Risky Decision Making in Psychological Disorders

By Melissa Buelow

Risky Decision Making in Psychological Disorders provides readers with a detailed examination of how risky decision making is affected by a wide array of individual psychological disorders. The book starts by providing important background information on the construct of risky decision making, the assessment of risky decision making, and the neuroscience behind such decision making. The Iowa Gambling Task, Balloon Analogue Risk Task, and other behavioral measures are covered, as are topics such as test reliability and the pros and cons of utilizing tasks that have strong practice effects. The book then moves into how risky decision making is affected by specific psychological disorders, such as addictive behaviors, anxiety disorders, mood disorders, schizophrenia, sleep disorders, eating disorders, and more.

• Explores how risky decision making is affected by different psychological disorders
• Examines risky decision making and ADHD, psychosis, mood/anxiety disorders, and more
• Synthesizes the research on risky decision making
• Discusses merits/limitations of the Iowa Gambling Task and other behavioral measures
• Covers risky decision making and its associations with other executive functions

• Language: English
• Publisher: Academic Press
• Release date: May 1, 2020
• ISBN: 9780128150030

Melissa Buelow

Associate Professor, Ohio State University, where she teaches Introductory Psychology, Research Methods in Psychology, and courses in clinical psychology and neuropsychology. Her research program focuses on decision making and other executive functions, specifically investigating emotionally-based decision making, its predictors, and its outcomes. She has authored dozens of papers, is a member of APA’s Division 40, the Association for Psychological Science, Society for Personality and Social Psychology, among other groups, and received the 2015 Thomas J. Evans Teaching Excellence Award for Junior Faculty at The Ohio State University. Associate Professor, Ohio State University, where she teaches Introductory Psychology, Research Methods in Psychology, and courses in clinical psychology and neuropsychology. Her research program focuses on decision making and other executive functions, specifically investigating emotionally-based decision making, its predictors, and its outcomes. She has authored dozens of papers, is a member of APA’s Division 40, the Association for Psychological Science, Society for Personality and Social Psychology, among other groups, and received the 2015 Thomas J. Evans Teaching Excellence Award for Junior Faculty at The Ohio State University.

Book preview

Risky Decision Making in Psychological Disorders

Melissa T. Buelow

The Ohio State University, OH, United States

Table of Contents

Cover image

Title page

Copyright

Acknowledgments

I: The assessment of risky decision making

Chapter 1. An introduction to risky decision making

Abstract

Decision making, risk, and uncertainty

Risk-taking behavior

Introduction to the remaining chapters

Chapter 2. Measurement methods

Abstract

Risky decision making measures

Risk propensity and risk attitude measures

Demographic factors in test performance

Modeling decision making on behavioral tasks

Chapter 3. Reliability and validity

Abstract

Reliability

Validity

Factors affecting reliability and validity

Conclusion

Chapter 4. Neuroscience and associations with other executive functions

Abstract

Executive functions: theories and constructs

Impairments in executive functions

Is decision making an executive function?

Correlations between measures of decision making and executive functions

Neuroscience of decision making

What neuroimaging teaches us about decision making processes

Conclusion

II: Risky decision making across psychological disorders

Preface to section II: Organization of the remaining chapters

Preface to section II: Organization of the remaining chapters

Chapter 5. Anxiety: state-dependent stress, generalized anxiety, social anxiety, posttraumatic stress disorder, and obsessive–compulsive disorder

Abstract

The current literature: generalized anxiety

The current literature: social anxiety

The current literature: posttraumatic stress disorder

The current literature: obsessive–compulsive disorder

The current literature: trait anxiety and other nondiagnosable types of anxiety

The current literature: state-dependent stress

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusion and future directions

Chapter 6. Disruptions of mood: positive and negative affect, depressive disorders, and bipolar disorders

Abstract

The current literature: risk-taking behaviors

The current literature: risky decision making

The current literature: delay discounting, and reward responsiveness

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusion and future directions

Chapter 7. Disordered eating behaviors: anorexia, bulimia, binge eating, and obesity

Abstract

The current literature: risk-taking behaviors

The current literature: risky decision making

The current literature: delay discounting and reward responsiveness

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusion and future directions

Chapter 8. Sleep deprivation and sleep-related disorders

Abstract

The current literature: risk-taking behaviors

The current literature: risky decision making

The current literature: delay discounting and reward responsiveness

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusion and future directions

Chapter 9. Impulsivity and attention-deficit/hyperactivity disorder

Abstract

The current literature: risk-taking behaviors

The current literature: risky decision making

The current literature: delay discounting and reward responsiveness

What factors could be affecting risky decision making in attention-deficit/hyperactivity disorder?

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusions and future directions

Chapter 10. Addictive behaviors: gambling and substances of abuse

Abstract

Pathological gambling

Other behavioral addictions

Alcohol use disorder

Nicotine or tobacco use disorder

Cannabis use disorder

Opioid-related disorders

Stimulant use disorders

Ecstasy or MDMA use

Theories of risky decision making across substances of dependence

Treatment implications

Conclusions

Chapter 11. Schizophrenia and delusional disorders

Abstract

The current literature: risk-taking behaviors

The current literature: risky decision making

The current literature: delay discounting and reward responsiveness

What factors could be affecting risky decision making in schizophrenia spectrum disorders?

