Bridging the Gap between Research and Practice
Mark R. Dixon & Alyssa WilsonSouthern Illinois University
The Road to Somewhere…..
• Casino laboratory• College students as
subjects in research• Computerized versions of
actual casino games
Bench
• Answering basic psychological processes with a clinical/sub-clinical sample
• In the field research or clients entered into lab
Translational • Real pathological gamblers• Real gambling problems• Real treatment that YOU
can take with you today
Bedside
Problem gambling is not the problem.
• Problem gambling is the outcome of deeper rooted clinical problem.
• Treatment should be designed to treat what the “cause” of the gambling is, not just the gambling itself.
• Life is not just “fine” except for problems with gambling.
Popular Treatment Approaches
• Gamblers Anonymous– Disease model– Client is a victim– You never “beat” the disease– No active treatment. Social support group.
• Self-Exclusion Programs– Self or court orders gambler to be banned from gaming establishments– No way to ban online or illegal local gambling
• Medication– Certain dopamine blockers can be effective at suppressing gambling for some people– Remove the medication, the problem returns
• Psycho-educational– Teach people about game odds– Teach about risk to self or others from repeated gambling
Classic Behavioral Treatments
• Aversive Conditioning
• Thought suppression
• Self-monitoring/reinforcement
Contemporary Behavioral Contributions
• Contingency-based Models
Response Cost
Behavioral Contributions
• Contingency-based Models• Language-based Models– External rules
Dixon (2000) – The Psychological Record
• Subjects: 5 Recreational roulette players• Baseline: Wagered on numbers they picked or the
experimenter picked.• Intervention: Provided rules to the subjects
– Roulette is easy to win; the more you play the more you win; the best way to win is to pick your own numbers
– Roulette is a losing game; the more you play the more you lose; the experimenter can not predict good/bad numbers
• Outcome: relative rise and decline in wagers while contingencies remained the same
• Conclusion: Rules matter – contingencies do not
Dixon, Hayes, & Aban (2000) – The Psychological Record
• Subjects: 45 Recreational roulette players• Baseline: Wagered on numbers they picked or the experimenter
picked.• Intervention: Provided one set of rules to the subjects
– Roulette is easy to win; the more you play the more you win; the best way to win is to pick your own numbers
– OR---– Roulette is a losing game; the more you play the more you lose; the
experimenter can not predict good/bad numbers• Outcome: relative rise and decline in wagers while contingencies
remained the same• Conclusion: Rules matter – contingencies do not
Behavioral Contributions
• Contingency-based Models• Language-based Models
• Delivered Rules• Self-Rules
Recent Attention Paid to Near-Miss
Procedure
• Participants – 18 recreational slot machine players• Setting - small room, computer, video camera,
observation mirror. Three computerized slot machines available concurrently.
• Method – – 100 trials w/ 20% chance of a win on every trial– 100 trials w/ 0% chance of a win on every trial
• Various densities of near-misses on each “slot machine”• Reinforcement densities were constant on each slot machine
What we know:
• Subjects will rate near-miss displays as:– Closer to wins– More pleasurable / less aversive to look at
• Subjects will prefer near-misses in concurrent operant preparations– Density effect of NM– Extinction conditions alter preference
• Neurological traces of the near-miss– Near-misses produce different levels of dopamine in brain– Pathological gamblers react neurologically different than non-
pathological
What we don’t know:
• What behavioral process produces a near-miss effect?
• Will the near-miss effect be demonstrated with other casino games?
• Can the near-miss effect be assessed independently of the by-chance reinforcers that occur during gambling
What Actually is the Near-Miss Effect?
• Product of Stimulus Generalization– Current display looks structurally similar to a
reinforced display, and thus it serves reinforcing function
• A Discriminative Stimulus– Signals the availability of an upcoming reinforcer
• Product of Verbal ConstructionOr, an interaction of all the above?
Almost winning…A verbal event
“Almost”
“9 + 4 = 14”
“9 + 4 = 14”
Antecedent
“Almost” + GCR
“What is 9 + 4 ?”Speaker
Listener
Math Time
Behavior Consequence
“What is 9 + 4 ?”
Verbal Construction
Note: GCR might be < for “almost” than for “correct”
Looking for House #34
See House #26 “Almost There” Arrive Soon at House #34
Antecedent Behavior Consequence
Looking for House #34
See Gas Station “Almost There” Arrive Soon at House #34
Antecedent Behavior Consequence
See Sign for Off Ramp
See Sign for Sunset Blvd
See House
#26
“Almost”
Desired OutcomeIn Close Proximity
Methods
• 16 participants with history of gambling• Rating of 100 various slot machine displays– Near miss - loss - win
HOW CLOSE IS THIS DISPLAY TO A WIN?
