close

Вход

Забыли?

вход по аккаунту

?

Joel Marks - Ought Implies Kant- A Reply to the Consequentialist Critique (2009)

код для вставкиСкачать
Ought Implies Kant
Ought Implies Kant
A Reply to the
Consequentialist Critique
JOEL MARKS
LEXINGTON BOOKS
A division of
ROWMAN & LITTLEFIELD PUBLISHERS, INC.
Lanham • Boulder • New York • Toronto • Plymouth, UK
LEXINGTON BOOKS
A division of Rowman & Littlefield Publishers, Inc.
A wholly owned subsidiary of The Rowman & Littlefield Publishing Group, Inc.
4501 Forbes Boulevard, Suite 200
Lanham, MD 20706
Estover Road
Plymouth PL6 7PY
United Kingdom
Copyright © 2009 by Lexington Books
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of the publisher.
British Library Cataloguing in Publication Information Available
Library of Congress Cataloging-in-Publication Data
Marks, Joel, 1949–
Ought implies Kant : a reply to the consequentialist critique / Joel Marks.
p. cm.
Includes bibliographical references and index.
ISBN-13: 978-0-7391-2877-0 (cloth : alk. paper)
ISBN-10: 0-7391-2877-9 (cloth : alk. paper)
eISBN13: 978-0-7391-3352-1
eISBN10: 0-7391-3352-7
1. Kant, Immanuel, 1724–1804. 2. Ethics. 3. Consequentialism (Ethics) I.
Title.
B2799.E8M365 2009
193—dc22
2008040442
Printed in the United States of America
⬁ ™ The paper used in this publication meets the minimum requirements of American
National Standard for Information Sciences—Permanence of Paper for Printed Library
Materials, ANSI/NISO Z39.48-1992.
To Huibing, who told me to write a book
Contents
Preface
ix
Acknowledgments
xiii
Introduction
1
1
Egoism
7
2
The Consequentialist Continuum
17
3
Let Us Boldly Go: The Case for Utilitarianism
27
4
Refutation of Consequentialism
39
5
Nonconsequentialism and the Consequentialist Critique
57
6
The Ethics of Ethics
71
Appendix I: What are we talking about? (What is ethics?)
89
Appendix II: A simple theory (What is theory?)
93
Appendix III: Animal ethics
97
Glossary
101
Bibliography
105
Index
111
About the Author
115
vii
Preface
There are more things in heaven and earth, Horatio,
Than are dreamt of in your philosophy.1
—Hamlet (I, v, 166–167)
There are more philosophies, I venture to suppose, than are dreamt of in
your philosophy. Certainly most people assume that the way they see
things is the way things are. But even philosophers, whose business is to
apprehend what lies beneath appearance, are likely to have a limited conception of the possibilities. Perhaps this is inevitable if the possibilities are
limitless. The possibilities of what? In this book I will be discussing ethical theories, the ways in which philosophers have sought to explain why
we have moral obligations, and why we have the ones we do. In Chapter 2
I will introduce a scheme for generating new ethical theories ad infinitum.
And for what purpose will I do this? I cannot deny: I find the inquiry to be
intellectually stimulating. But beyond my own motivation or even any intrinsic justification, I think the study of ethics is useful. For what? For
making us more ethical, that is, more likely to know what our most fundamental obligations are and to be motivated to fulfill them . . . or so I shall
argue in this book.
It might seem that having so many ethical theories would be a deficit. Don’t
we already have enough of them to leave us hopelessly confused about what
is the right thing to do or the best way to live or the kind of person we should
strive to become? Oh yes, if knowing which of them is The True Theory were
ix
x
Preface
necessary to accomplish these things. And that’s how it sometimes seems.
Here is a standard form of ethical reasoning:
The right thing to do is that which satisfies criterion C.
Of all the available options, only o satisfies C.
Therefore do o.
The first premise enunciates a theory of ethics, for example, utilitarianism,
according to which C is the maximization of good under the circumstances.
But if there are competing ethical theories, such as Kantianism and feminism,
then we are not in a position to know whether (we ought) to do o until we determine which theory is the correct one since the others might advise that we
need not or even ought not to do o. If we were now to multiply the number of
alternative theories indefinitely, the resolution of our particular ethical question about whether to do o would itself appear to recede indefinitely.
That is just the impression people can form when introduced to the philosophical study of ethics. Ethics (that is, its formal study) may strike us as useless, or worse: downright discouraging. Yet my own experience has been
quite the opposite; over a period of decades I have come to appreciate the ethical wisdom that can be acquired from the study of ethical theories . . . and
even of ethical theory as such, so-called meta-theory—regardless of whether
we can prove which (if any!) theory is true.2 So while I will indeed attempt
to demonstrate the superiority of one theory in particular, my main hope is
that by the end of this book you will join me in saying: I never met a theory
I didn’t like!3
NOTES
1. This passage is variously understood with the emphasis on “your” or “philosophy.” Either will do for my purposes.
2. The above paradigm of ethical reasoning has been challenged more directly as
putting the cart before the horse or being simply unnecessary. For example, Bryan G.
Norton (1955) argues for what he calls practical philosophy, which “works toward
principles by struggling with real cases rather than establishing theory and ‘applying’
it to real cases” (p. 126); furthermore, the principles need not be at the level of ultimate
theory but only consensus agreement among the disputants. Tzachi Zamir (2007) employs an even simpler methodology of establishing ethical conclusions by reflection on
and “extension” (p. 18) of “widely shared . . . general assumptions” (p. 16). I am completely sympathetic to such approaches to practical or urgent ethical problems as contributing to their resolution. However, neither Norton nor Zamir would disagree that
“pure theory” also has its place in the proper scheme of things, and this book will pro-
Preface
xi
ceed in that spirit. Indeed, let us not forget that theories of extreme abstraction can
themselves have profound impact, and not only in the physical sciences; for example,
Lee Hall (2006) notes that “Social justice movements everywhere find guidance in the
idea that another world is possible, and that once an idea can be conceived, it can be
achieved. Theories can indeed be put into practice overnight . . .” (p. 137). In sum: As
my first academic mentor—the late perception psychologist and theorist J.J. Gibson,
quoting his mentor, the Gestalt psychologist Kurt Lewin—used to say, “There is nothing so practical as a good theory” (personal communication).
3. Pace Will Rogers. As the author will indulge in the occasional play on words,
he hopes that by the end of this book you will not join Oliver Wendell Holmes in saying: “if a blow were given for [punning], and death ensued, the jury would be judges
both of the facts and of the pun, and might, if the latter were of an aggravated character, return a verdict of justifiable homicide” (The Autocrat of the Breakfast-Table,
ch. 1). I note that in the course of his argument the narrator himself makes a pun, to
wit: “Homicide and verbicide–that is, violent treatment of a word with fatal results to
its legitimate meaning, which is its life–are alike forbidden. Manslaughter, which is
the meaning of the one, is the same as man’s laughter, which is the end of the other.”
Acknowledgments
There is always a book behind the book, so how does one sum up four
decades of intellectual indebtedness in a few pages? For this reason the acknowledgments are difficult to write, but, nonetheless, it is my pleasure to
recognize so many good friends and colleagues who have accompanied me
on this philosophic exploration.
Since my entire professional career has been tending towards this book, its
genesis is the graduate school where my formal education in philosophy began. After all these years the faculty of the University of Connecticut
(“UConn”) continue to be my mentors, especially the three Js—Joel Kupperman, Jerome Shaffer, and Jack Davis—all of whom reviewed the manuscript
with the same critical eye, and practically parental heart, that they applied to
my doctoral dissertation a quarter-century ago.
Let me look back even further to acknowledge the enduring influence of
the late perception psychologist James J. Gibson, whose graduate seminar at
Cornell provided undergraduate me with my first exposure to the nitty-gritty
of theoretical work. There was also a subsequent interlude at Cornell when I
received my very first (and glorious) exposure to professional philosophy by
the good graces of Keith Donnellan, Gail Fine, Sidney Shoemaker, Robert
Stalnaker, and Allen Wood.
The years since UConn have been spent primarily at the University of New
Haven (“UNH”), where the normal teaching responsibility of eight introductory
courses per year was “too much of a good thing” and necessitated a creative
shift away from scholarly articles and monographs to “moral moments” for
newspapers and magazines. In this I have received an invaluable assist from the
good offices of Charles Kochakian, editorial page editor, and Abram Katz, science and health editor, both of the New Haven Register, and Rick Lewis and
xiii
xiv
Acknowledgments
Anja Steinbauer, editors of Philosophy Now magazine. The latter has furthermore provided a venue for explicitly philosophical reflection on every topic under (and above and including) the sun, and always under Rick and Anja’s sure
and pleasant stewardship. Much of the preliminary thinking for this book occurred on the pages of that magazine.
The double discipline of reaching out to non-majors in the classroom and
the general reading public beyond has made me a better writer and philosopher. The simple rule of thumb is never to assume that your interlocutor or
your reader knows anything at all about the subject you are discussing. The
effort to explain it therefore obliges you to understand it fully yourself. As a
result, questioning and clarifying become second nature.
Yet during the whole of this last quarter-century I have also been able to
maintain one fully “advanced” dialogue on one topic in particular with my
UConn classmate, Mitchell Silver. The topic is consequentialism, which
somehow consumed me. Mitchell actually took most of the initiatives to encourage our conversation, inviting me for stays at his home in Greater Boston
and, in summertime, visits to Camp Kinderland in rural Massachusetts; at
other times we would meet halfway at our old stamping ground of Storrs,
Connecticut. I wonder how many miles we walked as we talked all told?
But that was not the end of it. Mitchell went into high gear when he heard
of my intention to write a book on the subject. He urged me to start at once.
Amazingly I was already approaching the end of my UNH career and was
about to begin a sabbatical to boot, for which I thank the institution. Mitchell
felt that any other catching-up projects I might attend to instead should be put
on the back burner until this one was done. He had become convinced that my
take on the issue was valid and valuable and wanted me now to think it
through in sufficient detail and put it in the proper form to introduce to the
profession at large.
A particular day arrived when I was catalyzed to sit down at my computer
and begin to write. From that point on, until the completion of the first draft
and then the final one, Mitchell has been my constant interlocutor—by phone,
via email, and in person. As good a friend as Mitchell had always been, he
now came into his own—or we came into our own. He even took the whole
manuscript (I am tempted to write “MS,” since these are also Mitchell’s initials) on his vacation to Israel. I sincerely asked Mitchell if I should list him
as co-author, but he graciously replied that the book was mine.
But Mitchell was not my only constant interlocutor during this period. I
have been blessed with a second as well, Allan Saltzman. Our dialogue has
been informal and extra-professional, but has made up for whatever shortcoming that may seem to imply, with the continuity of Allan’s intuitive, incisive, and supportive commentary.
Acknowledgments
xv
For me the experience of writing this book has been utterly absorbing and
exhilarating: problem after problem arising and then each being solved . . .
until all the parts finally fit into place. I had expected the process to consume
at least a couple of years, but the first draft emerged in a little more than a
month as if on one continuous scroll, like Jack Kerouac’s On the Road.1 There
is no question that the ideas had been pent up inside for a decade and more
and so, even before I sat down at the computer, were pretty much “written”
except for the tapping of the keys. Of course in the end a couple of years have
elapsed after all as I revised, corrected, supplemented, and otherwise edited
the manuscript for publication.
The writing of the book has also coincided with a new affiliation for me
with the Interdisciplinary Bioethics Center at Yale University. I was first
drawn in, due to sheer interest in topics and speakers, to the monthly meetings of the working group on technology and ethics, chaired by Wendell Wallach. Wendell himself has now become a friend and teacher, who encouraged
me to write this book on what he considered to be a novel thesis. With his
guidance I was also able to initiate a colloquium series at the Center on animal ethics. The latter topic has been an enduring concern of mine, but now I
was given a regular forum to examine it; and that has in turn contributed to a
central theme of the present book. Carol Pollard, Associate Director of the
Center, and David Smith, Director, have been the most supportive allies
imaginable.
One of the most stimulating, and indeed provocative, presentations in Wendell’s group was given by Joshua Greene (now of Harvard’s Psychology Department). I owe him a special debt as this book fulfills his anticipation that I
would “of course, write volumes-long explanations for why this is a silly
question and why this [brain research on ethical decision-making] is all just
too simplistic” (as he put it in a personal communication).
Another invaluable interlocutor has been Lee Hall, legal director of the
U.S.-based Friends of Animals. Lee has taught me everything I know about
animal ethics, and our complementarity of practical and theoretical interests
has proved extremely productive. (Our dialogue is also testimony to the Internet Age, since Lee and I have only “met” via email.) I also benefited from
the opportunity to canvass my thoughts on the subject in a talk to the Hartford Ethics Group.
My thanks and appreciation also go to the editorial staff at Rowman and
Littlefield and Lexington Books, who have been unfailingly competent and
supportive. It was Patrick Dillon’s faith in this project that sustained it, and
Jessica Bradfield, Julia Loy, Paula Smith-Vanderslice, and Matthew McAdam
who saw it through to fruition. An anonymous reviewer made numerous astute observations and provided helpful suggestions.
xvi
Acknowledgments
Meanwhile I have depended on the extremely accommodating assistance
of the University of New Haven’s Marvin K. Peterson Library staff for access
to research sources. I would like to single out Evelina Woodruff, Library Reference Assistant, for her efficient handling of interlibrary loans.
Finally I want to thank the following good people for their review of the
manuscript and/or discussions through the years of relevant topics, as well as
for their steadfast encouragement: Tony Barrand, Howard Benditsky, Robert
Blair, David Brubaker, Mary Coykendall, Bill and Stella and Neda DeMayo,
Harvey Green, Darrell Harrison, Huibing He, John Lepore, Arthur S. McGrade, Michael Morris, Nora Porter, Stephen Rocketto, Erik Rosenthal,
Melanie Stengel, Tom Toleno, and my dear departed friends and colleagues
Tom Holahan and David Morris.
NOTE
1. As related in Gerald Nicosia, Memory Babe: A Critical Biography of Jack Kerouac (Berkeley: University of California Press, 1994). Alas, I doubt that my electronic
folder will fetch $2.43 million as Kerouac’s paper scroll did in 2001.
Introduction
“How often have I said to you that when you have eliminated the impossible, whatever remains, however improbable, must be the truth?”
—Sherlock Holmes in The Sign of Four (1890, ch. 6)
by Sir Arthur Conan Doyle
This book offers a defense of the ethical theory of Immanuel Kant, or Kantianism, and the extension of that theory to encompass moral regard for nonhuman animals; however, most of the book will be concerned with consequentialism. Consequentialism holds that the rightness or wrongness of an
action1 is determined by its consequences. Kantianism, by contrast, claims
that the core of ethics is to treat all persons (or, on my view, all living beings)
as ends-in-themselves. I say “by contrast” because the consequentialist criterion would seem to permit, indeed require, treating persons (not to mention,
other animals) differently, that is, not as ends-in-themselves, if this would result in a better outcome. For the Kantian, however, the end could never justify the means since the Kantian criterion is precisely never to treat any person or being as a mere means.2
The reason a book intended to defend Kantianism must be largely concerned with consequentialism is that any ethical theory that purports to be
based on something other than consequences must overcome the strong intuition that one ought always to do what is for the best. To take a classic example from Kant himself: If the choice were between, on the one hand, lying
about the whereabouts of your innocent friend to an evil killer who was intent upon murdering him, and thereby treating the killer merely as a means to
the good outcome of saving your friend’s life, and, on the other hand, revealing your friend’s hideout to the killer, who in his right (and moral) mind
1
2
Introduction
would opt for the latter?3 The intuition is so deeply felt that consideration of
the consequences must and should ground our ethical thinking, that even defenders of Kant retain it as a heartfelt dream. Somehow, somewhere (if only
in a heaven in which we may believe for precisely this reason), good will
come of our dutifulness!4
My grinch-like goal in this book is to demonstrate why the consequentialist intuition cannot be correct, and then also to explain why it had seemed to
be. Fully two-thirds of the book (Chapters 1-4) will be spent refuting consequentialism. The final third (Chapters 5 and 6) will then suggest a way that
Kantianism itself could satisfy, or at least address, the consequentialist intuition. In Chapter 6, I will also fulfill the promise of the Preface to justify ethicizing as worthy of pursuit even if my defense of Kantianism should fail (or
fail to persuade the reader), and even if there seemed little hope of successfully defending any ethical theory whatsoever.
The refutation of consequentialism is accomplished by first defending utilitarianism (in Chapters 1-3) as the most plausible form of consequentialism.
That takes up half the book, in large part because of the intrinsic interest of
relevant topics. It is also due to the introduction of several meta-ethical considerations, which will then be used throughout the book. But the main reason for the length of the treatment of utilitarianism is the novel approach I
take to explaining the nature of that theory by means of a device I have
dubbed the Consequentialist Continuum. The Continuum enables me to generate a host of “new” ethical theories, which are the source of the aforementioned intrinsically interesting topics, as well as (in keeping with my metathesis) ethical edification in their own right.
Having identified utilitarianism as the crown jewel of consequentialism, I
then proceed (in Chapter 4) to refute consequentialism by refuting utilitarianism. There are certainly many strong arguments against utilitarianism to be
found in the philosophical literature, but I have chosen to focus on one, which
I will refer to as the Epistemic Objection.5 I chose it because I am utterly convinced by it and can but marvel at the way utilitarians for the most part ignore
it.6 For example, in a standard anthology devoted exclusively to the subject
of consequentialism, we are told that utilitarianism “has been subject to three
particularly influential objections” (Scheffler 1988, p. 2), but the Epistemic
Objection is not one of them.7 It is not that it has gone unrecognized. Indeed,
it was put forward at utilitarianism’s very (modern) outset by John Stuart
Mill’s contemporary, William Whewell (1864). My own introduction to it
came via Alan Donagan’s wonderful book, The Theory of Morality (1979).
More recently it has been argued by Lenman (2000) and Feldman (2006)8
among others. I shall do my best to clinch the case.
Introduction
3
Having disposed of consequentialism by disposing of utilitarianism, I then
take up (in Ch. 5) what I call the Consequentialist Critique (or the Consequentialist Challenge), which asserts that any plausible nonconsequentialist
ethical theory must in fact rely upon some hidden consequentialism for its appeal. Note, then, that the stakes have become high indeed. For this means that
not only the fate of nonconsequentialism hangs in the balance, but also of
morality itself, given that (according to me) consequentialism is no longer an
option. I then do my best to defend Immanuel Kant’s theory of ethics (which
I take to be the most widely accepted nonconsequentialist theory) both as
genuinely nonconsequentialist and as true. I make my job more difficult by
altering the theory in one significant respect, namely, extending the moral domain to include nonhuman animals essentially. Others in the contemporary
literature, such as Julian H. Franklin, Christine M. Korsgaard, Tom Regan,
and Paul W. Taylor, have done this. My treatment is perhaps novel in that I
am incorporating the idea into a work of normative ethics, as opposed to a
special topic in applied ethics.9
Finally (in Ch. 6), I counter the Consequentialist Critique directly on behalf of Kantianism. I argue that any agreement of nonconsequentialist ethical
recommendations with consequentialist ones should not surprise us since any
prima-facie plausible ethical theory will naturally conform to our considered
ethical intuitions about particular cases. Otherwise it would not be primafacie plausible! In this way Kantianism retains its dual claim to be both true
and nonconsequentialist, since the theory has already been otherwise defended convincingly while the appeal of consequentialism has been neutralized by the Epistemic Objection.
But have I won the battle only to lose the war? For then why would ethical theory matter? If all prima-facie plausible ethical theories yield the same
ethical recommendations, why not simply (continue to) rely on our pretheoretic (albeit considered) intuitions? That question obliges me to defend the final (meta-)thesis, which is that ethical theorizing itself has ethical import. Notice that ethical theorizing has turned out to be problematic not for failing to
prove, or holding out little hope of proving, any particular ethical theory (the
problem of the Preface), but even for proving one. That is because, as I argued in my reply to the Consequentialist Critique, no great practical upshot is
to be expected from knowing which is the correct theory.
A note on the title: “Ought Implies Kant” is a pun on the dictum, associated
with Immanuel Kant,10 that “ought implies can.” The dictum asserts that a
person cannot be under an obligation to do something which that person is incapable of doing; for example, you can’t insist that the prisoner in the dock
ought to stand up when the judge enters the room if the prisoner has been tied
4
Introduction
to his chair! This sort of stricture on ethics is very general; it is intended to be
a defining feature of ethics, no matter which particular ethics one happens to
favor. I agree with this requirement, but over and above that, my punny rendition of the (supposedly) Kantian dictum is meant to convey my acceptance
of Kant’s ethics, which I defend in this book. Furthermore, a key part of that
defense is the refutation of consequentialism as violating the dictum,11 in that
it requires us to know something that we cannot.12
NOTES
1. And whatever else we have control over, such as a policy and perhaps even our
character.
2. Not even if doing so would result in a more Kantian society or world than not
doing so.
3. The answer: Kant (1799), apparently!
4. It is important to distinguish consequentialism from what we might call karma.
The latter is about rewards and punishments; thus, if you do the right thing, good
things will come to you. But that would be putting the cart before the horse for consequentialism, which holds that something is the right thing to do because it has good
results. For example, you jump into the water to rescue the drowning child and the
child is saved; hence you did the right thing because good resulted (and had you both
drowned, you did the wrong thing)—that’s consequentialism (without the refinements
that will be discussed in the sequel). But if you rescued the child and were given an
award for having done the right thing—that’s karma. It seems to me that our intuitions commonly conflate these utterly different phenomena.
5. Or the Epistemic Argument, as Lenman (2000) dubs it (p. 343). Donagan (1979)
calls it the Factor of Ignorance (sec. 6.5, p. 199).
6. As Donagan (1979) puts it, “perhaps the most puzzling thing about discussions
of [this objection] at the present time is that what they [Mill’s contemporaries, such
as William Whewell] pointed out is not commonplace” (p. 205).