Performance on other executive function tasks

Neuroimaging

Potential mechanisms

Conclusions and future directions

Chapter 12. Conclusions and future directions

Abstract

Etiologies of risky decision making across disorders

Current issues affecting understanding of risky decision making in psychological disorders

Treatment implications

Future directions for the field

References

Index

Copyright

Academic Press is an imprint of Elsevier

125 London Wall, London EC2Y 5AS, United Kingdom

525 B Street, Suite 1650, San Diego, CA 92101, United States

50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

Copyright © 2020 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

ISBN: 978-0-12-815002-3

For Information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Nikki Levy

Editorial Project Manager: Megan Ashdown

Production Project Manager: Omer Mukthar

Cover Designer: Victoria Pearson

Typeset by MPS Limited, Chennai, India

Acknowledgments

Several individuals provided support and encouragement during the development of this project. I would like to thank those undergraduate student researchers who helped with the literature review process (Celeste, Cortney, Alyssa, and Ashley) and Wes for providing feedback on drafts. I would also like to thank my department colleagues and friends for their encouragement: Brad, Chris, Jen, Jim, Julie, and Melissa. Finally, I would like to thank my family, Charlie and Paul in particular, for their support and encouragement throughout this process.

I

The assessment of risky decision making

Outline

Chapter 1 An introduction to risky decision making

Chapter 2 Measurement methods

Chapter 3 Reliability and validity

Chapter 4 Neuroscience and associations with other executive functions

Chapter 1

An introduction to risky decision making

Abstract

In this introductory chapter a brief introduction to decision making and risky decision making in particular is presented. Difficulties in defining these and related constructs, such as risk-taking behavior, are discussed. The origins of our knowledge that emotions affect decisions and the dual-process model as applied to risky decisions are presented. Finally, a brief introduction to the rest of the chapters is presented.

Keywords

Risky decision making; decision making; delay discounting; reward responsiveness; risk-taking

The act of making a decision permeates all facets of life. From seemingly mundane decisions, such as what to wear or what to eat, to larger scale decisions such as whether to undergo surgery or not, we face decisions every day that can have significant effects on both our present and future situations. How do we make these decisions? What factors go into the decision making process to lead us to act in a risk-averse versus risk-seeking manner? What happens in our frontal lobe and reward pathway as we are deciding to take a risk? Being able to think through both the immediate and long-term consequences of a given decision can lead to more or less optimal decisions. In addition, maintaining sight of both potential gains (benefits) and potential losses (risks) can lead to a more balanced evaluation of the available options. And, sometimes, having a particular psychiatric diagnosis can affect this decision making process. Individuals diagnosed with anxiety, depression, schizophrenia, and other psychological disorders need to make important decisions about treatment options, such as opting for a medication trial versus psychotherapy or in some cases a surgical treatment. But they also still engage in the same daily and larger scale decisions that individuals without these diagnoses make. How are decision making and risk-taking affected by having one or more psychological disorders? In the rest of this chapter, I will further examine the construct of decision making and risky decision making in particular. I will also discuss how decision making naturally relates to involvement in risk-taking behaviors.

Decision making, risk, and uncertainty

To start this journey, we first need to define the term decision making. At the most basic level, a decision involves a choice between two options. We engage in decision making when we are asked to judge if a letter that flashed on a computer screen was an X (yes or no decision). We engage in decision making when we decide which key card to match an item to on the Wisconsin Card Sort Task. We engage in decision making, though likely between more than two options, when we decide which car to lease or buy. But decision making involves a number of factors, including examination of not just potential gains and losses but when they could occur (immediate vs at some point in the future) and their probability of occurrence as well as external factors such as receiving feedback on the decision’s outcome and whether one is pressured to make a decision (e.g., time-limited decision making) (Defoe, Figner, & van Aken, 2015). To make the best possible decision, one must also accurately perceive the probabilities and risks associated with each option (Damasio, 1994).

Multiple theories exist to explain how we make decisions. Although an in-depth examination of these theories is beyond the scope of this chapter, reviews can be found in the following sources, among others: Gilovich, Griffin, and Kahneman (2002), Kahneman (2011), Tversky and Kahneman (1974, 1983, 1986) (see Busemeyer & Stout, 2002, for an application to the Iowa Gambling Task). Some of the earliest theories held that we choose the most optimal option based on it having the highest expected value. Let us use an example. Say we are going to roll a die to win some money. If we roll a 1, we would receive $1.00; roll a 2, $2.00; and so on. We could determine the probability of winning exactly $3.00 by taking the total number of winning rolls (in this case, just one since we only win $3.00 by rolling a 3) and dividing it by the total number of possible rolls (in this case, six). Our probability of winning $3.00 is 1/6, or 16.67%. Now let us say that we were asked to pay money to roll the die, in hopes of winning more money than we paid for this opportunity. What should we do? Per theories based on expected value, we should choose the option with the highest expected value. In this example, our new options are to pay or to not pay. For the pay option to have a higher expected value than the not pay option, our winnings would need to outweigh our costs. Let us say that a roll cost $1 and the winning amounts were the same as above (1=$1.00, 2=$2.00, etc.). What should we decide? If we roll the die one time, there are six possible outcomes. Five of those outcomes would result in our winning more money than we paid to play. In this case, we have a 5/6 chance (83.33%) of earning money by rolling and a 1/6 chance (16.67%) of breaking even. Most people would decide to roll the die for $1.00 in this example. What if we were asked to pay $4.00 to roll the die? In this case, we have a 2/6 chance (33.33%) of earning money by rolling a 5 or a 6. We have a 3/6 chance (50%) of losing money by rolling a 1, 2, or 3 and a 1/6 (16.67%) chance of breaking even by rolling a 4. In this case the probability of losing money outweighs the probability of winning money, leading to a higher expected value in the not pay option. Rational decision-makers should apply this logic to all decisions, always opting for the decision with the higher expected value. But our decisions are not always rational. In fact, we often choose an option that does not have the highest expected value. Why?

