The Art and Science of Making Up Your Mind

Rex V. Brown, Jonathan Baron, Karen Brown, Jonathan Baron, Karen Brown

• Publisher: Routledge
• Year: 2019
• eBook ISBN: 9781000024654
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

The Art and Science of Making Up Your Mind presents basic decision-making principles and tools to help the reader respond efficiently and wisely to everyday dilemmas.

Although most decisions are made informally (whether intuitively without deliberate thought, or based on careful reflection), over the centuries people have tried to develop systematic, scientific and structured ways in which to make decisions. Using qualitative counterparts to quantitative models, Rex Brown takes the reader through the basics, like 'what is a decision' and then considers a wide variety of real-life decisions, explaining how the best judgments can be made using logical principles.

Combining multiple evaluations of the same judgment ("hybrid judgment") and exploring innovative analytical concepts (such as "ideal judgment"), this book explores and analyzes the skills needed to master the basics of non-mathematical decision making, and what should be done, using real world illustrations of decision methods.

The book is an ideal companion for students of Thinking, Reasoning and Decision-Making, and also for anyone wanting to understand how to make better judgments in their everyday lives.

Chapter 1

Introduction

As author, my goal is to offer readers, whatever their aptitude, interest or background, the skills to master a basic set of non-mathematical decision-making tools. Those skills will enable them to make wiser everyday decisions, both large and small, which will lead them to more satisfying lives. This text is a first step along that path, with a narrower scope: choices among options for personal dilemmas, such as whether to get married or whom to vote for. And unlike traditional decision-analysis, this book aims to bypass the burdensome technical methods that often turn off the everyday decider.

It is designed for use in a college course, but can also serve as “self-help” for the general reader.

1.1 Potential need

The prevalence of poor decision-making has been well-established and documented in best-sellers such as Daniel Kahneman’s “Thinking, fast and slow” (2011). People make terrible decisions when they should have known better, and it may cost them dearly. They marry partners who are likely to make their lives miserable. They vote for a candidate even when they have clear evidence to the contrary that another candidate would govern more to their liking. There is plentiful research on how people make decisions already, but little on how people could make wiser decisions.

In the latter part of the 20th century, “Applied Decision Theory (ADT)” was developed, and the method was a substantial vogue in academic and professional circles. It was widely considered a universal key to wise decision making and widely adopted by business and major institutions. The idea was to analyze decisions in terms of options, the possible outcomes of choosing each option, their probabilities and a numerical measure of how much you value each outcome, its “utility.” In this way, you could calculate an “expected utility” for each option and then choose the option that would potentially have the most value for you.

And yet, by the end of the century, ADT had lost its shine, its usefulness coming under fire. ADT was no longer the wave of the future. As with most theories, the devil was in the details. I did not share this blanket dismissal; my reading was that the ADT logic was sound and that this was not what was being challenged. Wise decisions conformed to decision theory norms, but the most useful way to make a decision did not necessarily take the form of any theoretical model. Other approaches such as intuition and feedback from real-world decision-making practice might bring the decider closer to the ADT ideal. Ordinary people could not use the tool, as then developed, cost-effectively. It was difficult to understand, laborious to implement, and rarely outperformed common sense. Operational methods needed to be adapted to fit cognitive capacity and practical requirements. The tool and its application needed more than logic; it needed psychology and feedback from practice. The methodology and its practitioners needed threefold skills in logic, psychology and practice feedback, rarely found in the decision making profession.

A highly regarded ADT-based text book by Hammond, Keeney and Raiffa (1999) features different methodological devices and topics and its contribution is complimentary, not competing with this book. It expounds difficult ideas with rare clarity. These authors and I share the same decision theory perspective. All four of us were part of the Harvard team that originally developed ADT in the 1960s. However, a distinguishing feature of the present book is to enhance (rather than replace) the usual decider’s decision processes. Together, they could support a single course on prescriptive decision analysis.

1.2 Distinctive features of this text

• Qualitative counterparts to quantitative ADT models.

