Association for Behavior Analysis International

This symposium will highlight emerging trends in basic and translational gambling research. Presenters will discuss findings for choice selection on variable ratio schedules, effective use of behavioral skills training for teaching card counting, and the impact of verbal behavior on slot machine outcomes (e.g., losses disguised as wins) and rule following. Skinner first discussed variable ratio schedules as an addictive reinforcement schedule, and yet little is known about human preference to various types of schedules in a gaming context. Similarly, recent structural changes in slot machine reels, particularly losses disguised as wins, are under researched and therefore not understood behavior analytically. Furthermore, empirical evidence for rule formation and subsequent following are beginning to emerge within gambling contexts; yet further replication and extension of rule following across gaming activities are still needed. Therefore, the symposium will provide empirical support for how structural changes may impact gambling behaviors, both in terms of risk and magnitude of bet size, and in terms of rule formation and subsequent rule following. Implications of empirical findings and directions for future research will also be discussed.

Variable ratio (VR) schedules have been called "the addictive schedule of reinforcement" because they generate persistent high rate responding with little or no pausing. Because, like many gambling games, on VR schedules reinforcement is always uncertain- the very next response always could be reinforced (payoff). The more one responds the more likely responding will result in reinforcement - a win. For these reasons VR schedules are presented as a model of gambling contingencies (how slot machines are programmed, etc.). However in virtually every casino gambling game, increased responding does not increase the probability of winning (e.g. random probability, not VR schedules are used or "sampling with replacement") Nevertheless several studies with non-human subjects have shown that variable schedules are highly preferred over fixed schedules of reinforcement even when responding on the fixed schedule results much greater overall reinforcement. Variable schedules are preferred even when they are counterproductive. The present study examines human preference for VR versus fixed ratio (FR) schedules of reinforcement. Subjects remove blank cards and flip them putting cards marked win in one box and others in another box. Wins are paid cash. Subjects are given experience with both FR 5 (card color A), and VR 6 (card color B) schedules and then allowed to chose which color cards they would like to continue flipping. If the VR option is chosen, subject will be given experience with the FR 5 and a VR 7 and then allowed to choose again.

A behavioral skills training procedure consisting of video instructions, video rehearsal, and video testing was used to teach four college students a card counting strategy in blackjack. A multiple baseline design was used to measure card counting accuracy and chips won/lost across participants. Prior to any training, no participant counted cards accurately. Each participant completed all phases of the training protocol, counting cards fluently with 100% accuracy during slow, medium, and fast training exercises. Generalization probes were conducted while playing blackjack in a mock casino following each phase of training. After training, all four participants were able to accurately count cards while playing blackjack. In conjunction with count accuracy, total winnings were tracked to determine the monetary advantages associated with counting cards. After losing money during baseline measures, three of four participants won a substantial amount of money playing blackjack following the intervention.

The current study investigated preferences made by 73 college students when presented with 60 sets of images of slot machine outcomes. These images were categorized into three groups; (1) loss disguised as wins (LDW), (2) wins, and, (3) losses. Three preference tests were conducted (LDW vs. loss; win vs. loss; LDW vs. win) in a random sequence. Results yielded a statistically significant difference between the three preference tests F(2, 71) = 56.15, p < 0.001. A Sidak post hoc analysis demonstrated that there was a statistically significant difference, p < 0.001, between LDW vs. loss and the other two tests, not between win vs. loss and LDW vs. win. A chi-square goodness of fit test was conducted to evaluate if preferences in the LDW vs. loss group were acquired through chance. The results were statistically significant X2 (71) = 266.9, p < 0.001, indicating that the preference for LDW over losses were not acquired by chance.

In a replication of derived rule-following and subsequent rule following during a gambling activity, adult participants wagered on a roulette table before and after completing a discrimination task within a non-concurrent multiple baseline design. Participants were instructed to tact three arbitrary symbols that were placed above the roulette wheel. During roulette play, participants wagered one chip on either black or red to win. Following baseline, each participant was presented with a series of discrimination training and testing trials designed to create a three three-member stimulus class including the words "bet" "on" "red/black" depending on baseline response allocation. All participants were then instructed to complete a fill in the blank assessment and an open-ended tact assessment to determine if the rule (e.g., "bet on red/black") derived after training. Results suggested that all participants derived the rule, and altered their response allocations by betting more on the specific color trained. Implications for conceptual development of self-awareness in regards to self-generated rules will be discussed.