These questions led to the idea of expected utility and later prospect theory. The idea of utility dates back to Bernoulli who in 1738 (1954) introduced the idea that not all decisions are based on a rational examination of the expected value of each option. Instead, one’s subjective perception of the benefits of an option can influence decisions. Personal value, or utility, associated with an option can weigh just as heavily as expected value, in turn potentially biasing decisions. We should make decisions by evaluating the probabilities of the events and assess their expected value, choosing the one with the highest expected value. But we don't. Our decisions are biased and not entirely consistent with rational processes (Kahneman & Tversky, 1972). Rational decision making occurs when the same decision rules were applied in every situation. How we decided whether to pay $4.00 to roll a die should be the same process we use to decide which car to purchase or which college to attend. But, typically, we do not apply this same logic and reasoning to each decision making situation. The subjective value (utility) associated with different options in different decisions can bias—positively or negatively—the decision making process. The involvement of utility in decisions led to the development of prospect theory (Kahneman, 2011; Kahneman & Tversky, 1979). Prospect theory assesses how decisions are made based on how the outcome will differ from the individual’s current state, and how subjective value (utility) affects the processing of probabilities. Prospect theory focuses on what decisions will actually be made, rather than what decisions should be made based on expected values. In prospect theory, decisions are made as a function of the individual’s current state. Gains and losses (risks and benefits) are not viewed solely based on their magnitude and probability, but instead, on the likelihood they will improve or worsen one’s current state. If in the pay-to-play die roll example I only had $5.00 in my bank account, I would likely take fewer risks than if I had $5000.00 in my bank account. The subjective value of that $4.00 risk depends on my current financial state and can lead to a more risk-averse (not pay) versus risk-seeking (pay) decision.

Examinations of utility in general and prospect theory more specifically led to an understanding that how decisions are framed can affect outcomes. Specifically, framing a question in terms of gains versus framing a question in terms of losses can change how risk-seeking or risk-averse of a decision is made. In their classic Asian disease study, Tversky and Kahneman (1981) asked one group of participants to decide between two treatment options for an outbreak affecting 600 people: Option 1 would result in 200 of those affected being saved, whereas Option 2 had a 33% chance all 600 will be saved and a 67% chance no one will be saved. The second group of participants decided between Option 3, which results in the deaths of 400 of those affected, and Option 4, with a 33% probability of no deaths and a 67% probability of all dying. Reading these two scenarios, Options 1 and 3 are the same, and Options 2 and 4 are the same based on probabilities of success (life) and failure (death). But the participants made very different decisions in each set. The first group of participants were more likely to choose Option 1, whereas the second group of participants were more likely to choose Option 4. Why? Framing. When deciding between Options 1 and 2, the gain-frame that people will be saved (Tversky & Kahneman, 1981, p. 453) can lead to a more risk-averse decision making outcome than when deciding within a loss-frame [people will die (p. 453)].

Before talking more about risk and risky decisions, let us first examine other cognitive processes involved in decision making. Although expected value and utility, as well as an assessment of risks and benefits, are all crucial to decision making, other cognitive processes are just as important. In fact, Reyna and Brainerd (2011) outline the decision making process in such a way as to solidify its status as an executive function (a theme of Chapter 4: Neuroscience and associations with other executive functions). Decision making involves, at least in part, a process of storing knowledge from previous decisions and more generally, accessing that knowledge in the current decision making situation, then using that information in conjunction with one’s values and situational factors to arrive at a conclusion. In addition, Weber and Johnson (2009), in their review of the psychology and neuroscience of judgment and decision making, reiterate the importance of attention and memory in decision making. They also stress that emotions can affect decisions (see upcoming section). Other cognitive abilities affect decision making. Being able to hold onto the necessary information required to effectively evaluate different options requires attention and working memory skills. To the extent that attention may be limited or distracted in a given situation, decision making could suffer. Decision making also relies on longer term memory access to determine if and how a similar decision played out in a previous situation. Critically evaluating different options, such as by examining the relative pros and cons of each, relies on attention, memory, and executive functions. Although the main focus of this volume will be on decision making itself, decision making could not occur without the contributions of these other cognitive functions.

