ERIC EJ877527: Manipulating Slot Machine Preference in Problem Gamblers through Contextual Control

2010·0.13 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2010, 43, 125–129 NUMBER1 (SPRING2010)
MANIPULATING SLOT MACHINE PREFERENCE IN PROBLEM
GAMBLERS THROUGH CONTEXTUAL CONTROL
BECKY L. NASTALLY, MARK R. DIXON, AND JAMES W. JACKSON
SOUTHERNILLINOISUNIVERSITY
Pathological and nonpathological gamblers completed a task that assessed preference among 2
concurrently available slot machines. Subsequent assessments of choice were conducted after
variousattemptstotransfercontextualfunctionsassociatedwithirrelevantcharacteristicsofthe
slot machines. Results indicated that the nonproblem gambling group, but not the problem
gambling group, increased their responding toward the slot initially trained as greater than
following the initial training procedure, then decreased their responding toward that slot
followingthe reversal phase.
Keywords: contextualcontrol,pathologicalgambling,relationalframetheory,slotmachine
_______________________________________________________________________________
Increasing research has been conducted on shared properties with the function of the
the effect of contextual control on the choices contextual cue for greater than. This effect of
individuals make while gambling (Hoon, Dy- contextualfactors on preference ina two-choice
mond, Jackson, & Dixon, 2008; Johnson & gambling paradigm also has been demonstrated
Dixon, 2009; Zlomke & Dixon, 2006). For with children using concurrently available dice
example, Zlomke and Dixon measured partic- of differing colors in a board game situation
ipants’preferencefortwoconcurrentlyavailable (Johnson & Dixon, 2009).
slot machines of equal payout (one yellow and Although these studies demonstrate the effect
one blue). The experimenters then delivered of context on choice making in a gambling
reinforcement to participants when they select- procedure, they are not without limitations.
ed the comparison stimulus that was greater Perhaps one of the most glaring limitations that
thanthesamplewhentheyellowcontextualcue critics of behavioral research on gambling have
(background color) was present. Results indi- noted is that these studies have not been
cated that 8 of 9 participants demonstrated a
conducted with participants who regularly
higherpreference fortheyellowslotmachine in
gamble. Common characteristics of problem
the posttest compared with initial exposure to
gamblers have been evaluated that may differ
the slot machines. These results suggested that
from nongamblers (Petry, 2005), and there is
the contextual cue (slot machine color) exerted
evidenceofdifferencesinthewaythispopulation
control over responding.
valuesmoney(Dixon,Marley,& Jacobs,2003).
Hoonetal.(2008)attemptedtoreplicatethe
In addition, the pretest–posttest design used
resultsobtainedintheZlomkeandDixonstudy
in each of the studies discussed (Hoon et al.,
(2006) using only two comparison stimuli at a
2008; Johnson & Dixon, 2009; Zlomke &
time. The results indicated that following the
Dixon, 2006) often has been cited as a
nonarbitrary relational training and testing
procedural limitation. Therefore, the purpose
phases, 5 of the 6 participants demonstrated
of the following study was to extend the
higher responding on the slot machine that
literature already established in this area by
AddressallcorrespondencetoMarkR.Dixon,Behavior incorporating problem gamblers as participants
Analysis and Therapy Program Rehabilitation Institute, and adding a contingency reversal to the
Southern Illinois University, Carbondale, Illinois 62901
procedures (see Smyth, Barnes-Holmes, &
(e-mail: [email protected]).
doi:10.1901/jaba.2010.43-125 Forsyth, 2006).
125
126 BECKY L. NASTALLY et al.
METHOD credits were added to the participant’s ‘‘amount
won’’ and ‘‘total credits’’ windows. These
Participants and Screening
creditsdidnotcorrespondwiththeparticipant’s
Fourteen adults (7 men, 7 women), 18 to 60
overall compensation, and although they ap-
years old, who were actively recruited by the
peared on the screen, did not hold any other
experimenter, served as participants in the
meaning.Thenumberoftrialsinthisphasewas
study. Prior to being involved in the study,
randomly determined between 50, 70, and 90.
each participant was administered the South
This slot machine task also served as Posttest 1
Oaks Gambling Scale (SOGS; Lesieur &
and Posttest 2 and directly followed the
Blume, 1987), and scores ranged from 0 to
nonarbitrary relational training and testing
20, with at least a 5 or more being indicative of
phase as well as the nonarbitrary relational
apotentialpathologicalgamblingproblem.The
training and testing with reversal phase.
nonproblem gambling and problem gambling
Nonarbitrary relational training and testing.
groups were made up of 7 participants each.
