Breakthroughs in Decision Science and Risk Analysis E-Book
118,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Fachliteratur
- Serie: Wiley Series in Operations Research and Management Science
- Sprache: Englisch
Discover recent powerful advances in the theory, methods, and applications of decision and risk analysis Focusing on modern advances and innovations in the field of decision analysis (DA), Breakthroughs in Decision Science and Risk Analysis presents theories and methods for making, improving, and learning from significant practical decisions. The book explains these new methods and important applications in an accessible and stimulating style for readers from multiple backgrounds, including psychology, economics, statistics, engineering, risk analysis, operations research, and management science. Highlighting topics not conventionally found in DA textbooks, the book illustrates genuine advances in practical decision science, including developments and trends that depart from, or break with, the standard axiomatic DA paradigm in fundamental and useful ways. The book features methods for coping with realistic decision-making challenges such as online adaptive learning algorithms, innovations in robust decision-making, and the use of a variety of models to explain available data and recommend actions. In addition, the book illustrates how these techniques can be applied to dramatically improve risk management decisions. Breakthroughs in Decision Science and Risk Analysis also includes: * An emphasis on new approaches rather than only classical and traditional ideas * Discussions of how decision and risk analysis can be applied to improve high-stakes policy and management decisions * Coverage of the potential value and realism of decision science within applications in financial, health, safety, environmental, business, engineering, and security risk management * Innovative methods for deciding what actions to take when decision problems are not completely known or described or when useful probabilities cannot be specified * Recent breakthroughs in the psychology and brain science of risky decisions, mathematical foundations and techniques, and integration with learning and pattern recognition methods from computational intelligence Breakthroughs in Decision Science and Risk Analysis is an ideal reference for researchers, consultants, and practitioners in the fields of decision science, operations research, business, management science, engineering, statistics, and mathematics. The book is also an appropriate guide for managers, analysts, and decision and policy makers in the areas of finance, health and safety, environment, business, engineering, and security risk management.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 545
Veröffentlichungsjahr: 2015
Ähnliche
CONTENTS
Cover
Title page
Foreword
Preface
Contributors
Chapter 1: Introduction: Five Breakthroughs in Decision and Risk Analysis
Historical Development of Decision Analysis and Risk Analysis
Overcoming Challenges for Applying Decision and Risk Analysis to Important, Difficult, Real-World Problems
Chapter 2: The Ways We Decide: Reconciling Hearts and Minds
Do we decide?
Biology and Adaptation
Seu and Game Theory
Prospect Theory
Behavioral Decision Theory
Decisions with a Time Horizon
Morals, Emotions, and Consumer Behavior
Experimental Game Theory
Behavior Modification and Conclusions
References
Chapter 3: Simulation Optimization: Improving Decisions under Uncertainty
Introduction
An Illustrative Example
Optimization of Securities Portfolios
Simulation
A Simulation Optimization Solution Approach
Simulation Optimization Applications in Other Real-World Settings
Conclusions
References
Chapter 4: Optimal Learning in Business Decisions
Introduction
Optimal Learning in the Newsvendor Problem
Optimal Learning in the Selection Problem
Optimizing a Rule-Based Policy for Inventory Management
Discussion
References
Chapter 5: Using Preference Orderings to Make Quantitative Trade-Offs
Introduction
Literature Review
Estimating Attribute Weights from Ordinal Preference Rankings
Illustrative Case Study
Allowing for Negative Weights
Reliability of Partial Rank Orderings
Conclusions and Directions for Future Research
Acknowledgments
References
Chapter 6: Causal Analysis and Modeling for Decision and Risk Analysis
Introduction: The Challenge of Causal Inference in Risk Analysis
How to do Better: More Objective Tests for Causal Impacts
Predictive Models: Bayesian Network (BN) and Causal Graph Models
Deciding What to do: Influence Diagrams (IDS)
When is a BN or ID Causal?
Conclusions: Improving Causal Analysis of Health Effects
Acknowledgments
References
Chapter 7: Making Decisions without Trustworthy Risk Models
Challenge: How to make Good Decisions without agreed-to, Trustworthy Risk Models?
Principles and Challenges for Coping with Deep Uncertainty
Ten Tools of Robust Risk Analysis for Coping with Deep Uncertainty
Applying the Tools: Accomplishments and Ongoing Challenges for Managing Risks with Deep Uncertainty
Conclusions
Acknowledgments
References
Chapter 8: Medical Decision-Making: An Application to Sugar-Sweetened Beverages
Introduction
Medical Ethics and Autonomy
Multiattribute Utility for Preferences of Life and Consumption Under Uncertainty
Analysis Formulation
Case Example: Value to the Individual
Societal Analysis
Quality of Health Considerations
Conclusion
References
Chapter 9: Electric Power Vulnerability Models: From Protection to Resilience
Vulnerability-Analysis Methods
Modeling Cascading Failures in Electric Power Networks
Modeling Restoration Times
Summary
References
Chapter 10: Outthinking the Terrorists
Introduction
Eliciting Attacker Actions from Experts
Using Adaptive Decision and Game Theory
Natural Language Processingto Determine Terrorist Intent
Conclusions
References
Index
End User License Agreement
List of Tables
Chapter 03
Table 3.1 Summary statistics for selected stock prices
Table 3.2 Summary results
Chapter 05
Table 5.1 Attribute values for the 10 US cities with the highest expected terrorism losses
Table 5.2 Hypothetical expert rank orderings for the illustrative case study
Chapter 06
Table 6.1 Some examples of conflicting claims about health effects known to be caused by air pollution
Table 6.2 Some formal methods for modeling and testing causal hypotheses
Chapter 07
Table 7.1 Methods for decision-making with unknown models
Table 7.2 Decision problem for an optimistic scenario
Table 7.3 Decision problem for a pessimistic scenario
Chapter 08
Table 8.1 Notation used in the model
Table 8.2 The per beverage value ranges calculated for the case example parameters
Chapter 09
Table 9.1 Comparison of power network vulnerability models in the literature
List of Illustrations
Chapter 01
Figure 1.1 Plot of expected monetary value (EMV) for profit vs.
N
in the R&D example. Choosing
N
= 7 parallel projects (leads to investigate) maximizes expected net return.
Chapter 02
Figure 2.1 Zero-sum games.
