is supported by a recent study that demonstrated that five of six young children with autism (ages 4-5) learned new skills quicker when trials were delivered repeatedly with only a 1-s inter-trial interval (mass condition) as compared with 10-s inter-trial intervals (distributed condition; Majdalany, Wilder, Greif, Mathisen, & Saini, 2014). In the current study, in the discontinuous schedule, there were four 1-min breaks interspersed within the instructional trials, akin to Majdalany et al.’s distributed condition. Indeed, continued research is needed to understand why continuous and discontinuous schedules affect performance in the ways observed. The specific characteristics of schedules should be examined under different conditions, including non-token economy conditions, such as those reported in similar studies (e.g., Bukala et al., 2015).
The efficiency advantage of continuous schedules is important and has been replicated across the current study and two others (Bukala et al., 2015; DeLeon et al., 2014). Efficiency is important because the time saved by conducting continuous schedule sessions could result in more skill acquisition opportunities. DeLeon et al. found that participants responded faster during accumulated/continuous schedules. Bukala et al. found that one participant completed the work portion of the schedule faster with continuous schedules and that another participant has less transition time between work and reinforcer consumption. The current study only measured session duration at a broad level and did not break down duration into work and transition components. However, anecdotally, the two additional instructional staff noted that discontinuous schedules were more difficult to implement because of the required effort. That is, discontinuous schedules impose a number of transitions on teachers who have to excuse a child to consume the reinforcer, bring the child back to the instructional setting, re-orient him or her to the instructional materials, re-establish attending, and deliver the next trials. With continuous schedules, there is only one transition from instruction to reinforcer consumption before the end of the session. DeLeon et al. posited handling costs as a reason why a learner may not prefer discontinuous schedules—that the behavior of re-orienting to the reinforcer several times may diminish the value of that reinforcer. In fact, there may be handling costs for instructors, too, who are presenting instruction according to discontinuous schedules.
The current study expanded on previous research by measuring reinforcer engagement. This was important to measure because it provided insight into how the participants responded during different durations of reinforcer consumption. Ward-Horner et al. (2014) found that a participant preferred the continuous condition when the total duration of the reinforcer associated with the continuous schedule was 80% of the reinforcer associated with the discontinuous schedule. In the current study, we controlled for the total duration of reinforcer consumption and found that participants engaged with the larger duration reinforcer for fewer intervals than when smaller reinforcers were delivered. The participants engaged with the reinforcer during the discontinuous schedule (five 1-min consumption periods) during the entire interval between 88% and 97% of the recorded intervals but for only 68% to 88% of the intervals during the continuous schedule (one 5-min consumption period). One possible explanation for the lower reinforcer engagement during the continuous schedule is that the participants became satiated on the activities and engaged with them less when provided access for longer periods of time. This explanation may be particularly plausible for Chris because his activity-reinforcer was physically demanding. His engagement during the continuous schedule may have been lower because longer durations of riding the scooter were not as reinforcing as short bursts of physical activity. Although none of the participants in this study earned edibles for responding, it is also possible that the effects of large amounts of edible consumption could parallel the effects found with physical activity and result in an abolishing operation faster in a continuous schedule than in a discontinuous schedule.
Lower engagement during the continuous schedule for all participants may suggest that access to reinforcement for the full duration (5 min) may not be necessary which is supported by the findings of Ward-Horner et al. (2014). Future research should examine how skill acquisition would be affected if the magnitude of reinforcement during the continuous schedule was decreased. As long as a decrease in magnitude does not undesirably affect skill acquisition, the continuous schedule could be made even more efficient by reducing the total duration of access to reinforcement.
The current study expanded this area of research by demonstrating that work-reinforcer schedule preference is unique to the individual and may be more variable during skill acquisition. In the current study, Alex preferred the discontinuous schedule, Chris preferred the continuous schedule, and Lance did not demonstrate a consistent preference. These results differ from the results from previous research. Fienup et al. (2011), DeLeon et al. (2014), Ward-Horner et al. (2014), and Bukala et al. (2015) all demonstrated that the participants preferred the continuous schedule when the reinforcer was an activity. As there have been only a few studies that have investigated variables that influence preference (type of reinforcer, DeLeon et al., 2014; magnitude of reinforcer, Ward-Horner et al., 2014), it is unknown why there was variability in preference for work-reinforcer schedules. The schedules vary in both the arrangement of trials, frequency of reinforcement, and magnitude of reinforcement. Future research should examine which aspects of the work-reinforcer schedules influence preference (or lack thereof) for both skill acquisition and maintenance responses.
Aspects of this current study warrant further investigation. The study was conducted in a 1:1 teaching ratio at a behavioral school for children with autism. As such, the generality of these results to more common instructional settings, such as a classroom with a group of students and a single teacher, is unknown. In addition, although the work-reinforcer schedules operated under different contingencies regarding access to the preferred activity, both schedules maintained identical contingencies relative to token delivery (i.e., 2 checks for a correct response, 1 check for a prompted response, 0 checks for an incorrect response). The effects of skill acquisition without a visual, salient, and contingent schedule of reinforcement, is unknown. Finally, these participants had a long history of reinforcement with token economies and low levels of problem behavior related to the delay of a preferred item or activity. It is possible that discontinuous schedules are necessary for young learners who are learning to tolerate delays to preferred items and activities. Further research would be necessary to evaluate the effects of work-reinforcer schedules with learners who have less experience with token economies.
The current research suggests that work-reinforcer schedules have idiosyncratic effects on skill acquisition. These results indicate the importance of conducting similar assessments to determine the effect on an individual’s performance. Such an analysis could lead to an increased number of target behaviors that are mastered if practitioners identify schedule arrangements that lead to fewer sessions to mastery or sessions that require less time to complete. The two schedules examined in this study represent two points on a continuum that could also include many other points. Reinforcers can be delivered contingent on each response or after many responses and the reinforcers can be of varying magnitudes. The point at which reinforcers are delivered affect performance on both acquisition and maintenance tasks. If reinforcers are delivered too often, a number of handling costs are imposed on a session that can lengthen sessions and result in a lower rate of trials (Bukala et al., 2015; DeLeon et al., 2014). However, if too many responses are required for access to a reinforcer, a learner may experience ratio strain. Future research is needed to uncover points along the continuum that can serve as a starting point for practitioners to begin an analysis of how continuous and discontinuous schedules affect the learners they work with.
Footnotes
Acknowledgements
We thank Deb Thivierge and the staff at the ELIJA School.
Authors’ Note
This study was conducted by the first author in partial fulfillment of a master’s degree in applied behavior analysis at Queens College of the City University of New York.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.