• Uses the totality of the decider (D’s) knowledge rather than just what is called for by a single ADT analysis.

• Exercises students’ real judgments (as opposed to hypothetical or other people’s judgments used in traditional case studies).

• Illustrates argument with real examples.

• Combines multiple evaluations of the same judgment (“hybrid judgment”).

• Provides real-world illustrations of decision methods.

• Introduces innovative analytical concepts (such as “ideal” judgment).

This text can be primary support for a short course in decision-making, as part of a variety of programs, such as psychology, management and philosophy. It can complement other decision-aiding texts, notably Hammond et al. (1999), which shares the same perspective, or “descriptive” behavioral decision texts, such as Baron’s “Thinking and deciding” (2008).

Chapters are largely modular, and lend themselves to being taught as self-contained segments. Segments could be selective in emphasis: qualitative with or without quantitative treatment; personal vs. civic domains; factual vs. value judgments; case-studies vs. method exposition; more vs. less advanced material.

1.3 Vocabulary

The vocabulary here represents a significant modification to prevailing professional practice. I have found that students, clients and most others (including academics) who have not been indoctrinated in current practice are commonly confused or misled by much decision science language in common use. So, I have substituted, for current language, I hope, clearer language, which I have found communicates better. For example, I use “applied decision theory” in place of “decision analysis,” which misleadingly suggests a generic way of analyzing a decision.

Language has proven to be such an impediment to communication as to put people off from learning or using modern decision aids. I have painful memories, from my early consulting days, of presenting what I thought was the clear analysis of an investment option to a company president. After listening for ten minutes, he muttered “gobbledygook!” and stormed out of the room and had nothing further to do with our work. My goal is to be welcoming and inclusive with the language, and certainly to keep everyone in the room.

1.4 Historical background

Before the 20th century, major scientific advances oriented toward decision-making had tended to be descriptive more than prescriptive, focusing on how the world works rather than how to make it work better. Decision circumstances changed slowly from year to year, so that decision practice could take its time to improve by trial and error. By the beginning of the 20th century, however, technology and other fields had begun to change rapidly. The life-and-death perils of poor decisions in World War II spurred the development of the quantitative decision tools of Operations Research. They were special-purpose tools (e.g., for locating enemy submarines) that may well have been decisive in winning the war.

After the War, Operations Research was adapted to industry, with some success for certain situations. These tended to be where options were complex and consequences were clearly defined, and involved processes that could be mathematically modeled (such as in production scheduling and transportation logistics). Progress in applying quantitative methods to choices involving a few clear-cut options with messy outcomes was a good deal slower. Analysis here competed less effectively with unaided humans, and deciders often did better by backing their own judgment.

The mid 20th century saw the development of general-purpose statistical decision theory, which can readily adapt to changing circumstances and, in principle, analyze any choice whatsoever. It does so by quantifying a decider’s judgments about goals, options and outcomes, however ill-defined, and by inferring the preferred choice. Its practical application is Applied Decision Theory (ADT).

In the early 1960s, research groups at Harvard and Stanford developed and promoted ADT as a universal methodology for improving rationality in a world where poor decisions were damaging lives and communities. Leading corporations (like Dupont, General Electric and Kodak), and then government departments (like Defense and Energy) began to apply decision analysis to their most challenging (and controversial) decisions. Many impressive successes were reported, and those of us who had been in decision analysis “on the ground floor” viewed decision analysis with a missionary zeal.

Decision analysis has passed through several overlapping phases, characterized by distinctive modes of aiding, each building on the earlier ones.

• The theory phase laid the foundations of the parent discipline, statistical decision theory (about 1950 to 1970)

• The technique phase focused on specific modeling procedures and sought illustrative applications (about 1960 to 1980)

• The problem phase selected from among available decision analytic techniques and adapted them to a particular class of problem, such as capital budgeting or environmental protection (about 1975 to 1990)

• The use-and-user phase addresses all requirements of useful aid in a given context, in which the focus is on usefulness to a particular decider and context.

