ERIC EJ899108: Assessing the Value of Choice in a Token System

2010·0.13 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2010, 43, 553–557 NUMBER3 (FALL2010)
ASSESSING THE VALUE OF CHOICE IN A TOKEN SYSTEM
SANDEEP K. SRAN
UNIVERSITYOFTHEPACIFIC
AND
JOHN C. BORRERO
UNIVERSITYOFMARYLAND,BALTIMORECOUNTY
Responding of 4 children was assessed under conditions in which (a) no programmed
contingencieswerearrangedfortargetbehavior,(b)respondingproducedtokensthatcouldbe
exchanged for a single highly preferred edible item, and (c) responding produced a token that
could be exchanged for a variety of preferred edible items. After assessing the effects of these
contingencies,thepreferencesof3participantswereassessedusingaconcurrent-chainsschedule.
Preference for the opportunity to choose from the same or qualitatively different edible items
varied across participants, and findings were generally consistent with those of Tiger, Hanley,
andHernandez(2006).
Keywords: choice, concurrent chains,preference, token reinforcement
_______________________________________________________________________________
In a series of studies, Tiger, Hanley, and descriptive rules associated with the various
Hernandez (2006) assessed children’s academic options (e.g., ‘‘If you pick this option you can
performance during conditions in which they pick one of five identical edible items; if you
could obtain one highly preferred edible item pick that option, you receive this edible item’’).
contingent on task completion (no choice) or Although practical for academic settings and
select one identicaledible item from an array of very young children, the incorporation of rules
five identical edible items (choice). These makesitmoredifficulttodistinguishcontrolby
conditions were then compared to a condition experimentallyprogrammedcontingenciesfrom
in which no edible items were arranged to control by rules (Skinner, 1953). Further,
control for the simple presentation of edible completion of the response requirement in
items. Tiger et al. evaluated the opportunity to Tiger et al.’s study resulted in immediate access
choose using a concurrent-chains evaluation to the item. In application, this may not be the
(e.g., Luczynski & Hanley, 2009), and results most practical arrangement. An alternative
showed that some children preferred the strategy might involve the presentation of
opportunity to choose and others did not. tokens,whichhavebeenshowntobridgedelays
However, in the evaluations described by Tiger between completion of the response require-
et al., experimenters presented children with ment and receipt of primary or backup
reinforcers.
Thisexperimentwascompletedinpartialfulfillmentof The purpose of the current study was to
therequirementsoftheMasterofArtsdegreebythefirst
conduct a systematic replication of the proce-
authorfrom the University of the Pacific. We thank Ken
BeauchampandFloydO’Brienfortheircommentsonan dures described by Tiger et al. (2006) in the
earlier version of this manuscript. In addition, we thank context of a token system. The study was
members of the behavior analysis laboratory at the
designed to determine if the opportunity to
University of the Pacific for their assistance with data
collectionanddata preparation. choose backup reinforcers in a token system
DirectcorrespondencetoJohnC.Borrero,Department would influence response rates associated with
of Psychology, University Maryland, Baltimore County,
academic tasks and children’s preferences for
Baltimore,Maryland21250(e-mail:[email protected]).
doi:10.1901/jaba.2010.43-553 arrangements in which the opportunity to
553
554 SANDEEP K. SRAN and JOHN C. BORRERO
choose was or was not presented. Participants ratio to a percentage. These percentages were
werenotgivenanyverbalinstructionsaboutthe averaged to generate session agreement. Inter-
different conditions. Rather, nonverbal sched- observeragreementwasassessedforatleast32%
ule-correlated stimuli were associated with the of sessions across all conditions and exceeded
various operating contingencies. In this way, 80% for all participants.
performance was based on contact with the
Procedure
arranged contingencies.
During all sessions, the target response
consisted of tracing numbers and uppercase
METHOD
andlowercaseletterswithapencil.Priortoeach
Participants, Setting, and Materials session, the experimenter gave participants the
Fourtypicallydevelopingchildren(all4years following instruction: ‘‘Here are some letters
old) enrolled in preschool participated in this and numbers to trace. You can do as many as
study. Dylan, Mira, Milo, and Luke participat- youwant,asfewasyouwant,ornoneatall.’’In
ed in the preference (Fisher et al., 1992) and addition, during the first exposure to each
reinforcer assessments; however, Dylan did not condition, the experimenter implemented a
participate in the concurrent-chains evaluation. three-prompt instructional sequence when a
Experimental sessions were conducted at each participant did not emit a response in the first
participant’s preschool in an available space in 10softhefirstsession.Attheendofthe3-min
the classroom. session, the experimenter removed the work-
Session materials included academic work- sheet and said, ‘‘Okay, you’re finished.’’