Figure 2.2 The Prisoner’s Dilemma.
Figure 2.3 The Allais paradox. Each outcome has an associated probability.
Figure 2.4 A game of dominance.
Figure 2.5 A coordination game.
Figure 2.6 A four-round Centipede game. Payoffs in dollars. p, pass; t, take.
Chapter 03
Figure 3.1 Crystal Ball screen shot of lognormal probability distribution.
Figure 3.2 Black box approach to simulation optimization.
Figure 3.3 High-level process view.
Figure 3.4 Proposed process.
Figure 3.5 Map of personal claims process.
Figure 3.6 Performance graph for the optimization of the personal claims process.
Chapter 04
Figure 4.1 Process diagram for anticipatory decision-making.
Figure 4.2 Comparison of (a) probability densities and (b) expected profit curves for newsvendor models under exponential and gamma-exponential demand assumptions.
Figure 4.3 Comparison of (a) optimal production quantities and (b) shortage probabilities for different strategies in a problem with 10 planning periods under the same information.
Figure 4.4 Evaluation of the potential of (a) alternative #1 and (b) alternative #2 against the current best option.
Figure 4.5 Illustration of the behavior of a (
q
,
Q
) inventory policy.
Chapter 05
Figure 5.1 Elicited probability densities and point estimates for the attribute weights.
Figure 5.2 Elicited probability densities and point estimates for attribute weights when allowing negative weights.
Figure 5.3 Normalized distance between estimated weights using full rankings versus partial rankings.
Chapter 06
Figure 6.1 An Example Bayesian network (BN) causal model for benzene and leukemia risk.
Figure 6.2 An illustrative ID for net benefits of air pollution control.
Figure 6.3 ID output identifies the emissions reduction that minimizes expected net cost.
Chapter 07
Figure 7.1 A suggested taxonomy of uncertainties
Chapter 08
Figure 8.1 Population mortality by age and gender (data from the US Census; Arias, 2010).
Figure 8.2 General decision tree framework for a decision that affects the risk of death in future life years.
Figure 8.3 Decision tree framework for determining the value of a personal decision.
Figure 8.4 The effect of eta in the trade-off exponential utility model.
Figure 8.5 Effect of changes in relative risk of diabetes on consumption to compensate for risk.
Figure 8.6 Effect of risk tolerance and consumption on the value of daily sugar-sweetened beverage consumption,
η
= 1.5.
Figure 8.7 Expected medical costs associated with a 1.83 relative risk of type 2 diabetes.
Figure 8.8 Using a standard gamble question to determine the utility of a health state.
Chapter 10
Figure 10.1 An influence diagram for the MANPADS countermeasure decision.
Figure 10.2 An influence diagram for the container screening decision.
Figure 10.3 An influence diagram for the defender decision.
Figure 10.4 An influence diagram for the attacker decision.
Figure 10.5 A sequential game theory influence diagram for the attacker–defender decision.
Figure 10.6 An influence diagram using intelligent adversary risk analysis.
Figure 10.7 An influence diagram using adversarial risk analysis with defender uncertainty about attacker beliefs.
Figure 10.8 A Bayesian network of intelligence fusion from Paté-Cornell (2002).
Figure 10.9 A Bayesian network of a bag of words.
Figure 10.10 A Bayesian network of a hidden Markov model for information extraction.
Figure 10.11 Social network of the 9–11 terrorists.
Guide
Cover
Table of Contents
Begin Reading
Pages
ii
iii
iv
vii
viii
ix
x
xi
xii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
Breakthroughs in Decision Science and Risk Analysis
Edited byLouis Anthony Cox Jr.
Cox AssociatesNextHealth TechnologiesUniversity of Colorado-DenverDenver, CO, USA
Cover Image: iStockphoto © Tashatuvango
Copyright © 2015 by John Wiley & Sons, Inc. All rights reserved.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey.Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permissions.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.
For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com.
Library of Congress Cataloging-in-Publication Data:
Breakthroughs in decision science and risk analysis / [edited by] Louis Anthony Cox, Jr. pages cm Includes bibliographical references and index.
ISBN 978-1-118-21716-0 (cloth)1. Decision making. 2. Risk assessment. I. Cox, Louis A. T57.95.B74 2014 658.4′03–dc23
2014012640
Foreword
The Wiley Encyclopedia of Operations Research and Management Science (EORMS), published in 2011, is the first multi-volume encyclopedia devoted to advancing state-of-the-art applications and principles of operations research and management science. EORMS is available online and in print and serves students, academics, and professionals as a comprehensive resource on the field.
The articles published in EORMS provide robust summaries of the many topics and concepts that comprise operations research and management science. However, many readers need additional access to greater details and more thorough discussions on a variety of topics. In turn, we have created the Wiley Essentials in Operations Research and Management Science book series. Books published in this series allow invited expert authors and editors to expand and extend the treatment of topics beyond EORMS and offer new contributions and syntheses.
I am delighted to introduce Breakthroughs in Decision Science and Risk Analysis as the inaugural book in this series. It exemplifies how individual books will meet the goals of the series, setting a high standard for later volumes. Dr. Louis Anthony (Tony) Cox, Jr., the editor of Breakthroughs in Decision Science and Risk Analysis, has assembled a collection of renowned authors who contribute exciting syntheses and advances to an area that is critically important in both applications and research. This book is unique and important because it focuses on recent advances and innovations in decision analysis with an emphasis on topics that are not traditionally found in the decision analysis literature. I am confident that this book will be extremely useful to the operations research and management science community as well as to readers from economics, engineering, business, psychology, finance, environmental sciences, public policy, forestry, political science, health and medicine, education, and other social and applied sciences. We hope that you will enjoy it, and we welcome comments and suggestions for our new Wiley Essentials in Operations Research and Management Science series.
James J. CochranProfessor of Applied Statistics and the Rogers-Spivey Faculty FellowFounding Series Editor of Wiley Essentials in Operations Research and Management ScienceUniversity of Alabama
Preface
Decision and risk analysis can be exciting, clarifying, and fairly easy to apply to improve high-stakes policy and management decisions. The field has been undergoing a renaissance in the past decade, with remarkable breakthroughs in the psychology and brain science of risky decisions; mathematical foundations and techniques; integration with learning and pattern recognition methods from computational intelligence; and applications in new areas of financial, health, safety, environmental, business, engineering, and security risk management. These breakthroughs provide dramatic improvements in the potential value and realism of decision science. However, the field has also become increasingly technical and specialized, so that even the most useful advances are not widely known or applied by the general audience of risk managers and decision-makers who could benefit most from them. This book explains key recent breakthroughs in the theory, methods, and applications of decision and risk analysis. Its goal is to explain them in enough detail, but also with a simple and clear enough exposition, so that risk managers and decision-makers can understand and apply them.
