2018, Vol. 42(1) 9â€“33
Â© The Author(s) 2017
Reprints and permissions:
Special Issue Article
Leisure Engagement for
Children With Autism
Using Backward Chaining
Chelsea K. Edwards1, Robin K. Landa1,2,
Sarah E. Frampton1, and M. Alice Shillingsburg1,3
Research with individuals with disabilities has demonstrated the utility of
intervention approaches to address toy play, also referred to as functional
leisure engagement (FLE). Examples include prompting FLE, blocking
stereotypy, and differentially reinforcing appropriate FLE with social or
automatic (i.e., access to stereotypy) reinforcers. Backward chaining has
yet to be evaluated, but may be useful for establishing more complex
FLE. The current study employed a treatment package consisting of these
components with three school-aged children with autism in a therapeutic
classroom. Effects were evaluated during pretest and posttest sessions,
which consisted of free access to toys in a novel setting. The percentage
of session with FLE was evaluated using a multiple probe design across
participants. Results showed all participants demonstrated an increase in FLE
and two participants showed decreased stereotypy. Feasibility for classroom
implementation is discussed.
autism spectrum disorder, backward chaining, functional engagement,
functional play, toy play
1Marcus Autism Center, Atlanta, GA, USA
2Western New England University, Springfield, MA, USA
3Emory University, Atlanta, GA, USA
M. Alice Shillingsburg, Marcus Autism Center, 1920 Briarcliff Road NE, Atlanta, GA 30329, USA.
Email: [email protected]
699929 BMOXXX10.1177/0145445517699929Behavior ModificationEdwards et al.
10 Behavior Modification 42(1)
The development of play skills is crucial for children diagnosed with autism
spectrum disorder (ASD), as it functions as a platform for the development of
social interaction and social communication skills (Bloom, 1993; Piaget,
1952). Although exploratory engagement with items is seen in infancy, functional toy play typically begins to emerge later in the first year of life. This
higher level form of play has been delineated from its earliest precursors by
being characterized as playing with toys â€œaccording to their intended functionâ€ (Baron-Cohen, 1987, p. 142). Demonstrations of functional play skills
earlier in life have been correlated with the later development of language
with typically developing children (Bates, Benigni, Bretherton, Camaioni, &
Volterra, 1979; Ungerer & Sigman, 1984) and children with ASD (Sigman &
Ruskin, 1999; Toth, Munson, Meltzoff, & Dawson, 2006). However, children
with ASD engage in fewer spontaneous appropriate play acts than their typically developing peers (Lewis & Boucher, 1988; Wing, Gould, Yeates, &
Brierley, 1977). Sigafoos, Roberts-Pennell, and Graves (1999) found that
play skills for preschoolers with ASD showed little growth over a 3-year
period in comparison with the development of their other adaptive skills. This
deficit in play is of significant concern, as these failures to play may then
exacerbate failures to develop social communication skills. Thus, interventions that support the development of play skills are of critical importance for
learners with ASD (Jarrold, Boucher, & Smith, 1996).
Although a wide variety of interventions have been proposed to teach toy
play to children with ASD (see Lang et al., 2009; Luckett, Bundy, & Roberts,
2007; Stahmer, Ingersoll, & Carter, 2003), many, especially highly structured, behavior-analytic procedures, have been described as failing to produce genuine play, in part due to the reliance on external rather than â€œinternalâ€
rewards and limited spontaneity, flexibility, and generalization of targeted
play behaviors (Luckett et al., 2007). This criticism highlights several key
features of play that researchers and clinicians have emphasized including
that play is a behavior that is â€œfreely chosen, personally directed, and intrinsically motivatedâ€ (Hughes, 2012). Thus, instead of being highly structured
and adult directed, activities should be selected and initiated by the child.
Only interventions that evaluate play in an unstructured context, in which the
individual is free to engage with the toys or in other behaviors, would meet
this criteria (e.g., Eason, White, & Newsom, 1982; Nuzzolo-Gomez, Leonard,
Ortiz, Rivera, & Greer, 2002). Furthermore, play should not be motivated
exclusively by contingent reinforcers such as adult attention or other tangible
reinforcers. Thus, the intervention, including any contingent reinforcers,
must be fully withdrawn for this criteria to be met (e.g., Eason et al., 1982;
Kasari, Gulsrud, Wong, Kwon, & Locke, 2010; Nuzzolo-Gomez et al., 2002).
These characteristics of an intervention not only increase the acceptability of
Edwards et al. 11
outcomes, but also make it more sustainable in applied settings. For example,
a teacher serving 6 to 10 children with ASD will find it difficult to deliver
carefully scheduled tangible reinforcers during common free play contexts,
in the midst of other responsibilities.
Another key feature of play to consider is the importance of generalization. Play outcomes should also have generality across toys and settings.
Without these measures, there is little to suggest that the play intervention
will have effects beyond the intervention setting itself. This is crucial when
services are delivered outside of the childâ€™s home (i.e., in a treatment clinic or
school), because play is an ongoing, daily activity for children. Last, in addition to the quantity of play behaviors, the quality of play should be emphasized. Several studies have included dependent measures such as simple
object manipulation (Hanley, Iwata, Thompson, & Lindberg, 2000) and
touching toys (Nuzzolo-Gomez et al., 2002), which may have been appropriate for the participants in those studies but limits extensions to more complex
play behavior requiring a series of steps.
In addition to these key elements of what constitutes play, another factor
to consider when designing a toy play intervention in an applied setting is
whether the intervention will address sources of competing reinforcers.
Engagement in other behaviors that produce competing reinforcers have been
cited as a potential reason that children with ASD fail to develop appropriate
play skills, most notably when the competing behavior is stereotypy (e.g.,
Honey, Leekam, Turner, & McConachie, 2006; Wing & Gould, 1979). Thus,
many studies evaluating treatment packages for addressing toy play include
measures to specifically target the reduction of stereotypy. Some components
of these treatment packages include contingent access to tangible and social
reinforcement for appropriate play (e.g., Eason et al., 1982; Horner, 1980;
Nuzzolo-Gomez et al., 2002; Singh & Millichamp, 1987), blocking stereotypy (e.g., Hanley et al., 2000), and contingent access to stereotypy for appropriate play (e.g., Hanley et al., 2000; Potter, Hanley, Augustine, Clay, &
Phelps, 2013). Thus, not only will an effective treatment package need to
address the common criticisms of behavioral play interventions noted above,
it must also effectively address stereotypy.
