Sunteți pe pagina 1din 11

A consecutive case review of token systems

used to reduce socially maintained


challenging behavior in individuals with
intellectual and developmental delays
Mindy Scheithauer1, Tom Cariveau1, Nathan A. Call1, Hailey Ormand1,
SethClark2
Marcus Autism Center, Childrens Healthcare of Atlanta, Emory University School of Medicine, Atlanta, GA,
1

USA, 2Marcus Autism Center, Childrens Healthcare of Atlanta, Atlanta, GA, USA

Objectives: The current paper describes the use of token systems in a behavioral day-treatment unit for severe
challenging behavior using a consecutive case review spanning three years.
Methods: Experimenters reviewed 96 patient case files, 24 of which implemented some token system as a
component of the individuals treatment package. Aspects of each token system (including schedules of token
delivery and exchange; inclusion of token training and response cost; and types of backup reinforcers delivered),
and participant characteristics were coded.
Results: Token systems were most frequently employed during differential reinforcement for alternative behavior
(DRA, most commonly for compliance) or differential reinforcement of other behavior (DRO). Tokens were most
frequently used in an instructional context. Several commonalities were identified (e.g. restriction of backup
reinforcers between token exchanges and initial dense schedules of reinforcement). Compared to past reviews,
this sample had an over-representation of individuals with challenging behavior maintained by escape, multiply
maintained challenging behavior, and individuals for which the function was not identified in a functional analysis.
Treatment packages including token systems resulted in a reduction in challenging behavior for 91.67% of
participants, with 70.83% exhibiting at least an 80% reduction.
Conclusion: Our findings suggest that token systems may be a beneficial component of treatment plans used
to address challenging behavior. Some common components should likely be included in all token systems and
specific client variables may guide decisions related to idiosyncratic treatments.
Keywords: Functional analysis, Challenging behavior, Token economy, Token system

In 1968, Ayllon and Azrin described a series of studies preschool and elementary school classrooms (Boegli and
in which they utilized token-based reinforcement sys- Wasik 1978; Filcheck et al. 2004), special education set-
tems with individuals diagnosed with severe psychiatric tings (McCullagh and Vaal 1975), group homes (Handen
disorders in a hospital setting (Ayllon and Azrin 1968). et al.1984), and psychiatric hospitals (Birchwood et al.
Across six experiments, the authors demonstrated that 1989; Kazdin 1982; Matson and Boisjoli 2009).
providing participants with opportunities to earn tokens Simply put, in the context of reinforcement systems,
that could later be exchanged for tangible rewards or priv- tokens are objects or symbols that can be exchanged for
ileges increased a wide range of work and self-care tasks. goods or services (Hackenberg 2009). Within the behavior
Although other examples of token economies preceded analytic literature, tokens are generally considered to serve
this work (e.g. Cowles 1937; Kelleher and Gollub 1962; as conditioned reinforcers because they initially have no
Wolfe 1936), Ayllon and Azrin demonstrated the useful- inherent value. Instead, any reinforcing efficacy possessed
ness of token systems in applied contexts in a manner by the tokens has been acquired through having been paired
that garnered the interest of clinicians and researchers. with backup reinforcers. The fact that tokens can serve as
As a result, the use of token reinforcement systems has conditioned reinforcers has many advantages to clinicians.
since spread tremendously. Interventions that include For example, tokens are convenient because they can be
token systems are now nearly ubiquitous in settings like delivered frequently without interrupting ongoing activi-
ties and exchanged for a backup reinforcer at some later,
Correspondence to: Mindy Scheithauer, Marcus Autism Center, Childrens more convenient time (Hupp et al. 2002). Thus, research-
Healthcare of Atlanta, Emory University School of Medicine. Email: Mindy.
Scheithauer@choa.org ers originally viewed tokens as a way to bridge delays

The British Society of Developmental Disabilities 2016


DOI 10.1080/20473869.2016.1177925 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3 157
Scheithauer et al. A consecutive case review of token systems

