Crooked Thinking or Straight Talk?: Modernizing Epicurean Scientific Philosophy 9783030395469, 9783030395476, 3030395464

Table of contents :
Contents
1 Epicurus
1.1 Epicurus as a Guide
1.2 Epicureanism?
1.3 Life and Times
1.4 Epicurean Isms
1.4.1 Hedonism
1.4.2 Empiricism
1.4.3 Materialism
1.4.4 Atheism
1.4.5 Naturalism
1.5 How to Live?
1.5.1 Secular Puritanism?
1.6 Some Quotes from Epicurus
2 Rationality as Consistency
2.1 Social Engineering
2.2 Mechanism Design
2.3 Consistency
2.3.1 Revealed Preference
2.3.2 Accounting for Risk
2.3.3 Epicurean Reection
2.4 Revealed Belief
2.4.1 Bayesianism
2.4.2 Abusing the Theory
2.4.3 Revealed Knowledge?
2.5 Utilitarianism
2.5.1 Utilitarianism in Public Policy
2.5.2 Why Add Utilities?
2.5.3 A Foundation for Utilitarianism
2.6 Enforcement
3 Valuing Lives
3.1 Tree of Life
3.1.1 Game Tree
3.1.2 The End Justifies the Means?
3.1.3 Separating out Eve's Preferences
3.1.4 Choosing a Strategy for Living your Life
3.2 Ways of Valuing a Life
3.2.1 States of Mind
3.2.2 Utilitarian Discounting
3.3 Epicurean Valuation of a Life
3.3.1 Global Warming
4 Reciprocity
4.1 How to Get Along Together
4.2 Suppose Everybody Did That?
4.2.1 Categorical Imperative
4.2.2 Prisoners' Dilemma
4.2.3 Fallacies
4.2.4 Hamilton's Rule
4.2.5 Free Rider Problem
4.2.6 Tragedy of the Commons
4.3 Reciprocal Altruism
4.3.1 Reciprocating Strategies
4.3.2 The Reciprocity Principle
4.4 Social Contracts
4.4.1 Conventions
4.4.2 Constitutions
4.5 Global Warming?
5 Fairness
5.1 Evolutionary Ethics
5.2 Golden Rule
5.2.1 The Original Position
5.2.2 Deep Structure of Fairness
5.3 Utilitarianism or Egalitarianism?
5.4 Achieving a Social Consensus
5.4.1 Modeling Social Evolution
5.5 Social Contracts
5.5.1 Traditional Social Contracts
5.5.2 Widening the Scope of Fairness Norms
5.5.3 Applying Fairness Norms?
5.5.4 Universal Health Care
Index

Citation preview

KEN BINMORE

Crooked Thinking or Straight Talk? Modernizing Epicurean Scientific Philosophy

Crooked Thinking or Straight Talk?

Ken Binmore

Crooked Thinking or Straight Talk? Modernizing Epicurean Scientific Philosophy

123

Ken Binmore Department of Economics University of Bristol Bristol, UK

ISBN 978-3-030-39546-9 ISBN 978-3-030-39547-6 https://doi.org/10.1007/978-3-030-39547-6

(eBook)

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

To Danny and Christina In memory of Albi, where much of this book was written.

Preface This book is written for oddballs like myself, who like to do their own thinking for themselves without following the herd wherever it may go. Its theme is that modern philosophical ideas with a science base can help us structure the way we think about ourselves and our society in a practical way that is a million miles from the pretentious drivel offered by academic philosophers of the old school. The book isn’t some superficial self-help manual. It will be hard going in places for nearly everybody, although it contains no equations and very little jargon. It is the kind of book better not read at one gulp. After coming upon a new idea, it will pay to sit back and ask yourself whether what is said really makes sense—especially when it denies some entrenched orthodoxy. The kind of reader I am writing for will anticipate the prospect of such intellectual challenges with pleasure. The book has four deliberately interwoven strands, some more serious than others. The first strand is a series of diatribes against the complacency of traditional philosophy, making fun of the silliness of famous arguments once they have been stripped of complicated jargon. What a relief to get that off my chest! This line is intended to lighten up the second strand, which is a serious attempt to explain ideas from the theory of games and decisions that have the potential to improve both our personal lives—how we think about ourselves—and our societies; not in some great utopian leap, but in the kind of ways that our lives were improved by the telephone or the bicycle. The third strand is a collection of chatty and irreverent comments expressing my own views on the subject in hand—especially when it differs markedly from the opinion of the herd, which most people do not think to question. The fourth and last strand is a series of sometimes long footnotes placed at the bottom of the relevant page, and not inconveniently at the end of the book. Nobody need read a footnote, but good footnotes can sometimes be a lot of fun—like those of Edward Gibbon in his Decline and Fall of the Roman Empire. Otherwise my style is an attempt to emulate the incomparable David Hume in his Treatise on Human Nature. I quote from him a good deal too. It is hard to say anything better than Hume. A pity we were born into these barbaric times, rather than as gentlefolk in the eighteenth century! I want the book to be as little like an academic monograph as it is possible to be. I have therefore abandoned the practice of documenting quotations and referencing sources, which seems to me tiresomely pedantic now that we can google anything of interest, and get not only what we are looking for, but a variety of related stuff as well. So references are confined to occasional mentions of books in the footnotes.

vii

viii

Preface

As for myself, I began my professional life as a pure mathematician, writing research papers and textbooks in a conventional way. I was then fortunate enough to get into game theory when it was just taking off. It then made sense for me to switch to economics, where I not only pursued a standard academic career in both England and the USA, but also operated successfully as a business consultant, using my knowledge of game theory to give advice on economic regulation and auction design, notably the UK telecom auction of the year 2000 that made $35 billion. So much money attracted a lot of attention. Newsweek magazine described me as the ruthless, poker-playing economist who destroyed the telecom industry. It is true that I once enjoyed playing Poker for more than I could afford to lose, but I am not at all ruthless, and nor is the telecom industry destroyed. I no longer feel the need to update my vita, but the last revision is available at www.homepages.ucl.ac.uk/∼uctpa97/ I retired early to follow up my philosophical interests, but found quietude an elusive target, until I came across the work of the ancient Greek philosopher Epicurus. I hope my use of him as a putative guide to the book may help others who have similar problems in breaking free from the herd mentality that demands so much and offers so little in return. Ken Binmore Monmouth, Wales August, 2019

Contents 1 Epicurus 1.1 Epicurus as a Guide . . . . . 1.2 Epicureanism? . . . . . . . . 1.3 Life and Times . . . . . . . . 1.4 Epicurean Isms . . . . . . . . 1.4.1 Hedonism . . . . . . 1.4.2 Empiricism . . . . . . 1.4.3 Materialism . . . . . 1.4.4 Atheism . . . . . . . 1.4.5 Naturalism . . . . . . 1.5 How to Live? . . . . . . . . 1.5.1 Secular Puritanism? . 1.6 Some Quotes from Epicurus .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1 1 1 3 6 6 8 12 17 18 21 23 25

2 Rationality as Consistency 2.1 Social Engineering . . . . . . . . . . . 2.2 Mechanism Design . . . . . . . . . . . 2.3 Consistency . . . . . . . . . . . . . . 2.3.1 Revealed Preference . . . . . . 2.3.2 Accounting for Risk . . . . . . 2.3.3 Epicurean Reflection . . . . . 2.4 Revealed Belief . . . . . . . . . . . . 2.4.1 Bayesianism . . . . . . . . . . 2.4.2 Abusing the Theory . . . . . . 2.4.3 Revealed Knowledge? . . . . . 2.5 Utilitarianism . . . . . . . . . . . . . 2.5.1 Utilitarianism in Public Policy . 2.5.2 Why Add Utilities? . . . . . . 2.5.3 A Foundation for Utilitarianism 2.6 Enforcement . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

27 27 29 33 34 35 37 39 40 41 43 43 44 45 48 50

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

ix

x

CONTENTS

3 Valuing Lives 3.1 Tree of Life . . . . . . . . . . . . . . . . . . . 3.1.1 Game Tree . . . . . . . . . . . . . . . . 3.1.2 The End Justifies the Means? . . . . . . 3.1.3 Separating out Eve’s Preferences . . . . 3.1.4 Choosing a Strategy for Living your Life 3.2 Ways of Valuing a Life . . . . . . . . . . . . . 3.2.1 States of Mind . . . . . . . . . . . . . . 3.2.2 Utilitarian Discounting . . . . . . . . . 3.3 Epicurean Valuation of a Life . . . . . . . . . . 3.3.1 Global Warming . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

55 55 56 57 57 60 61 61 63 66 69

4 Reciprocity 4.1 How to Get Along Together . . . 4.2 Suppose Everybody Did That? . 4.2.1 Categorical Imperative . 4.2.2 Prisoners’ Dilemma . . . 4.2.3 Fallacies . . . . . . . . . 4.2.4 Hamilton’s Rule . . . . . 4.2.5 Free Rider Problem . . . 4.2.6 Tragedy of the Commons 4.3 Reciprocal Altruism . . . . . . . 4.3.1 Reciprocating Strategies 4.3.2 The Reciprocity Principle 4.4 Social Contracts . . . . . . . . . 4.4.1 Conventions . . . . . . . 4.4.2 Constitutions . . . . . . 4.5 Global Warming? . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

71 71 73 73 75 77 80 82 83 84 85 86 89 89 90 93

5 Fairness 5.1 Evolutionary Ethics . . . . . . . . . . . . . . 5.2 Golden Rule . . . . . . . . . . . . . . . . . . 5.2.1 The Original Position . . . . . . . . . 5.2.2 Deep Structure of Fairness . . . . . . 5.3 Utilitarianism or Egalitarianism? . . . . . . . 5.4 Achieving a Social Consensus . . . . . . . . . 5.4.1 Modeling Social Evolution . . . . . . 5.5 Social Contracts . . . . . . . . . . . . . . . . 5.5.1 Traditional Social Contracts . . . . . 5.5.2 Widening the Scope of Fairness Norms 5.5.3 Applying Fairness Norms? . . . . . . 5.5.4 Universal Health Care . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

95 95 97 99 100 104 108 111 113 114 117 121 124

Index

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

127

Chapter 1

Epicurus 1.1

Epicurus as a Guide

What a mess we have made of things! Not just of our personal lives, but of our societies, and our planet. We look to philosophy for guidance, but the schoolmen of today—the writers of footnotes to Plato—have nothing to offer. They idle away their time building philosophical castles in the air when in a good mood, or defining each other out of existence when they aren’t. Let us leave them to their empty posturing and crooked reasoning. The action is elsewhere on a rival channel, where an alternative but equally ancient branch of philosophy based on scientific principles offers genuine guidance, especially in the social world, where we are most in need of rational principles to light the way ahead. On this scientific channel, crooked reasoning is out and straight talking is in. This first chapter introduces the Greek philosopher Epicurus, who will be our guide on how to think straight. It surveys his life and times before examining his philosophical views, which turn out to be surprisingly up to date. The issues raised are then pursued in later chapters that show how modern techniques from game theory and elsewhere can put enough flesh on his philosophical bones to provide practical ways of structuring how we think about the many problems we face, both collectively and individually. Why don’t we take these social tools from their Epicurean box and see what we can do with them to improve our lives? In doing so, there is no need to weigh ourselves down with equations or heavy jargon. We can even laugh at ourselves a little along the way.

1.2

Epicureanism?

Which philosopher argued that the way to happiness is to give ourselves over to gluttony and lust? When asked this question, most people with an opinion go for Epicurus (341–270 BC). In congratulating the Church in having eliminated the last embers of his thought, St Augustine accordingly called Epicurus a pig. Modern

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6_1

1

2

CHAPTER 1. EPICURUS

dictionaries similarly define an epicurean as someone who luxuriates in the pleasures of the table and the flesh. This misrepresentation of Epicurus is perhaps the most successful example of the Big Lie ever. In fact, Epicurus is on record as being satisfied with no more than barley bread and water, with perhaps a little cheese as a special treat. As for sex, he thought it best avoided, although it probably does no harm. How come Epicurus got so badly misrepresented? The answer is simple. His philosophy was the big philosophical success story in the ancient world. Epicurus himself became something of a folk hero. So he was the major competition for the early Church. His chief offence in the eyes of the Church was the same Epicurean doctrine that made him popular with pagans—that there is no life after death. This doesn’t seem so splendid to us, but the ancient Greeks believed in a miserable after-life, in which their souls wandered forever as aimless ghosts through a dreary underworld. The Church therefore not only followed those Stoics who branded him a pig, but denounced him as a heretic. In Dante’s Inferno, he is the worst heretic of all, doomed to roast in his tomb forever in the sixth circle of Hell. His many books were so successfully suppressed that almost nothing he wrote himself survives. Rescue. His reputation was eventually rescued by the enlightenment philosopher, Pierre Gassendi. As a result, historians of philosophy now distinguish between the dictionary definition of epicureanism and the actual philosophy of Epicurus by calling the latter Epicureanism with a capital E. Philosophers of science similarly recognize Epicurus—along with Aristotle when in the mood—as a pioneer of the kind of scientific philosophy pursued by such luminaries as Thomas Hobbes, John Locke, Benedict de Spinoza, David Hume, Adam Smith, Jeremy Bentham, Bertrand Russell, and Karl Popper—of whom my own favorite is the incomparable David Hume. But such recognition cuts no ice with the majority of modern philosophers, who continue to regard themselves as writing footnotes to Plato. Although Epicurus didn’t write in explicit opposition to Jesus Christ (who wasn’t born until hundreds of years after his death), he did write in explicit opposition to Plato (who died only a few years before Epicurus was born). This makes Epicurus a secular heretic in the eyes of modern traditionalists. They are more civilized than their ancient ancestors in that they don’t invent fake news to tarnish his image, but they do him no favors when mentioning him in passing. In spite of this studied neglect, there has been a major resurgence of interest in Epicurus in recent years,1 to which I hope this book will be regarded as a no-holdsbarred contribution. The main aim of the current chapter is to list the down-toearth doctrines of Epicurus. The rest of the book develops his ideas, with numerous asides defending them against Plato and his metaphysical followers, whose airy-fairy 1 For example, Hans Dimetriadus: Epicurus and the Pleasant Life, George Strodach: The Art of Happiness Hermann Usener: Epicurea, James Warren: Facing Death: Epicurus and his Critics, Catherine Wilson: How to be an Epicurean. For historical perspectives, James Gaskin: The Epicurean Philosophers, A. A. Long and D. N. Sedley: The Hellenistic Philosophers.

1.3. LIFE AND TIMES

3

inventions are sometimes so ridiculous that one has to laugh out loud. As Spinoza put it when writing about their thoughts on politics and morality: Instead of ethics, they have generally written satire . . . such as might be taken for a chimera, or might have been formed in Utopia, or in that golden age of the poets . . .

Of course, Epicurus isn’t entirely free of absurdities himself—any more than Spinoza or anyone else. It is silly, for example, to suppose that our souls (whatever they may be) are made of atoms, but whoever got everything right?

1.3

Life and Times

There is surely something to the idea that creativity flourishes best in open societies, especially those that succeed in retaining their cohesion in the face of invasion or civil war. The eruption of all kinds of philosophical thought among the many city-states of ancient Greece certainly fits this story. Athens fits particularly well. History. Athens survived being burnt to the ground during a traumatic invasion of Greece by the Persian Empire, which was eventually defeated by a coalition of Greek cities led by Athens and Sparta. Athens seized the opportunity to grab a small empire of its own made up of Greek cities and islands previously held by Persia, but this led to the Peloponnesian War between Athens and Sparta (in which Socrates fought with courage and distinction for Athens). After Athens was defeated, Sparta replaced its democracy by the “tyranny of the thirty tyrants” (among whom followers of Plato were prominent). The democracy was restored, but didn’t survive the invasions of Philip of Macedon and his son Alexander the Great (famously tutored by Aristotle). Alexander went on to overthrow the Persian Empire altogether. His reign didn’t last long, but the Hellenic period that followed left Greek culture deeply embedded across the whole Near East and Mediterranean. Bunkum? Henry Ford was good at making cars, but he couldn’t have got things more wrong when he said that history is bunk. We should pay attention to the fact that the kind of direct democracy practised in ancient Athens—so much admired by modern politicians who neither know nor care what history has to teach—was a disaster in the Peloponnesian War. Populist demagogues of the day persuaded the assembly of all citizens to vote for too many ill-thought-out ventures that proved catastrophic. Nor was the assembly at all kind to the cities they liberated from Persia. It selfishly voted to squeeze them dry to finance its own architectural glory. Our own representative form of democracy is by no means perfect but does much better in keeping us out of trouble. So why are we currently moving in the direction of direct democracy with referenda and the like? In a referendum, each individual vote has a negligible impact, so there is no incentive for most voters to think hard about the issues (Section 4.2.3). So they let whatever enthusiasm is currently popular guide their vote, without thought or reflection—as though voting in a TV talent show. Even appointing a smallish body of representatives at random

4

CHAPTER 1. EPICURUS

to make decisions that are now settled by referendum would be better. At least the representatives would have an incentive to think the issues through. Even when we continue to let representatives decide for us, we do our best to reduce them to delegates—as in the electoral college that the founding fathers of the American republic set up to prevent the selection of the president becoming a referendum. We have nobody to blame but ourselves if populists get into power this way. Those who don’t learn from history are compelled to repeat it. Life. Epicurus was born in 341 BC to humble Athenian parents who were settled on the island of Samos when it was part of the brief Athenian empire. He was tutored by a Platonist but he reacted negatively to the teaching while still a teenage boy. To retain his Athenian citizenship, he served a military apprenticeship for two years in Athens from the age of eighteen, where he doubtless took what opportunity he could to listen to the leading philosophers of the day. After the death of Alexander in 323 BC, the Athenian colonists were forcibly removed from Samos to what is now the Turkish mainland, where Epicurus rejoined his family, to which he was always close. There he studied with a follower of the philosopher Democritus, who famously guessed right about matter being made of atoms. We don’t know how he eventually became sufficiently well-established to set up a philosophical school in the city of Mytilene on the island of Lesbos, but we do know that he and his followers were expelled from the city in 311 BC after their teaching led to civil strife. Epicurus relocated first to Lampsacus and then to Athens in 306 BC, where he established what became known as the Garden, in open opposition to Plato’s Academy. He is thought never to have married. He died what must have been a painful death from kidney stones at the age of 72 in 270 BC. Philosophical Schools. Plato’s Academy lay at walking distance to the northwest of the walled city of Athens. The Lyceum of the polymath Aristotle lay at a similar distance to the northeast. Zeno’s Stoa—where Stoicism began—was in the Agora, within the city itself.2 The Agora also hosted Diogenes, the philosopher who brightened things up by living in a barrel and poking fun at his fellow philosophers. On his return to Athens, Epicurus bought some land, outside the city on the road between the Agora and the Academy. The group of friends that met there supposedly tended the garden with their own fair hands, which is why their philosophical community became known as the Garden. It was actually the third Athenian philosophical school that we remember from that time. The Stoa was the fourth, having been established some years later. One of the surviving quotes of Epicurus says that the most important ingredient for a happy life is a circle of close friends. The Garden seems to have been an attempt to realize this ambition. However, all can’t have been entirely sweetness 2 The founder of Stoicism is Zeno of Citium, not the earlier Zeno of Elea, famous for his paradox of Achilles and the Tortoise. The Agora wasn’t only a market place, but the center of city life. The Stoa was a portico in which the Stoic philosophers would walk up and down exchanging views. Walking up and down was also popular at the Lyceum, for which reason the followers of Aristotle were called Peripatetics. Dante classifies Plato and Aristotle as noble pagans and places them untortured in the first of the nine circles of Hell. Zeno doesn’t get a mention.

1.3. LIFE AND TIMES

5

and light, since some of the slanders directed against Epicurus derived originally from erstwhile followers rejected from the community. Very unusually for ancient Greece, the Garden is said to have been open to both women and slaves on equal terms with free-born men. How true such claims may be is difficult to evaluate, although women were certainly part of the community, and the same is true of at least one freed household slave of his own.3 The obloquy that would have been directed against the Garden for this reason would have been mitigated by the community’s refusal to participate in the social and political life of the city—which Aristotle regarded as essential to the well-lived life. My own guess is that this feature of the Garden’s social contract was less a matter of principle than a reaction to whatever got Epicurus into serious trouble in Mytilene. Saints? This last observation accepts that Epicurus wasn’t a saint—and nor was the real Socrates (as opposed to the fictional Socrates invented by Plato to add authority to his later dialogues). They were just human beings, probably more vain than they thought themselves to be, but with much to be vain about. It would have been a lot of fun to meet them—especially Socrates.4 A serious argument with either would have been possible with no risk of anybody getting upset. What of the other Athenian giants? My guess is that Zeno was worthy but dull. Diogenes would have been good for a laugh, but too much of a show-off to be bearable for long. Plato’s sense of entitlement would have made him seriously uncomfortable as a companion. Aristotle is more of a mystery: affable but distant? Whatever the truth, we don’t need to treat them or anyone else with exaggerated respect. Epicurus wouldn’t have wanted it any other way. Surviving works. Epicurus wrote at least 300 books, only fragments of which survive. Much of his work would have been lost even without the hostility of the Church. We would have little even of the work of Aristotle—who was a favorite of the medieval Church—if a hidden cache of his writings hadn’t been discovered by chance in ancient times. Ironically, a small cache of Epicurean fragments similarly turned up in the Vatican Library. 3 Respectable Athenian women of the time seldom left their married quarters. So a female member of the Garden would have been suspected of being kept for immoral purposes. At a time when even Aristotle is on record as believing that barbarians were natural slaves, treating a slave or freedman as an equal was socially unacceptable. Even if women and slaves were not fully equal, Epicurus was therefore making himself a target by including them in his community at all. 4 Robert Burton’s magnificently eccentric Anatomy of Melancholy of 1651 has this to say of Socrates: “Theodoret in his tract manifestly evinces as much of Socrates, whom though that Oracle of Apollo confirmed him to be the wisest man then living, and saved him from the plague, whom 2,000 years have admired, of whom some will speak evil as soon as Christ, yet in reality he was an illiterate idiot, as Aristophanes calls him, a mocker and ambitious, as his Master Aristotle terms him, an Attic buffoon, as Zeno, an enemy of all arts & sciences, as Athenæus, to Philosophers and Travellers, an opinionative ass, a caviller, a kind of Pedant; for his manners, as Theod. Cyrensis describes him, a Sodomite, an Atheist (so convict by Anytus) hot tempered, and a drunkard, and prater &c., a pot companion, by Plato’s own confession, a sturdy drinker, and that of all others he was most sottish, a very mad-man in his actions and opinions.” His supposedly shrewish wife Xanthippe wasn’t pleased with him either.

6

CHAPTER 1. EPICURUS

Apart from surviving fragments of his own works, we have commentaries from other philosophers, although some of these are written by enemies and can’t be trusted. The Roman philosophical historian Diogenes Laertius was no enemy but nor was he very reliable. It is from him that we know the names of Epicurus’s books. His philosophical ideas are curiously preserved in a Latin poem by Lucretius called On the Nature of Things. Here is a prose translation of one of the poem’s less florid passages:5 When we have seen that nothing can be created out of nothing, then more rightly after that shall we discern that for which we search, both whence each thing can be created, and in what way all things come to be without the aid of gods.

In the second century AD, Zeno of Oenoanda had Epicurean principles carved into a wall in his home city in modern Turkey, some of which are still there. Finally, there is the library of the Villa of the Papyri in Herculaneum, buried in ash by the eruption of Vesuvius in 79 AD. Modern methods now sometimes allow the charred scrolls to be deciphered. Useful works by the Epicurean philosopher Philodemus (110-40 BC?) have been found, and perhaps original works by Epicurus himself might yet be unearthed. However, taken altogether, this is very little for a philosophy that was popular in ancient times for at least six hundred years.

1.4

Epicurean Isms

A glance down the following list of Isms espoused by Epicurus explains why he would still be in trouble with many philosophers if he reappeared today. In this chapter, only clarifications of the views he actually held are offered. Later chapters expand on these thoughts, taking the opportunity to discuss models or ways of thinking invented in modern times that develop the scientific approach he pioneered.

1.4.1

Hedonism

Everybody agrees that hedonists like Epicurus are in favor of seeking pleasure and avoiding pain. Some variant meanings offered by modern dictionaries are: selfindulgence, debauchery, rapacity—and, of course, epicureanism. There was a school of philosophers from Cyrene in North Africa led by Aristippus (a pupil of Socrates!) to which such epithets might apply,6 but they don’t even remotely describe the ascetic hedonism of Epicurus. How would Epicurus have defended his hedonism if someone I shall call Alice offered a challenge? I think he would have asked her for her current goals. For each goal, he would have then asked: Why do you want that? For each subsidiary goal offered in explanation, he would then have asked again: Why do you want that? 5 The full 1910 translation of Cyril Bailey can be obtained by googling his name. He ranks Lucretius alongside Virgil and Catullus as a Latin poet. Epicurus himself seems not to have cared much for poetry. 6 However, it was the city of Sybaris in South Italy that the Greeks thought of as embodying the ultimate in debauchery—particularly the famous Sybarite delightfully named Smindurides. Sybaritic is therefore yet another misleading synonym offered for hedonistic.

1.4. EPICUREAN ISMS

7

And he would keep on asking why until Alice was reduced to saying: Because it feels good. Or if talking about things to be avoided: Because it feels bad. Epicurus would then say that if seeking to feel good or avoiding feeling bad is your ultimate motivation, why not cut to the chase—admit that what you really want is to seek pleasure and avoid pain. Go straight for what you really want and perhaps you will get there eventually. Utilitarian? Philosophers sometimes identify hedonism with the utilitarianism of Jeremy Bentham and John Stuart Mill,7 so extending the concern of those they want to call hedonists from individual people to society as a whole. It is true that Mill described Epicurus as a pioneer utilitarian, but he meant no more by this than that Epicurus modeled people as individuals who seek pleasure and avoid pain. The utilitarian model of Bentham and Mill is naive in that one needs to envisage a dial in a person’s head that swings one way to measure increased pleasure or decreased pain, and the other to measure the reverse. It is true that rats will pass up food and sex altogether in favor of pressing a lever that delivers an electrical stimulus to a pleasure center in their brains—and who is to say that the same might not be true of you or me? However, modern neuroscience suggests that although the neural mechanisms for pain and pleasure overlap, one isn’t simply the negative of the other. Moreover, economic theorists reworking the foundations of their subject have redefined the notion of utility to make it independent of any psychological assumptions, thereby completely reversing its causal implications (Chapter 2). The confusion that results could be avoided by replacing Bentham’s utility by the term felicity—as he suggested would be better in his later work—but it is too late now for such a reform.8 What is natural? The hedonism of Epicurus was neither the self-indulgence of modern dictionaries, nor the utilitarianism of Bentham and Mill. It was no more than the claim that people act to increase pleasure and decrease pain. A modern defender of this proposition would probably say that people thereby merely accept that this is the way evolution made us, arguing that neuroscientific discoveries about dopamine, serotonin and the like have to be taken seriously. 7 Jeremy Bentham was a full-blown English eccentric of the nineteenth century. His will specified that his corpse be stuffed and mounted, so that he could always chair meetings of the governors at my old college in London. Nowadays, his corpse is kept on display in a glass cabinet. I sat next to it once as a guest at a dinner of the Jeremy Bentham Society. In several lively but undisciplined works, Bentham created utilitarianism more or less single-handed (although the Swiss philosopher Helvetius deserves mention too). John Stuart Mill is credited with refining Bentham’s theory into something intellectually respectable, but his efforts seem to me more directed at making Bentham’s ideas socially respectable than providing firm foundations for the theory. And is he dull! Although admittedly not so dull as when writing on economics. However, he was on the side of the angels as an early campaigner for women’s rights—and he deserves eternal credit for saying that people should be free to do what they want provided they don’t thereby infringe the freedom of others. He wrote a sour obituary of Bentham, comparing his work with the idle philosophical musings of the poet Coleridge. I guess Mill classified Bentham alongside his own eccentric father, who stole Mill’s childhood by turning him into a hothouse scholar from the age of three. 8 A forthcoming book The Pursuit of Happiness by Louis Narens and Brian Skyrms is an excellent survey of the long history of attempts to make sense of these issues.

8

CHAPTER 1. EPICURUS

But such arguments were not available in ancient times, and so the Stoics— who otherwise had a lot in common with Epicureans—were able to argue that virtue was the genuine natural objective of the human species. Nowadays, we recognize that the Stoic virtues are characteristically Roman, and that other societies in other places and times have prioritized different virtues. As David Hume comments, the Natural Laws invented by philosophers down the ages are actually artificial, and what is natural isn’t the norms themselves, but the fact that different societies should develop such norms. The unexamined life? In considering his style of hedonism, perhaps the most important thing to register is that Epicurus shared the view expressed by Socrates at his trial that the unexamined life isn’t worth living. He argued in particular that philosophical reflection can help us shape how our brains perceive pleasure or pain. So he would have sided with Mill against Bentham’s view that poetry is no better than the tavern game of push-pin as a source of pleasure. Epicurus argued—along with Aristotle and many other philosophers before and after—that one can learn, through adopting suitable habits or modes of thought, that practising certain virtues and shunning certain vices is a better way of seeking pleasure or avoiding pain than just letting experience wash over you. But what behavioral routines or habits of thinking are likely to work best? In his old age at least, Epicurus thought that avoiding pain—including painful thoughts and regrets—was much more important than seeking pleasure: that the tranquility to be so derived is a prize beyond measure. It doesn’t follow that he would have condemned our enjoying a good night out, or appreciating occasionally being the center of attention. He seems to have been a naturally kind person, not at all inclined to be judgemental. But he urges us to take proper account of the potentially heavy cost in the future of indulging ourselves overmuch in the present, completely unlike Aristippus, who explicitly advocated seizing all opportunities for pleasure whatever the future consequences might be! Epicurus himself reportedly took great pleasure in hanging out in the Garden talking philosophy with his circle of close friends—but surely not all the time. I like to think that they sometimes also told jokes, or made innocent fun of what the Platonists might be up to further down the road at the Academy. One might respond that achieving tranquility is going to be more difficult for most of us than for someone like Epicurus, who was apparently uninterested in sex and whose self-esteem was stoked by his success as a public intellectual. I guess he would have said that those of us with more to regret just have to try harder.

1.4.2

Empiricism

I learned yesterday of a group of artificial intelligence gurus who call themselves rationalists, because they believe in trying to resolve difficult questions in a rational way. However, when philosophers distinguish between philosophical empiricists and philosophical rationalists, they are far from suggesting that empiricists are irrational in the standard dictionary sense. They are talking about what the philosophers in question believe about where knowledge comes from.

1.4. EPICUREAN ISMS

9

Hume’s Fork. A full-blown empiricist believes that we get knowledge only through our senses: that we can’t really know anything about the world without examining whatever evidence we can uncover. As David Hume put it: If we take in our hand any volume; of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames: for it can contain nothing but sophistry and illusion.

For obvious reasons, this piece of wisdom is called Hume’s Fork. Immanuel Kant— who was reportedly awoken from his philosophical slumbers after learning of Hume’s work—thought otherwise. He held that a synthetic a priori is a factual proposition that we can know without needing any evidence. The synthetic here means that the proposition doesn’t satisfy the first prong of Hume’s fork. The a priori means that it doesn’t satisfy the second prong. Kant’s leading example of a synthetic a priori is that space is Euclidean—which we now know to be only approximately valid right here on Earth; and hopelessly wrong near a black hole. Models. I share the view on these matters generally held by scientists who care about such questions. We are certainly empiricists, but Hume goes too far for us. We certainly don’t want to commit Euclidean geometry to the flames! The same goes for Newtonian mechanics, even though it has been superseded by Einstein’s theory of relativity. Nor will we want to discard relativity if ever a theory is found that reconciles its approach with quantum physics. We don’t even want to consign theories to the flames that are as yet entirely speculative (like string theory so far), or even outright fictions (like the analysis of planetary orbits in the presence of an inverse-cube law that I was taught as an undergraduate). These are all examples of what scientists call models. Each comes with a (possibly empty) domain of application within which they work to a greater or lesser degree of approximation. There are philosophers of science who say that structural features of some models are so well-supported by the evidence that they must really reflect the way the universe is actually put together. But we don’t care very much what is called “real”. It is enough for us that we have models that work. We even follow Epicurus in tolerating two or more contradictory models at the same time, provided that both fit the data reasonably well: In declaring true every theory that does not contravene the evidence of the senses, Epicurus does not blink at the fact that the philosopher may arrive at more than one explanation for a given phenomenon . . . even at explanations that are mutually exclusive or contradictory—Lucretius

Indeed, when discussing “free will” and the like, one has little choice but to do so. 9 9 Economics offers a more mundane example. How do markets clear? How come a price is reached at which supply equals demand? Textbooks offer a simple model in which an auctioneer chooses the clearing price. But such an auctioneer is usually no more “real” than free will. A more realistic model has the market converging on a clearing price as buyers and sellers trade offers back and forward, but the textbook model works well enough much of the time. A similar split between different models explains the success of game theory (Section 2.2).

10

CHAPTER 1. EPICURUS

Asking for more is asking for too much—we can’t “know” things in the absurdly strong sense on which philosophers commonly insist. We can’t even trust our own perceptions, which are largely constructs of our brains operating on programs that we may never fully understand. Our memories aren’t just unreliable; recent research shows that we edit them each time we recall them. Even the data in scientific papers is mistaken or dishonestly presented much more often than is generally realized. We certainly often behave as though we know things by proceeding as if it is impossible that they could be wrong, which one might call knowledge-as-commitment. We already have the word faith, but its religious connotation makes it unusable, since religious folk insist that their faith generates certainty. But to behave as though one knows something isn’t the same as knowing it in the traditional sense of knowledge-as-certainty. Research on babies seems to show that we are genetically programmed to accord some propositions the status of knowledge-as-commitment. It may be that we are similarly programmed with the principle of scientific induction. David Hume is certainly very convincing in arguing that it isn’t entitled to the status of knowledgeas-certainty—the best one can do is to observe that it confirms itself. But the fact that evolution may have programmed us to treat some propositions as “synthetic a prioris” doesn’t make them true; it only shows that treating them as though they were true had survival value in ancestral times. Metaphysics? Such skepticism isn’t popular with traditionalists. They think that solid foundations for their metaphysical fables can be found by playing with words, but usually fail to notice when the words they use already take for granted the propositions they are trying to establish. Descartes’ “I think, therefore I am” is the most famous example.10 He is perhaps entitled to “Here is a thought, therefore thinking exists”, but his initial “I” already takes for granted the existence of a concept of the self that many neuroscientists nowadays regard as an illusion. Sometimes metaphysical systems are offered with no justification at all. As Bertrand Russell said of the celebrated Monadology of Leibniz, it is “a kind of fantastic fairy tale, coherent perhaps, but wholly arbitrary”. To use Plato’s admirable fable of the cave against its author, all we see are shadows on the wall. Metaphysicians think they can tell us what casts the shadows, but all we can genuinely say is that it must be something capable of casting the shadows that we see. Plato’s own fantastic fairy tale is that the shadows are cast by perfect prototypes of our earthly notions that somehow inhabit an abstract space of ideal forms transcending time and space. Diogenes aptly commented that he had seen Plato’s cups and tables, but neither his cupness nor his tableness. We don’t need to share Diogenes’ disgust at society in general, but we can surely sympathize with his view that Plato’s metaphysics are only good for a laugh. 10 I found the only metaphysical joke I know many years ago in the Beachcomber column of the Daily Express, which has nowadays been degraded into one of several British newspapers given over to telling populists what they want to hear. Among other whimsical inventions, it featured the non-existent philosopher Dr Strabismus of Utrecht, who went one better than Descartes by improving his cogito ergo sum to non cogito ergo non sum.

1.4. EPICUREAN ISMS

11

Reflective equilibrium? It seems likely that Plato’s fantastic fairy tale was inspired by the geometrical discoveries of earlier Greeks, of which Plato and his contemporaries were understandably very proud. Their underlying mathematical model—which idealizes such real-life geometrical shapes as triangles and circles— was later codified by Euclid. But Plato failed to see that the beautiful Euclidean universe is just a model. He thought the theorems of Pythagoras and others provided an insight into the nature of ultimate reality instead of simply being useful approximations.11 Euclidean geometry starts with formal definitions of ideal points and lines, and therefore metaphysicians typically think that further insights into the nature of ultimate reality must necessarily start with similar formal definitions, and then go on from there. The cult of “define your terms first” is typified by Plato’s tales of Socrates buttonholing folk in the street and demanding definitions of abstract notions of this and that. People then made inadequate replies without ever asking in what model the definition is to be embedded. What if the inventors of modern physics had proceeded in this way? First make up definitions of energy or momentum and the like, and then fit these a priori definitions together as best you can to make up some kind of ramshackle theory. This isn’t how scientific models are created! The philosopher John Rawls invented the term reflective equilibrium for the scientific approach. One tries out various possible models along with their associated definitions, to see how well they work. Only when one can’t see how to do better does one stop. Of course, metaphysicians really go through a process of reflective equilibrium too, although they are unlikely to admit it—even to themselves. However, their criterion of success isn’t how well their model predicts data, but how well it fits their prejudices. This leads to some bizarre definitions in which words are redefined to mean something quite unlike their usage in ordinary language. We have seen that rational is defined as a form of irrationality in which people behave as though they know things that can’t really be known for certain. We shall shortly note that moral philosophers similarly define objective to mean “absolute” rather than “following from generally agreed facts”. Perhaps the most famous example is Socrates’ definition of justice in Plato’s Republic as “minding your own business”. The list of such perverse definitions is very long. As the philosophical Bishop Butler cried out in frustration, “Everything is what it is, and not some other thing!” Authority. It is a fact that Epicurus followed Democritus, Aristotle and others in championing empiricism, but I don’t suppose it meant a great deal more for him than a reason to deny the authority claimed by the priests and metaphysicians of his time. More than two thousand years later, we are still plagued by their intellectual descendants, who threaten not only our freedom to think our own thoughts, but 11 To be fair to Plato, he is by no means alone in being seduced by the beauty of some theories into thinking they must be true. Dirac is the most famous physicist who expressed such a view. G¨ odel is the most famous mathematician. The poet Keats famously tells us: “Beauty is truth, truth beauty—that is all ye know on earth, and all ye need to know.” This may be a beautiful thing to say, but it isn’t true.

12

CHAPTER 1. EPICURUS

sometimes even our personal safety—as with Epicurus in Mytilene, Socrates in Athens, Galileo in Rome, and many more. Empiricists like Epicurus argue for doing your own thinking for yourself with as little reliance on authority as possible. This doesn’t mean that you have to follow Charles Darwin by reinventing evolution and the like for yourself. But when taking things on authority, you need to have good reason for thinking that your authority figure has done more than mindlessly accept the authority of someone else. Experiment, of course, is the ultimate authority. The Greeks somehow never made this last step.12 I have read that they would have regarded doing experiments as something suitable only for the lower classes.

1.4.3

Materialism

Epicurus taught that everything is made of atoms, a doctrine he inherited, along with much else, from the philosopher Democritus (460-370 BC?), who was a contemporary of Plato.13 On hedonism, Democritus says that a poor man makes himself rich by desiring little. On empiricism, he says that we know nothing in truth about anything; people who imagine they have a hotline to the truth are usually just repeating whatever it is currently fashionable to say. Atoms.

Here is what Democritus says of atoms:

By convention sweet is sweet, by convention bitter is bitter, by convention hot is hot, by convention cold is cold . . . in reality, only atoms and the void.

He envisaged a model that is familiar to us, but was revolutionary in its time. The universe is purposeless. It consists of nothing more than atoms of different shapes and sizes bouncing around in a vacuum, sometimes clinging together to make more complicated compounds. Since no evidence was then available to support his model, Democritus departed from the empirical principles that he argued for elsewhere, but what a magnificent leap of the imagination! Pythagoras similarly guessed that the world is round. Aristarchus that it orbits the sun. The evidence we have now tells us that we need to replace the idea that atoms are indivisible by the findings of quantum physics, but upgrading Greek physics in this way would make little difference to what comes next. 12 Although they often observed nature with close attention. Aristotle, for example, is credited with being the first biologist. Eratosthenes came up with a pretty good estimate of the diameter of the Earth by measuring the length of shadows at noon at different places in Egypt. 13 The great mathematician Archimedes credits Democritus with the discovery that the volume of a Euclidean cone is one third the product of its base area times its height—even when the base is irregular and the cone leans to one side. Aristotle says of Democritus that he seems to have thought of everything. Robert Burton has a lot to say about Democritus. In particular: “Lactantius, in his Book of Wisdom, proves . . . Democritus took all from Leucippus, and left, saith he, the inheritance of his folly to Epicurus, witless wisdom, &c. The like he holds of Plato, Aristippus, and the rest, making no difference between them and the beasts, saving that they could speak.” It is true that Leucippus deserves a good share of the credit for atomism.