Lang et al. (2014) provided an efficacious treatment package that met
these criteria with three young children with ASD who demonstrated low
levels of play and frequent stereotypy. To address concerns related to generality, the baseline and posttest measures were conducted in a different setting
from intervention. Furthermore, generalization to untrained toy sets was also
measured. To address concerns related to extrinsic reinforcers, the baseline
and posttest evaluations took place during unstructured play situations in the
absence of the teacher who conducted the treatment. Treatment consisted of
12 Behavior Modification 42(1)
redirecting stereotypy, prompting toy play, and contingent tangible reinforcement on a gradually faded schedule. All participants showed increased occurrences of appropriate play in the untrained setting, in the absence of the
teacher and extrinsic reinforcement. For one participant, play behaviors generalized to an untrained play set and stereotypy remained at low levels. The
other two participants still showed a moderate degree of stereotypy, though
below baseline levels, and generalization across toy sets was not observed.
For these participants, a lag schedule of reinforcement was added which
resulted in sustained, low levels of stereotypy and generalization to the
untrained play set. Overall, this study serves as a strong example of a treatment package to address play skills in an applied, school-like setting.
However, the children in this study were 3 years old and had emerging vocal
language so their responsiveness to this intervention may not represent findings for older, school-aged students with greater barriers to learning.
Furthermore, participants selected for the Lang study engaged in pretreatment repetitive toy play and were able to select toys within a paired stimulus
assessment, suggesting they already had some degree of interest in engaging
with toys. Two other studies also showed both a reduction in stereotypy and
increased play with generality across toys, one using video modeling
(Paterson & Arco, 2007) and the other, self-management (Stahmer &
Schriebman, 1992). While the participants in these studies were school-aged
(6-13 years), they were also able to communicate using complete sentences
and were reported to have generalized imitation repertoires (Paterson &
Arco). So, although these interventions addressed stereotypy and generalization of play behaviors, these approaches may not be effective for minimally
verbal, school-aged children with ASD.
Although several effective approaches to increasing toy play while also
reducing stereotypy have been established in the literature, no study has
shown how to teach more complex play sequences to minimally verbal,
school-aged children. This may be accomplished through the application of
backward chaining, a procedure often referenced in curricular manuals (e.g.,
Leaf & McEachin, 1999; Maurice, Green, & Luce, 1996) but yet to be widely
applied in the literature. In their review of the literature, Stahmer et al. (2003)
outlined general recommendations for teaching simple toy play to individuals
with disabilities using chaining procedures. The authors highlighted the
importance of breaking down a toy play sequence into steps, teaching the
individual steps, and then applying forward or backward chaining procedures
to ensure the sequence of steps is established so that the toy play is functional. Given the complexity of play skills and the likelihood that functional
play will require multistep sequences, backward chaining may be a valuable
procedure to include in treatment packages aimed at increasing play.
Edwards et al. 13
The purpose of the current study was to build upon prior research on
establishing functional play skills by evaluating the direct and indirect effects
of an intervention package consisting of (a) backward chaining, (b) blocking
stereotypy, and (c) differential reinforcement in the form of tangible items
and access to stereotypy. Furthermore, the current study addressed limitations of prior studies by including measures of play during free play, assessment of generalization across settings, and generalization to novel toys.
These components were included to most closely mimic those present in
most classroom settings. Importantly, the intervention was conducted with
three school-aged children with ASD with minimal verbal skills to evaluate
the effectiveness of the intervention within a population with more pronounced deficits than has been included in prior studies (e.g., Lang et al.,
2014; Nuzzolo-Gomez et al., 2002; Paterson & Arco, 2007; Stahmer &
Participants and Setting
All participants received one-on-one services within a therapeutic classroom
at a center-based intervention program, 5 days a week for 2 hr each day. The
Verbal Behavior-Milestones Assessment and Placement Program (VB-MAPP;
Sundberg, 2008) was administered to each participant upon their admission
into the clinic and was used for ongoing clinical programming. The clinical
case manager, a board certified behavior analyst (BCBA), in consultation
with the participantâ€™s parents determined that an intervention to support the
development of functional play was appropriate. Scores from the VB-MAPP
were used to confirm deficits with play, but no other inclusion or exclusion
criteria were required for participation in the study.
Ethan, a 4-year-old male, was diagnosed with ASD at the age of 3 by a
licensed psychologist. He was administered the Mullen Scales of Early
Learning (Mullen, 1995), which indicated his skills were in the 18- to
21-month range for receptive language, visual reception, and fine motor
skills while scores related to oral production were comparable to 14 months.
The Adaptive Behavior Assessment System (ABAS; Harrison & Oakland,
2003) indicated Ethanâ€™s overall functioning was below the 1st percentile.
Ethan had a VB-MAPP Milestone score of 46, a play domain score of 7.5
(18- to 30-month range), and an imitation domain score of 7 (18- to 30-month
range). Specifically, in the area of play, Ethan was observed to engage functionally with only a few items when specifically directed, but was unable to
assemble toys with multiple parts. Although Ethan engaged in cooperative
14 Behavior Modification 42(1)
play with others, he did not engage in sustained independent play. His parents
noted his limited independent play as a particular concern at home, as they
reported that he engaged in either impassivity or repetitive bids for attention
when asked to play by himself (e.g., such as when parental attention was
diverted toward cooking dinner). In regard to imitation skills, ongoing clinical data collection suggested that by the time Ethan began his participation in
the study, he could imitate any novel action presented by a therapist. Last,
Ethan had a score of 0 on the VB-MAPP Barriers Assessment for self-stimulation, indicating that he did not engage in stereotypy.