Risky decision making

What is risk? Risky decision making? Different definitions of each exist. Risk can be thought of as the opposite of a sure thing. If a sure thing has a guaranteed outcome, then a risk has a non-guaranteed outcome. We incorporate an understanding of risk into our decisions when we acknowledge that there is a probability of a particular event occurring and that this probability could vary depending on different decision options. Risk introduces uncertainty and the unknown into decision making. Taking a risk can be viewed as preferring a non-guaranteed or risky option to the sure thing or as choosing the option with the lower expected value over the option with the higher expected value (Reyna & Huettel, 2014a, 2014b). Risk can occur when individuals choose an uncertain over a certain option or choose an option with a greater level of variability in outcome (Chua Chow & Sarin, 2002; Figner & Weber, 2011; Weber, 2010). Taking a risk can also have a more clinical implication, as individuals take a risk when they decide to engage in a behavior knowing there is a potential for harm (e.g., Wallach, Kogan, & Bem, 1962). Or we may end up taking a risk because we just do not have all the information that we need. All in all, despite different definitions of risk, it does appear that risky decisions involve some element of uncertainty in terms of the expected outcome whereas riskless decisions are those with a known outcome (e.g., Kahneman & Tversky, 1984). Risky decision making, then, involves making a decision without full knowledge of the outcomes. Risky decision making involves at least some element of uncertainty and some element of risk. There may be a set probability of a large gain, but that outcome is unknown at the time of the decision and is likely offset by a probability of a large loss. A decision still needs to be made, leading to risk-seeking or risk-averse decisions. But what affects this uncertain or ambiguous decision making process? Situational factors, emotions, and cognitive resources, for starters. But some unconscious or automatic processes can affect decision making, leading to our understanding of dual-process models.

Dual-process models

As previously stated, emotions can affect decisions. One’s mood during decision making can affect the favorability of some options, and the potential for improvement in mood after a decision is made can affect the decision (e.g., Weber & Johnson, 2009). Emotions also underlie an automatic, unconscious form of decision making that is distinct from but related to the cognitive, conscious form of decision making. Several different terms for these two processes exist, but all fit under the overarching guideline of a dual-process model (e.g., Evans, 2008; Evans & Stanovich, 2013; Reyna, 2004; Tversky & Kahneman, 1983). The dual-process model states that decision making is guided by two different processes: a hot, Type I (System I) system and a cold, Type II (System II) system. System I is the rapid-response system. It occurs automatically when triggered by the current situation, acts without cognitive resources, and is described as unconscious, fast, implicit, automatic, impulsive, quick, requiring little effort, effortless, and innate (Kahneman, 2011; Metcalfe & Mischel, 1999; Stanovich, West, & Toplak, 2011; Zelazo & Muller, 2011). Emotions are involved in System I decision making, whereas cognition guides System II decision making. System II is more deliberate, requires cognitive resources, and is described as conscious, slow, processed, involving agency and choice, concentrated, reasoned, self-control, and limited in capacity.

System I may rely on heuristics to arrive at quick decisions. Heuristics are rules or guidelines that can be used to guide decision making, resulting in a more efficient process (Busemeyer & Townsend, 1993; Payne, Bettman, & Johnson, 1988). They can be beneficial, in part because they can decrease the amount of cognitive resources required to make a decision (Simon, 1955). But, heuristics can also introduce bias into decisions (Kahneman & Tversky, 1973). Tversky and Kahneman (1974) listed several heuristics that can introduce bias, including (1) representativeness (one may assess the likelihood of a particular result based on how similar the situation is to a previous one with the preferred result), (2) availability (one may base decisions in part on how readily available other examples/situations similar to the current one come to mind), and (3) adjustment and anchoring (one may bias decisions based on the initial starting point, or anchor, and resulting adjustments to it during decision making). These and other automatically initiated heuristics could lead to a nonoptimal decision by System I unless it is overridden by System 2. How does one override this automatic, mandatory System I? It needs to be interrupted and suppressed while a better, more reasoned response is created by System II (e.g., Stanovich et al., 2011). But not all is doom and gloom with regard to System I. There is evidence to suggest that this system is more efficient (De Neys, 2006) and less sensitive to working memory restrictions (De Neys, 2006; De Neys & Schaeken, 2007) than System II. When System II is overburdened, such as when multiple factors are overwhelming attentional resources, bias could actually creep in and impair the decision (De Neys, Schaeken, & d’Ydewalle, 2005). Thus there are cases in which the quick, efficient System I response leads to a better or more accurate decision than the longer, logical, System II response (De Neys & Pennycook, 2019). These two systems really do attempt their best to arrive at an efficient but also appropriate response to any decision making situation.

The somatic marker hypothesis and the neuroscience of risk

Risky decision making researchers often use the terms hot and cold decision making to refer to the System I and System II processes. Hot decision making emphasizes the idea that emotions can guide decision making. Sometimes these emotions can emerge as a gut feeling or instinct (e.g., the System I response; Bechara, Damasio, Tranel, & Damasio, 1997; Damasio, 1994). This idea sets the foundation for the somatic marker hypothesis (Damasio, 1994), which informed our understanding of the emotional side of decision making. At its most basic level, the somatic marker hypothesis states that we experience changes in our physiological and affective state as a function of different events in our lives. Representations of these emotional states are retained in long-term memory storage. When a similar situation arises in the future, the memories associated with the previous decision are accessed—along with those emotional states, now termed somatic markers. Somatic markers are then used as a kind of implicit knowledge that helps guide current decision making, most notably in situations where there is some level of ambiguity or uncertainty (e.g., Brand, Recknor, Grabenhorst, & Bechara, 2007; Maia & McClelland, 2004). In situations where the risks/probabilities are known versus uncertain, more explicit knowledge (i.e., cold decision making or System II) instead guides decisions. This interaction between hot and cold, implicit and explicit, and System I and System II is at the heart of any decision with some element of risk to it (Wood & Bechara, 2014).

