DRA Contingencies Promote Improved Tolerance To Delayed Reinforcement During FCT Compared To DRO and F

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Journal of Applied Behavior Analysis 2020, 9999, 1–14 NUMBER 9999 ()

DRA contingencies promote improved tolerance to delayed


reinforcement during FCT Compared to DRO and fixed-time
schedules
Melissa A. Drifke
May Institute

Jeffrey H. Tiger
Marquette University

Madelynn A. Lillie
The Pier Center for Autism

Following Functional Communication Training (FCT), clinicians often gradually expose newly
taught communicative responses to delayed reinforcement contingencies to prepare clients for
the normative environment in which requests are frequently reinforced after a delay. The intro-
duction of delays may result in the resurgence and maintenance of problem behavior and the
weakening of the newly trained communicative response. The current study compared delay tol-
erance with three individuals diagnosed with intellectual and developmental disabilities when
reinforcement for the functional communication response (FCR) was delivered following (a) the
passage of time, (b) the omission of problem behavior, or (c) the occurrence of an alternative
behavior. We measured delay tolerance in terms of minimizing problem behavior and
maintaining efficient FCRs. Outcomes support requiring alternative responding during delays to
attain optimal treatment results.
Key words: aggression, autism spectrum disorder, delayed reinforcement, delay tolerance,
functional communication training, problem behavior

Functional Communication Training (FCT) access the same source of reinforcement. In a


is the most thoroughly researched behavioral seminal demonstration, Carr and Durand
intervention for individuals with developmental (1985) identified individuals for whom prob-
disabilities who display problem behavior lem behavior was likely maintained by access to
(Tiger et al., 2008). Ideally informed by a func- attention and assistance with challenging
tional analysis (e.g., Iwata et al., 1982/1994), demands. They taught the requests, “Am I
FCT involves (a) arranging operant extinction doing good work?” and “I don’t understand” as
for problem behavior and (b) teaching a com- functionally equivalent alternatives to problem
municative alternative (referred to as a func- behavior. This intervention resulted in near-
tional communication response, or FCR) to zero levels of problem behavior across cases and
has been replicated with efficacious outcomes
This research was supported in part by a grant from in more than 90 studies (Tiger et al., 2008).
John and Lynn Schiek; we thank them for their generous At the onset of FCT, implementers arrange
support. reinforcement for the FCR to occur on a con-
[Correction added on 17 April 2020, after first online
publication: the affiliation of Jeffrey H. Tiger has been tinuous schedule. That is, each instance of the
corrected to ‘Marquette University’.] FCR results in immediate reinforcement.
Address correspondence to: Jeffrey H. Tiger, Ph.D., Ensuring such temporal contiguity and a strong
525 N 6th St. Milwaukee, WI 53203. Email: jeffrey.
tiger@marquette.edu contingency promotes the acquisition of the
doi: 10.1002/jaba.704 novel response (Vollmer & Hackenberg, 2001)
© 2020 Society for the Experimental Analysis of Behavior
1
2 Melissa A. Drifke et al.

and likely enhances the efficacy of extinction in reinforcer, weakening the association between
reducing problem behavior. However, such the behavior and its consequence. Second,
dense and immediate reinforcement schedules decreasing the density of reinforcement for the
cannot be maintained in normative environ- FCR by programming delays may result in the
ments when parents or teachers implement resurgence of problem behavior (Fuhrman
interventions. Even when caregivers can et al., 2016; Volkert et al., 2009). Third, given
respond to each instance of an FCR, reinforce- a chain of events in which (a) an FCR occurs,
ment delivery is likely delayed (e.g., when a (b) a delay is initiated followed by problem
parent must end a phone call before delivering behavior, and (c) reinforcement is delivered,
attention following an FCR). The introduction the occurrence of problem behavior is closer in
of delays between the FCR and the delivery of time to the delivery of reinforcement than the
reinforcement can result in multiple deleterious FCR that initiated the delay. Thus, there is a
effects, including the resurgence and mainte- possibility of problem behavior being adventi-
nance of problem behavior and the weakening tiously reinforced and therefore maintaining.
of the FCR (e.g., Fisher et al., 2000; Hagopian Finally, the absence of reinforcement following
et al., 1998; Hanley et al., 2001). an initial FCR may result in a burst of
Because delays are unavoidable, researchers responding (i.e., repeated instances of the FCR).
have focused on exposing client behavior to If these FCRs persist, the arranged contingency
delayed-reinforcement contingencies as part of may inadvertently strengthen repeated requesting
the therapeutic process. Specifically, researchers (Hanley et al., 2001). Some caregivers may con-
have attempted to create reinforcement histo- sider this repeated requesting as a form of nag-
ries during therapy that promote treatment ging (also an undesirable outcome).
maintenance in the normative environment. In response to these challenges associated
The predominant approach to developing this with time-based, progressive-delay fading,
“delay tolerance” has been a progressive, time- Ghaemmaghami et al. (2016) evaluated an
based, delay-fading model in which a therapist alternative fading process, which they termed a
introduces an initially brief, fixed-time delay contingency-based progressive delay. This pro-
initiated by the occurrence of the FCR and ter- cess was individualized for each participant, but
minated with the delivery of the functional involved (a) implementing reinforcement delays
reinforcer, while problem behavior results in no following only a subset of FCRs, (b) requiring
programmed consequence. Researchers then grad- participants to engage in a tolerance response
ually extend the duration of this delay. For following the FCR, which also resulted in
instance, Fisher et al. (2000) progressively immediate reinforcement on a subset of trials,
increased the delay to reinforcement for an FCR and (c) prolonging reinforcement delays follow-
from 0 s to 30 s. Although there have been a few ing the occurrence of problem behavior, such
successful demonstrations of this model, several that reinforcement was only delivered following
studies have shown that problem behavior is the omission of problem behavior. This differ-
likely to re-emerge and the FCR is more likely ential reinforcement of other behavior (DRO)
to decrease as delay durations increase (e.g., contingency would likely suppress problem
Hagopian et al., 1998; Hanley et al., 2001). behavior that resurges during delays by ensur-
An analysis of the programmed contingency ing it did not result in adventitious reinforce-
during time-based delay fading may elucidate ment. For some participants, the researchers
why treatment effects are fleeting. First, the also arranged differential reinforcement of alter-
introduction of a delay disrupts the continuity native behavior (DRA) contingencies in which
between the FCR and the delivery of the they delivered reinforcement when participants
Delay-Fading Comparison 3