Lance, a 4-year-old male, was diagnosed with a neurodevelopmental
disorder, which encompasses ASD, at age 2, by a neurodevelopmental
pediatrician. At that time, the Capute Scales (Accardo & Capute, 2005)
were administered, including both the Cognitive Adaptive Test (CAT) and
Clinical Linguistic Auditory Milestone Scale (CLAMS). Scores from the
CAT revealed overall functioning in the 16- to 18-month range. Scores
from the CLAMS revealed receptive language functioning in the 7- to
12-month range and expressive language functioning in the 8- to 14-month
range. Lance had a VB-MAPP Milestone score of 42, a play domain score
of 7 (18- to 30-month range), and an imitation domain score of 1.5 (0- to
18-month range). Specifically, in the area of play, Lance could engage
functionally with several items when specifically directed, but could not
assemble toys with multiple parts. In regard to imitation skills, Lance mastered his first gross imitation target after 3 months of teaching. Lance had
a score of 2 on the VB-MAPP Barriers Assessment for self-stimulation,
indicating that he engaged in stereotypy on a relatively frequent basis and
that it interfered with his completion of other activities. Lanceâ€™s parents
completed the Behavior Problems Inventory (BPI) and reported a variety
of stereotypic behaviors of varying intensity and frequency including pacing, head hitting, spinning, repetitive body movements, scratching, and
waving his arms.
Trey, a 7-year-old male, had been reportedly diagnosed with ASD, but no
additional testing or diagnostic information was available. Trey had a
VB-MAPP Milestone domain score of 20.5, a play domain score of 4 (0- to
18-month range), and an imitation domain score of 4 (0- to 18-month range).
Specifically, in the area of play Trey was observed to engage functionally
with several items when specifically directed and could assemble some toys
with multiple parts, such as puzzles and matching tasks, but did not do so in
the absence of adult direction. Trey did not receive a score on the VB-MAPP
Barriers Assessment for self-stimulation because he engaged in high rates of
other forms of problematic behavior during the initial assessment. However,
his therapists reported instances of self-biting, pacing, twirling, and bursts of
Edwards et al. 15
running which interfered with group play periods. Caregiver information
from the BPI was not available for Trey.
Baseline and treatment sessions took place in the clinic setting. Each participant had his own teaching area in a large classroom which consisted of a
table, chairs, preferred toys and foods, and other teaching materials. Pretests
and posttests were conducted in a separate room within the clinic, to serve as
a generalization environment. The room consisted of a table, three treatment
toys, three novel toys, and probe materials.
For all participants, a total of six toys were selected. Three toys were selected
for treatment: single inset puzzles, toy train and tracks, and toy piano. The
remaining three toys were assigned to generalization: car track, drum toy, and
pop-up toy. The toys were specifically selected to promote developmentally
appropriate, functional play. Conventional use of these toys produced a specific outcome (e.g., a song on the piano, a train pushed around the tracks),
making functional engagement more readily differentiated from stereotypy.
For example, when participants engaged in correct steps for playing the
piano, the toy lit up and replayed the designated song. Likewise, engaging in
the designated steps for the train set involved constructing the train track,
thereby enabling participants to push the train around the track. Finally, the
toys selected were deemed to be developmentally appropriate given the participantâ€™s developmental level (Lifter, Ellis, Cannon, & Anderson, 2005). The
VB-MAPP Milestones Assessment suggests that playing with toys with multiple parts and playing with toys according to design emerges within 18 to 30
months in development, which aligns with the scores observed for all participants. For these participants, preference for specific toys was not used
because none of these participants freely engaged or selected toys to engage
with either at home or on any occasion in the clinical setting.
Functional leisure engagement (FLE) was the primary dependent variable
evaluated during the pretest and posttest. FLE was defined as any instance in
which the participant was engaged in functional play responses specifically
identified for each activity. A list of appropriate actions was created for each
toy which would resemble appropriate FLE in a typically developing child
(e.g., toy train tracksâ€”connecting the track, putting the train on the track,
moving the train along the track). FLE was not scored if the participant was
not engaging with a toy, engaging with a toy in an inappropriate way (i.e.,
16 Behavior Modification 42(1)
mouthing, disruption, etc.), or engaging with the toy for less than 5 s at a
time. Further, FLE did not require play responses to occur in a specific order
or for completion of a sequence of actions. The duration of FLE was tracked
with a 5-s onset/offset. The number of seconds spent in FLE was then divided
by 360 (the total number of seconds in the 5-min session), resulting in the
percentage of session spent engaged in FLE.
Step completion served as the dependent variable during step completion
baseline and treatment probe sessions. Step completion was defined as the
independent completion of a step in the designated order. A task analysis was
created for each activity consisting of eight steps. Step completion was
defined separately for each toy and was scored each time the participant completed a step in the designated order, without prompts. Step completion was
not recorded if the participant did not engage with the toy or complete the
step in order, or required prompts to complete the step. For example, for the
train set containing six tracks and one train, 100% correct step completion
was recorded if the participant completed the following steps in order: Step
1â€“connect track, Step 2â€“connect track, Step 3â€“connect track, Step 4â€“connect
track, Step 5â€“connect track, Step 6â€“connect track, Step 7â€“put train on top of
track, Step 8â€“move train along the track. Observers recorded the total number
of steps completed, which was divided by the total number of steps and converted to a percentage. As backward chaining was used, the chain was completed by the therapist and only the targeted steps were probed.
Stereotypy was measured during pre- and posttests for Trey and Lance and
included a variety of behaviors. For Trey, stereotypy was defined as any
instance of mouthing, spinning, posturing, and bursts of running. Mouthing
was defined as any instance in which an object or Treyâ€™s fingers crossed the
plane of Treyâ€™s lips with a 5-s onset/offset. Spinning was defined as any
instance in which Trey rotated his body 360 degrees with a 5-s onset/offset.
