|
| Discounting: Reconsidering our Methods and Measures |
| Sunday, May 28, 2017 |
| 10:00 AM–10:50 AM |
| Hyatt Regency, Centennial Ballroom B/C |
| Area: EAB/PCH; Domain: Translational |
| Chair: Matt Locey (University of Nevada, Reno) |
| Abstract: Behavior analysts can and should be well-equipped to address the issue of how the value of reinforcement decreases as a function of delay, probability, and magnitude. Various methods have been developed (in human and non-human animal laboratories) to collect and compare data relevant to this issue of discounting. Empirical and conceptual work in this area will be discussed: including both (1) an evaluation of different discounting-related procedures and measures and (2) significant implications for future studies on discounting. |
| Instruction Level: Advanced |
| Keyword(s): Delay Discounting, Human Operant, Probability Discounting, Rodent |
| |
| A One-Week Method for Assessing Delay Discounting in Rodents |
| (Basic Research) |
| CARLA H. LAGORIO (University of Wisconsin-Eau Claire) |
| Abstract: Delay discounting describes how the value of an outcome is affected by how quickly it is delivered. How rapidly reinforcers are discounted differs across species as well as individuals, and this phenomenon has been widely studied over the past 30 years using a variety of approaches. Despite their differences, the methods used to assess discounting have impressive internal reliability and external validity. The current talk will compare several commonly utilized procedures, and will describe a methodological variant that has been effective in achieving orderly discounting curves in a week. Rats experienced these procedures in a counterbalanced order, including an adjusting delay procedure in which amounts to a larger reward were manipulated across conditions, and several increasing delay procedures in which the delay to the larger reward was manipulated either across days or within trial blocks in a session. Post-hoc analyses examined validity across measures and also whether reliable results could be attained in time spans as short as one week. Ideally, the results from the novel procedure can lend confidence in using abbreviated discounting assays when examining potential correlates with other time-sensitive variables in behavioral, pharmacological, or aging research. |
| |
| Verbal Behavior and Risky Choice in Humans |
| (Basic Research) |
| DAVID J. COX (University of Florida), Jesse Dallery (University of Florida) |
| Abstract: Previous research has shown the probability of contacting a reinforcer influences the relative value of that reinforcer. An interesting aspect of choice related to probabilistic outcomes is the description-experience gap. The description-experience gap refers to the human tendency to choose the uncertain alternative more frequently when the outcome is experienced compared to described. Previous researchers have analyzed this gap using single- or repeated-trial measures. However, no researchers have examined how single- and repeated-trial measures compare across description and experience conditions using within-subject methods. Experiment 1 sought to compare preference for the uncertain alternative between described and experienced outcomes using repeated-trial and single-trial measures. Data consistent with previous research was obtained only for single-trial measures of point gain with reverse patterns observed for repeated-trial point gain and both measures of point loss. Experiment 2 tested a psychophysical hypothesis for the results from Experiment 1 and included measures of probability discounting. The results of Experiment 1 and 2 are discussed relative to probability discounting in general. Given the correlation between probability discounting and various clinical populations, understanding what causes the description-experience gap and how it can be reduced or increased represents an important and open area for future behavior analytic research. |
| |
| AUC vs. k – Are Theories of Discounting Necessary? |
| (Theory) |
| MATT LOCEY (University of Nevada, Reno), Laura Barcelos Nomicos (University of Nevada, Reno), Courtney Smith (University of Nevada, Reno) |
| Abstract: Delay, probability, and social discounting functions have been described – with varying degrees of success – by a number of different mathematical models. Most of those models include free parameters (e.g., k) to capture individual differences in discounting functions. However, there has been increasing usage of a so-called “atheoretical” measure of discounting: Area under the Curve (AUC). In 2016, new measures of AUC for discounting were presented in the Journal of the Experimental Analysis of Behavior. The developers of the new AUC measures (and the developers of the original AUC measures for discounting) freely acknowledged some of the limits of the original AUC measures. What they failed to acknowledge is that all measures of AUC, these so-called “atheoretical” measures of discounting, are necessarily theoretical. Data from our lab (and others) allow a comparison of different AUC measures, k (and other free parameters), and a truly atheoretical measure of discounting. We conclude that theoretical measures are preferable to atheoretical measures. |
|
| |