|
| Methodological Advancements in the Study of Discounting Processes: Modeling More Complex, Naturalistic Choices and Exploring Opportunities in Longitudinal Big Data |
| Saturday, May 27, 2023 |
| 3:00 PM–4:50 PM |
| Hyatt Regency, Centennial Ballroom B |
| Area: EAB/CBM; Domain: Basic Research |
| Chair: Jeffrey S. Stein (Virginia Tech (FBRI)) |
| Discussant: Gregory J. Madden (Utah State University) |
| Abstract: Behavioral outcomes are devalued (discounted) as a function of delay, uncertainty, and effort, among other variables. Validated and reliable methods have been used to examine these forms of discounting and their possible roles in clinical disorders and maladaptive health behavior. However, considerable room for methodological innovation remains. Most existing methods arrange simple choices between discrete outcomes that differ only in magnitude and the specific variable under investigation (e.g., delay). Although indispensable, these methods may fail to model important complexities of real-world choice, which often produces multiple, non-discrete outcomes that accrue over time and can engage multiple forms of discounting simultaneously (e.g., delay and probability). Beyond these considerations, most knowledge about discounting and real-world human behavior comes from cross-sectional studies of adults featuring small to moderate sample sizes. In contrast, the Adolescent Brain Cognitive Development (ABCD) study features a large and dynamic longitudinal data set, providing an unprecedented opportunity to examine developmental trajectories in delay discounting, explore prospective associations with health behavior, and identify challenges and best practices in the analysis of “big data.” This symposium features four innovations that address the issues described above, including methods that model more complex, naturalistic choices and facilitate analysis of large-scale, longitudinal discounting data. |
| Instruction Level: Basic |
| Keyword(s): Behavioral Economics, Delay, Discounting, Probability |
| |
| Novel Commodity Discounting Tasks to Quantify Employment Value as a Function of Educational Requirements, Drug Testing Requirements, and Workplace Safety |
| MIKHAIL KOFFARNUS (University of Kentucky College of Medicine), Kenneth Silverman (Johns Hopkins University), Joseph Grzywacz (Florida State University), Haily Traxler (University of Kentucky) |
| Abstract: Commodity discounting tasks, while primarily used to study how delay and probability affect the valuation of money or consumable reinforcers, can be used to quantify how the value of most anything is devalued by any quantifiable process. Over a series of experiments, we have used discounting procedures to examine factors that may affect how employment opportunities are valued. Processes that we examined as factors that may modify the value of employment included the presence of drug testing requirements, the amount of additional training/education required to obtain the job, and the safety of the workplace of the prospective job. In each case, the hourly rate of pay was used as a measure of job value. Results confirmed that a discounting framework can be used to assess how realistic characteristics of hypothetical jobs impact the value of those jobs. For example, drug testing requirements reduce job value, but these reductions in value are almost exclusively limited to those with recent drug use. These results suggest that discounting tasks provide a convenient and adaptable framework to assess the valuation of non-consumable commodities like employment opportunities, as well as the factors that affect those valuations. |
| |
| Simultaneous Assessment of Delay, Probability, and Combined Delay × Probability Discounting: Extensions of a Multiplicative Model of Discounting Using Effective Delay 50 |
| JEFFREY S. STEIN (Virginia Tech (FBRI)), Allison Tegge (Fralin Biomedical Research Institute at Virginia Tech Carilion), Haylee Downey (Virginia Tech (FBRI)), Jeremiah Michael Brown (Fralin Biomedical Research Institute at Virginia Tech Carilion), Mary Jane King (Virginia Tech (FBRI)) |
| Abstract: Delayed outcomes are often probabilistic, but few studies have examined the combined effects of delay and probability on choice. Prior work shows that a multiplicative discounting model, in which subjective value is a function of the interaction between delay and probability, describes choice data well. However, these assessments can be burdensome (~125 trials), which limits their feasibility in multivariate assessment batteries and their ability to foster discovery. Here, we developed a briefer method (36 trials) to assess Effective Delay 50 (ED50; i.e., delay required to reduce subjective value by 50%) across a range of outcome probabilities. We also adapted the multiplicative model to describe this function, including two free parameters to estimate “pure” (univariate) rates of delay and probability discounting (k and h, respectively). We will present evidence that ED50 is well-described by this model, with estimates of k and h for monetary gains corresponding closely to those from standard discounting tasks. We will also discuss inclusion of a third free parameter (s) to describe individual differences in sensitivity to risk in the delay-discounting paradigm. Additional data and future directions will be discussed, including exploration of the generality of these findings across outcome types, associations between these measures and health behavior, and effects of experimental manipulations. |
| |
| Discounting Combination Outcomes: Immediate Losses (Gains) Followed by Delayed Gains (Losses) |
| Sara J. Estle (University of North Carolina at Chapel Hill), LEONARD GREEN (Washington University in St. Louis), Joel Myerson (Washington University in St. Louis), YU-HUA YEH (Fralin Biomedical Research Institute at Virginia Tech Carilion), Ke Ning (Washington University in St. Louis) |
| Abstract: Most discounting research has focused on relatively simple situations (e.g., choosing between immediate and delayed gains, or between immediate and delayed losses) and the relations among amount, delay, and subjective value in such situations are now well established. Many everyday choice situations, however, are more complex, involving alternatives that combine both gains and losses. We will present results from experiments in which participants discounted a monetary outcome that combined an immediate gain with a delayed loss or an immediate loss with a delayed gain. Consistent with the hyperboloid discounting function, the subjective value of the combination was approximately equal to the difference between the (undiscounted) amount of the immediate outcome and the (discounted) value of the delayed outcome. Taken together, these findings support the view that complicated choices like those common in everyday life can be understood within the discounting framework. Choice situations involving immediate gains (losses) followed by delayed losses (gains) pose iconic self-control problems, and the present findings support the application of the discounting framework to these important problems. |
| |
Evaluating Delay Discounting Across Childhood Within the Adolescent Brain Cognitive Development (ABCD) Study |
| JUSTIN CHARLES STRICKLAND (Johns Hopkins University School of Medicine), Richard Yi (University of Kansas), Julia Felton (Henry Ford Health), Brion Maher (Johns Hopkins University Bloomberg School of Public Health), Jill Rabinowitz (Johns Hopkins University Bloomberg School of Public Health) |
| Abstract: Research demonstrates that delay discounting is a modifiable risk factor for substance use throughout adulthood. However, limited work has evaluated delay discounting in early life leading to sizable gaps in the understanding of early developmental precursors needed to inform preventive health interventions. The Adolescent Brain Cognitive Development (ABCD) Study is an NIH-funded longitudinal cohort study of brain development and child health with access to curated data available through NIH data repositories. This cohort study collects health data from 11,880 children aged 9-10 at the time of study enrollment from 21 research sites across the United States. Core battery measures include delay discounting measures collected at biannual interviews as well as health interviews and brain imaging data. The nature of delay discounting assessments and constraints encountered in collection of these data in children present challenges in the analysis of ABCD delay discounting data. This talk will provide an overview of analytic considerations in the ABCD dataset and best practices. Data on handling non-systematic data, test-retest reliability of various analytic methods, and correspondence with health behaviors like substance use will be described. Open source materials will be provided to session attendees to facilitate analysis of ABCD data in their own research projects. |
|
| |