engaged with an alternative task during delay levels of problem behavior when reinforcers for
periods. This DRA contingency would likely problem behavior were delayed up to 10 min
also suppress problem behavior by strengthen- without prior exposure to delay fading.
ing an appropriate competing response, thereby The current study (a) compared independent
decreasing the likelihood of adventitiously rein- DRO and DRA contingencies during reinforce-
forcing other inappropriate behavior ment delays relative to time-based delays based
(e.g., repeated occurrences of the FCR) at the upon reductions in problem behavior, mainte-
end of a delay period. Across participants, nance of the FCR prior to delays, and minimiza-
contingency-based, progressive delays were tion of the FCR during delays, and (b) assessed
associated with lower levels of problem behav- the necessity of delay fading in conjunction with
ior and more efficient levels of FCRs relative to these contingencies. The latter goal was achieved
time-based, progressive delays. That is, arrang- by conducting terminal-delay probes prior to fad-
ing DRO contingencies for problem behavior ing when no alternative materials were available.
and DRA contingencies for engagement
resulted in greater delay tolerance.
Method
Although the combination of these contingen-
cies improved delay tolerance for these partici- Participants and Setting
pants, it was not clear from Ghaemmaghami Three children with intellectual and develop-
et al. (2016) if arranging both contingencies was mental disabilities, referred for the assessment
necessary to achieve therapeutic outcomes. That and treatment of problem behavior, partici-
is, arranging DRO contingencies may have been pated in this study. Logan was an 11-year-old
sufficient to eliminate problem behavior by boy diagnosed with autism spectrum disorder
reducing the likelihood of adventitious reinforce- (ASD) and attention-deficit/hyperactivity disor-
ment. If this were the case, arranging DRA con- der (ADHD). His target behaviors included
tingencies would require caregivers to maintain aggression and property destruction. Andy was
access to materials and to prompt engagement a 14-year-old boy diagnosed with ASD,
superfluously after the client is discharged. Alter- ADHD, and schizencephaly whose target
natively, prompting and reinforcing engagement behavior was property destruction. He also
with materials (i.e., DRA) may have competed engaged in pica, which was treated separately
with problem behavior during delays, rendering from this study. Tanner was a 9-year-old boy
the timing of DRO intervals unnecessary. DRO diagnosed with moderate intellectual disability.
contingencies may be particularly cumbersome His target behavior included aggression and
in that they require continuous monitoring of property destruction. Each participant experi-
the client to ensure integral implementation enced a functional analysis using a model simi-
(Vollmer et al., 1993). lar to Iwata et al. (1982/1994). Parents
It was also unclear if delay fading would be identified relevant topographies of attention,
necessary given the arranged contingencies in relevant leisure items, and challenging academic
Ghaemmaghami et al. (2016). Progressive delay tasks to include in test conditions via an inter-
fading is a lengthy process, and providing access view. This assessment tested the effects of indi-
to alternative activities and arranging DRO or vidual reinforcement contingencies and
DRA contingencies may have been sufficient to identified sensitivity to social positive reinforce-
promote tolerance at the terminal delay sans ment in the form of tangible items
progressive fading. For example, Sumter et al. (Logan &Tanner) or attention (Andy) as an
(in press) showed instances in which providing inclusionary criterion for this study. Therapists
access to alternative reinforcers sustained low conducted all experimental sessions in the
4 Melissa A. Drifke et al.