to reinforcement (Jackson and Hackenberg 1996; Kazdin reinforcer (e.g. one token equals one piece of candy), then
2001; Kazdin and Bootzin 1972; Stromer et al. 2000). the token-exchange is set at a FR 1. However, other unit
Additionally, tokens can generally be delivered on a prices could also be set (e.g. ten tokens equivalent to one
much denser schedule than primary reinforcers because unit of the backup reinforcer [FR 10]).
there is less chance of satiation for two main reasons Frequently, dense schedules for each of these compo-
(Moher et al. 2008). First, tokens are not consumed in nents are in place at the start of the intervention. In other
the same sense as many primary reinforcers. Second, when words, what the individual must do to earn a token may
used as generalized conditioned reinforcers, meaning they be minimal, tokens can be exchanged frequently or imme-
have been paired with a variety of backup reinforcers, diately, and/or large portions of the backup reinforcer are
tokens can increase the probability that there will be suf- provided for each token. However, systematic methods for
ficient deprivation for at least one backup reinforcer that thinning schedules of reinforcement are usually employed
will maintain desired behaviors (Skinner 1953, p. 77). so that the final treatment is more manageable (Hagopian
Another significant advantage of token reinforcement et al. 2011).
systems is their flexibility. That is, token systems are Similarly, variations exist in the type of backup rein-
highly adaptable to a wide range of behaviors and clini- forcers delivered following token exchanges. In some
cal needs. Most commonly, token economies have been cases, exchanging tokens produces access to a single
used with individuals with autism spectrum disorders and consistent reinforcer (Kahng et al. 2003). This approach
other developmental disabilities to increase prosocial and is most likely to be effective when challenging behavior
adaptive behaviors (Matson and Boisjoli 2009), such as is exclusively maintained by this particular reinforcer.
those related to school conduct (e.g. remaining in seat) For example, a treatment for challenging behavior that is
and academic performance (Cotler et al. 1972; Nay and maintained by negative reinforcement (e.g. a break from
Legum 1976; Tarbox et al. 2006). However, token econo- demands), might include a token economy in which
mies have also been frequently employed to reduce chal- tokens are always exchanged for breaks from tasks
lenging behaviors, such as disruptive behavior (Zlomke (e.g. McComas et al. 2000). However, in other cases a
and Zlomke 2003), aggression (LePage et al. 2003), inap- variety of backup reinforcers are available (Carton and
propriate sexual behavior (LeBlanc et al. 2000), and severe Schweitzer 1996; Zlmoke and Zlomke 2003), making
food refusal (Kahng et al. 2003). it possible to have different exchange schedules for
Although they are widely used to address challeng- different backup reinforcers (e.g. 1 token=candy, 10
ing behaviors, a recent review of token economies found tokens=a new toy). This procedure is referred to as a
that published accounts varied substantially in terms of menu of reinforcers.
reported effectiveness (Maggin et al. 2011). This incon- Another way in which token-based interventions can
sistency in the results of token economies could stem from vary is the degree to which tokens are explicitly paired
the many possible procedural variations. For example, in with the backup reinforcers. In many cases, conducting
a seminal review on token economies, Hackenberg (2009) specific token training ensures that the tokens are asso-
explained that a token economy comprised three intercon- ciated with the backup reinforcer, and therefore serve as
nected schedules of reinforcement, each of which can vary a conditioned reinforcer (Moher et al. 2008). However,
widely and influence the effectiveness of the entire token token training is not always necessary, as verbal stimuli
economy. The token-production schedule refers to how explaining contingencies may be sufficient for tokens to
long an individual must work or how much of a behav- become conditioned reinforcers (e.g. contingency-specify-
ior they must emit to earn a token. For example, a child ing stimuli; Mistr and Glenn 1992). Verbal stimuli may be
may earn a token after each behavior (e.g. after each math especially sufficient for individuals with a strong receptive
problem completed), after a set number of behaviors (e.g. verbal repertoire.
after every three math problems completed), or after a set Finally, token economies vary with respect to the
period of time (e.g. an entire class period without challeng- inclusion of contingencies in which tokens are lost or
ing behavior). The exchange-production schedule refers removed following inappropriate behaviors (referred to as
to when opportunities to exchange tokens for the backup response cost). Some token economies rely entirely upon
reinforcers arise. For example, a token economy can be this arrangement by providing a fixed number of tokens
arranged such that a student can exchange tokens as soon at set intervals (e.g. each morning), and then removing
as one is earned (Fixed Ratio (FR) 1 exchange-production one each time challenging behavior occurs. Any tokens
schedule), after they have accumulated 10 tokens (FR 10 that remain when the exchange-production schedule is met
exchange-production schedule), or at a certain time such can then be exchanged for backup reinforcers (Salend and
as the end of the school day (Fixed Time (FT) 6h). The Henry 1981).
token-exchange schedule refers to how much of the backup To date, existing reviews of token systems have gen-
reinforcer is delivered for each token. Thus, this schedule erally provided only broad overviews about procedural
can be considered the unit-price of the backup rein- aspects of token economies. However, specifics about
forcer. If each token is equivalent to one unit of the backup each of these variables, such as the exact schedules

158 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3


Scheithauer et al. A consecutive case review of token systems

at the start and end of the token economy as well as Table 1 Demographic characteristics of sample
when the exchange took place, are often not described. Characteristics n %
In addition, Maggin and colleagues (2011) identified Gender
limitations related to research in the field of token Male 17 70.83
Female 7 29.17
economies, including that the majority of studies do Age
not provide demographic information about the sam- 05 4 16.67
ple. Despite the fact that there are many examples of 610 6 25.00
1115 11 45.83
the effective use of tokens in the literature, it remains 1620 3 12.50
unclear how effective they are because of the possibility Ethnicity
African American 14 58.33
of publication bias (Sham and Smith 2014) and signif- Caucasian/White 9 37.50
icant variations in methodology. Asian 1 12.50
The current study expanded on past research related Diagnosis
Autism spectrum disorder 22 91.67
to token economies by providing a consecutive case Conduct disorder 5 20.83
review of all individuals seen in a day-treatment unit Attention deficit hyperactivity disorder 4 16.67
Disruptive behavior disorder 4 16.67
that included tokens in treatment development, elim- Developmental delay 2 8.33
inating the potential for publication bias. Specific Other 3 12.50
descriptive information is provided about the token
systems and participants demographic information,
including the function of their challenging behavior. A Table 2 Characteristics of Problem Behavior and General
broad description of the success of treatments including Treatment
tokens is also presented.
Characteristics n %
Primary concern
Methods Aggression 22 91.67
Participants and setting Disruption 19 79.17
Elopement and/or Dropping 8 33.33
Participants were individuals seen in a day-treatment Inappropriate vocalization 7 29.17
unit for the assessment and treatment of severe chal- Self-Injurious behavior 4 16.67
lenging behavior. Experimenters conducted a consecu- Spitting 2 8.33
Multiple primary concerns 20 83.33
tive case review of all patients admitted from January Function of challenging behavior
2012 to April 2015, which included 94 individuals. Escape 20 83.33
Tangible 15 62.50
Individuals were included if tokens were used at any Attention 10 41.67
point during the admission based on a review of the Automatic 1 4.17
treatment protocols and data, admission summaries, and Multiply maintained 15 62.50
Other treatment components
behavioral recommendations. This review resulted in Extinction 23 95.83
24 individuals who had received some form of token Least-to-most prompting 16 66.67
Functional communication training 12 50.00
treatment, with a mean age of 10.63years (range 318) Visual schedule 10 41.67
and the majority of whom were male (70.83%). All par- Multiple schedule 10 41.67
ticipants were diagnosed with an intellectual and devel- Response reduction 5 20.83
Differential reinforcement 4 16.67
opmental disabilities prior to admission (see Table 1 for Other 5 20.83
additional demographics). Note: Least-to-Most Prompting included vocal, model, and physi-
cal guidance prompts; Miltenberger, 2001.
Challenging behavior and function coding
Upon admission to the day-treatment unit, topographies of
When an analog functional analysis was conducted,
challenging behavior that were of primary concern to the
frequency data were collected on each instance of tar-
caregiver were identified through unstructured interviews
get behavior across a session and were then graphed
and observations. Operational definitions were written
across conditions. For three participants, the functional
for each topography targeted during assessment and/or
analysis was undifferentiated because no challenging
treatment in the participants admission (see Table 2 for
behavior was observed. For the remaining participants,
targeted topographies).
a panel of behavior analysts viewed the functional anal-
Early in the admission, a functional behavioral assess-
ysis graph(s) and came to a consensus on function (sim-
ment was conducted to identify the reinforcer maintaining
ilar to procedures used to identify a consensus by Roane
challenging behavior (i.e. the functional reinforcer). For
et al. (2013)). Any disagreements were discussed, and if
all but two participants, an analog functional analysis was
not resolved through discussion, a vote would have been
conducted following procedures similar to those described
taken with the majority vote being documented. In the
by Iwata et al. (1994), with modifications made based on
present study there were no disagreements, so no voting
idiosyncratic differences.