1.4. EPICUREAN ISMS

13

Ghost in the machine? One dictionary definition says the soul is the spiritual or immaterial part of a human being, regarded as immortal. Epicurus certainly didn’t mean this when translated as talking about souls. He probably meant no more than mind or self: whatever we are thinking of when we refer to ourselves as me—the name I call myself. Such a qualification is necessary when taking on the Epicurean doctrine that, just like everything else, even our souls are made of atoms. Nobody today believes our minds are made of atoms, and only those anxious to discredit materialistic explanations insist that a materialist must believe such nonsense, or else admit that spirits exist. One might as well argue that a football match must be made of atoms. Insofar as it can be said to be made of anything at a fundamental level, it is made of the motions of particles relative to each other. It isn’t anything material at all; it is an activity—like a wave upon the sea. Similarly, a thought is an activity in the brain in which electrochemical signals are transmitted through a complex network of neurons. It is true that neurons are made of atoms. Take them away and thought becomes impossible. Just as you can’t play football without a ball, so you can’t think without a brain. Or to say the same thing more controversially: there is no ghost in the machine. What people attribute to a ghostly presence is just the machine working. Why death isn’t to be feared. The atoms that make up our bodies are scattered when we die. Epicurus argued that the same is true of the atoms that he believed make up our minds. So our minds fall apart when we die, just like our bodies. So there is no life after death. Our “souls” are not immortal. We don’t need to believe that we have souls made of atoms to come to the same conclusion. We only need to believe that thought is impossible without a brain to do the thinking. Where is the evidence anyway that we have souls that survive our death? Objective investigation of the claims of spiritualists has been uniformly negative. It is presumably their fear of death that makes them so credulous. But Epicurus argues that the only thing to fear about death is the fear of death itself. Why should we not fear death? It may be rational perhaps to fear the act of dying, but it is irrational to fear death itself because it is impossible to experience death. We can’t experience death when we are alive—but nor can we experience death when we are dead, because we won’t be there to experience it. Epicurus would therefore have been totally unworried if his soul had been threatened in his own time with the torments of the sixth circle of Hell after his death. His great popularity in the ancient world is similarly attributed to his releasing his disciples from the same kind of anxiety about the imagined misery of Hades. Our souls can’t be tormented after our death, because they don’t exist when we are dead any more than they existed before we were conceived. Being dead is unimaginable. When Epicurus said that death is nothing to us, I think he meant more than that we can’t experience being dead. His argument implies that we can’t even empathize with a dead person (Section 2.5.1). We can’t put ourselves in the position of a dead person to see things from their point of view. We only imagine that we imagine how it is to be dead. I can’t even imagine what

14

CHAPTER 1. EPICURUS

people imagine they are imagining when they talk about being unborn. We can certainly imagine our bodies being dead and lying in a grave. We can entertain fantasies about immortal souls, just as we can fantasize about unicorns or leprechauns. But we can’t really imagine our brains working and not working at the same time—any more than we can genuinely imagine a square circle. People do indeed say “I wish I were dead” when feeling depressed. But such a comparison with how Alice feels now and how she would feel if she were dead makes no sense; she wouldn’t feel anything if she were dead because she wouldn’t be there to feel it. We could interpret her cry of despair to mean “I wish my dying moment were advanced” but I don’t suppose anyone would think such a scientific substitute adequate to express the anguish most of us feel at one time or another. I am making this point here because I think the Epicurean attitude to death implies more than that death isn’t to be feared—it also implies that the length of our lives is of no importance in itself (Section 3.3). What matters is what we make of our lives while alive, not how long we live. Plato on the immortality of the soul. Plato offers at least two arguments said to be proofs that the soul is immortal. The famous argument appears in Plato’s Meno. Socrates asks a slave boy a series of questions. The boy’s correct answers amount to a proof of the theorem of Pythagoras in a simple case.14 Socrates then claims that the boy must really have unconsciously known the proof all the time, since he wouldn’t otherwise have been able to answer correctly (thereby anticipating Kant’s assertion that Euclidean geometry is a synthetic a priori). But the boy had no opportunity to learn the proof in this life, and so he must have been born with a soul that knew the proof already. Therefore souls are immortal. This is a lot to deduce from so little! An unremarkable alternative explanation is that Euclidean geometry is a very close approximation to real geometry here on Earth, and so the boy picked up enough knowledge of Euclidean geometry in his daily life to give the necessary correct answers. Why else do clever schoolchildren usually greet Euclid’s axioms as obvious? Plato’s Phaedo also insists that learning is really remembering, and it is true that babies seem to be born with synthetic a prioris, but we can locate them in the baby’s genes rather than invent an immortal soul as their carrier. The pattern of a gene might be said to be theoretically immortal, but I don’t suppose Socrates would have been willing to call it a soul. Nor would Socrates have been willing to assign immortal souls to computers, although they can check the validity of modern mathematical proofs even better than slave boys. 14 The boy is intelligent, but would he really have followed the intricate argument that Socrates offers? Not without the diagrams that Socrates must have been drawing. Even then, the argument isn’t so transparent as is claimed. I wonder how many writers of footnotes to Plato have really understood it. I would tell the story this way if required to do without diagrams. We have to prove the theorem of Pythagoras about triangles with one right angle in the simple case when the shorter sides are equal. Imagine four copies of such a triangle and place them together to make one large square with their right angles all together at its center. Its area is the square on the hypotenuse. Now rearrange two of the triangles into a square by putting their hypotenuses together. Its area is the square on one of the sides. We can do this twice with our four original triangles, so the square on the hypotenuse is equal to the sum of the squares on the other two sides.

1.4. EPICUREAN ISMS

15

The second argument appears in Plato’s Republic. Socrates tells us that, just as bodies can only die of ills that afflict the body, so souls can only die of ills that afflict the soul. These ills consist of various kinds of wickedness. But we observe that very wicked people still have souls. Therefore souls can’t be destroyed. This wretched argument isn’t worth refuting. It is an insult to the memory of the real Socrates that Plato should put it in the mouth of his fictional Socrates. Suicide? Critics accuse Epicurus of hypocricy because he wrote a will. Why write a will if death is the end of your existence? For the same reason that people care for the future welfare of friends that they know they will never see again! The circumstances of his death also generate accusations of hypocricy. He died suffering badly from kidney stones or something similarly painful. We are told that he nevertheless died cheerfully while seeking relief in a hot bath. As a one-time victim of kidney stones myself, I understand why he might have actually opened his veins to put an end to his suffering. Others with similar suspicions charge him with thereby contradicting his strictures against suicide. One might say the same of Socrates if one thought that he really chose to drink the famous cup of hemlock because he preferred death to fleeing his beloved Athens at the age of 71. The ancient attitude to suicide was mixed. Some societies thought it admirable in appropriate circumstances and others condemned it. On the island of Cos, for example, there was a tradition of old people ending their lives peacefully to make room for others. Pompey is reported as attending a party in 18 AD that culminated in the aged hostess drinking hemlock. As with Socrates, she seems to have died without pain. It seems barbaric that most countries in the modern world still deny a similar easy exit even to those who cry out for relief from their suffering. Socrates did argue against suicide, but is the same true of Epicurus? Such an attitude doesn’t seem consistent with his general views on death. He certainly said that he had no respect for someone who gives more than one reason for committing suicide, but he probably just meant by this that nobody would give two reasons if they were able to give one decisive reason.15 In his surviving Letter to Menoeceus, he doesn’t advocate suicide, but he seems to accept that it can sometimes be a rational choice (Section 3.3). Epicurus died in pain, but how would he otherwise have responded to the slogan: Eat, drink and be merry for tomorrow you die? His argument that one should compare the benefits of the former against its potential future costs doesn’t apply if you have indeed conquered your fear of death, and tomorrow really means tomorrow. I think he would have said that in the tranquil state of ataraxia achievable by serious philosophical reflection, the benefits of making merry become negligible. However, ataraxia merits a section of it own later in the chapter (Section 1.5). Free will? We carry in our heads a standard model of ourselves in which the self freely chooses among the available actions to secure whatever outcome it likes best. But how can we choose freely in the totally deterministic world of atoms 15 So

he wouldn’t have approved of Lucretius giving six reasons why nothing can come of nothing.

16

CHAPTER 1. EPICURUS

proposed by Democritus? Epicurus sought to deal with this problem by suggesting that atoms occasionally swerve from what would otherwise be their pre-ordained paths. Modern writers sometimes appeal to quantum physics for the same reason. My own take on free will echoes Spinoza and many others before and after. Rather than follow the standard model in attributing our seeming capacity for free choice to a mysterious unmoved mover in our heads, it is almost trite nowadays to argue that we think our will is free because we observe our thoughts or actions but not how our minds come to think those thoughts or perform those actions.16 Imagine a computer programmed to solve quadratic equations. If it were also programmed to monitor its own calculations, it would know that the solutions of a particular equation were determined in advance of the calculation. But what if it were not so programmed? Before the calculation, it might then report itself as free to come up with any numbers at all within its computational range. After the calculation, it would simply report the choice of solution it had made. If programmed with some introspective capacity, it might reason that its program must be equipped with a calculating subroutine that it is unable to monitor. Doesn’t it make more sense for us to do the same rather than inventing some supernatural entity for which we have no direct evidence at all? For this simple reason, there is no necessary contradiction between our feelings of internal freedom and a deterministic world. There was therefore no need for Epicurus to gild the atomic lily with arbitrary swerves (or for modern thinkers to appeal to quantum phenomena). Multiple models. People commonly react badly to such attempts to reconcile their very strong feelings that our will is free with a deterministic world, partly because the assumptions of the standard model are embedded in our language, and partly because they think that acceptance of such arguments entails discarding our standard model of ourselves—whose essence is quite likely encoded in our genes. But if evolution wrote the standard model in our genes, it wrote it there because it works very well most of the time. So it would be stupid to throw the model away because it isn’t consistent with some immeasurably more complex model that fits the facts of physics better. One might as well throw away the idea that tables are solid because quantum physics shows that they consist mostly of empty space. It is true that one must then simultaneously maintain models that are inconsistent with each other, using in practice whichever model has worked out best in whatever environment one finds oneself. But so what? Isn’t this just how physicists proceed with the inconsistent theories of quantum mechanics and relativity? Can such an approach to the supposedly deep problem of consciousness really be adequate? When this question is asked, I am reminded of the fierce debates that raged in my youth about the nature of life. Nobody asks any more what kind of Life Force animates what would otherwise be a dead body. Scientific advances in biology have shown that the assumptions inherent in asking such questions are unfounded. Life isn’t the kind of thing that the debate took for granted. It isn’t a 16 This is a paraphrase of Spinoza. Extended discussions can be found in Daniel Dennett: Consciousness Explained, or Thomas Metzinger: The Ego Tunnel, and elsewhere.

1.4. EPICUREAN ISMS

17

thing at all. Advances in neuroscience suggest that the same is true of the current debate about consciousness. If so, we will never solve the problem of consciousness as currently understood because the manner in which it is posed is ill-conceived. Kant on free will. Epicurus would have lined up with Hume in arguing that philosophers always cheat when they claim to deduce an ought from an is.17 The problem is with what Kant called categorical imperatives—oughts that bind unconditionally. There is no problem in deducing hypothetical oughts, which are the only oughts that moral naturalists like Epicurus recognize. For example, it is ten past three—if you want to catch your train, you ought to leave now. Kant’s variant on the standard trick with the definition of ought is to claim that we have a transcendental variety of free will from the dictum: ought implies can. It is true that I wouldn’t tell you that you ought to leave now if you want to catch your train unless I thought that you would indeed catch your train by leaving now. The ought-implies-can principle therefore works fine for hypothetical imperatives, but Kant applies it to categorical imperatives. He thereby deduces the existence of a supernatural unmoved mover in our heads, who can somehow suspend the laws of physics when necessary, from the fact that philosophers like him say that there are things we ought to do regardless of the circumstances! (Section 4.2.1)

1.4.4

Atheism

Epicurus didn’t deny the existence of gods, but he maintained that they took no interest in human affairs. So he added to his popularity by arguing that there is no more reason to fear the gods than there is to fear death. This position leads commentators to call him a deist, although the eighteenth century philosophers who coined the term regarded God as a perfect unmoved mover who created the universe, whereas the “gods” that Epicurus envisaged were just part of the natural universe, and so made of atoms like the rest of us. I think the reason Epicurus invented such a minimal conception of the gods is that he didn’t believe in the supernatural at all. Not only didn’t he believe in gods or immortal souls, he didn’t believe in ghosts or spirits and the like (as a surprising proportion of professed atheists supposedly do today). He was an atheist for the same reason that the same is true of nearly all modern biologists—because there is no credible evidence. As Hume said of miracles: what is more likely, that a second-hand report of a miracle is true, or that it is mistaken or invented? The reason I think that Epicurus chose not to deny the existence of gods is that it can be dangerous not to go along with whatever religion is currently embraced by those around you. It wasn’t so long before he set up the Garden in Athens 17 Hume’s observation is often called the naturalistic fallacy, but it isn’t at all the same as the “rationalist” naturalistic fallacy of the philosopher Moore: that the Good is a transcendental concept that we all intuitively perceive, but can’t be defined in naturalistic terms. The reason Moore gives is that one can always ask of any complex of natural phenomena said to embody the Good, whether it is indeed the Good. One might similarly argue that if I am told that the lady with golden hair is Alice, I should deduce that she can’t really be Alice because I have doubts about whether she has been correctly identified!

18

CHAPTER 1. EPICURUS

that the trumped-up charges that led to the guilty verdict against Socrates included introducing strange gods. Epicurus probably had similar problems in Mytilene. For most of us, only our careers are nowadays at risk for holding unorthodox views. Hume, for example, was denied a Chair because he was held to be an atheist—although his magnificent Essays on Natural Religion was carefully only published after his death. Even in modern times, Bertrand Russell was hounded out of his Chair in New York because he was a professed atheist. A whole book could be written enumerating similar examples, especially if secular religions like Marxism are included. What kind of atheist was Epicurus? There are various definitions. Religious folk like to say that atheists are people who think they can prove conclusively that God doesn’t exist. Such atheists are indeed fairly common. For them, atheism is just another kind of religion. But Epicurus wasn’t one of these. Empiricists don’t think they can prove anything but a tautology conclusively: not even that there are no fairies at the bottom of their gardens. My dictionary merely says that an atheist disbelieves or lacks belief in the existence of God or gods. I think Epicurus probably fell into the second category: he lacked belief in the supernatural altogether because he saw no evidence for it. He did admittedly say that he regarded dreams about gods as evidence that gods exist, but this sounds no more than window-dressing for his professed deism. Problem of evil. Is God not able to prevent evil? Does He not want to prevent evil? But if He is neither omnipotent nor benevolent, why call Him God? This so-called Trilemma is attributed to Epicurus, but seems more likely to have been proposed by one of his later followers. I hope I will be forgiven for not following up such theological issues.

1.4.5

Naturalism

In biology, a naturalist is somebody like Aristotle, who gathered facts about living creatures, and sought to classify them within an orderly system without appealing to supernatural or metaphysical ideas. Epicurus tried to apply the same empirical principles to human morality. However, in traditional moral philosophy, a naturalist is defined as a philosopher who tries to find naturalistic definitions for the words people use when talking about morality. Traditional philosophers like this definition because it would make moral naturalism a hopeless task. One might as well try to understand pneumatics by starting with the phrase “Nature abhors a vacuum” and asking why an all-powerful lady called Nature dislikes vacuums so much. Epicurus understood that to make sense of how moral norms actually work, rather than how orthodox pundits say they should work, one has to pay close attention to what people do, and not just listen to what they say. Relativism. The pre-Socratic philosopher Xenophanes made an empirical observation which says everything that needs to be said about the supposedly universal

1.4. EPICUREAN ISMS

19

nature of the various supernatural entities that have been invented down the ages: “The gods of the Ethiopians are black and flat-nosed, and the gods of the Thracians are red-haired and blue-eyed.” The same is true of moral norms. The Greek historian Herodotus reports that a party of Greeks and a party of Indians were summoned to the court of the Great King of Persia in an early anthropological experiment. The Greeks were horrified to learn that the Indians ate their dead fathers. The Indians were no less horrified to learn that the Greeks burned theirs! We are similarly horrified when Plato tells us that someone who can no longer work in his supposedly ideal Republic loses the right to live. He would have been similarly horrified to learn that we think slavery is abhorrent. Norms aren’t absolute. Epicurus understood that the moral norms that actually operate in real societies are relative. Even when they are universal, as with our attitude to incest, they aren’t absolute.18 In speaking of fairness norms, Epicurus argued that justice isn’t a notion handed down by the gods or some Platonic absolute to be discovered by thinking hard—it is simply a useful idea that helps people avoid doing harm to each other. Significantly, he noted that what counts as fair is the same for everybody at some particular time and place, but need not be the same at different times and places. The idea that morality may differ between societies is denounced as relativism by all traditional religions. The Pope before last made a particular point of denouncing relativism in all its forms. It is therefore a delight when priests inadvertently reveal the truth that the moral doctrines of the Church evolve over time along with a society’s social contract. In England, where many babies are now born out of wedlock, the Archbishop of York of some years ago reassured us that even living in sin is no longer sinful! Nowadays, English gays who were until recently harried for a bit of harmless cottaging, are now arrayed in miter and crook, with the opportunity to harass those unfortunate straights who are still playing by the old rules. How come people are so ready to submit themselves to such threadbare sources of authority? Even to die for beliefs they have never examined? Surely nobody who had looked at accounts of the unholy compromise engineered by Constantine the Great at the Council of Nicaea that resulted in the incoherent doctrine of the Trinity would want to burn people at the stake for questioning the wording? What of the hilarious antics of Joseph Smith in the founding of the Mormon Church? What of the folk who worshipped the god Serapis with apparent sincerity, even though Serapis was openly and publicly invented by Alexander’s successor in Egypt to bring Greeks and Egyptians together? I read in the newspaper today of somebody who dug up a magnificent fossil of an icthyosaur that had been discovered and reburied by his grandparents because it offended their religious beliefs. What passes through the minds of such folk? What is the psychological mechanism that religious memes use to disable the critical 18 Absolute implies existing independently of human evolutionary history, both biological and social. Philosophers sometimes say objective instead, but I prefer its primary dictionary definition. The relative moral norms of Epicurus are objective in this sense by their very nature.

20

CHAPTER 1. EPICURUS

faculty of the minds they infect? Is the same mechanism responsible for the success of the memes that infect modern populists and conspiracy theorists? Can we be sure that our own minds aren’t crippled by some similarly clever meme? Lots of questions, but no answers! Social contract. A social contract sounds like something legal. It would have been better described as a social consensus that simply guides how the citizens of a society sometimes manage to get along together without doing each other any harm. Philosophers commonly quote Locke, Rousseau, and Kant as the chief proponents of this notion, but they would do better to return to Epicurus. Part of the reason why Epicurus is neglected is that his matter-of-fact approach denies that some invented source of celestial or metaphysical authority is necessary to enforce the rules of a social contract. David Hume used the word convention to describe the kind of rules that enforce themselves. They don’t need some outside party to enforce them because they coordinate behavior on one of the many equilibria in what modern game theorists sometimes call the “game of life”. Such equilibria are self-enforcing because nobody can gain by being the first to deviate from the behavior they specify (Section 2.2). David Hume anticipated this notion of an equilibrium in the following famous passage: Two men, who pull the oars of a boat, do it by an agreement or convention, though they have never given promises to each other. Nor is the rule concerning the stability of possession the less derived from human conventions, that it arises gradually, and acquires force by a slow progression, and by our repeated experience of the inconveniences of transgressing it. On the contrary, this experience assures us still more, that the sense of interest has become common to all our fellows, and gives us a confidence of the future regularity of their conduct: And it is only on the expectation of this, that our moderation and abstinence are founded. In like manner are languages gradually established by human conventions without any promise. In like manner do gold and silver become the common measures of exchange, and are esteemed sufficient payment for what is of a hundred times their value.

In a more homely style, Hume later offers a delightful account of the conventions observed by traffic in the narrow streets of the Edinburgh of his day. When, for example, does a sedan chair give way to brewer’s dray? But such examples only encourage traditionalists in their opposition. Is the way a society is organized no more than a consensus to drive on the left or the right? Such a reaction is similar to the traditional disdain for the idea that our brains might honor the laws of physics. Are all our fine thoughts no more than the grinding of gears in some chocolate-dispensing machine? One might as well argue that Jesus Christ had no more to say than that Samaritans sometimes help injured folk found by the roadside. His parables exist to illustrate moral principles in as simple a setting as possible—and so do the parables of empiricists like David Hume, and such modern successors as Tom Schelling and David Lewis.19 19 David Hume: Enquiries Concerning Human Understanding, Tom Schelling: Strategy of Conflict, David Lewis: Conventions. Also Ken Binmore: Natural Justice.

1.5. HOW TO LIVE?

21

No preaching. The idea that questions they thought lay in their domain are actually scientific is bad news for traditional philosophers; but worse is to come. They like to dismiss moral naturalism of the kind Epicurus promoted as a brand of moral subjectivism, in which everybody absurdly makes up their own morality for themselves. But whether something is held to be right or wrong in a given society at a given time isn’t at all subjective, it is a matter of objective fact that anthropologists do their best to record. For example, it is an objective fact that people drive on the left in Japan. Nobody thinks it is written in the stars that the Japanese should drive on the left, but it doesn’t follow that they each make their own mind up about the side of the road on which to drive! The real problem for traditionalists is that different societies don’t always have the same standards of right and wrong. So on what basis are they to preach to others on how people should live their lives? The answer offered by moral naturalism is a bitter pill to swallow. It undercuts the authority claimed by utopians who feel they have a mission to convert the world. It can offer a culture-free assessment of the kinds of social contract that are feasible for a human society, but it denies the existence of some privileged culture-free way of choosing an optimum outcome from this feasible set. It doesn’t say as traditionalists commonly insist that all social contacts are equally Good. It denies that there is such an absolute Good. Moral naturalists can tell their fellow human beings about the sort of society in which they would like to bring up their children. They can seek to infect others with the enthusiasm they feel for their personal vision of the good life (as Epicurus is about to do in the next section). They can urge others to join them in reforming society in order to realize those aspirations that are sufficiently widely shared (which Epicurus wouldn’t have done). What naturalists can’t do is to claim that their scientific knowledge converts their personal preferences over social contracts into a privileged set of moral values that trumps the personal preferences of others. Science can no more be an authority on what should count as the absolute Good or Right than such traditional skyhooks as Practical Reason, Moral Intuition, General Will, or Natural Law invented by a variety of famous philosophers as metaphysical substitutes for the gods of old.

1.5

How to Live?

What advice did Epicurus offer on how best to live one’s life? His disciple Philodemus is usually quoted as summarizing the wisdom of Epicurus in the following grandly named Tetrapharmakos: Don’t fear God. Don’t fear death What is worth having is easy to get. Suffering is easy to bear.

Epicurus himself would perhaps have added friendship as an essential. He might also have warned against mindlessly following the herd, rather than thinking your

22

CHAPTER 1. EPICURUS

own thoughts for yourself. But the immediate question is why Epicurus held that suffering is easy to bear. Ataraxia. Epicurus reportedly bore the pain he suffered when dying with an admirable equanimity. We have a better documented report of the similarly painful death of David Hume, probably from stomach cancer. Hume was visited shortly before his demise by a brash James Boswell wanting to know how it felt to be an atheist on his deathbed.20 Not only did Hume echo Epicurus in saying that he was no more concerned that he wouldn’t exist after his death than he was that he didn’t exist before his birth, he also amazed Boswell with his cheerfulness. Skeptics may doubt whether either Epicurus or Hume were able to sustain their cheerful show for any length of time, but the philosophical self-examination that both advocated isn’t so different from the kind of cognitive therapy that is nowadays accepted as sometimes being effective in managing pain. Their deathbed equanimity—even if partly assumed—is to be compared with the ranting of the poet Dylan Thomas: Do not go gentle into that good night, Old age should burn and rave at close of day; Rage, rage against the dying of the light.

Another of my heroes—John von Neumann, the inventor of game theory and much else—did indeed rage on his deathbed against the dying of the light. His mind was shattered and his pain was excruciating. But one can’t help but wonder whether he wouldn’t have coped better if he had learned to conquer his fear of death like Epicurus and Hume. Later Epicureans attributed the calm that Epicurus showed in dying to his having achieved ataraxia—an ultimate state of tranquility that can supposedly be achieved by following a self-administered program of intellectual cognitive therapy. They argued that achieving ataraxy immunizes us against being troubled by things going wrong in the external world. So pain becomes a distant irrelevance. This mundane interpretation translates ataraxia as tranquility, but a more popular transcendental interpretation identifies ataraxia with the Buddhist notion of enlightenment—an unimprovable state of true happiness that frees adepts from concern with events in the physical world, so that the joys and sorrows that trouble ordinary folk cease to be of any relevance to their states of mind. Time loses all significance: to achieve ataraxy for an instant is the same as for all eternity. To quote from the poets again, Omar Khayyam called for “infinite ecstasy, indefinitely prolonged”. The transcendental version of ataraxia would have fulfilled his wish, except that his ecstasy wouldn’t have been of the senses but of the mind. 21 Buddhism? How come ataraxia seems so like enlightenment in Buddhism? The philosophical reflection recommended by western philosophers is similarly related 20 James Boswell is famous as the biographer of Samuel Johnson, whom he worshipped as the great literary genius of his time. Ironically, we now find Johnson’s work tiresomely overwritten, but I for one continue to delight in the freshness of Boswell’s biography. 21 Did he need to specify that his ecstatic state be indefinitely prolonged? How could his ecstasy be infinite if he were anxious that it might be cut short?

1.5. HOW TO LIVE?

23

to the art of meditation widely practised in the East. Democritus is reported to have visited India not so long after the time of Buddha (563-483 BC?). Could he have brought the idea of enlightenment back with him to Greece? Even if he didn’t encounter Buddhists, he would surely have met Jainists, by whom Buddha must have been inspired to some extent.22 Neuroscience. I was more skeptical that ataraxia can transcend being a matterof-fact state of tranquility before reading some of the (admittedly speculative) neuroscience on reported spiritual experiences. Here is what the neuroscientists Newberg and d’Aquili have to say (Section 3.3): . . . As this cycle builds, it leads to the complete shutdown of the spatial awareness neural circuits, generating as a consequence, the burst of ecstatic liberation in which we seem to be united with the Infinity of Being that is so characteristic of being in a trancelike state.

My own guess is that ataraxy was no more for Epicurus than a hard-won state of tranquility, but perhaps he would have welcomed the exhilaration said to accompany Buddhist enlightenment (even if achieved through suspending one’s conscious rationality rather than focussing it). On the other hand, Epicurus wouldn’t have taken seriously d’Aquili’s suggestion that the neurological mechanism that supposedly allows us to experience enlightenment was implanted in our heads by God so that He could teach us what spirituality is like.

1.5.1

Secular Puritanism?

The transcendental version of ataraxia will perhaps seem too airy-fairy for some who are otherwise attracted by the down-to-earth philosophy of the Epicureans. But one doesn’t have to believe that the transcendental version is attainable or desirable to value what Epicurus suggests is a good way to lead one’s ordinary life. One might say that he advocated a kind of secular puritanism shorn of its fanatical element. I am reminded, indeed, of John Bunyan’s Pilgrim’s Progress, in which Christian seeks to relieve his troubled mind by following the strait-and-narrow path to the Celestial City. Epicureans would need to replace the latter by ataraxy (and the otherwise very similar Stoics by virtue). However, the difficulties along the way—particularly the Valley of the Shadow of Death—are to be surmounted with the same indomitable courage. But it is the temptation and persecution that Christian faced when he reached Vanity Fair that would perhaps engage Epicurus most if he could see what modern society has become; that and the anguish Christian suffered in the dungeon of the Giant Despair. Despair. To what extent were the ancients afflicted with the neuroses of the mind that bring so many of us low today? I suspect they were mostly too busy just 22 Pyrrho (360-270 BC?) similarly travelled to India (with Alexander the Great). Perhaps he learned there the brand of skepticism that was strong in both Jainism and Buddhism. Pyrrhonism was revived later by Sextus Empiricus (160-210 AD?), who held that tranquility is to be found by suspending judgement after fully exploring the arguments on both sides of a question.

24

CHAPTER 1. EPICURUS

getting by to wallow in the self-pity to which we tend to resort nowadays when things go wrong. But Epicurus taught us that we needn’t be helpless victims of misfortune. Epictetus the Stoic taught the same lesson even better. As the imprisoned Richard Lovelace put it in his 1642 poem when he wanted to tell the world that he was going to remain tranquil no matter what the world threw at him: Stone walls do not a prison make, Nor iron bars a cage.

Cognitive therapy? Nobody thinks the cognitive therapy advocated by both Epicureans and Stoics can substitute for more invasive treatment in serious psychiatric cases, but for lesser cases of mental unrest we might usefully follow Epicurus in not seeking to fight our biology, but to go along with it. We are programmed to seek pleasure and avoid pain, so let us use our reasoning power to this end. For example, everybody knows that evolution equipped us with the capacity to feel physical pain to make us pay attention when our body is damaged. So when you are in pain, focus your attention on its source. If you are successful, the pain may remain, but it won’t hurt so much. Similarly, when you feel anxious, you might conjecture that your body is telling you to pay attention to your feasible set of possible actions. So consider your feasible set very closely. Even when it turns out that no feasible action can help, the procedure can still be effective in lifting anxiety. What of regret? Is your brain not telling you to learn from your past mistakes? If the regret is constantly recycled, perhaps your brain is telling you that you haven’t learned the right lesson. So try repackaging the troublesome memory with a positive slant—yes, I am now learning something about myself that I should have learned before. Vanity Fair. Tranquility can’t be achieved without accepting the way your body is put together. This includes recognizing that we are social animals. It isn’t natural that we should be alone too much. We need the company of family and friends. But we didn’t evolve to live in a huge modern society. According to Robin Dunbar, we have room in our brains to model at most 150 other people sufficiently well to be able to get along with them without the risk of our mutual ignorance leading to strife. The demands of modern life therefore take us beyond both our biological competence and our biological needs. Epicurus dealt with this problem by founding the Garden as a small community of close friends that took little or no part in Athenian social life. Such a solution isn’t practical for most of us. We have no choice but to live the life of a herd animal. As part of a herd, we can get by most of the time by simply copying the behavior of those around us.23 Indeed, models of herd behavior constructed by economic theorists suggest that imitating our peers in this way can sometimes be optimal. However, such economic models are too simplified to capture how herd societies 23 Here is Robert Burton on herd behavior: Alexander stooped; so did his courtiers: Alphonsus turned his head; and so did his parasites. Sabina Poppaea, Nero’s wife, wore amber-colour’d hair, and so did all the Roman ladies; in an instant. To which one might add that Wittgenstein’s followers in Cambridge didn’t just copy his philosophy, they also famously copied his mannerisms and informal style of dress.

1.6. SOME QUOTES FROM EPICURUS

25

can drift into collective idiocies that may persist for a long time before the rational few succeed in applying a correction. The folly of the Athenian democracy during the Peloponnesian War is a good example. But things are no different now, as the current wave of populism shows. John Bunyan used the parable of Vanity Fair to illustrate the dangers that await those who think for themselves in a herd society. When Christian refuses to indulge in the excesses of gluttony and lust that have become fashionable in the town, he is tried by Lord Vainglory for his presumption in differing from the herd, and barely escapes execution. Following the fashion? What would Epicurus have made of Keeping up with the Kardashians? After recovering from his astonishment, I think he would say that we may have no alternative but to live as part of a herd, but we don’t need mindlessly to copy its goals and aspirations. What, for example, of the pursuit of status? Epicurus would ask whether raising your status level will indeed increase your pleasure or decrease your pain. Will you really achieve satisfaction by making lots of money to show off with? Is it really great to have lots of followers on Facebook or Twitter? Are pop stars or football heroes more content than other folk? How about gourmets and aesthetes? What of sexual athletes? Or even the winners of Nobel Prizes and the like? Tranquility—at least for older folk—is a greater prize than any of these, and we can achieve it by teaching ourselves to be satisfied with what we have already. To which one can only respond: easier said than done!

1.6

Some Quotes from Epicurus

This final section offers a sample in no particular order from surviving quotes plausibly attributed to Epicurus himself.24 He who is not satisfied with a little, is satisfied with nothing. The wealth required by nature is limited and is easy to procure; but the wealth required by vain ideals extends to infinity. The opinions held by most people about the gods are not true conceptions of them but fallacious notions, according to which awful penalties are meted out to the evil and the greatest of blessings to the good. If a person fights the clear evidence of his senses he will never be able to share in genuine tranquility. I have never wished to cater to the crowd; for what I know they do not approve, and what they approve I do not know. Natural justice is a symbol or expression of usefulness, to prevent one person from harming or being harmed by another. There is no such thing as justice in the abstract; it is merely a compact between men. Not what we have, but what we enjoy, constitutes our abundance. 24 For

much more, see Brad Inwood: The Epicurus Reader: Selected Writings and Testimonia.

26

CHAPTER 1. EPICURUS It is better for you to be free of fear lying upon a pallet, than to have a golden couch and a rich table and be full of trouble. The time when most of you should withdraw into yourself is when you are forced to be in a crowd. Of all the means to insure happiness throughout the whole life, by far the most important is the acquisition of friends. To eat and drink without a friend is to devour like the lion and the wolf. The art of living well and the art of dying well are one. Death, therefore, the most awful of evils, is nothing to us, seeing that, when we are, death is not come, and, when death is come, we are not. You don’t develop courage by being happy in your relationships everyday. You develop it by surviving difficult times and challenging adversity. No pleasure is a bad thing in itself, but the things which produce certain pleasures entail disturbances many times greater than the pleasures themselves. Every pain is easy to disregard; for that which is intense is of brief duration, and those bodily pains that last long are mild. He is of very small account for whom there are many good reasons for ending his life. It is impossible for a man who secretly violates the terms of the agreement not to harm or be harmed to feel confident that he will remain undiscovered, even if he has already escaped ten thousand times; for until his death he is never sure that he will not be detected. A happy day is this on which I write to you . . . the pains I feel . . . could not be greater. But all of this is opposed by the happiness which the soul experiences, remembering our conversations of a bygone time.

Chapter 2

Rationality as Consistency 2.1

Social Engineering

Engineering is usually understood to be the application of the physical sciences to practical problems in the physical world. Social engineering is the application of the social sciences to practical problems in the social world.1 Everybody knows that the social sciences lag far behind the physical sciences. The topics in social science that are sufficiently advanced to be genuinely useful are few in number, and so the practical problems to which they can be applied are correspondingly limited in scope. I have in mind such cases as the organization of a factory, or the design of a trade union constitution, or the regulation of the electricity industry. The big successes so far are in the design of big-money auctions and matching processes, like those in which children are assigned to schools. 2 Utopia? As usual, traditional philosophy stands in the way. What people expect from social scientists is more of the beautiful but useless maunderings offered by Plato in his famous Republic, which remains one of the “great books” that every educated person should have read. Like the bible, it is certainly an impressive work of literature, by far the best on this score than any of the other (equally impractical) utopias that have been written down the ages. But the intelligentsia are so dazzled by its fine writing, that they somehow contrive to overlook its appalling message— just as religious folk avert their eyes from the absurdity of their founding myths. The Republic is particularly unsuitable for American undergraduates, but it is normally compulsory reading for liberal arts students. It advocates an authoritarian state ruled by philosopher kings supported by a socialist class of guardians denied 1 Social engineering isn’t intended here to be understood pejoratively to refer to the attempts by self-appointed gurus to impose their prejudices on others. Jeremy Bentham called them ipsedixists, because they have nothing to offer but an invitation to follow where they lead. 2 Al Roth won a Nobel Prize in 2012 partly for saving lives by solving the difficult task of matching people needing a kidney with those who would donate a kidney to someone they love if they had the right blood group. I was involved myself in the design of big telecom auctions in Belgium, Denmark, Greece, Israel, Hong Kong, and the UK. The UK auction raised $35 billion.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6_2

27

28

CHAPTER 2. RATIONALITY AS CONSISTENCY

private property. Apart from their military function, the guardians supervise a proletarian class who do all the work not done by slaves. The proles are so closely supervised that they get to make virtually no decisions for themselves. The nearest we have come in modern times is the old Soviet Union, with its secret police and stultifying bureaucracy, Karl Popper’s Open Society and its Enemies says all these things, but insofar as his book is taken notice of at all, it is treated with an amused contempt on the grounds that he fails to appreciate that the Republic is of great literary and historical interest. But such critics miss the point. What matters for a philosopher of science is whether the arguments are sound, and whether the message offered to students is relevant in modern times. The latter is clearly false. Plato himself would have advocated censoring the Republic if he were Popper, as he proposed censoring the passages in Homer of which he disapproved. As for Plato’s arguments, where are they? There is dialogue only when the fictional Socrates says the obvious, but immediate endorsement whenever he makes an unjustified leap to something substantial. The only exception is when Glaucon— one of Plato’s older brothers—puts the case for treating justice as a social norm in the manner of the yet unborn Epicurus. But he is summarily dismissed, and then rolls over to expose a soft intellectual underbelly. Everybody agrees that the Republic is actually an idealized version of Sparta, whose military success was widely admired in ancient Greece. However, Socrates supposedly deduces its constitution from first principles, one of which is a definition of justice as “minding your own business and not interfering with other people”. Is this what justice means? Not today, nor in ancient Greece, if we are to judge by what Epicurus and other authors say. Neither do the guardians refrain from interfering with other people when they supervise the proles. Sparta held a whole class of conquered Helots in bondage, and so Plato has his guardians do the same. Plato was an aristocrat, who despised the Athenian democracy. His uncles Critias and Charmides, after whom two Platonic dialogues are named, were leaders of the Tyranny of the Thirty Tyrants forced on Athens after its defeat by Sparta. To his eternal credit and at some personal risk, the real Socrates refused to participate in the arrest of those singled out for the judicial murders in which tyrants delight. To his credit also, Plato himself was sickened by the repression. However, the story Plato tells about Socrates’ own judicial murder after the democracy was restored has to be taken with a pinch of salt. I suspect Socrates was really condemned because the aristocrats who ran the Tyranny regarded him as their intellectual guru. Power corrupts. The immediate point isn’t that Plato endorsed supposedly absolute moral norms that we nowadays find repulsive. Those of us who think that moral norms are relative believe that such differences from one society to another are inevitable. It isn’t hard to predict, for example, how the future will judge our refusal to let those suffering hell on earth to find release in an easy death. The point worth making here is that Plato’s Republic is hopelessly impractical. As Spinoza would have said, it would only work in some golden age of the poets. The Republic would fall apart for the reason explained by George Orwell in his

2.2. MECHANISM DESIGN

29

Animal Farm parable, in which the animals rebel against the farmer and establish a socialist republic with the pigs as guardians. But it isn’t long before the pigs become indistinguishable from the farmers. Plato goes on about how education will make each citizen satisfied with their assigned roles, but you can’t educate humans not to be human. In fact, Spartans given power outside their own society were notoriously corrupt. Nowadays, everybody understands that power corrupts. If we want organizations that work they have to be immunized against corruption as best we can. Hume already had the solution to this problem in 1739, except that he should have included women too: In constraining any system of government and fixing the several checks and controls of the system, every man ought to be supposed a knave and to have no other end in all his actions than private interest.

This isn’t to say that everybody is a selfish knave, which clearly isn’t true. But if you put people in power, history warns us that they or their successors will eventually invent reasons why it is somehow legitimate to exploit their power to their own advantage. Who hasn’t suffered from the insufferable abuse of their power by jacksin-office who no longer even pretend that they aren’t putting their own interests before those of the public they are paid to serve?3 We already use Hume’s principle in evaluating expense claims and tax returns. The idea of mechanism design shows how to do it better using ideas drawn from game theory.

