Association for Behavior Analysis International

The Association for Behavior Analysis International® (ABAI) is a nonprofit membership organization with the mission to contribute to the well-being of society by developing, enhancing, and supporting the growth and vitality of the science of behavior analysis through research, education, and practice.


41st Annual Convention; San Antonio, TX; 2015

Event Details

Previous Page


Symposium #450
Conditioned Reinforcement: Gambling, Information, and Token Economics
Tuesday, May 26, 2015
9:00 AM–10:50 AM
006D (CC)
Area: EAB/TPC; Domain: Basic Research
Chair: Travis Ray Smith (Southern Illinois University Carbondale)
Discussant: Travis Ray Smith (Southern Illinois University Carbondale)

The present symposium includes laboratory experiments that are tied together by a common theme: conditioned reinforcement. Smith & Jacobs will present data where rats preferences, in a rapidly-changing concurrent chains procedure, were influenced by stimuli correlated with probabilistic food delivery under conditions that ensured that both chained schedules delivered equal overall rates of food. Rice and Kyonka will present data from a concurrent chains procedure where preference for a probabilistic component was systematically affected by the probability that a food-correlated stimulus would intervene between the initial link and terminal link phases. These data will be related to the near win effect reported in the gambling literature. Cunningham, Craig, Filho, and Shahan will present data showing that response rates to produce an observing (food-correlated) stimulus were functionally related to the degree of informativeness (i.e., the ratio of the inter-reinforcer interval to the stimulus-reinforcer interval) of that observing stimulus. These data will be framed in terms of the Information Theory reported in the Pavlovian literature. Franceschini will report on the effects of increasing the cost of earning tokens on the demand for sucrose or water that are earned through token exchange. These data suggest that reinforcer demand mediated through token exchange may differ from procedures that do not use tokens to assess reinforcer demand in nonhuman subjects. Fantino will discuss the research from a conditioned reinforcement perspective.

Keyword(s): Conditioned Reinforcement, Gambling, Information Theory, Token Economics
Evidence for Conditional Reinforcement in Rats using Dynamic Concurrent Chained Schedules
TRAVIS RAY SMITH (Southern Illinois University Carbondale), Eric A. Jacobs (Southern Illinois University Carbondale)
Abstract: Rats’ lever pressing was maintained by a concurrent chains procedure in order to assess the influence of conditional reinforcement on preference. This occurred under conditions where preferences for conditional and unconditional reinforcement were not confounded. Initial links arranged different rates of terminal link entry, and terminal links provided different probabilities of food delivery. High (or low) rates of initial link transition corresponded with low (or high) probabilities of terminal link food delivery, such that both options delivered equal overall rates of food. In the signaled condition, different tone frequencies presented during the terminal link were correlated with the different probabilities of food delivery – putatively functioning as conditional reinforcement. In the unsignaled condition, the same terminal link tone frequency was presented regardless of the probability of food delivery. This experiment used a dynamic environment, where the relative rates of initial link entry (and corresponding probabilities of food delivery) were varied within-session. Overall, the generalized matching law provided an excellent description of the data, and the rats frequently favored the option providing higher rates of transition to the terminal link. However, this tendency was reduced in the signaled condition, suggesting modest influence by conditional reinforcement.
Conditioned Reinforcement Effects of Discriminative Stimuli in Concurrent-Chains Procedures
NATHAN RICE (West Virginia University), Elizabeth Kyonka (West Virginia University)
Abstract: In concurrent chains with probabilistic terminal links, sensitivity to relative probability of reinforcement was lower when terminal links ending with and without food were differentially signaled, a result known as the signaling effect (Mattson, Hucks, Grace & McLean, 2010). Our aim was to determine whether different arrangements of stimuli signaling the same outcome systematically affected response allocation. Pigeons pecked in concurrent chains with certain and probabilistic components. In both components, terminal links ending with food were preceded by a sample phase in which the active key was red. Terminal links in the probabilistic component that did not end with food were either preceded by a green active key in the sample phase, or a sample phase in which the key switched from red to green. Initial-link choice for the probabilistic component increased as a function of the proportion of non-food terminal links where the sample phase involved some presentation of the red key. This functional relation demonstrates that stimuli preceding food acquire conditioned reinforcing value at a temporal distance. We argue these results imply that the signaling effect may contribute to the development of problem gambling.
The Role of Temporal Informativeness in Observing: Conditioned Reinforcement and Information-Theory Revisited
PAUL CUNNINGHAM (Utah State University), Andrew R. Craig (Utah State University), Paulo Sérgio Dillon Soares Filho (Universidade de São Paulo), Timothy A. Shahan (Utah State University)
Abstract: Conditioned reinforcement is generally believed to result from Pavlovian conditioning. Recent applications of Information Theory to Pavlovian conditioning suggest that the temporal informativeness of a stimulus governs stimulus associability and can account for a variety of Pavlovian phenomena. Temporal informativeness is calculated as log2 (Cycle time/ Trial time), where cycle time refers to the average interval between reinforcer deliveries and trial time refers to the average interval between stimulus onset and reinforcer delivery. The purpose of the present experiment was to examine the role of the temporal informativeness of a stimulus in conditioned reinforcement by examining how observing response rates vary as a function of temporal informativeness. Response independent food was delivered according to a random-time schedule while pigeon’s key pecks produced a stimulus signaling those food deliveries. Temporal informativeness was varied across conditions by manipulating the C/T associated with the response-contingent trial stimulus. C/T values included: 32, 8, 4, 2, and 1.14. Results showed that observing rate was a hyperbolically increasing function of the temporal informativeness of the stimulus. These results are discussed in terms of the information-theoretic developments within the Pavlovian literature and are contrasted with previous applications of Information Theory to conditioned reinforcement and Delay-Reduction Theory.
Token Economies and Demand Elasticity Analysis: A Challenge for Behavioral Economic Explanations?
Abstract: Elasticity is a behavioral economic measure of the relation between consumption (in the laboratory setting: total reinforcers released per session) and price (response requirement per reinforcer). Consumption-price relations tend to form a downward sloping curve, showing that the number of reinforcers released decreases as the reinforcement schedule increases. Inelastic – or essential - goods are reinforcers whose consumption curves show only small decreases as the schedule increases. Elastic – or superfluous – goods show significant decreases in such situations. Elastic and inelastic relations have been repeatedly observed with laboratory rats and water (inelastic good) or sweetened water (elastic good). We built a token economy in which three rats had to spin a wheel (initial link) to produce tokens that were exchangeable for water or sweetened water (intermediate link) under a choice condition (terminal link). Increases in the token reinforcement schedule produced an immediate consumption decrease in both reinforcers, but the consumption of sweetened water returned to previous average levels in subsequent sessions. In economic terms, sweetened water became an inelastic good under this setting. The insertion of conditioned reinforcers (tokens) changed the relation between effort and consumption that is typically observed under non-chained procedures. This poses an interesting challenge for behavioral economic analysis.



Back to Top
Modifed by Eddie Soh