child’s bedroom (Logan), or in an otherwise 99.1% for problem behavior; 98.6%, 97.3%,
unoccupied living area of the home (Andy and and 94.5% for initial FCRs; 99.8%, 99.5%,
Tanner). and 99.1% for FCRs during delays; and
96.2%, 92.5%, and 84% for delay durations
for Logan, Andy, and Tanner, respectively.
Measurement and Interobserver
Agreement
Observers scored the occurrence of problem Procedures
behavior following each instance of aggressive FCT Evaluation
(i.e., hits, kicks, bites, or hair pulls) or destruc- Following a functional analysis indicating
tive (i.e., toward materials, furniture, or walls) that problem behavior was maintained by social
behavior. FCRs were scored following each positive reinforcement, therapists initiated the
unprompted instance of the utterance, “May I treatment evaluation. Prior to each 10-min ses-
have the iPad, please?” for Logan, “Can you sion, the therapist provided noncontingent
play with me, please?” for Andy, and each access to the target reinforcer for approximately
instance of a modified version of the manual 60 s. Baseline sessions began with the therapist
sign for “more” (i.e., bringing both hands disrupting access to this reinforcer. Following
together near the midline) for Tanner. each problem behavior, the therapist delivered
Observers scored delay periods using duration 30 s of access to the iPad for Logan and Tan-
measures by recording their onset after the ner, and 3 to 5 s of mild reprimands for Andy.
therapist acknowledged an FCR (stating, “In a The FCT sessions were similar to baseline
minute”) and their offset when a reinforcer was except that 5 s after disruption of the rein-
delivered. For the purpose of data analysis, the forcer, the therapist vocally prompted partici-
occurrence of problem behavior during each pants to emit an FCR, which resulted in 30 s
session was converted into a rate (responses per of access to the iPad for Logan and Tanner,
minute). The occurrences of FCRs were distin- and 30 s of conversational attention for Andy.
guished as those occurring prior to the onset of Problem behavior no longer resulted in rein-
a delay, or during a delay period (FCR during forcement during FCT sessions. Baseline and
delays). FCT sessions were compared in a non-
Experimenters assessed interobserver agree- concurrent multiple baseline design across par-
ment (IOA) by having a second observer simul- ticipants with reversals for Logan and Tanner.
taneously but independently collect data on
21%, 54%, and 18% of sessions for Logan, Delay-Tolerance Evaluation
Andy, and Tanner, respectively. IOA was calcu- Based upon consultation with each partici-
lated using a proportional agreement method pant’s family, therapists set goal delay durations
by comparing observers’ records within 10-s (i.e., 5 min for Logan and Andy, and 2 min for
intervals. Intervals scored identically were given Tanner) and identified target tasks to include
a score of 1 and intervals scored nonidentically during DRA periods (i.e., writing words for
were given a proportional agreement score by Logan, matching playing cards for Andy, and
dividing the smaller number of responses pointing to letters and numbers for Tanner).
(or durations) by the larger number of Therapists then conducted a brief, informal
responses (or durations). Scores were then pilot exposure by asking each child to complete
summed across intervals, divided by the total the task to determine the mean duration of
number of intervals, and converted into a per- item completion. The target duration
centage. IOA averaged 98.3%, 97.2%, and (e.g., 300 s) was then divided by the mean
Delay-Fading Comparison 5

duration of item completion (e.g., 30 s) to Time-, DRO-, and DRA-delay probes were
arrive at a target number of tasks (e.g., 10) to introduced sequentially for Logan and Andy to
the complete equal to the target duration. determine if the addition of contingencies dur-
No Delay (ND). Therapists conducted ND ing delays would enhance delay tolerance with-
sessions identically to FCT sessions. That is, out delay fading. Between each terminal probe
each 10-min session began by removing the phase, therapists conducted additional ND ses-
identified reinforcer, which was returned sions to re-strengthen the relevant FCR before
immediately following the occurrence of an exposing it to the subsequent delay probe. Tan-
FCR. Problem behavior resulted in no ner was exposed to Time-delay probes only as
programmed consequence. These sessions the addition of contingencies did not result in
served to ensure that the FCR had a recent additional delay tolerance for Logan or Andy.
history of reinforcement prior to exposure to Delay-Fading Comparison. Due to elevated
each delay condition. levels of problem behavior across all terminal
Terminal Probes. Therapists then conducted probes (Logan and Andy) and decreased levels of
probes of responding at terminal delays to deter- FCRs (all participants), therapists decreased delay
mine levels of the FCR and problem behavior. durations across all three delay-contingency con-
These sessions were similar to ND sessions except ditions and gradually extended (faded) delays as
that following an FCR, the therapist initiated a participants demonstrated delay tolerance. The
delay period by stating, “In a minute.” During initial delay value in the Time-based and DRO-
Time-delay probes, the therapist set a timer for based delay conditions was set as equivalent to
300 s (Logan and Andy) or 125 s (Tanner), the estimated duration of completing one task in
ignored all instances of behavior until the timer the DRA-based delay condition (i.e., equated
sounded, and delivered reinforcement for 30 s after across conditions). Sessions of each of three delay
the timer sounded. This timer remained in the procedures were alternated in a multielement
therapist’s possession; the digital display was not design in which delay durations increased inde-
visible to participants. During DRO probes, the pendently across each condition following three
therapist set a timer for 300 s (Logan and Andy) consecutive exposures to that contingency with
and ignored all behavior, except to restart the timer problem behavior occurring at less than 10% of
following an instance of problem behavior. When its baseline level.
the timer sounded, the therapist delivered rein- Time-Based Delay Fading. Following an
forcement for 30 s. Similar to Time-delay probes, FCR, the therapist indicated a delay
the display was not visible to the participant. Dur- (e.g., saying, “In a minute”), set a timer (ini-
ing DRA probes, the therapist presented task tially 15 s for Logan, 27 s for Andy, and 5 s for
materials and prompted completion of the task Tanner), and at the expiration of the timer,
using a three-step prompting procedure until the delivered access to reinforcement for 30 s. The
participant wrote 17 words (Logan). During therapist did not otherwise interact with partic-
Andy’s DRA probes, the therapist presented ipants during delays and problem behavior did
10 cards to match but did not provide any follow- not affect the duration of the delay. For Logan,
up prompts to engage with the items. Andy’s delays progressed from 15 s, to 30 s, 45 s, 60 s,
problem behavior was maintained by attention, so 75 s, 141 s, 177 s, 210 s, 264 s, and 300 s.1
therapists did not implement additional prompting The estimated duration for Logan was 17.5 s,
following noncompliance and problem behavior to
avoid adventitiously reinforcing these behaviors. 1
The estimated duration for Logan was 17.5 s, which
Following task completion, the therapist delivered was rounded to the nearest 15-s increment for the initial
reinforcement for 30 s. five fading steps.
6 Melissa A. Drifke et al.