Although covered in greater detail in an upcoming chapter (Chapter 4: Neuroscience and associations with other executive functions), we know that multiple cortical and subcortical structures are involved in these two decision making systems. The prefrontal cortex, the most anterior portion of the frontal lobe, is associated with both System I and System II cognitive processes. Although specific findings can vary, multiple neuroimaging studies support the idea that more cognitive-based (System II or cold) decision making occurs in the dorsolateral prefrontal cortex whereas more emotion-based (System I or hot) decision making occurs in the ventromedial prefrontal cortex and its connections to limbic system structures such as the amygdala (e.g., Adolphs & Tranel, 2004; Bechara, Damasio, Damasio, & Lee, 1999). How did we learn about the neuroscience of risky decision making? Case examples led researchers to hone in on the prefrontal cortex, with later neuroimaging supporting these initial suppositions.

The neuroscience of risky decision making and the somatic marker hypothesis has its origins in the case of Phineas Gage (e.g., Damasio, Grabowski, Frank, Galaburda, & Damasio, 1994; Harlow, 1848, 1868). In 1848 Phineas Gage was working as a railroad foreman in Vermont. One day, while preparing an explosive to remove rock in the path of the railroad, Gage was distracted and an explosion occurred. The tamping iron he was holding (approximately 2 m in length, 3 cm in diameter) was propelled through his skull, entering his left cheek and exiting the top-right portion of his skull (Figs. 1.1 and 1.2). Gage, still conscious, was taken to Dr. John Harlow for treatment. He survived this injury, living until the 1860s when he died of complications due to seizures (Damasio, 1994). It is a misnomer to say that Gage survived this incident intact. Although he lived, he was a very different person than he was before his injury. Per Dr. Harlow, Gage no longer acted like himself. He was able to walk, talk, and think, but he also was now impulsive, had difficulties planning ahead, and had difficulties controlling his emotions. Examination of Gage’s skull in the years since his death indicated that the tamping iron excised or otherwise significantly damaged large portions of the orbitofrontal cortex, an area we now know is linked with emotion-based and social decision making (Chapter 4: Neuroscience and associations with other executive functions).

Figure 1.1. Phineas Gage and the tamping iron. Originally from the collection of Jack and Beverly Wilgus, and now in the Warren Anatomical Museum, Harvard Medical School.

Figure 1.2. Dr. Harlow's (1868) depiction of the tamping iron's path.

A later case provided additional support for a link between the orbitofrontal cortex and emotion-based and social decision making. Patient EVR (Eslinger & Damasio, 1985) began experiencing personality changes that led to a diagnosis of orbitofrontal meningioma. The meningioma was surgically removed, but difficulties with real-world decision making persisted. He had difficulties holding down a job, failed to think through long-term consequences of his decisions, and experienced significant difficulties with daily activities including making minor decisions. However, his performance on clinical measures of different cognitive abilities, including executive functions, was within normal limits. There was no evidence of impairment on validated cognitive tasks despite evidence of significant real-world limitations. This case, among others, led to a rapid expansion of research into behavioral assessments of decision making. In particular, patient EVR, coupled with a better understanding of the cognitive changes and neurological damage experienced by Phineas Gage, led to the development of the somatic marker hypothesis and evidence that cognition-based and emotion-based systems (e.g., hot and cold) interact in decision making.

Risk-taking behavior

Risk-taking behaviors occur because we make a decision to engage in the behavior (Furby & Beyth-Marom, 1992; Reyna & Farley, 2006). Emotions, impulsivity, a failure to plan ahead—these and other reasons—can lead to greater involvement in risk-taking behaviors. A greater focus on the immediate, usually positive, outcomes and a lesser focus on the longer-term, potentially more negative outcomes is associated with greater rates of involvement in risk-taking behaviors (e.g., Bickel, Odum, & Madden, 1999; Bogg & Roberts, 2004; Madden, Petry, Badger, & Bickel, 1997; Mitchell, 1999). More generally, the tendency to increase the value of an immediate reward and steadily diminish the value of a more distant reward the further into the future it appears is termed delay discounting (Madden & Bickel, 2010). It can also appear as choice impulsivity, which is evident when an individual chooses a smaller–sooner reward over a larger–later reward (Hamilton et al., 2015). Impulsivity—the tendency to act on a whim (Eysenck & Eysenck, 1977)—can also affect risky behaviors. As we will see in Chapter 4, Neuroscience and associations with other executive functions, activation of the brain’s reward pathway can also lead to continued involvement in risk-taking behaviors (Jessor, 1991; Levy & Glimcher, 2011; Porcelli & Delgado, 2009b) due to changes and adaptations in how risks and rewards are processed.