7. As Hudson (1989) puts it, “It is remarkable that so many advocates of [the theory to maximize utility] do not notice how odd their views are” (p. 221). One explanation could be that the Epistemic Objection is so obvious and decisive, that the only
practical response from a dedicated utilitarian must be to ignore it! But I have also
noted that consequentialism has an appeal for everyone. Hence its oft-predicted demise may be put off indefinitely. As Scheffler (1982) notes of Bernard Williams’ closing remark in Smart and Williams (1973), “The day cannot be too far off in which we
hear no more of [utilitarianism]” (p. 150): “that day refuses to come” (p. 3). Therefore I will offer a more substantive account of why consequentialism continues to
have the hold on us that it does; see my discussion of the Consequentialist Illusion in
Chapters 5 and 6.
8. Indeed, there are striking similarities of argument and example between their
expositions and mine, which resulted from an independent investigation.
Introduction
5
9. See Appendices I and III for discussions of this distinction.
10. Although mistakenly, according to Baumgardt (1946, pp. 99–100)! Kant (1788)
did say, albeit in a very particular context, “A need of pure practical reason . . . is based
on a duty to make something . . . the object of my will so as to promote it with all my
strength. In doing so, I must presuppose its possibility . . .” (p. 142; please note that all
page references to Kant in the present work are to the [variously called Prussian, German, Berlin] Academy edition of Kant’s complete works).
11. I am indebted to Mark Cannon for first bringing to my attention this additional
significance of my pun.
12. Cf. Howard-Snyder (1997): “inability which involves ignorance” (p. 245).
Chapter One
Egoism
Do whatever will work out the best for yourself. Nice advice! So counsels the
ethical theory of egoism. This is the right way to live, according to philosophers like Epicurus. It seems straightforward and certainly appealing, given
that the injunction grants the imprimatur of morality to your doing just what,
presumably, you’d like to do anyway. But this calls for closer inspection. The
first thing we may note is that “egoism” is ambiguous. It can refer either to
ethical egoism or to psychological egoism. These are quite distinct theories1
or claims. Psychological egoism holds that human beings are incorrigibly
selfish, while ethical egoism holds that our proper concern is to take care of
ourselves, i.e., to be selfish. The most obvious difference between these
claims is that the former pertains to what is (asserted to be) the case, while
the latter concerns what (is asserted) ought to be the case.
We are accustomed to assign responsibility to ascertain the truth of claims
about what is to science, and claims about what ought to be to philosophy (or
religion), specifically, ethics. This sort of division of labor has been hotly
contested in recent (and even ancient) times,2 but for now let me only caution
that, just as “egoism” is ambiguous, so too is a term like “psychology.” For
we would want to know whether the reference is to the scientific study or the
philosophical study of psyche (or mind or human behavior or human nature
or whatever the subject matter of psychology is taken to be). I will have some
things to say about psychological egoism as well as ethical egoism, but in
both cases I will be speaking philosophically rather than scientifically (insofar as such a distinction can be maintained).3
So ethical egoism is about what we ought to do. What is an ought, anyway?
I think what pops first into one’s mind is that it is about obligation; to say that
7
8
Chapter One
one ought to do something is to say that one is obligated to do it. Ethical egoism, then, would be asserting that each one of us has an obligation to privilege oneself above all others. Does that strike you as odd—I mean not about
putting oneself first (we’ll get to that) but that we must do that? Consider that
ethics is commonly divided into two realms, with the moral having to do with
our interactions with others, and what is sometimes called the prudential having to do just with oneself. For example, if it’s a question of whether to smoke
in a public space, morality might be invoked to prohibit smoking since such
behavior could have a harmful effect on others and/or show disrespect for
their rightful prerogatives. But if one were walking alone on a deserted beach
and felt like taking a puff, would it not normally be considered strictly one’s
own business whether to light up, even if the habit could kill you?4
Prudence will still prescribe, for example, that one not smoke so as to live
a long and healthy life. More generally, prudence has to do with the good life:
how shall I live in order to enjoy the most pleasure and/or meaning and/or
value in my individual existence?5 But any such prescription would be a recommendation rather than a requirement; it would offer us wise counsel rather
than dictate what we must do. Oh yes, we do use the phrase “prudence dictates
that,” but should this not be understood as a kind of hypothetical imperative?
That is, it is shorthand for, “You must do x if you want things to turn out the
best for you” (or, as we say, “if you know what’s good for you”). But, as Immanuel Kant noted, morality is categorical6; it says, “You must do x . . . no ifs
. . . or cigarette butts!” So the prescription not to smoke in public would clearly
be a requirement, precisely because it affects or otherwise relates to others.
The advice of prudence, however, is yours to heed or ignore. Is not ethical egoism, then, an ethics of prudence? If so, then it would seem not to be about obligation and instead intended only as sound advice about how to live.
But that won’t work. Why not? Precisely because ethics does contain the
two components of prudence and morality. Ethical egoism purports to be a
theory of ethics; hence it must be telling us not only about how to “treat” ourselves but also how to treat others.7 And indeed it does, for the theory enjoins
us to watch out for our own interests above anybody else’s; so that right there
is telling us something about how to behave towards others. Hence ethical
egoism is a moral and not only a prudential theory.
But then this raises a problem for the non-obligatoriness of ethical egoism.
For look at how the theory goes about articulating its moral prescriptions,
viz., by demonstrating the dependence of the moral on the prudential. For example, if you treat others respectfully, keep your promises, don’t lie, and so
forth, then, the theory typically argues, you are more likely to benefit yourself because others will feel gratitude rather than enmity toward you. Of
course, if prudence is only a recommendation and not a command, then you
Egoism
9
are still free to lie, and so forth, thereby presumably bringing yourself to
grief; however, it would still be the case that ethical egoism prescribes truthtelling, and so forth. But if the egoistic prescriptions are only recommendations, then it turns out that you have no moral obligation to keep your promises. But of course you do (in general). Ergo: Ethical egoism is false . . . or
its prudential injunctions must be obligatory.
So what is it that ethical egoism would require us to do? “Be egoistic!” is
the obvious response. But we have not yet considered what that really means.
The name of the game in philosophy is ambiguity; that is, the goal is to disambiguate by drawing distinctions. What are different things that might be
meant by egoistic? This lands us in the realm of psychology, to which I had
promised to return. For the injunction “Be egoistic!” consists of an ethical
part, which we have just discussed—represented by the imperative “Be!”,
which is shorthand for “One ought to be”—plus an empirical part, which is a
certain way of being, namely, egoistic. This way of being belongs to psychology because it has to do with human beings and, more particularly, with
the way we behave or, speaking even more psychically, the way we are motivated.
The ambiguity is this: egoistic could mean a kind of selfishness or else
something more thoughtful. I suppose the former is the more natural reading
of the term, referring to a greedy, short-sighted attitude and activity of wanting and taking things for oneself. To be sure, humans often betray that manner of being. But obviously8 we can be a lot more sophisticated in our egoism and even in our selfishness. The Count of Monte Cristo devoted himself
over a period of decades to devising and carrying out a meticulous plan of revenge. Furthermore, either selfishness or egoism can be roundabout, even assuming the guise of its opposite; for example, you may lavish someone with
gifts for the purpose of securing her or his loyalty.
We still have not put our finger on the ethically decisive feature of non-selfish egoism because all of the above characterizations could apply to selfishness as well. It would seem obvious that ethical egoism does not advise us to
be selfish, for that is generally (although perhaps not universally) recognized
to be a self-defeating strategy of life. What we seek is sometimes called “enlightened self-interest.” There is supposed to be a way of being that does in
fact serve one’s own best interests and yet does not rub other people’s face in
it. Such things are not really that difficult to imagine. In fact, this form of egoism is commonly touted as the ideal ethic of the entrepreneur. A homely example would be sponsoring the local Little League team and thereby reaping
the gratitude and perhaps patronage of the players and their parents, some
“free” advertising when sponsors are mentioned at the games, and even some
personal pleasure if you happen to enjoy baseball or kids.
10
Chapter One
We can push this further. Notice that the way of being egoistic, especially
in the enlightened sense, has so far suggested a dichotomy between the appearance of behavior and the reality of motivation. For it looks like a good
way to endear oneself to others is to act as if one cared about them, for example, by being their benefactor, even if one’s actual intent were to benefit
oneself.9 Suppose this were true, that this is an effective strategy. Then might
it not also be true that an even more effective strategy would be really to care
about others? After all, if one wished to be a friend of humanity consistently
and reliably, thereby to maximize the benefits to oneself, wouldn’t it be better to be a whole-hearted rather than a phony friend? The egoistic friend is always having to summon up the motivational energy to do things he or she
would really rather not be doing; the genuine friend would therefore seem
more likely actually to do the friendly thing, because he or she wants to do it.
That has an immediate egoistic payoff in terms of satisfaction of the donor’s
desires (to do the very things she is doing), but also in terms of whatever benefits accrue to her from helping others, since she will be helping them more
than if she did not really care about them.
What I am suggesting, in a word, is that ethical egoism might urge that we
all strive to become altruists, that is, people whose motivation is altruistic
rather than egoistic. Let me now make a few observations about this conclusion. First, keep in mind that I have reached the conclusion by philosophic
means (i.e., adducing as evidence only things that are already known or plausibly believed). I am not a scientist. So my claim is more about possibilities
than actualities: I am arguing only that there are grounds to believe that each
of us would be better off if he or she moved in the direction of becoming less
focused on his or her individual welfare. To establish this as a truth we would
require a more extensive empirical investigation. Such an investigation might
still fall short of science; for example, I think it would be eminently useful for
each of us to test the hypothesis in our personal experience. Ultimately, however, some kind of controlled and comprehensive testing would be welcome.
Second: While the above remarks might seem to highlight the inadequacy
of philosophy as a method of inquiry, in fact I think they show the power of
philosophy. For without resort to time-consuming and expensive procedures
or apparatus, we have garnered plausibility for a hypothesis that is non-obvious, indeed, almost paradoxical-sounding: egoism may advise altruism! I
should also point out that the proposed scientific investigation of the hypothesis that altruistic motivation has an egoistic payoff would itself depend on
further philosophic reflection, for the question of what really is in one’s best
interest is not (only) a scientific one. In fact, this is a philosophical problem
par excellence: what is the (best) good for a human being?
Egoism
11
The scientist might demur by arguing as follows:
Science is best suited to investigate what people in fact prefer. Therefore science is best suited to determine what is the good for a human being.
But doesn’t that argument merely presume that the good for a human being is whatever human beings prefer? But why assume that? In fact it would
seem easy enough to refute such a contention; for example (if I may “remind” the reader, not to mention, the scientist), science itself has revealed to
us that people often, and perhaps usually, choose foods and pastimes and
even mates that are not good (or the best) for themselves. So preference does
not appear to be the best criterion of our personal good, at least not without
a great deal of articulation of how the notion should be conceived for this
purpose or, more likely, how it should be qualified. For example, it could be
that our “natural” preferences have been skewed by commercial interests,
which are indifferent to our personal best interests and “care” only about
their own (and that too, most likely, narrowly conceived); so a more plausible criterion of our good would be our preferences somehow “filtered” from
the prevailing influences of modern society and/or exposed to the best information.10
Science would certainly become involved in due course (or perhaps ideally,
hand-in-hand); for example, if the good does turn out to be some sort of qualified preferences, say, preferences that are formed under a condition of maximum exposure to relevant information, then we could imagine scientific psychology attempting to discover what those preferences actually are by putting
people into those circumstances and asking them questions.11 But equally certainly science will never supplant philosophy in the study of ethics or even
psychology since, as we have just seen, any scientific investigation would
only beg crucial questions if it proceeded without humanistic interpretation of
those questions.
So, finally, my third point about the present investigation is that it illustrates how the study of ethics can contribute to our becoming more ethical.
For if a part of being ethical is to do good things for others, then understanding that we may personally benefit from engaging in such behavior and perhaps even from undergoing a thorough revision of our own motivations,
could serve to move us in that direction. And note: This is the case no matter
whether ethical egoism turns out to be true. It is the discussion that has been
helpful.
Let us now resume that discussion. I have defended the claim that altruistic motivation can serve egoistic ends and might even be best suited to do so.
As noted, this is an empirical claim. But ethical egoism does not make an
12
Chapter One
empirical claim; rather it makes a normative claim about how one ought to
behave. At once we face yet another ambiguity: “normative” has quite distinct meanings, which has led to a great deal of confusion. It does have an
empirical sense, according to which the way people usually behave constitutes a norm. But ethical norms are not empirically based in this way; for example, even if all or most husbands beat their wives (and even believed that
was perfectly OK to do), it would still be wrong to do so.12
Another way to put this is that ethical theory is about justification rather
than mere explanation. Ethics asks: What makes it right (or wrong or permissible) to do x (or to feel y, or to be z, etc.)? So, we may agree that, all other
things equal, it is wrong to steal. But why is it wrong? Note: This is not to ask,
“Why do we believe it is wrong?” That is an empirical question; for example,
the explanation could be that we were taught this by our parents. But the
question remains: Is our belief true? That is, is it “in fact” wrong (as a rule)
to steal? We seek a sound argument to back up the belief. Ethical egoism, like
any ethical theory, offers such an argument, to wit:
One ought always to do that which promises to maximize one’s own welfare.
Stealing usually fails to maximize long-term benefit to oneself.
Therefore stealing is (by and large) wrong.
Having already analyzed ethical egoism to reveal its deeper meaning and
perhaps thereby enhanced its plausibility, I would now like to pose a question
that challenges the theory. Does it make any sense that what “makes it” right
that we not steal (murder, lie, be cruel, inconsiderate, etc. ad inf., and also that
we should be caring, charitable, helpful, etc. ad inf.) is that: it works out the
best for us, that is, for the individual agent (the potential thief, etc.)? Remember: this is not a claim about our psychology, about how selfish or egoistic we may be by nature (or inevitable nurture), but is rather a claim about
the justification of our behavior (and attitudes, character, etc.).
I submit that the claim is implausible on its face. Consider whether it’s OK
to slap a little girl who is crying until she becomes quiet. This seems outrageously wrong (in the normal run of cases). Does it make any intuitive sense
at all that the reason it is wrong is that somehow it would come back to
“haunt” you: because you’ll be thrown into jail, or the girl will grow up to
hate you and hurt you, or God will cast you into the eternal fires, and so forth?
Granted, many people are in fact motivated not to slap little girls by just such
considerations. But that is “only” an empirical fact. Ethics is interested in
something different, namely, why certain considerations, rather than explain
what you will do, explain why you should (or should not) do it. And “none of
Egoism
13
the above” egoistic considerations seems suited to that task. Therefore the
theory of ethical egoism, whatever virtues it may possess, cannot be true.
The ethical egoist has a reply, however.13 Perhaps it is “unpalatable” to accept egoism as the basis of ethics, yet it may be unavoidable. The reason is
that human beings are helplessly selfish to begin with—this is the thesis we
earlier dubbed psychological egoism—so it would hardly do to insist that we
act in any other way. This argument trades on the reasonable requirement that
ought implies can; that is, ethics only makes sense if it mandates actions that
we are capable of performing. That is why none of us is under a moral obligation to feed all of the starving people in the world since none of us is capable of doing that.14 Thus, if it is a fact of human nature that each of us is
only capable of caring about him- or herself, then no ethical theory could legitimately charge us to embrace the welfare of others as if it were our own.
It would seem to follow that ethical egoism must be true by default because
any alternative ethics would be premised upon our being something that we
are not.
But isn’t there something curious about that argument? If ethical egoism
must be true, then how could it be true? Put it this way: If we are by nature
selfish, then why would we need an ethical theory to tell us that we ought to
be? So it is not so much that ethical egoism would be false as that it would be
otiose; it would be hoisted by its own petard, the bombshell revelation15 that
we are selfish beings. But the theory is not so easily dismissed. We have already noted that egoism is not the same as selfishness, so even if it were true
that we are fundamentally egoistic, there might still be a lot to learn from rationally reflecting on how one could maximize one’s own welfare and, for
that matter, what constitutes genuine personal welfare. For clearly most of us
are wrong or confused about many of the elements of our own happiness, as
regards both the proper object or end of egoistic concern and the most effective means thereto. Thus, suppose that you were knowledgeable about what
constituted your welfare (the end)—say, health among other things; you
might still be in error about what conduces to that good (i.e., the means to that
end). For example, you might believe that a high-protein diet is good for your
health when in fact it may be more likely to undermine your long-term
prospects of a healthy life. Alternatively, you might be absolutely right about
the best way (the means) to secure a fortune for yourself, and yet wrong about
the value of being wealthy (the end).
But there is still a problem with the argument for ethical egoism that is
based on psychological egoism. Actually there are two problems or types of
problems: empirical and logical.16 The empirical problem is that the premise,
that is, psychological egoism, might be false. I think it is. But in keeping with
14
Chapter One
the science/philosophy division of labor I have chosen to observe in this investigation, I will leave the definitive determination of that question to the
scientists (in collaboration with “concept-checking” philosophers)17 and focus here on a logical problem with the argument. So, suppose we humans really are exclusively or at least predominantly concerned about our personal
good (more or less rationally conceived and pursued): it still does not follow
that we are obligated to put our personal concerns first. Why not? Because the
truth of that premise is compatible with our having no obligations whatever.
Recall the earlier discussion about whether ethical egoism is a theory of
obligation or only of recommendation. I there pointed out that morality is (at
least in part) about the former, and therefore so must ethical egoism be. In effect I was adducing another meta-ethical principle18; that is, just like ought
implies can, moral implies must19 is a criterion of counting as an ethical theory at all. But it is the job of so-called normative ethics to make the case that
any given ethical theory obligates us to carry out its injunctions. In other
words, any argument for the truth of a particular ethical theory, such as ethical egoism, ought20 to provide a justification for the “demand quality” of that
theory. For example: Why must21 I generally tell the truth, according to ethical egoism?
It seems to me that asserting psychological egoism fails to answer this
question. For even if it were true that we humans are well-nigh incapable of
caring about anybody except our personal selves, and also that, say, being a
reliable truth-teller is an excellent strategy for attaining our best long-term
personal good, I don’t see how it would follow that we have an obligation to
be reliable truth-tellers. Compare: Even if most men had an almost irresistible
desire to mate with almost any woman who is standing naked before them in
a private space (i.e., they are well-nigh incapable of not “taking advantage of”
her), and the most effective means to satisfy that desire were to be a rapist, it
would not follow that men have an obligation (nor even “permission,” of
course) to be rapists. In other words, can not-but does not imply ought.
The argument from psychological egoism, therefore, would require some
additional premise that explained why our irresistibly desiring our own welfare conferred a corresponding obligation on us to satisfy that desire. I leave
it to the reader and/or other theorists to try their hand at providing that; otherwise, consider an alternative ethical theory. Of course another possibility is
that ethics itself can22 be dispensed with23 (or radically reconceived). Then the
conclusion to draw from the (presumed) truth of psychological egoism would
simply be that various hypothetical imperatives could be justified as the best
ways to satisfy one’s basic motivation. But if one chose to ignore them—even
ones that might proscribe lying, stealing, and murder (albeit in one’s own best
interest)—one would not be doing anything wrong.
Egoism
15
NOTES
1. On what I mean by a theory, see Appendix II.
2. Regarding the philosophy/religion division, see Plato’s “Euthyphro.” Regarding
the science/philosophy division, see Quine (1951 and 1969). Regarding the science/
ethics division, see Wilson (1975).
3. For my conception of the difference between science and philosophy, see my
essay, “Science and Philosophy” in Philosophy Now no. 33 (Sept./Oct. 2001), p. 31.
In a nutshell: I argue that the two forms of inquiry have only one essential difference,
namely, that science adduces new evidence to support its claims, whereas philosophy
offers up “reminders” (which may include established scientific facts) to buttress its.
4. Of course it is plausible to suppose that others would be affected no matter
what; for example, if you were stricken with lung cancer, there would be economic
costs to society and emotional costs to your family. But if there is to be any division
at all between the public and the personal, the line must be drawn somewhere.
5. Insofar as that question has a general answer, it would be a topic of ethics, for
example, if it turned out that “pleasure is the good.” Then it would be left up to you
personally to figure out what constitutes pleasure for you or what sorts of pleasures
are available to you in your particular circumstances, etc.
6. Kant (1785), p. 416.
7. Ethical egoism must also incorporate a theory of the good, which is what the
ethical theory is enjoining us to maximize for ourselves. For example, Epicureanism,
a version of ethical egoism, stipulates that pleasure is the good, which is the theory of
hedonism. Hence Epicureanism advises us to maximize our own pleasure (and to hell
with anybody else’s if it would interfere with ours).
8. You see how I adduce reminders in my philosophical approach to psychology.
9. There is a delightful episode in Mark Twain’s autobiographical Roughing It
about an amiable travel companion who has perfected the skills of being helpful to
others and thereby reaps many benefits to himself (and young Sam).
10. Cf. the process of cognitive psychotherapy in Brandt (1979).
11. In recent times there have been interesting practical applications of this kind of
rationalized preferences, for example, in psychotherapy by Albert Ellis (1962) and in
social choice by Bruce Ackerman and James S. Fishkin (2004).
12. Thus, right and wrong are like truth and falsity in that, for example, everybody
could believe that the Earth stands still even though the Earth in fact moves.
13. Actually several, but I will consider the one I think most people would find
most plausible prima facie.
14. We might still have the obligation to do whatever we can to help feed all of the
starving people in the world.
15. “Commonplace reminder” is probably more apt since it is hardly news that humans are selfish.
16. This is one way to express the more general truth about reasoning that arguments can go wrong in two ways: one or more of the premises may be false and/or
the inference to the conclusion may be invalid (illogical or fallacious).
16
Chapter One
17. My personal opinion is that psychological egoism has been definitively refuted
by a philosopher, viz., C. D. Broad (1952), as well as by Bishop Joseph Butler (1726,
sermons I and XI) before him.