which was rounded to the nearest 15-s incre- goal of 5 min. Thus, therapists added addi-
ment for the initial five fading steps. For Andy, tional fading steps of 14, 18, 23, 27, 32, 36,
delays progressed from 27 s, to 55 s, 82 s, and 41 tasks. For Tanner, the work require-
109 s, 136 s, 163 s, 190 s, 218 s, 245 s, 273 s, ment progressed from 1, to 5, 10, 15, 20, and
and 300 s. For Tanner, delays progressed from 25 tasks.
5 s, to 25 s, 50 s, 75 s, 100 s, and 125 s.
DRO-Based Delay Fading. These sessions Experimental Design
were the same as the Time-based delay fading The present study compared the three delay
sessions except that every instance of problem procedures in a multielement experimental
behavior resulted in a resetting of the delay. design. After determining the most efficacious
For Logan, delays progressed from 15 s, to condition for attaining delay tolerance, we ter-
30 s, 45 s, 60 s, 75 s, 141 s, 177 s, 210 s, minated the multielement comparison and con-
264 s, and 300 s. For Andy, delays progressed ducted additional sessions of that condition to
from 27 s, to 55 s, 82 s, 109 s, 136 s, 163 s, assess maintenance without potential carry-over
190 s, 218 s, 245 s, 273 s, and 300 s. For Tan- effects (i.e., what Barlow & Hayes, 1979,
ner, delays progressed from 5 s, to 25 s, 50 s, referred to as independent verification).
75 s, 100 s, and 125 s. These values were iden-
tical to the time-based delay conditions such
that any improvement or worsening of problem Results
behavior under these conditions relative to the Figure 1 shows the results of the FCT evalu-
Time-based delay could be ascribed to the pres- ation for Logan, Andy, and Tanner. Logan (top
ence of the DRO contingency. panel of Figure 1) engaged in a mean rate of
DRA-Based Delay Fading. Following an 1.8 instances of problem behavior per minute
FCR, the therapist indicated a delay during baseline, which reduced to zero levels
(e.g., saying, “In a minute”) and presented during FCT, returned to a mean rate of 1.7
materials as described during the DRA terminal during a reversal to baseline, and reduced again
probe. For Logan and Tanner, compliance with to zero levels during a return to FCT. Andy
an instruction following a vocal or model engaged in a mean rate of 14.9 instances of
prompt resulted in reinforcement; for Andy, problem behavior per minute during baseline,
task completion resulted in reinforcement. which reduced to 1.1 per minute during FCT.
Problem behavior resulted in continued Tanner engaged in a mean rate of 2.0 instances
prompting for Logan and Tanner and resulted of problem behavior per minute during baseline
in no programmed consequence for Andy. Ini- which reduced to near-zero levels during FCT,
tially, one compliant response or task com- returned to a mean rate of 2.3 during a reversal
pleted produced a 30-s reinforcement period; to baseline, and decreased again to near-zero
this work requirement was then gradually levels during a return to FCT. Thus, the initial
increased across sessions. For Logan, the work exposure to FCT resulted in reductions of
requirement to produce reinforcement prog- problem behavior by 96%, 93%, and 99% of
ressed from 1, to 2, 3, 4, 5, 8, 10, 12, 15, and initial baseline levels for Logan, Andy, and
17 tasks. For Andy, the work requirement Tanner, respectively.
progressed from 1, to 2, 3, 4, 5, 6, 7, 8, 9, and The results of Logan’s delay-tolerance evalua-
10 tasks. Once Andy reached the terminal goal tion are shown in Figure 2. When therapists
of 10 tasks, his response efficiency increased delivered reinforcement immediately following
considerably such that the duration required to Logan’s FCR (first, third, fifth, and seventh
complete 10 tasks was consistently less than the phases labeled as ND), problem behavior
Delay-Fading Comparison 7