Actual ADT practice by no means fits neatly into these categories. Their edges and timings are much more blurred, but they may give some insight into ADT’s evolution. In some ways, I view this book as a manifesto for a “use-oriented” decision analysis revolution (rather than technique- or problem-oriented ADT). It draws equally on logic, people and practice skills (rather than on logic alone). I believe this development of emphasis can counter the pressures that have slowed down successful ADT practice in the past. ADT methodology, though useful as it stands, is still a work-in-progress.

1.4.1 ADT methods

At the time ADT was created, it was widely believed, in professional and academic circles, that any decider could improve his/her decision performance by acting on the implications of such a model. ADT modeling began to be taught to deciders-in-training throughout professional education (Brown, Kahr & Peterson, 1974). It was expected that it was only a matter of time before ADT would become standard practice in management and other applied domains. Indeed many of the nation’s major business and government or organizations began incorporating ADT into their decisions (Brown & Ulvila, 1982).

However, ADT, as originally and still generally practiced, has since lost its once-booming interest. Indeed, it is no longer a required course at Harvard Business School, where it originated. It has proven of little practical value, due to three fatal flaws: failure to adapt to human cognitive capacity; apparent cost in time of ADT analyses; and disregard of knowledge not called for by an ADT model. The action that ADT analysis favors is based on a single numerical model and largely ignores other sources of wisdom, such as intuition, others’ advice, and alternative analyses. As a result, ADT analyses do not usually improve on unaided decisions at least not enough to be worth the trouble.

The root cause of these technical ADT problems that impede useful prescription (as opposed to description) is that ADT tools have been developed, applied and taught by mathematically oriented statisticians and the like, who lack both relevant psychology training and familiarity with the real-world contexts. They can check for logical soundness, but not cognitive or organizational fit, or cost-beneficial balance. Academic career incentives do not motivate them to do otherwise.

1.4.2 My own ADT evolution

My first job out of college, in 1958, was in management consulting, which included trying to help clients make decisions in the 1960s. Seeking unsuccessfully some logical discipline to support my advice, I heard of relevant work on ADT led by statisticians Howard Raiffa and Robert Schlaifer (1961) at Harvard. I joined their group and spent five years absorbing their ADT methods, teaching these to MBAs.

When I tried these out on business clients, I was disappointed (as others were) to find that, in spite of ADT being logically compelling real, deciders rarely used it or found it useful. I attributed this to the fact that our version of ADT did not take into account human capacities and limitations.

So I moved to the University of Michigan for the next four years, to learn from psychologists, led by Ward Edwards, who were describing how people make decisions and especially their logical flaws. I found that enhanced knowledge improved my decision-aiding ability somewhat. I incorporated it into a decision-aiding course in collaboration with an interested psychologist and statistician, who complemented my earlier practical decision-airing experience (Brown, et al., 1974). This variant of ADT was an improvement, but it still lacked explicit adaptation of decision tools to human capacities.

In 1973, I returned to my original consulting career. I spent the next two decades aiding professional policy-makers (such as business managers and government officials), developing new decision tools (often funded by research agencies), teaching future managers (such as MBAs) and educating the general public (through the press and broadcasts). I interleaved consulting to practicing deciders and serving on relevant faculties, notably statistical decision theory (University College, London), behavioral decision theory (London School of Economics) and management (Harvard, Michigan, George Mason) and, where appropriate, collaborating with others within them.

By the time I retired from paid employment in the mid-1990s, I had come to believe that the private decider (the “common man”) stood to...

Table of contents

• Cover
• Title Page
• Copyright Page
• Table of Contents
• Editors’ preface
• 1 Introduction
• 2 Decision making
• 3 Term project: Evaluating a policy proposal
• 4 Qualitative decision aids
• 5 Hip surgery case
• 6 Quantitative ADT modeling
• 7 Family case study: C-section vs. “natural” child-birth?
• 8 Using ADT models
• 9 A civic case: Voting for president
• 10 Information value case: Life-saving diagnosis
• 11 Expanded view: Should we teach decision making in school?
• 12 Epilogue — What next?
• Bibliography
• Index