18. See Appendix I for an explanation of this notion.
19. That is, on at least some occasions, since morality probably also accommodates the permissible and the supererogatory (that which is “above and beyond” duty).
My point is that obligation is a necessary component of morality.
20. This is a meta-ethical “ought.”
21. That is, ethically not empirically speaking, of course. Ethical egoism is not
claiming that you “can’t help but” be a truth-teller!
22. Or should, meta-ethically speaking.
23. This would be analogous to exobiological theorists (meta-SETIcists, we might
call them) concluding that SETI (the search for extraterrestrial intelligence) should be
abandoned.
Chapter Two
The Consequentialist Continuum
Do whatever is for the best. Who could argue with counsel like that? And yet
it is fraught with perplexities. But to see this it is necessary (and proper) to understand exactly what such counsel amounts to and, indeed, how rife with possibilities it is. In fact there is a whole class of ethical theories that proffer this
advice. We examined one of those theories in Chapter 1: ethical egoism. They
go by the name of consequentialism because to focus on bringing about “the
best” is to be concerned with the consequences of one’s actions (and motivations and character, etc.). These theories state that it is the ultimate obligation
of us all to do our utmost to bring about the best outcome. Thus, if spending a
dollar on some candy would bring your child a few moments of pleasure but
donating that dollar to UNICEF would save another child’s life, then, ceteris
paribus (“all other things equal”), you ought to donate the dollar.
Immediately we begin to see how difficulties can emerge from the simplest
beginnings. Is one prohibited from ever buying one’s child a piece of candy?
After all there will always be, or as long as we can foresee in our lifetime,
other children living in desperate circumstances; so how could any indulgence be justified? And not only indulgences: since there are many degrees
of peril, it is easy enough to imagine how even helping to save the lives of
impoverished children could interfere with projects still more compelling,
such as devoting oneself to the elimination of nuclear weapons. So consequentialism can be questioned. But what alternative would fare better? And
how could we judge what would be better without begging the very question
at issue, since “better” itself sounds like a consequentialist criterion?
Thus, in a nutshell, is the controversy of consequentialism. It is really many
controversies and of different kinds. In the end (Chapters 4–6) I will be happy
to lay consequentialism to rest. But in the beginning (Chapters 1-3) I will do
17
18
Chapter Two
my “best” to reveal the richness of consequentialism. This is in keeping with
a theme of this book that there is much to be gained, of ethical value in particular, from the study of ethical theories, even when they turn out to be false.
I have spoken of a class of consequentialist theories. In the main among
ethicists one hears discussion about two: egoism and utilitarianism. Often
they are conflated1 for, as noted, they both advise maximizing the good. However there is a crucial difference between them. As its name suggests, egoism
advises2 that one maximize one’s own advantage or welfare or “utility” or
good. Utilitarianism is more catholic . . . indeed, maximally catholic . . . for
it commands maximizing good without regard to whose it is. So our neighbor
must also be taken into account, but also our enemy, and the stranger, and the
animals in the forest, and the Martians (no matter whether friendly or invading). Thus, each and every one of us is under the primary obligation to contribute as best we can to the abundance of good in the whole world (or universe). It’s a tall order, and to help us appreciate just how “tall” is my order
of business in this and the next chapter.
It occurred to me one fine day, after many years of pondering and teaching
egoism and utilitarianism as the exemplars of consequentialism, that they represent the extremes of a continuum. For between the one(self) of egoism and
the everyone of utilitarianism lie the some of . . .? The relevant variable by
which the continuum can be generated is: whose consequences matter? In the
case of egoism, only one’s own; in the case of utilitarianism, everyone’s. But
what about a theory that privileged, say, all of the members of one’s family?
“Thou shalt devote thyself to the welfare of thy family above all else and
above all others.” Would that be utilitarianism? No, because you would not
be obligated to support the welfare of any family other than your own, not to
mention the tigers in the forest; or if you did have an obligation to them, it
would only be indirect, only insofar as their welfare contributed to the welfare of your family. Indeed, it is conceivable that you would find yourself obligated in some situation to exterminate your neighbor’s family. Well, would
it be a form of egoism, then? Only trivially so in that it is your family you are
enjoined to cherish; but that it is not really a form of egoism becomes apparent from realizing that the assurance of your family’s welfare may well entail
significant sacrifice on your part personally.
In this chapter I would like to survey the ethical landscape(s) beyond egoism in more detail . . . and not as a mere exercise. Let me tout the utility of
ethical theorizing once again, for this business of generating a plethora of
“new” ethical theories can usefully provide us with a mirror for our mind, our
moral mind. “Who am I?” is a profound philosophical and existential question. At least part of the answer has to do with one’s ethical identity. I would
wager that many people of the world have not been able to discover them-
The Consequentialist Continuum
19
selves among the characterizations of theoretical ethics prevailing in academic philosophy. I think many will have a better chance of doing so in the
expanded set of theories I am about to propound. But let me also reiterate that
the continuum I shall be discussing will not exhaust the possibilities of consequentialism; for example, a theory that postulated the ultimate repository of
value in the universe, that is, the ultimate good, to be the artistic corpus of Picasso would be consequentialist but would not fall on this continuum since no
sentient being’s welfare would have intrinsic value, although some curators
and their support community might have instrumental value in maintaining
the corpus. The main purpose of this chapter is to help us appreciate the theory of utilitarianism, which is concerned with the welfare of sentient beings,3
and is presumably more plausible for doing so than one that wasn’t thus concerned.
Let me now characterize the continuum more precisely. I picture it as a
straight line in the horizontal, with egoism its left terminus and utilitarianism
its right. Thus, any point in between picks out a potential consequentialist theory, and each point or theory “contains” more people or sentient beings than
the one to its left.4 So, for example, the point representing family consequentialism or familism5 will presumably be rather close to the left or egoism terminus, while a point representing, say, sexism, which would be the ethical
theory that privileges one’s own sex above the other(s?), will be further to the
right since there are more individuals of a person’s sex than there are members of his or her family (supposing family here in the usual6 sense rather than
as in, e.g., “the human family”). Of course we would also have to stipulate
which kind of family (etc.)—“nuclear,” “extended,” etc.—in order to fix their
number, and then use, say, an average to represent the type in question, for
example, 3.57 persons in the human nuclear family.
As already noted, the continuum takes welfare (or happiness or rationalized-desire-satisfaction or however one would analyze the good of sentient
beings) as the proper object of ethical concern. “In theory” anything could
be, for the essence of consequentialism is just that the ethicality of our actions (and/or character and/or whatever else is under our control in some
meaningful way) is determined by their consequences. It is easy enough to
imagine, for example, someone maintaining that creativity (the human quality thereof) matters more than welfare; so a “creative consequentialism”
would advise (or command) us to maximize the amount of creativity (suitably defined) in the universe even if that meant making most or all of us miserable. And, as we saw with the Picasso example above, the “welfare” of
something other than sentient beings could be privileged. The “land ethic”
of Aldo Leopold (1949) and J. Baird Callicott (1989) is just such a theory,
for it assigns intrinsic value to Gaia, that is, the ecosphere as such. But I will
20
Chapter Two
leave it at that because the sentient-welfarist variety of consequentialism is
probably the most intuitive7 and hence can serve as the representative type
or “default” for illustrative purposes.
The consequentialist continuum has two additional general features of
note, which is really to say that consequentialist theories of ethics have these
two features. First, the “points” represent justifications of ethics and not, in
the first instance, types of ethical motivation. We saw the importance of this
distinction in Chapter 1, where the theory of ethical egoism turned out to have
the potential to justify altruistic motivation. Thus, it is an open question about
each theory in the continuum what kind(s) of motivation (or intention or
modus operandi) it would endorse. We would just have to see how that plays
out from the logic of the theory and the circumstances in which it is applied
(including human nature).8
Another feature of the continuum is that every theory applies universally.
Let me show you what I mean by that. Take the theory of ethical egoism,
which says: Maximize your own welfare. This means that Joe should maximize his, and Jane should maximize hers. It does not mean that Joe should
maximize his and Jane should also maximize Joe’s (or Joe, Jane’s, etc.)! This
may seem obvious once pointed out but is in fact a deep matter, to which we
shall return. A humorous but telling take on it comes from that marvelous—
alas, defunct—comic strip, Calvin and Hobbes. In one episode the lonely,
self-centered little boy, Calvin, explains to his imaginary friend Hobbes: “I’ve
discovered my purpose in life. I know why I was put here and why everything
exists. . . . I am here so everybody can do what I want.”9 Here we have an individual who personifies a perverse form of ethical egoism, who believes that
not only he but everyone else ought to have his, Calvin’s, happiness innermost at heart.
I call such an egoism ersatz egoism10 (as opposed to ethical egoism). It is
a mistake, but an interesting one. It is a mistake because it violates a metaethical principle we can add to our list,11 namely that to count as an ethics
“candidate” at all, a theory must pick out the object of ethical concern formally and not materially; that is, the specified object must be general, for example oneself or the citizens of one’s country or persons of one’s sex, and not
a particular individual, for example Calvin,12 or a particular group, such as
Americans or men. This is a subtle but crucial distinction, on which it is easy
to stumble because the theories we have been examining do privilege some
over others as proper objects of concern; for example, ethical egoism picks
out oneself and not others. But it does not run afoul of the cited meta-ethical
principle because the “oneself” is Joe for Joe and Jane for Jane.
Here again I proclaim the usefulness of ethicizing, for conflations abound
between ersatz and genuine ethics. To such an extent, in fact, that we could
The Consequentialist Continuum
21
generate an entire continuum of the ersatz variety of pseudo-ethical theories,
which would be in one-to-one correspondence with genuine (i.e., “candidate”) ethical theories on the kosher continuum. By such contrasts we will
better understand what the truly ethical theories have to offer us. So let us
postulate the two continua and designate them as the Ethical Consequentialist Continuum and the Ersatz Consequentialist Continuum. In the sequel,
however, “the continuum” will continue to refer to the former.
Let us now proceed by moving just to the right of egoism on the continuum
to consider the case of a two-person ethics, which we could call dyadism.
What would be an example of this kind of ethics? Could there be an ethics of
married couples? I think not. Why not? Because not all persons are married.
Thus, another meta-ethical principle suggests itself: Any ethical theory must
confer (ethical) agency status on all persons of sound mind. But we could
imagine a social/legal set-up where everybody is always a member of a (i.e.,
some) married couple, for example, by automatic assignments according to a
rule of precedence. In such a world, which is approximated in some places in
ours, every one of us would have as her or his primary obligation to be the
promoter of . . . what? There would seem to be two choices. The more obvious is: of his or her spouse’s (best) welfare.13
It’s a nice thought, I suppose, and is surely given lip service in many marriages. But as a thoroughgoing ethics it would just as assuredly raise questions. For example, would the welfare of one’s children always be secondary
to that of one’s spouse? We can certainly imagine that the welfare of one’s
children might have a big influence on the happiness of one’s spouse (even
though, keep in mind, your spouse’s primary obligation would be to your
welfare, not to your and/or his/her children’s) and thereby acquire indirect
significance among your concerns. But if your spouse happened not to care
that much about the children, then it seems you might be obligated to sacrifice their welfare to your spouse’s on some occasions.
This reasoning might seem “rigid,” but it points out an essential element
of ethical (or any) theorizing. The ethical conclusions drawn must follow
from the ethical premise; hence if other considerations are not explicitly or
implicitly to be found therein, they can have no bearing on one’s obligations.
So in our example, one might have thought that, “Oh yes, one’s primary obligation is to one’s spouse, but of course certain things are barred or understood, such as not harming the children.” But the whole point of devising an
ethical theory is to explain what is barred (and what is required) and what is
not. So if the theory said that your primary obligation is to promote the welfare of your spouse and said nothing about the children, then their welfare
simply would not “count” morally unless it somehow impacted on your
spouse’s.
22
Chapter Two
However, we have not yet really done justice to dyadism because, on closer
inspection, the theory we have just analyzed is not dyadism. For recall that
moving to the right on the consequentialist continuum means including more
beings (human or otherwise) within the scope of our moral concern. But the
“dyadism” so far represented contains only one such being, namely, one’s
spouse. Thus, this is in fact another form of moral monism, occupying the
same point on the continuum as egoism. Maybe we could call it “monistic altruism,” for one’s whole concern is supposed to focus on a single other person. A true dyadism would have each of us concerned about a pair of beings.
If restricted to a two-person marriage, therefore, it would apply to one’s
spouse and oneself.
Monistic altruism is not unique on the consequentialist continuum. Analogous theories will be found all along the continuum, just to the left of each
subsequent (non-altruistic) theory. For the key feature of altruism is that it focuses moral regard on everybody (in the relevant group) other than oneself.
Thus, for example, a triadic theory that stipulated moral regard for a threesome that included oneself would be shadowed by an altruistic dyadism that
directed one’s moral regard to the two other members of the threesome exclusively; so Moe would be responsible for Larry and Curly at whatever cost
to himself, while Larry would have the like responsibility towards Moe and
Curly, and mutatis mutandis for Curly.
In the subsequent discussion I will sidestep (or step over as I amble up the
continuum) all of the altruistic theories for this reason: their rationale looks
mysterious. If persons or sentient beings matter, then why wouldn’t oneself
matter too, no more but no less than any others in the relevant group?14 Perhaps, then, all of the altruistic consequentialist ethics, while not exactly ersatz, are nevertheless suspect enough that we could justifiably place them on
a continuum of their own. Let us do that for convenience. So now monistic
altruism and all other such “angels” will no longer have to crowd the pinpoints on the Ethical Consequentialist Continuum. Each point on the continuum will be understood to include oneself, the ethical agent, among the group
of individuals to whom one has one’s primary ethical obligation. This, as we
have seen, makes eminent sense because the agent is “one of them” in the relevant respect(s), for example, being a family member in the familist ethics.
Let me just note that the altruist continuum has its own ersatz continuum
running alongside it, on which we could locate, for example, a “Calvinist”
equivalent of the conjugal form of monistic altruism we have been considering. This would hold that both spouses have as their primary obligation promoting the welfare of one of them. For example, my fiancée’s cousin has informed me that in their culture, the man is supposed to sacrifice everything
for the welfare of his wife, while the wife has no corresponding obligation toward the husband. Thus, the altruism is non-reciprocal, one-way; the other
The Consequentialist Continuum
23
spouse is supposed to be an egoist. Indeed, if we carried this to its logical extreme, such that everybody’s primary obligation were to your spouse, it would
be indistinguishable from ersatz egoism.15
If we now consider dyadism proper, we can note that, among the group of
genuine candidates for “true theory of ethics,” it is the first one (“after” egoism) to make genuine self-sacrifice a real possibility. In egoism, genuine selfsacrifice is out of the question. Oh yes, one might submit to various slings and
arrows in order to achieve some goal, but that goal would always compensate
with at least as high a payoff for oneself as simply avoiding the slings and arrows (or else you did the wrong thing). If you “sacrifice” eating the second
ice cream sundae in a row, it is only because you know that doing so will retain your vital appetite for another one down the road, not to mention forestall a belly ache. One might even believe one was making a genuine sacrifice and yet be doing the right thing by egoist lights, according to my
argument of the preceding chapter, since a thoroughgoingly altruistic motivation could be just the ticket to personal happiness in the long term.
But dyadism is different, and this is so even though, as just determined, it
is not an altruistic theory and counts oneself among the proper objects of ethical concern. Thus, returning to the theory we might call conjugism, wherein
each of two spouses has as his or her16 primary obligation in life the furtherance of their joint welfare: is it not conceivable, even likely, that one spouse
would be called upon to reduce her or his own long-term prospects for the
betterment of the joint prospects of the two of them? For example, if you gave
one of your truffles, which you enjoy, to your spouse, who positively craves
them, then, all other things equal (for example, even assuming that your
spouse does not then reward you with a similar treat in return), the sum of
your happinesses would be increased.
However, just as with the altruistic form of dyadism, full-fledged dyadism
falls far short of displaying the scope of regard we would expect of a moral theory, for none of the members of other dyads warrants the slightest concern for
their sake. Each of us is only obligated to consider others when doing so would
have some effect on our own dyad’s welfare. Of course, analogous to what we
saw with ethical egoism, such inauthentic other-regard could still be significant
in everyday affairs and even be equivalent to heartfelt altruism for most practical purposes. Furthermore, and again as with ethical egoism, we might even
suspect that ethical dyadism might enjoin the cultivation of a sincere motivation of caring about the welfare of other dyads (or individuals singly or however grouped) . . . all the better to improve the prospects of one’s own dyad.
In the end, however, dyadism meets the same fate as egoism in ethical theory.
For what sense would it make to suppose that our concern about the welfare of
a little child lying gasping by the roadside . . . indeed, about our own child! . . .
could only be justified by its effect on the happiness of a couple that included
24
Chapter Two
oneself but happened to exclude that child? Surely17 that is not what makes stopping to aid the child the right thing to do. Thus, no matter how plausible some
form of dyadism might seem for whatever reason, nor how closely it may mimic
normal morality, even incorporating a sincere concern for some or all others, it
is shown by theoretical analysis to fail as an ethics. We are therefore propelled
further along the consequentialist continuum.
NOTES
1. It seems that people assume something like Adam Smith’s (1776) “invisible
hand” (bk. 4, ch. 2) such that to pursue one’s own good is ipso facto to pursue the general welfare.
2. Or commands, as was argued in Chapter 1.
3. Taking the cue from Mill (1863): “This, being . . . the standard of morality . . .
by the observance of which an existence such has been described might be, to the
greatest extent possible, secured to all mankind; and not to them only but, so far as the
nature of things admits, to the whole sentient creation” (ch. 2). Cf. also Hudson (1989):
“. . . ‘utility’ is happiness-minus-unhappiness for all sentient beings in the long run”
(p. 221; Hudson’s emphasis).
4. I am not going to worry overmuch about two different theories that happen to
contain the same number of individuals since I am only sketching a picture for the
practical purpose of talking about a concept. This matter does have some significance,
however, in the discussion of altruism later in this chapter.
5. A researcher who actually used the term “familism” to portray a moral outlook
was political scientist Edward C. Banfield (1967).
6. Albeit covering a wide range of variation, to be sure.
7. It can also account for valuing creativity et al. indirectly, viz., as contributing
to our welfare.
8. Railton (1984) limns this sort of distinction with great subtlety. Langenfus
(1990) calls it the “criterion/motivation separation” (p. 131), which he sees, however,
as deeply problematic. “Decision procedure” is another relevant term in the literature
for what I am calling motivation; cf. Crisp (1992).
9. Taken from a Calvin and Hobbes cartoon by Bill Watterson. (c) Watterson.
Used by permission of Universal Press Syndicate. All rights reserved.
10. I also sometimes refer to it as “Calvinism,” if you don’t mind the pun.
11. See Appendix I.
12. What about Yahweh or Brahman? Without turning this into a theological treatise, I will just remark that there seem to me to be grounds for an ethical interpretation of these supposed named individuals as placeholders for everyone. For example:
“Teacher, which is the greatest commandment in the Law?” Jesus replied: “‘Love the Lord
your God with all your heart and with all your soul and with all your mind.’ This is the
first and greatest commandment. And the second is like it: ‘Love your neighbor as yourself.’”(Matthew 22:36–39)
The Consequentialist Continuum
25
13. I shall omit from consideration the “third choice” of each spouse’s ethical regard focused on her- or himself since, so far as I can tell, it would be indistinguishable from ethical egoism. Cf. Erich Fromm’s (1956) critique of marriage (misconceived) as “egoism à deux.”
14. Remember that here I am talking about theories of ethical justification. Altruism as a motivation could still be important as an ethical strategy, as was suggested in
Chapter 1.
15. That is to say: with respect to the formal feature of focusing everybody’s ethical regard onto a single individual, the two theories would occupy the same point on
the ersatz continuum. Materially they would be different; for example, relative to
Adam, ersatz egoism would hold that he and Eve, as well as Cain and Abel, should favor him, while according to the ersatz form of my fiancée’s cousin’s theory, all must
favor Eve.
16. Logically this phrasing still applies in Massachusetts since the phrasing does
not preclude same-sex marriage.
17. “Surely” is surely not an argument. At this point let me therefore supplement
my account of philosophy’s reliance on reminders to include the drawing of attention
to things one might never have considered before (sometimes called “thought experiments”). Reasons must come to an end, that is, on pain of infinite regress (a fact to
which I draw your attention or of which I remind you); but, happily, for most practical purposes we need not argue beyond the reaching of shared assent. (Of course we
must also beware the danger of shared prejudices. Indeed, is not philosophy in the
very business of examining assumptions?).
Chapter Three
Let Us Boldly Go:
The Case for Utilitarianism
In our effort to understand what is for the best, we were immediately stymied
by the question: Whose best? We have found reasons to doubt that it could be
only one’s own or that of a couple to which one belonged. These two theories
(egoism and dyadism) are at the “near” end of a continuum of candidate consequentialist ethics that extends all the way to utilitarianism. Is the latter,
then, the true ethics? We might continue to balk at that conclusion because
utilitarianism is indeed at the other extreme. How could each one of us be responsible for every one of us? So let us first look to the middle ground to see
if a more plausible ethics could be found. Here we find a menagerie of theoretical beasts, including (in conjectural order of scope) communitism, nationism, ethnicism, religionism, racism, and sexism.1 Is this the proper range of
ethics?
Take nationism. This would hold that maximizing the aggregate good of
the citizenry of one’s country is the primary obligation of every person. Arguably, therefore, patriotism would be the most exalted virtue, treason the
meanest vice. Would war of each nation against all others be a logical consequence? Most unlikely. Even a two-front war is usually considered suicidal.