Posturing was defined as any instance in which Treyâ€™s knees were on the
floor with a 5-s onset/offset. Bursts of running were defined as any instance
in which Trey moved at a pace faster than walking around the room with a 5-s
onset/offset. For Lance, stereotypy was defined as any instance of head hitting, pacing, repetitive body movements, scratching, and spinning. Head hitting was defined as any instance or attempt in which an object Lance was
holding contacted his head from a distance of 3 in. (7.6 cm) or greater. Pacing
was defined as any instance in which Lance walked more than 3 ft. (.91 m)
from one point in the room to another, at least two times with a 5-s onset/
offset. A repetitive body movement was defined as any instance in which
Lance engaged in behavior at least five consecutive times with no functional
relation to the toys in the room with a 5-s onset/offset. Scratching was defined
as any instance in which Lanceâ€™s nails came into contact with his skin and left
Edwards et al. 17
red marks with a 5-s onset/offset. Spinning was defined as any instance in
which Lance rotated his body 360 degrees with a 5-s onset/offset. A total
duration measure was taken for all behaviors and converted into a percentage
Interobserver Agreement (IOA)
IOA data were collected by trained independent observers using pen and
paper and were collected for FLE during an average of 79.8% (range =
66.8%-100%) of sessions across all participants. IOA data were collected for
stereotypy an average of 34.8% (range = 33.3%-36.4%) of sessions across
two participants (Lance and Trey). The percentage of IOA data was calculated by taking the smallest duration measure recorded and dividing by the
largest duration measure, then multiplying by 100. The percentage of IOA
averaged 96.7% (range = 90%-100%) for FLE. The percentage of IOA averaged 87.1% (range = 72.8%-100%) for stereotypy. For step completion data,
data were collected for an average of 44.5% (range = 41%-50%) of sessions
across all participants. For these data, trial-by-trial IOA was calculated for
each participant by taking the number of trials with agreements and dividing
by the total number of trials with agreements and disagreements then multiplying by 100. The percentage of agreements averaged 98.9% (ranging =
Fidelity of Implementation
Treatment integrity data were also collected by a trained independent
observer. Step-by-step procedures were outlined in a checklist, and the independent observer recorded whether the therapist followed the procedures
exactly as detailed (scored as a â€œ+â€) or with any deviation (scored as a â€œâˆ’â€).
Scores were calculated by dividing the number of correctly implemented
steps by the total number of steps then multiplying by 100. The average percentage of sessions in which treatment integrity was scored across all participants was 61% (range = 44.5%-79.8%). The percentage of correct
implementation averaged 100%.
Experimental Design and Visual Display of Data
A multiple probe design across participants was used to evaluate FLE prior to
and following step completion treatment (i.e., pretest and posttest). The pretest and posttest sessions followed the procedures of a free operant preference
assessment (Roane, Vollmer, Ringdahl, & Marcus, 1998). Step completion
18 Behavior Modification 42(1)
was evaluated in a manner similar to a changing criterion design during baseline and treatment probe sessions. However, as backward chaining was incorporated into the treatment probe sessions, the participants were not free to
complete all of the possible steps. Thus, the number of steps that could be
completed in a treatment probe was limited to the target step and previously
Reinforcer selection. Therapists identified preferred items for all phases of this
study using a variety of formats. Because the participants in this study were
also being served in a clinic providing comprehensive intervention, careful
assessment of highly preferred items and activities using questionnaires
(Fisher, Piazza, Bowman, & Amari, 1996) and structured preference assessments (e.g., DeLeon & Iwata, 1996) were periodically conducted. The identified items and activities were then incorporated into sessions and moment to
moment verbal requests of those identified items were used to identify current preference. Therefore, a variety of preferred items identified via parental
interview, structured preference assessment, and requests were present on the
table and additional items were stored in bookshelves nearby. The therapist
either waited for participants to engage in a request or issued a vocal prompt
to occasion a request (e.g., â€œWhat would you like to work for?â€) before starting each session. Lance typically requested the iPad, cookies, and pretzels.
For Trey, an iPad was the only item requested as he did not exhibit sustained
interest in any other play activities or foods. Ethan typically requested potato
chips, candy, or the iPad. Ethanâ€™s clinical programming also included the use
of a token board. Once all tokens were earned, he would often clap his hands
and remove the tokens to start over, rather than exchange them for a backup
reinforcer. Thus, the token board was incorporated throughout all phases of
the study for Ethan. Tokens (Ethan) and other items (Lance and Trey) were
delivered at the end of each session. During these reinforcer intervals, therapists reinforced requests (including bids for attention) and allowed unrestricted access to stereotypy.
Pretest and posttest. Pretest and posttest sessions were used to evaluate the
percentage of session in which the participant demonstrated FLE and stereotypy. Sessions were 5 min and were conducted in a separate room inside the
clinic. Procedures followed those used in free operant preference assessments
(Roane et al., 1998). Toys were placed on a table in the room and were equidistant from each other and the participant. Using hand over hand prompts
provided by the therapist, the participant was exposed to all toys individually
Edwards et al. 19
for 10 s prior to every session. The therapist began the session by stating,
â€œPlay with whatever you want.â€ During the session, no prompts or other
forms of interaction with the participant were provided. If dangerous behavior occurred, such as elopement, this behavior was blocked, and minimal
attention was provided. The therapist tracked the total duration of FLE and
stereotypy on separate timers and a count of the number of toys the participant demonstrated FLE with during the session. After 5 min, the therapist
provided noncontingent access to tokens (for Ethan) or the previously identified preferred item (Lance and Trey) for a period of 1 min. Noncontingent
access was provided so that results of pre- and posttest sessions could be
evaluated in the absence of programmed contingencies. Furthermore, it was
hypothesized that these conditions would mimic those of a school routine.