There are several target audiences for this book. One is the operations research and management science (OR/MS) community. This overlaps with the audience for the Wiley Encyclopedia of Operations Research and Management Science (EORMS), including OR/MS professionals–academic and industry researchers, government organizations and contractors, and decision analysis (DA) consultants and practitioners. This book is also designed to appeal to managers, analysts, and decision and policy makers in the applications areas of financial, health and safety, environmental, business, engineering, and security risk management. By focusing on breakthroughs in decision and risk science that can significantly change and improve how we make (and learn from) important practical decisions, this book aims to inform a wide audience in these applied areas, as well as provide a fun and stimulating resource for students, researchers, and academics in DA and closely linked academic fields of psychology, economics, statistical decision theory, machine learning and computational intelligence, and OR/MS.
In contrast to other recent books on DA, this one spends relatively little space on “classical” topics such as the history of DA; the structure of decision problems in terms of acts, states, and consequences; and extensions of the traditional DA paradigm, such as fuzzy DA or multicriteria decision-making. Instead, the book is devoted to explaining and illustrating genuine breakthroughs in decision science—that is, developments that depart from, or break with, the standard DA paradigm in fundamental ways and yet have proved promising for leading to even more valuable insights and decision recommendations in practical applications. These breakthroughs include methods for deciding what to do when decision problems are incompletely known or described, useful probabilities cannot be specified, preferences and value trade-offs are uncertain, future preferences may conflict with present ones, and “model uncertainty” about the cause-and-effect relation between choices and their probable consequences runs too deep to permit any single decision support model, or small set of models, to be highly credible. In addition to explaining the most important new ideas for coping with such realistic challenges, the chapters in this book will show how these techniques are being used to dramatically improve risk management decisions in a variety of important applications, from finance to medicine to terrorism. This emphasis on new ideas that demonstrably work better than older ones, rather than primarily on expositions of and advances in traditional decision and risk analysis, is the essential unique contribution of this work. In addition, a pedagogical emphasis on simple, clear exposition (accessible to readers at different technical levels, with a minimum of mathematical notation and technical jargon), and important practical applications should help to broaden the practical value of the chapters that follow in making important advances in decision and risk analysis useful to readers who want to learn about them and apply them to important real-world decisions.
Louis Anthony Cox, Jr.January 2015
Contributors
Ali E. AbbasIndustrial and Systems Engineering,Center for Risk andEconomic Analysis of Terrorism EventsSol Price School of Public PolicyUniversity of Southern CaliforniaLos Angeles, CA, USA
Marco BetterOptTek Systems, Inc.Boulder, CO, USA
Vicki M. BierDepartment of Industrial and Systems EngineeringUniversity of Wisconsin-MadisonMadison, WI, USA
Louis Anthony (Tony) Cox, Jr.Cox AssociatesNextHealth TechnologiesUniversity of Colorado-DenverDenver, CO, USA
Fred GloverOptTek Systems, Inc.Boulder, CO, USA
Andrea C. HupmanDepartment of Industrial and Enterprise Systems EngineeringCollege of EngineeringUniversity of Illinois at Urbana-ChampaignUrbana-Champaign, IL, USA
Gary KochenbergerBusiness AnalyticsSchool of BusinessUniversity of Colorado-DenverDenver, CO, USA
Philip LeclercDepartment of Statistical Sciences and Operations ResearchVirginia Commonwealth UniversityRichmond, VA, USA
Jason MerrickDepartment of Statistical Sciences and Operations ResearchVirginia Commonwealth UniversityRichmond, VA, USA
Robert OlsenDeloitte ConsultingCenter for Risk Modeling and SimulationWashington, DC, USA
Edward C. RosenthalDepartment of Marketing and Supply Chain ManagementFox School of BusinessTemple UniversityPhiladelphia, PA, USA
Ilya O. RyzhovRobert H. Smith School of BusinessUniversity of MarylandCollege Park, MD, USA
Sinan TasInformation Sciences and TechnologyPenn State-BerksReading, PA, USA
Hristo TrenkovDeloitte ConsultingCenter for Risk Modeling and SimulationWashington, DC, USA
Chen WangDepartment of Industrial EngineeringTsinghua UniversityBeijing, P.R., China
Chapter 1Introduction: Five Breakthroughs in Decision and Risk Analysis
Louis Anthony (Tony) Cox, Jr.
Cox Associates, NextHealth Technologies,University of Colorado-Denver, Denver, CO, USA
This book is about breakthroughs in decision and risk analysis—new ideas, methods, and computational techniques that enable people and groups to choose more successfully when the consequences of different choices matter, yet are uncertain. The twentieth century produced several such breakthroughs. Development of subjective expected utility (SEU) theory combined with Bayesian statistical inference as a model of ideal, rational decision-making was among the most prominent of these. Chapter 2 introduces SEU theory as a point of departure for the rest of the book. It also discusses more recent developments—including prospect theory and behavioral decision theory—that seek to bridge the gap between the demanding requirements of SEU theory and the capabilities of real people to improve their decision-making. Chapters 5 and 8 address practical techniques for improving risky decisions when there are multiple objectives and when SEU cannot easily be applied, either because of uncertainty about relevant values, causal models, probabilities, and consequences; or because of the large number and complexity of available choices.
