Using computational psychiatry to understand decision-making processes in adolescents during the Iowa Gambling Task

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Cognitive Processes

Abstract

The Iowa Gambling Task (IGT) is widely implemented to assess decision-making deficits in adolescents, a group notable for risk taking and mental health concerns (Olson et al., 2007). While popular, there have been conflicting findings regarding overall performance (number of advantageous minus disadvantageous choices) in clinical populations (Buelow & Suhr, 2009) and what specific decision-making processes may yield these differences. To address these discrepancies, a variety of computational models have been developed to capture the cognitive processes that underlie decision-making during IGT. However, these models have been infrequently applied to adolescent decision-making and have only been recently used to address clinical populations. Furthermore, these studies take a diagnostic approach to groups, rather than observing changes in decision-making along broader spectra of psychopathology (Redish & Gordon, 2016).

The recently developed Value plus Sequential Exploration (VSE) model has been shown to fit participant data best and recover parameters more accurately relative to other computational models of the IGT, as well as meaningfully differentiate between healthy controls and patient groups (Ligneul, 2019). The VSE has yet to be applied to a developmental sample. For the pre-registered poster, we will analyze previously gathered data from 187 typically developing participants aged 9-23. For group-based analyses, these participants will be categorized by the following: no major internalizing or externalizing concerns, internalizing concerns, externalizing concerns, and both internalizing/externalizing concerns. For regression analyses, the spectra will be used as predictors in a model of the parameters including relevant control variables. Participant choices will be run through the VSE model using open-source code in MATLAB, resulting in estimates of parameters that reflect five decision-making processes: attention to gains and losses, how much attention depends more on recent versus past outcomes, exploration of decks regardless of the outcome (i.e., exploration bonus), the rate at which the exploration bonus influences exploration, and consistency of decisions. We will test whether internalizing and/or externalizing concerns are associated with behavioral decision-making differences on the IGT and proposed decision-making processes on the VSE via the above two analytic approaches. Previous research has generally indicated lower performance in patient groups for advantageous deck selections and we expect to replicate this finding. Externalizing behaviors have been associated with a preference for guaranteed immediate rewards versus higher uncertain rewards (Olson et al., 2007). We predict the externalizing group will show a higher exploration bonus and a quicker decline of directed exploration across trials, including a lower entropy value by unevenly exploring decks. Individuals with internalizing tendencies gradually choose less from advantageous decks, suggesting poorer memory (i.e., greater memory decay parameter) and attention and therefore lower consistency. Predicted differences among the groups will better specify the decision-making processes that lead to behavioral differences and contribute to the development of psychological disorders. Specific decision-making differences as a function of internalizing and externalizing may be promising treatment targets for more personalized intervention approaches (Patton et al., 2016). Data has been collected and the MATLAB code has been successfully adapted for our analyses, which will be completed before the conference.

Authors

• Joanne Jung, University of Minnesota- Twin Cities
  Presenting Author

• Brandon Almy, University of Minnesota - Twin Cities
  Non-Presenting Author

• Paul Collins, University of Minnesota
  Non-Presenting Author

• Monica Marie Luciana, University of Minnesota–Twin Cities
  Non-Presenting Author