For example, after the independent play time, all students rotate to recess or
snack (presumed highly preferred activities) regardless of performance during play time. Additionally, the participants were not blocked from engaging
in stereotypy during these periods. Pretest sessions were conducted until
Step completion baseline. Baseline sessions were conducted with each of the
six toys. Therapists presented one toy per session, in random order, for 2 min,
during which data collectors scored the number of steps completed. During
the session, no prompts or other forms of interaction with the participant were
provided. Dangerous behavior, such as elopement, was blocked, but the therapist avoided redirecting the participant to engage in any other behavior. At
the end of the session, therapists delivered tokens (Ethan, which he then
exchanged for his iPad) or noncontingent access to the previously selected
preferred items for 2 min. As noted above, during these periods of noncontingent access, stereotypy was not blocked. After the period of noncontingent
access, the therapist presented the next session. A minimum of three baseline
sessions per toy were conducted until data were deemed stable.
Step completion treatment probe. Percentage of steps completed for each of
the three toys assigned to treatment was evaluated during each step completion treatment probe session. The therapist randomized the toys and placed
them one at a time in front of the participant. Using backward chaining, the
therapist completed all nontargeted steps then provided the instruction, â€œPlay
with your toysâ€ and provided an independent response opportunity to complete the next step. The therapist then recorded the steps completed independently by the participant. If the participant errored or did not initiate step
completion within 10 s, the therapist used least-to-most prompting to complete that step and then moved on to the next step to be probed. Least-to-most
20 Behavior Modification 42(1)
prompting progressed through the following as needed: a verbal instruction
followed by a gestural prompt, followed by light physical guidance to move
the childâ€™s hands to the toy, followed by hand-over-hand guidance to complete the step. Upon independent completion of the behavior chain, the therapist provided 30-s to 60-s access to reinforcers depending on the number of
steps completed. Step completion treatment probe sessions were then followed by treatment sessions (described below). Mastery criteria ranged from
three to six consecutive probes with 100% independence on the targeted step.
Following mastery of a step, treatment began for the next step in the chain.
Mastery of the entire chain for each toy was 100% independent step completion of a predetermined number of consecutive probes; however, probes of all
steps continued following mastery to assess maintenance.
Step completion treatment. Following step completion treatment probes, step
completion treatment sessions were conducted with each toy. Procedures
were similar to step completion treatment probes; however, the therapist used
most-to-least prompting to help the participant to complete the target steps
and provided the preselected preferred items (e.g., tokens, tangible items)
contingent upon completion of the final step of the chain. If stereotypy
occurred during the treatment sessions, the therapist immediately blocked
and redirected by prompting the appropriate step in the chain. Each step completion treatment sessions typically lasted less than 2 min. Following three
step completion treatment sessions, a step completion treatment probe session was conducted. Of note, a treatment probe was not conducted until at
least 20 min had elapsed since the last step completion treatment session.
Results of pretests and posttest for all participants are shown in Figure 1 and
the results of step completion baseline and treatment probes are shown for
individual participants in Figures 2 to 4. The average percentage of sessions
(pretest and posttest) with occurrences of stereotypy for Lance and Trey is
shown in Figure 5. As shown in Figure 1, Ethan demonstrated no FLE during
pretest sessions (M = 0%). In Figure 2, the step completion baseline data
show he completed an average of 16.7% steps correctly for piano (range =
0%-25%), an average of 29.2% steps correctly for puzzles (range =
0%-87.5%), and no steps correctly for the trains. During step completion
treatment, Ethan reached mastery criteria, independently completing 100%
of steps correctly for at least three consecutive sessions, in 36, 34, and 46
sessions for piano, puzzles, and trains, respectively. During the posttests
(Figure 1), Ethan engaged in FLE for an average of 66.7% of the sessions
Edwards et al. 21
Figure 1. Results from pretest and posttest sessions are displayed for all
Note. The percentage of sessions in which the participants engaged in FLE with target toys
are shown with black bars. The percentage of sessions in which participants engaged in FLE
with generalization toys are shown with gray bars. The asterisks indicate that a session was
conducted, but 0% FLE was exhibited. The solid phase line indicates when SC treatment
occurred with toys assigned to treatment. SC = step completion; FLE = functional leisure
22 Behavior Modification 42(1)
(range = 0%-100%). During all posttest sessions, Ethan only engaged with
the generalization toys (e.g., pop-up toys, car track, and drum). No stereotypy
data were collected for Ethan.
Figure 2. Results of BL and step completion treatment probes across the piano,
puzzles, and trains for Ethan.
Note. The dashed line indicates the designated percentage of steps requiring independent
completion. BL = baseline.
Edwards et al. 23
As shown in Figure 1, Lance displayed FLE during the pretests for an
average of 9.1% of the sessions (range = 0%-15%). The percentage of sessions in which FLE included generalization toys averaged 5.4% (range =
0%-12.33%) and for treatment toys averaged 4% (range = 0%-11%). As
shown in Figure 5, during pretests Lance engaged in stereotypy for an
Figure 3. Results of BL and step completion treatment probes across the piano,
puzzles, and trains for Lance.
Note. The dashed line indicates the designated percentage of steps requiring independent
completion. BL = baseline.
24 Behavior Modification 42(1)
average duration of 35.5% of the sessions (range = 20%-46%). In Figure 3,
the step completion baseline data show he completed 0% of steps correctly
for piano, puzzles, and trains. During step completion treatment, Lance
reached mastery criteria in 41, 34, and 45 sessions for piano, puzzles, and
Figure 4. Results of BL step completion treatment probes across the piano,
puzzles, and trains for Trey.
Note. The dashed line indicates the designated percentage of steps requiring independent
completion. BL = baseline.
Edwards et al. 25
trains, respectively. During the posttests (Figure 1), Lance engaged in FLE
for an average of 75.2% of the sessions (range = 72%-79.3%). The percentage of session engaged in FLE with generalization toys averaged 23.1%
(range = 11.7%-32.7%) and for treatment toys averaged 52.1% (range =
39.3%-67.7%). As shown in Figure 5, during posttests Lance engaged in stereotypy for an average duration of 2.2% of the sessions (range = 0%-6.6%).