2.2

Mechanism Design

The basic idea behind mechanism design is as old as the bible. Two women were brought before King Solomon, both claiming to be the mother of a baby. To decide who should get the baby, he invented a simple game for the rivals to play. They were told that they could persist in their claims, or give up the baby to the other woman. If both persisted, the baby would be cut in half so that it could be shared between them. The real mother then gave up her claim but the false mother persisted. So Solomon awarded the baby to the mother who gave up her claim. This story wouldn’t have satisfied the inventors of modern mechanism design because it takes for granted that the rival claimants would believe Solomon when he lied to them about what game they were to play. Solomon would have done better to use his famous wisdom to find a more subtle punishment than cutting the baby in two to deter both claimants from persisting in their claims. Such a punishment would need to be chosen so that the true mother would be prepared to suffer it rather than lose the baby, but the false mother wouldn’t. The false mother would then predict that the true mother would persist in claiming the baby even if punished, and so retract her own claim rather than be punished herself. 3 In some countries, such abuse is part of an unofficial social contract—like tipping in a restaurant. The government can keep official salaries low because it is customary for people to pay a small bribe to oil the bureaucratic wheels. The officials perform adequately if bribed, and only get stroppy if not. But bribery is corrupt whether or not it is customary.

30

CHAPTER 2. RATIONALITY AS CONSISTENCY

Principals and agents. Economists say that Solomon is the principal in the preceding story and the two claimants are agents. Mechanism design was created to study what happens when a principal has no choice but to keep to the rules of the game he invents for the agents to play, and the agents play the game strategically. Three economists were awarded the Nobel Prize in 2007 for their contributions to modern mechanism design.4 They chose the term mechanism deliberately to draw attention to the role of mechanism design in social engineering. The mathematical details in particular cases can be seriously intricate, but the philosophy behind the idea is so simple that people usually think it obvious after it has been explained. Rules and regulations. We can’t avoid bureaucracy, but does the American internal revenue code (Title 26) need to be 6,550 pages long? One has to hire bureaucrats of one’s own to make sense of the layer upon layer of officialese invented by the bureaucrats hired by the government. My impression is that things are getting worse. How many of us have been afflicted with stupid boss syndrome? I have in mind the kind of boss who thinks the only way to manage is to micro-manage—to try to think up regulations that supposedly govern all possible contingencies, leaving their often more talented and better informed subordinates with no room to exercise their own initiative. We can’t do without rules and regulations in a modern society, but nor can we afford to stifle innovation and enterprise. Economists recognize the first of these two facts by accepting that some rules are necessary. The word mechanism in a social context actually means the assembly of all rules that govern an organization. The only caveat is that a rule must be enforceable. Everybody must therefore know what the rules are, and the anticipated cost of breaking a rule must be high enough that rational folk don’t break it. Economists take account of the second of the two facts by focussing attention on optimal mechanisms. If the criteria for optimality include a concern for efficient production, an optimal mechanism will then give at least some weight to the standard economic principle that decisions should be delegated down to the level at which the necessary knowledge and expertise resides. Adaptive behavior. My experience in working with governments is that, when the reform of a social mechanism is proposed, nobody wants to listen when it is explained that we don’t live in a golden age of the poets—that people will adapt their behavior to the new rules. They tend to laugh out loud when it is suggested that a simplified version of a very expensive reform be tested in a psychology laboratory to see how it works out with real people. 4 Leonid Hurwicz, Eric Maskin, and Roger Myerson. Myerson’s application to bargaining theory is a lot more impressive than the Judgement of Solomon. Suppose that a buyer and a seller are bargaining over the price of a house. Each has a valuation equally likely to be anything between one and two million dollars. If whatever bargaining procedure is used maximizes the expected total increase in wealth, how often will the house will be sold? Only when the buyer’s valuation exceeds the seller’s by a quarter of a million dollars! Of course, in a golden age of the poets, people would tell the truth when asked for their best offer, and so the answer would be: whenever the buyer values the house more than the seller.

2.2. MECHANISM DESIGN

31

In an example from real life, the new chair of the management committee of a health-care fund proposed abolishing the co-pay requirement for the fund’s clients. The extra demand on the fund’s resources was to be paid for by increasing next year’s subscriptions enough to cover the amount clients spent this year on co-pays. The economist on the committee argued that the subscription would need to be higher, but was in a minority of one when a vote was taken on “Whether people will visit their doctor when they don’t need to.” But the fund was in deficit the following year because a co-pay changes the behavior of people who are just on the edge of visiting or not visiting their doctor.5 The discipline of mechanism design comes with a built-in response to the phenomenon of adaptive behavior. The long-run prediction of the outcome to be expected from a reform assumes that the adaptation is over. Any corrupt or selfish behavior that the mechanism allows will then already have become established. The result is admittedly second-best as compared with the first-best outcome obtainable in a golden age of the poets. It accepts that some self-interested behavior may have to be tolerated because it can’t be realistically eliminated altogether. But the relevant comparison isn’t with the first-best outcome, but with the third-best or worse that we usually have to endure in the world as it is. Game theory. How do we predict how people will end up behaving after a new mechanism is installed? We treat the new regulations as the rules of a game to be played by the people who will be governed by the mechanism. They are then assumed to try out the various strategies at their disposal until their behavior stabilizes at an equilibrium, where nobody can profit by deviating unless somebody else deviates first. At such a Nash equilibrium, each player will be using a strategy that is a best reply to the strategies of the other players.6 The mechanism is evaluated in terms of what happens at this equilibrium. The designer then chooses whichever mechanism evaluated in this way best suits whatever design criteria are in use. 5 A mistake on a grander scale was made by the US Congress in 1990 when it passed an act intended to ensure that Medicare wouldn’t pay a lot more for its drugs than private health providers. The act said that a drug must be sold to Medicare at no more than 88% of the average selling price. However, an extra provision was tacked on saying that Medicare must also be offered at least as good a price as any retailer, but this would only work if drug manufacturers ignored the incentives it created for them. Why would they ever sell a drug to a retailer at less than 88% of the current average price if they must then sell the drug at the same price to a huge customer like Medicare? But if no drugs are sold at less than some fraction of the current average, then the average price will be forced up! 6 My Very Short Introduction to Game Theory is indeed short and has no equations. My Playing for Real provides the details while minimizing on mathematics. Game theory was invented by John von Neumann in 1928. It was perhaps the least of the scientific achievements of this extraordinary polymath. He was long dead when a Nobel Prize for game theory was awarded to John Nash, John Harsanyi, and Reinhard Selten in 1994. The equilibrium concept just described is named after Nash because he proved that all finite games have at least one such equilibrium if randomized strategies are allowed. Nash is one of the oddballs mentioned in Section 1.5.1. He was famously incapacitated for many years by a serious schizophrenic illness before the Nobel award. The movie A Beautiful Mind tells the story.

32

CHAPTER 2. RATIONALITY AS CONSISTENCY

What does a designer need to know? Like Epicurus and Hume, Thomas Hobbes was probably another secret atheist. He was certainly a materialist, as the following quote makes clear: What is the heart but a spring, and the nerves but so may strings, and the joints but so many wheels, giving motion to the whole body.

He is mentioned here because he put his finger on what a principal needs to know about the agents to work out a Nash equilibrium of a game in order to predict how they will learn to play it. Hobbes argues that people are characterized by their strength of body, their passions, their experience, and their reason. For the purposes of game theory, Alice’s strength of body translates into the strategies available to her in a game— what she can do. Her passions translate into her preferences—what she wants. Her experience translates into what she knows and what she believes. This chapter is mostly about modeling her preferences and beliefs. Her reason translates into the assumption that she acts rationally. Two models. The last of these assumptions should be distinguished from the claim that players always think rationally. It is true that when a rational solution of a game can be identified, it must be a Nash equilibrium because nobody would use their solution strategy if it weren’t the best thing to do when the other players use their solution strategies, but this will seldom be the reason that equilibria are played in practice. In real life, people get to equilibria by a process of trial-and-error. If such a process makes players better off as it proceeds, it can only stop when nobody can be made better off—at a Nash equilibrium. This dual interpretation of a Nash equilibrium is why game theory has been a success in evolutionary biology, even though genes—the players in an evolutionary game—are just molecules that can’t think at all. It allows biologists to skip the immensely complex task of tracing evolutionary dynamics back in time, and to focus instead on the end result, at which genes behave as if they had chosen rationally. Difficulties. It would be dishonest to seek to minimize the difficulties that arise in classifying players into types using Hobbes’ criteria. For example, nobody is likely to know the types of anybody else for certain. So how do we model their beliefs about each other? And what do they know about each other’s beliefs? And what do they know about what others know about what they know? This last question isn’t as exotic as it seems. For example, if Alice is a good boss, she delegates decisions down the chain of authority to Bob because she knows that he knows things she doesn’t know. Why care? We next survey the philosophy behind the way game theorists model their approach to the world. For those less interested in social engineering than the Epicurean approach to living your life, there is a reward in view from reading about these ideas. They can help organize your thoughts when seeking to implement the kind of rational self-examination almost universally recommended by philosophers.

2.3. CONSISTENCY

2.3

33

Consistency

Game theorists follow economists of the neoclassical persuasion in reducing rationality to consistency—and nothing more. They make the heroic assumption that the hard task of adjusting your preferences and beliefs until they become consistent has been completed before their analysis begins. What is denied altogether is that rationality can tell you a priori what you ought to want or what you ought to believe. As Hume put it: Reason is the slave of the passions. It is about means and not ends. As Hume extravagantly remarked: It is not contrary to reason to prefer the destruction of the whole world to the scratching of my finger.

The kind of rationality we are talking about is therefore a minimal concept. Assuming that rationality-as-consistency applies in some context doesn’t preclude the possibility that some wider notion of rationality might also apply (that might even include the Stoic virtues or Plato’s utopia as rational requirements if its proponents are sufficiently ingenious). It certainly doesn’t forbid the Epicurean idea that we should go along with our biology and seek pleasure and avoid pain, provided that we do so in a consistent way. Causal utility fallacy. Consistent choice behavior provides a new way of defining utility that differs from the happiness approach of Bentham and Mill, but supports their claims for utilitarianism much better. However, there is a price to pay that modern utilitarian philosophers don’t always appreciate. We can no longer reason that Alice prefers chocolate ice-cream to vanilla because she gets more utility from the former. This is sometimes called the causal utility fallacy in the new theory. The modern approach to utility doesn’t explain anything at all—it merely describes. It only says that Alice’s utility for chocolate should be chosen to be larger than her utility for vanilla because Alice is seen to choose chocolate ice-cream when she could have chosen vanilla. It therefore wouldn’t be much use if only applied to such binary comparisons. Large scale applications? What is coming next is a review of the consistency requirements on which the modern theory of utility is based. Its chief purpose is make it clear how innocuous they are. It really matters that the theory ought therefore to be regarded as uncontroversial, because without it we would have no scientific way of thinking about the big issues—like global warming or constitutional design. I am never sure whether its many critics genuinely don’t understand what they are criticizing, or whether they misrepresent the theory deliberately in promoting their prejudices. Either way, there is some small entertainment to be found in examining their views along the way. Reductionism? One line of criticism can be dismissed right away. Occam’s Razor tells us not to make models any more complicated than they need to be, and it is especially important to heed this good advice when giving examples. For

34

CHAPTER 2. RATIONALITY AS CONSISTENCY

this reason, casinos and horse races are standardly used as parables in talking about people’s beliefs. But it doesn’t follow that the theory only works in such parables. The big issues are put aside when explaining the theory, because otherwise the preconceptions that people usually bring to the table in discussing things like global warming get in the way. Holists condemn this attitude as reductionism, seemingly unaware that they thereby condemn science in general. But what did holists ever discover? As it says in the Book of Matthew: By their fruits ye shall know them.

2.3.1

Revealed Preference

The theory of revealed preference assumes that we know (or hypothesize) something of what Alice has chosen in the past, and uses this information to predict what she will choose in the future.7 The simplest model that allows such predictions assumes that we know what Alice would choose from all possible menus in some class of alternatives. For instance, one might take the alternatives to be flavors of ice-cream. I can then use as an example the time when I was first in America and found to my delight an ice-cream parlor offering a menu of 365 different flavors. My eventual choice was pistachio, which remains my favorite to this day. Consistency? If Alice is never indifferent between her choices, the consistency requirement for the theory of revealed preference is very simple. Consistency for this case is simply the following inoffensive requirement: If Alice sometimes chooses one flavor when a second flavor is available, then she never chooses the second flavor when the first flavor is available.8 When people are consistent in this sense, the alternatives they have chosen or would choose if given the opportunity can be ranked in order. We can then assign a number to each alternative so that higher ranked alternatives get bigger numbers. A number assigned to an alternative in this way is called its utility. If Alice’s choice behavior is stable, it can be used to predict her future choices: she will choose whichever available alternative has the highest utility. Leaving psychology to psychologists. This descriptive approach abandons the naive hypothesis that brains are machines for generating what Bentham and Mill called utility. On the contrary, the theory of revealed preference makes a virtue of 7 The theory was created by Paul Samuelson (another Nobel Laureate) for use in a market context. The simpler version to be described next works in all contexts, but is correspondingly weaker. I would prefer to call it the theory of attributed preference, because there is no need to assume that Alice has an unknown set of preferences that her choice behavior reveals. Alice might be a gene in an evolutionary game and so can’t have preferences at all (because genes are just molecules), but which behaves as though it had preferences because genes that didn’t consistently promote their own survival lost out to those that did. Only the bare bones of the theory are offered here, but my book Rational Decisions fills in the dull details for those who are comfortable with a little mathematics. 8 When indifferences are possible, the consistency criterion becomes: If Alice sometimes includes b in her choice set when a is feasible, then she never includes a in her choice set when b is feasible without including b as well.

2.3. CONSISTENCY

35

assuming nothing whatever about the psychological causes of our choice behavior. This doesn’t mean that the theory says that our choice behavior isn’t caused by what goes on in our heads: only that how it happens is a problem that can be left to psychologists (who can often nowadays be persuaded that the theory of revealed preference is a sufficiently weak kind of behaviorism that it needn’t be rejected out of hand). In particular, adopting the theory of revealed preference doesn’t entail abandoning Hume’s dictum that reason is the slave of the passions. Studies of brain-damaged people show that when our capacity for emotional response is impaired, we also lose our capacity to make coherent decisions. Nor is there any suggestion that we all make decisions in the same way. The theory of revealed preference accepts that some people are reckless, and others are cautious; that some care only about making lots of money, and others just want to stay out of jail. Economists say people are selfish? The theory doesn’t say that people are selfish, as the critics of neoclassical economics mischievously maintain—presumably because they think that utility is just a euphemism for money. It is true that everybody is assumed to maximize their own utility, but nothing says that they don’t get utility from helping the sick and the lame. How would the hard-boiled economists who give money to charity explain their own behavior to themselves if they thought only selfishness were rational? People are admittedly often selfish in real life, but the theory also has no difficulty in modeling even the kind of saintly folk who would sell the shirt off their back rather than see a baby cry. It does so by only taking account of what people do, and giving up any attempt to explain why they do it.

2.3.2

Accounting for Risk

It wouldn’t be surprising if you thought the preceding discussion to be a lot of fuss about very little, especially since it is obvious that it leaves room for many different ways of assigning utilities to the objects of choice. However, the next step adds more substance to the theory. Expected utility. Choices in real life are seldom as cut-and-dried as choosing between different flavors of ice-cream. We often have to choose without being certain what will happen next, as when choosing what bet to make at a Roulette table in Monte Carlo. Our choice then only determines the probability—the longrun frequency—of each possible outcome. For example, if you bet that the winning number will be odd and the wheel is fair, you will win half the time on average. So your probability of winning is one half.9 Utilities can be then be assigned to the outcomes so that an Alice who chooses consistently will act as though seeking to maximize her average utility. The standard jargon for the long-run average of something is its expectation. For a while, I shall 9 Or 9/19 in Las Vegas, where they take your money off the table if 0 or 00 comes up, so you only win 18 times out of 38.

36

CHAPTER 2. RATIONALITY AS CONSISTENCY

therefore say expected utility for the kind of utility we are talking about to emphasize that it isn’t the psychological utility of Bentham and Mill—which it seems can’t be emphasized enough. Working out an expected utility. My best flavor of ice-cream when faced with a menu of 365 flavors was pistachio. I can’t remember my worst flavor, perhaps rhubarb, which sounds seriously unpleasant. In between are all the other flavors like vanilla or chocolate. How are expected utilities assigned to all these alternatives? We are free to assign any utilities to Alice’s best and the worst outcomes as long as the best outcome gets a bigger utility than the worst outcome. Assuming Alice shares my preferences, let us therefore give rhubarb an expected utility of zero, and pistachio an expected utility of one. How does Alice then assign an expected utility to vanilla? We ask Alice to compare vanilla with bets in which she gets either pistachio or rhubarb with varying probabilities. As the probability of getting pistachio increases from 0 to 1, Alice will eventually find herself indifferent between the flavor vanilla and the bet between pistachio and rhubarb. The probability of pistachio at this indifference point is then Alice’s expected utility for vanilla.10 Consistency? What consistency assumptions are necessary? Alice’s choices must first honor the laws of probability. What this means in practice is that she mustn’t care that two bets look different if they both end up giving each outcome with the same probability. Secondly, she mustn’t care if one of the outcomes in a bet is replaced by another outcome if she is indifferent between the two outcomes. 11 Of course, most people don’t know what a probability is, let alone what the laws of probability are. Even if they did, they wouldn’t easily be able to recognize when two complicated bets are really the same. So it isn’t a surprise that the theory doesn’t work out very well in laboratories—although it works at least as well as alternative psychologically based theories when they are compared on equal terms. But philosophers won’t be walking into laboratories unprepared. We will have followed the advice of Epicurus by preparing ourselves in advance. Utility scales. It matters that an expected utility scale is essentially unique. It is like temperature in that you are only free to choose what values to assign to the freezing and boiling points on your scale. Our current version of expected utility assigns 0 to a worst outcome and 1 to a best outcome. But we could change our minds and assign 100 to the best outcome. Or we could assign 32 to the worst outcome and 212 to the best outcome. The 10 Von Neumann and Morgenstern utility introduced this idea in their 1944 book The Theory of Games and Economic Behavior. When they were writing this ground-breaking work, Morgenstern called on von Neumann to express his doubts about his assumption that players will act to maximize their long-run average payoff in a game. Unaware that the economics profession then thought it impossible to to defend such an approach, von Neumann sat down and wrote the theory currently being described before the afternoon was over! 11 Provided that the new outcome (which might itself be a bet) is independent of everything else. To emphasize this point, the second assumption is grandly called the Independence Axiom.

2.3. CONSISTENCY

37

new expected utility scales that result will then be related to each other in the same way that degrees Celsius are related to degrees Fahrenheit. We can even press the analogy further by inventing the term util to correspond to a degree on a temperature scale. The analogy with temperature is also useful when trying to compare how much utility two different people are getting. It makes no sense to say that your kitchen is hotter than your bedroom because its temperature is higher without first checking that the thermometers in the two rooms are using the same temperature scale. Similarly, it makes no sense to say that Alice is getting more utility than Bob without first finding some way to convert the utils on her expected utility scale into utils on his scale. But how are we to do this? The answer is put off until we need it when talking about utilitarianism.

2.3.3

Epicurean Reflection

Epicurus would probably have regarded social engineering as beneath the dignity of a gentleman. But I like to think that he would be pleased with the modern theory of revealed preference, which widens the potential scope of his hedonism to include anybody who behaves consistently—which includes anybody who succeeds in living their life according to rational principles. The task of delving into one’s own mind as philosophers almost universally recommend is notoriously difficult. David Hume is one of the few who have something to say on what he found there: basically a mess of fleeting thoughts coming from who-knows-where. My own experience is pretty much the same. We need the help of some organizing principle if we are somehow to reconcile the melange of conflicting impulses that block the road to tranquility. If we can somehow teach ourselves to overcome our internal conflicts, we will have learned to be consistent in our approach to the way we live: to manage our preferences and beliefs in a coherent way. Achieving consistency. We need to ask ourselves: What choice would I make if this contingency were to arise? What choice would I make if that contingency were to arise? What of all the other choices I might make under other contingencies that might be relevant to the choices I have provisionally made already? If all these hypothetical choices aren’t consistent with each other, we have to think again—and keep thinking again and again until consistency is achieved. It isn’t entirely trivial to achieve consistency when the world of relevant contingencies is restricted to menus of different flavors of ice-cream, but the world of all possible contingencies that might arise in Alice’s whole life is vast beyond all imagining. To get anywhere, she therefore needs to restrict the world of possible contingencies she needs to consider to something manageable—to a world that is small enough that the human mind is capable of dealing with it. Even if assisted by all the potential might of artificial intelligence, she would still be helpless unless the class of contingencies of which she elects to take account is somehow restricted to a small enough world.

38

CHAPTER 2. RATIONALITY AS CONSISTENCY

Here is where she needs the wisdom of those like Epicurus who seem to have successfully trod this road before. Of all the clutter in our minds that pushes us to make choices this way or that way, Epicurus tells us that what matters ultimately is how much pleasure or pain we genuinely derive from these choices. What worked for Epicurus will perhaps work for Alice too. If not, there are other sages who have plausibly claimed to have achieved tranquility in other ways. Comparability. Achieving consistency in our beliefs is the major problem, but it is worth pausing before examining how game theory handles the problem of beliefs to ask what simplifying lessons are to be learned from the theory of revealed (or attributed) preference. Of course, the objects for preference will no longer be flavors of ice-cream and the like, but the lives that we might live in the future depending on the decisions we make now and chance events in the world around us. This consideration further restricts what can be said now because I am leaving the problem of evaluating a whole life until the next chapter. However, one potentially useful observation follows immediately from the theory of revealed preference. If we succeed in achieving consistency in our own minds, then we will be able to judge matters in principle on an internal one-dimensional scale. That is to say, although our biological hardware doesn’t include a happiness meter, philosophical reflection that culminates in a state of internal consistency will create a software substitute: a one-dimensional index that measures our all-things-considered preferences. Philosophical concerns about various virtues being essentially incomparable can then be put aside.12 Making choices consistently allows expected utility to serve as an internal standard of comparison. For example, Epicurus tells us to seek pleasure and avoid pain. If we agree that it is a good idea not to fight our biology, we therefore have two dimensions to consider—and many more if we think it matters that there are different kinds of pleasure and pain. If we are learned in neuroscience, we might measure the brain states directly that correlate with our feelings of pleasure and pain: How much dopamine? How much adrenalin? What of all the other parameters that matter? If we can achieve internal consistency, we will have succeeded in somehow weighing all such factors against each other to produce a single expected utility scale. We may not agree with Epicurus that the pleasure factors should be treated as negligible compared with pain factors, but we will have found some way of trading them off, one against the other. Aristippus, for example, would treat pain tomorrow as negligible compared with pleasure today. So why not skip all this tiresome internal reflection—adjusting your hypothetical choices in a vast universe of possibilities until consistency has been achieved—and join the economics profession in simply asking yourself directly how you want to weigh one factor against another in your final expected utility index? 12 Is Mozart a better musician than Rodin is a sculptor? Traditional philosophers discuss such questions in moral philosophy a great deal at present. They certainly show that postulating a single absolute ethic requires squaring some difficult circles. However, if you choose to attend a Mozart concert instead of a Rodin exhibition, you are revealing a weak preference for the former over the latter—other things being equal.

2.4. REVEALED BELIEF

39

Small worlds. But a trap awaits the unwary, into which I believe the economics profession has fallen headlong—a trap that is even worse when preferences are replaced by beliefs. Are you entitled to assume that you would achieve consistency if you were to make the attempt? If your world of possible contingencies to be considered is large, you would most likely fail. If you then proceed as though you had succeeded by asking yourself what expected utility index suits you best, you will then be ranking consistency above the very considerations that would have driven you to a consistent outcome in the first place. This is why physicists live with the inconsistent theories of relativity and quantum physics. They could invent a (weak) consistent theory if they wanted, but they value consistency less than accuracy (within a given domain). We should at least be aware that if we follow the economics profession in prioritizing consistency, we are making a choice that physicists choose not to make. Only in a context in which we can somehow reduce the universe of relevant considerations to a manageable small world in which consistency is achievable does it make sense to assume that consistency can be achieved in advance of any analysis. However, in this plea for paying proper attention to the foundations of economic decision theory, I sometimes seem alone in the world. Fortunately, all these considerations are largely irrelevant in evolutionary applications (where game theory is empirically most successful), since evolution has sometimes had billions of years to force consistency in trying-to-survive on those genes that have survived.

2.4

Revealed Belief

Belief comes under the heading of experience in Hobbes’ breakdown of the nature of a human being. Hobbes gives himself away here as a thorough-going empiricist. He thinks that only through experience are we entitled to claim that we know or believe anything. However, the theory of revealed belief is neutral on such questions. Subjective probability. The theory of revealed belief says that if Alice chooses consistently in an uncertain environment, then she will choose as though maximizing her subjective expected utility. She may be at a horse race rather than a casino, but if she makes her choices consistently, she will still behave as though maximizing an expected utility index—but using probabilities conjured out of her own head when working out its average value rather than taken from the world around her . For example, if she bets on Punter’s Folly to win the Kentucky Derby when she could have bet on Gambler’s Ruin at better odds, then she behaves as though the probability that Punter’s Folly will win exceeds that of Gambler’s Ruin—even though neither horse might ever have run in a race before, and so no data at all is available with which to estimate their long-run frequencies of winning if the same race were run independently many times over under exactly the same conditions. Consistency implies subjective probability. What extra consistency requirement do we need to make the theory of revealed belief work? The inventor of

40

CHAPTER 2. RATIONALITY AS CONSISTENCY

the theory was Jimmy Savage.13 In his original version, the consistency requirement is called the sure-thing principle: If Alice chooses chocolate ice-cream rather than vanilla when the sun is shining, and also when the sun isn’t shining, then she will choose chocolate rather than vanilla when she doesn’t know whether the sun is shining or not. Who would have guessed in advance that teasing out the implications of such a seemingly innocent requirement would lead to the conclusion that Alice must operate as though equipped with subjective probabilities for all events that are relevant to her choice behavior?

2.4.1

Bayesianism

The Reverend Thomas Bayes was a nineteenth century clergyman who had no inkling of his future fame. He pointed out a way of working out conditional probabilities—a way of updating your prior probability of some event to a posterior probability after receiving some information. His method is called Bayes’ rule.14 For example, you are at home watching the Kentucky Derby on TV to see whether you have won any money by betting on Punter’s Folly. You thought your horse would win one time out of every three, and so your prior probability of winning was one third. But when you watched the finish, you only registered that the winning horse was brown. If Punter’s Folly is white, the posterior probability that it has won then becomes zero. But if Punter’s Folly is brown, the posterior probability becomes something larger than one third, because you no longer have to take account of all the horses that might have won if they hadn’t been the wrong color.15 Bayesian rationality? Bayesianism is the standard name for the principle that you should always maximize your subjective expected utility. Savage was led to this principle by assuming that Alice always chooses consistently, but Bayesianism pays no attention to the fact that consistency can be very difficult to achieve. It is called after Bayes because it is usually easiest to use his rule in working out how Alice will revise her subjective probabilities as she gets more information. Bayesianism was once the province of a small group of statisticians, but now it is a seriously hot topic with everybody climbing aboard: physicists, economists, psychologists—even journalists. Who would have thought that Bayesianism was capable of becoming a herd phenomenon? 13 The beginnings of what I am calling the theory of revealed belief anticipate the theory of revealed preference by quite a long time. It began with the brilliant Frank Ramsey in the 1930s, whose early death was a great tragedy. It was brought to its present state by Leonard (Jimmy) Savage in his 1954 book Foundations of Statistics. 14 Bayes’ rule is an entirely trivial consequence of the standard definition of a conditional probability, but people who call it a theorem commonly argue that to understand rational learning we need to know nothing more. Such naive folk don’t ask themselves why the definition of a conditional probability appropriate for objective probabilities is also appropriate for subjective probabilities. If they did, they would find that the small-world assumptions that Savage needed to make in order to justify using the same definition are seriously restrictive. 15 If half the original runners are brown, the probability that the winner is Punter’s Folly increases from one third to two thirds on learning that the winner is brown.

2.4. REVEALED BELIEF

41

Small worlds again. I am a Bayesian who doesn’t believe in Bayesianism, which leaves lots of room for confusion. I am a Bayesian because I think using Bayesian methods makes sense in the kind of environment for which they were designed. Savage went so far as to say that it would be preposterous and utterly ridiculous to use his newly invented theory outside such a small-world environment. But his strictures are almost universally ignored.16 Only in such a small world is it possible to achieve the consistency necessary to justify the theory (Section 2.3.3). I don’t believe in Bayesianism, which is the metaphysical doctrine that Bayesian methods are always valid everywhere, no matter how large the world in which they are applied.17 Its chief interest lies in its being a new branch of metaphysics with no ancient roots at all. It is simply wrong to announce that Alice is rational and therefore has no choice but to be the kind of naive Bayesian who believes that there is nothing more to rational learning than updating probabilities by Bayes’ rule. Rationality isn’t a magic wand that can be waved to banish all the difficulties of achieving consistency. Getting to rationality can be very hard work indeed. But Bayesian methods do make sense in small worlds, so let us use them in reflecting upon ourselves and our society when we can reduce the problems that beset us to a small enough compass that achieving consistency becomes feasible. Even piecemeal steps towards tranquility are better than nothing. Nor does anything prevent our using Bayesian models in large worlds—as long as it is understood that we do so only because no better model is available, and not because we imagine we are metaphysicians with a hotline to the truth.

2.4.2

Abusing the Theory

There are lies, damn lies, and statistics. The politician who said this certainly knew what he was talking about! And so do the journalists who spice up their articles by pretending to think that correlation is the same as causation. This subsection complains about the way expected utility theory is similarly abused by scientists who should know better. Its philosophical content is simply that care is necessary in choosing the model to which the theory is to be applied. Fitting instead of predicting. Carol can render any choice behavior whatever consistent by including enough parameters in her model. Almost any plausible model can be made to fit whatever data is available fairly well by including enough bells 16 I have tried showing naive Bayesians the page of Savage’s book (page 16) on which the strictures appear, but they mostly won’t look—rather like Galileo’s theological colleague who wouldn’t look through Galileo’s telescope lest he see the supposedly non-existent moons of Jupiter. 17 Everybody agrees that the probability the world seems designed given that God exists is bigger than the probability that the world seems designed given that God doesn’t exist. But what theologians who still defend the Argument by Design need is that the probability that God exists given that the world seems designed is bigger than the probability that God doesn’t exist given that the world seems designed. Does the latter proposition follow from the former? Applying Bayes rule, we find that we can deduce the required conclusion only when the prior probability that God exists is bigger than the prior probability that God doesn’t exist. Otherwise we are left in doubt. So Bayesianism even has an explanation of why religious folk are more ready to accept the Argument by Design than skeptics!

42

CHAPTER 2. RATIONALITY AS CONSISTENCY

and whistles—provided we can adjust their impact by turning their volume knobs up or down. It is a respectable activity if understood as merely a way of describing the data, but some researchers can’t resist going on as though their work provides an explanation of the data. But this isn’t science. If Carol wants to claim that her data-fitting exercise is explanatory, she must now use her model with the fitted parameters to predict some new independent data. What empirical support there is for the claim that real people are Bayesian is vulnerable under this heading. It is certainly possible to fit Bayesian models to suitable data given enough bells and whistles, but psychological experiments that test directly whether people update using Bayes’ rule are profoundly negative. This is hardly surprising, since ordinary people are notoriously bad at anything involving conditional probabilities. To maintain an enthusiasm for Bayesian models of real people, it is therefore necessary to overlook the experimental evidence that the Bayesian idol has feet of clay. But it bears repeating that it doesn’t follow that Bayesian models are useless—any more than the discovery that space isn’t Euclidean implies that Euclidean geometry should be thrown away. I think we should abandon the metaphysical pretence that only Bayesianism makes sense, and instead follow those Bayesian statisticians who defend their use of Bayesian models on the pragmatic grounds that they are better at describing data than any other model of comparable generality so far developed. Behavioral economics. There is an extreme faction of behavioral economists who get a lot of publicity by claiming to be refuting neoclassical economics when they fit unselfish utility indices to data obtained in laboratory experiments. But even if these unselfish utility indices were ever to succeed in predicting data from other independent experiments, neoclassical economics wouldn’t be refuted. On the contrary, such work would support the implicit neoclassical assumption that ordinary people can always be treated as though they are rational, and hence can be modeled as maximizers of some kind of expected utility index. The claim sometimes made that neoclassical economics is based on a “selfishness axiom” is a mischievous invention. But the behavioralists’ inadvertent defence of the implicit “neoclassical axiom” that people always maximize utility in laboratory games is perhaps more pernicious. Why is it so hard to accept that real people often behave irrationally when confronted with unfamiliar problems in a laboratory, or confuse them with real-life problems to which their resemblance is only superficial? With enough bells and whistles, you can fit a utility function to their behavior, but what is the point if each new data set requires a new utility function? Leaving things out. Problems can also arise when things that matter are left out of Alice’s choice model. We then have too few parameters instead of too many. Amartya Sen (yet another Nobel laureate) offers an example where Alice is an old lady about to accept an invitation to a tea party. But she changes her mind when the host observes that there will be an opportunity to snort cocaine. She is therefore inconsistent because the theory of revealed preference takes for granted that unchosen alternatives on her menu are otherwise irrelevant—and so we can’t

2.5. UTILITARIANISM

43

sensibly apply the theory to a model where they aren’t. (See Section 3.1.3 on Aesop’s fables.) However, the chief lesson of Sen’s example is that everything that matters must be included in whatever model we select from the many possibilities usually available. In particular, the old lady’s beliefs matter. She thought that the tea party was to be a respectable event and changed her mind when she discovered otherwise. So of course Sen was led astray when he left her beliefs out of his model.

2.4.3

Revealed Knowledge?

We don’t need to bother much about modeling knowledge, although game theorists often make a big fuss about what players need to know in order to play a game rationally. It is usual, for example, to say that the rationality of the players must be common knowledge,18 but evolutionary game theory wouldn’t work at all if animals had to satisfy such a requirement. To what extent can a spider be said to know even what game it is playing? Animals just do what they do—and people aren’t so very different much of the time. In any case, I avoid getting into epistemology by treating knowledge as knowledgeas-commitment rather than knowledge-as-certainty, or belief-with-probability-one, or (as is traditional in philosophy) justified-true-belief. It was knowledge as commitment that David Hume was talking about when he said that we are programmed to proceed as though the principle of scientific induction is true although it is impossible to justify. Knowledge-as-commitment requires that Alice—even if a spider—makes choices as though she were certain that various things are true whatever the evidence for or against may be. I could list consistency requirements for such a revealed-knowledge approach, but that would make more of the subject than it deserves.

2.5

Utilitarianism

Bentham and Mill’s utilitarianism says that, just as each individual naturally maximizes his or her own individual utility, so society should maximize the sum of all its citizens’ utilities. This statement treats utilitarianism as a device for shaping public policy and ignores the rival interpretation that regards it as a form of individual morality, in which Alice is told to ignore her own welfare in favor of the welfare of society as a whole—as in the collective choice fallacy of Chapter 4. It then becomes one of the more respectable attempts to persuade us that we should follow some invented notion of the absolute Good or Right rather than our own preferences. But to quote David Hume yet again: 18 Something is common knowledge between Alice and Bob if they both know it, and both know that they know it, and both know that they know that they know it, and so on. Game theorists may care to note that with knowledge-as-commitment the controversy over when common knowledge of rationality implies backward induction goes away (Ken Binmore, “Interpreting Knowledge in the Backward Induction Problem”, Episteme 8 (2011), 248-261).

44

CHAPTER 2. RATIONALITY AS CONSISTENCY What theory of morals can ever serve any useful purpose, unless it can show that all the duties it recommends are also the true interest of each individual?

Hume doesn’t mean by this that morals can never serve any useful purpose at all. They serve as a social device for selecting an equilibrium in real-life games that have many equilibria (Chapter 5). For example, when Epicurus observed that fairness norms exist to prevent conflict, he was saying that they do so by selecting a compromise equilibrium that the players both prefer to the alternative equilibria in the game they are playing when these involve the use of violent strategies.

2.5.1

Utilitarianism in Public Policy

What aims should Carol be required to follow if we hire her as a social engineer? In designing auctions, she may be told simply to maximize the resulting expected revenue in dollars. However, in reforming a tax code or designing a social benefit system, Carol will need to take account of the fact that an extra dollar is worth less to rich people than poor people, so something more sophisticated than dollars is needed to compare the welfare of different citizens. There are modern economists who follow Bentham and Mill in proposing indices of what makes people happy or sad, but it can’t be said that their approach inspires much confidence. In any case, for philosophical purposes, the first thing that comes to mind is to make Carol a utilitarian, using the scientific notion of expected utility instead of the dubious psychological notion of Bentham and Mill. To implement any findings of such an approach, one would need to find some observable correlate of expected utility, which might sometimes actually be dollars, or a happiness or healthiness index—or something else, depending on the context. But working with expected utility allows one to finesse all the difficulties that would arise from working directly with such a correlate while playing around with the conceptual issues. Social indices. The differences between people—whether they are rich or poor, healthy or sick, young or old—can then be taken into account by exploiting the fact that nothing says we have to proceed as though different people are using the same expected utility scales. For example, it may be that dollars are being used as a correlate for utils in some context where this makes sense, but if Alice is rich and Bob is poor, the utils on her expected utility scale will then be worth less than the utils on his expected utility scale. (Economists like to say that this is why rich folk take a cab when it rains but poor folk get wet.) A utilitarian Carol will then need to weight their utilities before adding them to accommodate this difference. Suppose there is a social consensus that 5 of Alice’s utils are worth the same as 9 of Bob’s utils (as would be the case if Alice’s utils were degrees Celsius and Bob’s utils were degrees Fahrenheit). We shall then say that their social indices are 5 and 9 respectively. Carol must then divide Alice’s utils by 5 and Bob’s utils by 9 before adding them. Or to say the same thing another way, Carol must weight Alice’s utils by one fifth and Bob’s utils by one ninth. So the weighted utilitarian sum of an outcome in which Alice gets 20 of her utils and Bob gets 27 of his utils is 7 (which is 15 × 20 + 19 × 27).

2.5. UTILITARIANISM

2.5.2

45

Why Add Utilities?

Why does utilitarianism add Alice and Bob’s weighted utilities? Why not combine them in some other way? If they are added, the implication is that all citizens are to be treated equally. But where do the equalizing weights come from? Why should 5 of Alice’s utils be worth the same as 9 of Bob’s utils? The answer is put off until Chapter 5. It is evaded here simply by assuming that Carol embodies an already established social consensus. Wealth has been used as a criterion in our examples so far, but it doesn’t follow that only wealth matters when comparing utilities. Any social consensus will reflect a whole kaleidoscope of factors determined by the history of the society in question. Is Alice fat or thin? Is Bob an illegal immigrant? Does he have a ring through his nose? Are they man and wife? Are they brother and sister? The list of possible cultural and contextual factors is obviously endless, but utilitarianism cuts through all the complexity by insisting that only the effect of such considerations on Alice and Bob’s social indices matters. This discussion of how utilities should be compared takes for granted that it is uncontroversial to compare Alice and Bob’s utilities. But who says that such comparisons make sense? Until recently, it was a dogma in neoclassical economics that such comparisons are meaningless. Bob may complain more than Alice in the dentist’s chair, but is he really suffering more? Bentham doesn’t evade this interpersonal comparison problem that later utilitarian philosophers largely ignore or brush aside as though it were of no importance: ’Tis vain to talk of adding quantities which after the addition will continue to be as distinct as they were before; one man’s happiness will never be another man’s happiness: a gain to one man is no gain to another: you might as well pretend to add 20 apples to 20 pears!

But, in typical style, he steamrollers on, leaving the problem for someone else to solve. John Stuart Mill isn’t that person; he ducks the issue altogether. The world had to wait for John Harsanyi to provide the firm foundations for utilitarianism that traditional philosophers continue to attribute to Mill. Why is Harsanyi’s work largely ignored by philosophers?19 A possible reason is that he used mathematics as a kind of shorthand, but his utilitarian ideas are so simple they can be explained without any equations at all, as in the discussion that follows. Empathy and sympathy. It is extraordinary that David Hume and his disciple Adam Smith seem to have been the first philosophers ever to have thought that empathy might be relevant to moral issues.20 19 John Harsanyi: Rational Behavior and Bargaining Equilibrium in Games and Social Situations. As a Jew in Hungary, Harsanyi twice contrived to avoid disaster. Having escaped the Nazi death camps by inches, he and his wife later fled into Austria from the Communists who followed. He then had to build his career again from scratch, starting with a factory job in Australia. As with many truly original minds, his talent was initially unrecognized. It took twenty-five years for economists to appreciate the ingenuity of his idea for dealing with incomplete information, for which he won a Nobel Prize in 1994. 20 David Hume: Treatise on Human Nature, Adam Smith: Theory of Moral Sentiments. My Natural Justice tries to bring their insights up to date.