 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3 159


Scheithauer et al. A consecutive case review of token systems

was required. The panel consisted of five individuals, all of day (e.g. at the end of the day), after a pre-specified
were Board Certified Behavior Analysts who interpreted number of tokens were earned (e.g. after every fifth token
functional analysis data on a regular basis. earned), or continuous schedules (i.e. exchange could
If an analog functional analysis was not conducted or occur after each token earned). In addition, tokens differed
if it was undifferentiated, the case manager at the time in whether they were exchanged for a specified amount of
of the admission concluded function based on records of time with the item/activity (e.g. one token=30 s of time on
the antecedents and consequences related to challenging an iPad, two tokens=60 s, etc.), a specified quantity of an
behavior and caregiver report. The case manager recorded item (e.g. one token=one piece of candy, two tokens=two
the hypothesized function in the clients behavioral rec- pieces of candy, etc.), or whether there was a menu of
ommendations and the experimenter pulled information items that the individual could select from and purchase
from this document. This was done for five participants with his/her tokens (e.g. one token=one piece of candy,
in this study. two tokens=5min of time on the iPad, 15 tokens=a trip
to a playground, etc.). All of these variations were coded
Token system coding for each participant. In addition to the token system, other
Each participants clinical file was searched to gather infor- aspects of treatment that were included in the childs final
mation about variables related to how tokens were used in treatment plan were also coded.
the childs treatment plan. Several variables were common
across all treatments. First, all tokens were exchanged for Interobserver agreement
backup reinforcers, and if the functional reinforcer was During all baseline and treatment sessions, an observer
identified, then this was included as one of the backup recorded the frequency of each instance of targeted chal-
reinforcers. In other words, if the participant engaged in lenging behavior using either a computerized data-col-
challenging behavior to get attention, then attention was lection system or paper-and-pencil recording. A second
one of the backup reinforcers. In addition, the items used independent observer also collected data for a portion of
as backup reinforcers were restricted across the day, while sessions for all participants except one for whom percent
the individual was at the day-treatment unit unless they reduction was based on injuries. Interobserver agree-
were earned by the exchange of tokens. So items used as ment (IOA) was collected for 34.58% sessions across the
backup reinforcers were not freely available outside of remaining participants (range 19.4057.03%). Agreement
the token system. This system, commonly referred to as was calculated using one of two methods. When data were
a closed-economy, is likely to result in a better response collected using a computerized system, IOA was calcu-
to the token system when compared to an open-economy lated by dividing the session into 10 s bins. The observers
when reinforcers are available outside of token-exchange record with the smallest frequency or duration of challeng-
times (Roane et al. 2005). We also coded whether a partici- ing behavior was divided by the other observers record
pant needed token training or not. We found that 45.83% of and multiplied by 100 to reach a percentage. When paper-
participants went through explicit token training to ensure and-pencil data were collected, the observers record with
that tokens would serve as a conditioned reinforcer and the smallest frequency or duration across the entire session
that the individual mastered the skills of token exchange was divided by the other observers record, and also con-
and production prior to starting treatment. verted to a percentage. We summarized mean IOA across
We also coded how tokens were produced; that is, both types of calculations. Average IOA across participants
what the participant was required to do to earn a token. was 96.67% (range 84.8899.92%).
Tokens were contingent on appropriate behaviors (differ-
ential reinforcement of alternative behavior [DRA]), the Effectiveness
absence of challenging behavior (differential reinforce- Effectiveness of the treatment that included a token system
ment of other behavior [DRO]), or a combination of the was determined using a percentage reduction calculation.
two (DRA/DRO). In addition, several participants had The rate of challenging behavior during the initial baseline
different production schedules in different settings (e.g. was compared to the rate observed on the last three days
a DRA during academic periods and a DRO during lei- that the token treatment was in place on the day-treatment
sure periods). The majority of participants started earning unit. The rate in treatment was subtracted from the rate
tokens on a dense schedule, but for some participants, this in baseline, divided by the baseline rates, and converted
schedule was gradually thinned to increase social validity to a percentage. The percentage reduction was calculated
and make the intervention more practical to use. Both the for all participants, with the exception of two who did not
starting and ending production and exchange values were exhibit challenging behavior in baseline. One participant
coded and reported when thinning occurred. had a different calculation because she engaged in covert
Participants treatment protocols differed in the self-injurious behavior that was not observed while she
exchange-production schedule (or when tokens could be was at the clinic. Skin-checks were conducted daily and
exchanged) and how the backup reinforcers were deliv- the number of new injuries caused by her self-injurious
ered. Possible exchange times included specific times behavior was recorded; thus, percentage reduction was