Following exposure to the slot machine task
There was a significant difference in SOGS
pretest, a nonarbitrary relational training pro-
scores among the problem gambling (M 5
cedure was presented to participants (see Hoon
10.71, SD 5 7.63) and nonproblem gambling
et al., 2008;Reilly, Whelan, & Barnes-Holmes,
(M50.29,SD50.49)groups,t(12)53.608,
2005). There were four separate sets of three
p 5.003592. Participants were closely matched
gambling-relevant stimuli varying from least to
on variables such as age, income, and education
most: jackpots (5 million, 10 million, 20
level.
million), the word BINGO (B-I, B-I-N, B-I-
Setting and Apparatus N-G-O), coins (nickel, dime, quarter), and
places (8th place, 3rd place, 1st place) in
Allsessionstookplaceinaquietsettingwhile
training and playing cards (4, 9, king), poker
the experimenter monitored the session about
chips ($25, $100, $500), dollar bills ($1, $10,
every 10 min. All experimental sessions were
$50),andbottlesofalcoholicbeverages(1,3,7)
conducted on a HP Pavilion laptop computer
in testing.
thatwas programmedin Microsoft VisualBasic
During a trial, either a blue or yellow screen
2005 and included a data-collection system.
appeared followed by two images positioned in
Experimental Design and Procedure the middle of the screen side by side. After the
The design was a within-subject pretest– participant clicked on one of the images, a
posttest group design with a contingency message appeared that read ‘‘correct’’ followed
reversal of the baseline discriminations. by a chime or ‘‘wrong’’ followed by a buzzer.
Slot machine task (pretest, Posttest 1, and The presence of the background color deter-
Posttest 2). A trial began with the participant mined which selection of the two stimuli was
reading the instructions on the screen and then reinforced. The set of contingencies each
clicking on a button to indicate a choice of participant was exposed to was determined by
playing on one of two concurrently available preferential response allocation ($70% of
simulated slot machines. The slot machines responses) in the pretest: If no preference
were identical except that the machines were emerged, the color chosen as the greater than
mostly blue or mostly yellow and were cuewasdeterminedrandomly,andtheopposite
programmed on equal schedules of random- color was chosen as the greater than cue if a
ratio(RR)reinforcementof0.5withmagnitude preference did emerge. There were a total of 48
ofreinforcementheldconstantatonecreditnet trials, and participants had to reach a criterion
gain or loss. Following a winning spin, two of 89%successive correct responding (43of the
PROBLEM GAMBLERS 127
Figure1. Respondingtowardthecoloredslotmachineinitiallytrainedasthegreaterthancontextualcueforboththe
nonproblem gambling (top) and problem gambling (bottom) groups in pretest (gray), Posttest 1 (black), and Posttest
2(white).
48 trials). The relational testing phase was and testing procedure described above, except
identicaltothetrainingphaseexceptfoursetsof that the initial contingencies in place were now
novel gambling-relevant images (listed above) reversed.
were presented, and no feedback was given. Slot machine task Posttest 2. After reaching
Slot machine task Posttest 1. If participants criterion responding in the testing phase,
now showed higher responding toward the participants were reexposed to the slot machine
color slot machine trained as greater than as task described above.
comparedtopretest, theywere reexposed to the
slot machine task. However, if they allocated
RESULTS AND DISCUSSION
the same number of responses or fewer toward
this slot machine, they were excused from the Figure 1 shows the data for nonproblem and
experiment at this point. problem gamblers. Consistent with previous
Nonarbitrary relational training and testing research, the nonproblem gambling group
with reversal. This phase of the experiment was quickly met criterion on the initial relational
identical to the nonarbitrary relational training training and testing phase, demonstrated a
128 BECKY L. NASTALLY et al.
predictable shift in preference for the color slot majority of their responses toward the color
machine that was trained as the greater than slot machine trained as the greater than
contextual cue during training (Hoon et al., contextual cue following the initial relational
2008; Zlomke & Dixon, 2006), and subse- training and testing phase (M 5 45, SD 5
quently reversed their preference for that slot 34.6). Of the 4 participants who advanced to
machine after the relational training and testing the second posttest, only 2 (P10 and P14)
with reversal phase. Specifically, 3 participants demonstratedareductionofresponseallocation
from this group (P1, P3, and P4) achieved toward this slot machine following the relation-
criterion for the initial relational testing within altrainingandtestingwithreversalphase(M5
one trial block, 3 (P2, P5, and P7) after two 40, SD 5 40.3).
trial blocks, and 1 (P6) following three trial The conceptual implication of these findings
blocks. All 7 participants in the nonproblem is additional evidence for the effect of context
gambling group met criterion in the relational on choice making in a gambling paradigm
training and testing with reversal phase within (Hoon et al., 2008; Zlomke & Dixon, 2006).