46

CHAPTER 2. RATIONALITY AS CONSISTENCY

Alice empathizes with Bob when she imagines herself in his shoes to see things from his point of view—very useful when seeking to predict what Bob will do next. If successful, she will then understand what his personal preferences are, but she will not necessarily sympathize with him, and so feel his joys and sorrows sufficiently strongly that they motivate her to some degree as they motivate him. For example, Alice might be a confidence trickster seeking to find what story will persuade Bob to part with his life savings. The fact that we share genes with our relatives explains why evolution has programmed us to sympathize with them (Section 4.2.4). The more closely related they are to us, the more we are likely to sympathize with them. Members of street gangs or army platoons similarly sympathize with each other, perhaps because their close proximity in dangerous situations tells their bodies that they are brothers. Most of us get a warm glow from giving a small fraction of our income to charity (and occasionally to beggars in the street who make eye contact), presumably because our ancestors didn’t know anybody who wasn’t at least a distant relative. The unselfish behavior that results from such sympathetic feelings is easily modeled using an appropriate expected utility scale. There are utopians who proceed as though the sympathy we usually feel for our near-and-dear can also be relied upon to create friendly relationships with the world in general. Such utopians should perhaps read the newspapers—or even examine their own feelings when they misrepresent neoclassical economists as mean-minded, money-grubbing misfits. Saintly folk do indeed exist, but we call them saints because they are exceptional. Anybody who locks their car or counts their change knows that strangers aren’t to be trusted without good reason. Hobbes said what needs to be said much better: Certain living creatures, as Bees or Ants, live sociably one with another . . . and therefore some man may perhaps desire to know why Mankind cannot do the same. To which I answer . . . amongst these creatures, the common good differeth not from the private.

To summarize, Alice’s sympathetic preferences—however weak or strong they may be—are built into her personal expected utility scale. Even St Francis of Assisi would have acted as though maximizing his personal expected utility when he did his good works, if he did them consistently. But we can’t build a workable society on the assumption that everybody is like St Francis. Empathetic preferences. All this fuss about sympathetic preferences is to emphasize that empathetic preferences are something different. Carol expresses an empathetic preference when she says that she would rather be Alice eating an icecream than Bob stroking his cat. We say this kind of thing all the time. It is understood to mean that Carol has imagined herself first in Alice’s position, and then in Bob’s and prefers the former to the latter. But we can’t have deduced this from her individual choice behavior, because it is impossible that Carol could actually choose between being Alice or Bob. So why has evolution equipped us with the expensive mental capacity to feel empathetic preferences? What evolutionary advantage do they convey?

2.5. UTILITARIANISM

47

Fairness norms. Harsanyi’s answer is that we need them as inputs when using fairness norms. In his favorite example, Carol has an opera ticket that she can’t use and so plans to give it to either Alice or Bob. Since she would rather be Alice at the opera than Bob, she gives the ticket to Alice. My own favorite example anticipates the egalitarian alternative to utilitarianism to be discussed in Chapter 5. Carol is hired to arbitrate a fair settlement between Alice and Bob. Her solution to this arbitration problem is to choose whatever compromise makes her indifferent between being Alice or Bob. I find that people don’t think to question the fairness of such a compromise, but we shall shortly see that it isn’t always consistent with utilitarianism. However, the immediate point is that Carol can’t do without empathetic preferences in making fairness judgements. If one thinks that fairness matters, it is therefore also necessary to pay attention to empathetic preferences. One could even put together a theory of revealed empathetic preference that is based on Carol’s fairness decisions—provided she is consistent and we know what fairness criterion she is using. Does fairness matter? Behavioral economists hit an important nail on the head when they criticize traditional neoclassical economists for treating fairness as an irrelevance. Traditional textbooks offer various stratagems to explain why fairness doesn’t matter. My favorite defines social optimality as economic efficiency, which means that nobody can be made better off without making somebody else worse off.21 So why worry about fairness? Efficiency is already optimal! But it is efficient when some fat cat gets all the gravy, leaving none for anybody else. Do we really want to call this optimal? A more respectable trick claims that there is a necessary trade-off between equity and efficiency, to be settled in favor of efficiency. There is an argument for this that works in markets—but markets aren’t the only way to distribute commodities! Finally, it is said that interpersonal comparisons between different people are impossible, or even meaningless. Therefore fairness is a chimera. The only argument offered is that such comparisons aren’t justified by the theory of expected utility, as so far developed. But lots of other things aren’t justified by the theory either: for example, that hot air rises. In brief, traditional economists are nearly as good as traditional philosophers at inventing phony arguments that lead to conclusions that favor their prejudices. Adam Smith, the founder of modern economics, took fairness seriously, but nobody worries about fairness nowadays, so why should I? Jimmy Savage, the founder of Bayesian decision theory, thought his ideas only make sense in a small world, but so what? Jeremy Bentham, the founder of utilitarianism, thought that interpersonal comparison is a problem, but who cares? However, I shall continue to follow the wisdom of Epicurus, dating from long 21 This definition of efficiency is due to the Italian polymath Vilfredo Pareto (1848–1923). His support for the fascist regime of Mussolini has clouded his very substantial scientific contributions across a whole range of disciplines.

48

CHAPTER 2. RATIONALITY AS CONSISTENCY

before economics was born, and perhaps the economics profession will eventually learn the same lesson. Economists of the behavioral persuasion are certainly doing their best to persuade their more conservative fellows that fairness might really matter after all. How to make interpersonal utility comparisons. Harsanyi’s simple solution trivializes the interpersonal comparison problem. We first need that Alice and Bob’s personal preferences can be represented by an expected utility scale, and that the same is true of Carol’s empathetic preferences. Here Carol is to regarded as a personification of a community consensus. Next comes the heroic requirement. When Carol identifies with Alice or Bob, she does so with full success. For example, Carol herself may choose chocolate icecream over vanilla, but if Alice chooses vanilla over chocolate, then Carol accepts that, if she were Alice, then she would also choose vanilla over chocolate. 22 When identifying with Alice, Carol will therefore choose as Alice chooses, and so reveal exactly the same preferences. But there is no reason why they should both be using the same scale to measure their expected utilities. So when Carol is identifying with Alice, her empathetic utils needn’t be equal to Alice’s personal utils. Suppose that 5 of Alice’s personal utils and 9 of Bob’s personal utils each turn out to be worth the same as 1 of Carol’s empathetic utils. Then we know how to compare Alice and Bob’s expected utilities: 5 of Alice’s utils are worth the same as 9 of Bob’s utils. But this is all we need to weigh Alice’s personal utils against Bob’s—provided Carol is standing by to provide a standard of comparison. Conclusion. Some kind of empathetic preferences seem necessary for any reasonably sophisticated fairness norm to work, but part of the reason they often don’t work is surely that Harsanyi’s heroic assumption of total empathetic identification is too demanding. Some better theory is needed.23 But how are we going to get anywhere if we follow the herd by ignoring these issues altogether?

2.5.3

A Foundation for Utilitarianism

For what comes next we must imagine that Carol is what Adam Smith called an an impartial spectator, who is an imaginary philosopher king equipped with sublime wisdom in fair arbitration. Such remarkable folk from the golden age of the poets are more usually called ideal observers.They are a standard ploy in traditional attempts to get a handle on moral problems. With the help of an ideal observer, Harsanyi’s next argument is even easer than for interpersonal comparison. 22 Alexander the Great is an authority on this point! Here is an abbreviated report of a conversation in which he and his general Parmenio are discussing a peace treaty proposed by the Persians: Parmenio: If I were Alexander, I would accept this treaty. Alexander: If I were Parmenio, so would I! 23 Harsanyi’s theory of incomplete information can be used to put together a model in which empathetic identification is only partial, but my own view is that the Bayesian foundations of Harsanyi’s theory make such an approach impractical for applications to real people.

2.5. UTILITARIANISM

49

Without fear or favor, our incorruptible Carol chooses an outcome for the whole of society. In doing so, she is required to act as though she were equally likely to end up as any one of the citizens of the society on whose behalf she is acting. She then chooses whichever available outcome then maximizes her expected empathetic utility—which is simply a weighted sum of all the citizens’ personal utilities if Carol empathizes successfully with each type of citizen. But summing a weighted sum of everybody’s personal utility is what utilitarianism recommends. Of course, using an ideal observer to represent some postulated consensus on how different types of citizen are to be compared is seriously problematic, but Harsanyi has a second argument reviewed in Chapter 5 that can be made consistent with an evolutionary approach. Utilitarianism isn’t always egalitarian. Critics of the utilitarianism for which foundations have just been offered love the following fable of the cannibals and the missionaries. Ten missionaries are held at bay by cannibals and offered their freedom if they turn over one their number to be cooked and eaten. If suitable utilities are assigned to the various possibilities, utilitarianism will then sometimes recommend the sacrifice of one of the missionaries because his loss is outweighed by the gain of his companions. The point here is that utilitarianism can sometimes result in very unequal outcomes although we tell ourselves that we live in an egalitarian society. Even Jeremy Bentham was conflicted on this issue, sometimes writing as though utilitarianism will necessarily generate equal outcomes for all. I wonder how he would have responded to the story of the cannibals and missionaries? Status quo? Part of the reason that people react badly to the fate of the sacrificed missionary is that they don’t accept that he has a moral duty to honor the utilitarian norm. They ask what would happen if he refused to acquiesce in his sacrifice. But utilitarianism takes no account of the status quo—what would happen if an attempt to improve a society were to fail. It takes for granted that the utilitarian outcome will somehow be enforced. By contrast, the egalitarian norm to be discussed on Chapter 5 assumes that a status quo exists that will persist unless everybody agrees to an improvement. That is to say, reforms require mutual consent. We then have to worry—not only about how to compare the units on Alice and Bob’s utility scales—but where to place the zero on their utility scales as well. We deal with this problem here and elsewhere by simply assigning a utility of zero to the status quo on everybody’s scales. When we talk about how much utility an egalitarian outcome yields to Alice of Bob, we are therefore referring to their gains over the status quo. Sharing a sack of flour. A less dramatic example than the story of the cannibals and missionaries may help to illuminate how egalitarianism and utilitarianism differ (Section 5.3). Suppose Carol has to share a sack of flour between Alice and Bob that is worth 100 utils on Carol’s empathetic utility scale. The utilitarian solution requires that

50

CHAPTER 2. RATIONALITY AS CONSISTENCY

Carol should maximize her empathetic utility. But Carol’s empathetic utility is just a weighted sum of Alice and Bob’s personal utilities. So if Alice has a social index of 5 and Bob has a social index of 9, the utilitarian solution to the sharing problem is that Alice gets the whole sack of flour and Bob gets nothing. With the egalitarian norm, Carol assigns equal utilities to Alice and Bob on her empathetic scale. If the status quo is that neither Alice nor Bob get anything, Carol then splits the 100 utils at which she values the sack of flour equally between them. In terms of Alice and Bob’s personal utility scales, Bob then gets more than Alice because each of Carol’s utils is worth 9 of Bob’s utils and only 5 of Eve’s utils. If flour is an acceptable correlate for their personal utilities, each will then receive a share proportional to their social index. So Alice will get about 36% of the sack of flour and Bob will get the remaining 64%. However, Epicurus would have us also consider the possibility that Alice and Bob might fight over who gets the whole sack of flour if Carol weren’t available as an arbitrator. A fight would be bad for both, but perhaps Bob’s prospects of victory would be better than Alice’s. If so, the new asymmetric status quo favors Bob. So when Carol assigns equal gains over the new status quo to Alice and Bob in terms of her utils, Alice will get even less than the 36% of the sack of flour that she was assigned when the status quo was symmetric. All is therefore not sweetness and light with either utilitarianism or egalitarianism. Liberals tend to prefer utilitarianism because it ignores the current distribution of power and favors citizens with smaller social indices—too bad if some folk need to be sacrificed. Conservatives tend to prefer egalitarianism because it only permits reforms on the status quo in which everybody gains, and those with larger social indices gain more—too bad if the poor continue to suffer. The question left for Chapter 5 is: which does evolution favor?

2.6

Enforcement

The previous section made Carol—our mechanism designer—into a philosopher king. We instructed her to find the best available mechanism that satisfies an even-handed objective, and she obliged by designing a system along utilitarian principles using her own empathetic preferences in making interpersonal comparisons. How might this conceptual model be a useful guide in practice? Designing a tax code. Imagine that Carol has been been employed to reform a tax code. Economists say that Carol then has to solve a principal-agent problem in which she is the principal and the taxpayers are the agents. How will she reason if she proceeds along utilitarian lines? Carol might first argue that money will serve as adequate correlate for expected utility over the smallish range of differences citizens will face when comparing the reformed tax code with the code it is replacing. Next she will observe that rich and poor people will not have the same expected utility scales because an extra dollar is worth less to a rich person than to a poor person.

2.6. ENFORCEMENT

51

Carol will therefore assign different weights to different types of citizen depending on their wealth. Harsanyi’s theory says that she gets these weights from her empathetic preferences, but in reality they are determined by some political compromise that depends only in part on any consensus that society may have reached on what counts as fair (Chapter 5). However, it seems to be universally accepted that it would be unfair for rich people to be assigned bigger weights than poor people. Carol would therefore end up taxing income using a progressive system that isn’t so different from what happens in practice now. How much more the rich are supposed to pay per extra dollar of income than the poor varies from one country to another, but such relativism is in the nature of real social contracts. If we want to know what currently counts as fair in a particular society at a particular time, it is pointless to study the works of metaphysicians, where we are unlikely to find any answer at all. The absolute answers that they would give if they thought the problem worthy of their attention would be useless in any case. The real answers will vary with place and time. To find them out, it is necessary to do some serious sociological research (or to run an election in which fairness is a central issue). Designing a constitution. Suppose that Carol were hired not just to redesign a tax code, but to redesign a whole constitution—as Bentham offered to do for President James Madison in 1811, presumably along utilitarian lines. But the utilitarian creed attributed to Carol in designing a tax code is a firstbest criterion. It is sensible to use it when working for a government because the designer is then expected to proceed as if the problem of why the citizens of her society will honor the rules and regulations of her mechanism doesn’t exist. The putative reason is that the might of the state will be harnessed to this end; cheaters will be punished so heavily that nobody will cheat. Here theory and practice diverge quite a lot! The seriously rich contrive to pay essentially no tax at all. Even the tax returns of moral philosophers aren’t always to be trusted. However, the immediate point is that we have no almighty state standing by to enforce the laws that Carol might write into a constitution. It is sometimes argued that we can rely on the morality of our citizens to uphold the laws, but we have already rejected this optimistic application of the collective choice fallacy when discussing the interpretation of utilitarianism as a recipe for individual behavior. Carol will have to address the enforcement problem somehow. Guarding the guardians? Plato invented a class of Guardians for this purpose. Hitler had the Gestapo. Stalin had the KGB. The Thirty Tyrants of Athens had something similar in mind when they tried to recruit Socrates as one of their guardians. But any guardians will themselves be citizens (as will Plato’s philosopher kings). So they are also players in the game that Carol’s mechanism will create for the citizens to play. Carol will anticipate the equilibrium that will eventually be reached in this game, and use this equilibrium to evaluate her mechanism. As she is to act as a utilitarian, she will work out each citizen’s weighted expected utility at the equilibrium and add them up. Then she will choose whatever constitution maximizes this utilitarian

52

CHAPTER 2. RATIONALITY AS CONSISTENCY

valuation. The result will depend on all kinds of stuff, but the sort of police state idealized by Plato isn’t on the cards. If a large underclass ends up being exploited by a corrupt overclass, their misery will outweigh the doubtful bliss that their masters may enjoy. To avoid such a suboptimal outcome, a utilitarian Carol will have to find some way of keeping corruption to a minimum in her ideal state. The basic question is: Who guards the guardians?24 To which the game theory answer is that the game Carol creates when choosing a constitution must have a Nash equilibrium in which those citizens who take up the duties of a guardian find themselves guarding each other (Section 4.4). How can the guardians guard each other? The famous lobster fishery in Maine provides an example. When it became obvious that over-fishing and pollution were destroying the industry, the lobster fishermen got together and agreed to conservation rules that they then enforced on each other by giving cheaters a hard time— sometimes a very hard time. This is a small-scale model for a constitution that Carol would do well to study, but large-scale models that can’t avoid having specialized enforcement agencies are much more difficult. For example, who monitors the honesty of the regulators of a regulated industry? However, the immediate point is that paying attention to such examples undercuts Carol’s motivation for using a utilitarian criterion to choose between constitutions. Harsanyi’s metaphysical defence of utilitarianism in terms of an invented ideal observer becomes decidedly lame as soon as game theory appears on the scene. Somehow the criteria we use to evaluate what is fair must take account of the fact that we live in second-best societies that evolution (both biological and social) washed up on the beach—and not in some golden age of the poets. American constitution? The founding fathers of the American Republic weren’t sold on utilitarianism, but their debates nevertheless provide a good example of the type of reasoning Carol would need to follow in attempting to use mechanism design to plan a constitution. Imagine the dismay of Jefferson or Madison if they could see how their carefully considered system of checks and balances is being dismantled in favor of a populist nightmare. The unconstitutional right of the president to sometimes act like a king dates all the way back to Andrew Jackson. The supreme court has gradually been politicized. Nobody nowadays even pretends otherwise. Some congressional districts have been gerrymandered to the point of absurdity. Billionaires pay no tax. The election of the president has been essentially reduced to a referendum by a systematic emasculation of the electoral college. Who is going to guard the guardians in the kind of state our modern populists are creating? 24 The

phrase derives from a poem of the Roman satirist Juvenal:

Pone seram; cohibe: Sed quis custodiet ipsos custodes? Cauta est, et ab illis incipit uxor. The italicized phrase translates as: Who guards the guardians? Who they are guarding and why the guards might prove untrustworthy is best left untransated.

2.6. ENFORCEMENT

53

As Napoleon Bonaparte said in 1796 of the love of liberty espoused by the French Revolution he was about to foreclose: It is a wild dream, with which the French are infatuated, like so many before it. They must have glory, the satisfactions of vanity. But as for liberty, they understand nothing about it.

Looking at the mess the Athenians made of direct democracy, Epicurus might have said the same. But we have the example of the long fight for liberty in Britain leading up to the Glorious Revolution of 1688 and finally culminating in the American Constitution of 1787. Fraternity may be beyond our reach, but history shows that some measure of liberty and equality are not. Why throw away all that our ancestors fought for so long and so hard?

Chapter 3

Valuing Lives 3.1

Tree of Life

In a much-loved creation myth, Eve tempts Adam with the forbidden fruit of the tree of knowledge of good and evil. I wonder whether the tree of life mentioned elsewhere in the bible is the same tree? Either way, it is pleasing to apply the same term to the mundane notion of a game tree when talking about evaluating lives—not just the life of an individual, but of a society, or our species. Risk and time. Our immediate need is to separate our concerns about the risks we face in looking to the future from our attitude to the timing of the costs and benefits we will suffer or enjoy as our life advances. I shall argue that we do best to look forward from now when assessing risk, and backwards from our prospective death when assessing time. As the philosopher Kierkegaard observed: “Life can only be understood backwards; but it must be lived forwards.” Looking backwards from your death implies that there is then no forwards to assess—as Epicurus famously argued when denying that there is anything to fear about death. When we get to it, I shall therefore call the backward-looking methodology Epicurean to distinguish it from the forward-looking utilitarian orthodoxy. Global warming. I think we need to be seriously skeptical about our unthinking use of utilitarian methods in valuing possible futures—-especially in such vital areas as global warming (Section 3.2.2). In particular, it doesn’t make sense to worry about the possible extinction of the human race while simultaneously proceeding as though the well-being of our descendants is worth only a fraction of our own welfare. For this reason, I have discussed the Epicurean alternative much more carefully than I originally intended.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6_3

55

56

CHAPTER 3. VALUING LIVES

3.1.1

Game Tree

A game tree is used when we need to take account of time, which is nearly always for the games humans play in real life.1 Game trees are just like real trees except that they don’t have a trunk and have only one leaf at the end of each twig. The branches spring directly from its root and then divide and divide until we reach a leaf. When using the metaphor of a tree to talk about a life, we must imagine Eve climbing the tree all the way up from the root (where she is born) to a leaf at the top of the tree (where she dies). Her life is the path that she follows in climbing the tree. To go along with the bible story, I shall depart from the usual terminology and replace the leaves by apples. We can then envisage Eve imagining herself at the end of her life tasting the apple she finds there to gain knowledge of the good and evil she has experienced in the life she led in climbing the tree. Where the tree forks. A fork in the tree, where a branch splits into one or more subsidiary branches, is identified with a move in the game. This move is made by one of the players of the game, or by Chance. It determines which subsidiary branch will be taken next in an ascent of the tree. If Chance makes the move, we have to know the probabilities with which each subsidiary branch is chosen. Eve’s life—her path up the tree—is therefore not only determined by herself but by the vagaries of chance and by the choices of the other players. Chess. The shortest life for White in Chess is the Fool’s Mate sequence (two Chess moves translate into four moves in game theory): 1. Pawn–f3 2. Pawn–g4

Pawn–e5 Queen–h4 (checkmate)

We could solve Chess in principle by working backwards from all possible endgames, but this is impossible in practice because the game tree of Chess is complicated beyond all measure. The number of strategies in Chess is greater than the estimated number of electrons in the known universe! But Chess is absurdly simple compared with the real game that Eve plays every day while getting on with her life. Poker. More progress is possible with Poker, although the root of its game tree is a chance move corresponding to shuffling and dealing the cards. Things are further complicated by the fact that Eve never knows all the cards the other players are holding until after the game is over. 1 A game is specified in extensive form when its game tree is given explicitly. Theoretical biologists usually work with the strategic or normal form, in which time is abstracted away. The price of using such a simplifying model is that one can’t study phenomena that depend directly on the time structure of a game. Ambitious students of biology might therefore consider evolutionary behavior in extensive-form games as a thesis topic. Evolution then finds it easy, for example, to select Nash equilibria that aren’t ESS in the sense of Maynard Smith and Price. See the paper: “Suicidal punishment in the ant Acromyrmex Versicolor”, Evolutionary Ecology Research 14 (2012), 1–21, on which I am a junior author.

3.1. TREE OF LIFE

57

Von Neumann dealt with the second problem by inventing information sets, which list all the possible forks of the tree Eve might be at when she makes a move. (Only a genius would think that such a simple expedient might solve such an apparently enormous problem!) For example, after the deal in Poker, Eve will only know that the game so far has led to an information set in which her cards are what she sees in her hand. The information set will therefore list all the possible ways the rest of the cards could have been dealt to the other players.2

3.1.2

The End Justifies the Means?

Some light relief is provided by considering the charge sometimes made by moral philosophers that game theorists are naive consequentialists, and so can have nothing worth saying about moral norms. This reflects a debate between consequentialist philosophers like the utilitarians, and deontologists like Immanuel Kant. One might summarize the issue by saying that the former think the Good is fundamental—morality should prioritize ends—and the latter that the Right should take precedence—morality should focus on means by specifying what actions are permitted or forbidden. For example, Mill follows Plato’s Republic in arguing that citizens can be lied to “for their own good”, so endorsing the doubtful principle that the ends justify the means. For once, my sympathies are more with Kant, who naively argues that lying should be forbidden altogether. The reason that game theorists are thought to be consequentialists is that, when Eve gets to the top of the Chess tree, she is assumed only to care about whether the apple she then gets to taste tells her whether she won, lost or drew the game. In Poker she is assumed only to care about how much money she lost or won. So only consequences are taken to matter in such parlor games. After all, nobody is going to congratulate Eve on her brilliant strategy in Chess if she loses. But we don’t need to bundle the apples at the top of a game tree into simple categories like win or lose. Each apple can be considered uniquely, as is necessary when evaluating a life. Means and ends are then not separated at all. When tasting her apple, Eve will take account not only of its flavor but of the particular ascent of the tree that brought her to this particular apple, and not to some other apple that she would have reached if she had lived another life.

3.1.3

Separating out Eve’s Preferences

Epicurus would perhaps be impatient at all this talk of Chess and Poker. He would want to know how it might help a rational Eve choose how to live her life. How will Eve make an optimal choice of a strategy for climbing her tree of life, starting from wherever she finds herself now? Such a strategy will tell her which onward route to take at every fork in the tree.3 2 Von Neumann’s assumptions about information sets are essentially equivalent to those logicians attribute to their necessity operator. So they fit well with knowledge-as-commitment, in which things said to be known are impervious to correction. 3 She must take the same action at all forks within a particular information set, and so we can reduce the number of her strategies by only telling her what to do at each of her information sets.

58

CHAPTER 3. VALUING LIVES

Our immediate task is to separate Eve’s preferences over possible lives from all the other issues she faces when trying to plan her life in an uncertain world. We will then be able to address the problem of how she might value different lives without being distracted by her beliefs about what the future might hold. Aesop. Long before Epicurus was born, Aesop (620–564 BC) knew that rationality requires separating our preferences from our beliefs. For example, the kind of pessimism that makes Alice predict that it is sure to rain now that she is enjoying the sunshine is irrational.4 Aesop also knew that our perception of what is feasible mustn’t influence our preferences or beliefs—or vice versa. For example, in his fable of the fox and the grapes, the fox can’t reach the grapes and so alters his beliefs about whether the grapes are ripe. Aesop’s principles are all we really need to separate out Eve’s preferences, but without Bayesian rationality we wouldn’t be able to quantify these preferences. So we need to assume that Eve is Bayesian rational. Economists take this for granted. Anyone willing to do the same can skip forward to Section 3.1.4. Small worlds. Bayesian rationality—rationality-as-consistency—can’t really be taken for granted. Appealing directly to Bayesian rationality is a ramshackle expedient. It can only properly be defended in a small world. The best we can say in its defence when using it in large world is that better models aren’t usually available. In helping Eve to reflect on her own state of mind and plans for the future, we shall therefore assume that she adopts a model for her game of life that counts as a small world. Consistency will then be achievable, so justifying her acting as a Bayesian. She will then judge a life-strategy in terms of its expected utility, using subjective probabilities to quantify her uncertainties, updating them by Bayes’ rule as she becomes better informed in later life. Epicurus after Mytilene. To limit Eve to a small-world model isn’t a genuine restriction, because I doubt that any human being is capable of doing anything more sophisticated when assessing their future prospects. Even when Eve represents the human race in an analysis of global warming and the resources of the whole world are brought to bear, we won’t be able to do much better. Perhaps the science of artificial intelligence will eventually crack this problem, but I am not optimistic. In any case, I don’t think anyone alive today will be able to do any better than the type of internal colloquy attributed to Epicurus below after his violent expulsion from Mytilene. Epicurus will not only have asked himself questions, but questions conditional on the answers, and questions conditional on the answers to the conditional questions, and so on. A small sample of one fanciful line of enquiry inspired by the fate of 4 Voltaire was making fun of the optimism of religious folk when he observed that if God didn’t exist, it would be necessary to invent Him. Actually, psychologists find some degree of optimism is normal in our species. If you want an objective assessment of an issue, ask someone who is classified as mildly depressed.

3.1. TREE OF LIFE

59

Socrates is shown below. It shows how quickly things can get out of hand if one tries to cover too much ground. How would I respond if put on trial in Athens like Socrates? ◦ Follow Socrates in not treating the trial seriously. (1) ◦ Plead guilty and beg for mercy. ◦ Lie about my teaching and beliefs. (2) If (1), how would I respond if condemned? ◦ Drink the hemlock. (3) ◦ Flee abroad. If (2), what lies would I tell? ◦ That I follow the state religion of Athens. ◦ That I am a deist. If (3), how would I behave? ◦ Remain tranquil. Death is nothing to us. ◦ Rage against the dying of the light.

Each such question is a possible fork in the model of a tree of life that Epicurus is envisaged as building. The possible answers are the branches at that fork. Epicurus might perhaps have pursued large numbers of such lines of enquiry, asking himself how he would react if put in the position of this or that historical or contemporary person, paying particular attention to the possibility that events from his own past—like the disaster in Mytilene—might be repeated in the future. The collection of all Epicurus’s actual answers in the final tree (that takes account of all his lines of enquiry) is a strategy for playing his model of the game of life. This example is perhaps adequate to show that the kind of internal reflection necessary to build a model of the game of life is within anybody’s grasp, although some people will build much more elaborate models than others. But even those born with a talent for such reflection, like Epicurus or Socrates, will only be able to put together something more like a parable or a fable than the real game of life. Birth and death. When we imagined Epicurus constructing a highly simplified model of his tree of life in the preceding discussion, he was envisaged as reconsidering his life after what must have been a traumatic experience in Mytilene. The possible lives he was evaluating began from that time. He wouldn’t have liked the comparison, but he might be said to have been “born again” at this time, in the style popular with fundamentalist Christian sects. That is to say, we aren’t stuck with evaluating whole lives from our physical birth to our physical death. We can choose what counts as our “birth” and our “death”. For example, if Eve could ever achieve a sufficient degree of ataraxia that time ceases to be relevant, she might as well stop assessing her life at that point. Similar considerations are particularly important when Eve is taken to represent the human species contemplating its possible futures. We obviously want to start from now, but how far forward does it make sense to look? Not necessarily to the eventual extinction of our species, although some possible eventualities make this a distinct

60

CHAPTER 3. VALUING LIVES

possibility in the not-so-distant future. (If it is any comfort, what Epicurus has to say about the death of individual people also applies to the death of our species). In other possible futures in which our species survives our current woes, we only have to look forward far enough to a future in which we attain some kind of equilibrium in our relationship with our planet and each other. Normal science. A small-world model of the game of life won’t fit all the facts of the real world because Eve will only be able to construct a child’s toy of a tree. Her prior subjective probabilities are particularly likely to be wildly wrong. Even if Eve’s model of her game of life were more refined, things will still happen in the real world that can’t be accounted for by her model. Such discrepancies will come as no surprise, and so Eve will stick with her model as long as the discrepancies don’t get out of hand—a pattern of behavior called normal science by the philosopher Thomas Kuhn. What happens if the discrepancies do get out of hand? Then Eve will have to rework her model. She will be moved to do so anyway, even if things go relatively smoothly. Even in the little chunk of self-questioning attributed to Epicurus above, it is evident that all kinds of relevant questions and possible answers have been omitted.5 But Eve can’t spend her whole life thinking about how to live her life. As Aristotle said of happiness, tranquility will always evade those who seek it obsessively. Ataraxia must bring freedom, not only from the urge to follow the herd, but also from the urge to chase after your own tail.

3.1.4

Choosing a Strategy for Living your Life

How will Epicurean reflection help Eve in choosing a strategy for living her life? A lot depends on how she models other human beings. If people who matter to her life are likely to be as strategic as she is planning to be, then we are into serious game theory—where things get very technical once time enters the picture, and controversies about how best to model various problems are still unsettled. So I plan to leave this case aside, and focus on the case when other people aren’t strategic. It isn’t a bad assumption anyway. Few people are as reflective as game theory assumes. When game theory works in practice, it is usually only because evolutionary forces eventually make people behave as if they had thought everything out in situations that occur over and over again. Since the behavior of the other players is assumed not to be strategic, Eve can model them as (complicated) chance moves in her game of life.6 Eve will then be the only strategic player in her model of the game of life. Her problem is then reduced to finding the strategy that maximizes her expected utility in this model. 5 A very naive Bayesian might say that Eve should anticipate all questions at time zero and attach probabilities to all possible answers to be updated as she receives more information. But this proposal is absurdly impractical. 6 Immanuel Kant would have a fit to see her thereby treating other folk as means-to-her-end rather than ends-in-themselves. I am not sure how this view is reconciled with his being regarded as the archetypical deontologist.

3.2. WAYS OF VALUING A LIFE

61

Working backwards. Eve can proceed by backward induction, which is easiest to explain in the simple case in which she is never uncertain about which fork of the tree she has reached. At any final fork of the tree, she would then simply pluck the apple available there that she likes best. She can then imagine chopping off the branches at this final fork and replacing it by the apple she just plucked. Now she has a smaller tree to which she can do the same thing again—and so on until she is standing at the root of the tree with her most preferred apple in her hand. When uncertainties exist, everything is the same except that Eve must work with expected utilities of apples instead of the apples themselves. Separation achieved! We have at last reached our initial objective: to separate Eve’s problem in evaluating a life from the issues of uncertainty that always confuse discussions that fail to distinguish between Eve’s problem in choosing a strategy for living, and her problem in comparing what lives may be there to be lived. We can now focus without distraction on why she prefers one apple to another at the top of her tree of life—where each apple corresponds to the life Eve would actually have experienced as she climbed the tree and plucked that particular apple.

3.2

Ways of Valuing a Life

In evaluating lives, we adopt the viewpoint of Eve at her calm and rational best looking back over possible lives that she might live from the perspective of her dying moment in those lives. She will then be evaluating a life from the imaginary vantage point attributed to St Peter at the Pearly Gates. Was this a good life to have lived? Would it have been better or worse to have lived a different life? Uncertainty resolved. From St Peter’s vantage point, all the uncertainty that troubles our daily lives is resolved. Perhaps Eve lost much more in a particular life being studied than she could afford when playing Roulette on her thirty-second birthday. Perhaps she accepted Bob’s offer of marriage the next day. Perhaps Bob betrayed her with another woman the following year. But all these events that might have gone another way, went this particular way in this particular life. The putative lives she is looking back on from their end-points are therefore deterministic. Our separation of Eve’s preferences and beliefs allows all the questions of how lives come about to be put aside until after we have finished assessing what lives might conceivably be lived.

3.2.1

States of Mind

What matters in a life? Economists who have lost track of the foundations of their subject typically treat a life as an income stream—so much money lost or received each day. But followers of Epicurus are more sophisticated. We don’t think that money is the only thing that matters. We don’t care whether you own a fancy car, or if your house is bigger than Maisie’s down the road. Nor do we care how many letters you have after your name, or followers on Facebook. It is your attitude to

62

CHAPTER 3. VALUING LIVES

the world and yourself that counts. Eve may be poor in worldly goods, but rich in her heart if she doesn’t hanker after what she hasn’t got. A model that allows us to say such things treats a life as a stream of states of mind. You may be a billionaire, but what good is all your money if you are suffering from clinical depression? How does it help if your beloved children have gone to the bad? It is what is going on inside your head that matters—how you cope with both good times and bad. We need in any case to move to a model in which expected utility is attached to states of mind rather than physical objects to avoid beliefs intruding into our life-evaluations through a back door. If asked to choose between an umbrella and an ice-cream, most people would say that they prefer the umbrella on a rainy day and the ice-cream on a sunny day—thus making their choice contingent on the state of the world. But with states-of-the-mind introduced alongside states-of-the-world, we no longer need to complicate our thinking with such contingent preferences. We simply list four states of mind that specify how it feels to have an umbrella on a rainy day; or an umbrella on a sunny day; or an ice-cream on a rainy day; or an ice-cream on a sunny day. For what follows, we must also use the same technique to make sure that our formulation of a state of mind is itself free of contingencies. In particular, a state of mind must be described so that it can be evaluated independently of states of mind that precede or follow it. Otherwise, we would have to take account of when states of mind occur before asking ourselves how much we like them. Utility streams. The last point is important for the next step, which is to argue that we can assign expected utilities to states of mind in the style described in Chapter 2. We can then follow the financial experts in regarding a life as an income stream, but in which income is measured in utils rather than dollars. How is such an expected utility stream to be evaluated? The next section takes this problem on, starting with the utilitarian orthodoxy in economics. But first we need to pause to consider how the possibility of philosophical reflection alters the economic approach. Philosophical reflection. We have been led to measure a life in terms of a stream of expected utilities corresponding to states of mind rather than physical objects. However, there is a further reason for avoiding assessing a life in terms of physical objects. If we follow Epicurus in believing that philosophical reflection can alter our future states of mind, Eve needs to build this facility into her model of her tree of life. She may begin by putting together a model of her tree of life that doesn’t take account of the possibility that philosophical reflection can be life-changing. But she may then realize that if she is in certain states of mind at certain forks of the tree, she can create new branches of the tree by thinking certain thoughts. Along such new branches, states of mind could become possible that wouldn’t have been possible in her original tree. Epicurus would probably have said that he did indeed create such new branches

3.2. WAYS OF VALUING A LIFE

63

on his tree of life along some of which tranquil states of mind could be reached that were inaccessible in his original tree, and that we can all do the same if we examine our inner selves sufficiently closely. I have achieved some degree of quietude in following his advice, but I don’t have any idea how to model the possibility of using your mind to alter your tree of life except in very simple examples. Aside from the practicalities, there are always theoretical problems when trying to model a decision system—like the human brain—that has itself as a possible object of its decisions. So here is another reason why we have to make do with toy models that we know are inadequate to capture the enormous complexity of the real world.

3.2.2

Utilitarian Discounting

How do we evaluate a stream of expected utilities? The orthodoxy in economics is to value the stream as a weighted sum of the utilities in the stream. That is, we treat the succession of states of mind we are talking about as though they were the citizens in a society to be evaluated along utilitarian lines. The justifications offered for proceeding in this way with utility streams are so similar to the case for societies that they aren’t worth repeating. Discount factors. The only difference of any significance is that something substantial can be said about the weights if one is willing to accept that Eve shouldn’t change her mind about when she prefers payments to be made as she gets closer to the dates on which the payments are due. The weights can then be described using a discount factor, which might perhaps be 50%. What this means is that each weight is 50% of its predecessor. So the weights decrease in size very quickly. The first weight is one. The second weight is one half. The third weight is one quarter. The fourth weight is one eighth. And so on.7 So the discounted sum of the utility stream in which Eve gets 8 utils at the beginning of each of four years is 8 + 12 × 8 + 14 × 8 + 18 × 8 which comes to 15 utils. Interest rate. Economists are fond of the utilitarian approach to valuing utility streams because it mimics how financiers value a dollar income stream when the rate of interest is fixed. An interest rate of 25%, for example, corresponds to a discount factor of 80%.8 7 Such a fast rate of decrease is said to be exponential. How well does such exponential discounting do in predicting how real people handle income streams in laboratories? Not very well—presumably for the same reason that people find compound interest difficult. Various psychologically based alternatives said to be hyperbolic are claimed to do better. 8 Suppose Alice has $1,000 to fund some ongoing activity. She might set aside ten $100 bills to be made available now and at the beginning of the following nine years. But she could do better by putting each $100 bill in the bank to earn interest while it isn’t needed. If the interest rate is a very generous 25%, then her second $100 bill will earn an extra $25 before it needs to be spent. So she only needed to set aside $80 to have $100 available at the beginning of next year (because $80 plus 25% of itself is $100). So $100 to be paid in a year’s time is equivalent to 80% of $100 now. Alice’s discount factor is therefore 80%, which makes the present discounted value of an expenditure of $100 per year for ten years only about $440. Alice could therefore have spent $660 of her initial $1,000 on something else.

64

CHAPTER 3. VALUING LIVES

What determines the interest rate? The exact figure is determined by a market in which borrowers and lenders are the buyers and the sellers. However, the fundamental consideration is the level of risk involved in making a loan. If Alice lends money to Bob, she knows there is a chance the money will never be repaid. The greater this chance, the higher the interest rate, and so the lower the corresponding discount factor. But what determines Eve’s personal discount factor when she makes a utilitarian valuation of a utility stream? Global warming? This last question is brought into focus by the use of utilitarian discounting in evaluating studies of global warming. In such studies, Eve is replaced by Homo sapiens, but otherwise nothing is changed. The Stern Report of 2007 is an example.9 It was intended to take account of supposedly low-probability disasters neglected in previous studies, but its impact was blunted by a debate that arose among leading environmental economists over its choice of a social discount factor. Why was this particular discount factor chosen and not another? What is the correct social discount factor? It is significant that such a metaphysical question could even be asked. How could there be a “correct” social discount factor? Is it to be found in Plato’s realm of perfect forms, of which everything in the real world is a mere shadow? What an absurd thought! As in Section 2.5.3, we could adopt the unsatisfactory expedient of inventing a metaphysical ideal observer whose personal discount factor would serve as our social discount factor. Such an ideal observer supposedly represents a social consensus, but how can there be a social consensus on a discount factor when virtually nobody knows what a social discount factor is? In practice, the social discount factor in use is determined by a kind of intellectual market in which environmental gurus compete to get their personal preferences institutionalized. But such an way of proceeding isn’t remotely scientific. Why are we using a discount factor anyway? Don’t future generations count as much as us? Are our children and grandchildren somehow of lesser worth? Almost unbelievably, one response to such questions is that we need a discount factor so that we can guarantee that a potentially infinite series of weighted utilities will converge! Is our attitude to the potential extinction of the human race in the not-so-distant future to be made dependent on our mathematical convenience? What determines Eve’s personal discount factor? Two factors are usually mentioned, but without much exploration of their implications: Risk? The first factor reflects the fact that interest rates in the world of finance are determined in principle by the degree of risk in making a loan. So one interpretation of Eve’s discount factor is that it represents the risk that an opportunity in the future may not be there when the time comes for Eve to experience it. She might perhaps have died in the interim, or her bank may have gone bust, or some other disaster may have occurred. The discount factor is then simply the (supposedly constant) probability that no such event occurs in any time period. 9 Nick

Stern: The Economics of Climate Change: The Stern Review.