Historical Development of Decision Analysis and Risk Analysis
Perhaps the most audacious breakthrough in twentieth-century decision analysis was the very idea that a single normative theory of decision-making could be applied to all of the varied risky decisions encountered in life. It may not be obvious what the following problems, discussed in ensuing chapters, have in common:
Investment decisions:
How should investors allocate funds across investment opportunities in a financial portfolio? (
Chapter 3
)
Operations management decisions:
How should a hospital emergency room be configured to make the flow of patients as easy and efficient as possible? How should an insurance company staff its claims-handling operations? (
Chapter 3
)
Inventory management and retail decisions:
How much of an expensive, perishable product should a business buy if demand for the product is uncertain? (
Chapter 4
)
Trial evaluation and selection decisions:
How much trial, testing, and comparative evaluation should be done before selecting one of a small number of costly alternatives with uncertain consequences, for example, in choosing among alternative new public policies, consumer or financial products, health care insurance plans, research and development (R&D) projects, job applicants, supply contracts, locations in which to drill for oil, or alternative drugs or treatments in a clinical trial? (
Chapter 4
)
Adversarial risk management decisions:
How should we model the preferences and likely actions of others, in order to make effective decisions ourselves in situations where both their choices and ours affect the outcomes? (
Chapters 2
,
5
,
9
, and
10
)
Regulatory decisions:
When experimentation is unethical or impractical, how can historical data be used to estimate and compare the probable consequences that would be caused by alternative choices, such as revising versus maintaining currently permitted levels of air pollutants? (
Chapter 6
)
Learning how to decide in uncertain environments:
Suppose that not enough is known about a system or process to simulate its behavior. How can one use well-designed trial-and-error learning to quickly develop high-performance decision rules for deciding what to do in response to observations? (
Chapters 4
and
7
)
Medical decision-making:
How should one trade off the ordinary pleasures of life, such as consumption of sugar-sweetened drinks, against the health risks that they might create (e.g., risk of adult-onset diabetes)? More generally, how can and should individuals make decisions that affect their probable future health states in ways that may be difficult to clearly imagine, evaluate, or compare? (
Chapter 8
)
That the same basic ideas and techniques might be useful for decision-making in such very different domains is a profound insight that might once have excited incredulity among experts in these fields. It is now part of the canon of management science, widely taught in business schools and in many economics, statistics, and engineering programs.
Decision analysis views the “success” of a decision process in terms of the successes of the particular decisions that it leads to, given the information (usually incomplete and possibly incorrect or inconsistent) that is available when decisions must be made. The “success” of a single choice, in turn, can be assessed by several criteria. Does it minimize expected post-decision regret? Is it logically consistent with (or implied by) one’s preferences for and beliefs about probable consequences? In hindsight, would one want to make the same choice again in the same situation, if given the same information? The giants of twentieth-century decision theory, including Frank Ramsey in the 1920s, John von Neumann in the 1940s, and Jimmy Savage in the 1950s, proved that, for perfectly rational people (homo economicus) satisfying certain mathematical axioms of coherence and consistency (i.e., complete and transitive preference orderings for outcomes and for probability distributions over outcomes), all of these criteria prescribe the same choices. All imply that a decision-maker should choose among risky prospects (including alternative acts, policies, or decision rules with uncertain consequences) as if she were maximizing subjective expected utility (SEU). Chapters 2 and 7 introduce SEU theory and some more recent alternatives. Decision-making processes and environments that encourage high-quality decisions as judged by one of these criteria will also promote the rest.
However, real people are not perfectly rational. As discussed in Chapter 2, homo economicus is a fiction. The prescriptions of decision theory are not necessarily easy to follow. Knowing that SEU theory, the long-reigning gold standard for rational decision-making, logically implies that one should act as if one had coherent (e.g., transitive) preferences, and clear subjective probabilities are cold comfort to people who find that they have neither. These principles and limitations of decision theory were well understood by 1957, when Duncan Luce and Howard Raiffa’s masterful survey Games and Decisions explained and appraised much of what had been learned by decision theorists, and by game theorists for situations with multiple interacting decision-makers. Chapter 2 introduces both decision theory and game theory and discusses how they have been modified recently in light of insights from decision psychology and behavioral economics.
During the half-century after publication of Games and Decisions, a host of technical innovations followed in both decision analysis and game theory. Decision tree analysis (discussed in Chapters 8 and 10) was extended to include Monte Carlo simulation of uncertainties (see Chapter 3). Influence diagrams were introduced that could represent large decision problems far more compactly than decision trees, and sophisticated computer science algorithms were created to store and solve them efficiently. Methods of causal analysis and modeling were developed to help use data to create risk models that accurately predict the probable consequences of alternative actions (see Chapter 6). Markov decision processes for dynamic and adaptive decision-making were formulated, and algorithms were developed to adaptively and robustly optimize decision rules under uncertainty (see Chapter 7). SEU theory was generalized, e.g., to allow for robust optimization with ambiguity aversion when probabilities are not well known. Practical constructive approaches were created for structuring and eliciting probabilities and utilities, as discussed and illustrated in Chapters 5, 8, and 10.
These technical developments supported a firmly founded discipline of applied decision analysis, decision aids, and decision support consulting. The relatively new discipline of applied decision analysis, developed largely from the 1960s on, emphasized structuring of decision problems (especially, identifying and solving the right problem(s)); clearly separating beliefs about facts from values and preferences for outcomes; eliciting or constructing well-calibrated probabilities and coherent utilities; presenting decision recommendations, together with sensitivity and uncertainty analyses, in understandable ways that decision-makers find useful; assessing value of information and optimal timing of actions; and deliberate, careful learning from results, for both individuals and organizations. The 1976 publication of the landmark Decisions with Multiple Objectives: Preferences and Value Tradeoffs by Ralph Keeney and Howard Raiffa summarized much of the state of the art at the time, with emphasis on recently developed multiattribute value and utility theory and methods. These were designed to allow clearer thinking about decisions with multiple important consequence dimensions, such as costs, safety, profitability, and sustainability. Chapters 5, 8, and 10 review and illustrate developments in elicitation methods and multiattribute methods up to the present.
While decision analysis was being developed as a prescriptive discipline based on normative theory (primarily SEU theory), an increasingly realistic appreciation of systematic “heuristics and biases” and of predictable anomalies in both laboratory and real-world decision-making was being developed by psychologists such as Amos Tversky, Daniel Kahneman, Paul Slovic, and Baruch Fischhoff, and by many other talented and ingenious researchers in what became the new field of behavioral economics. Chapter 2 introduces these developments. Striking differences between decision-making by idealized, rational thinkers (homo economicus) and by real people were solidly documented and successfully replicated by different teams of investigators. For example, whether cancer patients and their physicians preferred one risky treatment procedure to another might be changed by presenting risk information as the probability of survival for at least 5 years instead of as the probability of death within 5 years—two logically equivalent descriptions (gain faming vs. loss framing) with quite different emotional impacts and effects on decisions. Many of these developments were reflected in the 1982 collection Judgment under Uncertainty: Heuristics and Biases, edited by Kahneman, Slovic, and Tversky. Chapter 10 summarizes key insights from the heuristic-and-biases literature in the context of eliciting expert judgments about probabilities of adversarial actions.