160 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3


Scheithauer et al. A consecutive case review of token systems

calculated using these values. For another participant, Table 3 Components of Token Treatments
in-seat behavior served as the primary target. A percentage Characteristics n %
increase in latency from the start of session to the time he Context
left his seat was calculated from baseline to the last three Demands 18 75.00
Non demands (e.g. Leisure) 1 4.17
days of treatment. Demands & non demands 5 20.83
Percentage reduction in challenging behavior was com- Token training
Yes 11 45.83
pared across participants with different components of the No 13 54.17
token economy including, how reinforcement was earned Response cost
(DRO or DRO/DRA), how reinforcement schedules were Yes 7 29.17
No 17 70.83
thinned (production-exchange schedule thinned; both pro- Target behaviors
duction-exchange and production schedules thinned; or no DRO 7 29.17
DRO & DRA 17 70.83
thinning), whether a response cost was included, whether Schedule thinning
token training was conducted, and variables related to the Exchange-production 5 20.83
cost of backup reinforcers (constant, varied, or the use Production & exchange-production 11 45.83
No thinning 8 33.33
of a menu). Unit price
In addition to percentage reduction, the type of experi- Fixed 13 54.17
Varied 9 37.50
mental design used with each participant was also coded. Menu 2 8.33
For several participants, experimental control was demon-
strated by delivering reinforcement immediately following
compliance or the absence of challenging behavior, with When tokens were delivered based on compliance, all
tokens added in at a later generalization step as reinforce- participants began with a FR 1 schedule of reinforcement
ment schedules were thinned. For these individuals, the (i.e. one instance of compliance resulted in one token).
initial design was coded in addition to when tokens were Similarly, when tokens were delivered based on time
added into treatment. without challenging behavior, the schedule was initially
relatively brief (e.g. 60 s or less for 63.63% of participants
Results with a time-based schedule). Schedule thinning occurred
Challenging behavior and function for the majority of participants (66.67%).
The most common topographies among participants A menu of reinforcers was used for 8.33% of partic-
whose treatment packages included tokens were aggres- ipants. These participants were given a reinforcer menu
sion and disruptive behavior (see Table 2). The majority with several items/activities listed and the number of
of participants treatments targeted multiple topographies tokens that had to be exchanged for each. During the token
of challenging behavior. Regarding the function of the exchange, the participant vocally stated which item he/she
behavior treated with token systems, we found that an would like to purchase. Both of these individuals had
analog functional analysis was conducted with 91.67% of a vocal repertoire that allowed for reliable requests. For
cases. The most common function was escape (83.33%), participants not using a menu, the unit price was either
followed by tangible and attention (see Table 2). Only one constant (meaning one token always resulted in the same
participant engaged in challenging behavior (specifically backup reinforcer) or varied (cost gradually increased
disruption) maintained by automatic reinforcement, other based on when tokens could be exchanged). The most
topographies were maintained by escape. For most par- frequent backup reinforcer used was a break (83.33% of
ticipants, challenging behavior was multiply maintained. participants).
A functional analysis was not conducted for two partic- Finally, only 29.17% of token systems included a
ipants because their treatment goals were completion of response cost, and all participants but one were able to
academic work. Results from the functional analysis were re-earn removed tokens. See Appendix A for individual
undifferentiated for an additional three participants. Thus, participant information related to the token treatments.
a function was hypothesized for these five participants
based on their FBA which is reported in Table 2. Treatment effectiveness
Token systems were included in the final treatment pack-
Common token system components age for all but two of the participants because simpler
All but two participants earned tokens in a demand con- interventions worked just as well for these individuals. The
text. For all participants, tokens were earned for the mean percent reduction across participants was 70.18%
absence, or low rates, of challenging behavior. Procedural with a median of 92.94% (see Table 4). We also com-
differences were present based on whether the participants pared percentage reduction across participants based on
earned tokens for the passage of time without challenging a variety of variables. These included the type of backup
behavior (DRO) or whether they were required to comply reinforcers used (menu, unit price constant, or unit price
with demands (DRA and DRO; see Table 3). varied); how the reinforcement schedule was thinned

 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3 161


Scheithauer et al. A consecutive case review of token systems

Table 4 Results of Treatment Plans the targeted behavior for earning reinforcement, with the
Outcome n % DRO/DRA resulting in a slightly higher percentage reduc-
Tokens used in final treatment tion (M=73.22%) than a DRO alone (M=62.80%). There
Yes 22 91.67 were no differences based on the thinning method (mean
No 2 8.33
Experimental design reductions of 82.20 and 81.89%); however, both were
ABAB generalization 14 58.33 superior to times when reinforcement was not thinned
AB 6 25.00
Multiple baseline 3 12.50
(M=46.57%). Lastly, there was negligible difference for
ABAB 1 4.17 whether or not token training was conducted (average
Change in problem behavior reductions of 74.12 and 66.84%, respectively). None
Increase 2 8.33
150% Decrease 3 12.50 of these comparisons resulted in statistically significant
5075% Decrease 2 8.33 differences.
75100% Decrease 17 70.83
For 75% of participants, experimental control was
Note: A=Baseline, B=Treatment, ABAB-Generalization=tokens achieved through either a reversal (ABAB) or multiple
added after the reversal.
baseline design (see Table 4). However, for most partici-
pants, the tokens were added in later in treatment, using an
(exchange-production schedule, exchange-production and ABAB design with tokens added in during generalization
production schedule thinning, or no thinning); how tokens (see Appendix B for individual participant information).
were earned (passage of time without challenging behav-
ior only [DRO] or whether compliance was also required Discussion
[DRO/DRA]); whether token training was conducted; and The current report suggests that interventionists may select
whether response cost was included (see Fig. 1). Results from a wide range of potential components to include in
from each of these analyses are incorporated below. a token system. Specifically, our sample included vari-
Based on visual analysis of the mean percentage reduc- ations in token production, exchange, and the use of
tion of challenging behavior, inclusion of a response-cost backup reinforcers; which clinicians may alter based on
resulted in higher reductions in challenging behavior individual need or the presenting problem. Some of the
(M=84.56%) than when a response-cost was not included consistent aspects of the token systems used were that
(M=64.26%). There was also an advantage in using a clinicians (a) restricted reinforcement outside of the token
menu of reinforcers compared to using a constant unit- system while participants were at the treatment center, (b)
price, which resulted in moderately higher reductions delivered tokens on a predetermined and consistent sched-
than when a constant reinforcer was utilized with a varied ule, (c) clearly defined exchange time, (d) made backup
unit price (mean reductions of 85.17, 74.09, and 61.20%, reinforcers consistently available at exchange times, and
respectively). Similar differences were found based on (e) utilized assessment tools (i.e. functional analysis or

Figure 1 Between-group comparison of token system components. Gray bars represent the mean percentage reduction in
challenging behavior for each group and the error bars represent one standard deviation (.5 SD above and .5 SD below the mean).