Just as we saw there is no universal definition of risky decision making, there is also no one definition of risk-taking behavior (Schonberg, Fox, & Poldrack, 2011). Taking a risk, to an economist, can refer to deciding on an option with uncertainty or some level of ambiguity to it. To a clinician, risk-taking may be more focused on the potential negative health consequences of a behavior (e.g., Defoe, Figner, & van Aken, 2015). For the purposes of this volume, risk-taking behavior will be defined in this clinical realm, as a behavior that can potentially result in negative consequences for physical and mental health in the future. Risk-taking behaviors can vary across individuals but are thought to include such behaviors as overuse of alcohol or binge drinking, use of illegal substances, use of substances in potentially dangerous situations (such as driving), engaging in sexual activity without protection against pregnancy or sexually transmitted infection, skydiving, mountain climbing, reckless driving, and playing the stock market. This is by no means an exhaustive list but instead reflects the breadth of different behaviors classified as risky. What will become evident in the remaining chapters is that the definition of risk-taking, and how it is assessed, can vary widely by study.

Introduction to the remaining chapters

Both self-report and behavioral measures are frequently used to study risky decision making and real-world risk-taking behavior. In Chapter 2, Measurement methods, I provide information about some of the most common measurement methods for assessing risky decision making and two factors involved in the process: (1) attitudes and (2) propensity toward risk. As will be evident, numerous measures of these constructs exist, with differences in the specific type of decision making assessed by each one. For example, the behavioral measures often vary as a function of the assessment of risk and uncertainty (De Groot & Thurik, 2018) and whether decisions made at the start of the task have implications for later within-task decisions (i.e., the extent to which learning occurs). In Chapter 3, Reliability and validity, the evidence for or against reliability and validity of these measures is examined. A common theme across measures is whether lab-based assessments of risk-taking correlate with real-world risk-taking, or the extent to which tasks show ecological validity versus this validity being assumed. Chapter 4, Neuroscience and associations with other executive functions, tackles questions of whether decision making is an executive function, whether task performance correlates with measures of other executive functions, and which brain structures are involved in hot versus cold (or System I vs System II) decision processes. After that, the remaining chapters delve into the research regarding risky decision making and risk-taking behavior as a function of several primary categories of psychopathology. The current status of the literature will be reviewed, with an emphasis on our current understanding of risk-taking behavior, risky decision making, and neuroimaging of risk-related constructs in these disorders. Each chapter will present several theories that underlie risky decision making, as well as examine other factors that could be affecting decisions (e.g., disease severity, medication, and other treatment status). The final chapter will tie together the common threads from each chapter and present information about the next steps in risky decision making research, including cognitive modeling techniques and how our knowledge can inform the treatment process.

Chapter 2

Measurement methods

Abstract

This chapter delves into the history of assessing risky decision-making by examining some of the most common behavioral tasks. Descriptions of tasks, such as the Angling Risk Task, Balloon Analogue Risk Task, Choice Dilemmas, Framing Spinner, Iowa Gambling Task, Risky Gains, and Wheel of Fortune, are presented. In addition, several common assessments of risk-taking propensity and risk attitudes are also reviewed. Finally, the influence of demographic factors on task performance is examined and cognitive and mathematical modeling approaches are introduced.

Keywords

Risky decision-making; delay discounting; reward responsiveness; risk-taking; age; sex; mathematical modeling

Multiple measures exist to assess risky decision making, or at least a portion of the risky decision making construct. Some of these measures are study-specific, utilized in only one study. Others are more prevalent in the research literature—standardized, implemented across studies, and in some cases developed into clinical instruments. Both self-report and behavioral measures will be examined, as will assessments of the related constructs of risk propensity and risk attitude. Measures purported to assess decision making under risk, delay discounting, and reward responsiveness are included as they all tap into a component—small or large—of the overall risky decision making construct. Some measuresdiscussed could be classified as measures of impulsivity and/or measures of risky decision making. They are included in order to provide a context for later discussion of risky decision making in various psychological disorders. Finally, the content of these tasks will be addressed in the present chapter, whereas task psychometrics (reliability, validity) will be addressed in Chapter 3, Reliability and validity, and the neuroscience of the tasks in Chapter 4, Neuroscience and associations with other executive functions. What will become readily apparent across these three chapters is that tasks believed to assess risky decision making do not necessarily correlate with one another, leading to concerns about how the construct is defined across studies. It should be noted that this is not an exhaustive review of all previously utilized risky decision making measures, but rather a compilation of some of the most commonly used tasks in the risk-taking and risky decision making literature.

Risky decision making measures

Adult Decision making Competence

The adult decision making competence scale (ADMC) was developed to assess accuracy and consistency in decisions (Bruine de Bruin, Parker, & Fischhoff, 2007; Parker, Bruine de Bruin, Fischhoff, & Weller, 2018; Parker & Fischhoff, 2005). Originally containing seven subscales, the standard ADMC contains six intercorrelated scales. Resistance to framing has participants respond to a series of dilemmas, choosing which of the two options constitutes the best decision. The questions are first posed within a gain frame, then the same questions are posed within a loss frame. The consistency in responses despite the question framing is assessed. Recognizing social norms asks participants if it is sometimes OK to engage in a series of different actions (Bruine de Bruin et al., 2007). A correlation is then calculated between responses to these items and to the same items in which the participant estimated the percentage of 100 similarly aged individuals who would say the behavior is OK. To assess under/overconfidence, participants respond to a series of true/false statements then rate their confidence in the decision. Applying decision rules examines the accuracy of individual’s decisions, assessing the number of correct responses to a series of decisions. On the consistency in risk perception scale, consistency in judgments of the likelihood of events occurring in the next year versus next five years is assessed. Finally, the resistance to sunk costs scale has participants respond to a series of scenarios that pit past expenditures against new circumstances.