3.2. WAYS OF VALUING A LIFE

65

There are at least two problems with this interpretation of a personal discount factor. The first is that it offends against Aesop’s principle that preferences and beliefs should be separated. The second is that the approach developed in this chapter has already separated preferences and beliefs, so there is no room left for such an interpretation in terms of risk, even if it were beefed up to be more realistic. True time preference? The second factor thought to require a discount factor— usually offered alongside the first—is that we need to register how Eve feels about time clicking away on the clock. This seems reasonable at first sight because most people would rather be young and vital than old and feeble. But time is only an instrumental factor in such a preference. What matters is to be young and vital and not the time at which this happens. Even if there are people who like contemplating the time on a clock for its own sake, we don’t need to look beyond their states of mind to assess their preferences, because a person’s state of mind includes what they believe about the current time on a clock or the date on a calendar. There is then nothing left to be included in a discount factor. Comparing lives? Suppose that Alice lives for 90 years and Bob for 80 years. The utilitarian method of comparing their lives computes the life-time sum of their discounted utilities. Whoever has the greater sum is then said to have had the better life. Problems with interpersonal comparison of utilities are normally ignored, but we dealt with this problem in Chapter 2. The problem we shall worry about here is that this method of comparison assigns a utility of zero to Bob during the ten years when he is dead and Alice is alive. Such an attitude to death isn’t consistent with the foundations of utilitarianism offered in the previous chapter (Section 2.5.3). An ideal observer like Carol can’t empathize with a dead person, because a dead person—let alone an unborn person—isn’t there to be empathized with. What is worse, we are choosing to assign a special role to a state of mind that not only doesn’t exist, but which is unimaginable—because we can’t genuinely imagine a mind that is working and not working at the same time (Section 1.4.3). So we are saying that some states of mind feel better than being dead and others feel worse than being dead, without being able to say how it would feel to be dead because a corpse doesn’t feel anything. To assign a zero utility to being dead (or unborn), is therefore to introduce something new and unjustified.10 It isn’t consistent anyway with von Neumann’s analogy between an expected utility scale and a temperature scale (Section 2.3.2). It denies that we can choose the zero and the unit on our scale however we like. 10 This metaphysical misuse of zero tempts me to lighten things up with some metaphysics of my own. Some physicists think they can answer the most basic of philosophical questions: Why is there something and not nothing? When their answers are rejected by philosophers, they become frustrated: You say our definition of nothing isn’t any good, so what is your definition? But if you could define nothing, it wouldn’t be no thing.The definition would make it a thing. Like zero—zero on what scale? Like the empty set—empty of what?

66

CHAPTER 3. VALUING LIVES

The repugnant conclusion. At the very least, someone assigning zero to the unborn and the dead in a utilitarian analysis needs to explain how this can make sense, but the assumption is usually just taken for granted. For example, the philosopher Derek Parfit seeks to discredit utilitarianism by arguing that it leads to a repugnant conclusion: namely, that it doesn’t matter how low the utility of the citizens in a society may be if there are enough of them. Whatever society you propose, I can make a repugnant society whose sum of utilities is greater than yours by making its population size very large. A standard response is to say that the correct utilitarian criterion is to take the average utility in a society as one’s measure of its welfare (rather than the sum of utilities), but this ad hoc defence is no better than Parfit’s attack because the usual justifications of utilitarianism have nothing to say about population size. One could also respond that the citizens of a repugnant society wouldn’t like living in such a society any more than we would, so as we increase the population size, the attainable utility possible for citizens would decline fast enough to keep the optimal size of a utilitarian society within bounds. However, the reason for mentioning Parfit is to complain about his assumption that unborn citizens—citizens who have never existed and never will exist—should be assigned a utility of zero. This offends the Epicurean ethos because Epicureans regard the unborn as being no different from the dead. It offends the utilitarian ethos because of the unjustified special role assigned to zero.

3.3

Epicurean Valuation of a Life

When economists take account of time, only objective time is ever considered—the time shown on a clock, or the date when your monthly salary is due, or a loan is to be repaid. But we have just finished explaining why Eve is unlikely to care about the time shown on a clock for its own sake. What matters to her is subjective time—time as perceived in her own head. Subjective time. When finding your way around a new city, the way from one place to another commonly seems to take a long time at first, but then turns out to take very little time at all once it has become familiar. Psychologists attribute this phenomenon to our brains measuring time internally by the number of intervening incidents to which we need to pay attention. This is presumably why old folk say that time now rushes by, although a single day seemed to last forever when they were children. Subjective time is therefore not to be measured in terms of hours or years. It needs to be measured in terms of relevant incidents. In particular, if nothing distinguishes two adjacent objective time periods, Eve is likely to lump them together as part of the same subjective time period. That is to say, subjective time periods are elastic in their length when translated into objective time periods measured by the ticking of a clock.

3.3. EPICUREAN VALUATION OF A LIFE

67

Milnor’s model. This last observation about subjective time periods has powerful implications when fully exploited. We first dispense with discounting by treating all periods the same. Since states of mind have been defined so that they can be evaluated independently of when they occur, we can then reshuffle the order in which they appear without altering our attitude to a life. It then makes sense to demand that two lives be valued the same if two time periods with the same expected utility in one life are subsumed into one time period with the same expected utility in the other life—everything else being kept the same. The surprising conclusion is that any valuation of a life can then take account only of the best utility and the worst utility in the expected utility stream used to represent it. This is a strong conclusion—perhaps too strong. If neuroscience were able to tell us more about how we perceive time, we might possibly to able to relax the assumptions of what I shall call Milnor’s model to yield something more sophisticated.11 However, for want of anything better, the rest of this section follows up its implications. The immediate point is that there are ways of valuing a life other than the utilitarian approach that I have been criticizing. At least one advantage of the Epicurean approach can be identified right away, whether using the Milnor model or not. When looking back from the end-point of a life so that it can be evaluated as a whole, there are only a finite number of periods to consider, so no problems of convergence arise. Different highs and lows? How are we to compare two lives with different highs and lows? Milnor leaves this question open. Epicurus himself would seem to favor paying attention only to the worst utility in a life. Given a choice between lives, he would therefore have sought to maximize his minimum utility. Aristippus would have paid attention only to the best utility in a life, and so tried to maximize his maximum utility. Most of us would probably be more comfortable assessing a life using some weighted average of its best and worst utilities. Length of your life? One consequence of the Epicurean method of valuing a life is that the length of a life ceases to be directly relevant. Abraham Lincoln felt the same: “In the end, it’s not the years in your life that count. It’s the life in your years.” Or Victor Hugo, in more dramatic style: “It is nothing to die; it is frightful not to live.” Why do we care how long we live anyway? Some people want to delay their death because of their fear of being dead, but even if Eve can’t see that there is nothing to fear about being dead, she still has to ask herself whether an infinite future of being dead is made less by dying later. It is true that evolution has 11 The conclusion is obtained by reinterpreting a 1954 theorem of the famous mathematician John Milnor. (Milnor is best known for his book Differential Topology, and his friendship with the even more famous game theorist, John Nash.) My paper “Life and Death”, Economics and Philosophy 32 (2016), 75-97 adapts his work to the context of time, and explains his much rediscovered argument without any equations, but I am ashamed to admit that the paper fails to question the utilitarian orthodoxy of assigning a utility of zero to being dead or unborn.

68

CHAPTER 3. VALUING LIVES

programmed us to fear injury, but this is taken into account when Eve imagines looking back over a life and assigning low utilities to days when she suffers pain or distress—including the pain or distress she may experience in dying. In any case, the Epicurean attitude to death says not only that Eve has nothing to fear about the prospect of being dead, but also that she has nothing on which to congratulate herself on still being alive. So why disturb your tranquility by worrying about the fact that you are going to die? As the poet Swinburne put it: From too much love of living, From hope and fear set free, We thank with brief thanksgiving Whatever gods may be That no life lives for ever; That dead men rise up never; That even the weariest river Winds somewhere safe to sea.

Suicide? Albert Camus claimed at the beginning of his Myth of Sisyphus that the only important philosophical question is suicide.12 If you can’t be punished after your death, why not? I recently found the same question asked in an internet discussion on Epicurus, why not live it up, and then commit suicide? In fact, some young folk do choose this path—especially pop stars and the like—although their suicides are usually not cold-blooded, but consist of living so dangerously that an early death is almost inevitable. Journalists like to say that they flash across the sky like a meteor. Such young suicides create much unhappiness in those left behind, and so suicide is forbidden by the major religions and moral systems—but not universally. The Stoics favored suicide when honor was lost, as did the Samurai of medieval Japan, for whom a ritual suicide was mandatory in certain situations. Epicurus didn’t encourage suicide but nor did he condemn suicide outright. He certainly didn’t condemn those who choose suicide to escape unbearable pain, as he might have done himself. But nor would he have condemned the pop stars who flash like meteors across the sky. He would instead be saddened that they died as herd animals—that they didn’t think harder about what makes life worth living—that they thought poetry no better than push-pin—that they thought snorting cocaine better than ataraxia. As for those who deny an easy death to those for whom being alive is continuous suffering, I hope he would have the same opinion as me. At the very least, we should insist that the opponents of assisted dying put their cards on the table, and admit that they oppose suicide for religious reasons, instead of pretending that there are rational reasons why people should be made to suffer tortures like those Dante imagined for Epicurus in the sixth circle of Hell. 12 In an ancient myth, Sisyphus was condemned to roll a rock up a hill but never to complete the task, because the rock was fated to roll back again when nearly at the top. Camus compares the life of most people with that of Sysiphus. Why don’t they commit suicide? In a bleak reply to himself, Camus says that we must imagine Sisyphus happy.

3.3. EPICUREAN VALUATION OF A LIFE

69

Ataraxia. How would Epicurus have responded to Milnor’s model in which only the best and worst outcomes in a utility stream are relevant to its value? The only lives feasible for him are those that he could actually have lived by climbing his tree of life from wherever he chose to be born again. So Epicurus will face feasibility restrictions on the best and worst outcomes available to him. What preferences would he have over these feasible lives? We know that he thought avoiding pain more important than seeking pleasure, and so we can model him as preferring lives with better worst outcomes. But as we consider lives with better and better worst outcomes, the corresponding best outcome in a feasible life will eventually decrease until it equals the worst outcome. 13 If nothing blocks this process, all the utilities in the resulting utility stream will be the same, and hence this ideal life will be one of perfect tranquility—the ataraxic life. We were promised bliss as well, but perhaps bliss is indeed a natural consequence of organizing your thoughts so that time becomes irrelevant, as argued by the neuroscientists Newberg and d’Aquili (Section 1.5). Ignoring pleasure? Did Epicurus succeed in finding a way to use philosophical reflection to manipulate his future physical and mental states to make this ideal life available—if not thwarted by the enmity of the mob or some other irresistible force? Is a similar intellectual rebirth available to us? I am willing to believe he got a long way down this road, and perhaps we might be able to follow him at least some of the way. However, I remain doubtful that bliss automatically accompanies tranquility, and I can’t go along with the notion that a blissless tranquility by itself justifies neglecting all the innocent pleasures of life—and perhaps some not-so-innocent pleasures too. Epicurus himself certainly took pleasure in talking philosophy with his friends, and, although his tastes were doubtless ascetic, I wouldn’t be at all surprised if he were also sometimes naughtier than a saint ought to be. Changing your mind? Nothing says what Eve should count as birth and death in her model of a life. But suppose she changes her mind about this choice as her actual life proceeds? Won’t she then be inconsistent? It is true that Eve will then sometimes be inconsistent, but so what? The consistency championed in Chapter 2 is consistency in how Eve reasons. But reason is the slave of the passions, and we are talking about what Hume called the passions when we contemplate Eve changing her preferences. For example, it would be ridiculous to feel stuck with your childhood preferences for fear of being inconsistent over time. The whole point of philosophical reflection is to change how you feel about your life.

3.3.1

Global Warming

Nothing more was promised at the beginning of this book than that a scientific approach to philosophy would help us structure how we think about important 13 A similar argument appears in Section 5.3, where it is explained why the philosopher John Rawls is regarded as an egalitarian

70

CHAPTER 3. VALUING LIVES

problems—and what could be more important than the possible early extinction of the human race due to global warming. You may feel that the mountain heaved a great deal to bring forth such a tiny mouse, but the conclusion I want to emphasize is that the utilitarian orthodoxy of valuing a possible future using social discounting is badly flawed and likely to lead us wildly astray. The problem isn’t so much that we don’t know the “right” social discount factor—one could after all report the outcome of a study by listing the results with a whole range of social discount factors so that disputants could make their own choice from the list. The problem is that we shouldn’t be using a social discount factor at all. The alternative proposed here of replacing the utilitarian orthodoxy by an Epicurean approach in which one looks backward from the end of a life instead of forwards from its beginning is almost absurdly simple, but it avoids the problems that social discounting can’t escape. Milnor’s model—in which only the best and worst outcomes are taken as relevant—is admittedly crude, but it shows that there are viable alternatives to the utilitarian orthodoxy. In particular, every generation now and in the future can be treated the same. Implementing even Milnor’s crude model would create immense technical difficulties, but isn’t the potential reward worth the effort?

Chapter 4

Reciprocity 4.1

How to Get Along Together

Confucius (Kong Qiu) was asked by one of his followers for the secret of human sociality. His reply is usually translated as reciprocity. How do we succeed in getting along together much of the time? You scratch my back and I’ll scratch yours. This piece of wisdom has been repeatedly rediscovered down the ages, but we never seem to learn the lesson. This chapter tries again, using game theory to explain why reciprocity is so important to our societies, and why its cohesive power is lost when we interact through the internet instead of face-to-face. Iron laws? As always, we first need to clear away the crooked thinking that blocks the path to realistic assessments of how societies work. They aren’t held together by iron laws of morality without which they would collapse. They aren’t in need of moral philosophers to explain what kind of social glue holds them together. They don’t need any glue. If our societies had to be held together by the glue that writers of footnotes to Plato invent, they would have fallen apart long ago. Our societies hold themselves together like a masonry arch or a drystone wall, with each stone kept in place by its neighbors and reciprocating in turn by helping to hold its neighbors in place. They don’t need any help from the outside to survive because everybody is acting in their own self-interest when an equilibrium is in place.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6_4

71

72

CHAPTER 4. RECIPROCITY

Jekyll or Hyde? It was David Hume1 who first explained how societies hold themselves together without any glue (Section 1.4.5).The metaphor of a masonry arch is his. But not even Hume understood how far his own insight can take us. He thought that we still need to postulate some built-in benevolence in the human psyche to explain how we manage to get along together so well much of the time— that we may be mixtures of the sympathetic Dr Jekyll and the selfish Mr Hyde, but it is Dr Jekyll who is predominant (Section 2.5.2). However, the modern theory of repeated games shows that we can sometimes rely on the existence of Nash equilibria in which it is in everybody’s self-interest to cooperate to the maximum degree possible without the need for any benevolence at all. Even when Mr Hyde is predominant, we can still cooperate perfectly well because nothing but the selfish self-interest of the players is required. The following quote from Dostoyevsky’s autobiographical House of the Dead describing his experience as a political prisoner in a Czarist concentration camp shows that this observation isn’t just an idle piece of theory but can work in the real world too: The majority of these men were depraved and hopelessly corrupt. The scandals and the gossip never ceased; this was a hell, a dark night of the soul. But no-one dared to rebel against the endogenous and accepted rules of the prison; everyone submitted to them. To the prison came men who had gone too far . . . the terror of whole villages and towns . . . [but] the new convict . . . imperceptibly grew resigned and fitted in with the general tone.

Why doesn’t everybody know this already? Part of the reason is that mathematicians can’t get their minds around the notion that most people think a theorem is something to be shunned with horror. So the mathematicians who rediscovered Hume’s insights in the 1950s called their result a theorem—the folk theorem— although calling it a theorem makes it seem something grand and difficult, instead of an almost trivial consequence of the idea of a Nash equilibrium. However, the main reason is that traditional philosophers had already claimed the high ground by insisting that the kind of moral imperatives to be considered next are necessary for cooperation to be possible. They thereby confused the situation by failing to see—both that reciprocity is vital—and that it can only work in repeated situations. Their hopeless attempts to invent reasons why the cooperative behavior that they observed working in repeated games can also work in one-shot games remain an obstacle to progress that sometimes seems insuperable. 1 I have delayed a potted biography of David Hume (1711–1776) until reaching the point where his influence matters most. He was born the younger son of a decayed family of Scottish gentlefolk. So he had to make his own way in the world. He did so by living in penury while developing his philosophical ideas, which he thought would bring him wealth and fame when eventually published as the Treatise on Human Nature. But the book famously fell “dead-born from the press” as with most books by oddballs who do their own thinking for themselves. He rescued himself financially by writing a history of England which proved very popular. Eventually he became a philosophical hero of the Scottish Enlightenment. While working at the British embassy in Paris, he was lauded by the French intellectual elite, which must have been good for his self-esteem. But he was never offered a Chair in Scotland because of his suspected atheism. As a person, he was like Epicurus, good-humored and companiable. Like Epicurus again, it is sad that he seems not to have been interested in sex. It is said that he enjoyed a love affair with the flamboyant Countess de Bouffler in Paris, but their published letters don’t read like billets-doux to me.

4.2. SUPPOSE EVERYBODY DID THAT?

4.2

73

Suppose Everybody Did That?

The greatest philosopher of all time is commonly said to be Immanuel Kant. Following the herd often takes us to strange places, but seldom to such a monstrous misjudgement as this! We have already looked at his absurd attempt to deduce the existence of a metaphysical kind of free will from his claim that rationality compels us to honor his categorical imperative (Section 1.4.2). The immediate plan now is to look at his failure to give any serious argument in support of the claim that rationality demands that the categorical imperative be honored.

4.2.1

Categorical Imperative

This section looks at what passes for an argument for the categorical imperative in Kant’s Groundwork of the Metaphysic of Morals:2 1. Kant first asks why Nature has equipped us with reasoning powers. (Who is this Nature?) It can’t be to pursue our happiness (or presumably our evolutionary fitness) because it would have been better for this purpose if we were stimulus-response machines programmed with the optimal happiness response for all possible stimuli. (What of the impossible complexity of such a proposal?) 2. The true function of our reason must therefore be to “produce a will which is good, not as a means to some end, but in itself”. (He doesn’t say why.) 3. A good will is then said to entail a duty to “act out of reverence for the law”. (No reason is given.) The law here is nothing legal, but a transcendental notion that rational beings make for themselves. 4. Kant then says that no ingenuity is necessary to deduce what rationality now prescribes: in choosing what strategy to follow, rational beings need only ask themselves if it would be a good idea if everybody else were to follow it too. If not, throw it out. (No reasons are given.)

After emphasizing the difference between hypothetical and categorical imperatives, Kant then asserts that there is only one categorical imperative: Act only on the maxim through which you can at the same time will that it be a universal law.

One consequence of this claim is Kant’s insistence that it doesn’t count as moral if you help others because you want to—because you sympathize with your fellow 2 Kant’s Groundwork is standard reading for philosophy undergraduates. It is certainly a lot easier reading than his three Critiques—especially in the original German. I am told that even native Germans choose to read Kant in an English translation for this reason. I suspect that Kant is so respected partly because of the obscurity of his prose—so unlike the clarity of his rival David Hume, who is said to have awoken Kant from his philosophical slumbers. Kant (1724–1804) was apparently seriously obsessive. People are said to have set their watches from the time that he passed their houses in his daily walk through his home city of K¨ onigsberg (now Kaliningrad), which he never left. The one time he planned a trip elsewhere, he stopped the departing coach on the outskirts of the city and went home. Why do we let neurotics like Kant—and his hero Jean-Jacques Rousseau—tell us how to organize our lives, rather than better adjusted folk like Epicurus or Hume?

74

CHAPTER 4. RECIPROCITY

human beings. To be moral, you must do whatever you do because this is what the categorical imperative demands. How about some examples? We are told that suicide is absolutely forbidden, no matter how great your suffering. Also making false promises, no matter how pure your motives. Lying is never allowed—even to a mad axeman looking for his intended victim. This last is as crazy as it sounds. It is supposedly derived by showing that the maxim “Lie when it suits you” can’t satisfy the categorical imperative, because when made a universal law it leads to a contradiction. The reason given is that there would be no point in lying if nobody ever told the truth. But the result of making “Lying when it suits you” into a universal law wouldn’t be that nobody would ever tell the truth. It would only be that people sometimes don’t tell the truth—as in the world right now. The real reason people mostly tell the truth is much simpler. We explain it to our children with the story of the Boy who Cried Wolf. If you get a reputation for being a liar, nobody will believe you when you need them to. Shibboleth? One can’t really criticize Kant’s argument for the categorical imperative because even the loosest of arguments involves some connection between a set of premises and a conclusion, but Kant just offers a string of assertions conjured from nowhere. If you don’t believe me, check the text yourself; it is only nine pages long (pages 60–68 in Paton’s admirable translation). No wonder I have never found a mathematician who takes Kant seriously. No wonder undergraduates exposed to such pretentious drivel so often end up thinking of philosophy as just idle word play, with no genuine significance to how we live our lives. What of their teachers? As it happens, the theory of repeated games allows us—as an optional extra—to understand how they contrive to treat acceptance of Kantian metaphysics as a shibboleth: an arbitrary rite of passage necessary for those who want to be regarded as serious philosophers (Section 4.3.2). Kant and my mother. I think Kant really derived his categorical imperative from the same source as my mother, who was also keen on the idea, although she never heard of Immanuel Kant. When I was a naughty boy, she would rebuke me by saying, “Suppose everybody did that?” And I would secretly say to myself, “I’m not everybody, I’m just me.” I guess everybody’s mother gives their children the same sample of bad reasoning. There is even a cartoon strip in which Dilnot as a child silently protests the illogic of his cartoon mother in promoting the same error. Everybody is vulnerable. Even Spinoza succumbed when denying that treachery can ever be rational: What if a man could save himself from the present danger of death by treachery?. . . If reason should recommend that it would recommend it to all men.

4.2. SUPPOSE EVERYBODY DID THAT?

75

A hundred wise men have only one opinion. Not everybody follows Spinoza into the trap of thinking that it makes no sense to do something if it would be bad if everyone did it, as the following story from the court of Akbar the Great demonstrates.3 To test a proverb, he ordered a hundred sages to bring a pail of milk to fill a dried-up well, but each sage risked his displeasure by bringing instead a pail of water that cost him nothing. The well was then filled with water instead of milk. So each sage did the thing that would be bad if everybody did it. But Akbar wasn’t as displeased as expected, since the proverb was the heading of this paragraph. As this story from India indicates, it just isn’t true that something can’t be rational for Alice if the result of everybody doing it would be bad. Each sage reasoned correctly that his bucket of water would make no detectable difference to how much milk went down the well, so why spend money on milk? I think people are usually ready to accept the logic of the argument in this case, because it isn’t immoral for the sages to cheat Akbar, who has put them to the expense of buying milk merely to satisfy a whim. But the logic is the same whether the behavior is moral or immoral. It is simply wishful thinking to argue that rationality should always lead to what traditional philosophers say is moral behavior.

4.2.2

Prisoners’ Dilemma

Our next task is to explain how game theory drives home the unreason of Kant’s categorical imperative using its most famous toy game—the Prisoners’ Dilemma. Its multi-person version is called the Tragedy of the Commons because of its relevance to environmental issues. No wonder we make so little progress on cleaning up our planet when even the simple logic of this game isn’t understood by the movers and shakers of this unhappy world! Chicago times. The traditional story used to motivate the Prisoners’ Dilemma is somewhat tedious, but can’t be evaded. Alice and Bob are gangsters in the Chicago of the 1920s. The District Attorney knows that they are guilty of a major crime, but can’t convict either unless one of them confesses. He orders their arrest, and separately offers each the following deal: If you confess and your accomplice holds out, then you go free. If you hold out but your accomplice confesses, then you will be convicted and sentenced to the maximum term in jail. If both confess, you will each be convicted but the maximum sentence won’t be imposed. If both hold out, you will each be framed on a lesser tax evasion charge for which a conviction is certain.

The story becomes more telling if Alice and Bob have agreed to keep their mouths shut if ever put into such a situation. Should they break their promise by confessing? Or stand by their promise by holding out? 3 Akbar the Great (1542–1605) was the Mogul Emperor of India some years before Spinoza (1632–1677) was born. I have been fascinated by his antics ever since I visited his ruined palace in Agra to see the giant marble board on which he played the board game Parcheesi (Ludo in England) using beautiful maidens as pieces.

76

CHAPTER 4. RECIPROCITY

Kant tells us that they should hold out. Confessing is a bad idea if both do it. Being framed on a tax evasion charge is better than a long spell in jail. Game theory says the opposite. If Alice and Bob each seek to minimize their years in jail, each should confess. It is best for Alice to confess if Bob keeps his mouth shut, because she then goes free. It is best for Alice to confess if Bob confesses, because she thereby avoids the maximum sentence. So it is always best for Alice to confess—and the same goes for Bob. Game theorists say that confessing is a dominant strategy because it is best whatever the other player may do. Games usually don’t have dominant strategies, but when they do it is obviously rational to play them. Each player will then be making a best reply to the strategy chosen by the other, and so the outcome will be a Nash equilibrium. As we shall see, people don’t like the game theory analysis at all. They think that society would collapse if people reasoned like Alice and Bob—or the hundred sages in the Akbar story. They therefore raise all kinds of objections to how game theorists model the Chicago gangster story, and it is true that our toy model leaves out all kinds of things that would have mattered to Al Capone. So it will be as well to tell the story again in a more abstract setting so that all such objections can be seen to be irrelevant. Abstracting the Prisoners’ Dilemma. Alice and Bob are to play an abstract version of the Prisoners’ Dilemma in which each has two strategies, dove and hawk, between which each must choose without knowing the choice of the other. If you like, we can interpret dove as behaving in a socially approved manner—like keeping your word or honoring a promise. Hawk is the aggressive alternative.. But it doesn’t matter how we interpret the strategies. What matters is the assumption that if Alice knew that Bob had chosen dove she would prefer to choose hawk, and if Alice knew that Bob had chosen hawk she would also prefer to choose hawk. These preferences are assumed to take account of everything that matters to Alice and Bob, including their attitude to telling lies or breaking promises. What is the rational choice for Alice when she doesn’t know what choice Bob will make? Here we have the same issue for which the sure-thing principle was invented (Section 2.4). The answer is obviously that she should choose hawk. If the game looks the same to Bob, he should also choose hawk. Now assume that the result of both choosing hawk is a lot worse than the result of both choosing dove. Kant would then think it requires no ingenuity to see that rationality must entail choosing dove—and not only Kant. A whole generation of modern scholars fell into the same trap, calling the game theory analysis a paradox of rationality and seeking to find some way around it. The error made by such scholars is to imagine that the Prisoners’ Dilemma is adequate as a toy model of society. Game theorists think this is a huge mistake. The dice are as loaded against the emergence of cooperation in the Prisoners’ Dilemma as they could possibly be. If the human game of life were like the one-shot Prisoners’ Dilemma, we wouldn’t have evolved as a social species! When we come across situations that can be modeled as the Prisoners’ Dilemma– or its multi-player cousin called the Tragedy of the Commons—the lesson game

4.2. SUPPOSE EVERYBODY DID THAT?

77

theory has to teach is that we mustn’t rely on the usual way we manage cooperative ventures in ordinary life. We have to look for ways to change the game. And we desperately need to learn this lesson because looking after our planet is an example of the Tragedy of the Commons.

4.2.3

Fallacies

Some of the fallacies invented by scholars trying to escape the logic of the Prisoners’ Dilemma are reviewed below. People are nice. It is said that Prisoners’ Dilemmas are impossible because people are nice. Tell that to Akbar the Great! Tautologies don’t matter! This criticism applies particularly well to the abstract version of the Prisoners’ Dilemma given above. It is said that nobody needs to pay attention to the game theory analysis because it reduces the proposition to a tautology—something that can’t fail to be true because of the way things are defined. They presumably feel the same about the tautologies of mathematics, like 2 + 2 = 4. I wonder how they manage their tax returns. Collective rationality fallacy. Some critics are sold on the idea that rationality resides in groups rather than individuals. So they think that rational behavior on the part of an individual player lies merely in agreeing to whatever is best for the group of players as a whole. So it is supposedly rational to cooperate in the Prisoners’ Dilemma.The biological version of the mistake is called the group selection fallacy. 4 Philosophers who regard utilitarianism as a recipe for individual morality are victims of the collective choice fallacy (Section 2.6), but its most famous exponent is Karl Marx. He treated Capital and Labor as though they had the single-minded and enduring aims of individual people. Marx was wrong on this count because his coalitions don’t satisfy the consistency requirements that are needed to justify treating an entity as a player. They aren’t even sufficiently cohesive not to break apart if stressed. But sometimes coalitions do come close to satisfying the requirements of a rational decision-maker. After all, our bodies are just a coalition of the cells of which we are made, but they act together as a unit most of the time and so it makes sense to model them as an individual. On the other hand, it is sometimes necessary in evolutionary biology to go down a level, so that individual genes become the players (Section 4.2.4). Sometimes it makes sense to go up a level to governments, as in the Global Warming Game when China and the USA are treated as single players along with all the other nations of the world. 4 The crooked thinkers have triumphed here. To discredit writers like Richard Dawkins and his famous book The Selfish Gene, they have succeeded in convincing the world that the group selection fallacy isn’t a fallacy, because groups really matter in evolution. They are right that groups matter—of course they matter. Nobody ever said otherwise. But the group selection fallacy of the biologist Wynne-Edwards is that genes will be selected that favor the species as a whole rather than the gene itself.

78

CHAPTER 4. RECIPROCITY

There are good reasons why the cells of our body or the citizens of a society sometimes act in concert. The collective rationality fallacy consists of making the same assumption without feeling the need to give any reason why. External enforcement unnecessary? A common complaint is that our analysis of the Prisoners’ Dilemma ignores the existence of cooperative game theory. Such critics are unaware that cooperative game theory is about the contracts that would be written if players could make legally binding pre-play commitments on how to play a game. In the Prisoners’ Dilemma, they would then both agree to play dove. But without an external enforcement agency, neither player can afford to trust the other to honor any agreement they might make before the game is played. Put not thy trust in princes. People don’t like the Machiavellian suggestion that it may sometimes be rational to cheat and lie. Where would we be if we couldn’t trust our friends and neighbors? Why can’t game theorists see how much better the world would be if we all trusted each other? We could then use cooperative game theory without the need for a legal system to enforce our agreements. But game theorists don’t say that rational people should never trust each other. They only say that Alice needs a good reason to trust Bob. Only in the golden age of the poets would it be possible to take it on trust that strangers are trustworthy. As a dealer in curios put it in the New York Times of 29 August 1991 when asked whether he could rely on the honesty of the owner of the antique store that sold his finds on commission: “Sure I trust him. You know the ones to trust in this business. The ones who betray you, bye-bye.” Transparent disposition fallacy. The preceding objections aren’t serious. They are just ways of saying that some form of words should be found that allows game theory’s unwelcome analysis of the Prisoners’ Dilemma to be laid to rest somewhere out of sight. The transparent disposition fallacy and the fallacy of the twins coming next are more interesting, since they attempt a rival analysis. The transparent disposition fallacy5 asks us to believe two doubtful propositions. The first is that rational people have the willpower to commit themselves in advance to playing games in a particular way. The second is that other people can read our body language well enough to know when we are telling the truth.6 If we truthfully claim that we have made a commitment, we will therefore be believed. If these claims were true, our world would certainly be very different! Poker would be impossible to play. Rationality would be a defense against drug addiction. Politicians would be incorruptible. But the logic of game theory would still apply. 5 The philosopher David Gauthier—mentioned again in Chapter 5—was one of the inventors of the transparent disposition fallacy. However, I find that his article in Martin Peterson’s recent The Prisoner’s Dilemma abandons his own fallacy in favor of the collective choice fallacy. 6 Charles Darwin actually considered this popular claim in his Expression of the Emotions in Man and Animals. As he pointed out, actors can make themselves weep when they aren’t sad by imagining how it is to be sad.

4.2. SUPPOSE EVERYBODY DID THAT?

79

Imagine two possible mental dispositions called clint and john. The former commemorates the character played by Clint Eastwood in the spaghetti westerns. The latter derives from a hilarious movie I once saw in which I think it was John Wayne who played the part of Genghis Khan. To choose the disposition john is to announce that you have committed yourself to play hawk in the Prisoners’ Dilemma no matter what. To choose the disposition clint is to announce that you are committed to play dove in the Prisoners’ Dilemma if and only if your opponent announces the same commitment. Otherwise you play hawk. It is a Nash equilibrium for both Alice and Bob to choose clint in the Film Star Game, in which each must choose between clint and john. Each will then be making a best reply to the other. Their choices then play the Prisoners’ Dilemma, with the result that dove gets played on behalf of both Alice and Bob. But the Film Star Game isn’t the Prisoners’ Dilemma. The transparent disposition fallacy is typical of a whole strew of similar fallacies in which some other game is substituted for the Prisoners’ Dilemma, and its solution said to be a solution of the Prisoners’ Dilemma. How will we ever understand how reciprocity works if we get ourselves into a muddle on how it works in such simple cases as this? The disposition called clint is a failed attempt to introduce reciprocity into the one-shot Prisoners’ Dilemma, but we shall see that reciprocity actually does work in the repeated Prisoners’ Dilemma. We therefore need to distinguish between one-shot interaction and repeated interaction in the real world. The last thing we need to do is to invent arguments that confuse the two. Fallacy of the twins. This fallacy replaces the two-player Prisoners’ Dilemma with a one-player game in which Alice and Bob might as well be the same person. Alice reasons that Bob is her twin, since he is in the same position as herself. So he will do whatever she does. If she chooses dove, he will choose dove. If she chooses hawk, he will choose hawk. Of these two possibilities, she prefers that they both choose dove. Bob reasons the same, and so their expectations are realized in their both choosing dove. It is true that if there is a unique rational solution of a symmetric game, then a rational Alice and Bob will independently both choose the same strategy. But it doesn’t follow that whatever Alice chooses, Bob will choose the same. Alice will certainly think about playing irrational strategies to check that they are indeed irrational. But she won’t think that Bob’s thoughts move in lockstep with hers. His thoughts are independent of hers. Their end-product will be that he chooses the rational strategy because he is assumed to be rational. She should therefore choose whatever strategy is her best reply to his rational strategy. Since the same reasoning works for Bob, the result will be a Nash equilibrium—each will make a best reply to the other’s choice by playing hawk. Myth of the wasted vote. Two fallacies always appear at election time in Europe, where lots of parties usually compete. The fallacies are combined in the claim that “every vote counts” unless wasted on a minority party that “has no chance of winning”. But if a wasted vote is one that doesn’t affect who wins the

80

CHAPTER 4. RECIPROCITY

election, then all votes are wasted—except in the very unusual case when the winner and the runner-up turn out to be separated by only one vote. A version of the fallacy of the twins is brought to bear in reply to this objection. It is said that Alice is wrong to count only the impact that her own vote will have on the outcome of the election—she should count instead all the votes cast by people who think and feel as she thinks and feels, and hence will vote as she votes. If Alice has 1,000 such soul-mates or twins, her vote wouldn’t then necessarily be wasted because the probability that an election will be decided by a margin of 1,000 votes or less is often big enough to matter. This argument is wrong for the same reason that the twins fallacy fails in the Prisoners’ Dilemma. There may be large numbers of people who think and feel like Alice, but their decisions on whether to go out and vote won’t change if she stays home to wash her hair. Epicurus disdained real-life voting as a herd phenomenon. He would have explained that the myth of the wasted vote is only persuasive because people actively enjoy the prospect of having joined with others to vote for the winner. Game theorists like me vote for different reasons. We enjoy using our vote rationally. We ask what are the most likely ways that chance might lead other people to vote so as to equalize the votes for two of the candidates. We then vote assuming that one of these very unlikely events has occurred. Often one of the two parties in contention will be a minor party. The same gurus who tell us that every vote counts also tell us that a strategic vote for such a minor party is wasted, but they can’t be allowed to have it both ways!

4.2.4

Hamilton’s Rule

The fallacy of the twins would actually be correct if Alice and Bob were identical twins and their choice of strategy were genetically determined. This observation turns out to be the tip of a biological iceberg. Altruism? Skeptical traditionalists ask how altruism could conceivably evolve. Biologists have two answers. We leave discussion of the reciprocal altruism of Robert Trivers until Section 4.3. It is the work of Bill Hamilton on altruism within the family that is immediately relevant.7 His famous rule solves Hobbes’ puzzle of why the “common good differeth not from the private” among ants and bees. Most genetically determined behavior is controlled by a whole bundle of genes, so a mutation in one relevant gene will usually only modify the behavior somewhat. But to keep things simple, suppose a mutant gene appears that modifies playing 7 Hamilton died after contracting a tropical disease during an expedition to the Congo to investigate the origins of HIV. His Epicurean attitudes are captured by a statement he made about his anticipated death: “I will leave a sum in my last will for my body to be carried to Brazil and to these forests. It will be laid out in a manner secure against the possums and the vultures just as we make our chickens secure; and this great Coprophanaeus beetle will bury me. They will enter, will bury, will live on my flesh; and in the shape of their children and mine, I will escape death. No worm for me nor sordid fly, I will buzz in the dusk like a huge bumble bee. I will be many, buzz even as a swarm of motorbikes, be borne, body by flying body out into the Brazilian wilderness beneath the stars, lofted under those beautiful and un-fused elytra which we will all hold over our backs. So finally I too will shine like a violet ground beetle under a stone.”

4.2. SUPPOSE EVERYBODY DID THAT?

81

hawk in the Prisoners’ Dilemma to playing dove when it recognizes that it is playing a relative. When can such a mutant gene invade the population? Hamilton’s rule answers this question, not just for the Prisoners’ Dilemma, but for games in general. The mutant will succeed in invading the population if the benefit of playing dove times the degree of relationship between Alice and Bob exceeds the cost of playing dove.8 Degree of relationship. It is worth expanding on the degrees of relationship possible within human families, since this will matter in the next chapter when we come to consider the evolution of fairness norms (Section 5.2.2). The degree of relationship between two animals is the probability that a newly mutated gene in one animal is also present in the other. (One can’t simply talk about the probability that a gene in one animal is also in the other; most of their genes will be exactly the same. We share more than 99% of our DNA with chimpanzees.) Identical twins have all their genes in common, and so their degree of relationship is one. Siblings have a degree of relationship of one half, because a newly mutated gene will derive either from their mother or their father, who each have an equal chance of passing a particular gene to their offspring. Half-siblings have a degree of relationship of one quarter. Full cousins have a degree of relationship of one eighth. And so on for second-cousins-once-removed and the like. Utilitarianism within the family. Bill Hamilton is responsible not only for his rule, but also for the insight that identifying utility in a traditional game with fitness in an evolutionary game makes it almost a tautology that evolution will end up at a Nash equilibrium. Evolution acts to maximize the fitness of the players in an evolutionary game, and so they they will eventually behave as though rationally maximizing their utility in a traditional game.9 Fitness here is taken to be the expected number of extra offspring that each player will produce as a result of the way the game is played. The players are taken to be the genes that modify the choice of an animal’s behavior in the game. In a game played between relatives, a relevant gene won’t be confined to a single body. So when a gene is modeled as deciding what strategy to use in an animal it 8 It is the behavior of the workers in colonies of bees and ants that is crucial in determining how a colony copes with its problems. So it is their genetic relationship to which Hamilton’s rule needs to be applied. The workers are sterile but genetically female. It is therefore the relationship between sisters—and hence between workers and the queen—that matters. In the order Hymenoptera, which includes ants and bees, the degree of relationship between sisters is three quarters, compared with only one half in our species. So evolution had a better chance of creating unselfish behavior within the order Hymenoptera than within our own. 9 William Hamilton: The Narrow Roads of Geneland. I have delayed a tribute to Bill Hamilton until this point because it is his recognition of this point—seemingly obvious in retrospect—that bridged the gap between evolutionary biology and formal game theory. Hamilton is another oddball who worked alone and unacknowledged for many years, but his story has a happy ending, since he ended up as an academic superstar. His work on why some species have different sex ratios to ours is one of the most convincing arguments for the theory of evolution that I know. (Homo sapiens has approximately equal sex ratios because whichever sex is in shorter supply will have more opportunity to replicate their genes.)