162 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3


Scheithauer et al. A consecutive case review of token systems

preference assessment) to guide the selection of backup this recommendation in that a fixed unit-price was superior
reinforcers. Given the success of the majority of the treat- to a unit-price that changed across the study. However,
ment packages in this case review, these basic components including a menu of reinforcers was superior to both of
may be important to include in token systems. these options.
In the current case review, clinicians generally deliv- It is interesting to note that challenging behavior was
ered tokens either contingent on a specific behavior or for maintained by escape for 83.33% of participants. This per-
the passage of time without challenging behavior. This centage is higher than reviews on published functional
decision was often matched to the setting, with tokens analyses conducted prior to a variety of function-based
delivered for a specific behavior common in demand con- interventions (32.2%; Beavers et al. 2013). Token sys-
texts where there is a socially appropriate behavior that is tems may fit easily into treatments for escape maintained
often incompatible with challenging behavior (i.e. com- challenging behavior because reinforcement may be pro-
pliance). Contrarily, token delivery based on the passage vided for a discrete behavior in the form of compliance
of time without challenging behavior was common in lei- or task completion. In addition, token systems allow for
sure contexts, where there may be a variety of alternative the delivery of intermittent reinforcement and signal the
appropriate behaviors that could be the target for a DRA amount of work required to earn the backup reinforcer
intervention. When challenging behavior served multiple while still allowing for uninterrupted work. Further, the
functions and tokens were used to address multiple sce- delivery of a token is likely less disruptive than stopping
narios, clinicians often used different contingencies based work to consume a backup reinforcer. This finding aligns
on the setting. For example, a DRA in work settings (for with DeLeon and colleagues (1996) description regarding
challenging behavior maintained by escape) and a DRO accumulated vs. distributed reinforcement arrangements,
in a leisure context (for challenging behavior maintained as well as research related to fluent vs. disfluent work
by access to attention or tangible items). Our results sug- schedules (Fienup et al. 2011). This explanation likely
gest a slight difference in treatment effectiveness based contributes to the frequent use of token systems in school
on how tokens were delivered, with an advantage to use settings (Boegli and Wasik 1978; Filcheck et al. 2004;
a combination of DRA and DRO procedures as opposed McCullagh and Vaal 1975).
to a DRO alone. The current study found that the majority of partic-
For several participants, tokens were initially delivered ipants challenging behavior served multiple functions
and exchanged on a dense schedule of reinforcement that (62.50% of participants compared to 18.9% identified by
was later thinned. This thinning procedure is important Beavers et al. 2013). In addition, despite multiple itera-
in that it allows for token delivery on a leaner schedule, tions and modifications to the analog functional analysis,
requiring less effort from the caregiver during general- results were still undifferentiated for three participants.
ization, which may increase caregiver fidelity (Stocco This could be a facet of publication bias in that functional
and Thompson 2015). In addition, providing tokens on analyses demonstrating challenging behavior that is mul-
an intermittent schedule may promote generalization tiply maintained or with inconclusive results may be less
and maintenance if integrity errors do occur (Kazdin and likely to be included in reviews of published work (Sham
Polster 1973; Stokes and Baer 1977). The results from and Smith 2014), or clinicians working with individuals
the current study demonstrate that treatment remained presenting with multiply maintained challenging behavior
effective even when the schedule of reinforcement was may be more likely to utilize token systems. The latter
reduced or thinned over time. Treatment was substantially is likely, given that token systems are particularly useful
less effective when no schedule thinning took place. It intervention for individuals with an undifferentiated or
is somewhat difficult to interpret these results, however, unclear function for challenging behavior as clinicians can
because the criteria to thin reinforcement were a series of include a variety of potential backup reinforcers in a menu
consecutive sessions with low rates of challenging behav- format. This is supported by the superior results in the cur-
ior, preventing thinning when criteria were not met at early rent study for treatment plans using a menu of reinforers.
stages of treatment development. Despite this, the results Another interesting finding is only one participant
demonstrate a compelling argument that schedule thinning engaged in challenging behavior maintained by access
was helpful for the majority of participants. to automatic reinforcement, possibly because tokens are
There were also variations in whether the unit price for most easily exchanged for social and tangible forms of
reinforcement was held constant (e.g. where one token reinforcement. Commonly recommended interventions for
was always exchanged for one piece of food) compared automatically maintained challenging behavior (including
to participants for whom the unit price varied based on response interruption and redirection, competing items,
the exchange-production schedule (e.g. one piece of food and sensory extinction; Vollmer 1994) are not easily con-
delivered whenever the exchange-production schedule verted to a token system.
was met). Past research has suggested that keeping the Another important outcome from this study is that treat-
unit price constant is adventitious when schedule thinning ment plans including response-cost components were more
(Roane et al. 2007). Results from the current study support effective than treatment plans without a response-cost,