As shown in Figure 1, Trey displayed FLE during the pretests for an average of 5.9% of the sessions (range = 0%-29.3%). Trey engaged only with the
treatment toys during the pretest. As shown in Figure 5, during pretests Trey
engaged in stereotypy for an average duration of 40.7% of sessions (range =
3.6%-65%). As shown in Figure 4, the step completion baseline data show he
completed an average of 0% steps correctly for piano, an average of 16.67%
Figure 5. Results from pretest and posttest sessions are displayed for Lance and
Note. The percentage of sessions in which the participants engaged in stereotypy are shown
with white bars. The asterisks indicate that a session was conducted, but 0% stereotypy was
exhibited. The solid phase line indicates when SC treatment occurred. SC = step completion.
26 Behavior Modification 42(1)
of steps correctly for puzzles (range = 12.5%-25%), and an average of 33.3%
of steps correctly for trains (range = 0%-62.5%). During step completion
treatment, Trey reached mastery criteria in 94, 36, and 71 sessions for piano,
puzzles, and trains, respectively. During the posttests (Figure 1), Trey engaged
in FLE for an average of 52.11% of the sessions (range = 0%-69.7%). The
percentage of session engaged in FLE with generalization toys averaged
14.7% (range = 0%-23%) and for treatment toys averaged 37.4% (range =
0%-55.4%). As shown in Figure 5, during posttests Trey engaged in stereotypy for an average duration of 17.6% of the sessions (range = 0%-38.4%).
All three participants in the current study demonstrated an increase in functional play in treatment settings (i.e., step completion treatment), which
resulted in an increase in FLE in a free play setting (i.e., posttest). Two of the
three participants engaged with both targeted and generalization toys in the
posttests. The results of this study extend upon prior research on teaching
functional toy play to children with autism by demonstrating the efficacy of
backward chaining in teaching more complex play behaviors involving multiple steps in a sequence. These results also add to the body of research that
demonstrates the generality of behavior-analytic interventions for teaching
toy play and address some of the limitations regarding behavioral approaches
to teaching play. By assessing toy play in an unstructured setting in which the
intervention was fully withdrawn, we were able to evaluate whether activities
were selected and initiated by the child, who was free to either play with
items or engage in other behaviors (e.g., stereotypy, impassivity, or attempts
to gain access to other reinforcers). The inclusion of generalization toys
enabled us to assess treatment generality to novel items and allowed us to
identify the occurrence of novel play behaviors. Our results indicated that,
following treatment, participants continued to engage in play in the absence
of programmed contingencies, thereby meeting the definition of play set
forth by Hughes (2012) and addressing the concern proposed by Luckett
et al. (2007) that highly structured approaches result in reliance on external
rewards. The increase in untargeted play actions with novel toys is also notable given Luckett et al.â€™s concerns regarding limited spontaneity, flexibility,
and generalization associated with highly structured behavior-analytic interventions. Although similar effects have been previously demonstrated (Lang
et al., 2014), our results provide further support to the generality of behavioral interventions for play by including children who exhibited limited interest in toys prior to intervention (i.e., children who either did not approach
toys or did not engage in sustained interaction with toys) and with more
Edwards et al. 27
pronounced intellectual and verbal deficits than in other studies (Paterson &
Arco, 2007; Stahmer & Schriebman, 1992).
The generality of play behaviors to untargeted toys and the persistence of
play even when external reinforcers were removed suggests that something
about the learned play behaviors themselves were reinforcing. This is an ideal
outcome. The use of a backward chaining procedure to teach multistep functional play potentially lead to the child experiencing for the first time, the reinforcing experience of playing with toys. In contrast, previous studies that
targeted simple object manipulation or merely touching toys may not have
taught behaviors that ultimately were automatically reinforcing. In the current
study, it is possible that teaching the participants what to do with the toys
required reinforcement for the behaviors to be learned, but did not require
external reinforcement for the behaviors to be maintained. Further, the generalization toys may have had similar features to the trained toys that made engaging in a sequence of responding possible with those toys even though they were
novel. For example, the car and car track have similar features to the train and
train track in that the car and train both have wheels and are vehicles and the car
track and train track are both flat tracks with grooves (though different shapes
and colors). Thus, in the presence of the novel car track, similar responses such
as pushing the car along the car track may have been evoked. Teachers and
parents hoping to teach generalized toy play skills may want to consider
whether the toy play behaviors will be automatically reinforcing once taught
and may consider having other toys with similar features available.
The treatment package we employed involved several components of previously established interventions for increasing functional engagement such
as prompting and differential reinforcement of appropriate play behaviors
with social reinforcers (e.g., Nuzzolo-Gomez et al., 2002), restricted access
to stereotypy during â€œplay timeâ€ (e.g., Hanley et al., 2000; Potter et al., 2013),
and access to nonsocial reinforcers contingent on appropriate play (i.e., contingent access to stereotypy or other behaviors; e.g., Potter et al., 2013),
which were embedded within a backward chaining approach. Although this
intervention sounds quite intensive, one backward chaining session with one
targeted toy typically took fewer than 2 min to conduct. Therefore, while
other interventions, such as self-management or in-vivo or video modeling,
might seem relatively more feasible to implement in settings such as the
classroom, the current intervention package is fairly simple to implement and
could be interspersed with other teaching activities throughout a school day.
The intervention employed in our study was efficacious for three participants who exhibited limited to no play in unstructured settings, with limited
imitation skills, and who engaged in stereotypic behaviors (Lance and Trey),
suggesting that this package may be particularly suited for children with
28 Behavior Modification 42(1)
more pronounced deficits and without the prerequisites (i.e., imitation skills)
to benefit from other empirically supported interventions such as video modeling or self-management. In addition, the finding that it resulted in generalization of play to contexts with minimal to no oversight suggests that, once
the toy play skills are taught, parents or teachers may see increases in play
during periods in which attention is not continuously available, without having to intervene in or otherwise restructure the typical play context. This may
be ideal for caregivers whose attention must be diverted during the times in
which the child is expected to play independently. In a classroom, for example, free play contexts are often associated with lower instructor to child
ratios as the teacher is preparing for the next instructional block.