So some sort of “peaceful co-existence” would seem to be called for, as a
rule, on rational nationist grounds, with, I suppose, the occasional defensive
or offensive war thrown in, preferably in league with allies and against a single enemy. Indeed, analogously to the case of ethical egoism, we might look
for active cooperation among “enlightened” nations, or even hold out for the
possibility of a heartfelt camaraderie among “selfless” nations that would
tend to benefit all of them optimally and severally. We can also ponder the
distinction between a potentially ethical nationism and an ersatz one. In both
cases a given nation would properly be pledged to the paramountcy of its own
27
28
Chapter Three
welfare. However, the ethical nation would also recognize that every other
nation has the same primary obligation to itself, whereas the pseudo-ethical
nation would expect every other nation to support only its (the pseudo-ethical nation’s) welfare.
At first the ethical nation might seem to be at a curious disadvantage to the
unethical; after all, if one’s own welfare were the most important to oneself,
then why would one accord any weight to another’s, even in their estimation?
Wouldn’t it be a conflict of interest to respect another’s like prerogative in any
meaningful way, given its potential to negatively impact one’s own welfare?2
Now, insofar as this question is about the practical application of the theory, we have already answered it: the nationistically ethical nation cannot be
at any disadvantage since, ex hypothesi, it will adopt whatever attitude towards other nations best promotes its own best interest. That attitude, depending on the circumstances, could range anywhere from pure altruism to
pure jingoism.
The deeper question, though, is how to justify nationism, that is, the dividing up of loyalty by national boundaries. Viewed sub specie rationis (from the
viewpoint of theory) such division may seem completely arbitrary, for remember that the object of ethical concern for any theory on the continuum is
sentient beings.3 Thus, it is not “the nation” as such (whatever that might
mean) that one cares about but rather all those persons who are its citizens.
On the impersonal conception, it is conceivable that the nation might be better off under certain circumstances if many or even most of its legal citizens
were eliminated! We could imagine a Nazi believing this if she conceived the
nation as a geographico-“spiritual” entity that could only be embodied by
“Aryans,” who happened to constitute a minority of the current residents, and
even if this resulted in the diminution of the surviving Aryans’ general level
of personal welfare (economic, medical, etc.). Or citizen/persons might be altogether irrelevant, for example, if “the nation” were equivalent to a set of
ideals on the order of “give me liberty or give me death” or “live free or die,”
such that the welfare of the nation might on some tragic occasion entail the
sacrifice of its entire citizenry! Again one thinks of the Nazis, some of whom
took their own lives—even their children’s (e.g. the Goebbels)—rather than
submit to communist rule. But this is not ethical nationism as herein conceived.4
So the question remains: if the brand of nationism we are talking about is
on the continuum and hence is not “justified” by the value of a set of ideals
or some mystical notion of the nation or the like, then what would be the reason for stipulating that our (direct) concern for persons ends at the national
border (or, more precisely, is constrained by the legal definition of national
citizen)? Remember also that in order to count as ethical, as opposed to er-
Let Us Boldly Go
29
satz, the justification would have to apply to all national citizens as such
rather than as possessors of whatever feature(s) might happen to make one’s
own citizenry unique (such as their being “Aryan” or their valuing liberty
above all else). So what could that justification be? My answer is of course
that there is none! Just as with egoism and dyadism, there seems to be no
good reason to limit the ethical regard of any truly ethical agent to this limited set of sentient beings.
Then why have I “wasted” (y)our time with it? I do not think I have wasted
it. Indeed, the utility of pondering these theories lies precisely in the appreciation of their inadequacy, their lack of theoretical appeal. For my intent is to
demonstrate the need for a utilitarian form of consequentialism. And this effort is itself justified for at least a couple of reasons: (1) Utilitarianism may
seem at first utterly implausible because of the overwhelming nature of its apparent moral demand that we value all sentient beings and (2) Many people
do in fact feel a moral “tug” toward one of these less inclusive ethics, of
which they therefore need to be disabused or otherwise edified.
Interestingly it may be possible to resolve both the implausibility and the
“tug” at the same time. First consider the latter: many people feel allegiance
exclusively to the members of a group in such a way that other groups are recognized to have the same legitimacy. Sports teams are like this. Membership
is somewhat arbitrary; certainly for fans it is often simply the accident of
birthplace, as with nations. Yet the allegiance can be heartfelt; as would be the
disrespect for a player or fan of an opposing team who did not give his or her
“all” to their team. Analogously in the moral realm, for example, nationism:
do not many people feel it would be their highest duty to defend a fellow citizen, regardless of that citizen’s character (or even allegiance to the same
cause or group), solely because they are compatriots? And do not many of
these also fully appreciate the like duty of citizens of other nations, and indeed, see their own duty as exemplifying this broader virtue that all humans
owe to their respective countrymen? I think it is obviously so. All of us can
understand (without necessarily sharing or endorsing) the feeling that there is
nothing worse than a traitor, even if it is to the country of one’s enemy.
But such phenomena can find a more plausible justification in terms of utilitarianism than in nationism, I maintain. Consider again the analogy of sports
teams (putting aside commercial interests, although analogous considerations
might apply there as well). Is it paradoxical to want your opponents to be
good players and motivated to win even though you want your own team to
win? Not at all. For what would be the point, where is the “glory” in beating
a lackluster opponent? It is also good to be challenged to excel. But presumably what justifies caring about these things (excellence and winning in highlevel competitions) is not just the happiness of your teammates and yourself,
30
Chapter Three
since teams are relatively arbitrary associations, but their value for human beings as such. It’s just that such values could not be realized (to the max, anyway) if human beings were not broken up into competing teams. Just so with
ethics, for example, nationism: nations may not matter (just as particular
sports teams may not matter) so much as people do. Therefore nations may
be the means to the end of maximizing the welfare of people, who, due to
their nature and the ways of the world, may be more likely to thrive if they
form local allegiances than if it were always a case of “one for all and all for
one.”
Thus, it could be that something similar to, yet also somewhat the reverse
of, what we suggested for egoism holds for utilitarianism: ethical egoism
(everyone’s primary duty is to oneself) may recommend motivational altruism (the ethical agent should therefore care about others), and ethical utilitarianism (everyone has a duty to everyone) may recommend motivational nationism (the ethical agent should therefore care only about one’s
countrymen5). Similarly with religionism, ethnicism, and so forth: all of these
may fail as ethics, that is, as justifications of behavior, but play some justified
role (leading or supporting) in the conscious motivation of the ethical agent
in furtherance of the utilitarian end of maximizing universal welfare.
To continue to explore how this plays out, let’s now take a look at religionism (for God is in the details). This is the ethical theory6 that every person has as his or her primary responsibility the welfare of one’s co-religionists and oneself.7 A special issue for such a theory would arise from
proselytizing, since this kind of activity appears on its face to disregard the
respect that a truly ethical theory should accord the members of other groups
to watch out for their own interests. After all, is not the purpose of proselytizing to grow the ranks of one’s own religion? Yet in its very nature that
would, if successful, diminish the ranks of others,8 thereby bringing them
harm from the loss of strength in numbers.
Perhaps the distinction between an ethical theory and an ersatz one will
help us once again.9 Presumably the latter would proselytize (if this were in
the best interest of its adherents) with an attitude of indifference to the welfare (and derivative rights) claims of nonbelievers (except insofar as they affected the proselytizers). But ethical proselytizers would recognize the rightful prerogative of other religions to compete for converts in an agora of
religious ideas.10 As always, however, such predictions beg the circumstances
if they assume a straightforward mapping of the ethical theory onto motivation and behavior; for it is surely conceivable, for example, that even a religion on the ethical continuum might see fit to frustrate another religion’s public access and welfare, while an ersatz religion could find it useful to behave
magnanimously. Also keep in mind that what matters to the determination of
Let Us Boldly Go
31
the right thing to do are the facts and not merely the “perceptions” of the
agents; so for example the ersatz religionists could be mistaken that they
should behave magnanimously in some particular situation. The sole criterion
at all times would be: what is most likely, that is, in fact, to have the best
(long-term or over-all) outcome for one’s religious community?
Proselytizing also makes us puzzle in a new way about just whose welfare
matters since presumably the proselytizer is trying to save the souls of members of other groups. Again, this could only be justified by ethical religionism
if it were for the sake of the members of one’s own group, that is, religion;
and, as already noted, it is easy enough to imagine how increasing one’s own
ranks could benefit the membership. But this does then mean the “saving of
souls” rationale would be hypocritical, however sincerely this altruistic impulse might be experienced by the proselytizers. Could it be “theoretically
sincere” as well? What if the nonbelievers (i.e., the members of other religious communities) were conceptualized as potential believers or converts?
This would make them analogous to infants (not to mention, human fetuses)
and future generations, whose membership in a group is more or less problematic in any consequentialist ethics. In other words, does the injunction to
be concerned first and foremost about the members of one’s own religion (or
nation, race, sex, etc.) apply to its future and potential members as well as the
current ones? I leave that for you to ponder. Of course the answer has tremendous practical import, as such issues as the national debt, the exhaustion of
natural resources, and global warming attest.
Let me reiterate the importance of emphasizing the factual nature of outcomes in consequentialist ethics (even the ersatz ones). For example, this is
how we can resolve another ambiguity, namely, what it means to speak of the
welfare of a religion (or of a nation, etc.). Suppose Religion A decided that
proselytizing would not be a good idea since it would likely antagonize the
surrounding population and lead to pogroms (and otherwise have no compensating good effects for the religion or its adherents). This may be a rational assessment of the reality, and so the conclusion would be approved by ethical religionism: Thou (i.e., Religion A) shalt not proselytize. But suppose Religion
B also believed (correctly) that pogroms would result from proselytizing but
believed in addition that the best thing that could possibly happen to any of its
adherents would be to die a martyr’s death in furtherance of the faith? Would
religionism then endorse their decision to proselytize? I think the answer can
only be: By their lights, yet, but in fact, no. Who am I to decide what the facts
are? I’m me, the only person who can decide (for me) what I believe is true;
and I have no other access to truth except through my own beliefs. I believe
that martyrdom for the purpose of eternal reward does not in fact serve the best
interests of the (deluded) believers.11 Hence I (if a consequentialist) would find
32
Chapter Three
their ethical conclusion to be mistaken as well, since it is based on a false (factual) belief.
What if the purpose of the martyrdom were to help the religion increase its
power in the local region, for example, by inspiring admiration for the martyrs and instilling resentment against their persecutors? This could be a plausible claim to back up the religiously ethical rightness of proselytizing that
would result in such martyrdom. However, even here it is appropriate to inquire regarding the facts, that is, all the relevant facts, and not only about the
accuracy of the one prediction. For suppose martyrdom would indeed lead to
Religion B’s strengthening its grip on the society and so promote its welfare
in that sense; would this mean that religionism’s ethical demand had been
met? Not necessarily. What if the entrenchment of this religion had the net result (i.e., all other things, such as the economy of the region, remaining pretty
much the same) of, say, eroding women’s rights, and hence welfare, among
the membership? The believers, even many of the women (if only after a
time), might think that was a good thing. But, once again, they would be mistaken, and so what they thought was right would in fact be wrong.
Note that the relevant facts of the last example are themselves ethical in nature. What this shows is that, in consequentialism, determinations of right and
wrong depend on determinations of good and bad. Thus, proselytizing was
believed by Religion B to be the right thing to do because it led to martyrdoms that led to the reinforcement of the religion and its illiberal ideas and
hence to an illiberal society where men were free to beat their wives, and so
forth, and which its adherents could look at and pronounce to be good. My
disagreement with them would be on the assessment of the illiberal society,
for I see it as a decrease in value from the existing liberal society; in a word,
it is bad, or less good than the alternative. Hence I (if a consequentialist)
would conclude that the proselytizing that would lead to it is, even by religionism’s lights, wrong.
How far can we carry this line of reasoning? For example, what about the
idea that the ultimate dissolution of one’s religion (or nation, etc.) could be in
the believers’ (or citizens’, etc.) best interest? This smacks of paradox, like
the American major’s sanguine judgment that “it became necessary to destroy
the town [Ben Tre, Vietnam] in order to save it.” But a business bankruptcy
can be straightforwardly in the owners’ best interest even if the business
thereby ceases to exist. In our case, however, it does become problematical
how to identify the beneficiaries in the terms of the theory. Thus, suppose
abolishing a religion (not to mention, all religions) would be in the best interest of all of its adherents: could religionism as such mandate this? To what
extent does the welfare of a nation or a religion, and so forth, reside in its people as opposed to the ideals or doctrines it embodies? Would one be throwing
Let Us Boldly Go
33
the baby out with the bathwater to attempt to promote the people’s happiness
at the cost of their beliefs and allegiances? One thinks of the fate of Soviet
citizens, whose country dissolved before their very eyes. Supposedly this was
for their long-term betterment; could a good Soviet nationist have approved it
even so?
My sense is that we have indeed come up against a kind of paradox, but
which suggests a conclusion (as opposed to a resolution). Contemplation on
one theory after another on the continuum brings us always to the same suspicion, that only something wholly arbitrary keeps us from moving on to the
“next” one. Why should ethical boundaries be maintained around nations or
religions, and so forth, any more than around individual selves? So we kept
throwing the net wider and wider; for once we saw the reason to do it the first
time, what reason was there to dissuade us from doing it indefinitely? That is,
until the net encompassed everyone. Hence utilitarianism.
However, there is still an apparent objection to the argument that all groupings short of everyone (that is, all relevantly12 sentient beings) fail to be ethical due to arbitrariness. Even if that were true of theories like conjugism, nationism, and religionism, where human associations come about by personal
choice or social fiat, it would not be obviously true, goes the objection, of
such ethics candidates as ethnicism, racism, and sexism. The latter groupings
have a biological basis and hence cannot be considered arbitrary. So let’s take
a closer look at this question. First let us dispense with an automatic but misguided reply, which is that there is nothing ethical about racism, sexism, and
the like. I call this misguided not because it is question-begging but because
it is probably conflating the ersatz forms of these theories with the ethical
ones.13 What is morally obnoxious about Aryanism and discrimination
against women, and so forth, is, in the first instance, their counting of their
targets or victims as valueless. That is, they are deemed to be without intrinsic value—for Aryans can certainly find members of other ethnicities or races
to be instrumentally valuable or useful, for example, as slave labor, just as
men can find women valuable for sex, child-bearing, and household chores.
But no theory on the ethical continuum could get away with privileging the
members of any one group over some or all of the other like groups.14
And it is precisely because of such historical (and, alas, contemporary) patterns of abuse that a case could be made for the ethical counterparts of these ersatz theories. For given that the distinguishing features of the biologically based
groupings that ethics such as racism and sexism privilege are relatively conspicuous and ineradicable (putting aside such phenomena as racial passing and
sex-change operations), it is understandable why any one such group would
want to assert and accept the responsibility of watching out for their own, lest
they fall victim to abuse by other groups. Note, then, that the argument is not
34
Chapter Three
that biology by itself justifies these divisions of ethical valuing. Indeed, their biological basis has been questioned; for example, it is oft noted that a wider genetic variation obtains among the members of any one (supposed) race than, on
average, between races.15 Nor is the argument that their conspicuous and ineradicable nature justifies counting them as ethically significant; after all, adult
height is both but has not been widely regarded as having ethical significance.
Rather, it is that certain such traits or sets of traits have been used extensively
as the basis for discrimination, exploitation, and even genocide and therefore,
because they are conspicuous and ineradicable (possibly because biological),
constitute good grounds for the relevant populations to cultivate solidarity.
Thus, would it not be perfectly reasonable and fair and ethical to maintain that
every person’s primary obligation is to the welfare of (all the members of) her
race or sex, and without prejudice against the like duty of every other person?
An “ethical Aryan” of this sort would not be a racist since he would (at least sub
specie rationis) accord the same prerogative of racial-prioritizing to blacks and
Jews, and even, however grudgingly, respect them for it.
So why not, then, recognize the more “biological” groupings, such as race
and sex, as legitimate bases of ethical regard? I begin my reply in earnest by
pointing out that the issue is probably a red herring. Human begins have a remarkable penchant for persecuting or otherwise disadvantaging any sort of
difference from the norm16—indeed, even from the apparent norm (hence the
widespread use of “beauty” products by women to “enhance” appearance,
and the widespread hiding of academic prowess from one’s school peers by
teenage boys to “diminish” appearance). And certainly the “non-biological”
phenomena of religion and nationality constitute majors norms. Furthermore,
they too leave their “conspicuous and ineradicable” marks. Right now, for example, being an American is dangerous in some parts of the world; and to try
to disguise that identity would be well-nigh impossible for most Americans,
given their language, accent, manner of dress, behavior, knowledge (i.e., ignorance) of the world, and so forth. Similarly for religion: in a scene from the
movie Perlasca: The Courage of a Just Man (based on actual events), a Jew
who has been able to hide his ethnicity from the Nazis because of his atypical appearance is killed on the spot by a guard when he is unable to answer a
question about the Lord’s Prayer.
My argument thus far may seem an about-face, for it appears to justify ethical groupings, biological or otherwise, short of everyone after all. But the
real point is that allegiances to couples and families and communities and nations and religions and races and so forth could well have their places in the
best possible scheme of things for assuring general human thriving.17 As a
typical argument goes, one’s knowledge of needs and means is usually superior at the (relatively) local level; therefore one can help “the world” more ef-
Let Us Boldly Go
35
fectively by focusing on oneself and those who are close(-by) rather than on
strangers overseas, and so forth.18 Furthermore, even competition can be beneficial, as we saw in the discussion about sports. Capitalism, of course, is another telling example of that kind of thinking, for the motivationally selfish
(or egoistic) behavior of individual profit-seekers, turns out to be just what
the doctor of economics ordered for a thriving society; hence its justification
would be utilitarian.19 Now we have seen yet another reason to justify “local”
prioritizing of concern: protection from group prejudice. So people will continue to put themselves or their families or their countries and so forth ahead
of everything and everybody else, but what may make it right that they do so
is that everyone’s first obligation is to do their best to secure the maximum
welfare of all, and this division of ethical labor is the best means to achieve
that end.20
But wait: have we settled the matter yet—the matter of whose good matters? Or could there still be a parochialism about our ethics? Exactly where
have we got to? The progression from ego to dyad to family to community to
nation to ethnicity to religion to race to sex seems to leave us at the doorstep
of humanity. So is utilitarianism the same as humanism? Nyet. The greatest
proponents of utilitarianism, Jeremy Bentham and John Stuart Mill, saw at
once that that was only another arbitrary boundary when it came to considerations of genuine ethical import.21 For humans are animals, whose common
lot is our liability to pain and suffering (and perhaps as well a capacity for
pleasure and happiness). And surely suffering lays claim to ethical concern.
Ergo: “everybody” must include all sentient22 animals, including but not limited to humans.
In fact, I would argue, humanism falls flat on its face in its attempt even to
be in the running for an ethical theory. I think our diagrammatic scheme helps
us to see that humanism belongs on the ersatz continuum. For to the right of
sexism23 would properly be speciesism, would it not? And according to this
ethics, that is, on the ethical continuum, everyone’s first allegiance is to the
welfare of their species. So it would be perfectly proper for us humans to be
oriented first and foremost to the furtherance of our own welfare; however,
we could not deny other species the same (respective) prerogative.24 In addition, to replay a familiar theme: the best strategy for obtaining our, in this
case, human welfare could well be to suppress any explicit sense of our
species privilege.25
Does utilitarianism therefore equal animalism?26 For all practical purposes,
perhaps . . . for the time being. But utilitarianism is truly a Star Trek ethics.27
Strictly speaking, our concern for ourselves as one among many species could
be seen as falling short ethically of a concern for tellurians, that is, the members of our planetary biota and hence only one among the possible myriad that
36
Chapter Three
are sentient in the relevant sense. But then the same logic would oblige every
entity capable of being an ethical agent to be responsive to the ethical claims
of the totality of sentient life in the universe.28,29 That is utilitarianism.
NOTES
1. I have tried to avoid loaded terms, like “communism,” “communitarianism,”
and “nationalism,” but could not in every case. Perhaps the full label would sufficiently disambiguate, for example, ethical nationalism.
2. Frankena (1973) discusses this kind of problem with respect to ethical egoism
(pp. 18–20).
3. Again: following Mill’s lead, since the purpose of the continuum is to elucidate
utilitarianism.
4. The examples given do, however, have their equivalents on the consequentialist
continuum, that is, where the object of ethical concern is persons and not some mystical or idealist “nation.” Thus, one could imagine (and certainly find exemplified in
history and probably also at this very minute) ethical nationists who would choose
death, even auto-genocide, over treason or even assimilation, from a sense that their
own welfare as citizen-persons was indeed at stake. From this point of view, “better
red (or whatever) than dead” would make as little sense as “better warehoused in a
nursing home than dead” does to many people who prize the quality of life over the
quantity of their years of life. My entire peer generation may reasonably be contemplating mass suicide, as it were, for our own welfare in the face of the increasing likelihood of Alzheimer’s disease with the increase of average life expectancy (since the
condition is said to strike half of those over age 85).
5. Remember that this could also involve directly treating the citizens of other
countries well. A stark example is the reason currently be given for why even America-Firsters should support America’s observing the Geneva Conventions—in case
our American soldiers are captured by the enemy.
6. I am not claiming that religions are themselves ethical theories, although typically they will incorporate one. For example, Christian ethics would presumably hold
that one should treat others as one would like to be treated (or if that requires further
justification, then the theory might be that the right thing to do is whatever God commands, and God commands the Golden Rule; but that theory strikes me as ten times
more problematic than the Golden Rule itself). This is obviously quite different from
religionism, which privileges one’s co-religionists (unless one gives a narrow interpretation of Jesus’ enunciation of the Golden Rule as applying only to one’s Jewish
“neighbors”).
7. To satisfy our meta-ethical principle that an ethical theory must confer (ethical)
agency status on all persons of sound mind, let us number agnosticism and atheism
among the world’s religions. This is not so far-fetched; cf. Armstrong (1994) and Silver (2006).
8. Assuming the mutual exclusivity of religions, contra “Jews for Jesus,” Protestant
practitioners of yoga, Buddhist Shinto, etc.
Let Us Boldly Go
37
9. Please keep in mind that I am speaking here with respect to consequentialist
ethics.