Twenty-five years later, the 2007 collection Advances in Decision Analysis: From Foundations to Applications, edited by Ward Edwards, Ralph Miles, and Detlof von Winterfeldt, took stock of the thriving and increasingly well-developed field of decision analysis, which now integrated both normative (prescriptive) theory and more descriptively realistic considerations, e.g., using Kahneman and Tversky’s prospect theory. This collection looked back on decades of successful developments in decision analysis, including the field’s history (as recalled by founding luminaries, including Ron Howard of Stanford and Howard Raiffa of the Harvard Business School), surveys of modern progress (including influence diagrams, Bayesian network models, and causal networks), and important practical applications, such as to engineering and health and safety risk analysis, military acquisitions, and nuclear supply chain and plutonium disposal decisions.
Overcoming Challenges for Applying Decision and Risk Analysis to Important, Difficult, Real-World Problems
Despite over five decades of exciting intellectual and practical progress, and widespread acceptance and incorporation into business school curricula (and into some engineering, statistics, mathematics, and economics programs), decision analysis has limited impact on most important real-world decisions today. Possible reasons include the following:
Many real-world problems still resist easy and convincing decision-analytic formulations.
For example, a dynamic system with random events (i.e., patient arrivals, departures, and changes in condition in a hospital ward) with ongoing opportunities to intervene (e.g., by relocating or augmenting staff to meet the most pressing needs) cannot necessarily be represented by a manageably small decision tree, influence diagram, Markov decision process, or other tractable model—especially if the required transition rates or other model parameters are not known and data from which to estimate them are not already available.
Chapters 3
,
4
,
6
, and
7
present breakthroughs for extending decision and risk analysis principles to such realistically complex settings. These include increasingly well-developed
simulation–optimization
methods for relatively well-characterized systems (
Chapter 3
) and
adaptive learning
, statistical methods for estimating causal relations from data (
Chapter 6
), model ensemble, and robust optimization methods for settings where not enough is known to create a trustworthy simulation model (
Chapters 4
and
7
).
It has often not been clear that individuals and organizations using formal decision analytic models and methods outperform and outcompete those who do not.
Chapters 4
,
6
, and
7
emphasize methods for causal analysis, adaptive learning, and empirical evaluation and comparison of alternative choices. These methods can help decision-makers make choices that demonstrably outperform (with high statistical confidence) other available choices in a wide variety of practical applications.
While decision analysts excel at distinguishing clearly between matters of fact and matters of preference, real-world decision-makers often prefer to fall back on judgments that conflate the two
, perhaps feeling that no matter what academic theory may say, such holistic judgments give more satisfactory and trustworthy recommendations than calculations using hypothetical (and not necessarily clearly perceived or firmly believed) subjective utilities and probabilities. (This tendency may perhaps explain some of the popularity of simplistic decision aids that use ill-defined concepts, such as “relative importance” of goals or attributes, without clear definition of what “relative importance” means and of how it should reflect interdependencies.) Too often, there is simply no satisfactory way to develop or elicit credible, defensible, widely shared probabilities and utilities for situations or outcomes that are novel, hard to imagine, or controversial.
Chapters 5
and
8–10
discuss innovations for alleviating this problem with new methods for eliciting and structuring utilities, value trade-offs, and probabilistic expert beliefs.
Most real-world decisions involve multiple stakeholders, influencers, and decision-makers,
but SEU is preeminently a theory for single-person decisions. (Extensions of SEU to “social utility” for groups, usually represented as a sum of individual utilities, can certainly be made, but an impressive list of impossibility theorems from collective choice theory establish that
homo economicus
will not necessarily provide the private information needed for collective choice mechanisms to produce desirable, e.g., Pareto-efficient, outcomes.) Less theoretically, the notorious Prisoner’s Dilemma, discussed in
Chapter 2
, illustrates the tension between individual and group rationality principles. In the Prisoner’s Dilemma, and in many other situations with externalities, individuals who make undominated or otherwise individually “rational” choices will thereby collectively achieve Pareto-dominated outcomes (a hallmark of collectively suboptimal choice), meaning that everyone would have been better off if all had made different (not “rational”) choices.
Chapter 2
discusses both classical game theory and its behavioral modifications to better apply to real people, who often cooperate far better than theories for merely “rational” individuals would predict.
Chapters 9
and
10
consider applications of game theory and alternatives for defending electrical grids (
Chapter 9
) and other targets (
Chapter 10
) against terrorists or other adversaries, including natural disasters in
Chapter 9
.
In addition to these major conceptual challenges, there are also purely technical challenges for making decision-analytic principles more widely applicable. For example, decision trees (Chapter 5) are well suited to model (and if necessary simulate) alternative possible sequences of events and decisions when there are only a few of each. However, they are far from ideal when the number of choices is large or continuous.
Example: Searching for a Hidden Prize
Suppose that a prize is hidden in one of 100 boxes, and that the cost of opening each box to see whether the prize is in it, as well as the prior probability that it is, are known (say, c(j) to open box j, which has prior probability p(j) of containing the prize). Then in what order should the boxes be opened to minimize the expected cost of finding the prize? (This is a very simple model for sequential search and R&D problems.) It would clearly be impracticable to create a decision tree describing the 100! possible orders in which the boxes might be opened. Yet, it is easy to determine the optimal decision. Simple optimization reasoning establishes that the boxes should be opened in order of descending probability-to-cost ratio (since interchanging the order of any two boxes that violate this rule can readily be seen to reduce expected cost).
The remainder of this book explains and illustrates breakthrough methods to help real people make real decisions better. It presents ideas and methods that the authors and editors believe are mature enough to be highly valuable in practice and that deserve to be more widely known and applied. The starting point, developed in Chapter 2, is a candid acknowledgment that:
SEU and classical Bayesian decision analysis together provide a logically compelling model for how individual decisions ideally should be made; but
Real people have not evolved to be always capable of producing (or agreeing on) the crisply specified, neatly decoupled subjective probabilities and utilities that are required (and implied) by SEU.