As noted, two of the participants in our study demonstrated reductions in
stereotypy in free play contexts following intervention. Given the likelihood
that stereotypy competes with appropriate functional play, this result is encouraging. However, it is important to note that these participants engaged in stereotypy at low to moderate levels in baseline. Whether the treatment package
employed in our study would also be successful for individuals who engage in
high-rate or continuous stereotypy is unclear and other procedural components
may be necessary to facilitate practical maintenance of low rate stereotypy for
these individuals (see Slaton & Hanley, 2016, for an example and further discussion). Thus, replication across participants is necessary to evaluate this
across a wide variety of children with ASD. Future researchers should also
consider conducting component analyses to determine the critical components
of the intervention package. Last, additional research is also needed to evaluate
whether similar results would be achieved when the intervention package is
implemented by teachers or parents and whether caregivers deem the procedures and outcomes associated with this intervention as socially acceptable.
A few limitations and other procedural aspects of our study indicate several other areas for future research. First, we assessed generalization across
only one setting and did not evaluate maintenance in the free play, posttest
context. Therefore, in addition to researching the generality of this intervention across participants, researchers should examine the extent to which this
intervention facilitates both generalization and maintenance across a wider
variety of situations. Second, we included a component (i.e., backward
chaining) that has not been previously evaluated in the literature on toy play.
Though effective in the current study, it is not clear if other procedures
would be more effective and efficient. Therefore, future researchers might
investigate the relative efficacy of backward chaining compared with other
teaching procedures, such as forward chaining or differential reinforcement
with schedule thinning, in promoting generalization and maintenance of toy
play among children who demonstrate limited interest in toys. Furthermore,
Edwards et al. 29
it may be the case that the relative efficacy of backward chaining is dependent on the type of play skills being taught and, therefore, researchers should
also consider assessing the relative efficacy of backward chaining using a
wide variety of play items, including those that are commonly associated
with a specific sequence of actions (e.g., a puzzle set) and those that are not
(e.g., dolls). Third, we did not formally assess preference for the toys prior
to selecting them for inclusion in the study. Given that the participants did
not engage in any play prior to this study, selecting toys based on preference
was not feasible. However, if a child shows interest in toys or engages in
some approximations of play with toys, conducting a preference assessment
and including higher preferred toys is warranted. Finally, we did not collect
data on whether participants would have engaged in steps above and beyond
that which we specifically reinforced during the backward chaining intervention. In other words, participants were given the opportunity to complete
only a certain number of steps at the time of the treatment probeâ€”whether
they would have exceeded the target requirement is unclear, but is possible,
given the similarity of some steps (e.g., putting one vs. another piece in a
puzzle). Therefore, during treatment, future researchers and practitioners
may also consider interspersing terminal probes that enable students to
engage in multiple play actions, to determine whether play is conforming to
the contingencies in place.
In summary, the current treatment package was effective in teaching functional toy play for three participants diagnosed with autism and may be useful
for parents and teachers who are interested in promoting play. When considering this strategy for use in a school setting, it may be useful to select toys
that promote developmentally appropriate, functional play and that result in
a specific outcome when used correctly. Therefore, it is important that the
selection of toys matches the developmental level of the child and, when
feasible, takes into consideration the childâ€™s preferences for the types of toys
and activities used. It is also important to understand that external reinforcement may be needed to teach play behaviors and that once the natural reinforcement of the new play behaviors is experienced, those external reinforcers
can be removed. Last, teachers may appreciate the current strategy in that
each teaching session only required a few minutes making this procedure
something that might easily fit in with other activities that occur during the
school day. Though not explicitly evaluated, it may be easier to teach step
completion in a few brief 1:1 sessions throughout the day.
The authors would like to thank Mandi Kaur and Heidi Morgan for their assistance
with data collection.
30 Behavior Modification 42(1)
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.
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Accardo, P. J., & Capute, A. J. (2005). The Capute Scales: Cognitive Adaptive Test/
Clinical Linguistic & Auditory Milestone Scale (CAT/CLAMS). Baltimore, MD:
Paul H. Brookes.
Baron-Cohen, S. (1987). Autism and symbolic play. British Journal of Developmental
Psychology, 5, 139-148.
Bates, E., Benigni, L., Bretherton, I., Camaioni, L., & Volterra, V. (1979). The emergence of symbols: Cognition and communication in infancy. New York, NY:
Bloom, L. (1993). The transition from infancy to language: Acquiring the power of
expression. New York, NY: Cambridge University Press.
DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple-stimulus presentation format for assessing reinforcer preferences. Journal of Applied Behavior
Analysis, 29, 519-533.
Eason, L. J., White, M. J., & Newsom, C. (1982). Generalized reduction of selfstimulatory behavior: An effect of teaching appropriate play to autistic children.
Analysis and Intervention in Developmental Disabilities, 2, 157-169.
Fisher, W. W., Piazza, C., Bowman, L. G., & Amari, A. (1996). Integrating caregiver
report with systematic choice assessment to enhance reinforcer identification.
American Journal of Mental Retardation, 101, 15-25.
Hanley, G. P., Iwata, B. A., Thompson, R. H., & Lindberg, J. S. (2000). A component analysis of â€œstereotypy as reinforcementâ€ for alternative behavior. Journal
of Applied Behavior Analysis, 33, 285-297.
Harrison, P. L., & Oakland, T. (2003). Adaptive behavior assessment system: Second
edition. San Antonio, TX: The Psychological Corporation.
Honey, E., Leekam, S., Turner, M., & Mcconachie, H. (2006). Repetitive behavior
and play in typically developing children and children with autism spectrum disorders. Journal of Autism and Developmental Disorders, 37, 1107-1115.
Horner, T. M. (1980). Testâ€“retest and homeâ€“clinic characteristics of the Bayley
Scales of Infant Development in nine- and fifteen-month-old infants. Child
Development, 51, 751-758.
Hughes, B. (2012). Evolutionary playwork: Reflective analytic practice (2nd ed.).
New York, NY: Routledge.