Figure 1 maintained at zero levels (top panel of


Results of the FCT Evaluation for Logan, Andy, and Tanner. Figure 2) and the FCR occurred at optimal
levels (second panel of Figure 2). When
exposed to terminal-delay probes, Logan
engaged in bursts of problem behavior and
ceased to engage in the FCR during Time-
based, DRO-based, and DRA-based delay con-
ditions (second, fourth, and sixth phases,
respectively). These data indicated delay fading
would be necessary to maintain lower levels of
problem behavior and higher levels of the FCR.
The fading comparison appears in the eighth
phase of the evaluation. Problem behavior
maintained at near-zero levels across each of the
delay fading conditions, except for two DRA-
based delay sessions. FCRs also maintained
across each of the delay-fading conditions. In
addition to the overall rates of FCRs, we also
examined FCRs that occurred during delay
periods, which were elevated during Time-
based and DRO-based delay conditions but
remained at near-zero levels in the DRA-based
delay condition (third panel of Figure 2). After
delay fading, terminal duration criteria were
achieved with Time-based, DRO-based, and
DRA-based delay conditions.
The programmed work requirements in the
DRA condition were selected in an attempt to
yoke the duration of the delay values
programmed for the equivalent fading steps of
the Time-based and DRO-based delay condi-
tions. To determine whether these values were
equal, obtained delay intervals were calculated
from each session and compared to the
programmed durations of each step. Twenty-
two of 47 sessions met or exceeded the
programmed delay values (47%). The mean
difference between the programmed and the
obtained delay values across steps was 9.5 s.
Dividing the difference between obtained and
Note. The rate of problem behavior (PB) is shown by
programmed delays by the duration of the
open squares and the rate of the FCR is shown by the
filled squares programmed delay yielded a mean deviation of
obtained delays at 4% above the programmed
value (range, 25% underestimation to 100%
overestimation). These data indicate that DRA
8 Melissa A. Drifke et al.

Figure 2
Results of the Delay-Fading Evaluation for Logan.
ND TP ND TP ND TP ND Fading Comparison
Problem Behavior per Min
25 1 2 3 4 5 6 7 8 9 10
Time
Fading
1 2 3 4 5 6 7 8 9 10
DRO
DRA Step
20 1 2 3 4 5 6 7 8 9 10

15
10
5
0
16
FCR (per min)

12

0
FCRs during Delay (per min)

5.0
4.0
3.0
2.0
1.0
0.0
50 100 150
Sessions
Note. Problem behavior (top panel), FCRs (middle panel), and FCRs during delays (bottom panel) across No Delay
(ND), Terminal Probe (TP) and Fading Comparison conditions. The fading progression (i.e., steps) during the fading
comparison are indicated via timelines above the top panel.

conditions were not associated with briefer behavior (first, third, fifth, and seventh phases).
delays and this should not be considered a con- When exposed to terminal delays (300 s), Andy
found for Logan’s evaluation. Given the differ- engaged in bursts of problem behavior regard-
ential efficacy of DRA contingencies at less of whether those delays ended based upon
maintaining low levels of problem behavior and time, DRO, or DRA contingencies. Therefore,
FCRs during delays, therapists conducted an Andy’s behavior was exposed to the delay-
additional 23 sessions with DRA-based delays fading comparison. Exposure to the Time-
yielding near-zero levels of problem behavior based and DRO-based delays resulted in imme-
with delays averaging near 5 min. diate bursts of problem behavior, whereas
Figure 3 shows the results of Andy’s delay- DRA-based delays were associated with an
tolerance evaluation. When therapists delivered immediate reduction to zero levels. The differ-
reinforcement immediately following an FCR, ences between these conditions were most pro-
Andy exhibited near-zero levels of problem nounced during the first 10 exposures to each
Delay-Fading Comparison 9

Figure 3
Results of the Delay-Fading Evaluation For Andy.
Terminal
Probe
Problem Behavior per Min ND Fading Comparison
10 1 2 3 4 5 Time
Fading
DRO
1 2 3 4 5
DRA Step
8 1 2 3 4 5 6 7 8 9 10 11 12 13 14

6
4
2
0

2.0
FCR (per min)

1.5

1.0

0.5

0.0
FCRs during Delay (per min)

1.0
0.8
0.6
0.4
0.2
0.0
30 60 90 120 150 180 210
Sessions
Note. Problem behavior (top panel), FCRs (middle panel), and FCRs during delays (bottom panel) across No Delay
(ND), Terminal Probe (TP) and Fading Comparison conditions. The fading progression (i.e., steps) during the fading
comparison are indicated via timelines above the top panel.

condition. Following this initial burst, problem DRA-delay value met or exceeded the
behavior persisted at low, variable rates across programmed value in only one of 32 sessions
conditions for the remainder of the evaluation. (3% of sessions). The mean difference between
Overall, the FCR occurred at similar rates the programmed and obtained values was a
across conditions with low levels during delays. 98-s underestimation or 63% under the
Andy met the completion criterion for the programmed value. At the terminal goal of
DRA condition at session 192 and had not yet 10 tasks, Andy required only about 3 min on
met criteria in the Time-based delay (at step average (programmed goal of 5 min). There-
5, a delay of 109 s) or DRO-based delay fore, therapists continued to increase the num-
(at step 6, a delay of 163 s) conditions. ber of tasks required to terminate delays under
Through the first 10 fading steps, the obtained DRA-based delay conditions. Delay durations
10 Melissa A. Drifke et al.

continued to increase until they met the 300-s The results of Tanner’s delay-tolerance com-
goal. Problem behavior continued at similar parison are depicted in Figure 4. When thera-
low, but variable levels. Although there was a pists delivered reinforcement immediately
slight upward trend in problem behavior at the following an FCR, Tanner exhibited near-zero
end of this evaluation, Andy’s disruptive behav- levels of problem behavior (first and third
ior was still reduced by greater than 90% of his phases) and high levels of FCR. When exposed
initial baseline (M = 14.9 rpm). His parents to terminal delays (125 s), Tanner did not
were satisfied with this reduction and requested resume problem behavior, but the FCR
that therapists focus their efforts on addressing decreased to zero levels under delayed-
his pica. reinforcement conditions. When exposed to the

Figure 4
Results of the Delay-Fading Evaluation for Tanner.
Terminal
Probe
ND ND Fading Comparison
Problem Behavior per Min