10. Note that this would not imply a relativism of religious truth, any more than
does scientific debate imply a relativism about the nature of the physical world.
11. I suppose an exception might be in cases where the deluded belief itself brings
believers who are in truly hopeless straits such joy that it overwhelms whatever
worldly prospects these poor souls might have.
12. That is, capable of happiness or suffering.
13. Cf. analogous comments by Appiah (1990). I would like to thank Robert C.
Jones for bringing this article to my attention.
14. Although again keep in mind that ethical theories have to do with justification
and not motivation. Thus, as always, it is logically possible that empirical considerations could endorse an attitude that is prima facie contradictory to the consequentialist theory in question in certain circumstances.
15. Cf. Stolberg (2001):
“I have a standard teaching exercise that I do for my class every year,” says H. Jack
Geiger, a professor of community medicine at the City University of New York who studies racial disparities in health. “I ask two African-American male students to stand up, and
I ask two Caucasian male students to stand up. Then I point out that there is more genetic
variation between the two African-Americans than there is between them and the two
whites. And jaws drop all over the place.”
16. I was hard-put to come up with an example in the previous paragraph of a difference that has not been used as a basis for prejudiced discrimination, etc. Height
was the best I could do, but of course it too has been thus (ab)used.
17. Robert Wright (2000) tells this kind of just-so story.
18. “Think globally; act locally.”
19. Interestingly this is just the opposite (or rather the complement?) of my claim
in Chapter 1 that ethical egoism might justify altruist (or we could also say utilitarian)
motivation. Both are of course empirical claims, which might be false.
20. And ethical theory, therefore, may function like a Gestalt shift (as when the
figure of a duck changes into a rabbit), transforming our vision utterly even though,
materially speaking, everything stays the same; for family, national, and other such
loyalties may still turn out to be justified, but on a completely different basis from
what one had pre-theoretically supposed. One encounters this kind of phenomenon
across the philosophical spectrum, from the analytic, as we have just seen, to the existential, e.g., Kierkegaard’s (1843) Knight of Faith, who although a religious hero
may outwardly appear to be a common burgher sitting at his dinner table, to the mystical, e.g., the Zen oxherder, who looks like any other oxherder at the end of his extraordinary “journey” of Enlightenment (see Reps 1989, pp. 131–55). The late, great
comedian Victor Borge told of a man asking for a round-trip ticket at the train station
ticket window. “To where?” asked the clerk. “To here, of course!” answered the man.
21. E.g.: “The day may come, when the rest of the animal creation may acquire
those rights which never could have been withholden from them save by the hand of
tyranny” (Bentham 1789, ch. 17: sect. 1: part 1: note 1) and, as previously quoted,
38
Chapter Three
“This, being . . . the standard of morality . . . by the observance of which an existence
such has been described might be, to the greatest extent possible, secured to all
mankind; and not to them only but, so far as the nature of things admits, to the whole
sentient creation” (Mill 1863, ch. 2).
22. In the relevant sense, i.e., capable of suffering.
23. Or indeed, if one were broad-minded enough, even to its left!
24. Indeed, obligation, although what that would mean for the other animals on
this planet is problematic . . . perhaps calling for a reworking of the very notion of obligation? More modestly, we could simply say of other animals that they automatically behave in a way that would be ethically prescribed if they had the capacity to
be moral agents. Such a resolution would not be acceptable to Immanuel Kant, who
argued for an identity between moral agents or “subjects” and moral “objects.” But
Kant was not a consequentialist. Furthermore, I myself will argue for a different form
of Kantianism in Chapter 5.
25. Quinn (1995) highlights the oddity of humanity’s having assumed the unique
prerogative among all species of disregarding the others’ welfare at its own peril. I
would like to thank Sam Garner for bringing this novel to my attention.
26. And cf. Appendix III.
27. Hence my chapter title, since the title sequence of this popular science-fiction
television series was: “Space—the Final Frontier. These are the voyages of the starship Enterprise. Its five-year mission: To explore strange new worlds. To seek out
new life and new civilizations. To boldly go where no man has gone before.”
28. The question naturally arises, therefore, whether utilitarianism is compatible
with, not to mention implies, Starfleet’s General Order #1, the so-called Prime Directive: Do not interfere with the internal affairs of other civilizations.
29. And beyond? According to contemporary physics, there may be other universes
besides ours. However, we might be off the hook in their regard if we could not possibly affect them, or at least not know that or how we were affecting them. Isaac Asimov’s science fiction novel The Gods Themselves (1972) trades on this conceit.
Chapter Four
Refutation of Consequentialism
The injunction to do whatever is for the best has turned out to be not so simple to execute or even to understand. The situation is about to get a googolfold worse. We have seen that there is an indefinitely large number of ethical
theories on the consequentialist continuum. I have argued that utilitarianism
is the most plausible. Yet since I myself favor a nonconsequentialist ethics, I
shall now proceed to refute utilitarianism. There have been countless attempts
to do so since the theory was given its most rousing and articulate defense by
Jeremy Bentham and John Stuart Mill in the seventeenth and eighteenth centuries resp.1 The argument I find most persuasive, however, has, as I noted in
the Introduction, been relatively neglected in the critical literature, although
it was put forward forcefully at the outset.2 It is simply this. Utilitarianism is
predicated on our ability to know what the effects of our actions (and attitudes, etc.) will be3 . . . or at least to have a pretty good idea, or a justifiable
hunch. For according to this view the right thing to do is precisely that which
will have the best consequences.4 But it seems clear, once one reflects, that
all claims to such knowledge, even only with some degree of probability, are
mere pretensions.
Let me be clear about what kind of knowledge this objection is repudiating, which is to say, what kind of knowledge utilitarianism would require us
to have. There is a sense in which I no more know the result of my setting out
to drive downtown to meet a colleague for lunch than I know the man on the
moon, despite my having a specific appointment to be at a particular place
which I fully intend to keep. On what basis would I (or some scientist or statistician) be able to formulate even the likelihood of my showing up where I
am supposed to be? There are so many potentially relevant factors (a flat tire?
a traffic jam? a stroke? an incoming asteroid?) that it seems ludicrous even to
39
40
Chapter Four
begin to make the attempt to calculate their influence. However, that is not the
sort of knowledge my objection denies.5 For, despite our inability to apply a
number to my chances of keeping the appointment, it is a commonplace that
I am entitled to be fairly confident of doing so. The “exception proves the
rule,” as it were; I will be where I expect to be in enough such cases to entitle us to label my belief that I am likely to keep the appointment knowledge.
But the effects of an action do not halt after an hour has gone by, or a day,
or a week, or a year, or a millennium; indeed, they only grow.6 So I keep the
appointment. So what? Do I thereby know “the result of my setting out to
drive downtown”? Remember that we are talking about utilitarianism (although my remarks would apply as well to other theories on the consequentialist continuum mutatis mutandis). It seems to me obvious that the results
are incalculable and unknowable.7 For example, somebody seeing me driving
downtown—even somebody seeing me but mistaking me for somebody
else—could draw certain conclusions that lead to who-knows-what down the
road (literally and figuratively). And/or the person I am meeting ends up
meeting somebody at a party a week later he would not otherwise have done
(since he showed up at a slightly different time from when he would have,
had he not met with me) and . . . wedding bells two years later and . . . their
great-great-grandchild becomes a religious fundamentalist who carries a suitcase bomb into downtown Manhattan and blows it up. But that isn’t the end
of the story either, is it? For among the people wiped out in that explosion is
a person who would have become the great-great-grandmother of a psychotic
genius who would have invented a doomsday device that would have destroyed the entire planet. Finis? “It depends,”8 doesn’t it? If there happen to
be alien explorers millennia later who . . . well, you get the picture. Thus, the
issue is not whether we have total ignorance of the future (we do not), but ignorance of the total future (we do).
More to the point: it is not only the identity of the outcome(s) but also their
assessment, that is, as good or bad in the relevant sense, that is indefinitely
postponed. Thus, did my driving downtown have a good result? Yes and no:
it had both. But what matters for a utilitarian analysis is the relative net outcome. That is, first one needs to assign a positive or negative value to each
outcome; for example, it was good that I met my colleague, but it was bad
that the other driver who recognized me drew the wrong conclusion about
why I was on the road at that time, and it was also bad (indeed, much much
worse) that Manhattan was destroyed, etc. Then one adds up those values to
generate the net outcome of my having driven downtown: in this case let’s
say it has a net negative value. But we are still not through. Then one must do
the same for all of the alternative possibilities; that is, one must consider all
of the options that were available to me other than driving downtown, then
Refutation of Consequentialism
41
identify their likely outcomes, then assess every one of those outcomes, and
then derive a net outcome for every alternative option.
More simply, one could set up a single alternative, for example, not driving downtown. So in this case, had I not driven downtown, I would not have
met my colleague (give that a negative for the disappointment we both experienced), and the entire planet would eventually have been destroyed (give
that a very very big negative), and so the net outcome would have been very
very9 negative. Finally, one notes which outcome has the higher net value: in
this case although the net outcomes of both my driving downtown and my not
doing so were negative, driving downtown had a lower negative value and
hence a higher relative net value.
But of course it is beyond ridiculous that I or anyone, including any Cray
computer, could ever draw that or any other conclusion about the relative net
value of my driving downtown—the calculation would be so vast. Discovering (not to mention, assessing) the entire outcome of my action would tax our
knowledge abilities fantastically, and predicting the entire outcome(s) of the
alternative action(s) that I do not perform is an exercise in counterfactual omniscience that is well-nigh meaningless. In addition all of this needs to be
done speedily if it is to be of any practical use for advising me which option
I ought to choose, that is, what is the right thing to do. In sum: the utilitarian
determination of the rightness or wrongness of my driving downtown (or of
any other action, etc.) would be indefinitely postponed and hence useless,
which is, to say the least, ironic for a theory that is all about utility, and, ironic
or not, quite decisively eliminates the theory from contention as a guide to
life,10 that is, in “real time,” and hence as an ethics.11
Of course the utter contingency of existence comes as no news. But its relevance to the issue of consequentialism has been bypassed or minimized by
most theorists. Perhaps it is assumed that the epistemic objection has been answered by one or another decisive consideration. Let us look, then, at some
possible replies.
REPLIES TO THE EPISTEMIC OBJECTION
The Objection Trades on a Fluke
A utilitarian might protest that my example was simply bizarre, so that I have
biased the case by choosing a highly unrepresentative type of situation. For it
is not ordinarily that the fate of a major city or of the whole world hangs on
one’s decision to drive downtown to meet a colleague for a chat. But my
counterreply is that the example is not bizarre in the slightest. Why not? I say
42
Chapter Four
to you in total honesty that I believe everything in existence is at least as
“bizarre” . . . and hence nothing is (in the relevant sense).12
For example, it seems to me literally true that I, and you if you were born
after, say, 1950, almost certainly have Hitler to thank for our very existence.13
I am not referring to the Baby Boom, although that too could be adduced, but
more fundamentally to some plausible assumptions about human reproduction and personal identity. Since spermatozoa are as individualized as
snowflakes, if someone so much as read a newspaper headline about der
Führer (or was however indirectly affected later by someone who had), then,
at a minimum, the moment of their offspring’s conception was almost certainly different from what it otherwise would have been, hence also which
sperm successfully penetrated the egg to become the male gamete, hence also
the identity of the offspring. Therefore by, say, the year 2100, I surmise, and
thenceforward till the end of time, every living human being will owe his or
her existence to Hitler. But of course I should also say: to Hitler’s grandparents, and even more to the point, to the woman who said guten Morgen to
Hitler’s grandmother on that fateful day, thereby delaying grandmother’s going to bed that night by one minute, and so on. That is surely bizarre in some
sense, yet it is also par for the course.
Furthermore, I need not postulate that something momentous is always
at stake. My end-of-the-world scenario was only for dramatic illustration.
The refutation of utilitarianism would still be accomplished if even only
trivial outcomes predominated, so long as we could never determine which
of opposing options would have the more or less (“trivially”) good outcome. It seems to me, then, that the real bizarrerie belongs to utilitarianism for its holding hostage the ethical assessment of any action until the
end of time . . . or even for a short span. I have the image in my mind of a
person jumping into the water to save a drowning child: is it at all consistent with our notions of ethics and morality that we must suspend judgment
rather than approve this action on the spot, and more to the point, that the
would-be rescuer cannot decide at once that it’s the right thing to do? Must
we wait until we ascertain whether on balance the long-term outcome will
be better for the child’s having been saved? . . . not to mention, whether the
identity of the child was Adolf Hitler!14
We Are the Proof
The utilitarian replies: There is a glaringly obvious proof of the long-term efficacy of ethicality—none other than our own prevalence on this planet. From
having been rodents scurrying about at the feet of dinosaurs, we have evolved
physically, mentally, and morally to a position of dominance of the globe.
Where is the uncertainty of outcome?15
Refutation of Consequentialism
43
My counterreply: In the first place the argument may be beside the point if
not question-begging since it is far from clear that humans, not to mention our
ancestors, have been behaving ethically all the while. Ethics is about how we
ought to live and not in the first instance how we do live. Secondly, even if we
were to grant the premises of the argument—that our behavior has been by and
large ethical, and it has brought us to where we are today as a species—
the conclusion that utilitarianism has proven its mettle would not follow because the second premise refers only to the outcome for humanity and not the
totality of sentient life on this planet (not to mention, in the universe). And I
dare say that the lot of most animals has been diminished by human ascendancy; other species either are being extinguished at a rapid clip16 or at the
least decimated and the surviving members then often restricted in their range
of movement or activity or outright imprisoned and put on display for human
entertainment, or else have been domesticated and, except for pets in some
wealthy nations, their members are being raised under atrocious factory farming conditions before being sent to slaughter in the prime of their life.17 Third,
even ignoring the (negative) welfare of other animals, is it accurate to describe
the human condition today as one of (optimal) thriving? It is certainly true
(pace the Buddha) that some of us lead enviable existences. But if one were to
take an objective measure of human welfare and suffering around the globe,
would it come out net positive, or at any rate better than any conceivable alternative scenario? I am skeptical.
But finally, even supposing the current state of affairs really is the best in
the relevant sense (at least for humans), the utilitarian conclusion is (as, I
maintain, it must always be) premature; for we have not reached the end of
time.18 Doomsayers are never hard to find. Today they are mainstream: global
warming, exhaustion of finite resources, the proliferation of nuclear weapons,
and so forth. Each of these is a world catastrophe waiting to happen; and if
one or a combination finally does us in or at least reduces us to universal and
perhaps long-term misery . . . what then will have been the utilitarian conclusion regarding the relative net result of either (i) our actual behavior or (ii)
our presumed ethical behavior? I conclude that the human condition does not
constitute an obvious proof that moral behavior brings about the best outcome, nor, a fortiori, that the justification of moral behavior is that it would
do so.
Long-term Consequences Cancel Out
Even accepting my premise that the indefinitely far future is unknowable, the
utilitarian could draw a different implication, namely, that its bearing on a utilitarian decision can be discounted because the differences among the utility
values of the total consequences of the available options would be negligible
44
Chapter Four
over such a long span of time and hence the values essentially equal. Meanwhile there could be quite significant known or probable differential consequences in the shorter term, such as the likelihood of halting World War II and
the Holocaust if the attempt to assassinate Hitler on July 20, 1944, had been
successful. Therefore utilitarianism is perfectly “do-able.”19 But my counterreply is that the long-term calculations cannot be ignored for the following reason: according to consequentialism if the net result of doing Act A has a value
of googol units of good and of not doing Act A has a value of googol-plus-one,
then doing A would be wrong and not doing A would be obligatory. Therefore
Q.E.D. because even a negligible difference would therefore be decisive, but
even a gross difference could never be known. I call this the Camel’s Back Argument—as in “the straw that broke . . .”—since it shows that a single straw,
as it were, would make the difference between right and wrong according to
utilitarianism, which therefore20 refutes the theory.21
What I think the utilitarian has overlooked in his reply is that World War II
and the Holocaust would themselves be included among the consequences of
one of the options whose total consequences could never be assessed. Hence
they could never be compared to those of the other option(s), equally unknowable and, a fortiori, unassessable; hence there could never be a utilitarian calculation; hence there could never be a moral judgment or decision
based on them.
Rationality Requires Knowing the Future
Despite the questionableness (I would say, refutations) of the preceding attempts to discount the unknowable in utilitarian accounts of morality, a further argument for doing so is that rationality itself requires the same tactic. As
one devil’s advocate put it:22 “Your epistemological objection to consequentialism is strong. But is it too strong? Does it not undermine practical reason
as well? Any goal-directed behavior, ‘If you want x, the rational thing is to
do y,’ runs into the same objection that you put against ethical consequentialism, namely, ‘To whatever degree we are able to know the consequences of
our behavior, such knowledge will always fall laughably short of what we
would need to know in order to assess that behavior.’ So no course of action,
even practical action, can be said to be rational.”23
My counterreply is that adducing rationality actually works in favor of the
epistemic argument. Consider an engineer who conscientiously designed a
bridge which satisfied all existing regulations but which nevertheless collapsed due to a freak earthquake. Was the engineer irrational? Certainly not
prima facie. A perfectly rational decision or behavior can result in unexpected
disaster. For that matter, total lunacy can have surprisingly successful conse-
Refutation of Consequentialism
45
quences, but remains irrational for all that. Mutatis mutandis for ethicality. A
perfectly ethical decision or behavior can result in unexpected disaster. Meantime, total malevolence can have surprisingly benign consequences, but it remains evil all the same. Therefore it cannot be that the consequences determine the morality of the act. Just so for the rationality of the act. Thus neither
ethics nor practical reason has the infinite pretensions of consequentialism.
As I previously noted, the issue is not whether we have total ignorance of the
future (we do not), but ignorance of the total future (we do).
The utilitarian might further reply: But surely practical reason is all about
making sure that bridges don’t collapse. So isn’t there still a disanalogy between practical reason and ethics on your account if you do not hold ethics
accountable for its consequences? I reply in conclusion: I am only trying to
explain ethics and practical reason as we take them to be. It is wrong to torture a baby no matter what the overall consequences might be. It is not irrational to construct an imperfect bridge; indeed it would be irrational to think
one could construct a perfect one. The goals of practical reason are always
limited; but to be justified by consequentialism, a moral judgment would
have to be backed up by virtually infinite knowledge. Furthermore, it would
be a toss of the coin, for all you and I could tell now, whether the final judgment would be that it is or is not wrong to torture the baby.
Only Foreseeable Outcomes Matter
Ah! exclaims the utilitarian, there you have it: the answer to the epistemic
objection to utilitarianism is that it has dragged a red herring across our
path. For the utilitarian determination of the rightness or wrongness of an
action does not depend on the infinite and impossible determination and assessment of the ultimate outcomes of all of our options. It depends only on
the reasonably foreseeable outcomes.24 Your driving downtown for your
appointment was the right thing to do if it resulted in a pleasant occasion for
you and your colleague, and the alternative was your sitting at home being
lazy and your colleague being disappointed. These outcomes are easily
enough determined and easily enough “weighed.” (Mutatis mutandis for
saving the drowning child and for torturing the baby.) The analysis is utilitarian simply because it considered not only your own welfare but also your
colleague’s (and no one else seemed relevantly involved). And even if you
ended up being driven off the road by a semi and injured (and missed the
appointment), the decision could still be judged to have been the right one
if the accident were a fluke not possibly foreseeable by you (other than as
a statistical [un]likelihood). So the epistemic objection is no more than a
Zeno’s paradox, which means not a paradox at all because it only seems to
46
Chapter Four
make impossible what is a commonplace actuality (such as, in Zeno’s case,
a faster runner overtaking a slower one who started first).
My counterreply is that it is the above argument rather than my original objection that misrepresents utilitarianism. My interlocutor is quite correct that
people do commonly judge the rightness or wrongness of actions by their relatively foreseeable outcomes. But when they do so, are they behaving in a
utilitarian fashion? Let us even suppose that people usually consider the welfare of all known affected parties and that their (informal) “calculations” are
correct or rationally and empirically grounded: would this make them utilitarian? Only in a mimic way, I maintain.25 Recall that an ethical theory is
about justification—what makes an action right or wrong—and not, first and
foremost, about motivation, intention, or behavior. Thus, it may even be true
that people are behaving ethically when they judge actions on the basis of
their foreseeable consequences, but it would still not follow from such a fact
that that manner of judging displayed its underlying justification on its face.
A separate argument would be required, namely, that by judging things on the
basis of their foreseeable consequences, people are bringing about a better
world than if they didn’t judge in that way. But then it remains a completely
open question how one is to judge whether a better world would result from
such manner of judging. And it would quite beg the question to suppose that
the answer is to consider only the relatively foreseeable consequences of doing so!26
I can also refute the reply more substantively. For I ask: what would be the
point of grounding right and wrong in consequences if one were not concerned about the actual, that is, long-term relative net, consequences? Why
should the merely “reasonably foreseeable” consequences matter if one were
a consequentialist? The foreseeable consequences have practically nothing to
do with the relevant, that is, long-term relative net, consequences;27 my enjoyable meeting with my colleague contains not one single clue28 about the
subsequent destruction of Manhattan or, a fortiori, the potential Armageddon
fiend. So what possible consequentialist justification could there be for considering, not to mention privileging, a subset of consequences that can in no
way provide evidence for the relevant consequences of one’s actions?