82

CHAPTER 4. RECIPROCITY

occupies, its estimate of its fitness must take account not only of the fitness of the animal it occupies, but of all the other animals in whose bodies it might be lodged. The result of such a calculation of the inclusive fitness of a gene is quite pleasing. Translating fitness into utility terms, we find that the expected utility to be attributed to a gene in Alice’s body that controls her behavior in a game played with Bob and other relatives, is a weighted sum of the individual utilities of all the players—where the weights are Alice’s degrees of relationship with the other players. For example, if Bob is Alice’s brother, his utility gets a weight of one half. Perhaps this is why some people promote utilitarianism so strongly as a recipe for individual morality—rather than as a tool of public policy (Section 2.5). It works in families, so why not use it in whole societies? Because the degree of relationship among strangers is negligible! We must beware lest we fall into the same trap in the next chapter, when considering whether fairness norms that evolved for use in small-scale interactions can usefully be applied to large-scale problems. Not unless we first understand how they work in small-scale interactions! Who is my brother? Ants can detect their relatives by their smell. In a celebrated dirty T-shirt experiment, human women similarly found the T-shirts of genetically distant men more attractive than those with a closer relationship. But nobody would claim that humans can reliably tell an adopted daughter from a genetic daughter just by smell. We presumably learn who our relatives are by their role in our family dynamics. That boy must be my brother because he is always present at the dinner table making a silly fuss about nothing. That girl is my cousin because when we eat at her house, our mothers treat each other as sisters. This story would also explain the sympathetic relationships found among street gangs and army platoons. They treat each other like brothers because they are thrown together like a family. The traditional ceremony in which young men declare themselves blood brothers would then be more than a simple metaphor.

4.2.5

Free Rider Problem

Public goods are commodities like street lights that can’t be provided without everybody being able to consume them. No matter who pays, everybody has access to a public good. Some public goods are funded by voluntary subscription. Utopians argue that all public goods should be funded this way. Economists then worry about the free rider problem. For example, if people can choose whether or not to buy a ticket when riding on trains, will enough people pay up to cover the cost of running the system? Utopians shrug off this problem by arguing that people will see that it makes sense to pay, because otherwise the train service will cease to run. A Public Goods Game is a version of the Prisoners’ Dilemma invented to explore the free rider problem. Each player can contribute (play dove) or free ride (play hawk). In our example, the contributions are added and then doubled. The resulting sum of money is then returned in equal shares to all the players—including any free

4.2. SUPPOSE EVERYBODY DID THAT?

83

riders. The game is like the Prisoners’ Dilemma because it is a dominant strategy to free ride when there are three or more players, but it would be better for everybody if everybody were to contribute rather than everybody choosing to free ride.10 Experiments. The game theory analysis of the Prisoners’ Dilemma suggests that everybody will free ride, but what do real people do? The evidence from laboratories is no more optimistic than the history of real-life utopian schemes. They work fine for a while, but eventually collapse as people gradually become disillusioned with the behavior of some of their fellows, and start free riding. A huge number of laboratory experiments have been carried out on the Prisoners’ Dilemma and Public Goods Games in general. It is commonly said that real people usually cooperate in these experiments, but this isn’t true.11 Nearly half of all subjects (49%) play hawk, even when they are playing the Prisoners’ Dilemma for the very first time. When Public Goods Games are played repeatedly against new opponents each time, most subjects (90%) are contributing nothing after ten trials.

4.2.6

Tragedy of the Commons

A social dilemma in psychology is a multi-player version of the Prisoners’ Dilemma of which Public Goods Games are examples. But psychologists are more interested in real-life versions. Social dilemmas. How often have you waited seemingly forever in a long line while whoever is currently being served transacts their business in slow motion? What of the tiresome game played at airline carousels, when everybody edges forward so that it becomes necessary to stretch one’s neck in looking out for your bags? What of the more seriously inconvenient game played by people who bring oversize bags aboard claiming them to be hand luggage? You can probably list a whole raft of similar social dilemmas. When a social dilemma concerns environmental issues, it is usually called a Tragedy of the Commons after Garret Hardin’s example of a collection of peasants who graze their goats on common land. It is always in the interest of each individual peasant to put one more goat to graze but each goat thereby gets less grass to eat, and so each goat produces less milk and the common is eventually devastated. The Tragedy of the Commons captures the logic of a whole spectrum of environmental disasters that we have brought upon ourselves. Deserts expand because the pastoral peoples who live on their borders overgraze their marginal grasslands. Developed nations jam their roads with cars. We poison our rivers. We fell the rainforests. We have plundered our fisheries until some fish stocks have reached a level from which they may never recover. 10 The two-player case is just on the edge of being a Prisoners’ Dilemma. If Bob contributes, Alice ends up with the amount of the contribution whether she contributes or not. If Bob free rides, she gets nothing either way. 11 See John Ledyard’s survey article in Kagel and Roth’s: Handbook of Experimental Game Theory. Also David Sally’s independent “Conversation and Cooperation in Social Dilemmas: A Meta-Analysis of Experiments from 1958 to 1992”, Rationality and Society, 7 (1995), 58–92.

84

CHAPTER 4. RECIPROCITY

What is to be done about the Tragedy of the Commons? Nobody likes where the logic of the game theory argument leads, but it doesn’t help to insist that the logic must therefore be wrong. One might as well complain that arithmetic must be wrong because seven loaves and two fishes won’t feed a multitude. Social engineering? Game theorists prefer a more positive approach. When they are convinced that they have got the game right but don’t like the answer to which its analysis leads, they ask whether it may be possible to change the game. For example, in the original Tragedy of the Commons, a social engineer might confiscate all the milk produced by goats grazing on the common and redistribute it in equal shares to all the peasants. The peasants then have a common interest in maximizing the total amount of milk produced. Too few goats will produce too little milk, and so will too many goats. A Nash equilibrium of the new game will therefore produce just the right number of goats (which is something the peasants are likely to know a lot better than the social engineer).

4.3

Reciprocal Altruism

Moral philosophers say that we ought to be altruistic even to people we do not count among our near-and-dear. Deontologists say we ought to be altruistic because it is Right. Consequentialists say we ought to be altruistic because it is Good. They have been telling us this for thousands of years, but we remain what evolution made of us. Their approach is to be compared with the evolutionary origins of what the biologist Robert Trivers calls reciprocal altruism, which explains how people and animals can get along together without their needing to love each other. Reciprocal altruism works so well it can work when everybody is nasty. It is even said to be at work when unrelated vampire bats share blood. Biologists are totally sold on the idea that it also works with baboons and other primates. But we are primates too, and reciprocity works for us for the same reason it works for baboons. As Charles Darwin wrote in a notebook, “He who understands baboon would do more towards metaphysics than Locke.” When do nasties act nice? How are beneficial social norms to be sustained in a society whose citizens aren’t necessarily benevolent? No answer is to be found in looking at one-shot games like the Prisoners’ Dilemma—games that are played just once after which the players never interact again. Nowadays, we play such games in our cars and on the internet, but we didn’t play such games at all when we were evolving as social animals. For reciprocity to work, we need the future to matter so that favors and disfavors can be returned. The simplest case is when the same game is played over and over again without a definite end in sight. Why repeat the Prisoners’ Dilemma? It turns out that it doesn’t matter very much which game is repeated for the issues that need to be discussed, so we shall use the repeated Prisoners’ Dilemma to make the one-shot case as awkward as

4.3. RECIPROCAL ALTRUISM

85

possible. The aim is then to find Nash equilibria of the repeated game that result in fully cooperative behavior. This sounds as though it ought to be difficult, but is actually almost absurdly easy. A second more difficult aim is to protest against the rising tide of populism by replacing the crooked thinking of authoritarian gurus with a more reasonable explanation of how authority is really sustained in our societies.

4.3.1

Reciprocating Strategies

David Hume already understood how reciprocity works in 1739: I learn to do service to another, without bearing him any real kindness, because I foresee, that he will return my service in expectation of another of the same kind, and in order to maintain the same correspondence of good offices with me and others. And accordingly, after I have serv’d him and he is in possession of the advantage arising from my action, he is induc’d to perform his part, as foreseeing the consequences of his refusal.

I am most interested in the relevance of Hume’s insight to whole societies: to the question of what constitutions are viable in the long run. But it is helpful to consider first how it works on a very small scale. The hold-up problem. As a little boy, I can remember wondering why the shopkeeper handed over the candy when I had handed over my money. Why not just keep the money? Hume’s answer is that even dishonestly inclined shopkeepers honor the convention that they supply the candy after being payed because they would otherwise risk being punished. In my case, the shopkeeper would have lost more than my custom. I would have told all and sundry about my mistreatment. The damage to his reputation would then be out of all proportion to the small gain in cheating a little boy. How does this idea work in the indefinitely repeated Prisoners’ Dilemma? With only Alice and Bob, we can’t capture the effect that cheating by one or the other might have on the cheater’s reputation with third parties, but the punishment that can be administered by the cheated party alone is often enough to deter cheating. Repeated Prisoners’ Dilemma. Alice and Bob are to play the Prisoners’ Dilemma repeatedly. Each time they play, there is a small probability that this is the last time. To end up always cooperating in the repeated game, they can simply use the grim strategy: Play dove until your opponent deviates by playing hawk, after which switch permanently to hawk yourself. All problems with achieving cooperation in the one-shot Prisoners’ Dilemma then disappear, because it is a Nash equilibrium for Alice and Bob to play the grim strategy. Alice can’t gain overall by deviating from grim because her immediate gain will be swamped when Bob is triggered into playing hawk forever—which is very bad for Alice. Bob can’t gain by deviating for the same reason. They are therefore locked into a cooperative state from which nobody can gain by cheating. An observer who sees Alice and Bob cooperating all the time in this cosy fashion might be tempted to deduce that they have put aside the Machiavellian scheming

86

CHAPTER 4. RECIPROCITY

that game theorists supposedly promote in favor of an open-hearted policy of trust and good fellowship. But we are now looking at a case where there is no contradiction between what game theory recommends and the yearnings of utopians. It isn’t rational to cooperate in the one-shot Prisoners’ Dilemma, but two grim strategies constitute a Nash equilibrium for the indefinitely repeated Prisoners’ Dilemma and so cooperation becomes rational. Tit-for-tat. Axelrod’s Evolution of Cooperation tells the same story using the strategy tit-for-tat instead of grim, except that he didn’t express the argument in terms of Nash equilibria, of which he was unaware. The strategy tit-for-tat requires that a player begin by playing dove and continue by copying whatever action the opponent chose last time. If both Alice and Bob play tit-for-tat, the result will again be a Nash equilibrium in which both play dove each time the Prisoners’ Dilemma is repeated. It is unfortunate that Axelrod overplayed the virtues of the particular strategy tit-for-tat to such an extent that a whole generation of social scientists has grown up believing that tit-for-tat embodies everything that they need to know about how reciprocity works.12 As an example of the pitfalls that follow from swallowing this line, we have the claim that alliances in chimp societies can’t be explained by reciprocity because if one chimp turns on an ally, the injured ally will be in no position to retaliate. But other chimps will be watching. The chimp who betrayed his partner will then be shunned when he seeks new allies, for nobody wants to to be allied with someone who betrays his allies. Evolution and cooperation. Although Axelrod’s claims for the particular strategy tit-for-tat are overblown, his conclusion that evolution is likely to generate a cooperative outcome seems to be genuinely robust. We therefore don’t need to pretend that we are all Dr Jekylls in order to explain how we manage to get on with each other fairly well much of the time. Even a society of Mr Hydes can eventually learn to coordinate on an efficient equilibrium in an indefinitely repeated game.

4.3.2

The Reciprocity Principle

An infinite number of strategies share with grim and tit-for-tat the property of sustaining cooperation as a Nash equilibrium in the indefinitely repeated Prisoners’ Dilemma. In fact, the general result shows that reciprocity is applicable across the board in almost any situation where people interact repeatedly without a finite horizon in view: 12 Axelrod’s book might be a case study showing that neither experiments with humans nor computer simulations can be relied upon without serious robustness tests to check that biases haven’t been introduced into the framing of a problem. Robust evolutionary simulations actually generate strategies that only cooperate after trying to exploit their opponent. Nor is it true that tit-for-tat is evolutionarily stable (ESS). A population of tit-for-tats can’t even repel an invasion by mutants who always play dove no matter what. See my review of Axelrod’s later Complexity of Cooperation: http://jasss.soc.surrey.ac.uk/1/1/review1.html.

4.3. RECIPROCAL ALTRUISM

87

If the players care about the future sufficiently, then an approximation to any outcome in a one-shot game on which the players might agree if they could write enforceable contracts is available as a Nash equilibrium outcome in the indefinitely repeated game.

In repeated situations under ideal conditions, we can therefore do without external enforcement altogether and suffer no serious handicap as a consequence. What could be more important to understanding how human societies work? The application of the reciprocity principle in multi-player cases is indeed so significant for political philosophy, that I have put its statement in a box.13 Identifying deviants? When the indefinitely repeated Prisoners’ Dilemma was introduced, it was taken for granted that every time Alice and Bob chose between dove and hawk, they both knew how they had chosen in the past. With only two players, it would be odd to assume otherwise. But what if we were playing a repeated version of the Tragedy of the Commons? Would Alice necessarily know how many goats Bob was grazing on the common last week? The assumption that everybody’s history of play is known to everybody else makes good sense in the small bands of hunter-gatherers in which evolution taught us to be social. As in the small towns of today, presumably everybody knew everybody else’s business. But people often don’t even know who their neighbors are in modern city life. So in the anonymity of a big city, it isn’t possible to sustain the tight social contracts of small societies, because it isn’t possible to detect and punish deviants often enough to deter cheating. We do our best with policemen and tax inspectors, but everybody knows that our efforts in this direction are nowhere near adequate. The current problem with internet trolls is a relatively new example for which we haven’t yet come up with any workable answers at all as yet. A minimum requirement to be able to punish a deviant is that you know who the deviant is. Sometimes people punish anybody that resembles the deviant in some way—by race or clan or social class—but this is a seriously inefficient way of policing a social contract. We evolved as social animals in societies where people met face-to-face, and this is when reciprocity works best. Now a great deal of our contacts are essentially anonymous exchanges through the internet. We have 13 The larger the continuation probability after every repetition, the better the approximation. The restriction that only outcomes in the one-shot game on which the players might agree if they could write contracts excludes uninteresting possibilities in which some player would refuse because he or she can do better on their own. It is because of such tiresome details that mathematicians call the result a theorem, although its proof in the general case is no different in essence from the trivial argument given for the repeated Prisoners’ Dilemma. I don’t suppose that Bob Aumann (another Nobel laureate) or any of the other game theorists who discovered the folk theorem in the 1950s knew anything about David Hume. The biologist Robert Trivers was equally ignorant of Aumann’s work when he proposed the idea of reciprocal altruism in 1972. It wasn’t until the publication of Bob Axelrod’s Evolution of Cooperation in 1984 that the idea finally stopped being rediscovered—in much the same way that America ceased being discovered after the voyage of Columbus in 1492.

88

CHAPTER 4. RECIPROCITY

to learn that reciprocity doesn’t work in such an environment any better than the iron laws of morality we were taught in school. We can only exploit the reciprocity principle to the extent that we can identify and punish cheaters adequately. Emperor’s new clothes. Hans Christian Andersen’s fable of the emperor’s new clothes offers some amusement on the side. Some conmen assure the emperor that they will weave him magic clothes that are visible only to the pure of heart. When the emperor parades naked though the town to show off what he thinks are his new clothes, the crowds cry out in admiration at his magnificent attire. (Perhaps this reminds you of the admiration that traditional philosophers direct at Plato or Kant?) Nobody but an innocent child is willing to say that the emperor is naked, because everybody thinks that they would thereby reveal their impurity, and so become an object of social disapproval. The story is much loved, because it captures a truth that usually goes unrecognized. The reciprocity principle doesn’t just support outcomes that we all welcome, it can also support the kind of public lies and idiocies with which the media is packed. How do you get people to go along with some piece of foolishness? Use the grim strategy. Punish deviants if they don’t conform. Make fun of their folly in not agreeing with everyone else. Don’t listen to their explanations. Don’t hire them. Promote less able people over their heads. But things can get much worse. The story of the naked emperor is the classic exposition of how various bizarre memes have defended their existence throughout human history, but it fails to capture the horror of the millions of burnt and tortured bodies left in their wake. The urge to laugh at the ridiculous nature of the dogmas that have caused so much suffering somehow only makes it worse. Mild punishments. Talk of torture and burning at the stake is dramatic, but few punishments are so draconic in ordinary life. Indeed, nearly all punishments are administered without either the punishers or the victim being aware that a punishment has taken place. No stick is commonly brandished. The carrot is simply withdrawn a little. Greetings are less warm. Eyes look over your shoulder for someone more acceptable. These are warnings that you are in for more serious disapproval if you don’t mend your ways. We think of ourselves as sophisticated, but the higher stages of punishment used in the academic world to keep rogue thinkers in line are no different from the punishments anthropologists report as being used in hunter-gatherer societies. First, we all laugh at the rogue. If this doesn’t work, the next stage is boycotting. Nobody speaks to the rogue—or refers to their research. Finally, persistent offenders are expelled from the group—or can’t get their work published. The role of the emotions. Emotions were once dismissed as irrational urges left over from our evolutionary history. The socially aroused emotions associated with pride, envy, and anger are still counted among the seven deadly sins. But if these emotions are as self-destructive as tradition holds, how come evolution equipped us with them? I share the now widely held view that tradition is plain

4.4. SOCIAL CONTRACTS

89

wrong in seeing no useful social role for the emotions.They evolved to help police primeval social norms, and they remain useful to us for this purpose. For example, the prototypical scenario for the expression of anger arises when Bob treats Alice unfairly. She may then launch an angry attack on Bob. In doing so, she is unconsciously carrying out what would be her part of an equilibrium if she were playing an indefinitely repeated game with Bob. But what if the game she is playing with Bob isn’t repeated? Epicurus might respond that chimps can’t help falling victim to their emotions, but we aren’t chimps. He would say that we need to think harder about the primitive urge to retaliate that most people feel when treated badly. Evolution programmed us to feel rage when treated unfairly because it makes sense to punish such behavior in the repeated games that were the only games our ancestors played. But it doesn’t make sense to seek vengeance in one-shot cases. Bob may have treated Alice unfairly, but what does she gain by punishing him—or someone like him—if she will never see Bob again? Taking vengeance may relieve her feelings, but wouldn’t it be better not to let your feelings get out of hand in the first place?

4.4

Social Contracts

So far we have explored how the reciprocity principle works in everyday situations in real life. The plan is next to look at its large-scale implications for the social contracts of whole societies and its possible role in constitutional design.

4.4.1

Conventions

Some critics complain that the reciprocity principle allows too much. What use is a principle which says that any outcome whatever on which Alice and Bob might want to agree is available as an equilibrium when the situation is repeated indefinitely often? What we want is a prediction of which equilibrium we should expect to see in operation. Such criticism fails altogether to understand the function of the reciprocity principle. Its function isn’t to predict a particular outcome. Its purpose is to say what outcomes are possible in a self-policing society. If it didn’t allow a wide variety of possibilities, it wouldn’t be any use as part of a model of how societies work, because a wide variety of possibilities is what anthropologists actually observe in the world. To predict which equilibrium a particular society will use in a particular game, we need to know what conventions or social norms the society operates. Social contracts. A convention is a common understanding to operate a particular equilibrium in one of the many games that the citizens of a society play (Section 1.4.5). For example, there are two efficient equilibria in the Driving Game: Drive on the left or drive on the right. It is conventional in France and the USA to drive on the right. It is conventional in Japan and the UK to drive on the left. It is a pity that we are stuck with the term social contract for the collection of all such conventions in the myriad of subgames like the Driving Game that together

90

CHAPTER 4. RECIPROCITY

make up the game of life of a particular society. But it is far too late to propose that we use some more suitable term like social consensus for an idea whose whole point is that it operates without any contractual obligations in the absence of any external enforcement. A social contract is an assemblage of many different conventions. They range from the etiquette of fancy dinner parties to the significance of the paper money in our wallets. From the side of the road on which we drive to the meaning we attach to the words in the language we speak. From paying gambling debts to speaking the truth. From keeping up with the Kardashians to respecting other people’s property. From how much we tip in restaurants to honoring the authority of public officials. The seemingly profound half of each of these pairings is usually attributed to iron laws of morality, but Epicurus would say that the differences between the trivial and the profound halves in the pairings are differences only of degree. There are no iron laws beyond those encoded in our genes. We are bound only by a thousand gossamer threads woven from our own beliefs and opinions.

4.4.2

Constitutions

Some of our cultural conventions are codified as laws, but the legal system and the constitution of a modern society are relevant to a social contract only to the extent that they are actually honored in practice. If it is customary to give and take bribes, then bribery is part of the social contract, whatever the law may say. As for the authority of constitutions, all of us would be delighted to enjoy the civil rights supposedly guaranteed by the constitution of the old Soviet Union. As this and numerous other examples show, the brute fact is that a constitution is just words on a piece of paper. For a constitution to be more than a set of pious hopes, its coordinating function needs to be embedded in the social contract of a society—to be part of the equilibrium the society operates. Authority. There is a lot of crooked thinking about authority. Those in power have an interest in promoting the myth that they have a god-given right to rule. Or that their ancestry or their money makes them superior beings. But they are just human beings like you and me. How come they get to tell us what to do? As David Hume put it: Nothing appears more surprising to those who consider human affairs with a philosophical eye, than the ease with which the many are governed by the few, and the implicit submission with which men resign their own sentiments and passions to those of their rulers. When we inquire by what means this wonder is effected, we shall find that, as Force is always on the side of the governed, the governors have nothing to support them but opinion. It is therefore on opinion only that government is founded, and this maxim extends to the most despotic and most military governments as well as to the most free and most popular.

It is a bad mistake to allow the trumpery of the modern state—its constitution, its laws, its pomp and ceremony—to blind us to the fact that it all rests on nothing more

4.4. SOCIAL CONTRACTS

91

than the self-confirming beliefs of its citizens. Laws and directives, even if written on tablets of stone, are simply devices that help us coordinate on an equilibrium in the game of life. The fact that judges are instruments of the Law no more excludes them from our social contract than the Great Seal of State that symbolizes supreme power excludes a king or a president. Efficient coordination in a modern social contract demands that leading roles be assigned to certain individuals, but the power we lend to the mighty actually remains in our collective hands. Since the mighty can’t be trusted not to abuse their privileges, a liberal social contract makes effective provision for this collective power to be brought quickly to bear when corruption threatens the state. In short, the authority of public officials just a matter of convention and habit. Alice obeys the king because it is the custom to observe this norm—and the custom survives because the king will order Bob to punish Alice if she fails to obey. But why does Bob obey the order to punish Alice? In brief, who guards the guardians? Immanuel Kant absurdly thought that to answer this question would necessarily initiate an infinite regress unless the chains of responsibility end at some ultimate source of authority—in his case, the King of Prussia. But the reciprocity principle allows us to see that the chains of responsibility can be bent back on each other so that we can all guard each other. But for this purpose, we need to observe that the reciprocity principle can be strengthened so that its Nash equilibria become perfect. Perfect equilibria. A perfect equilibrium is a Nash equilibrium that remains a Nash equilibrium when attention is restricted to all subgames of the original game— whether or not these subgames will actually get to be played.14 The revised Judgement of Solomon of Section 2.2 will serve as an example. Suppose the false mother is the first to be asked whether she persists in claiming the baby. In a perfect equilibrium, she says no because she predicts that the true mother will say yes in the one-player subgame that would follow her saying yes. She therefore assumes that a Nash equilibrium would be played in a subgame that won’t actually be reached if she follows the perfect equilibrium in the whole game. Why punish? If a deviation forces us into a subgame off the equilibrium path, it remains optimal in a perfect equilibrium to stick to your original strategy provided everybody else does. In particular, if your strategy tells you to punish the deviant at some cost to yourself, it will be optimal for you to carry through on the punishment. If you deviate yourself by trying to escape your duty to punish, you will take us to yet another subgame where it is optimal for some other player to punish you for your dereliction of duty. If he fails to discharge his duty, we go to yet another subgame—and so on forever. With only a finite number of players, these chains of responsibility are necessarily closed in the manner that Kant didn’t consider. In the case of a social contract with a monarch, Alice obeys the king because 14 Reinhard Selten was the first to look at perfect equilibria. Some confusion is possible because he changed the definition offered in the text to something more general in his later work. When it is necessary to distinguish between the two definitions, the original definition is called subgame perfect and the later definition is called trembling-hand perfect.

92

CHAPTER 4. RECIPROCITY

she fears Bob will otherwise punish her. Bob would obey the order to punish Alice because he fears that Carol will otherwise punish him. Carol would obey the order to punish Bob because she fears that someone else will otherwise punish her. Since this someone else might be Alice, we are looking at a spiral of self-confirming beliefs, which may seem too fragile to support anything solid. It is true that the beliefs go round in a circle, but the reciprocity principle shows that their fragility is an illusion, since the behavior generated by the beliefs holds together as a perfect equilibrium. Guarding the guardians again. It is all very well to observe that popes, presidents, kings, judges, and the police derive what power they have from a social convention which says that ordinary citizens should accept their direction. It is true that if they were ignored in the same way that the citizens of Naples ignore traffic signals, they would be totally powerless. But it isn’t easy for people oppressed by despotic or corrupt officials to bring their power to bear. This is why guardians need to be guarded. This is why the Roman Republic introduced Tribunes to represent the plebs in 494 BC. This why bad King John of England was forced to sign the Magna Carta in 1215. This is why there was an English Civil War (1642–1651) and a Glorious Revolution of 1688. This is why there was an American War of Independence (1775–1783). This is even why the American Constitution of 1787 was amended to include the right to bear arms. More generally, this is why the constitutions of free peoples have built-in checks and balances. In an ideal world, all these provisions would hang neatly together in an equilibrium that we all operate together. But without the reciprocity principle it would all fall apart more or less overnight. Why have leaders at all? Anthropologists tell us that food was gathered and distributed in pure hunter-gatherer societies largely according to the Marxian principle that each contributes according to his ability and benefits according to his need. These small societies operated without any leaders at all. The urge to dominate by powerfully built individuals was held in check by the knowledge that they would face a coalition of opponents if they stepped out of line. In the foraging societies that survived into modern times, everybody was careful not even to seem to be telling others what to do in case they were punished for their presumption. So why do we need leaders in our big societies? Can it be that leaders are necessary because they know what is good for us better than we know ourselves? This may have been true of some Roman emperors like Marcus Aurelius, but authoritarian leaders like Joseph Stalin—once admired by naive westerners as Uncle Joe Stalin—are much more common. But even bad leaders can be useful as a coordinating device for solving the equilibrium selection problem in games for which the traditional methods are too slow or uncertain. On a sailing ship in a storm or in a nation at war, one can’t afford to wait for due process to generate a compromise acceptable to all. The purest examples come from primitive societies—as when Rome made Cincinnatus into a temporary dictator, or the native Americans of the great plains nominated a temporary war chief. What is the social mechanism that secured obedience

4.5. GLOBAL WARMING?

93

to such leaders during their period of tenure, but not before or after? Traditional theories of how authority works are remarkably threadbare. There is the divine right of kings. There is the Hobbesian social contract in which citizens supposedly trade their rights to self-determination for security. There is Rousseau’s argument purporting to prove it rational to subordinate one’s own desires to the General Will as perceived by those blessed with sublime wisdom. And so on. But the real reason that authority can work is that a leader’s role can be idealized as simply pointing to a perfect equilibrium of the game being played. If the convention is to honor the leader’s choice, then everyone will be optimizing when they do so. No individual will therefore have an incentive to deviate. The fact that it is only deviations by individuals that are suboptimal is important here, because leaders are unseated by coalitions whose members act together in destabilizing a social contract. As James Madison explained, a leader who chooses an equilibrium too far from what a society considers fair therefore risks creating a focal point around which an opposition can coalesce. Rival leaders then appear who appeal to the sense of fairness of their prospective followers.

4.5

Global Warming?

Evolution didn’t make us bees or ants. So there is no point in bleating that we are all members one of another—that no man is an island—that when the bell tolls, it tolls for thee. We can’t rely on people caring for each other to rescue us from global warming. If we could, we wouldn’t have got into the mess we have made of the world in the first place. A social engineer will urge us to change the Global Warming Game, as in Section 4.2.5, when all the milk produced in the Tragedy of the Commons was redistributed equally. But there is no way we can change the rules of the planetary game of life to align the interests of each nation, because no external enforcement agency is available to enforce whatever rules may be proposed to this end. The social engineer might more plausibly advocate our abandoning the current patchwork of independent nations that history has washed up on our evolutionary beach in favor of a world government. But the prospects for such a development seem gloomy to say the least. Even if a reincarnated and newly benevolent Akbar the Great were put in charge, people would still put water in their pails rather than milk when they could get away with it. Scientists may change the game with some major technical advance, but they haven’t got much time left. It looks as though we have to play the game that social evolution has stuck us with whether we like it or not. But if we can understand the techniques that we currently use to get along together within our separate societies, perhaps we can turn them to advantage on the larger stage. Game theorists say that our societies are held together by a vast complex of reciprocal understandings that keep the system in equilibrium most of the time. How they work has been the subject of this chapter. Might this insight help in getting us together to solve the global warming problem? We can but try, as no other remedy is in sight. But I am pessimistic, as we have

94

CHAPTER 4. RECIPROCITY

run out of time for the necessary reciprocal arrangements to get established. We can add our voices to those who cry out that it doesn’t look like there are going to be any quick fixes, and so whatever efforts are made have to made urgently. But the populists won’t listen. They will continue to regard the science of global warming as fake news. As for game theory, since it can’t be explained in a soundbite, it might as well not exist at all.

Chapter 5

Fairness 5.1

Evolutionary Ethics

Traditional philosophers don’t think evolution is relevant to morality. Sometimes they even warn against the heresy represented by evolutionary ethics. So where do they think moral principles come from? Theologians say they come from God. Various metaphysical substitutes for a God, such as Practical Reason, Moral Intuition, General Will, or Natural Law are popular with more secular philosophers. Immanuel Kant thought that he could deduce his categorical imperative from Rationality alone. According to him, all Rational Beings will therefore honor the same moral code.1 Followers of Epicurus have little patience with the kind of empty playing with words in such fairy stories. We don’t believe that morality somehow exists in some Platonic limbo independently of the physical world. We think that the moral norms found by anthropologists in all societies everywhere are products of the evolutionary history—both biological and social—of our species. They aren’t even unique to Homo sapiens. The sharing of blood among vampire bats is the most exotic example, but chimps even follow humans in operating norms that sometimes differ between different societies as a consequence of their differing social histories. Anthropology is helpful in sorting out what is universal in the human species from what differs from one society to another. It is surely no accident that Kalarahi bushmen, African pygmies, Andaman islanders, Greenland eskimos, Australian aborigines, Paraguayan indians, and Siberian nomads and other pure hunter-gatherer societies that survived into the twentieth century all operated social contracts in 1 Nobody can deny that Immanuel Kant was creative! In his youth, he was inspired by the scientific advances of Isaac Newton. His speculations about its astronomical implications are sometimes seriously insightful. With his Nebula Hypothesis, he was perhaps the first to suggest that stars and planets are created by the accretion of gas and dust through gravitational attraction. What a pity that he failed to see that evolution would also shape the behavior of the rational aliens that he conjectured inhabit the other planets of our system. If we ever encounter real aliens, they are unlikely to honor the same moral principles as Homo sapiens because their evolutionary history will differ from ours.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6_5

95

96

CHAPTER 5. FAIRNESS

which food, especially meat, was shared on a markedly egalitarian basis. But such small societies lived in a wide variety of diverse environments. What their social contracts have in common is therefore presumably biologically determined, leaving their many differences to be explained by social evolution. But even when such universal traits exist, they aren’t absolute—they might have been different if evolution had gifted us with a different set of genes. Sore thumbs. Traditionalists respond that evolution can’t be responsible for morality, because Nature is red in tooth and claw. How can the survival of the fittest be compatible with loving your neighbor? But only in the golden age of the poets can people be relied upon to emulate the Good Samaritan in loving strangers as they love themselves. Of course evolution won’t generate the kind of saintly behavior that traditionalists identify with morality. But we don’t have to accept their criteria for what counts as moral. Evolution didn’t shape our minds to operate the utopian fantasies invented by metaphysicians; it created the moral rules actually used by our ancestors in real life. We still use the same moral rules when solving everyday coordination problems today. The sort of coordination problems I have in mind are those that we commonly solve without thought or discussion: Who gives way when cars are maneuvering in heavy traffic? Who gets that parking space? Whose turn is it to wash the dishes? Who goes through that door first? How long does Bob get to speak before Alice gets a turn? Who moves how much when an elderly lady encounters a hip-hop boy in a narrow passage? What share should Alice appropriate of that strawberry pavlova? How much should Bob contribute toward Alice’s retirement present? These are trivial problems, but if conflict arose every time they needed to be solved, our societies would fall apart. When interacting with people from our own culture, we commonly solve such coordination problems so effortlessly that we don’t even think of them as problems. Our moral programming then runs well below the level of consciousness, like our internal routines for driving cars or tying shoelaces. As with Moli`ere’s Monsieur Jourdain, who was delighted to discover that he had been speaking prose all his life, we are moral in small-scale situations without knowing that we are moral. Just as we only take note of a thumb when it is sore, we tend to notice moral rules only when attempts are made to apply them in situations for which they are ill-adapted. However, it is precisely from such sore-thumb situations that I think traditional moralists unconsciously distil their ethical principles. We discuss these situations endlessly, because our failure to coordinate successfully brings them forcefully to our attention. Why did fairness norms evolve? The Epicurean answer is that they provide a convention that allowed us to coordinate on a compromise equilibrium in the Sharing Game played by our prehuman ancestors. Why did evolution care about sharing? Because sharing is a means of insuring against hunger. Why didn’t the big guys grab the lot? Because of reciprocal altruism in the repeated Sharing Game. We are by no means the only species that shares food, but each species has its

5.2. GOLDEN RULE

97

own way of sharing. Vampire bats have their way of sharing and we have ours. We call our method of sharing fairness, and use it these days for a lot more than simply sharing food. How do fairness norms work? Writers of footnotes to Plato think that science is helpless in the face of such questions, but this chapter offers a possible answer. It is doubtless inadequate or plain wrong in places, but its point is to make it clear that we needn’t heed the traditionalists when they tell us that evolutionary ethics is an absurd impossibility. The answer offered is very unlikely to be the final word on the subject, but it shows that evolutionary ethics is neither absurd nor impossible—merely unspectacular. As Epicurus was perhaps first to appreciate, justice isn’t some grand notion like the laws of physics; it is merely a device that evolved to solve equilibrium selection problems in our game of life. It is basically for this reason that such a view is unacceptable to traditional moral philosophers. They want justice to be a substitute for power rather than a means of balancing power. They are determined not to understand that it is the reciprocity principle that keeps our urge to dominate others in check, and not our respect for their invented moral principles. Fair social contracts? Fairness is a social device washed up on our evolutionary beach for use in solving small-scale coordination problems. Can we adapt it for use in constitutional design? I think it hopelessly utopian to think that we can redesign the constitutions of whole societies from ground zero, but there remain opportunities for social engineering on a lesser scale. The design of fair health-care systems is a case that cries out for such attention. But we won’t get anywhere in seeking to apply our current fairness norms to such large-scale problems if we don’t understand how they currently work in solving small-scale problems. So this is our first objective.

5.2

Golden Rule

The golden rule—do as you would be done by—seems to be a universal moral principle. A list of sages who endorse the golden rule would be endless. But here is what the superstars have to say: Zoroaster: That nature alone is good that refrains from doing to another what is not good for itself. Buddha: Hurt not others in ways that you yourself would find hurtful. Confucius: Do not unto others what you would not have them do unto you. Hillel: Do not to others that which you would not have others do to you. Jesus: Do to others whatever you would have them do to you. Mohammed: As you would have people do to you, do to them; and what you dislike to be done to you, don’t do to them.

The golden rule captures something of the reciprocity principle of the previous chapter, but it is also a fairness criterion. As such, it has been much criticized

98

CHAPTER 5. FAIRNESS

because it fails to recognize that different people like different things. For example, it may be that Alice likes being shaken awake before dawn for a cold shower and a ten-mile run. Bob prefers a gentle awakening at a late hour with a cup of coffee and a newspaper. So Alice doesn’t want Bob to do to her what he would like her to do for him. Bob would like it even less if Alice were to do to him what she would like him to do for her. Empathetic preferences. The sages who argued for the golden rule would probably have regarded such objections as nitpicking. They would perhaps have said that they were taking for granted that the golden rule needs to be modified when people have different tastes. The qualified version is simply: Do as you would be done by—if you were the person to whom something is to be done. If Alice wants to implement this qualified golden rule, she needs to be equipped with the empathetic preferences we met when discussing utilitarianism (Section 2.5.2). She not only needs to be able to put herself in Bob’s position to see things from his point of view—she needs to be able to compare how she feels herself with how she would feel if she were Bob. This seems so obvious in retrospect that it is hard to understand how moral philosophers could have overlooked the relevance of empathetic preferences for so long. Apart from anything else, speaking explicitly of empathetic preferences forces us to ask: How come people from the same culture are broadly in agreement on how to compare each other’s welfare, so that Alice and Bob can be modeled as having the same empathetic preferences? Chapter 2 dodged this question by allowing a metaphysical ideal observer called Carol to make fairness decisions on behalf of Alice and Bob. But we can’t allow fanciful notions from the golden age of the poets in an evolutionary approach. We have to dispense with Carol and try to understand how the social consensus that she supposedly represents might evolve. Equal or proportional? Dispensing with Carol creates some difficulties in language, because traditional discussions always correspond to adopting the viewpoint of an ideal observer. In our terms, the weights in a traditional utilitarian sum are then all equal because the utils to be summed are identified with Carol’s empathetic utils. Similarly, in traditional egalitarian discussions, Alice and Bob gain equally because the utils to be equalized are Carol’s empathetic utils. In her absence, we only have Alice and Bob’s personal utility scales to work with. It is then necessary to introduce weights into a utilitarian sum. In the egalitarian case, Alice and Bob’s utility gains are only proportional instead of equal (Section 2.5.2). To tackle the social consensus problem and other questions left hanging in the air by the traditional approach to the golden rule, we need to think harder about how the golden rule works in practice.

5.2. GOLDEN RULE

5.2.1

99

The Original Position

The original position was invented in the 1950s by John Rawls, who is justly regarded as the leading moral philosopher of the last century.2 He describes it as an operationalization of the categorical imperative. John Harsanyi proposed the same idea independently at about the same time. Both were Kantians interested in using the idea to characterize an ideally fair society. It is ironic that Harsanyi was thereby led to argue for utilitarianism, and Rawls—who didn’t approve of utilitarianism at all—to argue for a kind of egalitarianism. I see the original position as a device we use in ordinary life to implement the golden rule in the qualified form that takes account of the fact that Alice and Bob may have different tastes. So we need to put aside for the moment its possible role in designing a fair constitution. We shall just talk about Alice and Bob solving some kind of sharing problem. We also need to abandon the Kantian perspective of both Harsanyi and Rawls in favor of an evolutionary approach. It is true that most people find the idea of the original position intuitively attractive when they first hear of it, but this isn’t because we all have a hidden talent for metaphysical inquiry. I think the idea of the original position hits the spot for the mundane reason that we recognize a device that we unconsciously use whenever appealing to a fairness norm in ordinary life. Veil of ignorance. When a fairness judgement is to be made using the original position, Alice and Bob imagine themselves behind a veil of ignorance that conceals their identity. Behind the veil of ignorance, Alice thinks that she may turn out to be either Alice or Bob with equal probability, and the same goes for Bob. In this hypothetical state of ignorance, they imagine themselves bargaining about how to reach a compromise in whatever sharing problem they may face. One conclusion follows immediately. Whatever compromise they reach will be efficient as economists understand the term; Alice and Bob won’t agree on a wasteful inefficient deal because they could both get more by switching to a deal they both prefer. Those traditionalists who think that equity is so important that it should trump efficiency will therefore find no comfort in Rawls’ theory. People think that an efficient agreement reached in the original position will be fair because the distribution of advantage in whatever compromise Alice and Bob reach behind the veil of ignorance will seem determined as though by a lottery. Devil take the hindmost then becomes an unattractive principle, since you yourself might end up with the lottery ticket that assigns you to the rear. How might the original position have evolved? Why should it be thought to implement the golden rule? How does the golden rule relate to our previous discussion of utilitarian and egalitarian norms? These questions shape the agenda for what comes next. 2 John Rawls: Theory of Justice. I owe a debt of gratitude to Rawls for encouraging me to work on the ideas outlined in this chapter. If everybody were like John Rawls, we wouldn’t need to worry so much about how morality works.