sheets (letters and numbers that could be The reinforcer assessment was conducted
traced), colored pencils, poker chips, and a using a combined reversal and multielement
smock worn by the experimenter. Each of the design. The baseline condition (A) was intro-
items was one of four different colors (see duced in the first phase. The second phase
below). The colors were included to enhance consisted of a multielement arrangement of the
discrimination. Trained undergraduate or grad- no-choice condition (B), the single-choice
uate students acted as the experimenter for all condition (C), and the varied-choice condition
the sessions. (D). During baseline, participants worked with
All sessions lasted 3 min, and data were a green worksheet. Participants received no
collected on the target response, delivery of tokens during these sessions. Instructions were
tokens, and condition selection (when partici- provided to participants at the beginning and
pants had the opportunity to choose). Approx- end of the session, as described previously.
imately three sessions were conducted per day, Prior to the introduction of the token
and data were collected manually or with conditions, the experimenter gave the partici-
handheld computers. pant two to four tokens noncontingently and
asked him or her to hand the experimenter one
Response Measurement and tokenatatime.Eachtokenexchangeresultedin
Interobserver Agreement one bite of the food. When all tokens were
Interobserver agreement was obtained for at exchanged, the session commenced. During
least 25% of sessions across all conditions. The sessions, the experimenter placed tokens in a
partial agreement within intervals method was small receptacle positioned next to the work-
used to determine interobserver agreement, sheets.
which was calculated by dividing the smaller For the no-choice condition, participants
response frequency by the larger response worked with a red worksheet and received a
frequency per 10-s interval, and converting the tokenaccordingtoafixed-ratio(FR)5schedule
VALUE OF CHOICE 555
(i.e., the token-production schedule). After Each pile of like-colored worksheets consisted
completion of each 3-min session, the partici- of schedule-correlated stimuli in the initial link
pant exchanged his or her tokens for only one of the chains and was arranged in a row, with
type of reinforcer identified as the most each worksheet equidistant from the other. The
preferreditemviaapriorpreferenceassessment. colors corresponded to the same conditions to
The item was put on one plate, and each time whichparticipantshadbeenexposedduringthe
one token was exchanged for a reinforcer, the reinforcer assessment. The experimenter in-
plate was replenished with the same item until structedtheparticipanttochoosetheworksheet
the participant had exchanged all of the tokens. by saying, ‘‘Go to the table and pick one color
In the single-choice condition, the participant you would like to work with.’’ The experi-
workedwithayellowworksheet,andthetoken- menter changed the placement of the colored
production and token-exchange schedules were items on the table before the start of the next
the same as those described in the no-choice sessionandgavetheparticipanttheopportunity
condition. During token exchange, the experi- to make another initial-link response.
menter arranged five identical edible items on a Afteraparticipantchosetheworksheetcolor,
plate (e.g., fruit snacks). The same highly the experimenter put on a colored smock that
matched the selected worksheet, and a 3-min
preferred item chosen for the no-choice condi-
session commenced. During the token-ex-
tion was used. The experimenter gave the
change period, programmed consequences for
participant the opportunity to choose one item.
the terminal link were available (i.e., no choice,
Although we attempted to ensure that all
single choice, varied choice). The experimenter
stimuli were identical in the single-choice
presented three opportunities to make initial-
condition, it is possible that slight differences
link color selections each day, except for Luke,
in size, color, or shape may have influenced
who received only one or two opportunities to
responding.Aftertheparticipantchoseanitem,
make a selection (due to unavailability).
theexperimenterreplenishedtheplatesuchthat
the same five items remained on the plate at all
times. RESULTS AND DISCUSSION
In the varied-choice condition, the participant
Figure 1 depicts response rates during the
worked with a blue worksheet, and the token-
reinforcer assessment for all participants. These
production and token-exchange schedules were
data represent response rates for situations in
the same as those described in the single-choice
which the experimenter selected particular
condition. During token exchange, the partic- conditions (e.g., no choice, varied choice).
ipant could exchange each token for one item When reinforcers were arranged for target
from the five most highly preferred items (e.g., behavior in the no-choice, single-choice, and
one fruit snack, one piece of candy, one cookie, varied-choice conditions, responding increased
one piece of chocolate, one marshmallow). The relative to rates obtained in the prior baseline
experimenter placed each item on a plate and condition. Dylan’s performance in the no-
replenished the item if the participant chose it choice condition was the exception. When
during the token exchange. These three condi- compared to the no-reinforcement baseline
tionswerepresentedonceeachdayinarandom condition (M 5 0.44 responses per minute),
order.AfterthemultielementB-C-Dcondition, no change in mean level of responding was
a reversal to baseline (A) was introduced. observed in the last three sessions of the no-
During the concurrent-chains evaluation, choice condition (M 5 0.44); this suggested
each participant was permitted to select a that there was no reinforcement effect when
condition as described by Tiger et al. (2006). edible items were available but he could not
556 SANDEEP K. SRAN and JOHN C. BORRERO
Figure 2. Mean percentage of selections for Mira,
Milo, and Luke during the concurrent-chains evaluation.