The essential intuition of a consequentialist is that the only sensible basis
for making decisions is whether one is helping, or at least attempting, to bring
about the best outcome. Would one be doing that if one had no idea what the
ultimate outcome will be?29 Would one even be attempting to bring it about
if it were practicably indeterminable? One might as well say that one is doing one’s best to predict the future by flipping a coin.30 This seems to me like
playing a game. But ethics is not a game. We do not want merely to pretend
that we are doing the right thing, or to act as if we were good people. Ought
implies can, not appear-to-be-able-to!31
Refutation of Consequentialism
47
Utilitarianism Is about Tendencies
The utilitarian replies: What makes consequentialism do-able is that our
knowledge of the future is based on tendencies. Mill’s original formulation of
utilitarianism made this explicit: “[A]ctions are right in proportion as they
tend to promote happiness.”32 What has become known as the rule-utilitarian
approach therefore allows for the occasional exceptional circumstance, such
as winning the lottery, or an innocent child becoming the monster of the
world. These contingencies do not make it right to have bet the farm on the
winning number, or wrong to have saved the drowning child. Thus, a rule like
“Thou shalt not lie” gains its purchase from the observation that universal deceit is self-defeating: if everybody were lying all the time, nobody could ever
believe anybody, and all business and even personal relations and hence civilization would collapse. Of course there are occasions when lying will have
better consequences than telling the truth, but it would be better to tell the
truth even then33 since utility also accrues to maintaining the force of the rule
(not to mention the consideration you have stressed: that one never really
knows the ultimate outcome of any individual act).34
My counterreply: The justification of moral rules on the basis of their predicted total efficacy is no more feasible than the like attempt to justify individual acts.35 Is lying wrong? As always with utilitarianism, the answer is: “It
depends.” The question would have to be refined indefinitely for an answer to
have any hope of correctness, that is, of due sensitivity to circumstances; for
example, “is it wrong to lie solely out of self-interest?” is obviously a different issue from “is it wrong to lie in order to lend significant assistance to an
innocent other?” That is why some philosophers have argued that rule-utilitarianism reduces to act-utilitarianism in the end,36 for articulating the circumstances of application sufficiently to warrant the moral assessment of a rule
may be tantamount to describing an individual act; in which case we would be
back where we started, trying to figure out the long-term relative net consequences for all sentient beings of A’s doing x on some particular occasion.
It could be argued, however, that this problem would not even arise for a
thoroughgoing utilitarianism since the prohibitive costs of inculcating and applying an indefinitely large set of indefinitely long rules (comparable to the set
of all possible acts) would rule out propounding such a set on utilitarian
grounds. Instead a genuinely utilitarian set of moral rules would in its very nature be manageable, indeed, optimally so.37 Even so, however, it is the entire
set of rules (that is, some degree of compliance with or acceptance of them)
whose utility we would need to ascertain before we were justified to adopt it
over all alternative manageable sets; and this would be just as daunting—I
would say, impossible—an epistemic task as that posed by any other form of
utilitarianism.38
48
Chapter Four
“But surely,” the utilitarian might insist, “it is absolutely reasonable to suppose that the common rules of morality regarding killing, lying, promisekeeping, and the like, contribute to a better society than one that was bereft
of them.”39 Yes, I could agree. However, my constant theme is that a proper
consequentialism must be concerned, not with the appearances, but with the
reality of outcomes.40 And, for all anyone ever could determine, the preservation of a well-functioning and even happy society on this planet could lead
to a worse consequence for the Milky Way Galaxy41 than its alternative, by
some fluky (i.e., contingent, i.e., normal) sequence of begettings and other
events—and no matter whether the consequence were catastrophically worse,
à la the Holocaust resulting from a highly civilized Germanic culture, or infinitesimally worse, à la the Camel’s Back Argument.42
A last-ditch attempt to defend a utilitarianism based on rules in light of the
Epistemic Objection is to argue that the rules establish blameworthiness
rather than wrongness.43 For example, if one followed a rule not to lie but as
a consequence brought about an avoidable calamity, one might then have
done wrong (because the outcome was non-optimal) but would not necessarily be held morally accountable.44 This approach does at first seem to answer
the epistemic objection, which is that the relevant consequences cannot be
known and hence the ethical value of the act (as right or wrong or permissible) is unknowable and hence utilitarianism is false, since it undercuts that
last inference. But what kind of a “save” would this really be? First of all, on
what basis would the rules of blameworthiness be established? Ex hypothesi
it could not be the moral value of the acts performed since that is unknowable; but by my arguments above, it could not be the (relative net) utility of
the rules either, for that is also unknowable. Second and even more tellingly,
what then would be the point of wrongness45 (or rightness or permissibility)
if it had been conceded to be unknowable46 even in a quotidian sense of
know? An answer might be that moral value serves as a regulative ideal, giving us a target at which to aim. But, to repeat my refrain: How can something
utterly unknowable serve as any kind of guide to action or life? Ethics itself
would thereby become pointless, or else just in the business of establishing
blameworthiness, albeit on a mysterious or irrelevant basis.
Egoism Redux
I think the utilitarian has given his best shot, but there might still be one last report from the consequentialist camp. Perhaps it is precisely the utilitarian interpretation of consequentialism that has been the source of the difficulties. Utilitarianism demands that we consider the welfare of all and into the indefinite
future. That would indeed involve an immense “calculation” beyond the power
Refutation of Consequentialism
49
of any conceivable supercomputer. But that is why a more “modest” consequentialism is plausible; and the most modest of all is none other than utilitarianism’s chief consequentialist alternative, egoism. The latter theory stipulates
that each individual consider the consequences of his or her acts only for himor herself. That is do-able. We are able to make useful predictions regarding our
personal welfare: the time span is short enough, and we are in the best position
to know the relevant considerations. Betting the farm on the lottery is highly unlikely to work out the best for oneself; hence it would be wrong to do it.
A twofold reply is in order. First, it is arguable whether self-prediction is
superior to social prediction. When I think about my own life and key events
that have shaped my fate, I feel completely at a loss to decide whether things
have turned out for the best. I may feel very sure that they have turned out
well or poorly; but how could I ever know that they might not have turned out
better or worse47 (not to mention, how to assess them in light of future contingencies) had I done otherwise?48 A fortiori for predictions on which to base
future behavior. As Kant (1785) put it:
[M]en cannot form any definite and certain concept of the sum of satisfaction of
all inclinations that is called happiness. Hence . . . there is no wonder that a man,
e.g., a gouty patient, can choose to enjoy what he likes and to suffer what he
may, since by his calculation he has here at least not sacrificed the enjoyment of
the present moment to some possible groundless expectations of the good fortune that is supposed to be found in health. (p. 399; cf. also p. 418)
I conclude that some if not all of the major determinants of my happiness and
suffering are quite beyond my ken at those times when I might do something
about them. This applies as well to rules of prudence, which seem to me no
more empirically grounded than rules of morality. Can we with any more assurance predict that going to college will lead to a happier life than that legalizing abortion will lead to a happier society?
But there is a more fundamental problem with egoism, as I argued in
Chapter 1: it provides an implausible justification of morality. The idea that
only one’s own welfare determines the value of one’s conduct is ridiculous
on its face. All that has now been put forward to counter that is its (purported) do-ability (relative to utilitarianism). But a variant of the further objection of Chapter 1 remains relevant, to wit: unlike the very plausible metaethical criterion that ought implies can, its inversion—can implies ought—is
surely insupportable. Granted, the do-ability argument now rests on epistemic rather than psychological grounds. But just because one can do something is hardly a good reason for asserting that one ought to . . . indeed,
even49 for being ethically permitted to. Therefore ethical egoism is no “save”
for consequentialism.
50
Chapter Four
This, then, is the essence of the argument against consequentialism: To
whatever degree we are able to know the consequences of our behavior (and
attitudes, etc.), such knowledge will always fall laughably short of what we
would need to know in order to assess that behavior on a utilitarian basis.50
This is because the reasonably knowable consequences of our behavior bear
a minuscule relation to the relevant consequences, that is, the long-term relative net consequences of our actions. But utilitarianism is the most plausible
consequentialist theory. Therefore consequentialism is defeated.
But is there a viable alternative?
NOTES
1. The canonical texts are Bentham’s Principles of Morals and Legislation (1789)
and Mill’s Utilitarianism (1863). A review of objections can be found in Scheffler
(1988).
2. Indeed, before the “outset”; for I also find it in Kant’s (1785) discussion of happiness (pp. 418–19).
3. Strictly speaking the premise is that the rightness (or permissibility or wrongness) of acts (etc.) is knowable, for otherwise a utilitarian could bite the bullet and accept the implication that, her criterion of rightness being never within our epistemic
reach, rightness can never be ascribed. (Cf. the discussion of blameworthiness in the
sequel.) Thus, and in keeping with the underlying significance of the punning title of
this book, my argument presumes that ought implies can, not only in the usual practical sense of ruling out prescribed acts that one is unable to perform, but also in an
epistemic sense, namely, of ruling out acts whose ethical value one is unable to know
(as well as anything can be known). Otherwise philosophy (or, technically speaking,
philosophical ethics) could not hope to be “the guide to life.” Hudson (1989) makes
the same point (p. 227), although he allows that acts might have two different kinds
of ethical value, namely, the right/ought and the best/ideal, in which case the practicability requirement would apply only to the former.
4. Marcus G. Singer (1977) points out that it is only with Henry Sidgwick and G.
E. Moore at the beginning of the Twentieth Century that utilitarianism acquires the
additional stipulation that any act having suboptimal results is wrong. He argues that
prior to that the act having the best consequences was taken to be the best thing to do,
but things having middling consequences might still not be wrong, the latter designation being reserved for whatever had the worst consequences. That sounds at first
blush like a more plausible consequentialism; however, I don’t see that it would put a
serious dent in my objection (to follow) since our knowledge would not have any purchase on whether a given act would have even middling results. If one were to reply
that most things are “in the middle,” so therefore it is reasonable enough to infer that
one’s acts do have middling results and hence are morally passable (cf. physicist J.
Richard Gott III (1997)’s analogous “Copernican formula” of prediction), my reply
would be that that argument is really a reductio because it would make morality point-
Refutation of Consequentialism
51
less, for on that account just about anything would be permissible.
5. Cf. Hudson (1989): “The skepticism I am relying upon here is very modest. . . .
This is merely a matter of common sense, and does not suggest any broader philosophical skepticism” (p. 229).
6. Cf. the phenomenon in chaos theory known as the butterfly effect, after a scientific paper by the mathematician and meteorologist Edward Lorenz (1972).
7. Cf. Frazier (1994): “By considering the nature of causality, we would conclude that it is practically impossible to ascertain all of the consequences of any
action” (p. 47).
8. The consequentialist refrain.
9. I am using these qualitative descriptors to avoid the absurdity of trying to assign numbers to these values.
10. Cf. Hudson (1989): “Thus for human agents the theory is not really ‘actionguiding’: it does say what one should do, but it gives this information in a practically
unusable way” (p. 221).
11. The example suggests a related objection to utilitarianism. For given that
the consequences of driving downtown to meet somebody for lunch are potentially
(or even actually) momentous, any act whatsoever (including “non-acts,” like staying in bed instead of getting up), no matter how trivial on its face, would become
a matter of serious ethical concern if ethics were consequentialist. Cf. Langenfus
(1990): “Morality, according to this conception of the right, is utterly ‘pervasive’
in the sense that each and every occasion of an agent’s life is subject to actual, occurrent . . . moral requirements” (p. 133). But of course the vast majority of things
we do simply do not rise to the level of ethical issues; therefore Q.E.D. Call this
the Leviticus Objection, after the book of the Torah that raises even some of the
most trivial acts to the level of abominations. Interestingly, however, as our knowledge of consequences increases in the global village, we begin to see the kind of
ethical world a utilitarian might envisage; thus, we now question the casual drive
downtown for lunch in light of global warming. (Langenfus, however, goes on to
argue that taking the long view of consequences—what he calls the “sub specie aeternitatis perspective”—would in practice have the opposite effect of reducing the
apparent moral significance of almost every act to practically nothing (p. 138). I
think this is a variant of my Illusion Argument in the sequel.)
12. In a similar vein, Thich Nhat Hanh (1988) points out that a piece of paper
“contains everything in the universe” (p. 5); and “Please remember,” he asserts, “that
we are not talking about philosophy [here in the sense of idle speculation, I suspect—
JM]; we are talking about reality”(p. 22). Or as I myself put it elsewhere: “I view the
popular conception of philosophers as ones who have their heads in the clouds to be
quite the opposite of the truth. The true philosophers are precisely those who have
their feet planted firmly on the ground; they are not dealing with airy abstractions, but
with reality, while unreflective folk flit about among the appearances” (“Thought for
Food—Re: Veal” in Philosophy Now no. 62, July/August 2007, p. 48).
13. And also, if you are an analytic ethicist of a certain age, for your favorite
knockout thought experiment. I call mine “Hitler’s Humanity.” The next generation
of American philosophers will no doubt rely on Osama bin Laden for theirs.
52
Chapter Four
14. Smart (in Smart and Williams 1973) famously bites the bullet on this one—
“He [the rescuer] would have done the wrong thing, for he would have saved the
world a lot of trouble if he had left Hitler below the surface” (p. 49)—although he attempts to compensate by asserting we should nonetheless praise the rescuer. (Cf. the
discussion of blameworthiness in the sequel.) I believe I have read that there actually
was some such event in Hitler’s childhood. But of course it is also not unusual for a
child’s life to be “saved”—indeed, it happens every time you watch the baby in the
bathtub.
15. Cf. the following passage from Dawkins (2006), which seems to imply a general affinity between natural selection and utilitarianism:
If a wild animal habitually performs some useless activity, natural selection will favor rival individuals who devote the time and energy, instead, to surviving and reproducing. Nature cannot afford frivolous jeux d’esprit. Ruthless utilitarianism triumphs, even if it
doesn’t always seem that way. (p. 191)
16. See, e.g., Wilson (2002).
17. See, e.g., Singer and Mason (2006).
18. I owe this point to an actual interlocutor, but I cannot remember who it was!
Lenman (2000) makes an analogous point (p. 354). This is also why I cannot fully accept the otherwise fiendishly clever refutation of utilitarianism by Marcus G. Singer
(1977), which begins with the observation that (what he calls) actual consequence utilitarianism is “retrospective, it looks back, and provides at best a basis for judging what
ought to have been done. Before the event we can only judge with some probability
what the consequences will be, and there is always some appreciable chance that we
shall be wrong. It is only after the event that we can determine, with reasonable certainty, what they actually were” (p. 70). But I am arguing that there is no such thing as
a time “after the event” when any final inventory of its consequences might be made.
19. Peter Singer presented this argument to me (in conversation on 14 September
2007). Smart (in Smart and Williams 1973) also says as much: “. . . we do not normally in practice need to consider very remote consequences, as these in the end approximate rapidly to zero like the furthermost ripples on a pond after a stone has been
dropped into it” (p. 33). Moore (1903) makes much the same argument (§ 93). Kagan
(1998) similarly dismisses the Epistemic Objection in a parenthesis: “. . . there will
always be a very small chance of some totally unforeseen disaster resulting from your
act. But it seems equally true that there will be a corresponding very small chance of
your act resulting in something fantastically wonderful, although totally unforeseen.
If there is indeed no reason to expect either, then the two possibilities will cancel each
other out as we try to decide how to act” (p. 65). Meanwhile, Frazier (1994) refutes
Moore (pp. 46–49), and Lenman (2000) gives Kagan a shellacking.
20. Because such minuscule differential consequences could not be known, and
ethicality requires knowability.
21. Lenman (2000) takes a different tack when he refutes both “peters out” and
“cancels out” types of argument, emphasizing instead that the differential outcomes
are likely to be large (pp. 350–59).
Refutation of Consequentialism
53
22. An authentic advocate of the same point is Mason (2004).
23. Jerome A. Shaffer (personal communication, 20 May 2008).
24. Feldman (2006) focuses on this claim and finds it wanting (with, let me add, a
methodological deadpan that can be hilarious, e.g., p. 61). Feldman’s main point, I
think, is that “the expected utility of a given action on a given body of evidence is an
objective matter of fact” (p. 64); hence the presumed gain of practicality over “actual
utility” is simply mistaken. Indeed, he argues, “the determination of expected utility
is even more problematic [than of actual utility]” (p. 56). My refutation takes a different tack, to wit: foreseeable utility is irrelevant to a proper consequentialism.
25. I call such calculations faux utilitarianism, which I will discuss at greater
length in Chapter 6.
26. Cf. Hudson’s (1989) analogous argumentation: “Isn’t it obvious that [utilitarianism] tells the uncertain agent to try to maximized [sic] utility? No, that would be
quite a misunderstanding of the plain content of the theory. The theory says nothing
explicitly about trying; hence it must be applied to tryings as to any other acts. One
ought to try to do X, according to the theory, just in case this would maximize utility”
(p. 221). Hudson believes, as do I, that maximization of utility is impractical as an
ethics; however, he is more sanguine about the prospects of a subjective form of utilitarianism, wherein the criterion of morality is not utility but expected utility. As a descriptive analysis of the empirical phenomenon of human morality, his case might
have traction. But as a justification of morality, I think it falls before the “substantive”
objection of my immediately following paragraph.
27. Harvey Green (in a personal communication) has commented that the more
precise assertion is that “the foreseeable consequences have no knowable relation to
the relevant, i.e., relative net, consequences.” The second clause of the compound
sentence (about “not one single clue”) is more to that point.
28. After writing this I discovered that “cluelessness” is similarly the cornerstone
of Jimmy Lenman’s (2000) delightful refutation of consequentialism. As he writes,
“The worry is not that our certainty is imperfect, but that we do not have a clue about
the overall consequences of many of our actions” (p. 349). I would go further, which
he modestly declines to do (p. 359), and say “of any of our actions.”
29. Lenman (2000) puts it this way: “For a consequentialist the point of maximizing the goodness in the visible consequences of our actions has to be as a means to
maximizing the goodness in their overall consequences. But . . . we have only the
most feeble of grounds to suppose that means—or any feasible other—is a remotely
reliable means to this end” (p. 364).
30. Note that I do not mean to be suggesting that flipping a coin has no place in
ethical decision-making. Cf. DeLong (1991).
31. Interestingly Moore (1903) states my main case precisely: “Our utter ignorance of the far future gives us no justification for saying that it is even probably right
to choose the greater good within the region over which a probable forecast may extend” (p. 153). Of course Moore then tries to remove the sting of that apparent death
blow to utilitarianism, but surely fails, as I argued above with my Camel’s Back Argument.
32. Utilitarianism (1861), ch. 2; my emphasis.
54
Chapter Four
33. Alternatively lying could be endorsed on such occasions (if per impossible the
ultimate outcome could be known to be for the best!) precisely because the theory is
about rules of ethics rather than laws, for only the latter would be exceptionless.
34. Curiously this seems to be the very argument Kant gives in his “On the Supposed Right to Lie because of Philanthropic Concerns” (1799), even though he was
an avowed anti-consequentialist.
35. Cf. Shaw (2000): “. . . the difficulty of knowing the full consequences of our actions cuts two ways: both against the claim that sticking to the rule will have good results and against the claim that violating it will” (p. 16). See also Hudson (1989), p. 223.
36. See Hooker (2008) for a review of the relevant literature and arguments.
37. Hooker (2008) makes this point.
38. Whewell (1864) stressed this point.
39. Moore (1903) argues in this way; e.g., “On any view commonly taken, it seems
certain that the preservation of civilized society, which these rules are necessary to effect, is necessary for the existence, in any great degree, of anything which may be
held to be good in itself” (p. 158).
40. Cf. Marcus G. Singer’s (1977) refutation of “actual consequence utilitarianism,” which is that it is incoherent (and not merely impossible, as I argue) to base a
decision or even a judgment on the best actual consequences of an act since the only
thing that could have such consequences is the act that is (actually) performed (and
mutatis mutandis for a rule, etc., I would say); hence no comparison would be possible with any alternatives, since, not having been performed, they will not have any
(actual) consequences; hence there is no basis for arguing that the act performed had
the best consequences.
41. Or “the Universe,” to cite Moore’s (1903) ultimate criterion of rightness (e.g.,
p. 147).
42. In his Hitchhiker’s Guide to the Galaxy (e.g., in the novel of that name published by Pan Books, 1979), Douglas Adams suggested that our planet was in the way
of a hyperspace bypass. We could imagine that its construction is critical to the welfare of the galaxy as a whole, or even might only make the galaxy, i.e., its sentient
population, slightly better off than a galaxy containing Earth but no bypass.
43. Although strictly speaking this distinction can also be applied to acts; e.g., see
Sider (1995).
44. Railton (1984) speaks approvingly of this strategy, although he does not commit himself to rule-utilitarianism. Also keep in mind that, however comforting such
an approach may sound when couched in terms of exoneration, it also implies, of
course, that one would be blameworthy for doing the right thing if it happened to involve violating a rule.
45. Both Hooker (2008) and Lenman (2000) make the same observation.
46. Of course a typical utilitarian would more moderately hold that it is only on
some occasions that the relevant consequences are (reasonably) unknowable, as opposed to the epistemic objection that we never have even an inkling of such knowledge. Yet the literature of utilitarianism does contain the extreme position, for example, in Moore (1903), thus:
Refutation of Consequentialism
55
In order to shew that any action is a duty, it is necessary to know . . . all the events
which will be in any way affected by our action throughout an infinite future. . . . and
further we must know accurately the degree of value both of the action itself and of all
these effects. . . . And not only this: we must also possess all this knowledge with regard to the effects of every possible alternative. . . . Accordingly it follows that we
never have any reason to suppose that an action is our duty: we can never be sure that
any action will produce the greatest value possible. (§91)
47. For this reason some consequentialists have defended a criterion of “satisficing” rather than “maximizing”. Our goal in life should be to make things well enough
rather than optimal. See e.g., Byron (2004). But even if this might “work” as practicable, it would still leave unaddressed the second, more telling objection to ethical
egoism (to follow).
48. I do not even address the conceptual issue of what constitutes human happiness, which would only make the consequentialist’s predicament more intractable.
49. Unless one has no other option? But there is certainly a viable alternative to
consequentialism, or so I shall argue in the sequel.
50. Thus it is another glorious but failed empiricist project, on a par with the attempt to construct all human experience out of sense-data, e.g., Russell (1912). The
special irony of the utilitarian project is that one of its supposed advantages was precisely to turn ethics into a precision, scientific tool; but my refutation shows that it
fails utterly as a practical means of pinpointing the right thing to do, far more so than
even the most unreflective intuition, not to mention a sophisticated approach like
Kantianism (to which I now turn).