Instead, decision-makers in real-world organizations and institutions typically have access to some imperfect data and knowledge bearing on the causal relations between alternative choices and their probable consequences. From this partial information, and through processes of deliberation and analysis, they may construct preferences for alternative actions, with supporting rationales that are more or less convincing to themselves and to others.
The following five breakthroughs, explained and illustrated in the chapters that follow, can help to understand and improve these decision processes.
Breakthrough 1: Behavioral Decision Theory and Game Theory
It is now known that different neural pathways and parts of the brain are activated by different aspects of risky prospects, such as probabilities versus amounts of potential perceived gains or losses; immediate versus delayed consequences; positive versus negative emotional affects of cues used in describing them; trust versus suspicion of others involved in joint decisions; and moral reactions to risks and to imposing risks on others. To a useful first approximation, heart and head (or, more formally, “System 1” and “System 2,” referring to the quick, intuitive and slower, more cognitive aspects of decision-making, respectively) may disagree about what is best to do, using different parts of the brain to evaluate alternatives and to arrive at these conclusions. Chapter 2 further describes these aspects of the divided decision-making self, situating the problem of wise decision-making (as well as moral and social judgment-making) in the context of competing decision pathways and emphasizing the need to use both emotional and intuitive judgments and rational calculations in making effective decisions. Preferences, judgments, and beliefs are often transient and contingent on context and cues (some of which may be logically irrelevant) when they are elicited. Assessing a single coherent utility function for a person whose preferences arise from multiple cues and competing pathways may not yield a reliable basis for prescribing what to do if the goal is to minimize post-decision regret.
Recognizing these realities of human nature and behavior motivates behavioral decision theory and behavioral game theory. These address how to use well-placed “nudges,” design of occupational and consumer environments, and other practical methods to help real people make better decisions while taking into account their heuristics, biases, inconsistencies, limited attention span and will-power, irrational altruism, moral aspirations, perceptions of fairness, desire for social approval, and other realities of motivation and behavior. “Better” decisions can no longer necessarily be defined as those that are consistent with SEU axioms for preferences and beliefs, as assessed at a given moment and in a given context. What constitutes desirable decision-making must be defined afresh when preferences and beliefs are seen as being constructed on the fly and subject to motivated and self-serving reasoning, wishful thinking, salience of cues and their emotional affects, priming by context, and other biases. For example, “good” choices might be defined as those that are consistent with the guidance or principles (more formally, with the if-then “decision rules” mapping available information to available actions) that one would ideally want one’s self to follow, if one were given time, resources, and ability to develop such principles outside the context of short-run distractions, passions, and temptations. Such reflective and reflexive considerations, which have a long tradition in deontological, utilitarian, and virtue ethics, are gaining new currency and an applied focus through behavioral decision theory and game theory. The core breakthrough in Chapter 2 is the insight that advice on how to make decisions, to be most useful, should be rooted to an understanding of real human behavior and realistic possibilities for changing it.
Breakthrough 2: Simulation–Optimization of Risk Models
For decades, one vision of applied decision analysis has been that knowledge and information about the system or situation that a decision-maker seeks to influence via her decisions should be represented in an explicit risk model relating decisions (controllable inputs) and uncertainties (e.g., modeled as random inputs from the environment, not controlled by the decision-maker) to resulting probabilities of different outputs (consequences). If expected utility, or any other “objective function” whose expected value is to be maximized, is used to evaluate the probabilities of consequences induced by alternative choices of controllable inputs, then the decision problem of selecting those inputs can be decomposed into the following two technical tasks:
Simulate
output (consequence) probability distributions, for any choice of inputs; and
Optimize
inputs, that is, identify a combination of input values to produce the most desirable probability distribution of outputs, as evaluated by the objective function (e.g., expected utility).
If the risk model and simulation–optimization process can be trusted to model adequately the real system or situation of interest and to automatically find the best combination of controllable inputs to choose for that system, then the decision-maker is freed to focus on specifying the controllable inputs and the objective function to be used in evaluating results. Appropriate subject matter experts and modelers can focus on developing and validating the risk model describing the probabilities of consequences caused by different choices of controllable inputs (together with the uncontrollable ones chosen by “nature” or by others). The simulation–optimization engine can handle the details of solving for the best choice of inputs, much as software products such as the Excel Solver or Wolfram Alpha can solve simper problems, freeing users to focus on model development and input specification.
Example: Optimal Level of R&D
Suppose that a pharmaceutical company can invest in investigating any of a large number of new leads (molecules and molecular signaling pathways) in parallel for developing a drug to treat a disease in patients with a certain genotype. Each lead costs $5M to investigate, and each has a probability 0.1 of proving successful within 5 years. The value of a success within 5 years is $100M. The company must decide how many leads to investigate in parallel. What number of leads should be investigated to maximize the expected profit? This objective function, in units of millions of dollars, is given by the formula:
where N is the number of leads investigated—the decision variable in this problem—and p is the probability of success for each investigated lead. (The probability that all N investigated leads are unsuccessful is (1 − p)N, and hence the probability of success for the whole effort, that is, the probability that not all fail, is (1 − (1 − p)N).) The expected profit-maximizing value of N can readily be found in this simple example by searching over a range of values of N. For example, for those familiar with R, the following code generates Fig. 1.1: N = c(1:20); EMV = 100 × (1 − 0.9^N) − 5 × N; plot (N, EMV). The number of leads that maximizes the objective function is N = 7.
Figure 1.1 Plot of expected monetary value (EMV) for profit vs. N in the R&D example. Choosing N = 7 parallel projects (leads to investigate) maximizes expected net return.
Now, suppose that the problem were more complicated, with unequal success probabilities and different costs for the different leads, and with annual budgets and other resource constraints limiting the number of projects (investigations) that could be undertaken simultaneously. Then instead of searching for the best solution over a range of values for N, it would be necessary to search a more complicated space of possible solutions, consisting of all subsets of projects (lead investigations) that can be investigated simultaneously (i.e., that satisfy the budget and resource constraints). If, in addition, the objective function could not easily be described via a formula, but instead had to be estimated by simulating many realizations of the uncertain quantities in the model for each choice of inputs, then efficient search and evaluation of different input combinations might become important, or even essential, for finding a good solution to the decision problem. Simulation-optimization provides technical methods for efficiently searching complex sets of feasible decisions, performing multiple simulation-based evaluations of alternative combinations of controllable inputs to identify those that (approximately) optimize the user-specified objective function.