Althoughnotdepictedinthefigure,participantswerealso
giventheopportunitytoselectthebaselinecondition(this
never occurred).
Figure 1. Results of the reinforcer assessment for all
participants. Response rates are depicted by the open respectively.Resultsindicatedpreferenceforthe
triangles for the no-choice condition, open circles for the
varied-choice condition (42% of selections) for
single-choice condition, and open squares for the varied-
Mira, a strong preference for the varied-choice
choicecondition.
condition (81% of selections) for Milo, and a
choose the item. Following the reinforcement modest preference for the varied-choice condi-
condition, baseline was reintroduced, and tion (39% of selections) for Luke.
response rates for all participants decreased to Results of the present study replicate those
low levels, particularly during the last three reported by Tiger et al. (2006), in that
sessions of the condition. preference for the opportunity to choose varied
Figure 2 depicts the percentage of selections across participants. The findings of Tiger et al.
for each of the conditions available, for all appear to be sufficiently robust to the extent
participants who completed the concurrent- that findings of the current study were similar
chains evaluation. Mira, Milo, and Luke were andwereobtainedusingtokenreinforcersinthe
given 24, 21, and 24 choice opportunities, absence of rules. In addition, Tiger et al.
VALUE OF CHOICE 557
focused exclusively on participant selections for down, or both) to assess how reinforcer rate
various initial- and terminal-link options. This might affect preference.
is useful information as it relates to preference. Choice could be a simple classroom alter-
However,ofsimilarimportanceisasenseofthe ation designed to improve the performance of
amount of work (e.g., response rate) that is some students (Romaniuk et al., 2002). Future
completed under the various conditions. studies could assess the practical implications of
One limitation of the present study is choice by investigating the effects of parametric
apparent in the number of opportunities to increases in work requirements on preference
choose. Preference for one condition may have foravarietyofchoice(orno-choice)conditions.
been more discernible given more than three
opportunities to select a condition. Another
REFERENCES
limitation involves the absence of a clear
evaluation of the role of the tokens. In the Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L.
P., Owens, J. C., & Slevin, I. (1992). A comparison
presentstudy,tokenswereneverremoved;thus,
of two approaches for identifying reinforcers for
definitive conclusions regarding the role of personswithsevereandprofounddisabilities.Journal
tokens cannot be reached (Foster & Hacken- of AppliedBehaviorAnalysis,25,491–498.
berg, 2004). Future research may involve Foster, T. A., & Hackenberg, T. D. (2004). Unit price
and choice inatoken-reinforcement context. Journal
procedures similar to those described in the
of theExperimental Analysisof Behavior, 81,5–25.
present study, but should include token-with- Luczynski, K. C., & Hanley, G. P. (2009). Do children
drawal conditions. prefer contingencies? Anevaluation ofthe efficacy of
and preference for contingent versus noncontingent
A third limitation of the current study may
social reinforcement during play. Journal of Applied
be found in the reinforcement schedule. An FR
Behavior Analysis,42,511–525.
5 schedule was in place that necessarily Romaniuk,C.,Miltenberger,R.,Conyers,C.,Jenner,N.,
decreased reinforcement rate compared to the Jurgens,M.,&Ringenberg,C.(2002).Theinfluence
of activity choice on problem behaviors maintained
common FR 1 schedule implemented in
byescapeversusattention.JournalofAppliedBehavior
preference and reinforcer assessment research.
Analysis,35,349–362.
It is possible that an FR 1 schedule would have Skinner, B. F. (1953). Science and human behavior.
established higher rates of responding in one Oxford, UK:Macmillan.
Tiger,J.H.,Hanley,G.P.,&Hernandez,E.(2006).An
condition than another, thereby exaggerating
evaluation of the value of choice with preschool
preference. However, an FR 5 schedule was children. Journal of Applied Behavior Analysis, 39,
used to reduce the possibility of satiation. 1–16.
Future research may involve decreasing session
Received March 26,2008
duration to control reinforcement rate or
Final acceptanceJuly 2,2008
titrating the response requirement (either up, Action Editor,James Carr