5.0 1 2 3 4 5 6
Time
Fading
DRO
1 2 3 4 5 6 Step
4.0 1 2 3 4
DRA

3.0
2.0
1.0
0.0

2.0
FCR (per min)

1.5

1.0

0.5

0.0
FCRs during Delay (per min)

1.0
0.8
0.6
0.4
0.2
0.0
10 30 50 70
Sessions
Note. Problem behavior (top panel), FCRs (middle panel), and FCRs during delays (bottom panel) across No Delay
(ND), Terminal Probe (TP) and Fading Comparison conditions. The fading progression (i.e., steps) during the fading
comparison are indicated via timelines above the top panel.
Delay-Fading Comparison 11

delay-fading comparison, Tanner’s problem minimizing the occurrence of problem behavior


behavior was low across delay conditions (one relative to progressive time delays.
DRA session occasioned problem behavior Unlike during terminal probes, FCRs
commensurate with prefading, terminal-probe maintained at similar levels as in no-delay condi-
levels). FCRs maintained across each of the tions across participants and across conditions
delay conditions, but DRA-based delays were during the initial stages of the fading comparison.
associated with fewer FCRs during delay The overall rates of the FCRs declined as delays
periods than were DRO-based delays or Time- increased, but these response rates never
based delays. Tanner first met his terminal approached zero, as was the case during terminal
delay of 125 s in the Time-based delay condi- probes. Perhaps this maintenance of the FCR
tion. At that time, fading progressed to Step was sufficient to minimize resurgence of problem
5 (100-s delay) in the DRO-based delay condi- behavior across conditions, notably for Logan
tion and Step 3 (10 tasks) in the DRA-based and Tanner who exhibited very low levels of
delay condition. Due to the slightly elevated problem behavior across the fading evaluation
levels of problem behavior during DRA-based with or without the introduction of DRO and
delays, therapists conducted an additional DRA contingencies.
12 Time-based delay sessions. Problem behav- For Andy, there was more variability in the
ior and the FCR during delays reduced to near occurrence of the FCR during his initial exposure
zero levels. to progressive Time-delay conditions in the fad-
ing comparison (e.g., he did not engage in a sin-
gle FCR in his first two Time-delay sessions).
Thus, it is not surprising that he engaged in rela-
Discussion tively higher levels of problem behavior during
One of the primary concerns raised by these initial sessions. Of the three participants,
arranging delayed reinforcement for communi- his data set may provide the best insight into the
cative responses following FCT is that tempo- impact of DRO and DRA contingencies on
rally separating the delivery of reinforcement problem behavior during these delays. If Andy’s
from the FCR may generate conditions that problem behavior was contacting adventitious
weaken the FCR and create a resurgence of reinforcement at the conclusion of the delay
problem behavior. This pattern of increased intervals, then the inclusion of DRO contingen-
problem behavior, despite programmed extinc- cies would presumably disrupt the contiguity
tion for these responses, was apparent for both between problem behavior and the onset of the
Logan and Andy during their initial exposure reinforcement period, and thus reduce problem
to delayed contingencies during terminal pro- behavior. However, the DRO contingency did
bes; increased problem behavior was not not appear to have any reductive benefits for
observed during Tanner’s terminal probe Andy relative to progressive time-delays. Both
despite a weakening of the FCR. Although time-delay and DRO conditions evoked similar
resurgence is a transient phenomenon (Briggs levels of problem behavior early in the compari-
et al., 2018), the delivery of reinforcement at son and problem behavior progressively decreased
the end of a fixed-time delay interval may pro- to similar, lower levels.
duce adventitious reinforcement of problem There are at least two possible explanations
behavior, and thus, maintenance. The current for the similarities in responding between the
study evaluated the effects of arranging isolated Time-delay and DRO-delay conditions. First,
DRO and DRA contingencies during progres- and perhaps most simply, it may be that
sive delay fading in sustaining the FCR and despite the potential for temporal contiguity,
12 Melissa A. Drifke et al.