100

5.2.2

CHAPTER 5. FAIRNESS

Deep Structure of Fairness

Recognition of the golden rule seems to be universal in human societies. Is there any reason why evolution should have written such a principle into our genes? Some equilibrium selection devices are obviously necessary for social life to be possible, but why should something like the golden rule have evolved? If the golden rule is understood as a simplified version of the device of the original position, I think an answer to this question can be found by asking why social animals evolved in the first place. This is generally thought to have been because food-sharing has survival value. Sharing food. The vampire bats mentioned earlier provide an example. Unless a vampire bat can feed every sixty hours or so, it is likely to die. The advantages of sharing food among vampire bats are therefore strong—so strong that evolution has taught even unrelated bats to share blood on a reciprocal basis. By sharing food, the bats are essentially insuring each other against hunger. Animals can’t write insurance contracts in the human manner, and even if they could, they would have no legal system to which to appeal if one animal were to hold up on their contractual obligation to the other. But the reciprocity principle tells us that evolution can get round the problem of external enforcement if the animals interact together on a repeated basis. By coordinating on a suitable equilibrium in their repeated game of life, two animals who are able to monitor each other’s behavior sufficiently closely can achieve whatever could be achieved by negotiating a legally binding insurance contract. It will be easier for evolution to find its way to such an equilibrium if the animals are related, but the case of vampire bats shows that kinship isn’t necessary if the evolutionary pressures are sufficiently strong. Insuring against hunger. What considerations would Alice and Bob need to take into account when negotiating a similar mutual insurance pact? Imagine a time before cooperative hunting had evolved, in which Alice and Bob foraged separately for food. Like vampire bats, they would sometimes come home lucky and sometimes unlucky. An insurance pact between them would specify how to share the available food on days when one was lucky and the other unlucky. If Alice and Bob were rational players negotiating an insurance contract, they wouldn’t know in advance who was going to be lucky and who unlucky on any given day on which the contract would be invoked. To keep things simple, suppose that both possibilities are equally likely. Alice and Bob can then be seen as bargaining behind a veil of uncertainty that conceals who is going to turn out to be Ms Lucky or Mr Unlucky. Both players then bargain on the assumption that they are as likely to end up holding the share assigned to Mr Unlucky as they are to end up holding the share assigned to Ms Lucky. I think the obvious parallel between bargaining over such mutual insurance pacts and bargaining in the original position is no accident. To nail the similarity down completely, we need only give Alice and Bob new names when they take their places behind Rawls’ veil of ignorance. Since we are offering a foundation myth in

5.2. GOLDEN RULE

101

competition with the bible, Alice and Bob will be called Adam and Eve when they are behind the veil of ignorance. Instead of Alice and Bob being uncertain about whether they will turn out to be Ms Lucky or Mr Unlucky, our story about the original position requires that Adam and Eve pretend to be ignorant about whether they will turn out to be Alice or Bob. It then becomes clear that the original position is simply a restatement of an insurance problem—except that the players must imagine sometimes finding themselves in the shoes of somebody else rather than in the shoes of one of their own possible future selves. If Nature wired us up to solve the simple insurance problems that arise in foodsharing, she therefore also simultaneously provided much of the wiring necessary to operate the original position. Of course, in an insurance contract, the parties to the agreement don’t have to pretend that they might end up in somebody else’s shoes. On the contrary, it is the reality of the prospect that they might turn out to be Ms Lucky or Mr Unlucky that motivates their writing a contract in the first place. But when the device of the original position is used to adjudicate fairness questions, then Adam knows perfectly well that he is actually Alice, and that it is physically impossible that he could become Bob. To use the device in the manner recommended by Rawls and Harsanyi, Adam therefore has to indulge in a counterfactual act of imagination. He can’t become Bob, but he must pretend that he could. How is this gap between reality and pretence to be bridged without violating the Linnaean dictum that Nature doesn’t make jumps? Expanding circles. As argued earlier, I think that human fairness norms arose from Nature’s attempt to solve certain equilibrium selection problems. But Nature doesn’t jump from the simple to the complex in a single bound. She tinkers with existing structures rather than creating hopeful monsters. To make a naturalistic origin for the device of the original position plausible, it is therefore necessary to give some account of what tinkering she might have done—just as evolutionary biologists explain the evolution of the eye by a sequence of little steps starting from light-sensitive spots on an animal’s skin. In Peter Singer’s Expanding Circle, the circle that expands is the domain within which norms are understood to apply. For example, if you ask Swedes how they came to drive on the right like the rest of continental Europe, they will say it is because that has been the law in Sweden since 3 September, 1967. But the law doesn’t need enforcing because driving on the right is an equilibrium in the Driving Game. Jesus sought similarly to expand the domain of the principle that you should love your neighbor by redefining a neighbor to be anyone at all, but he met with only limited success because it is only rarely that acting as though you love a stranger is an equilibrium strategy. How does evolution expand the domain within which a moral rule operates? My guess is that the domain of a moral rule sometimes expands when players misread signals from their environment, and so unconsciously apply a piece of behavior or a way of thinking that has evolved for use within some inner circle to a larger set

102

CHAPTER 5. FAIRNESS

of people, or to a new game. When such a mistake is made, the players attempt to play their part in sustaining an equilibrium in the inner-circle game without fully appreciating that the outer-circle game has different rules. For example, Alice might treat Bob as a sibling even though they are unrelated. Or she might treat a one-shot game as though it were going to be repeated indefinitely often. A strategy profile that is an equilibrium for an inner-circle game won’t normally be an equilibrium for an outer-circle game. A rule that selects an equilibrium strategy in an inner-circle game will therefore normally be selected against if used in an outercircle game. But there will be exceptions. When playing an outer-circle game as though it were an inner-circle game, the players will sometimes happen to coordinate on an equilibrium of the outer-circle game. The group will then have stumbled upon an equilibrium selection device for the outer-circle game. This device consists of the players behaving as though they were constrained by the rules of the innercircle game, when the rules by which they are actually constrained are those of the outer-circle game. Family life. Everybody agrees that the origins of human sociality are to be found within the family. A game theorist will offer the explanation that the equilibrium selection problem is easier for evolution to solve in such games. The reason why is to be found in Hamilton’s rule, which explains that animals should be expected to care about a relative in proportion to their degree of relationship (Section 4.2.4). Family relationships therefore provide a natural basis for making the kind of interpersonal comparison of utility that is necessary to operate the device of the original position. The circle was then ready to be expanded by including strangers in the game by treating them as honorary or fictive kinfolk, starting with outsiders adopted into the clan by marriage or cooption. Indeed, if you only interact on a regular basis with kinfolk, what other template for behavior would be available? Putting yourself in another’s shoes. We are now ready to return to the question of how the original position gets to be used—not just in situations in which Alice and Bob might turn out to be Ms Lucky or Mr Unlucky—but when they proceed as though it were possible for each to occupy the role of the other. To accept that I may be unlucky may seem a long way from contemplating the possibility that I might become another person in another body, but is the difference really so great? After all, there is a sense in which none of us are the same person when comfortable and well fed as when tired and hungry. In different circumstances, we reveal different personalities and want different things. To pursue this point, consider what is involved when players consider the various contingencies that may arise when planning ahead. The players then compute their expected utility as a weighted average of the utilities of all the future people—lucky or unlucky—they may turn out to be when all the uncertainties they face in the future have been resolved. When choosing a strategy in a family game, players similarly take their utilities to be a weighted average of the fitnesses of everybody in their family, where the weights correspond to their various degrees of relationship. In order to convert our ability to negotiate insurance contracts into a capacity

5.2. GOLDEN RULE

103

for using fairness as a more general coordinating device in the game of life, all that is then needed is for us to hybridize these two processes by allowing players to replace one of the future persons that they might turn out to be by a person in another body. The empathetic preferences that are needed to assess this possibility then require nothing more than that they treat this person in another body in much the same way that they treat their brothers, cousins, or aunts. Deep structure? The preceding account of the possible evolution of the original position has a similar status to Chomsky’s deep structure of language. That is to say, the original position models the deep structure of human fairness norms. So why do different societies have different views—not on the importance of fairness itself—but on the details of what counts as fair? How come, for example, that Aristotle thought slavery was fine as long as only barbarians were enslaved? Or that Spinoza thought that women should be treated as a lesser breed? To find an answer, we need to ask where Alice and Bob get their empathetic preferences from in the original position. They can’t do without them—otherwise they wouldn’t be able to weigh up prospects in which they have to compare being Alice in one situation with being Bob in another when deciding what is fair. Why else do we have empathetic preferences? Why else did evolution equip us with the expensive mental capacity to make such empathetic comparisons, if not to provide the inputs we need when using fairness norms? But the particular empathetic preferences Alice and Bob may have in a particular time and place are social constructs. We learn to adopt one set of empathetic preferences rather than another by imitating the behavior of those we hope to emulate when we see them acting fairly. This is why Aristotle assigned a small social index to barbarians and Spinoza to women. They thought they were simply acting naturally, but they were actually the victims of the prejudices built into the social consensus of their respective societies—just as we are similarly the victims of the social prejudices built into our own social consensus today. Analogy with language. The deep structure of language is universal in the human species and therefore presumably coded in our genes.3 But the languages spoken in different countries vary because the particular language spoken in a particular country is a product of social evolution—it depends on the cultural history of the people who speak it. The deep structure of language with which we are biologically programmed is necessary to allow babies to learn a language, but the particular language they learn depends on where they are brought up. I think the same is true of fairness. We are biologically programmed with the deep structure of fairness, but the shape of the particular fairness norms we learn as we grow up is determined by the standard of interpersonal comparison that social evolution has crafted in the particular society in which we live. David Hume said much the same thing about social norms in general. The Natural Laws to which writers of footnotes to Plato appeal are actually artificial in 3 Steven

Pinker: The Language Instinct.

104

CHAPTER 5. FAIRNESS

the sense that they are products of social evolution. What is natural is that we have such artificial laws. Similarly empathetic preferences are social constructs. What is determined biologically is that we are able to entertain such empathetic preferences.

5.3

Utilitarianism or Egalitarianism?

If Alice and Bob use the original position to make fairness judgements, will they act like utilitarians or egalitarians? Harsanyi says the former and Rawls the latter. Both Harsanyi and Rawls assume the existence of a social consensus on how interpersonal comparisons are to be made. This assumption trivializes the bargaining between Adam and Eve in the original position. They both have the same empathetic preferences and so will simply agree on whatever they both like best. We therefore might as well replace them by Carol, as in Section 2.5.3. The result will then be utilitarian. So how come Rawls is led to an egalitarian conclusion? His method is to replace the assumption that Alice and Bob maximize expected utility by the assumption that they use the maximin criterion (so that Adam and Eve seek to maximize the utility of whichever of Alice and Bob ends up with the smaller utility in their agreed outcome). Egalitarianism follows immediately, provided that the set of utility pairs from which a fair outcome is to be selected is convex. Roughly speaking, this means that viable compromises between rival alternatives are always available.4 New compromises can then be found that sacrifice some of advantages of the better-off classes to increase the welfare of the least-well-off class. The welfare of the least well-off class will then eventually coincide with the welfare of the class that was previously ranked above them. The same reasoning can then be applied to the larger least-well-off class just created. And so on, until only one class is left. I guess Rawls didn’t understand how iconoclastic it is to reject Bayesian decision theory. If it is to be rejected in the small world of the original position, it would need to be rejected across the board. One might as well reject arithmetic because you don’t like what it says about how much tax you owe! The real difference between utilitarianism and egalitarianism is to be found in asking why people honor fairness norms at all. However, before explaining why, a diversion is necessary to discuss what traditional philosophy has made of John Rawls’ creative contribution. Prioritarianism? Fairness is obviously relevant to the provision of public health care, but health professionals who take an interest in philosophy have been persuaded by writers of footnotes to Plato that their problem in structuring how they think about fairness is a Trilemma, in which their choices reduce to naive versions of either Utilitarianism, Egalitarianism, or what they call Prioritarianism. The leading exponent of this misleading reduction is perhaps the late philosopher Derek Parfit. He thinks it wrong to regard Rawls as an Egalitarian; we should instead apply Rawls’ maximin criterion directly as the moral principle of Prioritarianism. Perhaps he didn’t know that favoring the least well-off will always lead 4 The

set of equilibrium outcomes of an indefinitely repeated game is automatically convex.

5.3. UTILITARIANISM OR EGALITARIANISM?

105

to an egalitarian outcome in a convex environment, in which case Prioritarianism becomes merely a step along the way to achieving an egalitarian outcome.5 He certainly didn’t know the argument coming up later in this section, in which the equalizing principle is central to resolving Rawls’ original position without any need to replace orthodox decision theory with the maximin criterion. However, the immediate point is that what Parfit calls Egalitarianism is very distant from the egalitarian norm discussed in this chapter. Along with other traditionalists, he implicitly assumes the existence of a transcendental Good perceived through an implicitly assumed capacity for Moral Intuition that is explored through a sequence of contrived examples in which all problems of interpersonal comparison are assumed away.6 He asks whether Equality is to be regarded as a Good in itself, or whether it is to be favored because it assists in achieving some other aspect of the Good. Either way, his own Moral Intuition finds in favor of Prioritarianism. This empty kind of discussion is very popular because anybody can join in. No foundations are necessary! You have your Moral Intuition and I have mine—and nothing else matters. The idea that our evolutionary history might be relevant would be dismissed as laughable if considered at all. In particular, the thought that egalitarianism might be a theory about how people compromise when splitting the available gains on their current status quo would be dismissed out of hand. Parfit’s appeals in passing to a popular retelling of the fable of the Missionaries and the Cannibals (Section 2.5.3). What morality should apply if an eye could be surgically removed from a sighted person and implanted in a blind person? Utilitarianism implies that some unfortunate two-eyed people should then give up an eye so that the blind can see, but so does Prioritarianism. According to Parfit, there are philosophers who argue that Egalitarianism requires that everybody be blinded! But even a more reasonable version of Egalitarianism, in which efficiency takes precedence over equality, isn’t necessarily compatible with the notion of an egalitarian norm defended in this chapter. The latter, requires that departure from the status quo should be by mutual consent—and who but a saint is likely to consent to the surgical removal of an eye? Indeed, who but a saint would regard the current status quo as irrelevant? Strains of commitment. Having distinguished between the Egalitarianism of traditionalists and the egalitarianism of this chapter, it is time to return to the question of when egalitarianism trumps utilitarianism. Allegorical statues representing justice take the form of a blindfolded matron holding aloft a pair of scales and a sword. Her blindfold can be taken to correspond to the veil of ignorance. She needs her pair of scales to weigh Alice’s worth against Bob’s. She needs her sword to chastise Alice or Bob if they decline to accept her judgement about what is fair. 5 Rawls draws figures with non-convex feasible sets in which applying the maximin criterion yields non-equal outcomes, but situations where compromise is impossible are rare. 6 Derek Parfit: Equality or Priority, which is a lecture whose transcript can be googled. The issues of measurement and comparison that traditionalists like Parfit ignore are surveyed in the forthcoming The Pursuit of Happiness by Louis Narens and Brian Skyrms.

106

CHAPTER 5. FAIRNESS

Thomas Hobbes bluntly explained this last point by saying: Covenants without the sword are but words. Traditionalists don’t like such talk at all. They think power has no place in a moral discussion. Harsanyi and Rawls thought the same. Their answer to the question of why fairness norms should be honored is to invent yet more skyhooks. Harsanyi calls his skyhook Moral Commitment. Rawls calls his Natural Duty, but later questions his own invention by writing at length on the “strains of commitment” that such notions impose on Alice and Bob. I am with Hobbes on this question. Gurus may tell us that we have a Natural Duty to play fair, but why should we listen? If Natural Duty doesn’t promote fitness, evolution will pay no attention at all. It is enforcement that distinguishes when egalitarianism applies rather than utilitarianism. Utilitarianism applies when the imaginary sources of authority invented by writers of footnotes to Plato are replaced by a real-world external enforcement agency that compels obedience to whatever a fairness norm specifies. This is why it made sense for Carol to be a utilitarian when asked to reform the tax code (Section 2.6). She had the government standing by to police her new code. How many people would report their taxes honestly if we relied on a taxpayer’s sense of natural duty for this purpose? Cheating is widespread even with the government doing its best to stop it. Self-policing fairness. Why is egalitarianism appropriate when external enforcement is unavailable? In talking about expanding circles, it was proposed that we sometimes play reallife games as though bound by rules that don’t actually apply (Section 5.2.2). Such (largely unconscious) behavior can survive if it results in an equilibrium being chosen from those available in the game actually being played. However, the rules of the inner-circle game that survives as an equilibrium-selection device in the outer-circle game that is currently being played need to be realistic. We mustn’t follow the tradition in philosophy of making up whatever rules confirm our prejudices. The inner-circle game must be a game that our ancestors would recognize as making sense in the environment in which they lived. In particular, it must operate in the absence of any external enforcement. In examining the device of the original position from this point of view, there are three points at which external enforcement needs to be excluded: Equilibrium outcomes: The first point requires an appeal to the reciprocity principle of the previous chapter. The agreements envisaged as being reached in the original position must be viable as equilibrium outcomes in the ongoing game of life that Alice and Bob are actually playing. In particular, we must remember that property rights are a human invention. As David Hume explains, it is only by convention that you own your home, or your car, or anything else. John Locke tells us that we have a Natural Right to our own bodies, but as someone who invested money in the slave trade of his time, he should have known better. Realistic bargaining: The bargaining envisaged in the original position must be realistic. We mustn’t cheat by inventing some fair process imported from the golden age of the poets. However, we can leave this issue until the next section, since

5.3. UTILITARIANISM OR EGALITARIANISM?

107

our current assumption that Alice and Bob have the same empathetic preferences trivializes the bargaining problem. Realistic randomizing? It matters how we model the random event that determines which of Adam and Eve will turn out to be Alice and Bob when they supposedly emerge from behind the veil of ignorance. This very simple issue turns out to be the fulcrum on which a defence of egalitarianism turns, and so it is worth discussing at some length. The lot causeth contention to cease. The Book of Proverbs endorses the tossing of a coin to settle disputes. Adam and Eve do the same when considering who will be Alice and who will be Bob when they leave the original position. But the coin tossed in the original position is only hypothetical. When is this phantom coin tossed? How do they know which way it falls?7 In the following example, Alice and Bob both need a heart transplant, but only one heart is available. If the lives of both are regarded as equally valuable, then a utilitarian will be indifferent between giving the heart to Alice or Bob. But Alice would regard it as grossly unfair if the heart were then given to Bob on the grounds that he is a man—or because he is white or rich. Nor would she be at all mollified if told that she had an equal chance of being a man when the egg from which she grew was fertilized in her mother’s womb. Alice therefore cares a lot about when the phantom coin is tossed. If such a random event is to be used to determine who gets the heart, she will argue that a real coin should be tossed right now. If the coin falls in favor of Bob, she will be tempted to find arguments why this particular toss is unfair. So how is it possible to gain her unforced consent to honoring the fall of a coin that is only hypothetical? The very simple answer is that for the original position to be viable without external enforcement, Adam and Eve must be indifferent between the coin falling heads or tails. Otherwise, if the original position were played for real, either Adam or Eve would reject the outcome and call for the coin to be tossed again. Egalitarianism justified! When Adam and Eve have the same empathetic preferences—as we are currently assuming—the outcome of using the original position will therefore lead to the standard conception of an egalitarian sharing rule. Alice and Bob will get whatever shares equalize their empathetic utilities. When their shares are measured in terms of their personal utilities, these utilities will therefore be proportional to their social indices. So we begin and end with Aristotle. He tells us first that the “sources and springs of friendship, political organization, and justice” lie within the family, and second that “What is fair . . . is what is proportional.” 7 The actual probability with which these two alternatives occur isn’t significant, because we can absorb any deviation from our ongoing assumption that the two alternatives are equally likely into Alice and Bob’s social indices. Being a mighty hunter in ancestral times would therefore have been one of the criteria that generate a larger social index. It would have been the equivalent of being a modern pop star or football hero.

108

CHAPTER 5. FAIRNESS

When are people actually egalitarian? The preceding argument for fairness norms having an egalitarian structure applies particularly well to the ancestral bands of hunter-gatherers in which evolution shaped the human way of being social. 8 Anthropologists are solidly of the view that the sharing of food—especially meat— in such groups was very egalitarian. But we don’t know to what extent such sharing could be modeled as proportional to a person’s social index. It was only late in the day that anthropologists thought to measure how much different people got when meat was shared, and the available data is inadequate to settle the issue. There are also laboratory experiments carried out by psychologists under the heading of “modern equity theory”. Their theory is the same as the egalitarianism of this chapter, but derived empirically.9 The theory is said to be modern to distinguish it from Aristotle’s observation that what is fair is what is proportional. Do people honor egalitarian norms in the laboratory? The evidence is mixed. The theory does well in cases when Alice and Bob are said to have jointly invested in a business project and the time has now come to split the profits—which are then split in proportion to the size of each player’s investment. The worst cases I know occur when Alice and Bob are friends or lovers, and so likely to have sympathetic preferences. If any psychologists are reading this, I hope they will consider new experiments on modern equity theory that pay closer attention to where the status quo is located, what counts as a reasonable correlate of utility in different contexts, and what social indices may be in use. Finally, there are theoretical results from cooperative game theory in which axioms are proposed in an attempt to characterize what counts as fair. All the axiom systems of which I know that postulate full interpersonal comparison—of both the zero and the unit on Alice and Bob’s personal utility scales—yield the same conclusion, variously called the egalitarian or proportional bargaining solution because it is identical to what we have been calling an egalitarian norm.

5.4

Achieving a Social Consensus

Why has evolution equipped us with the capacity to have empathetic preferences? So that we can make fairness judgements. When do they take the form of a weighted sum of Alice and Bob’s personal utilities? When both are completely successful in their attempt to empathize with the other. So why aren’t our fairness norms utilitarian? Because utilitarian norms require external enforcement. Why are our fairness norms egalitarian instead? Because when a social consensus on empathetic preferences exists, the only agreement in the original position that doesn’t somehow rely on external enforcement equalizes their empathetic utilities. Why do we need the original position at all? Because it still works when there is no social consensus, and so Alice and Bob may have different empathetic preferences. Why will social evolution remove differences between our empathetic preferences when the original position is in use? That is the subject of this section. 8 Christopher

Boehm: Hierarchy in the Forest: The Evolution of Egalitarian Behavior. broadly philosophical discussions, see Graham Wagstaff: An Integrated Psychological and Philosophical Approach to Justice, and George Homans: Social Behavior: Its Elementary Forms. 9 For

5.4. ACHIEVING A SOCIAL CONSENSUS

109

No secrets. There is no depth to the answers given to these questions except for the last—provided that we are willing to go along with the assumption that Alice and Bob have no secrets from each other. This is a strong assumption in a modern society and one reason that we aren’t more successful in using fairness to solve disputes is doubtless that it often fails. However, we shall continue to maintain the assumption in this section because without it we wouldn’t be able to appeal to the following reasonably realistic model of bargaining when trying to work out what deal Adam and Eve will reach behind the veil of ignorance when they have different empathetic preferences. But we first need to be more careful about what kind of sharing problem Alice and Bob need a fairness norm to solve. Sharing Game. A kind of sharing game was used as an example when comparing egalitarianism and utilitarianism in a previous chapter (Section 2.5.2). We looked at the very different results of using these two conceptions of how fairness norms work in the case when Carol is asked to share a sack of flour between Alice and Bob. Essentially the same model of sharing is recycled here, but without Carol to provide a common standard of interpersonal comparison. Anthropologists report that meat obtained by cooperative hunting in huntergatherer groups was commonly shared very fairly. But the problem of assessing the costs and benefits of such cooperative behavior will be abstracted away. When the benefits don’t outweigh the costs, cooperative behavior won’t evolve. With this caveat, we might as well proceed as though Alice and Bob have come across a dead animal, and the question is how they are to share the meat. What would happen if they were unable to agree on how to share then matters. We have to deal with the problem envisaged by Epicurus when he said that fairness exists to coordinate behavior on some compromise as an alternative to getting into a fight. It is necessary to specify how Alice and Bob evaluate the (possibly violent) consequences of a failure to agree. We can summarize these considerations by saying that they reduce the sharing question to what economists call a bargaining problem. This simply consists of a set of possible agreements and a disagreement outcome—often called the status quo—from which Alice and Bob can depart only by mutual consent (Section 2.5.3). Both the possible agreements and the disagreement outcome are represented by the expected utilities assigned to them by Alice and Bob. Bargaining. The bargaining envisaged as taking place behind the veil of ignorance must be the ordinary kind of bargaining to which Alice and Bob are accustomed in real life.10 How else could they predict what its outcome would be? So we don’t want any fancy bells or whistles. We especially don’t want any suggestion that only fair bargaining is allowed—otherwise our attempt to explain how fairness norms work would be circular. Plato’s Thrasymachus would therefore have approved 10 I am making a fuss about this obvious point because the nearest orthodox philosophy comes to the approach described here is David Gauthier’s Morals by Agreement, in which he not only argues that cooperation is rational in the Prisoners’ Dilemma, but invents his own theory of bargaining.

110

CHAPTER 5. FAIRNESS

of the bargaining theory needed: the bargainers negotiate face-to-face, bringing to bear whatever power they may have at their disposal.11 The economic theory of bargaining invented by John Nash and Ariel Rubinstein for the case when the bargainers have no secrets from each other is therefore just what is required. To cut what could be a long story short, both their approaches lead to what is called the Nash bargaining solution.12 Rubinstein’s bargaining model is the more realistic. In a simple version, Alice and Bob alternate in making offers on how to split whatever is available. The bargaining ends either when an offer is accepted, or some small probability event outside anyone’s control interrupts the bargaining, in which case Alice and Bob are stuck with the disagreement outcome. This bargaining game has a unique perfect equilibrium that approximates the Nash bargaining solution when the interruption probability after each refusal of an offer gets small. A simple example may help give the flavor of how the Nash bargaining solution works. Suppose that Alice and Bob are a divorcing couple in dispute over who gets custody of their baby without the opportunity to appeal to a court of law. If they can’t agree, the baby goes into an orphanage, which each regard as the same as the other getting the baby. Various compromises are available. They could agree to some time-sharing arrangement, but it is simpler to assume that Alice and Bob only consider agreements that say who gets the baby with what probability. The Nash bargaining solution unremarkably says that each will then get the baby with probability one half. For this reason, some authors assess the Nash bargaining solution as if it were an attempt to characterize a fair arbitration scheme, but this is to miss the point altogether. It is totally without virtue as a fairness criterion because it remains the same however we compare Alice and Bob’s personal utility scales. This is easy to see in the preceding example. Alice may love the baby very dearly and Bob hardly at all. If so, it wouldn’t be fair that Bob should have an equal chance of getting the baby—which is why King Solomon had to find a way of determining who loved the baby best in his famous judgement (Section 2.2). 11 In Plato’s Republic, Thrasymachus absurdly suggests that talk of justice simply disguises the unbridled exercise of power. 12 John Nash is famous for his 1951 definition of a Nash equilibrium, although he was anticipated by Auguste Cournot in 1838 (Section 2.2). But Nash was truly original in his 1950 thoughts on bargaining, which economists at the time dismissed as a neglected branch of psychology. He gave a list of rationality assumptions that determine a unique outcome of a bargaining problem called the Nash bargaining solution—which isn’t at all the same thing as a Nash equilibrium. He supported this result by analyzing a simple bargaining model called the Nash demand game, in which Alice and Bob make simultaneous demands, which are implemented if jointly feasible and which result in the disagreement outcome otherwise. With a little uncertainty about what potential agreements are actually feasible, he found that all Nash equilibria of his demand game approximate the Nash bargaining solution. Economists largely ignored this discovery until 1982 when Ariel Rubinstein reported that his much more realistic alternating-offers bargaining game has a unique perfect equilibrium. My own contribution was to show that this unique bargaining outcome implements a generalized version of the Nash bargaining solution. All this is explained in my book Playing for Real with a minimum of mathematics, and in my Natural Justice with some diagrams substituting for equations. Since we aren’t using mathematics at all here, much of what I shall be saying will have to be taken on trust.

5.4. ACHIEVING A SOCIAL CONSENSUS

111

Why play fair when you could bargain? In modern times, people use fairness norms when the number of players is too large or communication is too difficult for bargaining to be practical, or else because the cost of bargaining in time and trouble is too high. Sometimes an attempt at bargaining would be punished by the disapproval of onlookers—as for example at a dinner party when a dish in short supply is being shared. Fairness is often mentioned in questions of public policy, but this is mostly just for rhetorical purposes. Things were more interesting in ancestral times. It seems likely that fairness is as ancient as language—perhaps more ancient. If so, then bargaining wasn’t available as an alternative coordinating device. The other obvious alternative to fairness is leadership, but the hunter-gatherer societies that survived into modern times had no leaders. In fact, the punishments inflicted on tough guys who sought to dominate their fellows were sufficiently severe that anthropologists report that everybody understood that it was wise not even to give the impression of being bossy. Fairness was therefore much more important in human ancestral societies than it is to us—which goes a long way to explaining why their egalitarianism came as a surprise to early anthropologists.

5.4.1

Modeling Social Evolution

How might social evolution generate a social consensus on how interpersonal comparisons are made? Nothing profound is required. People mostly have the same empathetic preferences for the same reason that teenagers who hang out together dress similarly and like the same kind of music—or that traditional philosophers all admire Plato and Kant—because the way to count as a somebody is to copy those who are counted as a somebody already. The following outline of a model fleshes out this story, but nothing much will be lost by skipping forward to Section 5.5. Memes. Richard Dawkins introduced the word meme to serve as a substitute for gene when biological evolution is replaced by social evolution. I don’t think it helpful to pursue the biological parallel too closely, so my use of the term is better understood in the sense in which it has entered popular culture. I treat empathetic preferences as memes propogated by imitation. Alice and Bob unconsciously alter their empathetic preferences if they would do better in terms of their personal utilities by copying the empathetic preferences of others when the original position is used. The players’ strategies in the underlying evolutionary game then correspond to their (largely involuntary) choices of different standards of interpersonal comparison. At a Nash equilibrium of this evolutionary game, neither Alice nor Bob will have an incentive to switch from their current standard of interpersonal comparison to one they see being operated by others.13 13 My book Natural Justice explains why different ways of modeling the imitation process can lead to different definitions of an empathy equilibrium. The more refined definition considered there requires that neither Adam nor Eve in the original position have an incentive to misrepresent their empathetic preferences when these are evaluated using the empathetic preferences that Alice and Bob actually hold. The refined definition leads to the same conclusion as the cruder version of the text.

112

CHAPTER 5. FAIRNESS

Social consensus achieved! It turns out that the evolutionary process just described results in Alice and Bob ending up with the same empathetic preferences, and so they will agree on what social indices should be assigned to different types of people. We can therefore recycle our earlier analysis in which the existence of a social consensus was taken for granted. In particular, we should expect to see egalitarian fairness norms in operation in situations where no external enforcement agency is available. Evolution erodes the moral content of norms? An equilibrium in empathetic preferences encapsulates the cultural history of a society that led people to adopt one standard of interpersonal comparison rather than another in a particular context. Traditionalists don’t like the fact that this history will largely be shaped by the way in which power is distributed in whatever society is under study. In their fairy stories, power can’t be relevant to how moral norms are shaped—morality is a substitute for power! Their worst fears are apparently confirmed when they learn the conclusion to which our current evolutionary model leads. If the underlying sharing game Alice and Bob play in real life were always the same, social evolution would shape their empathetic preferences so that the egalitarian norm coincides with the Nash bargaining solution of the sharing game. 14 But the Nash bargaining solution has no virtue as a fairness norm. Does this mean that the traditionalists are right to argue that evolution and morality are like oil and water in that they can’t be mixed? It certainly does mean that we can’t expect the kind of morality that evolved for use in real-life situations to coincide with traditional fantasies, but it doesn’t imply that we can throw away fairness norms in favor of implicit bargaining models. The reason is that we use the same fairness norm to solve many different sharing games in real life. In our simple model of social evolution, Alice and Bob’s empathetic preferences will evolve so that the egalitarian norm of the historically average sharing game coincides with the Nash bargaining solution. In other sharing games—especially those that are unusually asymmetric or that incorporate relatively new technologies—the egalitarian outcome won’t look like the Nash bargaining solution at all. Traditionalists who think that empathy matters at all may therefore continue to complain that they don’t like the way empathetic preferences are shaped by social evolution, but they can’t simultaneously complain that they have no role in the fairness norms we use in real life. The needy will get more in contexts where hunger is an issue. The rich will pay higher taxes. And so on.15 If such conclusions didn’t follow, we wouldn’t have come anywhere near explaining how evolution shaped our intuitions of how fairness works. For those with a scientific background, the model has the additional merit that it 14 In which case it is a theorem that both also coincide with the utilitarian norm. So social evolution will lead both egalitarians and utilitarians to the same standard of interpersonal comparison. Perhaps this is why Harsanyi and Rawls favored similar reforms in private conversation, although espousing very different ideas on how societies should be organized. 15 My Natural Justice pursues the impact of changes in need, effort, ability and status on a person’s social index by looking at how altering suitably defined versions of these notions changes the historical average sharing game.

5.5. SOCIAL CONTRACTS

113

provides a theoretical way of predicting what standards of interpersonal comparison different evolutionary histories will generate. All we need to do in principle is to work out the Nash bargaining solution of the historically average sharing game, and then choose Alice and Bob’s social indices so that this coincides with the egalitarian outcome. As with much else, easier said than done! Local justice. As Epicurus explained, justice is the same for all at any particular time and place, but may vary between different times and places—a fact that has been widely documented.16 But one doesn’t need to look very far to see that we take this for granted when making fairness judgements. For example, everybody agrees that need is what matters when assigning food stamps, but merit is what counts when awarding Nobel Prizes. In our evolutionary model, this phenomenon is captured by asking who Alice and Bob copy when adjusting their empathetic preferences. If they only copy people they see operating the original position in a particular context, then the standard of interpersonal comparison that evolves in that context needn’t resemble the standard that evolves in other contexts.

5.5

Social Contracts

I hope the discussion so far has at least made it plausible that evolution could have generated the fairness norms that actually get used in real life for solving small-scale coordination problems. Just as it would have been better if evolution had made us stronger and cleverer, so it would perhaps have been better if evolution had made our social life more like that of bees and ants, but we are stuck with what evolution actually made of us. But we aren’t altogether helpless victims of our evolutionary history. Our big brains evolved partly so that we could chip flint to make stone tools, but we used these tool-making skills to create modern technology. Our big brains also evolved to allow us to live amicably together in small societies without anyone ordering us around. Perhaps we can similarly use these social skills to learn to live together more amicably in large societies. Scaling up? Can we learn to use the fairness norms that evolved for small-scale applications to solve large-scale problems? In doing so, we must put aside the grandiose aspirations of writers of footnotes to Plato. They have been telling us to ignore what evolution made of us for more than two thousand years without any notable success. The fairness norms we actually use in real life aren’t shadows cast by some absolute noumenal world; they are simply social tools washed up on our evolutionary beach. If enough of us sufficiently near the levers of power want to use them to improve how big societies work, let us just get on and do it. We don’t need 16 For example, Jon Elster: Local Justice and Peyton Young: Equity. It is usually taken for granted that the standard of interpersonal comparison is universal, and so the differences are attributed to the method used to compute a fair outcome. In the approach defended here, the original position is assumed to be universal, and the differences attributed to variations in the standard of interpersonal comparison.

114

CHAPTER 5. FAIRNESS

to invent the kind of metaphysical justifications that traditionalists think necessary. It is enough that we want to do it. Perhaps we would want to do something else if our histories—both personal and social—had been different, but so what?

5.5.1

Traditional Social Contracts

Before exploring the possibility of using egalitarian fairness norms to improve our current social contract, it may be helpful to look at some of the traditional approaches to thinking about social contracts. Two aims need to be distinguished. Epicurus would have approved of the first. It seeks to explain how societies came to be as they are without inventing metaphysical skyhooks. He would have been less enthusiastic about the second aim, which is to justify replacing our current social contract with some utopian alternative. He would have been even less enthusiastic about how the utopian aim sometimes gets muddled with the historical aim. Social contracts as historical constructs. The original philosophical approach was to suggest that the rules which govern our moral and political behavior were agreed at some ancient conclave, whose authority we are still somehow obliged to honor. David Hume wrote an essay condemning this notion of an “original contract”. Why should we be bound by agreements made at such a conclave? Where is the evidence for such an ancient conclave anyway? Hume accepts that talk of an ancient conclave may stand in for a history of lesser coordinating innovations that gained general acquiescence over time, but points out that the histories of actual constitutions are largely accounts of usurpation, conquest and rebellion. He has no time at all for the fiction that we could somehow be bound by some historical agreement of whose very existence we have no knowledge. How are Hume’s criticisms of the idea of the original contract to be reconciled with his much admired views on reciprocity and the evolution of conventions? We must remember that Hume didn’t have modern game theory at his disposal, and so didn’t realize the extent to which the reciprocity principle can be used to sustain cooperative equilibria in repeated games. When operating such cooperative equilibria, we don’t need to invent metaphysical reasons to justify our behaving cooperatively because, most of the time, it is in our long-run self-interest to do so. Nor did Hume know of Darwin’s theory of evolution. He says somewhere that species never change. So when he speaks of history, he isn’t referring to evolutionary history. If we ignore ephemeral aspects of social contracts, like which family counts as royal or which side of the road on which to drive, his analysis is therefore unproblematic for the scientific approach to social contracts pursued in this book—which I hope goes some way toward fulfilling the aspirations of the early social contract theorists like Grotius and Pufendorf whose ideas Hume was criticizing. Social contract theory as a political tool. The idea of a social contract was a hot topic for David Hume because of its use by the Whigs of his day in opposing the Tories.The Tories were conservatives who favored retaining the traditional British monarchy because they thought this is how things had always been. The Whigs

5.5. SOCIAL CONTRACTS

115

argued that governments operate by mutual consent of the governed, and so kings can legitimately be replaced if this consent is lost.17 The Whigs were inspired by the social contract ideas of John Locke. The Tories didn’t feel the need for any philosophy, but they could have appealed to Thomas Hobbes, whose earlier social contract theory defended the authority of kings. Looking back, it is plain that social contract theorists like Hobbes and Locke began with their favored conclusion and invented whatever philosophical arguments took them to where they wanted to go—as with Plato when he tacked on as an afterthought the idea that his idealized Spartan constitution could be deduced from a contrived definition of justice, or Rawls when he discarded Bayesian decision theory in favor of the maximin criterion. In the disputations that followed, the explanatory aspirations of the earliest social contract theorists were lost sight of altogether. Hobbes’ social contract. Thomas Hobbes (1588–1679) was a royalist who thought the rebels in the English Civil War irrational to prefer the brutality of the war to submission to the king. His Leviathan idealizes this political position by comparing the safety to be found in accepting the role of a cog in an authoritarian social machine with what he called the state of nature, which he famously characterized as a “war of all against all” in which life is “poor, nasty, solitary, brutish, and short”. Everybody would doubtless agree if this were indeed the choice! Locke’s social contract. John Locke (1632–1704) was even more of an empiricist than Hobbes. His idea that the human mind is a blank slate on which experience can write anything whatever continues to be popular with utopians in spite of all the objective evidence to the contrary from biology and neuroscience.18 His contribution to social contract theory can be seen as a philosophical justification for replacing the outdated morality of medieval Christianity by a morality more suited to the new commercial age in which he lived. In politics, these ideas became the bedrock on which British opposition to the unbridled authority of kings was based. Locke’s fable of the origins of the social contract is perhaps best seen as a story of how a liberal constitution might have arisen in an ideal world. He introduces the skyhook of Natural Rights according to which “No one ought to harm another in his life, health, liberty, or possessions”. Property rights can then only be transferred in a civilized way. They are originally acquired by mixing one’s labor with property in a state of nature that precedes ownership, provided “enough and as good is left in common for others”. 17 The Whigs are traditionally associated with the Glorious Revolution of 1688, in which the Catholic and authoritarian James II was expelled in favor of the Protestant and constitutionally minded William III. American history also boasted a Whig party, broadly similar in character to its British counterpart. It was vocal in its opposition to Andrew Jackson’s authoritarian innovations in the use of the presidential veto. Before joining the newly emergent Republican party, Abraham Lincoln was a Whig, but modern Republicans have largely forgotten their whiggish roots. 18 John Locke: Two Treatises of Government. Steven Pinker: The Blank Slate: The Modern Denial of Human Nature.