 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3 163


Scheithauer et al. A consecutive case review of token systems

but there are some aspects to consider when including a are used. However, the effectiveness outcomes should
response-cost. In this review, all but one participant was be evaluated as exploratory or preliminary findings that
able to re-earn tokens after they were lost. When includ- require verification. In addition, although no statistically
ing a response-cost contingency in a token economy, it is significant differences were found between groups based
important to ensure that the individual does not lose all on aspects of the token systems, this study was not ade-
opportunity to access the backup reinforcer. If the child quately powered to identify minor differences between
is unable to earn tokens back, there are no contingencies groups. Future studies should replicate these comparisons
in place for appropriate behavior after all tokens are lost. with more participants.
Therefore, it is important that he/she can earn the tokens Finally, the participants in this study were selected
back or that another strategy is in place to promote appro- based on a retrospective analysis of clinical case series
priate behavior in this scenario. In addition, using response and the reason a token system was selected varied
cost token systems alone may not promote appropriate across participants. While this allows for a compre-
behavior. In several of the examples discussed in this case hensive evaluation of how token systems were used
review, a DRA system was used to reinforce appropriate in a naturalistic clinic context, it creates difficulties in
behavior. However, if a response cost system were to be corroborating the data for the purpose of evaluating
used in isolation, we would expect reductions in chal- effectiveness. Future research should expand on this
lenging behavior without promoting alternative behavior by creating guidelines or manualizing the interventions
unless the two are incompatible (e.g. losing a token for to guide development of token systems in a consistent
noncompliance would promote compliance given that way. This would also allow for token systems to be
these are incompatible responses). So while response cost evaluated using designs with more experimental rigor
appears to make a token system more effective, it is crucial (e.g. random control trials) and lend further support to
to take these issues into consideration. the use of token systems in clinical work.
Although token systems were used in the vast majority
of participants final treatment packages and these final References
treatment packages resulted in reductions in challenging Ayllon, T. and Azrin, N. H. 1968. Reinforcer sampling: A technique
for increasing the behavior of mental patients. Journal of Applied
behavior, these results should be interpreted with caution Behavior Analysis, 1, 1320. doi:http://dx.doi.org/10.1901/
due to a few limitations. First, token economies were not jaba.1968.1-13.
implemented exclusively; other treatment components Beavers, G. A., Iwata, B. A. and Lerman, D. C. 2013. Thirty years of
research on the functional analysis of problem behavior. Journal of
were also included and varied across participants, mak- Applied Behavior Analysis, 46, 121. doi:http://dx.doi.org/10.1002/
ing it impossible to ascertain the specific contribution jaba.30.
Birchwood, M., Hallett, S. and Preston, M. 1989. Schizophrenia: An
that token systems alone had to reductions in challenging integrated approach to research and treatment. New York, NY: New
behavior. Future research should conduct component anal- York University Press.
Boegli, R. G. and Wasik, B. H. 1978. Use of the token economy system to
yses to determine the direct effect of the tokens themselves intervene on a school-wide level. Psychology in the Schools, 15(1), 72
on challenging behavior with and without other treatment 78. doi:http://dx.doi.org/10.1002/1520-6807(197801)15:1<72:AID-
PITS2310150114>3.0.CO;2-Q.
components. It is important to note that the results of the Carton, J. S. and Schweitzer, J. B. 1996. Use of a token economy to
comparisons of percentage reductions across different increase compliance during hemodialysis. Journal of Applied
components of the token system were not statistically sig- Behavior Analysis, 29(1), 111113. doi:http://dx.doi.org/10.1901/
jaba.1996.29-111.
nificant. However, given our sample size, the differences Cotler, S. B., Applegate, G., King, L. W. and Kristal, S. 1972. Establishing
create an interesting area for future research and should a token economy program in a state hospital classroom: A lesson in
training student and teacher. Behavior Therapy, 3, 209222. doi:http://
be further studied. dx.doi.org/10.1016/S0005-7894(72)80081-9.
In addition, while the reinforcement contingencies were Cowles, J. T. 1937. Food tokens as incentives for learning by chimpanzees.
Comparative Psychological Monographs, 12, 196. doi:http://dx.doi.
implemented in a manner that allowed for single-subject org/10.1037/14268-000.
experimental control for most participants, tokens were DeLeon, I. G. and Iwata, B. A. 1996. Evaluation of a multiple-stimulus
presentation format for assessing reinforcer preferences. Journal
often added in later in treatment. In other words, an ABAB of Applied Behavior Analysis, 29, 519533. doi:http://dx.doi.
design demonstrated the effectiveness of delivering rein- org/10.1901/jaba.1996.29-519.
forcement immediately without using tokens, and tokens Fienup, D. M., Ahlers, A. A. and Pace, G. 2011. Preference for fluent
versus disfluent work schedules. Journal of Applied Behavior
were added in to assist with thinning the schedule of rein- Analysis, 44, 847858.
forcement. Modifying treatment to promote schedule thin- Filcheck, H. A., McNeil, C. B., Greco, L. A. and Bernard, R. S. 2004.
Using a whole-class token economy and coaching of teacher skills in
ning after experimental control has been demonstrated with a preschool classroom to manage disruptive behavior. Psychology in
a more dense and immediate schedule of reinforcement is the Schools, 41, 351361. doi:http://dx.doi.org/10.1002/pits.10168.
Fisher, W. W., Thompson, R. H., Hagopian, L. P., Bowman, L. G. and
not uncommon (e.g. Fisher et al. 2000; Hanley et al. 2001), Krug, A. 2000. Facilitating tolerance of delayed reinforcement during
but it creates a limitation in interpreting the effectiveness functional communication training. Behavior Modification, 24, 329.
doi:http://dx.doi.org/10.1177/0145445500241001.
of the token system itself. Conclusions drawn from this Hackenberg, T. D. 2009. Token reinforcement: A review and analysis.
review create an informative guideline of different ways Journal of the Experimental Analysis of Behavior, 91, 257286.
in which token systems may be implemented, as well as a doi:http://dx.doi.org/10.1901/jeab.2009.91-257.