The vision of decision-making as optimization of an appropriate objective function subject to constraints, corresponding to a model of how choices affect consequence probabilities, is fundamental in economics, operations research, and optimal control engineering (including stochastic, robust, and adaptive control variations). However, to make it practical, both the simulation and the optimization components must be well enough developed to apply to realistically complex systems. This is an area in which huge strides have been made in the past two decades, amounting to a breakthrough in decision problem-solving technology. Sophisticated stochastic simulation techniques (e.g., Gibbs sampling, more general Markov chain Monte Carlo (MCMC) techniques, Latin Hypercube sampling, importance sampling, and discrete-event simulation methods) and sophisticated optimization methods that work with them (e.g., evolutionary optimization, simulated annealing, particle filtering, tabu search, scatter search, and other optimization meta-heuristics) are now mature. They have been encapsulated in user-friendly software that presents simple interfaces and insightful reports to users, who do not need to understand the details of the underlying algorithms. For example, the commercially available OptQuest simulation-optimization engine discussed in Chapter 3 is now embedded in software products such as Oracle, Excel, and Crystal Ball. Its state-of-the-art optimization meta-heuristics make it practical to easily formulate and solve decision problems that once would have been formidable or impossible.
The basic breakthrough discussed in Chapter 3 is to extend to realistically complex decisions the key decision and risk analysis principles of (i) predicting probable consequences of alternative actions (using stochastic simulation-based risk models of the relation between actions and their probable consequences, which may include complex, nonlinear interactions and random transitions) and (ii) finding the “best” feasible actions, defined as those that create preferred probability distributions for consequences. This is accomplished via simulation-optimization models, in which computer simulation models are used to represent the probable behavior of the system or situation of interest in response to different choices of controllable inputs. Powerful heuristic optimization methods, extensively developed over the past two decades, then search for the combination of controllable inputs that produces the most desirable (simulated) distribution of outcomes over time. Modern simulation-optimization technology substantially extends the practical reach of decision analysis concepts by making them applicable to realistically complex problems, provided that there is enough knowledge to develop a useful simulation model.
Breakthrough 3: Decision-Making with Unknown Risk Models
Simulation-optimization technology provides a breakthrough for deciding what to do if the causal relation between alternative feasible actions and their probable consequences—however complex, nonlinear, dynamic, and probabilistic it may be—is understood well enough to be described by a risk model that can be simulated on a computer. But suppose that the relation between controllable inputs and valued outputs is unknown, or is so uncertain that it cannot usefully be simulated. Then different methods are needed. One possibility, discussed in Chapters 4 and 7, is to learn from experience, by intelligent trial and error. In many applications, one might dispense with models altogether, and experiment and adaptively optimize interactions with the real system of interest, in effect replacing a simulation model with reality. (This is most practical when the costs of trial and error are relatively small.) For example, a marketing campaign manager might try different combinations of messages and media, study the results, and attempt to learn what combinations are most effective for which customers. Or, in the domain of healthcare, a hospital might try different drugs, treatment options, or procedures (none of which is known to be worse than the others) with patients, collect data on the results, and learn what works best for whom—the basic idea behind clinical trials. Chapter 4 considers optimal learning and anticipatory decision-making, in which part of the value of a possible current decision is driven by the information that it may reveal, and the capacity of such information to improve future decisions (the “value-of-information” (VOI) concept from traditional decision analysis). Recognizing that any current “best” estimated model may be wrong, and that future information may lead to changes in current beliefs about the best models, and hence to changes in future decision rules, can help to improve current decisions. Chapter 7 also discusses “low-regret” decision-making, in which probabilities of selecting different actions, models to act on, or decision rules are adaptively adjusted based on their empirical performance. Such adaptive learning leads in many settings to quick convergence to approximately optimal decision rules.
A second approach to decision-making with initially unknown risk models is possible if plentiful historical data on inputs and outputs are available. This is to estimate a relevant causal risk model from the available historical data. The estimated model can then be used to optimize decisions (e.g., selection of controllable inputs, or design of policies or decision rules for selecting future inputs dynamically, based on future observations as they become available). Chapter 6 briefly surveys methods of causal analysis and modeling useful for constructing risk models from historical (observational) data; for testing causal hypotheses about the extent to which controllable inputs (e.g., exposures) actually cause valued outputs (e.g., changes in health risks); and for estimating the causal impacts of historical interventions on outcomes. Chapter 7 discusses what to do when more than one possible model fits available data approximately equally well, making it impossible to confidently identify a unique risk model from the data. In this case, model ensemble methods, which combine results from multiple plausible models, can give better average performance than any single model.
Finally, a third possible approach to decision-making with unknown or highly uncertain models, also discussed in Chapter 7, is to seek “robust” decisions—that is, decisions that will produce desirable consequences no matter how model uncertainties are eventually resolved. Of course, such a robust decision may not always exist. But a rich theory of robust optimization (and of its relations to related innovative concepts, including coherent risk measures and to older techniques such as stochastic programming) has been developed relatively recently, and this theory shows that many risk management decision problems of practical interest can be formulated and solved using robust optimization techniques. Chapter 7 and its references discuss these recent developments further.
Taken together, these relatively recent techniques for dealing with model uncertainty in decision-making constitute a distinct improvement over earlier methods that required decision-makers to, in effect, specify a single best-estimate model (typically based on subjective probabilities). Allowing for model uncertainty leads to new and useful principles for adaptive learning from data and for low-regret and robust optimization. These hold great promise for a wide variety of practical risk management decision problems, as illustrated by examples in Chapters 4 and 7.