Chapter Five
Nonconsequentialism and
the Consequentialist Critique
If sufficient justification of right actions and the good life cannot be derived
from the consequences of acting and living in the recommended ways, then
what alternative is there? After all, consequentialism was plausible on its
face; so if it fell flat on that face, can we expect to find a more smiling
prospect elsewhere? At least we do have a ready name for the alternative:
nonconsequentialism. But that is a mere negation: a frowning face. What is
its positive content? It turns out, as with consequentialism, there is a variety
of possibilities to explore.
But first I should point out the obvious: dividing the domain of ethical theories into consequentialist and nonconsequentialist ones was a choice. Of
anything whatever it can be said that it is either x or not-x. So one could postulate of ethical theories that each and every one of them is widely called in
English by a name having the same number of letters in it as the word “asparagus,” or else it isn’t, and then assigning all theories to those two camps.
But what would be the point? More plausibly, one could pick a more germane
or salient feature of ethical theories, such as their relative emphasis on feeling (e.g., “moral sentiment”), and declare that all ethical theories are either
(predominantly) “affective” or “nonaffective.” I have chosen to cleave the
realm at consequences for two reasons: (1) the analytic tradition of philosophy in which I was trained does so, and (2) perhaps for that reason or due to
some personal quirk or because the distinction itself beckons those who study
ethics, the issue simply captivates me, and has for decades. So: no justification, nor even an explanation, but just to advise you that it was a choice. If
you find the present study to be profitable, then consider that a kind of justification ex post facto. But might an alternative cleavage be even more profitable? Be my guest.
57
58
Chapter Five
However, perhaps I can suggest a reason for this choice after all, since even
as we turn our gaze to the nonconsequentialist theories, consequentialism
haunts us. By this I mean two things. First is the overwhelming intuitive appeal of the essential consequentialist command: Do whatever is for the best
(or more precisely, do your best to bring about whatever is for the best). Although I have argued that this falls apart upon closer scrutiny, it remains enticing. I would compare the effect of my argument to discovering that one is
seeing a visual illusion, for example, pulling a stick out of the water and observing that it is not, despite its previous appearance, bent but straight; and
yet when the stick is re-immersed, it still looks bent. An even more visceral
example is peering over the rail of a bridge to gaze on the flowing water below and feeling oneself move in the opposite direction, despite one’s full
knowledge of the scene. Just so, even if the authority of the consequentialist
command has been turned to mincemeat, it could still strike us unreflectively
as irresistible.1
Second, the consequentialist is ready with her objection to any nonconsequentialist theory we might propose, to wit: insofar as any ethical theory is
powerful, it owes that power to consequentialism. If a theory is overtly nonconsequentialist, then the consequentialist agenda is hidden. I will call this
the consequentialist critique. I won’t attempt to apply the critique to the
panoply of nonconsequentialist theories simply because that would steer us
too far afield. Nor have I discovered an ordering scheme for the nonconsequentialist theories, analogous to the consequentialist continuum, which
would have made it easier to summarize the consequentialist critique. Furthermore, again unlike the continuum series, which I mostly made up myself
solely for illustrative purposes, the nonconsequentialist theories have been
fully developed as serious contenders by their respective proponents. These
include: divine command theory (do as God tells you to), karma yoga (fulfill
your caste duty2), Confucianism (cultivate ren: humanity or gentility), Taoism
(act without acting or wu-wei), Buddhism (develop non-attachment and compassion), Zen (be here now), the Golden Rule (do unto others as you would
have them3 do unto you), virtue theory (build up character), Kantianism
(never treat anyone merely as a means), existentialism (make your own meaning), feminism (one variant: practice caring), and the beat goes on.
What I shall do, therefore, is illustrate the consequentialist critique with the
archetypal nonconsequentialist theory, divine command theory, and then I will
turn to the defense of the theory I myself (along with, I estimate, a plurality of
contemporary analytic philosophers) consider to offer the most effective alternative to consequentialism, namely Kantianism (or some variation thereon,
which may, as well, contain elements of other nonconsequentialist theories).
Nonconsequentialism and the Consequentialist Critique
59
DIVINE COMMAND THEORY
Most likely the granddaddy of nonconsequentialist theories is divine command theory. This holds that right and wrong are laid down by God by fiat,
or as if by fiat since God may have good reasons that are ultimately incomprehensible or unknowable (or even unacceptable!4) by us. The archetypal
image of this theory is of course Moses coming down from the mount with
the Decalogue on the tablets of stone. But as the consequentialist (not to mention, just about any other sort of nonconsequentialist) sees it, this conception
of the ground of ethics risks making ethics arbitrary, since it would be based
on the free will of a being who acts in mysterious ways, and hence falls short
of an intelligible justification. But the consequentialist does in fact see two
obvious ways to make sense of divine command theory that are none other
than the alpha and omega of the consequentialist continuum: egoism and utilitarianism.
The egoist parsing of divine command theory would be that God’s promulgation of the moral law(s) is a thinly veiled, or sometimes not at all veiled,
threat: obey or be damned! Thus, the arbitrariness, or arbitrariness-for-all-weknow, of ethical injunctions would remain, but the justification of our heeding them would become clear: our personal welfare depends on it. In fact this
could be pushed further along the consequentialist continuum since, according to the Old Testament anyway, God’s wrath can be visited upon whole
cities that contain sinners, or a whole people, and as well “unto the n-th generation”—indeed, with Adam and Eve, all the way to utilitarianism, since
their violation of God’s commandment had consequences for the whole of humanity as well as snakes and all the other animals of the (no-longer) peaceable kingdom.
The divine command theorist would reply that the above represents a degenerate interpretation of the theory. One ought to obey God’s commands because they are God’s commands and not in order to avoid God’s eternal wrath
in the hereafter and enjoy eternal bliss as a reward. Divine command theory
proper offers no justification of the authority of ethics other than its having
been promulgated and imposed by God, however much some of the world’s
religions may seek to motivate ethicality via promises of eternal reward and
punishment. But of course that is precisely where the consequentialist critique of nonconsequentialism makes its first inroad; for if divine command
theory is not ethical egoism in disguise, then the authority of ethics looks to
be simply arbitrary. What kind of justification is that?
Socrates pretty much demolished divine command theory, or a variant of it
enunciated by the priest Euthyphro, according to whom the right way to behave
60
Chapter Five
is set by “what the gods love.” In his querulous and querying way Socrates
asked Euthyphro, “Is the pious loved by the gods because it is pious, or is it pious because it is loved by the gods?”5 It is clear that Socrates favored the former. In other words, the perfect goodness of God (epitomized for Socrates by
Zeus, “the best and most just of the gods”6) resides in his unfailingly approving
what is good, but it is not the case that whatever he happens to like is by definition good. Therefore adducing the divine does not settle the matter of ethics’
justification, for we would still not know what makes something “pious” or
good or right.
Granted, on this account, a knowledge of God’s preferences—if such were
to be had, for example, by studying certain holy books—might inform us of
what is truly good and right. And that might be the best we mortals could
hope for; so our justification for being ethical would be that we rely on the
say-so of a being whom we trust absolutely. But then of course the question
would become: are we justified to trust God? And this really is a set of questions: (1) Do we know there is such a being? (2) If so, do we know which
words (or other indicators) handed down or spoken to us are his? and (3) Can
we trust our own understanding or interpretation of those words (or whatever)? But even answering all of those in the affirmative would still beg the
question of ethics, which is: what justifies it in the first place such that (and
not because) “God saw that it was good”?
The consequentialist critique continues: If the divine command theorist rejects the consequentialist interpretation that bases ethics’ authority on our
(egoistic) fear of God’s displeasure at our ignoring or disobeying his (arbitrary) dictates, then, on pain of arbitrariness or emptiness, she must have
some other kind of consequence in mind. For there must be some reason why
God loves the good that bears on our good, or else why would we interpret
God as good for having given us the moral law? What falls out quite naturally
(if not inevitably) from that line of thinking is utilitarianism.7 Recall that the
downfall of utilitarianism (according to me) was that it is impossible to know
the relevant actual and hypothetical future(s). But presumably this would not
be a problem for God, who is omniscient. Hence we could rely on God, who
is not only all-knowing but also all-good and so8 has our best interest unfailingly at heart, to be a reliable informant of how we should behave. Divine
utilitarianism is therefore the answer to our ethical prayers: the right thing to
do is that which has the best consequences, but we rely on God to tell us
which acts, and so forth, have the best consequences (using the shorthand
“are right” or “thou shalt”).9 God would on this account be a kind of supercomputer: we know in general what makes something right but only God has
the ability to work out the details. “The details are in God,” you might say.
Nonconsequentialism and the Consequentialist Critique
61
It’s a great idea, or I should say, story. But, again, what reason is there to
believe it? Its perfect interlocking of parts or consistency and its strong appeal to our deepest yearnings and highest aspirations—intellectual, ethical,
aesthetic, salvific—make it a marvelous candidate for mythology. But the
philosophical ethicist wants to know: Where’s the evidence? In any case, the
consequentialist has argued that divine utilitarianism is not divine command
theory understood as an alternative to consequentialism; it only provides for
a division of labor in figuring out what is right or wrong according to a consequentialist scheme. A fortiori, we lack sufficient reason to accept the nonconsequentialist version. But, Q.E.D., its undeniable appeal relies on consequentialism; in the last analysis, no other basis for ethics makes sense, argues
the consequentialist.
THE KANTIAN ALTERNATIVE
I turn now to Kantianism, the moral theory of Immanuel Kant, which, I shall
argue, does make sense as an alternative, nonconsequentialist basis for ethics.
But first yet another item of complexity: there are several versions of Kantianism. Kant himself presented at least three, which he claimed to be equivalent.10 Critics disagree. Perhaps the most widely cited of the three is the formulation of universal law, which I shall refute forthwith. According to it, one
should “Act as if the maxim of your action were to become through your will
a universal law of nature.”11 This is why, for example, one ought never to lie
for personal advantage, since the maxim of such an action would be “Lie
whenever you think it will work out well for you,” but to will that to be a universal law would be nonsensical since nobody would ever be able to believe
anybody and so lying itself would become pointless since it could never
achieve its purpose. But the obvious rejoinder12 is: “Why should I (or anyone)
care whether the universal adoption of my action (i.e., its maxim) would bollocks things up since the very reason I can get away with it is that it isn’t a
law of nature?” In other words, even if it mapped exactly onto the set of moral
actions, the formula of universal law would not seem justificatory; it would
not explain why one ought to refrain from actions that fail a purely hypothetical test. Hence it is not suitable as an ethical theory.
The formula of universal law does not concern me, however, since I do not
see it as equivalent to the formulation that I favor.13 And so without further
ado, here is what I take to be the essence of ethics: We should treat all beings
that are ends-in-themselves with due concern. Call this the formula of endsin-themselves, but it is my variation on Kant’s formula of humanity.14 The latter goes like this: “Act in such a way that you treat humanity, whether in your
62
Chapter Five
own person or in the person of another, always at the same time as an end and
never simply as a means.”15 Let us now consider these two (or two-for-one)
more closely.
Kant’s “formula of humanity” is a misnomer even on its own account since
it is really about rational beings, and humans may or may not be the only rational beings. As Kant makes clear in his discussion of the “kingdom of
ends,”16 he believes that ethics comes about when rational beings freely
choose what standards of behavior they themselves shall be governed by.
Since the only such beings we know about are humans, referring to “humanity” is a practical abbreviation of what or whom he is talking about. But it
should not mislead one into thinking that humans matter in virtue of their being humans. They, that is, we, matter, according to Kant, in virtue of our
power to be ethics-makers, you might say; and only a being who is rational
and free could make ethics, that is, laws whose authority and power to control behavior derive only from one’s own reason and will. This kind of theory
has the distinct advantage over divine command theory of making our assent
to ethics’ authority reasonable and rational ex hypothesi, for it is none other
than we who decide ethics’ content under ideal circumstances (i.e., rationally
and freely). So there is nothing arbitrary or mysterious or imposed from without. The justification of morality is transparent and intrinsically motivating.17
It’s an elegant idea, really, as spelled out by Kant (despite his or his tranlators’ at-times belabored prose), but Kant draws a further implication, which I
do not accept,18 viz., that there is no place for duties to beings who are not rational and free in the requisite sense. Kant did argue that there might still be
reasons to treat them kindly or decently or even respectfully, but such considerations or “duties” would not be owed these others on account of their inherent worth, since, being neither rational nor free, they have none, but are
only conferred upon them by our rational will. For example, other animals
might be exempted from gratuitous torture by humans, but only because it
would ill serve our human purposes and welfare to permit such behavior (because, for instance, torturing other animals makes humans more likely to
harm other humans, as has been shown by some studies19 and as Kant himself believed20).
But such a conception of pseudo-duties to non-human animals is counterintuitive, I maintain. They have a right to be free of gratuitous torture by us
that springs from their nature,21 not ours,22 just as we have the same right for
the same reason, that is, as arising from our own nature and not a potential
torturer’s. Therefore it is just not true that the set of ethical objects or patients,
that is, those who merit direct moral concern and consideration, is necessarily limited to the membership of the set of ethical subjects or agents, that is,
those who are bound by ethics (i.e., to be ethical).23 Thus, the realm of ethi-
Nonconsequentialism and the Consequentialist Critique
63
cal concern includes other animals, not because there may be other beings
who are rational and free, but because there are other beings who are ends-inthemselves.
That latter is of course a technical notion, which comes from Kant, which
is why I would still refer to the theory I am discussing as “Kantian” even
though I have rejected aspects of Kant’s own conception of it. What are endsin-themselves? Because of the use of the terms “end” and “means” in Kant’s
formula of humanity, it is tempting to think of ends-in-themselves in those
terms. We could illustrate the general notion with countless examples. Suppose you owned a precious Chinese vase. It is of course possible that you
could “treat” it as a means, for example, to hold flowers. But you might very
well forswear any such use of the vase and value it simply for itself. So it is
tempting to say that you would then be treating the vase as an end-in-itself.
There is no further end being served, such as holding flowers. And again: suppose you went for a walk. It is perfectly possible that you did so, even though
you disliked walking or any form of exercise, simply to maintain your health;
so you would be treating the activity as a means to the end of being healthy.
However, you might not even be thinking about your health and instead just
love to walk, “for its own sake,” as we say. Therefore you seem to be appreciating walking as an end-in-itself, with no need of further justification.
I do not agree with that interpretation of “ends-in-themselves,” however.
The concept embodied in the examples is a different one, namely, intrinsic
value. A means has instrumental value. Thus, a vase can be used as an instrument or tool or means to hold flowers, and going for a walk can be a way
of preserving health. But a vase and a walk can also be valued for themselves,
as we have seen; hence they would (instead or in addition, as the case may
be) have intrinsic value. In general, all instrumental value derives from intrinsic value, for there would be no means without ends. Now an end, as we
have seen, can also be a means, but then it would be for some further end.
Eventually the referral of means to ends must come to an end, however, or
else nothing could have any kind of value at all, not even merely instrumental value.24 So, again, it is tempting to refer to the kind of end that is intrinsic
as an “end-in-itself,” for the buck stops there.
But that is not what Kant means by an end-in-itself. For consider that the
ends in our examples—the preciousness of the vase and the enjoyment of the
walk—are relative to a human being or human beings in general. Even though
we refer to their value as “intrinsic,” we do so only to preserve the distinction
from their being used for some further purpose. But their value is not intrinsic
in the sense of being self-sufficient. The intrinsic (not to mention instrumental) value of a vase or a walk in the woods would evaporate instantly if there
were no human being(s) to appreciate it. Thus, it is we who bring value into
64
Chapter Five
the universe. It is not that we have value, although that is true too (both kinds);
but for ethical purposes what matters most is that we are value-makers. It is we
who have purposes, create meanings, and so forth.
And who are “we”? Not only humans, surely. We are, at the very least, animals: all animals are value-makers. A cat can appreciate, for example, the
meaning of being in pain. Therefore a cat is a being that is an end-in-itself as
much as you are.25 Again, this is not because pain may have negative value
“intrinsically,” as a hedonist or utilitarian would have it, but because it is the
cat to whom that value has meaning—the cat who brings that very value into
existence by its own being. A cat may be valued as useful because of her ratcatching ability; and she may be valued intrinsically by you for her loveableness. But the cat is also an end-in-itself because things are valued by her.26
This is an order of magnitude beyond simply having value, even intrinsic
value.27
That, fundamentally, is the justification of ethics, I submit. That is why beings who are capable of recognizing and responding to an ethical imperative—beings who are rational and free as we are—have obligations to all beings who are ends-in-themselves, which includes us but not only us because
a being does not need to be rational and free, or what I called an ethics-maker,
in order to be a value-maker. You could say: The reason human beings have
ethical responsibilities is that we are human (i.e., rational and free, ethicsmakers), but the reason human beings have ethical responsibilities to themselves is that we are animals (value-makers, ends-in-themselves).
And what obligations do we owe ourselves and others? To treat all “with
due concern.” I use the word “concern” to encompass the typically Kantian
word respect and also the characteristically feminist word caring.28 Thus we
would say, cats are not mere objects, like boxes, but are “objects” or beings
that merit moral concern. We could also say: Concern is the attitude it is appropriate to harbor towards a being who has concerns, albeit not necessarily
of the ethical kind. One can be ethically concerned about not stepping on the
cat, who is (non-ethically) concerned about not getting stepped on. Both are
a kind of valuing; but the one is an ethical obligation, while the other is a fact
of the cat’s psychology and being. The former depends on my will as a rational and free being; the latter need not be willed, although even that kind of
concern can be, as when a human being in the existentialist mode freely values something he must do, for example, pushing a rock up a hill in the triumphant manner of Camus’ Sisyphus.29 In either case I would be called upon
to respect and care about these beings who have their respective concerns.
Thus, I would (i.e., ought to) be careful where I step in the vicinity of the cat,
or lend a helping hand to Sisyphus, or not if Sisyphus valued pushing the rock
by himself. These beings should matter to me because they do matter “in
Nonconsequentialism and the Consequentialist Critique
65
themselves.” They matter in themselves because things matter to them. Ergo,
these beings should matter to me because things do matter to them.
Do I therefore have an obligation to help the fiendish assassin locate her innocent victim?30 Shall I help her hold his head steady as she struggles to rip
open his neck with her knife? I think not! But why not? For two interrelated reasons. One is that I ought to be concerned about the victim as well. The other is
that ethical concern is always “due” concern. Ah, there must be a great deal
packed into that little word! Note that we owe concern to beings first and foremost and not to their (or our) concerns. Thus, I can be genuinely and fully ethically concerned about a being who has various concerns without respecting or
caring about all of her concerns. Think of the difference between treating a prisoner properly and treating a prisoner in the way some Americans did at Abu
Ghraib.31 In neither case is the prisoner’s concern to be released being “respected.” But in the former case the prisoner is being respected. In the assassin
example, I would argue that the assassin’s failure to respect her victim, by so
grossly ignoring his concerns, warrants my putting the victim’s concerns above
the (potential) assassin’s. The point is that I do not simply discount the assassin
or even her concerns but show my respect for her by weighing hers in the balance, even though I may ultimately reject them and even try to frustrate them.
But what is this “balance”? asks the consequentialist. Must it not be of relative net outcomes? I will consider the consequentialist critique of Kantianism in particular in the next chapter, but for now let me advise that all I could
hope to mean by due concern, that is, appropriate respect and caring, is something non-quantitative. It is always meet in ethics to recall Aristotle’s observation that “Our discussion will be adequate if it has as much clearness as the
subject-matter admits of, for precision is not to be sought for alike in all discussions . . . .”32 Thus, as far as I am concerned, explaining exactly how one
is to implement the Kantian Supreme Injunction will not require an algorithm
or the assurance that it will be easy to arrive at particular ethical conclusions,
not to mention always arrive at consensus even among committed Kantians.
My aim in this book is to argue the general point that an attitude of respect
and caring trumps attempting to ascertain consequences and calculate their
relative net values. It is not my aim to pin down precise definitions of “respect” and “caring,” a fortiori to explain how we might adjudicate their occasional conflicts, either in themselves, as with the assassin example, where
concern for one person conflicted with concern for another, or between them,
for example, by assigning priority either to respect or to caring under suchand-such circumstances, or, alternatively, trying to demonstrate that one of
these attitudes is basic and the other one derived from it.33 Instead, examples
should suffice to convey the idea; and the rest is then a matter of life practice
for each of us (with aid from the collected wisdom of the ages and sages).
66
Chapter Five
Should the consequentialist cry foul because this seems to present a double
standard of rigor demanded of our opposing theories, I reply (once again)34:
it is simply not humanly possible to acquire or even approximate the kind of
knowledge that would be required to implement a genuine consequentialism
since the needed knowledge would be, for all practical purposes, infinite. The
kind of judgment required for Kantianism, on the other hand, calls “only” for
committed study, reflection (especially of the peripatetic and dialogic sorts,
i.e., walking and talking), experience, and meditation—in a word, philosophical living.35
NOTES
1. The consequentialist would say that we owe her an explanation for such a powerful phenomenon if we consider it to be illusory. Evolutionary theory would suggest
that it is probably “there” for a reason having to do with our survival, which is a kind
of consequence. I will take up this subject again in the next chapter.
2. In its most general form karma yoga has very close parallels to Kantianism in
their shared emphasis on acting in accordance with duty without respect to consequences. Cf. the Bhagavad-Gita:
. . . if one performs prescribed action
because it must be done,
relinquishing attachment and the fruit,
his relinquishment is a lucid act. (18:1)
3. I would like to propose what might be call the Animal Golden Rule (for reasons
which will be further explicated later in this chapter), where the word “them” is
changed to “others,” thus: “do unto others as you would have others do unto you.” In
other words, the others unto whom you are “doing” need not be the same others who
might “do” to you . . . not even the same kind of others. The former is probably already an implication of the Golden Rule, but the latter would be innovative. So for
example one would be enjoined to refrain from confining a rat to a cage in an experimental laboratory because one would not like to be confined in a similar way oneself
by laboratory scientists, even though one could not even imagine the rat reciprocating in any way whatever.