one trial block. Figure 1 also shows that all More specifically, the contextual functions of
participantsinthenonproblemgamblinggroup the colors yellow and blue were transferred
(P1 through P7) allocated a majority of through multiple-exemplar training of the
responses toward the color slot machine that relations greater than and less than to the slot
was trained as the greater than cue in Posttest 1 machines demonstrated by the preferential
(M 5 80, SD 5 14.5). During Posttest 2, all responding that emerged following exposure
participants(exceptP6)allocatedlessthan 45% to onesetof trainingcontingencies initiallyand
of their responses toward this same color slot thenareversalofthosecontingencies.Thestudy
machine (M 5 25, SD 5 22.3). extended prior research by demonstrating a
In contrast to prior research and our data difference in the way these contextual factors
obtained from the nonproblem gambling affected participants based on their history of
group, the problem gambling group required problem gambling.
on average five times as many trial blocks to Toinvestigatedifferencesthatmayemergein
meet criterion in the initial relational training he choice making of problem and nonproblem
and testing phase and did not demonstrate gamblers during gambling experiments, future
predictableshiftsinpreference forthecolorslot studies should examine the role of verbal
machine trained as the greater than cue behavior, because both groups in this study
following each training and testing phase. In were exposed to identical programmed contin-
the problem gambling group, 3 participants gencies. There has been evidence elsewhere of
(P9, P10, and P11) achieved criterion on the an increased self-governed rule adherence in
initial relational training and testing phase clinical populations (Wulfert, Greenway, Far-
within two trial blocks, 2 (P12 and P13) after kas, Hayes, & Dougher, 1994), and some
six trial blocks, 1 (P8) after eight trial blocks, researchers have suggested that problem gam-
and 1 (P14) after nine trial blocks. It is blers may be especially susceptible to maladap-
interesting to note that all participants in the tive rule formation such as superstitious
problem gambling group who advanced to the behavior or skill, rather than chance, producing
relational testing with reversal phase (P8, P9, positive outcomes (Delfabbro, 2004; Walker,
P10, P13, P14) reached criterion responding 1992). For example, future studies could
following exposure to one trial block. The data employ methods such as a talk-aloud procedure
show that only 4 of the 7 participants in this or postexperimental interviews to further exam-
group (P10, P11, P13, P14) allocated a ine the role of verbal rules. In addition, to
PROBLEM GAMBLERS 129
investigate the relation between derived rela- Johnson, T. E., & Dixon, M. R. (2009). Influencing
children’s pregambling game playing via conditional
tional responding and verbal rule formation
discrimination training. Journal of Applied Behavior
related to gambling behavior, future studies Analysis,42,73–81.
should incorporate procedures that require the Lesieur, H. R., & Blume, S. B. (1987). The South Oaks
participanttoderivesomeoftherelationsinthe gambling screen (SOGS): A new instrument for the
identification of pathological gamblers. American
absence of direct training.
Journal ofPsychiatry,144,1184–1188.
One obvious limitation in the current study Petry, N. M. (2005). Pathological gambling: Etiology,
was the small number of participants in each comorbidity, and treatment. Washington, DC: Amer-
ican Psychological Association.
comparison group. To further increase the
Reilly,T.,Whelan,R.,&Barnes-Holmes,D.(2005).The
external validity of these findings, it would be
effectoftrainingstructureonthelatencyofresponses
advantageoustoincorporatebiggersamplesizes. to a five-term linear chain. The Psychological Record,
However,demonstratingasignificantdifference 55,233–249.
Smyth,S.,Barnes-Holmes,D.,&Forsyth,J.P.(2006).A
between the two groups may indicate that the
derived transfer of simple discrimination and self-
type of participant (problem or nonproblem reported arousal functions in spider fearful and
gamblers) can make a difference when conduct- nonfearful participants. Journal of the Experimental
Analysis ofBehavior,85,223–246.
ing behavioral investigations on gambling. This
Walker, M. B. (1992). Irrational thinking among slot
study provides preliminary evidence that re-
machine players. Journal of Gambling Studies, 8,
searchers should be cautious when employing a 245–261.
nonclinical sample to investigate a clinical Wulfert,E.,Greenway,D.E.,Farkas,P.,Hayes,S.C.,&
Dougher, M. J. (1994). Correlation between self-
phenomenon.
reported rigidity and rule-governed insensitivity to
operant contingencies. Journal of Applied Behavior
Analysis,24,659–671.
REFERENCES
Zlomke,K.R.,&Dixon,M.R.(2006).Modificationof
Delfabbro, P. (2004). The stubborn logic of regular slot-machine preferences through the use of a
gamblers: Obstacles and dilemmas in cognitive conditional discrimination paradigm. Journal of
gambling research. Journal of Gambling Studies, 20, Applied BehaviorAnalysis,9, 351–361.
1–21.
Dixon, M. R., Marley, J., & Jacobs, E. A. (2003). Delay
discounting by pathological gamblers. Journal of
AppliedBehavior Analysis, 36,449–458.
Hoon,A., Dymond, S.,Dixon, M.R., & Jackson, J. W.
(2008).Contextualcontrolofslotmachinegambling: Received October1, 2008
Replicationandextension.JournalofAppliedBehavior Final acceptanceMarch 7,2009
Analysis, 41, 467–470. Action Editor,Chris Ninness