Breakthrough 4: Practical Elicitation and Structuring of Probabilities and Multiattribute Utilities
The first three breakthroughs—behavioral decision and game theory, simulation-optimization, and methods for learning risk models from data and, in the interim, for making decisions with unknown or highly uncertain risk models—represent substantial enhancements to or departures from traditional SEU-based decision analysis. Breakthrough 4 consists of methods for making SEU theory more applicable to complex and difficult real-world problems by eliciting preferences and beliefs via techniques that impose less cognitive load on users and/or that achieve greater consistency and reliability of results than older methods. Chapter 5 considers state-of-the-art methods for developing multiattribute utility functions. This is a potentially painstaking task that once involved estimating multiple trade-off parameters and verifying quite abstract independence conditions for effects of changes in attribute levels on preferences. Chapter 5 presents much simpler and more robust methods, developed largely in marketing science, to enable relatively quick and easy development of multiattribute utility functions from simple preference orderings.
Breakthrough 5: Important Real-World Applications
The final category of breakthroughs consists of applications of decision and risk analysis principles to important and difficult fields that have historically relied on other methods. Chapters 5, 9, and 10 discuss applications of expert elicitation, game theory, decision tree analysis, Bayesian networks (for text mining of natural language), and machine learning techniques to the challenges of modeling and defending against adversarial actions. Chapter 8 illustrates the application of multiattribute utility theory to medical decision-making problems, at both the individual and the societal levels, by assessing utility functions for making trade-offs between consumption of sugar-sweetened beverages and risks of morbidity (type 2 diabetes) and early mortality. Chapter 9 discusses vulnerability, resilience, and defense of complex systems—specifically, electric power networks—and compares the insights gleaned from game-theoretic models and considerations to those from less sophisticated methods, concluding that the more sophisticated methods are often very worthwhile. Chapter 6 suggests that many public health decisions that are based on attempts to interpret associations causally would be much better served by applying more objective (and now readily available) methods of causal analysis and risk modeling.
In each of these application areas, and countless others, decision support methods have long been used that do not incorporate the precepts of SEU theory, modern decision analysis, simulation, optimization, optimal learning, or analysis and deliberation using causal risk models of causal relations. In each of these areas, adopting improved methods can potentially achieve dramatic objective improvements in average outcomes (and in entire probability distributions of outcomes). This point is made and illustrated by dozens of examples and case studies in the chapters that follow. Adopting the methods discussed in this book, implementing them carefully, and monitoring and learning from the results can yield breakthrough improvements in areas including marketing, regulation, public health, healthcare and disease risk management, infrastructure resilience improvement, network engineering, and homeland security. The conceptual and methodological breakthroughs presented in the following chapters were selected because they are ready for practical use and because they have been found to create great benefits in practice. Opportunities to apply them more widely are many, and the likely rewards for doing so are great.
Chapter 2The Ways We Decide: Reconciling Hearts and Minds
Edward C. Rosenthal
Department of Marketing and Supply Chain Management, Fox School of Business, Temple University, Philadelphia, PA, USA
The story of Odysseus and his encounter with the Sirens is one of the oldest and most psychologically captivating tales in the Western literary tradition. The sea god Poseidon, you may remember, cursed Odysseus to wander the sea for 10 years, and he and his crew faced countless and remarkable ordeals—among them, getting past the Sirens. The Sirens were known for their beautiful singing; in fact, all sailors who came within earshot of them would compulsively steer their boat closer, only to perish on the rocky shore.
Knowing of the Sirens’ reputation, Odysseus was determined to avoid disaster, but he desperately wanted to hear the songs for himself. His solution was to have his men plug up their ears with wax and tie him to the mast, under strict orders not to release him until they were well out of danger. Spellbound by the beautiful voices, Odysseus vehemently insisted that the crew untie him, but his men obeyed the plan and freed him only after they had sailed away.
What does this mean for us? Homer’s tale is the first known account of someone who, in a cold and unemotional state, was able to anticipate his urges when in an aroused state and take action to prevent those urges from being acted upon, thus avoiding a disastrous outcome. Even thousands of years ago, we realize, people recognized the very real conflict between the emotional and unemotional, or the hot and the cold, facets of our decision making. However, even today, the coexistence and interplay between these very different drivers of our behavior are not as well understood as they could be.
Decision theorists, as well as psychologists, are eager to answer the question, How do we make decisions? This chapter explores the two different sources to which we typically attribute our decisions when challenged to explain them—hearts and minds—in terms of their roles, their interactions, and the conflicts between them. First, we will explore some biological and psychological underpinnings: Do we actually “make” decisions, or are our choices and actions determined for us by unconscious deterministic processes? What are the Systems 1 and 2 that psychologists often mention? How might evolution have shaped our decision processes?
Next, we will look at scientific attempts to provide a sound foundation for decision making. We’ll peek at the underpinnings of subjective expected utility theory (SEU)—a powerful model meant to formalize our preferences among different actions when their outcomes are uncertain. After that, in order to understand strategic behavior, we will quickly review some important results from theoretical and empirical game theory. Game theory and SEU together provide a prescriptive methodology for how to act in well-defined situations with the following features: different outcomes (or consequences) of choices are possible; the different outcomes are not all equally desirable (since otherwise it would not matter what one did); the outcomes are uncertain (and their probabilities depend on what actions are taken); there may be multiple decision makers whose choices are jointly responsible for bringing these different outcomes about, or at least affecting their probabilities; and the preferences of different decision makers might conflict. For the remainder of this chapter, I use the word “rational,” to refer to decisions and behaviors that would be suggested by an SEU or game-theoretic model of a particular situation.
Game theory, together with SEU, forms an intimidating edifice. This normative approach represents the cold and unemotional side of life. And since SEU and game theory are scientifically formulated to provide the best responses and, loosely speaking, best overall outcome probabilities when faced with a decision, it seems reasonable to think that no other actions can be superior to the rational ones prescribed by the theory. But, after having seen the highlights of these approaches, we will explore their limitations—the chinks in the armor, so to speak. How do our emotions enter into our decision making? How should they? Can emotionally charged decisions work out better than the rational ones? To explore these questions, we will review prospect theory. This descriptive theory shows that the normative framework of SEU does not seem to accurately describe or explain actual human behavior, and identifies and describes systematic differences between actual and SEU-consistent decisions. We then go beyond prospect theory to observe other important examples of how homo sapiens seem to interpret their world and act on certain systematic biases. These examples include recent findings in the fields of neuroeconomics, behavioral economics, experimental game theory, moral decision-making, intertemporal choice, and consumer behavior.
While prospect theory and behavioral decision theory describe our preferences in certain situations, they do not address how we do or should choose in situations of conflict or cooperation