Angling risk task

The Angling Risk Task (ART; Pleskac, 2008) was developed to assess the influence of learning on decision making on the Balloon Analogue Risk Task (BART) and similar tasks. Pleskac argued that the theory behind the Devil’s Task, BART, and Iowa Gambling Task (IGT), for example, failed to account for participants learning from their experiences on previous selections. Multiple factors are varied in the ART, allowing for an examination of the specific influence of learning in the decision making process. The specifics of the ART are similar to the BART. Participants are told that they will be participating in a fishing tournament in which they will earn money for caught fish. The pond is stocked with blue and red fish. If a red fish is caught, they will earn 5 cents. If instead the blue fish is caught, the trial will be over and all money earned on that trial will be lost. In order to keep the earned money, participants must stop fishing and click the Collect button prior to catching a blue fish. On the BART the number of possible pumps per balloon and explosion points is unknown to the participants. On the ART, this information may or may not be known to the participants. Each game trial starts with one blue fish and n−1 red fish (Pleskac, 2008). If the participants are fishing on a sunny/clear day, then they see the number of fish available and do not need to learn this information. If instead the participants are fishing on a cloudy day, they must use trial-and-error feedback to learn the general probability of catching a blue versus red fish. Two additional manipulations can occur. If the tournament is catch-and-keep, the participants sample fish without replacement and a running tally of the fish caught is shown on the screen. If the tournament is catch-and-release, the participants sample fish with replacement and the tally is not shown on the screen. In keeping with the similarities to the BART, the total number of fish was set to 128, with the maximum beneficial number of caught fish per trial set to 64. The number of trials is variable but typically set to 30 (Pleskac, 2008). The primary outcome variable assessed is the average number of casts per trial, adjusted just for the trials in which the blue fish was not caught (i.e., the trial did not end early). In the original validation study, Pleskac (2008) found that participants made more selections/took more risks when it was sunny/clear than cloudy and in the catch-and-keep versus the catch-and-release condition. He argued that this task showed participants react to changes during the decision making process, in turn changing their representations of the task to reassess the current decision making strategy.

Balloon analogue risk task

The BART (Lejuez et al., 2002) was designed to assess risk-taking behavior in adolescents and young adults. Participants are tasked with earning money by pumping up a series of 30 balloons. Each pump of the balloon earns 5 cents, but participants will lose this earned money if the balloon pops. In order to keep the earned money, participants should stop pumping the balloon before it pops and click the Collect $$$ button to bank the earned money for that balloon. Unknown to the participants, balloons can pop after 1–128 pumps, with an average break point of 64 pumps. No explicit information about the probability a given balloon will pop at a particular point is given to participants, and in fact, one of the initial balloons is typically set to pop very early (i.e., under 10 pumps) to show that the balloons do, in fact, pop. As the pumps progress on a given balloon, the relative benefit of each pump decreases, while the relative risk increases. Thus, risky decision making occurs when participants continue to pump up the balloons (Lejuez et al., 2002). That said, risky decision making is conflated with monetary gain. Participants engage in riskier behavior the more times they pump up a particular balloon, but they also earn a greater monetary reward for a greater number of pumps (provided the balloon does not pop).

Several modifications of the BART are seen across studies. In the original validation study (Lejuez et al., 2002), and utilized in several studies afterward, three different colored balloons were used (90 trials total). Each balloon color represented a different average break point (small, medium, and large) so that researchers could examine how risk-taking changed with greater opportunities to take risks. An automatic version of the BART was utilized across several neuroimaging studies. On this version, participants do not pump up the balloons themselves but instead watch while the BART is completed for them after indicating how many times to pump up a balloon. Finally, some versions of the BART utilize a set number of trials (usually 30), whereas others consist of as many trials as possible completed in a set time period.

Blackjack task

The Blackjack Task mimics a real-world game of blackjack. Although the specific details of the task vary by study, a general format is followed. Participants receive a hand of blackjack and see one of the two cards the dealer was dealt. Based on the visible cards, participants bet on the outcome of the hand while deciding to hit or stay. The outcome could be a win (monetary gain), a loss (monetary loss), or a push (no change in monetary status).

Bomb risk elicitation task

On the Bomb Risk Elicitation Task (Crosetto & Filippin, 2013), participants are tasked with collecting boxes without collecting the one with a bomb in it. Participants see an array of 100 boxes, one of which contains a bomb. They are asked how many boxes they want to collect. The location of the bomb is randomly assigned at the start of the trial. Two versions of the task exist. In the static version, participants just indicate the total number of boxes to collect. The presence/absence of the bomb is revealed at the end of the trial. In the dynamic version, one box is removed per second until participants hit Stop. The presence/absence of the bomb is not revealed until the end of the entire task. For both the versions the risk-neutral or risk-avoidant individual can maximize potential by choosing 50 boxes on each trial.