Edwards et al. 31
Jarrold, C., Boucher, J., & Smith, P. K. (1996). Generativity defects in pretend play in
autism. British Journal of Developmental Psychology, 14, 275-300.
Kasari, C., Gulsrud, A. C., Wong, C., Kwon, S., & Locke, J. (2010). Randomized
controlled caregiver mediated joint engagement intervention for toddlers with
autism. Journal of Autism and Developmental Disorders, 40, 1045-1056.
Lang, R., Machalicek, W., Rispoli, M., Oâ€™Reilly, M., Sigafoos, J., Lancioni, G., . . .Didden,
R. (2014). Play skills taught via behavioral intervention generalize, maintain, and
persist in the absence of socially mediated reinforcement in children with autism.
Research in Autism Spectrum Disorders, 8, 860-872.
Lang, R., Oâ€™Reilly, M., Rispoli, M., Shogren, K., Machalicek, W., Sigafoos, J., &
Regester, A. (2009). Review of interventions to increase functional and symbolic play in children with autism. Education and Training in Developmental
Disabilities, 44, 481-492.
Leaf, R., & McEachin, J. (1999). A work in progress: Behavior management strategies and a curriculum for intensive behavioral treatment of autism. New York,
NY: DRL Books.
Lewis, V., & Boucher, J. (1988). Spontaneous, instructed and elicited play in relatively able autistic children. British Journal of Developmental Psychology, 6,
Lifter, K., Ellis, J., Cannon, B., & Anderson, S. R. (2005). Developmental specificity
in targeting and teaching play activities to children with pervasive developmental
disorders. Journal of Early Intervention, 27, 247-267.
Luckett, T., Bundy, A., & Roberts, J. (2007). Do behavioural approaches teach children with autism to play or are they pretending? Autism, 11, 365-388.
Maurice, C., Green, G., & Luce, S. C. (1996). Behavioral intervention for young children with autism: A manual for parents and professionals. Austin, TX: PRO-ED.
Mullen, E. M. (1995). Mullen Scales of Early Learning (AGS edition). Circle Pines,
MN: American Guidance Service.
Nuzzolo-Gomez, R., Leonard, M. A., Ortiz, E., Rivera, C. M., & Greer, R. D. (2002).
Teaching children with autism to prefer books or toys over stereotypy or passivity. Journal of Positive Behavior Interventions, 4, 80-87.
Paterson, C. R., & Arco, L. (2007). Using video modeling for generalizing toy play in
children with autism. Behavior Modification, 31, 660-681.
Piaget, J. (1952). The origins of intelligence in children. New York, NY: International
Potter, J. N., Hanley, G. P., Augustine, M., Clay, C. J., & Phelps, M. C. (2013).
Treating stereotypy in adolescents diagnosed with autism by refining the tactic
of â€œusing stereotypy as reinforcement.â€ Journal of Applied Behavior Analysis,
Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Evaluation of
a brief stimulus preference assessment. Journal of Applied Behavior Analysis,
32 Behavior Modification 42(1)
Sigafoos, J., Roberts-Pennell, D., & Graves, D. (1999). Longitudinal assessment of
play and adaptive behavior in young children with developmental disabilities.
Research in Developmental Disabilities, 20, 147-161.
Sigman, M., & Ruskin, E. (1999). Continuity and change in the social competence of
children with autism, Down syndrome, and developmental delays. Monographs
of the Society for Research in Child Development, 64(1), 1-114.
Singh, N. N., & Millichamp, C. J. (1987). Independent and social play among profoundly mentally retarded adults: Training, maintenance, generalization, and
long-term follow-up. Journal of Applied Behavior Analysis, 20, 23-34.
Slaton, J. L., & Hanley, G. P. (2016). Effects of multiple versus chained schedules
on stereotypy and item engagement. Journal of Applied Behavior Analysis, 49,
Stahmer, A. C., Ingersoll, B., & Carter, C. (2003). Behavioral approaches to promoting play. Autism, 7, 401-413.
Stahmer, A. C., & Schriebman, L. (1992). Teaching children with autism appropriate
play in unsupervised environments using a self-management treatment package.
Journal of Applied Behavior Analysis, 25, 447-459.
Sundberg, M. L. (2008). VB-MAPP: Verbal Behavior Milestones Assessment and
Placement Program. Concord, CA: Advancements in Verbal Behavior Press.
Toth, K., Munson, J., Meltzoff, A. N., & Dawson, G. (2006). Early predictors of
communication development in young children with autism spectrum disorder:
Joint attention, imitation, and toy play. Journal of Autism and Developmental
Disorders, 36, 993-1005.
Ungerer, J. A., & Sigman, M. (1984). The relation of play and sensorimotor behavior
to language in the second year. Child Development, 55, 1448-1455.
Wing, L., & Gould, J. (1979). Severe impairments of social interaction and associated
abnormalities in children: Epidemiology and classification. Journal of Autism
and Developmental Disorders, 9, 11-29.
Wing, L., Gould, J., Yeates, S. R., & Brierley, L. M. (1977). Symbolic play in
severely mentally retarded and in autistic children. Journal of Child Psychology
and Psychiatry, 18, 167-178.
Chelsea K. Edwards received her BA in psychology at Agnes Scott College in
Decatur, GA, and her MA in applied behavior analysis at Ball State University in
Muncie, IN. She currently resides in Duluth, GA and is a practicing BCBA.
Robin K. Landa, MS, BCBA, is a doctoral student at Western New England
University. Her current research interests include assessment and treatment of problem behavior and procedures for promoting generalization and maintenance of adaptive skills.
Edwards et al. 33
Sarah E. Frampton, MA, BCBA, is a behavioral clinical practitioner at the Marcus
Autism Center. Her current research interests include procedures for promoting emergence of untrained skills and developing social communication skills for children
with autism spectrum disorder.
M. Alice Shillingsburg is the director of the Language and Learning Clinic at the
Marcus Autism Center. She is an assistant professor in the Department of Pediatrics
at Emory University School of Medicine. She is a licensed psychologist and board
certified behavior analyst.