116

CHAPTER 5. FAIRNESS

Rousseau’s social contract. Bertrand Russell is doubtful whether Jean-Jacques Rousseau (1712–1778) can be considered a philosopher at all. But whatever his genuine merits as a philosopher, his writings have arguably been more influential than the writings of the rest of the philosophy profession put together. He was for the Jacobins of the French Revolution what John Locke was for the earlier Whigs across the English Channel, but written very much larger. Perhaps equally important, he marked the moment in which a wave of irrational romanticism began to displace the values of the enlightenment. Rousseau’s arguments in support of his social contract theory seem to me largely cosmetic. In his romantic state of nature, we were noble savages wandering at peace with the world in the primeval forest, but although we were “born free”, we are now “everywhere in chains”. His solution to our decline from this blissful state of nature is to surrender our autonomy to the General Will. To articulate the General Will, we need someone with sublime wisdom: otherwise we might get stuck with the Will of All, which isn’t at all the same thing. On the contrary, the wills of all citizens must be brought into accord with the will of this mythical sage.19 I guess this is what Robespierre thought he was doing when presiding over the Terror after the French Revolution. Kant’s social contract. Immanuel Kant absurdly idolized Rousseau as the “Newton of the moral world”, but I hope I will be excused from discussing what Kant’s categorical imperative supposedly implies about social contracts. He advocates what we nowadays regard as an enlightened form of political freedom, but the fact that we like his conclusion shouldn’t be allowed to obscure his failure to provide any serious foundation for the claim that Rationality can lead nowhere else. Kant really thought that his metaphysics were more than just rhetoric, but is the same true of other social contact theorists? For example, the tyrannical Maximilien Robespierre tells us that “Any law that violates the inalienable rights of man is essentially unjust and tyrannical; it is not a law at all.” But such talk of inalienable or imprescriptible Natural Rights in documents like the 1789 French Declaration of the Rights of Man is pure rhetoric: to quote Jeremy Bentham, it is “nonsense upon stilts” if taken literally. A much-quoted passage from the American Declaration of Independence similarly says, “We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.” But if these admirable aspirations were really self-evident, it wouldn’t be necessary to say so. As in the French case, we are really just being told the principles on which the founders of the American Republic planned to proceed. They probably understood themselves that their metaphysical rhetoric was just propaganda. 19 Jean-Jacques Rousseau: The Social Contract. David Hume selflessly helped Rousseau when this seriously neurotic show-off was banished from France, but his reward was to be accused of plotting against him. Can it really be true, as Rousseau’s Confessions tells us, that he abandoned his five babies one by one on the doorstep of an orphanage? But perhaps this is why his book Emile is so hilariously unrealistic about the realities of bringing up a child to be a good citizen.

5.5. SOCIAL CONTRACTS

117

Rawls’ social contract. John Rawls tells us that his theory of justice is in the tradition of Locke, Rousseau, and Kant. Hobbes is always omitted from this orthodox list—presumably because of his disdain for metaphysics. In the rest of this chapter, I argue that Rawls’ moral intuitions are better seen as idealizations to whole societies of the fairness norms that we observe being used to solve small-scale coordination problems in our everyday lives, and to which the early part of this chapter was devoted. I think that all moral philosophers really get their intuitions from the same source, but Rawls is one of the few who isn’t led badly astray by their belief that they have a hotline to some metaphysical world available only to those gifted with sublime wisdom.

5.5.2

Widening the Scope of Fairness Norms

The early part of this chapter offers a putative evolutionary history of the fair social contracts of the small societies of hunter-gatherers into which the human species was divided ten thousand years or more ago. My contention is that we still use the same fairness norms in modern times to solve small-scale coordination problems. I now move on from this historical approach to add my voice to the rhetorical efforts of social contract theorists like Hobbes and Locke. I propose seeking to use the egalitarian fairness norms that I think we use to solve small-scale problems to address the large-scale problem of improving at least some aspects of our social contract. The idea will have political appeal only to the extent that it accords with the fairness intuitions that ordinary people acquire as they live their ordinary lives. In particular, the standard of interpersonal comparison employed must be whatever convention people actually operate in real life, rather than some utopian ideal preached from a metaphysical pulpit. Does history matter? Neither Harsanyi nor Rawls saw any need to propose a state of nature in defending their utopian aspirations. They proceed as though history doesn’t matter. Thomas Paine thought the same in his fiercely contested debate with Edmund Burke after the French Revolution.20 Paine argued that we should throw away our old social contract—root and branch—and design a new social contract that owes nothing whatever to the past. Burke argued that a social contract is much more than Robespierre and his fellow revolutionaries understood. You can write words on a piece of paper and call it a constitution, but you waste your time if the citizens of your society aren’t ready to accept your piece of paper as a coordinating device. What determines how much the citizens are willing to accept? The attitudes they inherit from the past. Ignoring history seems not just unwise, but on the edge of incoherence—rather like the Irish peasant who was asked the way to Dublin and replied that if he were going to Dublin he wouldn’t start from here. We have no choice but to start from where we are now. Insofar as my approach to social contract issues can be said to 20 Edmund Burke: Reflections of the Revolution in France and An Appeal from the New to the Old Whigs. Thomas Paine: Rights of Man and Common Sense.

118

CHAPTER 5. FAIRNESS

have a state of nature, it is therefore the social contract we are operating right now in the society in which we currently live. In the debate between Thomas Paine and Edmund Burke, I therefore think Epicurus would have been firmly on the side of Burke. The real question is: How can we improve what we have got already? What gains are possible relative to the current status quo? Efficiency. Fairness isn’t the only thing that matters when writing idealized models of social contracts. The first priority is that a social contract be sufficiently stable to survive for long enough to be recognized as a social contract. An attempt to meet this requirement is made by insisting that only equilibria of the game of life played by a society are viable as possible social contracts. The second priority is that a social contract be efficient, which means that nothing is wasted—no reform is possible in which everybody can gain. An anecdote about a one-lane bridge in Ithica, NY may help to explain why efficiency is relevant. Professors from nearby Cornell University liked to tell how cars behaved very fairly at this bridge by taking turns to cross, shrugging off the observations from foreign professors that cars on the one-lane bridges with which they were familiar did the efficient thing in terms of expected waiting times by pausing until a gap appeared in the flow of oncoming cars and then crossing until a gap appeared in their own flow. However, the last time I heard a Cornell professor on this subject, he reported that the current norm was for four cars to go one way before four cars went the other. In the face of increasing congestion, social evolution was on the way to generating an efficient solution to the Ithaca bridge-crossing problem! Evolution and efficiency. A fable may help explain how evolution can generate efficient social contracts. Suppose that many identical small societies are operating one of two social contracts, busy and idle. If busy makes each member of a society that operates it fitter than the corresponding member of a society that operates idle, then here is an argument why busy will eventually come to predominate. To say that a citizen is fitter in this context means that the citizen has a larger number of children on average. Societies operating social contract busy will therefore grow faster. Assuming societies cope with population growth by splitting off colonies which inherit the social contract of the parent society, we will then eventually observe large numbers of copies of societies operating social contract busy compared with those operating contract idle.21 But this is how evolution works. Fairness comes after efficiency. Fairness norms, and other norms like traffic conventions or ownership norms, only arise at the third level of priority in this 21 Although selection takes place among groups, the argument isn’t an example of the group selection fallacy, because a social contract is identified with an equilibrium of the game of life played by each of the competing societies (Section 4.2.3). But selection among equilibria doesn’t require that individuals sacrifice anything for the public good, because every individual in every group is already optimizing his or her fitness by acting in accordance with the social contract of their society. The paradigm of the selfish gene is therefore maintained throughout.

5.5. SOCIAL CONTRACTS

119

idealized approach. They exist as selection devices when multiple efficient equilibria are available—which is all of the time when repeated games are involved. Utopian reform? It may be helpful to review what would be involved in consciously attempting a fair reform of some aspect of a society, focussing on the difficulties along the way, both theoretical and practical. Stability: The first step is to identify what reforms are feasible—which reforms lead to outcomes that won’t fall apart because they don’t respect the nature of human nature. A reform that requires people to neglect their own self-interest may work for a while, but it will gradually unravel as people invent reasons why it is acceptable to cheat a little here and there. In our idealizations, this stability requirement is met by requiring that only equilibria of the human game of life be considered. It must then be remembered that the politicians who run our governments and the officials charged with implementing their decisions are also players in the game of life. When can they be trusted? Only when guardians are in place—along with guardians to guard the guardians. In brief, things need to organized so that the guardians guard each other. It is tempting to respond that even this first step in implementing a fair reform is an impossible task, and it is true that the best one can hope for in real life is some temporary stability. Who knows where social evolution will eventually take a society after a reform has been implemented? But one can at least make the effort to look ahead and attempt to predict how a new system might be gamed. The alternative—which we usually see in practice—is equivalent to strapping on a pair of ramshackle wings and jumping off a high building in an attempt to fly. Efficiency: The next step is to question the efficiency of possible reforms that pass the stability test. Is there some way we could improve on the proposed reform so that everybody gains? It is important here to recognize that creativity and enterprise mustn’t be stifled as in the old Soviet Union. We are going the same way too, with rules and regulations coming out of our ears. There is also a major philosophical question: When it is said that efficiency requires that everybody should gain, who counts as everybody? What of mutual consent in general? Whose consent are we talking about? I have heard it argued that chimpanzees should count. If so, why not animals in general? What of young children, or people suffering from dementia or some other disabling mental illness? What of the economically powerless—the huddled masses famously promised liberation by the verse engraved on the Statue of Liberty? What of people of another religion or a different color? Here is yet another problem that writers of footnotes to Plato are reluctant to confront. The unwelcome truth is that only those who are able to influence the outcome actually count. If a group that is currently discriminated against can’t organize itself to bring its collective power to bear, then it will have to rely for help on the doubtful sympathetic preferences of those who can. A proper analysis of such questions will have to await a better understanding of both human psychology and the game theory of coalition formation. In the interim, we have to live with the unpleasant fact that mutual consent in practice

120

CHAPTER 5. FAIRNESS

actually means the consent of those sufficiently powerful—either individually or collectively—to influence what reforms have a chance of getting implemented. Finally, there is the problem posed by the fact that people often don’t even know who their neighbors are in a large modern society. The reciprocity principle says that a multitude of efficient equilibria is always available in an indefinitely repeated game—provided that everybody’s business is an open book. But what happens when people have secrets from each other? The best theory has to offer at present are examples in which the extent to which reciprocity works depends on how much the players in a game know about each other. However, the currently unregulated invasion of our privacy through the internet may paradoxically ease this difficulty (although I no more keen on my own privacy being invaded than anybody else). In any case, even when the kind of first-best efficiency available when information isn’t a problem is beyond our reach, we still need to strive for whatever second-best kind of efficiency can be achieved when information is a problem. Much to the distress of some writers of footnotes to Plato, markets will be part of the solution to this efficiency problem.22 When working well, they can be marvellously effective in overcoming informational difficulties. However, it is necessary to be skeptical about the claims of right-wing economists that markets are always the solution to everything.The theoretical results on which they rely are subject to all kinds of restrictions—there must be a large number of small buyers and sellers, goods must be divisible, preferences must be convex, and so on. Enthusiasts for free markets typically overlook these restrictions, but when they aren’t satisfied—which is much of the time—markets need to be regulated. It is true, as right-wingers complain, that current regulation is often appallingly bad. But it would be a lot less bad if not left in the hands of lawyers, who are often entirely ignorant of the most elementary economic principles.23 Sometimes regulation is entrusted to the very capitalists whose activities are supposedly being regulated. It is very frustrating for game theorists when they know how a particular industry needs to be regulated but their expertise on such social engineering is called upon only when lawyers want to discredit the arguments of other lawyers. Fairness: The proposition defended in this chapter is that fairness norms evolved to select an efficient equilibrium when many are available. Because they work in everyday subgames of the human game of life, perhaps people might be persuaded to honor them if applied to the social contract problems that arise in the game of life as a whole. Politicians might reflect that Nash equilibria are only proof against 22 Here is Robert Burton on markets: What’s the market? A place (according to Anacharsis) wherein they cozen one another, a trap: nay, what’s the world it self? A vast chaos, a confusion of manners, as fickle as the air, domicilium insanorum, a turbulent troupe full of impurities, a mart of walking spirits, goblins, the theatre of hypocrosie, a shop of knavery, a nursery of villainy, the scene of babbling, the school of giddiness, the academy of vice—and much more of the same! 23 I was once employed as a consultant by a leading lawyer in economic regulation, to whom I was trying to explain that the short-haul UK package-holiday business might usefully be modeled as Bertrand-Edgeworth competition. To explain the idea, I began by saying that the idea is a hybrid of the Cournot and Bertrand models of oligopoly. He then asked: What is an oligopoly? But an oligopoly is the standard textbook term used by economists to describe an industry dominated by a few powerful companies—the kind of industry for which anti-trust laws were created to regulate.

5.5. SOCIAL CONTRACTS

121

deviations by individuals. James Madison’s observation that operating an unfair social contract allows fairness to act as a focal point around which destabilizing coalitions can gather is therefore of practical relevance. However, using fairness to select among potential social contacts is unlikely to work if it doesn’t correspond to how fairness is actually used in small-scale interactions. In particular, the ineffective rhetorical appeals to justice with which we are currently regaled need to be put aside. Jon Elster’s study of collective wage bargaining in Sweden says everything that needs to be said on this score. Whereas the rest of the world bargains about how much money per hour, the Swedes bargained about which fairness norm to apply. Elster counted 24 different possibilities that had been proposed at one time or another! My own experience is that people don’t think to question the deep structure of egalitarian norms for resolving fairness problems. They need prompting to consider other possibilities. The controversial issues are how the utilities of different types of people should be compared, and what should count as a correlate for utility in different contexts. These aren’t questions to which metaphysics can make any useful contribution. They are practical questions requiring serious social research, both in psychological laboratories and in the field. You can preach as much as you like about what people ought to regard as fair for this or that metaphysical reason, but people won’t be persuaded if what they are actually accustomed to regard as fair in their daily lives is something different.

5.5.3

Applying Fairness Norms?

This section offers some examples in which the scope of the fairness problem considered is gradually expanded. Its focus is on the obstacles that stand in the way of institutionalizing realistic approaches to fairness. Broken partnerships. Alice and Bob have fallen out after running an antique shop together for a number of years. What is the fair way to divide the assets of the business if they are still on good enough terms to attempt an amicable settlement? The status quo for their problem is their expectation of the outcome of a legal arbitration. Neither party is likely to find such a resort to law attractive unless their erstwhile partner holds out for a split of the assets that the other perceives as unfair. But what factors are relevant to what counts as fair? Here are some of the questions that may need an answer. Who invested how much when the business was set up? How were the yearly profits divided? How many hours did each partner work per day? Who brought what skills to the enterprise? Who attended the rotary luncheons? Who made unwelcome advances to the customers? What are their future prospects after the partnership is dissolved? Who has children to support? The possible issues that one or the other might raise are clearly endless. Alice and Bob may not agree on the answers to all such questions, but they will be better informed than any judge, who can only take into account the plausibility of what evidence is presented in court. If they have been splitting the yearly profits

122

CHAPTER 5. FAIRNESS

equally, a judge may well rule that it is equable that the remaining assets be split proportionally to their initial investment. But it can be seriously costly to go to law—not only because lawyers are expensive, but because of delays and uncertainties about the judgement. It is the potential gains from avoiding such legal costs that are available to be split according to an egalitarian fairness norm—provided that Alice and Bob share a social consensus on how all the issues they think relevant should be summarized by social indices that will be more refined than those a judge can use. Most of us have little or no idea of precisely how our cultural programming allows us to translate a set of contextual parameters into appropriate social indices, because the process operates beneath the conscious level, but we can at least be sure that whatever social consensus Alice and Bob may share will depend on where and when their antique shop was located. It won’t be the same in New York as in Teheran: nor in either city fifty years ago. Regulation. Alice and Bob weren’t assumed to be married in the preceding discussion of the breakdown of a partnership because there seems little prospect of reforming legal practice on financial settlements in divorce cases—even though it sometimes generates outcomes that are so spectacularly unfair that they get headline treatment in the tabloid press. The next example shows that the same obstinate conservatism is often true of economic regulation in those contexts in which the legal profession has become deeply entrenched. There is a legal requirement that bargaining over royalties for the use of patents for smartphones and the like should result in a fair outcome. The relevant judgement calls for ”fair, reasonable, and nondiscriminatory (FRAND) royalties for standardessential patents for smartphones and other wireless devices”. Judge Posner is a public intellectual informed by the famous Chicago School, where neoclassical economics was reborn in the 1930s. He has brought economic enlightenment to the legal profession across a whole range of regulatory issues, but the fact that the Chicago School were dismissive about the relevance of fairness to economic issues perhaps lingers in his 2012 judgement that the word “fair” adds nothing to “reasonable and nondiscriminatory” in the definition of FRAND. 24 Perhaps Posner’s judgement is sound while no relevant legal definition of fair is available for this context, but the way forward isn’t to pretend that fair doesn’t mean anything beyond nondiscriminatory, but to establish a viable definition of what should count legally as fair by building up suitable precedents on what social indices are appropriate in different contexts on a case-by-case basis. The FRAND case typifies a general move toward taking account of fairness in economic regulation, but lawyers are dismissive of the idea that the role of social history in determining social indices needs to be replaced by an artificial legal history that determines what social indices are relevant for different industries. They want a simple fairness formula that can be applied across the board. But you can’t 24 Richard Posner made this judgement sitting as trial judge in Apple, Inc. v. Motorola, Inc. See Gregory Sidak: “What Makes FRAND Fair? The Just Price, Contract Formation, and the Division of Surplus from Voluntary Exchange” 4 Criterion Journal on Innovation 701 (2019).

5.5. SOCIAL CONTRACTS

123

appeal to people’s intuitions of what counts as fair without paying attention to the empirical fact that people regard different things as fair in different contexts. National constitutions? Jeremy Bentham got nowhere when he offered to rewrite the American constitution for James Madison. He was treated with more respect by Simon Bolivar when he volunteered to write a constitution for the megastate Bolivar hoped to create in South America, but the political realities of the time trumped their utopian aspirations. I think the same will always be true in attempting social engineering on such a grand scale, but I would like nevertheless to sketch briefly how Epicurus might have structured the problem. Von Clausewitz was a Prussian military theorist who famously argued that war is merely diplomacy carried on by other means. Epicurus wouldn’t have agreed at all. For him, diplomacy was an alternative to war. He would therefore have said that a constitution is ultimately an attempt to provide a structure for the peaceful settlement of differences that will inevitably arise within a nation in the future. From a game theory perspective, a constitution can be seen as part of the equilibrium selection device that a society uses to refine the details of its social contract to meet new challenges and changing attitudes. But a constitution is itself part of a society’s overall social contract. Honoring the constitution must therefore be part of what is involved in operating whatever equilibrium in the game of life has been chosen for a society by its history. Constitutional issues require thinking about a two-stage process. The first stage consists of a choice of constitution. In seeking some compromise to avoid strife, conflicting coalitions recognize that unpredictable events are likely to empower different groupings in the future. Each group therefore accepts that it will need to tolerate other groups when lucky enough to be in power in return for being tolerated when out of power. But this is the setting for Rawls’ original position (Section 5.2.1).25 The outcome will therefore necessarily have some of the trappings of a fairness norm, but after a civil war, nobody will be surrendering any bargaining power they may have at their disposal in favor of a fair settlement. It is only when a constitution is being put together by like-minded folk who have learned the lessons of history—as in the case of the American constitution—that an appeal to fairness is likely to have any traction. It is here that the egalitarian norm has some chance of being applied—with whatever standard of interpersonal comparison history has gifted the population at that time and place.26 The second stage that follows the choice of constitution is determined not only by the particular laws passed by a nation’s legislative bodies and the precedents established in its courts of law, but also in large part by the workings of social history in changing our attitudes—for example, to slavery or homosexuality. It would be necessary to take on the almost impossible task of modeling this second 25 James Buchanan and Gordon Tullock: Calculus of Consent. Buchanan and Tullock speak of a veil of uncertainty rather than a veil of ignorance because rival groups don’t have to pretend that they might sometimes be in or out of power. People sometimes wonder why only Buchanan was rewarded with a Nobel Prize in 1986. 26 The original American constitution notoriously counted slaves as three-fifths of a free person for the purpose of calculating representation in the House of Representatives.

124

CHAPTER 5. FAIRNESS

stage in order to approach the first stage in a scientific way. Instead of evaluating the life prospects of a single individual, as in Chapter 3, we would need to evaluate the life prospects of a whole society. When contemplating the immense difficulties in following through such a program scientifically, it is tempting to throw in the towel. But the observations of Section 3.1.3 apply here too. Nobody will be remotely satisfied with the vastly simplified models necessary to make a rational analysis feasible, but the alternative is to abandon rationality altogether when thinking about constitutional issues. Global warming? Rationality isn’t likely to get us anywhere on this biggest of all issues. Even the the current intellectual consensus on how to evaluate the possible futures we might face is dubious (Section 3.2.2). Pointing out that global warming is an example of the Tragedy of the Commons merely generates more denials of the logic of the Prisoners’ Dilemma (Section 4.2.6). As for fairness being relevant to any serious international effort that might eventually be made to do something effective about global warming, one might as well cry for the moon. Something will doubtless be done eventually—much too little and much too late to avert disaster—and rhetorical appeals will be made to fairness when decisions need to be made on which nations contribute what share to the enormous capital cost. But there is no consensus on what social indices should apply, and even if there were, fairness won’t be allowed to displace the realpolitik of hard-nosed bargaining. To add to this gloomy prediction, the theory of rational bargaining with a deadline doesn’t even guarantee that agreement will be reached at all. The best we can hope for is some major technical advance, but time is getting very short.

5.5.4

Universal Health Care

To end on a more upbeat note, this final section considers the relevance of fairness norms to health care. On an adventure holiday in Inner Mongolia, one of our party fell off her camel and cracked her ribs. She was charged $25 in cash for treatment at an excellent hospital not too far away. The native Chinese were apparently also charged up-front, but had organized themselves into local cooperatives that insured them against illhealth or accidents. Such insurance arrangements are a real-life example of what Section 5.2.2 argues lies behind the notion of the original position; Alice and Bob pay their premiums, not knowing in advance who is going to turn out to be lucky or unlucky in the health lottery. In many countries, these Chinese cooperatives have been expanded to provide universal coverage to every citizen, with the premiums absorbed into the general taxation system. How should such universal health-care systems be organized? Independent health-care institutions? It isn’t impossible that some enlightened country might set up a such a universal health service largely funded by the government, but operated independently of government. Why independent? For the same reason that central banks nowadays usually operate independently of

5.5. SOCIAL CONTRACTS

125

government—so that they can pursue their long-term responsibilities without being used as footballs to be kicked about for short-term political advantage. An independent health service would need a constitution. How might such a constitution be designed? In considering the question, I shall be guided by a recent report from the World Health Organization (WHO) that offers a minimal consensus on what health gurus think acceptable.27 Governance: With an election in the offing, the governing party in my own country just touted the idea of building 40 new hospitals. If they were really serious, why spend the money on hospitals, and not on whatever else health professionals think should have priority? Because new hospitals are what attract the floating voter. Voters are similarly sensitive to delays in treatment, so fines are imposed on hospitals whose patients are made to wait for more than some specified period. Hospitals are then forced to divert money from other priorities to meet an arbitrary target chosen for populist reasons. The WHO report offers no suggestions on how to cope with irresponsible political interference in the provision of health services beyond observing that “public accountability and participation should be institutionalized”. My own preference for the governance of an independent health-care institution would have the governing board selected by an electoral college as originally envisaged in the American constitution. But populists would prefer the board to be directly elected, and so such a college would eventually be corrupted into a convocation of delegates with irrational axes to grind. The alternative is for the board to consist of health gurus appointed by the government—an arrangement that works well enough with central banks provided the appointments are for periods that are long by political standards. As always, these appointed guardians would themselves need to be guarded (Section 2.6). We tend to think of doctors as beyond criticism, but their professional organizations look after their self-interest no less assiduously than other trade unions. However, I don’t want to dwell any more on the possible governance of a heath-care institution beyond observing that medical decisions need to be made by medics and not by administrators in some remote office (Section 2.1). Efficiency: The WHO report talks about cost-effectiveness, but doesn’t recognize that co-pays—which it disapprovingly calls out-of-pocket payments—are necessary to discourage irresponsible demands for publicly provided health care, although nobody would want co-pays to be larger than strictly necessary (Section 2.2). However, the immediate point is that efficiency in the economic sense needs to prevail over fairness considerations. Fairness: I am most interested in how a national health-care institution would use fairness or equity considerations to assign priorities to the treatment of different health problems. Sometimes utopians insist that all perceived needs for health care should be met, but the demand for health services is always going to exceed the supply. It is 27 Trygve Ottersen, Ole Norheim, and twenty others: Making Fair Choices on the Path to Universal Health Coverage. Also instructive was my service with a ginger group whose conclusions are reported in Will Hutton’s: New Life for Health: The Commission on the NHS.

126

CHAPTER 5. FAIRNESS

currently rationed in the United States largely by price. But how is rationing to be organized when the state provides health care supposedly for free? In Britain, it is customary to deny that there is any rationing at all, while actually rationing health care by making people wait for treatment. However, I think people would perhaps accept the need to ration health care if it were done on what ordinary people regard as a fair basis. Among other positive factors, the indices of healthiness currently in use are reasonably objective, and would serve well as a correlate of expected utility. On the other hand, empirical research would be necessary into how ordinary people make interpersonal comparisons of utility—how they would assign social indices to different types of people when making small-scale fairness judgements. Should an old lady’s hip operation take priority over open-heart surgery for someone who has lived recklessly all his life? Should mental health treatment for teenage dropouts get more resources per head than regular gynecological care for pregnant middle-class housewives? Such questions are difficult, but it is better they be addressed directly rather than simply allowing the dice to roll where they will. But should the assignment of priorities be made on a utilitarian or an egalitarian basis? On this subject, the WHO report joins a number of other commentaries in falling victim to the idea that the choice really lies between naive versions of Utilitarianism, Egalitarianism, or Prioritarianism proposed by metaphysicians who believe that their academic status renders their Moral Intuitions more worthy than those of ordinary folk (Section 5.3). Worse still, the report can’t make up its mind whether it favors Utilitarianism or Prioritarianism, since it argues both for the need to “maximize total benefits across all people in society” and also to offer “priority to people with the least coverage”. Neither of these alternatives is likely to appeal to the patients whose interests a universal health-care institution is intended to serve, because they will evaluate the fairness of various possible rationing schemes using the fairness norms they actually use in their everyday life. If I am right about how fairness norms work, they will therefore evaluate the gains (or losses) over the status quo resulting from a reform using the egalitarian norm with whatever social indices they regard as appropriate for different kinds of patient. In particular, they are likely to reject the ruthless application of Utilitarian principles, for the same reason that people react badly to the story of the missionaries and the cannibals (Section 2.5.3). They certainly don’t like current services being cut back to fund what administrators regard as more urgent priorities. They are likely to accept that priority should be given to whoever needs treatment most, but only because this is a step on the road to achieving an egalitarian outcome for all. In brief, if people are to regard a national health-care institution as existing for the personal welfare of ordinary folk like themselves, it needs to operate using the same egalitarian norms that they use in ordinary life.

Index absolute, 11, 19 Academy, 4 Aesop, 58 Agora, 4 Akbar the Great, 75 Alexander the Great, 3, 48 altruism, 84 American Declaration of Independence, 116 Andersen, Hans Christian, 88 anger, 89 Archimedes, 12 Argument by Design, 41 Aristarchus, 12 Aristippus, 6, 8, 38, 67 Aristotle, 2, 5, 12, 60, 103, 107 ataraxia, 15, 22, 59, 69 Athens, 3, 4 atoms, 12 Aumann, Robert, 87 authority, 11, 90 Axelrod, Robert, 86, 87 backward induction, 43, 61 Bailey, Cyril, 6 bargaining problem, 109 Bayes’ rule, 40 Bayes, Thomas, 40 Bayesian decision theory, 104 Bayesian rationality, 58 Bayesianism, 40–42 behavioral economics, 42, 47 behaviorism, 35 Bentham, Jeremy, 2, 7, 27, 33, 43, 45, 116, 123 Bishop Butler, 11 blank slate, 115 Boehm, Christopher, 108 Bolivar, Simon, 123 Boswell, James, 22 Buchanan, James, 123 Buddha, 22, 97 Bunyan, John, 23 Burke, Edmund, 117 Burton, Robert, 5, 12, 24, 120 Camus, Albert, 68 categorical imperative, 17, 73, 116

Catullus, 6 causal utility fallacy, 33 Chess, 56 Chicago School, 122 Chomsky, Noam, 103 Cincinnatus, 92 Coleridge, 7 collective rationality fallacy, 77 common knowledge, 43 Confucius, 71, 97 consequentialism, 57, 84 consistency, 33, 34, 36 constitution, 90, 117 convention, 20, 89 convex, 105 Cornell, 118 Council of Nicaea, 19 Cournot, Auguste, 110 d’Aquili, Eugene, 23, 69 Dante Alighieri, 2 Darwin, Charles, 12, 78, 84, 114 Dawkins, Richard, 77, 111 death, 13, 55, 59, 67 deep structure, 103 degree of relationship, 81 deism, 17 Democritus, 4, 12, 23 Dennett, Daniel, 16 deontology, 57, 84 Descartes, Ren´ e, 10 Dimetriadus, Hans, 2 Diogenes, 4, 5, 10 Diogenes Laertius, 6 Dirac, Paul, 11 discount factor, 63 divine right of kings, 93 dominant strategy, 76 Dostoyevsky, Fyodor, 72 Driving Game, 89, 101 Dunbar, Robin, 24 efficiency, 47, 99, 118 egalitarian bargaining solution, 108 egalitarianism, 49, 104, 106, 107 electoral college, 4

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020 K. Binmore, Crooked Thinking or Straight Talk?, https://doi.org/10.1007/978-3-030-39547-6

127

128 Elster, Jon, 113, 121 emotions, 88 empathetic preferences, 46, 98, 103, 111, 112 empathy, 13, 46 empiricism, 8 enlightenment, 22 Epictetus, 24 Epicureanism, 2 Epicurus, 1, 2, 4, 6, 15, 17, 20, 21, 24, 25, 37, 44, 55, 58, 62, 67, 89, 114 equilibrium, 20, 31 Eratosthenes, 12 Euclidean geometry, 9, 11, 14, 42 evolutionarily stable strategy, 86 evolutionary ethics, 95 expectation, 35 expected utility, 36, 39 exponential discounting, 63 extensive-form game, 56 external enforcement, 106 fairness norm, 44, 47, 112, 113 faith, 10 fallacy of the twins, 79 first-best, 31, 51 fitness, 81 folk theorem, 72 Fool’s Mate, 56 Ford, Henry, 3 FRAND, 122 free rider problem, 82 free will, 16 G¨ odel, Kurt, 11 Galileo Galilei, 12, 41 game of life, 20, 118 game theory, 31 game tree, 56 Garden, of Epicurus, 4, 8, 24 Gaskin, James, 2 Gassendi, Pierre, 2 Gauthier, David, 78, 109 General Will, 21, 95, 116 ghost in the machine, 13 Glaucon, 28 global warming, 55, 64, 69, 93 Global Warming Game, 93 Glorious Revolution, 115 golden age of the poets, 3 golden rule, 97, 100 Good, the, 17, 21, 43, 84, 105 grim strategy, 85 Grotius, Hugo, 114 group selection fallacy, 77, 118 guarding the guardians, 52, 91, 92 Hamilton’s rule, 80, 102

INDEX Hamilton, William, 80, 81 Hardin, Garret, 83 Harsanyi, John, 31, 45, 48, 99, 104 health care, 124 hedonism, 6, 37 Helvetius, 7 hemlock, 15 herd behavior, 24 Herodotus, 19 Hillel, 97 Hobbes, Thomas, 2, 32, 46, 80, 93, 106, 115 holism, 34 Homans, George, 108 Hume’s Fork, 9 Hume, David, 2, 8, 9, 17, 20, 22, 29, 33, 37, 43, 45, 72, 85, 90, 114 Hurwicz, Leonid, 30 Hutton, Will, 125 Hymenoptera, 81 hyperbolic discounting, 63 hypothetical imperative, 17 ideal observer, 48, 64 income stream, 61 Independence Axiom, 36 information set, 57 interest rate, 63 interpersonal comparison, 45, 48, 108, 113, 126 Inwood, Brad, 25 ipsedixist, 27 Ithica, 118 Jackson, Andrew, 52, 115 Jacobins, 116 Jainism, 23 Jesus, 2, 20, 97, 101 Johnson, Samuel, 22 Judgement of Solomon, 29, 91 justice, 19, 28, 113, 117 Juvenal, 52 Kagel, John, 83 Kant, Immanuel, 9, 20, 57, 60, 73, 91, 95, 116 Kardashians, 25, 90 Keats, John, 11 Kierkegaard, Soren, 55 knowledge-as-certainty, 10, 43 knowledge-as-commitment, 10, 43, 57 Kuhn, Thomas, 60 language, 103 leadership, 111 Ledyard, John, 83 Leibniz, Gottfried, 10 Leucippus, 12 Lewis, David, 20 Lincoln, Abraham, 67, 115

INDEX Locke, John, 2, 20, 84, 106, 115 Lovelace, Richard, 24 Lucretius, 6, 15 Lyceum, 4 Madison, James, 51, 93, 121, 123 Marcus Aurelius, 92 Marx, Karl, 18, 77 Maskin, Eric, 30 maximin criterion, 104 means and ends, 57 mechanism design, 30 Medicare, 31 meme, 19, 111 Meno, 14 metaphysics, 10 Metzinger, Thomas, 16 Mill, John Stuart, 7, 33, 43, 45, 57 Milnor’s model, 67, 70 Milnor, John, 67 miracle, 17 model, 9, 16 modern equity theory, 108 Mohammed, 97 Moore, G. E., 17 Moral Commitment, 106 Moral Intuition, 21, 95, 105 Morgenstern, Oskar, 36 mutual consent, 49, 109, 115 Myerson, Roger, 30 myth of the wasted vote, 79 Mytilene, 4, 18 Napoleon Bonaparte, 53 Narens, Louis, 7, 105 Nash bargaining solution, 110, 112 Nash equilibrium, 31, 76 Nash, John, 31, 110 Natural Duty, 106 Natural Law, 8, 21, 95, 103 Natural Right, 106, 115 naturalism, 18 naturalistic fallacy, 17 Nature abhors a vacuum, 18 nebula hypothesis, 95 neoclassical economics, 42, 47 neuroscience, 23 Newberg, Andrew, 23, 69 Newton, Isaac, 95 Nobel Prize, 27, 30, 31, 34, 87, 123 Norheim, Ole, 125 objective, 11, 19 Occam’s Razor, 33 oligopoly, 120 Omar Khayyam, 22 one-shot game, 84

129 original contract, 114 original position, 99, 103, 106 Orwell, George, 28 Ottersen, Trygve, 125 ought implies can, 17 Paine, Thomas, 117 Parcheesi, 75 Pareto, Vilfredo, 47 Parfit, Derek, 66, 104 Paton, Herbert, 74 Peloponnesian War, 3, 25 perfect equilibrium, 91–93, 110 Peripatetics, 4 Peterson, Martin, 78 Phaedo, 14 Philodemus, 6, 21 Pinker, Steven, 103, 115 Plato, 1, 5, 10, 14, 19, 27, 28, 64, 115 Poker, 56 Pompey, 15 Popper, Karl, 2, 28 Posner, Richard, 122 Practical Reason, 21, 95 Price, George , 56 principal-agent problem, 30, 50 Prisoners’ Dilemma, 75, 76, 78, 79, 81 probability, 35, 36 problem of evil, 18 proportional bargaining solution, 108 public good, 82 Public Goods Game, 82, 83 Pufendorf, Samuel von, 114 Pyrrhonism, 23 Pythagoras, 12, 14 Ramsey, Frank, 40 rationalism, 8 rationality, 33, 77 Rawls, John, 11, 69, 99, 104, 115 reciprocal altruism, 84, 96 reciprocity, 71, 79, 85, 87 reciprocity principle, 86, 89, 91, 100, 106, 114 reductionism, 33 referendum, 4 reflective equilibrium, 11 regulation, 122 relative, 19 relativism, 18, 19 repeated game, 72 repeated Prisoners’ Dilemma, 85 Republic, Plato’s, 11, 15, 19, 27, 57, 110 Republicans, 115 repugnant conclusion, 65 Right, the, 21, 43, 84 Robespierre, Maximilien, 116, 117 Roth, Alvin, 27, 83

130 Rousseau, Jean-Jacques, 20, 93, 116 Rubinstein, Ariel, 110 Russell, Bertrand, 2, 10, 18, 116 Sally, David, 83 Samuelson, Paul, 34 Savage, Leonard (Jimmy), 40 Schelling, Tom, 20 scientific induction, 10, 43 second-best, 31, 52 Selten, Reinhard, 31, 91 Sen, Amartya, 42 Serapis, 19 Sextus Empiricus, 23 Sharing Game, 96, 109 Sidak, Gregory, 122 Singer, Peter, 101 Sisyphus, 68 Skyrms, Brian, 7, 105 small world, 39, 41, 58 Smindurides, 6 Smith, Adam, 2, 45 Smith, John Maynard, 56 Smith, Joseph, 19 social consensus, 98, 108, 112 social contract, 20, 89, 113, 118 social dilemma, 83 social discount factor, 64, 70 social engineering, 27, 84, 93, 97 social index, 44, 112, 122 Socrates, 5, 8, 11, 14, 28, 59 soul, 13, 14 Sparta, 3, 28, 115 Spinoza, Benedict de, 2, 3, 16, 74, 103 St Augustine, 1 St Francis of Assisi, 46 Stalin, Joseph, 92 state of mind, 61 state of nature, 115, 116, 118 state of the world, 62 status, 25 status quo, 109, 118, 121 Stern report, 64 Stoa, 4 Stoicism, 4, 8 strains of commitment, 105 strategy, 57 Strodach, George, 2 subgame, 91 subjective time, 66 suicide, 15, 68 sure-thing principle, 40, 76 Swinburne, Algernon, 68 Sybaris, 6 sympathetic preferences, 46 sympathy, 46 synthetic a priori, 9, 14

INDEX tax code, 50 Tetrapharmakos, 21 Thomas, Dylan, 22 Thrasymachus, 109 three-fifths compromise, 123 tit-for-tat strategy, 86 Tories, 114 Tragedy of the Commons, 75, 77, 83 tranquility, 8, 22, 25, 37, 69 transparent disposition fallacy, 78 tree of knowledge, 55 tree of life, 55 Trivers, Robert, 80, 84, 87 Tullock, Gordon, 123 tyranny of the thirty tyrants, 3, 28 Usener, Hermann, 2 util, 37 utilitarianism, 7, 43, 48, 49, 104 utility, 34 utility stream, 62 vampire bats, 84, 95, 100 veil of ignorance, 99 Victor Hugo, 67 Villa of the Papyri, 6 Virgil, 6 virtue, 8 Voltaire, 58 von Clausewitz, Carl, 123 von Neumann, John, 22, 31, 36, 57, 65 Wagstaff, Graham, 108 Warren, James, 2 wasted vote, 79 Whigs, 114 Wilson, Catherine, 2 Wittgenstein, Ludwig, 24 World Health Organization (WHO), 125 Wynne-Edwards, Vero, 77 Xenophanes, 18 Young, Peyton, 113 Zeno of Citium, 4, 5 Zeno of Elea, 4 Zeno of Oenoanda, 6 Zoroaster, 97