1 (not at all) 5 10 (very much like a win)
Methods
• Phase 1:– Rate slot machine images
• Phase 2: – Develop 3 three member stimulus classes– Attempt to derive “almost” to non-near miss display
• Phase 3:– Repeat exposure to Phase 1 task
Stimulus 1 Stimulus 2 Stimulus 3
A Stimuli
WIN Almost Lose
B Stimuli
C Stimuli
Stimulus 1 Stimulus 2 Stimulus 3
A Stimuli
WIN Almost Lose
B Stimuli
C Stimuli
0
2
4
6
8
10
76 112 113 115 123 169 195 211 339 357 769 771 772 773 774 776
0
2
4
6
8
10
76 112 113 115 123 169 195 211 339 357 769 771 772 773 774 776
Participant Number
Lik
ert
Rat
ing
s (1
= l
oss
--
10 =
win
)
Pre-Test
Post-Test
Loss Near-Miss
Win
0
2
4
6
8
10
76 112 113 115 123 169 195 211 339 357 769 771 772 773 774 776
0
2
4
6
8
10
76 112 113 115 123 169 195 211 339 357 769 771 772 773 774 776
Participant Number
Lik
ert
Rat
ing
s (1
= l
oss
--
10 =
win
)
Pre-Test
Post-Test
Loss Near-Miss
Win
More than Slots
• Many more types of near misses occur while gambling:
– Blackjack
– Roulette
– Craps
Blackjack
Near Miss: Blackjack• Participants:
– 5 undergrads with history of playing cards for money
– Paid 50 dollars in lotto drawing based on # of chips left
• 50 trials (1o practice trials)
• Data Collection• Self-recorded data
• Experimenter IOR on 30% trials
• End of trial – circle number 1-9 on how close their hand was to a win – 1 = no chance ; moderate
chance; good chance (as anchors)
– Record their score, dealer’s score and if they won or not on that given hand
Results• 2 factor Near-Miss Effect– Non-bust loss– Mathematical difference between dealer and
player
Minimal Difference between player and
dealer cards
Non-Bust (under 21)
NearMiss
Minimal Difference between player and
dealer cards
Bust(over 21)
No Near Miss
Average Loss Trials
All Loss Trials (all players combined)
Roulette
Near Miss: Roulette
• Participants:– 28 College Undergraduates (run concurrently)
– Extra credit value based on winnings• First 5 students to hit a number = 10 x points• Next 5 = 5 x points• Remainder of students = 1 x point
• Played 60 trials of roulette– 1 single bet on a single number (1:38 odds of winning)
– Rating of outcome• “How close to a win was this outcome for you?”• Scale 1 to 10
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 310
1
2
3
4
5
6
7
8
9
10
- numerical value of bet and win
Avg
Sub
jecti
ve R
ating
1 2 3 4 5 6 7 8 9 100123456789
10
- between Location of Bet and Win on Board
Avg
Sub
jecti
ve R
ating
Alternative Methods
• Self-reports of:– How close to win– How much do you like
• Preference for near-misses during gambling– Interaction between display and superstitious
reinforcement
• Can we show a “preference” for near-misses absent of the reinforcement interaction?
Paired-Choice Near- Miss
• Participants– 34 College
Undergraduates– Awarded course extra
credit– Randomly assigned to 2
groups of 17– Instructed to choose
between two slot images.• “Which one would you
rather see if you were playing a slot machine?”
• Procedures– Exposure to 120 trials of 3
trial types• Win vs Loss• Win vs Near Miss• Near Miss vs Loss
– Experimental Group• 5 min intervention
– Control Group• 5 min break in hallway
Intervention Details
• Prior research suggests that rules are effective ways of altering gambling behavior– Dixon (2000); Dixon, Aban, & Hayes (2000)– Dixon & Delaney (2006)
• Prior research also suggests that the deliteralization of language can alter the current functions of a specific verbal stimulus– Aka: defusion in therapy contexts
Experimental Intervention: (one slide)
• Almost winning is not winning at all
• Almost winning is a trick played on you by the slot machine
• Almost winning makes you feel good, but it is false feeling
• Losing is losing is losing is losing is losing is losing– Repeat for 2 minutes
Which One??
A B
Which One??
A B
Which One??
A B
Pre Post0
20
40
60
80
100
ExpControl
% C
hoic
es fo
r Win
Pre Post0
20
40
60
80
100
ExpControl
% C
hoic
es fo
r N.M
.
Pre Post0
20
40
60
80
100
ExpControl
% C
hoic
es fo
r Los
s
Variations of Effect
• The Near-Miss effect varies• Not based exclusively on physical characteristics of the
stimulus
• Core behavioral process rests on altering of psychological function of the stimulus (stimuli)
• Altering psychological function will alter the type of stimulus that is considered a near-miss
Variations of Assessment
• The Near-Miss Effect can be assessed with novel methods and produce similar effects
•Verbally based interventions for gamblers who are under control of near-misses appear promising
Nastally and Dixon (2011): The Psychological Record
• N=3 Pathological gamblers • MBL across participants• Baseline Computerized slot machine play– 50, 70, 90 trials at baseline– Report out loud how close each outcome was to a
win– 1 (very far from a win) to 10 (very close to a win)
SIMULATION
SIMULATIONVisible Symbols
on Wheel
Treatment
• Intervention– ACT intervention targeted each of the 6 components– Intervention delivered via PowerPoint presentation
each 5 min in length– Slides consisted of words/pictures in form of
directions + experiential exercises– Each component was delivered at equal length of
time
• Return to computerized slot machine play
Self asContext
Contact with the Present Moment
Defusion
Acceptance
Committed Action
Values
Psychological
Flexibility
Baseline Play and Self-Ratings
Baseline Play and Self-Ratings
Baseline Play and Self-Ratings
30 min ACT
30 min ACT
30 min ACT
Post-Treatment Play and Self Ratings
Post-Treatment Play and Self Ratings
Post-Treatment Play and Self Ratings
Time
Client 1
Client 2
Client 3