problem behavior was evoked, but not adventi- if those replacements were not the initial “problem
tiously reinforced under time-delay arrange- behavior.”
ments. The progressive decrease in problem One such form of problem behavior
behavior during both Time-delay and DRO- observed in the current study was the occur-
delay conditions suggests that the process of rence of repeated requesting via the FCR dur-
extinction was taking place. Second, it is possi- ing delays. Across participants, these requests
ble that despite programmed discriminative were elevated during both time-based and
stimuli (therapists wore unique colored shirts DRO-based delay conditions, relative to DRA
in each condition), the programmed contin- conditions. Although such repeated requesting
gencies across conditions were not sufficiently is not as severe a problem as aggression and
salient to promote discriminated responding. property destruction, these response patterns
Thus, different behavioral responses to have been noted (Ghaemmaghami et al., 2016;
programmed contingencies may have been Hanley et al., 2001; Landa & Hanley, 2016) as
obscured due to carry over effects in which sources of concern for clinicians and families
problem behavior extinguished across condi- and should be accounted for in intervention
tions. We did not attempt to separate these planning. To the extent that one intervention
effects, in part due to the clear reductions in may reduce problem behavior, but produce a
problem behavior when DRA contingencies novel alternative problem behavior, the social
were arranged. validity of that intervention is likely limited.
The presumed logic supporting DRA-based Despite the value of including DRA contin-
delay contingencies is that following the FCR, gencies in delays, this approach may raise some
DRA contingencies reinforce a chain of appropri- concerns as well. For example, Tanner engaged
ate behavior that is incompatible with problematic in slightly higher levels of problem behavior
behavior. For Andy, the card-matching response during DRA-based delays, relative to DRO or
was strengthened to the exclusion of aggressive time-based delays. It is reasonable to assume
and destructive behavior and continued to com- that presenting nonpreferred tasks may evoke
pete with problem behavior as the delay duration problem behavior during delays for some indi-
increased. During all delay conditions, a partici- viduals. We did not attempt to identify partici-
pant does not simply stop behaving during delay pants’ preferences for the tasks included in the
intervals. Even in cases in which the individual DRA condition and instead relied on parental
does not engage in overt problem behavior during requests. Problem behavior may be further
delays, he or she will engage in some behavior that minimized and engagement enhanced by
may be reinforced by the end of the delay interval. including high-preference mediating tasks (Call
Time-based and DRO-based delays do not specify et al., 2009). However, many naturally occur-
what behavior must occur during this delay; thus, ring delays may also require participants to
these contingencies may select that which occurs engage with nonpreferred activities
due to happenstance (i.e., that which is evoked (e.g., putting away one’s iPad to complete a
during delays; see Sidman, 1960, pp. 341-392 for nighttime routine). The role of task preference
a discussion of adventitious reinforcement during and caregiver acceptability of alternative tasks
delayed-reinforcement contingencies). It may be in promoting delay tolerance will be an impor-
the case that some individuals engage in desirable tant avenue for future research.
mediating behaviors when faced with a delay Ghaemmaghami et al. (2016) also intro-
(e.g., an adult reads a magazine while waiting at a duced DRO and DRA contingencies during
doctor’s office), but it may be that some individ- delays to reinforcement following FCT, but did
uals fill the gap with less desirable behaviors, even not attempt to separate the impact of DRO
Delay-Fading Comparison 13

and DRA contingencies. The results of the cur- have described other schedule thinning proce-
rent study suggest that progressive delay fading dures, such as arranging a multiple schedule
is sufficient to maintain FCRs when delays are (e.g., Greer et al., 2016; Hagopian et al., 2011;
introduced, and DRA contingencies promote Saini et al., 2016). If the clinical goal of inter-
greater engagement in appropriate behavior vention is solely to thin the schedule of rein-
while competing with target and nontarget forcement for an FCR, then the two
forms of problem behavior relative to time- procedures may be viewed similarly. However,
based and DRO-based delays. We did not if the goal is to prepare clients and their care-
include combined DRO and DRA conditions givers to maintain low levels of problem behav-
similar to Ghaemmaghami et al. Although the ior, multiple schedule and delay tolerance
results do not demonstrate behavioral improve- procedures address two very distinct challenges
ments when examining DRO in isolation, it is to behavioral interventions.
possible that DRO contingencies may enhance Multiple schedule procedures have most com-
the delay tolerance produced by DRA. This monly addressed challenges in which a reinforcer
remains an important area for future research. either is or is not available. For instance, Hanley
The current procedures also differed from et al. (2001) arranged a multiple schedule that
Ghaemmaghami et al. (2016) in that the alter- alternated between conditions in which an FCR
native task was available only during DRA con- would result in immediate reinforcement or would
ditions of the current study, but remained result in extinction. Each of these conditions was
available throughout both Time-Based and correlated with a salient signal (e.g., a colored card
Contingency-Based progressive delay conditions on the table). The intervention was considered
of Ghaemmaghami et al. It is possible that pro- effective when (a) problem behavior remained low
viding an alternative activity during delays across both conditions, (b) appropriate FCRs
alone may be sufficient to promote delay toler- occurred when the positive signal was presented,
ance (Austin & Tiger, 2015). Thus, benefits of and (c) FCRs did not occur when the negative
DRA-based delays observed in the current signal was presented. However, when the positive
study resulted from the availability of this com- signal was presented, caregivers were required to
peting activity rather than the contingencies provide immediate reinforcement following each
arranged to promote engagement. FCR. In practical application, caregivers may still
Ghaemmaghami et al. did not see treatment inadvertently impose delays to reinforcement
effects associated with providing material access delivery during periods in which a positive signal
without supporting contingencies with their is presented. It is not clear how likely integrity
participants, but such results are particularly with such contingencies can be maintained out-
likely if high-preference activities are identified side of a structured clinical evaluation or how
for delivery during delays. Future researchers integrity failures may impact the clinical outcomes
should include material access during delay of multiple schedule interventions. Although a
probes prior to arranging contingencies. larger quantity of research has shown that multiple
At its onset, FCT involves arranging dense schedules can produce discriminated response pat-
schedules of reinforcement to support a novel terns, this does not obviate the need to include
communicative response. However, the success- delay tolerance training.
ful transition outside of a clinic environment
will require thinning the schedule of reinforce-
ment for that communicative response while REFERENCES
maintaining low levels of problem behavior. In Austin, J. E., & Tiger, J. H. (2015). Providing alternative
addition to progressive time delays, researchers reinforcers to facilitate tolerance to delayed
14 Melissa A. Drifke et al.

reinforcement following functional communication Analysis, 31, 211-235. https://doi.org/10.1901/jaba.