review of the characteristics of children for which tokens

164 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3


Scheithauer et al. A consecutive case review of token systems

Hagopian, L. P., Boelter, E. W. and Jarmolowicz, D. P. 2011. Reinforcement McComas, J., Hoch, H., Paone, D. and El-Roy, D. 2000. Escape behavior
schedule thinning following functional communication training: during academic tasks: A preliminary analysis of idiosyncratic
Review and recommendations. Behavior Analysis in Practice, 4, establishing operations. Journal of Applied Behavior Analysis, 33,
416. 479493. doi:http://dx.doi.org/10.1901/jaba.2000.33-479.
Handen, B. L., Apolito, P. M. and Seltzer, G. B. 1984. Use of differential McCullagh, J. and Vaal, J. 1975. A token economy in a junior high special
reinforcement of low rates of behavior to decrease repetitive speech in education classroom. School Applications of Learning Theory, 7, 18.
an autistic adolescent. Journal of Behavior Therapy and Experimental Retrieved from http://psycnet.apa.org/psycinfo/1975-30623-001
Psychiatry, 15, 359364. doi:http://dx.doi.org/10.1016/0005- Miltenberger, R. G. 2001. Behavior modification: Principles and
7916(84)90102-2. procedures. Belmont, CA: Wadsworth/Thompson Learning.
Hanley, G. P., Iwata, B. A. and Thompson, R. H. 2001. Reinforcement Mistr, K. N. and Glenn, S. S. 1992. Evocative and function-altering
schedule thinning following treatment with functional communication effects of contingency-specifying stimuli. The Analysis of Verbal
training. Journal of Applied Behavior Analysis, 34, 1738. doi:http:// Behavior, 10, 1121.
dx.doi.org/10.1901/jaba.2001.34-17. Moher, C. A., Gould, D. D., Hegg, E. and Mahoney, A. M. 2008. Non-
Hupp, S. D., Reitman, D., Northup, J., OCallaghan, P. and LeBlanc, generalized and generalized conditioned reinforcers: Establishment
M. 2002. The effects of delayed rewards, tokens, and stimulant and validation. Behavioral Interventions, 23, 1338. doi:http://dx.doi.
medication on sportsmanlike behavior with ADHD-diagnosed org/10.1002/bin.253.
children. Behavior Modification, 26, 148162. doi:http://dx.doi.org/ Nay, W. R. and Legum, L. 1976. Increasing generalization in a token
10.1177/0145445502026002002. program for adolescent retardates: A methodological comment.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E. and Richman, Behavior Therapy, 7, 413414. doi:http://dx.doi.org/10.1016/S0005-
G. S. 1994. Toward a functional analysis of self-injury. Journal of 7894(76)80076-7.
Applied Behavior Analysis, 27, 197209. doi:http://dx.doi.org/10.190/ Roane, H. S., Call, N. A. and Falcomata, T. S. 2005. A preliminary
jaba.1994.27-197. Reprinted from Analysis and Intervention in analysis of adaptive responding under open and closed economies.
developmental Disabilities, 2(320), 1982. Journal of Applied Behavior Analysis, 38, 335348. doi:http://dx.doi.
Jackson, K. and Hackenberg, T. D. 1996. Token reinforcement, choice, org/10.1901/jaba.2005.85-04.
and self-control in pigeons. Journal of the Experimental Analysis of Roane, H. S., Falcomata, T. S. and Fisher, W. W. 2007. Applying the
Behavior, 66, 2949. doi:http://dx.doi.org/10.1901/jeab.1996.66-29. behavioral economics principle of unit price to DRO schedule
Kahng, S., Boscoe, J. H. and Byrne, S. 2003. The use of an escape thinning. Journal of Applied Behavior Analysis, 40, 529534.
contingency and a token economy to increase food acceptance. doi:http://dx.doi.org/10.1901/jaba.2007.40-529.
Journal of Applied Behavior Analysis, 36, 349353. doi:http://dx.doi. Roane, H. S., Fisher, W. W., Kelley, M. E., Mevers, J. L. and Bouxsein,
org/10.1901/jaba.2003.36-349. K. J. 2013. Using modified visual-inspection criteria to interpret
Kazdin, A. E. 2001. Bridging the enormous gaps of theory with functional analysis outcomes. Journal of Applied Behavior Analysis,
therapy research and practice. Journal of Clinical Child & 46, 130146. doi:http://dx.doi.org/10.1002/jaba.13.
Adolescent Psychology, 30, 5966. doi:http://dx.doi.org/10.1207/ Salend, S. J. and Henry, K. 1981. Response cost in mainstreamed
S15374424JCCP3001_7. settings. Journal of School Psychology, 19, 242249. doi:http://
Kazdin, A. E. (1982). The token economy: A decade later. Journal dx.doi.org/10.1016/0022-4405(81)90043-1.
of Applied Behavior Analysis, 15, 431445. doi:http://dx.doi. Sham, E. and Smith, T. 2014. Publication bias in studies of an applied
org/10.1901/jaba.1982.15-431. behavior-analytic intervention: An initial analysis. Journal of Applied
Kazdin, A. E. and Bootzin, R. R. 1972. The token economy: An evaluative Behavior Analysis, 47, 663678. doi:http://dx.doi.org/10.1002/
review. Journal of Applied Behavior Analysis, 5, 343372. doi:http:// jaba.146.
dx.doi.org/10.1901/jaba.1972.5-343. Skinner, B. F. 1953. Science and human behavior. New York, NY: Simon
Kazdin, A. E. and Polster, R. 1973. Intermittent token reinforcement and and Schuster.
response maintenance in extinction. Behavior Therapy, 4, 386391. Stocco, C. S. and Thompson, R. H. 2015. Contingency analysis of
doi:http://dx.doi.org/10.1016/S0005-7894(73)80118-2. caregiver behavior: Implications for parent training and future
Kelleher, R. T. and Gollub, L. R. 1962. A review of positive conditioned directions. Journal of Applied Behavior Analysis, 48, 417435.
reinforcement. Journal of the Experimental Analysis of Behavior, 5, doi:http://dx.doi.org/10.1002/jaba.206.
543597. doi:http://dx.doi.org/10.1901/jeab.1962.5-s543. Stokes, T. F. and Baer, D. M. 1977. An implicit technology of
Leblanc, L. A., Hagopian, L. P. and Maglieri, K. A. 2000. Use of a generalizationl. Journal of Applied Behavior Analysis, 10, 349367.
token economy to eliminate excessive inappropriate social doi:http://dx.doi.org/10.1901/jaba.1977.10-349.
behavior in an adult with developmental disabilities. Behavioral Stromer, R., McComas, J. J. and Rehfeldt, R. A. 2000. Designing
Interventions, 15, 135143. doi:http://dx.doi.org/10.1002/(SICI)1099- interventions that include delayed reinforcement: Implications
078X(200004/06)15:2<135:AID-BIN51>3.0.CO;2-3. of recent laboratory research. Journal of Applied Behavior
LePage, J. P., DelBen, K., Pollard, S., McGhee, M., VanHorn, L., Analysis, 33, 359371. doi:http://dx.doi.org/10.1901/
Murphy, J., Lewis, P., Aboraya, A., and Mogge, N. (2003). Reducing jaba.2000.33-359.
assaults on an acute psychiatric unit using a token economy: A 2-year Tarbox, R. S. F., Ghezzi, P. M. and Wilson, G. 2006. The effects
follow-up. Behavioral Interventions, 18, 179190. doi:http://dx.doi. of token reinforcement on attending in a young child with
org/10.1002/bin.134. autism. Behavioral Interventions, 21, 155164. doi:http://dx.doi.
Maggin, D. M., Chafouleas, S. M., Goddard, K. M. and Johnson, A. H. org/10.1002/bin.213.
2011. A systematic evaluation of token economies as a classroom Vollmer, T. R. 1994. The concept of automatic reinforcement:
management tool for students with challenging behavior. Journal Implications for behavioral research in developmental disabilities.
of School Psychology, 49, 529554. doi:http://dx.doi.org/10.1016/j. Research in Developmental Disabilities, 15, 187207. doi:http://
jsp.2011.05.001. dx.doi.org/10.1016/0891-4222(94)90011-6.
Matson, J. L. and Boisjoli, J. A. 2009. The token economy for children Zlomke, K. and Zlomke, L. 2003. Token economy plus self-monitoring to
with intellectual disability and/or autism: A review. Research reduce disruptive classroom behaviors. The Behavior Analyst Today,
in Developmental Disabilities, 30, 240248. doi:http://dx.doi. 4, 177182. doi:http://dx.doi.org/10.1037/h0100117.
org/10.1016/j.ridd.2008.04.001.

