The author presents a number of strategies for making decisions based on desires or values which are incompatible or which conflict with one another in various ways. Cases discussed include conflicts of first and second order desires, conflicts between desires for present and for future ends, problems deriving from anticipated changes of desire, risk-taking problems, and coordination problems. One central claim of the book is that the same dilemma-managing strategies can be applied to all of these. The book also argues that many of the characteristics of moral dilemmas appear in non-moral decision-making. The relations between these strategies and utility-maximizing decision rules are subtle, and are explored throughout the book. To some extent the strategies apply to cases which are too complicated for utility-maximization to apply. Some of them also apply to the early stages of decision-making where utility-maximization does not enter, for example, in selecting a list of options for serious consideration. In some tidy cases, though, the strategies give different recommendations. This book is meant to have both a theoretical and a practical appeal, deriving from our need for ways of making decisions that do not force us to find trade-offs between goods or values which are hard to compare. The strategies presented in the book are meant to be usable in situations which seem to force decision-makers to balance very different quantities, and the discussion of them is meant as a contribution to debates about incomparable values, moral dilemmas and rational decision.
Here is how the eco-mystics of South Devon make decisions. They make a pendulum with a crystal - different people argue strongly for different crystals - and they swing it from a string held between the thumb and first finger of the left hand (to engage the right hemisphere). Then as the pendulum swings they think hard about the actions they are choosing between and associate each with a simple cyclic pattern - circle, ellipse, figure of eight, and so on. Then the crystal begins to follow one of these patterns. That tells which action to perform. Often the crystal chooses an action that the person had not thought very promising or reasonable, but often when the action is performed it turns out to have hidden virtues and unexpected advantages.
Here is how the really tough-minded economists of the world make decisions. They list all the courses of action they are choosing between: for example putting their money in the bank, in a government bond, in a secure investment trust, in a speculative unlisted stock. And they list all the natural and human events that might affect the outcome of the action taken: the possible rates of inflation, the possible behaviour of the stock market, the possible bank rates. Then for each outcome and each possible action they list the probability of that outcome's occurring if the action is performed, and the relative desirability of the resulting situation. For example there may be a 0.2 chance that after I have made a speculative investment there will be a stock market crash, and the desirability of the resulting situation might be measured as minus -100 on a scale in which making a fortune counts as 100 and coming out even counts as 0. Then each action is given a ranking according to the sum of all these probabilities each multiplied by the corresponding desirability. The top-ranked action is chosen.
Each of these methods may seem mad, in very different ways. Decision-by-pendulum seems to allow no room for careful consideration of the consequences of possible actions, and it seems to allow arbitrary whims of hand and brain to govern the direction of one's life. On the other hand it gives a way in which unconscious preferences and unconscious knowledge can affect a decision, and it allows a certain feedback between the course of the decision and the preferences on which it is based. (I mean: if the pendulum begins to trace a figure eight, and your heart sinks because that means you must stay with your spouse, the sinking of your heart will no doubt cause a crystal of any intuitive power at all to jump quickly into another pattern.) Decision-by-calculation, in contrast, seems thoroughly methodical and to take the consequences of actions thoroughly into account, but it presupposes a superhuman knowledge of probabilities, of one's own desires, and of what the consequences of one's actions would really be like.
This is a book about decision making. I describe some decision-making strategies and I argue that they are both usable and give good answers. They may well seem more in the spirit of decision-by-calculation than decision-by-pendulum. Probably they are. But they are meant to connect with intuition as well as calculation, and they are meant to take into account the fact that we usually have no very clear ideas about how likely the world is to go one way rather than another and how we would like it if it did.
The most important claim I make for these decision-making strategies is that they apply in situations where one has to make a decision on the basis of 'incomparable desires'. That is a term I shall use to cover a variety of cases in which desires are not ranked in a simple linear way: one cannot say 'I want A more than I want B'. There are both trivial and heartbreaking examples. You may have to choose between a film and a concert, or between your self-respect and your career, and you don't seem to have a simple preference between them. There are very different cases here, and I try to distinguish them, but in all cases I try to develop 'non-balancing' strategies: strategies which do not force one to find a balance or a tradeoff between very different things.