training. Journal of Applied Behavior Analysis, 48, 1998.31-211
663–668. https://doi.org/10.1002/jaba.215 Hanley, G. P., Iwata, B. A., & Thompson, R. H. (2001).
Barlow, D. H., & Hayes, S. C. (1979). Alternating treat- Reinforcement schedule thinning following treatment
ments design: One strategy for comparing the effects with functional communication training. Journal of
of two treatments in a single subject. Journal of Applied Behavior Analysis, 34, 17-38. https://doi.org/
Applied Behavior Analysis, 12, 199-210. https://doi. 10.1901/jaba.2001.34-17
org/10.1901/jaba.1979.12-199 Iwata, B. A., Dorsey, M. F., Slifer, K. J.,
Briggs, A. M., Fisher, W. W., Greer, B. D., & Bauman, K. E., & Richman, G. S. (1994). Toward a
Kimball, R. T. (2018). Prevalence of resurgence of functional analysis of self-injury. Journal of Applied
destructive behavior when thinning reinforcement Behavior Analysis, 27, 197-209. https://doi.org/10.
schedules during functional communication training. 1901/jaba.1994.27-197
Journal of Applied Behavior Analysis, 51, 620–633. Landa, R., & Hanley, G. P. (2016). An evaluation of
https://doi.org/10.1002/jaba.472 multiple-schedule variations to reduce high-rate
Call, N. A., Pabico, R. S., & Lomas, J. E. (2009). Use of requests in the picture exchange communication sys-
latency to problem behavior to evaluate demands for tem. Journal of Applied Behavior Analysis, 49,
inclusion in functional analyses. Journal of Applied 388–393. https://doi.org/10.1002/jaba.285
Behavior Analysis, 42, 723–728. https://doi.org/10. Saini, V., Miller, S. A., & Fisher, W. W. (2016). Multiple
1901/jaba.2009.42.723 schedules in practical application: Research trends
Carr, E. G., & Durand, V. M. (1985). Reducing behavior and implications for future investigation. Journal of
problems through functional communication train- Applied Behavior Analysis, 49, 421–444. https://doi.
ing. Journal of Applied Behavior Analysis, 18, org/10.1002/jaba.300
111–126. https://doi.org/10.1901/jaba.1985.18-111 Sidman, M. (1960). Tactics of scientific research. Authors
Fisher, W. W., Thompson, R. H., Hagopian, L. P., Cooperative Inc.
Bowman, L. G., & Krug, A. (2000). Facilitating tol- Sumter, M. E., Gifford, M. R., Tiger, J. H.,
erance of delayed reinforcement during functional Effertz, H. M., & Fulton, C. J. (in press). Providing
communication training. Behavior Modification, 24, noncontingent, alternative, functional reinforcers dur-
3–29. https://doi.org/10.1177/145445500241001 ing delays following functional communication train-
Fuhrman, A. M., Fisher, W. W., & Greer, B. D. (2016). ing. Journal of Applied Behavior Analysis.
A preliminary investigation on improving functional Tiger, J. H., Hanley, G. P., & Bruzek, J. (2008). Func-
communication training by mitigating resurgence of tional communication training: A review and practi-
destructive behavior. Journal of Applied Behavior cal guide. Behavior Analysis and Practice, 1, 16-23.
Analysis, 49, 884–899. https://doi.org/10.1002/ https://doi.org/10.1007/BF03391716
jaba.338 Volkert, V. M., Lerman, D. C., Call, N. A., & Trosclair-
Ghaemmaghami, M., Hanley, G. P., & Jessel, J. (2016). Lasserre, N. (2009). An evaluation of resurgence dur-
Contingencies promote delay tolerance. Journal of ing treatment with functional communication train-
Applied Behavior Analysis, 49, 548–575. https://doi. ing. Journal of Applied Behavior Analysis, 42,
org/10.1002/jaba.333 145–160. https://doi.org/10.1901/jaba.2009.42-145
Greer, B. D., Fisher, W. W., Saini, V., Owen, T. M., & Vollmer, T. R., & Hackenberg, T. D. (2001). Reinforce-
Jones, J. K. (2016). Functional communication train- ment contingencies and social reinforcement: Some
ing during reinforcement schedule thinning: An anal- reciprocal relations between basic and applied
ysis of 25 applications. Journal of Applied Behavior research. Journal of Applied Behavior Analysis, 34,
Analysis, 49, 105–121. https://doi.org/10.1002/ 241–253. https://doi.org/10.1901/jaba.4001.34-241
jaba.265 Vollmer, T. R., Iwata, B. A., Zarcone, J. R.,
Hagopian, L. P., Boelter, E. W., & Jarmolowicz, D. P. Smith, R. G., & Mazaleski, J. L. (1993). The role of
(2011). Reinforcement schedule thinning following attention in the treatment of attention-maintained
functional communication training: Review and rec- self-injurious behavior: Noncontingent reinforcement
ommendations. Behavior Analysis in Practice, 4, 4-16. and differential reinforcement of other behavior. Jour-
doi/10.1007/BF03391770 nal of Applied Behavior Analysis, 26, 9-21. https://doi.
Hagopian, L. P., Fisher, W. W., Sullivan, M. T., org/10.1901/jaba.1993.26-9
Acquisto, J., & LeBlanc, L. A. (1998). Effectiveness
of functional communication training with and with- Received March 14, 2018
out extinction and punishment: A summary of Final acceptance January 11, 2020
21 inpatient cases. Journal of Applied Behavior Action Editor, John Borrero

You might also like