 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3 165


Scheithauer et al. A consecutive case review of token systems

Appendix A. Information about individual participants token systems


Production schedule Exchange-production schedule
Token Response Backup
# training Cost Context Target behavior Start End Start End reinforcer
1 No No Demands Absence of PB 60 s 60 s 1 Token (T) 1 T 30 s break
2 Yes Yes Demands Compliance & No PB 1 step of 1 full task 1 T 10 T 1=2min break
dem
3 No No Demands No PB 60 s 60 s 1T 1T 1=1 piece of
food
4 No No Demands No PB 60 s 60 s 1T 1T 1=1min break
5 No No Life Skills Compliance & No PB 1 dem 1 dem 1T 4T 30 s break
Academic Compliance & No PB 1 dem 1 dem 1T 10 T 30 s break
6 No No Demands Compliance & No PB 1 dem 1 dem 1T 10 T 30 s break
7 Yes No Leisure No PB 5min 515min 1T Specific times Menu
Demands Compliance & No PB 1 dem 13 dem 1T Specific times Menu
8 No No Demands No PB 5min 4min 2T 8T 1=15 s break
Breaks from No PB & Restarting Duration Duration Based on Based on Item Item Selected
Demands Work of Break of Break Item after Previous
Exchange
9 Yes No Demands Compliance & No PB 1 dem 1 dem 12 T 515 T 1=1min break
10 No No Demands Compliance & No PB 1 dem 12 dem 1T 15 T 30min break
11 No No All Contexts Low rates of PB All Day All Day 1T 1T Activities on a
menu
12 Yes Yes Academic Compliance & No PB 1 dem 1 dem 2T 5T 1=30 s break
Life Skills Compliance & No PB 1 dem 3 dem 2T 5T 1=30 s break
13 Yes Yes Demands Compliance & No PB 1 dem 1 dem 1T 5T 1=30 s break
14 No No Car Rides No PB 10min 10min End of Car End of Car Ride 1=piece of
Ride food
15 No Yes Demands Compliance & No PB 1 dem 1 dem 520 T 520 T variable time
break
16 No No Demands Compliance & No PB 1 dem 2 dem 1T 10 T 2min break
17 No No Demands Compliance & No PB 1 dem 2 dem 1T 30 T 5min break
18 Yes No Demands Compliance & No PB 1dem 4 dem 1T 6T 1=20 s break
Leisure No PB 30 s 2min 1T 6T 1=20 s break
19 Yes Yes Demands On-Task & No PB 1015 s 1015 s 5T 5T 1=25 s break
20 Yes No Demands Compliance & No PB 1 dem 1 dem 1T 1T 1=30 s break
21 No Yes Demands In Seat Checked Checked 1T 10 T 2min break
at 60 s at 60 s
22 Yes No Demands Compliance & No PB 1 dem 2 dem 1T 5T 1=4min break
23 Yes No Demands Compliance & No PB 1 dem 3 dem 2T 10 T 3min break
24 Yes Yes Demands Compliance & No PB 1 dem 10 dem 1T 10 T 1=30 s break
Leisure No PB 30 s 30 s 1T 10 T 1=30 s break

Appendix B. Results of treatment plans for individual participants

Type of Experimental Control and Percentage Reduction Values


Included in final Percentage
# treatment Experimental design First baseline Last 3 days change
1 Yes AB .07 .12 (+) 71.78
2 Yes AB .04 .00 () 90.69
3 Yes ABAB Generalization 1.73 .23 () 86.71
4 No ABAB Generalization 1.85 .65 () 64.86
5 Yes ABAB Generalization 5.10 .15 () 97.06
6 Yes ABAB Generalization 2.63 .00 () 99.87
7 Yes N/A 1.40 new injuries per day .00 new injuries per day () 100
8 Yes AB .05 .00 () 100
9 Yes MBL 1.90 .00 () 100
10 Yes ABAB Generalization 2.70 2.26 () 16.30
11 Yes ABAB Generalization 1.18 .35 () 70.34
12 Yes ABAB Generalization 1.91 .07 () 96.34
13 Yes ABAB Generalization 1.85 1.59 () 14.05
14 Yes ABAB Generalization .19 .00 () 89.47
15 Yes ABAB Generalization .76 .00 () 100
16 Yes MBL .43 .03 () 30.23
17 Yes ABAB Generalization .92 .13 () 85.87
18 Yes AB 2.56 .13 () 94.92
19 Yes ABAB Generalization 1.86 .17 () 90.86
20 No AB .19 .30 (+)57.89
21 Yes ABAB 25.56 s latency to leave seat 417.27 s latency to leave seat (+)1532.51
22 Yes ABAB Generalization .87 .06 () 93.10
23 Yes ABAB Generalization 2.68 .18 () 93.28
24 Yes ABAB & MBL 1.15 .00 () 100

Note: Unless otherwise denoted, challenging behavior in baseline and treatment are reported in responses per minute.

166 International Journal of Developmental Disabilities 2016VOL. 62 NO. 3


Copyright of International Journal of Developmental Disabilities is the property of Taylor &
Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.

S-ar putea să vă placă și