Non-balancing strategies are useful also in situations in which one does have clear, or at any rate clearer, preferences and some grasp of probabilities. I discuss some issues about risk-taking and about coordination problems in order to show this. Moreover, non-balancing strategies apply to moral dilemmas as well as to non-moral dilemmas. And as a result, or so I claim, when one begins to think in terms of them the distinction between moral and non-moral ('practical') decision making begins to break down. Chapter 5 discusses a very basic and very puzzling question, essentially 'Are there decisions one should avoid making?', just because it arises in both moral and non-moral contexts. And chapter 11 isin fact an attempt to argue that the moral/non-moral distinction is a very unclear one.
(I use the word dilemma to mean what it usually means: a situation where it is very hard to decide what to do. So a moral dilemma is a situation in which it is very hard to decide what one ought morally to do. In recent moral philosophy the word has come to be used rather differently, to mean a situation where one will regret whatever one does. I will not use the word this way, though I discuss a number of moral, and non-moral, problems exhibiting the double regret trap.)
Though this is a book about making decisions it does not bear a simple relation to what is usually called 'decision theory'. Decision theory tries to give precise procedures by which, given a fixed list of options and relatively clear beliefs and desires (or probabilities and utilities), one can choose a 'best' action to perform. The heart of decision theory lies in its analysis of the relation between degrees of desirability of simple and of complex situations. (So it should tell you, for example, how much it is reasonable to want a gamble offering a 0.6 chance of soup or bread and a 0.4 chance of a ride on the roller coaster, compared to how much you want soup, bread, and a ride on the roller coaster.) The emphasis in this book is rather different. I am concerned with strategies for getting complex, vague, incomparable, or otherwise recalcitrant beliefs and desires into a form in which one could apply the procedures of decision theory. I am interested in an earlier stage of the process.
In fact, one of the aims of the book is to defend the possibility of a helpful theory of the 'middle level' of decision making. The bottom level, on this metaphor, consists of fairly precise reasoning about how, given a list of options, one is to choose one to perform. That is the domain of traditional decision theory. And the top level is the search for promising options. Thinking of options is like any other kind of creative thought: it is unpredictable, mysterious, and not governed by any very obvious rules. And while no doubt the central clue to actually making good decisions is to think of the right options, it seems no more likely that there are simple how-to-do-it rules for thinking of the right options than that there are simple rules for making scientific discoveries or finding gripping plots for novels.
But there is a level between these two, between the mechanical and the mysterious. At this middle level of decision making we do such things as: judge which ideas about what to do are live options and which are non-starters, think through what criteria are to be used in a decision, and assess the reliability of the beliefs that will have to be brought to bear on the question. It is this middle level that concerns me, and my claim is that there is a lot to be said about it.
(It is not like doing arithmetic, or verifying a formal proof. Nor is it like finding a proof in the first place. More like solving equations. Like solving differential equations, in fact, in the presence of complicated boundary conditions.)
Some of the topics I discuss, for example coordination problems and decisions involving risk, are traditional concerns of decision theory proper. But, I think, we miss some important points if we do not see that some of the difficulties of decision making in these areas arise at the middle level: from the difficulty of getting our beliefs and preferences into a usable form. And sometimes what decision theory tells us is that there is no ideal solution to such a problem. That still leaves us with the problem of cobbling together a way of getting on with life given that rationality cannot give all that we want. (It is a bit like the problem of putting together a workable political system given that every voting procedure has serious flaws.)
The book is written in short chapters, which are meant to be relatively independent of one another. Moreover, it is divided into three sections, which are also relatively independent of one another. The middle section, 'Theory', is a little heavier going than the other two. Some readers may want to skip it at first reading. Since I wanted the book to be accessible to a wide range of readers I have tried to keep the main text as free as possible from technical terms and free from very local issues from moral philosophy, economics, decision theory, and statistics. For the same reason I have avoided discussing particular people's work in the main text. But this may give the impression that I am claiming more originality for some of the ideas than I should. The notes at the end of the book make connections with more technical questions and with the work of others.1