Cambridge (Rogers) gambling task

The Cambridge Gambling Task (CGT) was developed to take into account some of the concerns with the IGT (Rogers et al., 1999). Participants see an array of 10 boxes/squares, some of which are red and others are blue. They are tasked with guessing which color box a token is hidden in and then placing a bet that their guess is correct. The ratio of red to blue squares varies in each round (1:9, 2:8, 3:7, 4:6, 5:5, 6:4, 7:3, 8:2, and 9:1). Participants start with 100 tokens and can bet 5%, 25%, 50%, 75%, or 95% of their earnings on each trial.

Child version: Cake Gamble (Gambling) Task. The Cake Gambling Task is a child-friendly version of the Cambridge (Rogers) Gambling Task (Van Leijenhorst, Westenberg, & Crone, 2008). The boxes holding a token are replaced with two different flavors of cake, one representing a low-risk gamble and the other a high-risk gamble. Participants bet on which flavor of cake will be randomly selected, with the low-risk bet remaining constant but the high-risk bet varying on each trial. Probabilities of selecting the higher risk gamble (17%, 33%, and 50%) are varied throughout the task.

Card-guessing task

On one variation of the Card-Guessing Task, participants guess whether a playing card presented facedown is higher or lower than a 5 (Delgado, Nystrom, Fissell, Noll, & Fiez, 2000). Correct guesses earn the participant money, whereas incorrect guess can cost the participant money (e.g., Tricomi, Delgado, McCandliss, McClelland, & Fiez, 2006). On a different variation of the Card-Guessing Task (Van Hoorn, Crone, & Van Leijenhorst, 2016), participants are instead dealt two cards. One appears faceup, while the other appears facedown. Participants first guess whether the second card will be a higher or lower card than the first one, then place a bet on their guess. A correct guess results in a gain of double chips, while an incorrect guess results in a loss of the chips that were bet.

Chicken game

The chicken game (Gardner & Steinberg, 2005; Steinberg et al., 2008) is similar to the soon-to-be-described Stoplight Task and is based at least in part on the real-life game of chicken that car drivers were at one time known to play. Participants earn money by driving a virtual car as far as possible but stopping before the clock/timer/meter runs out. In some versions of the task, this timer is represented by a yellow light that can turn to a red light. In other versions, there is instead a visual meter running during the task (Bjork, Momenan, Smith, & Hommer, 2008). Continuing to go past the red light or end of the meter results in a loss on the trial.

Choice dilemmas questionnaire

On the Choice Dilemmas (Kogan & Dorros, 1978; Kogan & Wallach, 1964), participants view a series of real-world dilemmas. In each, the participant is tasked with providing their friend with feedback about which of two options they should take. One option involves a safer option that may not be optimal (e.g., working for an established corporation with job security but less independence), whereas the other option involves risk but potentially more rewarding outcomes (in this same example, opening one’s own law firm with preferred cases, but the possibility of no new clients; Pruitt & Cosentino, 1975). Participants choose between these two options, indicating whether they would tell the friend to choose the riskier option with a 1/10, 3/10, 5/10, 7/10, 9/10, or 10/10 chance of success associated with that option. The lowest probability with which they would advise the friend to choose the risky option is considered the participant’s level of risk on this task. Modifications were made to the task since its creation in order to decrease gender bias in the original scenarios (e.g., Kogan & Dorros, 1978).

Columbia card task

The Columbia Card Task (CCT; Figner & Voelki, 2004, and revised in Figner & Weber, 2011) assesses both hot and cold components of the risky decision making process via parallel versions of the task. In the hot version of the task, participants view a series of 32 cards that when turned over reveal either a smiley face (win) or a sad face (loss). On each trial, participants are given information about the number of loss cards (1 or 3), the amount gained per win card (10 or 30 points), and the amount to be lost if a loss card is turned over (250 or 750 points). Participants turn over cards, one at a time, earning points for each win card. If a loss card is turned over, the trial ends and the loss amount is applied to the accrued points. To bank the most points on a particular trial, participants should stop turning over cards before a loss card is selected and instead click a button to end the trial. The goal of the task is to win as many points as possible. It is believed that participants should utilize the information about the potential wins and losses on a given trial to maximize their likelihood of success (e.g., banking more points) on that trial. Due to the immediate feedback given to participants after turning over each card, and the knowledge of an increasing likelihood a loss card will be selected, the CCT-hot version is thought to measure emotionally based, hot or Type I decision making processes (Figner, Mackinlay, Wilkening, & Weber, 2009). On the parallel version of the task, the CCT-cold, participants again view a series of 32 cards that contain both loss and win cards. They are again given information about the number of loss cards, amount of losses, and amount of grains. But, participants do not turn over any cards on the CCT-cold. Instead, they indicate how many cards, 0–32, they would like to turn over on a given trial. No immediate feedback is given as to how many win/loss cards they turned over prior to the next trial starting. Thus it is believed that this version of the CCT focuses instead on calculated, deliberative, cold or Type II decision making processes (Figner et al., 2009).

Cups task

The Cups Task was designed to assess risky decision making in children (Levin & Hart, 2003) but has been used across a variety of participant ages. Participants choose between a set of boxes (cups) on the left side of the screen and a set of boxes on the right side of the screen (the terms boxes and cups are used interchangeably as researchers use either term when describing the task). There could be an array of 2, 3, or 5 cups