Modelling Dynamic Processes in the Hot Columbia Card Task
Keywords
Loading...
Authors
Issue Date
2019-07-09
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Despite the ubiquity of risk-taking, we lack a precise understanding of how it is shaped by
experience. Here, we analyze the dynamics of risk-taking in a recently developed
experimental paradigm for studying sequential risky choice, the hot Columbia Card Task
(CCT), which combines a dynamic structure with a systematic variation of outcome and
probability. Although prior models based on Cumulative Prospect Theory offered insights
into the cognitive processes underlying risk-taking in the hot CCT, they cannot account for
the dynamics occurring in the task. In this project, we propose two dynamic extensions to
existing approaches. The Bayesian Probability model assumes a confluence of description and
experience, and aims at capturing a gradual decrease in risk-taking across the CCT. The
Prior-Influence model assumes a direct effect of the outcome of the prior game round on
choices made in the current game round. Model comparison showed that these extensions
offer no clear improvement in explanatory power over static models, suggesting that they
cannot capture the actual dynamics occurring in the hot CCT. Our analyses also pointed
towards several problematic aspects in the parameter estimation and recovery of the models;
and should thus be interpreted with extreme caution. We close with a discussion of the
potential causes and solutions to this problem.
Description
Citation
Faculty
Faculteit der Sociale Wetenschappen