4. Ivan Karamazov gives famous expression to this last: “And if the sufferings of
children go to swell the sum of sufferings which was necessary to pay for truth, then
I protest that the truth is not worth such a price. I don’t want the mother to embrace
the oppressor who threw her son to the dogs! She dare not forgive him! Let her forgive him for herself, if she will, let her forgive the torturer for the immeasurable suffering of her mother’s heart. But the sufferings of her tortured child she has no right
to forgive; she dare not forgive the torturer, even if the child were to forgive him! And
if that is so, if they dare not forgive, what becomes of harmony? Is there in the whole
Nonconsequentialism and the Consequentialist Critique
67
world a being who would have the right to forgive and could forgive? I don’t want
harmony. From love for humanity I don’t want it. I would rather be left with the unavenged suffering. I would rather remain with my unavenged suffering and unsatisfied indignation, even if I were wrong. Besides, too high a price is asked for harmony;
it’s beyond our means to pay so much to enter on it. And so I hasten to give back my
entrance ticket, and if I am an honest man I am bound to give it back as soon as possible. And that I am doing. It’s not God that I don’t accept, Alyosha, only I most respectfully return him the ticket” (Fyodor Dostoevsky, The Brothers Karamazov, trans.
Constance Garnett, pt. 2, bk. 5, ch. 4.: “Rebellion”).
5. Plato’s Euthyphro, 10a.
6. Ibid., 5e.
7. Or at least ethical humanism, which is what utilitarianism was typically taken
to be before Peter Singer reminded us of its roots . . . and ours as sentient beings. Cf.
Singer (2004).
8. Again, from the humanist standpoint at any rate . . . or even the tellurian standpoint, which is still parochial according to the argument of Chapter 3.
9. Cf. this remark from The Nine Tailors (Harcourt Brace, 1934), one of Dorothy
L. Sayers’ Lord Peter Wimsey mystery novels:
“My dear boy,” said the Rector, “it does not do for us to take too much thought for the
morrow. It is better to follow the truth and leave the result in the hand of God. He can foresee where we cannot, because He knows all the facts.” (p. 307)
10. Kant (1785), p. 436.
11. Kant (1785), p. 421.
12. That is, even putting aside the direct criticism that some perfectly moral actions would fail the test and some obviously immoral actions would pass it.
13. I suggest an alternative possible use of the formula in a note to Appendix I.
14. Franklin (2005) makes a very similar move, proposing the following variant on
the formula of humanity: “Act in such a way that you always treat sentience, whether
in yourself or in the self of any other, never simply as a means but also at the same
time as an end” (p. xiii, my emphasis; and cf. p. 42).
15. Kant (1785), p. 429.
16. Kant (1785), pp. 433-39.
17. The consequentialist would no doubt argue that the only thing that could make
ethics appealing and motivating under such a regime would be the perceived (personal or mutual) advantage of accepting it. But that claim seems merely tendentious
and, as argued in Chapter 1, empirically dubitable.
18. That is, not even as an implication; and neither does Korsgaard (2005) accept it.
19. See “Animal Abuse and Human Abuse: Partners in Crime” in the bibliography.
20. See Kant (1993): “[H]e who is cruel to animals becomes hard also in his dealings with men” (p. 240 of this edition).
21. If the reader does not share this intuition, I can offer no additional argument—
the reasons must come to an end. However, I would recommend research (about other
animals), experience (with other animals), and the cultivation of moral imagination
68
Chapter Five
and empathy through meditation (e.g., on the Animal Golden Rule). Cf. Korsgaard
(forthcoming): “Given how pressing the moral questions are, I would encourage
everyone to read about non-human animals, to find out what they are actually like. A
growing scientific literature suggests that many of us have underestimated the intelligence and emotional complexity of the other animals.”
22. Cf. Feinberg’s (1974) dismissal of another kind of indirect concern for animals:
Professor Louis B. Schwartz finds the rationale of the cruelty-to-animals prohibition in its
protection of animal lovers from affronts to their sensibilities. “It is not the mistreated dog
who is the ultimate object of concern, . . .” he writes. “Our concern is for the feelings of
other human beings, a large proportion of whom, although accustomed to the slaughter of
animals for food, readily identify themselves with a tortured dog or horse and respond
with great sensitivity to its sufferings” (“Morals, Offenses and the Model Penal Code,”
Columbia law Review, Vol. 63 (1963), p. 673). This seems to me to be factitious. How
much more natural it is to say with John Chipman Gray that the true purpose of crueltyto-animals statutes is “to preserve the dumb brutes from suffering” (The Nature and
Sources of the Law, 2nd ed. [Boston-Beacon Press, 1963], p. 43).
And then there is Mencius (1970) with his odd advice (?) to King Hsüan of Ch’i:
“The attitude of a gentleman towards animals is this: once having seen them alive, he
cannot bear to see them die, and once having heard them cry, he cannot bear to eat
their flesh. That is why the gentleman keeps his distance from the kitchen” (IA7).
23. Cf. Martha C. Nussbaum’s (2006) assertion about contract theories (which
hold that morality or justice reflects an implicit contract among equals): “. . . the fact
that such theories conflate the question ‘Who frames the principles of justice?’ with
the question ‘For whom are these principles framed?’ means that they cannot include
animals in the group of subjects for whom the theory is devised, given that animals
do not partake in the making of contracts” (p. 21). Thus, contract theories imply an
identity between moral agents and moral patients since only the contractors, the
“framers” of justice or morality, can be either; but Nussbaum and I (and others cited)
would agree that morality can (and should) certainly accommodate moral patients
who are not moral agents.
24. Hence de Sousa (1986): “Some things can be both valuable and useful, but
only what is useless can be wholly and purely valuable. This, rather than the usual
disingenuous mumblings about careers in law, is the proper retort to students who
question the usefulness of philosophy” (p. 100).
25. Christine Korsgaard (2005) has provided an alternative account of the Kantian
“end-in-itself” that also accommodates nonhuman animals, but in a different way. For
Korsgaard it is not enough that things matter to a being, but that the being matter to
itself; furthermore, it must also be deemed considerable by a being that is capable of
doing such a thing, which on this planet may be limited to human beings, who have
the requisite rationality and freedom for such a cognitive stepping-back and acting on
it. (The second premise of her argument is that we humans consider ourselves considerable in the very act of deeming our desires considerable, and Premise Three is
that a subset of those desires constitutes our “animal nature”; ergo, animal nature is
worthy of moral respect.) While her analysis is truer to Kant’s ingenious moral meta-
Nonconsequentialism and the Consequentialist Critique
69
physics, albeit drawing a conclusion he himself did not, I stand by mine as truer to the
value of (all) animals. An acknowledged implication of Korsgaard’s analysis is that
animals (including ourselves) are valuable because we (i.e., the type of animal we are)
value them. I reject any such dependency of moral considerableness on another kind
of creature’s valuation.
26. I owe Miller Brown a fuller analysis of valuing once I decide whether I believe
nonsentient entities, e.g. plants, also value things. I note that Taylor (1986) develops
a speculation about their having a point of view:
. . . organisms like trees and one-celled protozoa do not have a conscious life. They are not
aware of a world around them. They have no thoughts or feelings and hence no interest in
anything that happens to them. Yet they have a good of their own around which their behavior is organized. . . . Under this conception of individual living things, each is seen to
have a single, unique point of view. (p. 122)
27. This distinction is routinely overlooked even by thinkers who place great emphasis on the notion of ends-in-themselves. For example, Callicott (1989) writes:
Something is intrinsically valuable if it is valuable in and for itself—if its value is not derived from its utility, but is independent of any use or function it may have in relation to
something or someone else. In classical philosophical terminology, an intrinsically valuable entity is said to be an “end-in-itself,” not just a “means” to another’s ends. (p. 131)
I think Callicott here conflates being valuable-for-itself (i.e., having intrinsic value)
with being valuable-in-itself (like a true end-in-itself). Francione (2007) makes the
same mistake, as in the following passage:
When we say that a person has inherent or intrinsic value, what we mean is that she has
value that is not solely extrinsic or conditional. We recognize that she has value because
she values herself, even if no one else values her” (p. 37).
The first sentence of that passage is talking about value that is non-instrumental,
again, something’s being valued “for its own sake.” The second sentence is talking
about the “value” or worth of a valuer.
Even Tom Regan, the most prominent animal advocate of this school of thought, is
not as clear as could be desired about the distinction, and even though he spends a
chapter discussing it in his classic book (1983, ch. 7). Regan does acknowledge a distinction between what he calls intrinsic value and inherent value, terms which Francione uses interchangeably (as above). The latter is not merely non-instrumental but
belongs to “subjects-of-a-life,” who have “welfare” interests “logically independently
of their [the subjects’] being the object of anyone else’s interests” (p. 243)—what I
have referred to as things “mattering” to an individual. Still Regan does not explain
that inherent value belongs precisely to the sort of being, whom I have called a “valuer,” that is capable of valuing something intrinsically.
28. Kant (1797) himself holds that we have the double duties of respect and “love”
(see especially his discussion of duties of virtue to others that begins at 6:448).
29. See “The Myth of Sisyphus” in Camus (1955).
70
Chapter Five
30. I have in mind of course the notorious example in Kant (1799).
31. See e.g. Hersh (2004).
32. Nicomachean Ethics (bk. 1, ch. 3).
33. For example, Franklin (2005) argues that “while the moral imperative is autonomous, the ethic of care is not. The former must, accordingly, be primary” (p. 84).
Hall (2008) recognizes the opposite possibility: “wouldn’t a truly caring ethic include
consciously transcending dominance and fostering respect?” Kant (1797), meanwhile, when introducing a detailed discussion of the distinction between respect and
“love,” says of their relation, “But they are basically always united by the law into
one duty, only in such a way that now one duty and now the other is the subject’s principle, with the other joined to it as accessory” (6:448).
34. As in Chapter 4.
35. This is my reply to Hudson’s (1989) claim that “. . . the Categorical Imperative
is defective in an irremediable way . . . [namely, that it does] not allow hedging in the
face of uncertainty” (p. 228). The need for hedging is far less critical for Kantianism
than for utilitarianism precisely because the former’s epistemic status in far less dire.
Chapter Six
The Ethics of Ethics
If utilitarianism is the most plausible form of consequentialism but is impossible, and if Kantianism is the most plausible form of nonconsequentialism
but succumbs to the consequentialist critique that it is itself really a form of
consequentialism, indeed, utilitarianism, then ethics is futile.1 I have proposed that we are under the ultimate obligation to treat all beings that are
ends-in-themselves, including of course oneself, with due concern. The utilitarian has rejoined that “with due concern” could only mean with regard to
relative net outcomes; hence my proposal, which I consider to be the most
plausible form of Kantianism, would really be utilitarianism in disguise. Let
us now consider the utilitarian’s objection in detail.
THE CONSEQUENTIALIST CRITIQUE OF KANTIANISM
The utilitarian claims that if something is right (or else permissible) by Kantian lights, then one of two things will be true: either there is a utilitarian
analysis that endorses it, or else it is in fact wrong; and mutatis mutandis for
something deemed wrong by Kantianism. More particularly, the utilitarian argues that Kantianism amounts to a set of rules, which are themselves endorsed by the optimality of their outcomes (that is, the consequences of following them or else of accepting them); however, they are just that—rules, as
in “rules of thumb.” So, unlike laws, they have exceptions; and the only way
to account for those exceptions is by appealing to consequences, since Kantianism will have exhausted its resources by offering its rules. Thus, Kantianism errs in taking the rules as fundamental and hence exceptionless. The
rules are indeed justified, but by their consequences, not by themselves;
71
72
Chapter Six
hence when consequences advise breaking them they should be put aside. So
consequentialism has two advantages in this debate because (1) it can account
for the apparently nonconsequentialist rules of the Kantian while at the same
time (2) it avoids the absurd rigidity of Kantianism in the face of catastrophe
(as in the infamous Kantian cri de coeur: fiat justitia ruat caelum or “Let justice be done, though the heavens fall”).2
For example: A current controversy in the United States is whether torture
is ever justified. This debate has been brought on by the advent of modern terrorism,3 which has the potential to do damage on scales hitherto unimaginable. One would naturally expect a consequentialist to approve torture whenever doing so seemed likely to have better results overall than failing to, for
example, by averting some imminent terrorist act; while the Kantian would
prohibit it because it would violate the categorical imperative never to treat
anybody merely as a means, even if as a means to averting great harm.
But the consequentialist could say: Not so fast! The given analysis is shallow: one must consider also the “deep” or long-term consequences of employing torture. For example, if our country used it, then wouldn’t other countries be that much more likely to use it, perhaps at times even against our own
citizens? Furthermore, if we used it, wouldn’t we require a torture “infrastructure” that actually teaches such methods, etc.? And might this not
weaken the strictures we have in place, including the emotional ones, not to
use torture even on our own citizens? In other words, it could lead us down a
slippery slope, opening also the prospect of an ever-increasing number of innocent people becoming subjected to it. Even more generally, it could begin
to unravel the whole structure of our moral aversion to treating people in various ways that are now deemed immoral and simply “not done.” So in the end
this could have even more disastrous consequences for us or for the whole
world than failing to avert the occasional terrorist act.4 Therefore the supposed Kantian intuition to refrain from the use of torture even when there
promises to be a clear advantage to using it can itself be seen to have consequentialist backing. (Analogous arguments have been made regarding capital
punishment,5 the use of nuclear weapons, pre-emptive war, and so forth.)
However, it remains problematic whether any such consequentially backed
prohibition would be absolute. So it is still conceivable that the consequentialist would oppose the United Nations Convention against Torture, which
states: “No exceptional circumstances whatsoever, whether a state of war or
a threat of war, internal political instability or any other public emergency,
may be invoked as a justification of torture” (Article 2). The Kantian, however, would endorse it, that is, as absolute, since the Convention rests on “the
inherent dignity of the human person” (preamble), which is sacrosanct precisely because it is the Kantian bottom line (just as consequences are the con-
The Ethics of Ethics
73
sequentialist’s bottom line). To the consequentialist this could appear insane
if the perceived threat were dire enough.
Thus, the problem with Kantianism is that it has no wiggle room even to
consider that a particular situation might warrant an exception to one of its
rules. Again, note, the consequentialist also has the option of supporting an
absolute prohibition on torture; but for the consequentialist it is an option,
whereas for the Kantian it is mandated. From the consequentialist’s point of
view, it just seems a simple category mistake to graduate the result of a deliberation to the status of an axiom. It is analogous to the biblical literalist’s
considering species to be fundamental existents and essences as opposed to
evolved via survival under ever-changing circumstances. Just so: our ethics
must be sensitive to changing (and different) circumstances. Insofar as some
circumstances may be constant throughout our history (and across cultures),
certain ethical precepts, such as “Thou shalt not torture,” may prove enduring
(and universal). But to assume that they must endure no matter what is a
recipe for disaster (not to mention, missed opportunities) should those hitherto constant circumstances ever change in hitherto unimagined but relevant
ways, for example, with the prospect of nuclear terrorism and the advent of
biotechnology, the Internet, etc. ad inf.
REPLY TO THE CONSEQUENTIALIST CRITIQUE
The consequentialist critique of Kantianism is a powerful argument indeed.
But it has two Achilles’ heels. The first, as considered at length in Chapter 4,
is that it simply cannot work. Utilitarianism presumes a kind of knowledge of
outcomes that is unavailable to us. So as wonderful as might be the ability to
base out decisions on consequentially justified rules and finely tuned exceptions to them, in reality this is only an appealing myth (in appealing in its own
way as a perfectly just universe is to the religious-minded), especially if the
consequentialism in question is utilitarianism. The second problem is that it
caricatures Kantianism. It is simply not true that Kantianism amounts to a set
of absolute rules, although it does have that reputation. But I think it is clear
from a close reading of the Grundlegung (Kant 1785) that Kant intended for
there to be only one categorical imperative.
I am not here denying that there might be alternative “formulations” of the
categorical imperative, of which Kant notoriously claimed there are at least
three, as we noted. My point is that Kant intended the categorical imperative
to be a meta-ethical notion, such that it applies to all ethical theories (i.e., candidate ethical theories) and not only his.6 It is the requirement that any genuine ethics or ethical theory consist in a supreme injunction or principle of
74
Chapter Six
morality.7 This is the case even for utilitarianism,8 where there is a single
supreme injunction, namely, Maximize utility. This is categorical because it is
derived from nothing else; it is the ethical bottom line. All other
utilitarian/ethical “commandments” are derived from it, and hence are hypothetical in that sense; for example, “If lying (and only lying) promises to maximize utility in such-and-such situation, then lie.”
Thus, Kant’s categorical imperative—Never treat anyone merely as a
means—is simply the meta-ethical notion applied to his ideas.9 I have argued
for a modified Kantian categorical imperative: Treat all beings that are endsin-themselves with due concern. But this would be the only categorical imperative of Kantianism.10 All other Kantian imperatives would be derived
from it and hence hypothetical in that sense.11 Thus, for example, I would argue that there might even be Kantian torture, provided “only” that one has ascertained that due concern (that is, for all concerned, including of course the
being who is being tortured) dictated or permitted torture on some particular
occasion or even in some type of situation. Although, just as I argued with respect to consequentialism, there might be something about torture, and the
constants of the human or animal condition, that militate against permitting it
on any occasion. It all depends—that supposedly consequentialist refrain!
We appear, then, to have reached a stalemate. The utilitarian sought to defeat Kantianism with the consequentialist critique that any ethical theory with
intuitive appeal must in the end depend on a consideration of consequences
for that appeal; and insofar as a given (supposedly) nonconsequentialist theory diverged from that underpinning, its recommendations would simply be
incorrect. I, although a Kantian, have seemingly strengthened that objection
by arguing that Kantianism, correctly understood, has the capacity to match
utilitarian ethical recommendations in every particular. However, I do not see
this as a vindication of consequentialism; rather, it is in the first instance just
what we should expect of all ethical theories “in the running,” namely, that
their recommendations (or mandates) will conform to our pre-theoretic ethical intuitions. After all, it is by the latter that we “test” theories in our search
for the correct one; ethical intuitions are our “data.” A given theory need not
support every such intuition, but it would not have been taken seriously even
as a candidate had it not appeared to support the most basic ones.
Let me try to make this point decisively by a close analysis of a classic
thought experiment, which supposedly demonstrates a clear difference between
the recommendations of the two theories (utilitarianism and Kantianism), to wit:
A family is hiding in the attic from the Nazis. The baby is restless and likely to
cry. If he does, the entire family is doomed, including several children. Should
the mother stifle the baby’s cry, thereby likely suffocating him?
The Ethics of Ethics
75
The utilitarian does a simple calculation: Stifle the baby—one person dies.
Don’t stifle the baby—everybody dies. Ergo: stifle the baby. The Kantian reasons that stifling the baby would be treating the baby “merely as a means” and
not at the same time as an end-in-itself. Ergo: don’t stifle the baby. Q.E.D.:
the theories offer quite distinct prescriptions.
Now, first I want to emphasize that the thought experiment illustrates an
extreme situation. Therefore even if we were to accept the analysis, it would
not necessarily show a significant difference between the recommendations
of the two theories in general. In other words: ethics is not like physics, where
one theory can be overthrown by another because of a difference that only
shows up under extreme circumstances; for example, Einstein has overthrown Newton, even though everyday life remains Newtonian, because Einstein’s predictions are superior at near-light speeds and minuscule measurements. Perhaps an even more apt analogy is the complete breakdown or
silence even of Einsteinian theory inside a black hole. Certain thought experiments could be moral black holes where no ethical theory, even the correct
one, is even applicable. In such an unfortunate situation it is not that anything
is permissible but that the moral reasoner is left to her own devices, without
“guidance,” as it were. When forced to “compute” when the resources are
lacking, a human being could become unhinged, like Sophie as a result of
having to make her terrible “choice.”12
But I do not accept the analysis of the case at hand since I believe it shortchanges the actual process that both theories would employ to justify their
conclusions.13 First as regards utilitarianism: it can be argued that an essential part of human morality is cultivating the right habits.14 This is not only so
that we will more reliably do the right things when there is insufficient time
to reflect, but also so that we will have the strength of motivation to counteract strong countervailing temptations. Mother and child would seem to be the
archetypical case in point: In order to motivate a human being to make the
sort of continual and difficult sacrifices of motherhood (beginning with nine
months of carrying a heavy burden and then undergoing an extremely painful
delivery), for the betterment of babe and society as a whole something pretty
darn strong has to be “wired in.” Thus, utilitarianism could not only be about
assessing the morality of individual acts but must also rate motivational inclinations or habits of feeling, thinking, and so forth.15 The upshot for the
thought experiment is that it would be wrong for the mother even to have the
motivational capacity16 to intentionally smother her own child, no matter
what the circumstances.17
Meanwhile as regards Kantianism: genuinely Kantian motivation comes
from the will, not habit.18 Furthermore, a genuinely Kantian analysis and intention would have to consider all affected parties. Thus, the mother could not
76
Chapter Six
simply ignore her other children nor the other people present in the attic. Due
concern for her other children and other companions (and of herself) as endsin-themselves would certainly require her to consider their interests in surviving as well as her baby’s. This could translate into her taking steps to silence the baby. That does not mean she would be intending to kill her baby
since silencing, even stifling, does not have killing as its intent, not to mention death as its guaranteed outcome.19 But even with the high risk of death,
the mother might be implicitly considering the baby’s interest in having as
“comfortable” a death as possible under the circumstances, given that he was
equally likely to die at the “less considerate” hands of the Nazis. Finally, due
concern for the Nazis’ interests would be not to harm them gratuitously or
maliciously, but surely her trying her best to protect everyone in the attic is
doing no such thing.
Thus, there are good Kantian and utilitarian reasons to draw whichever
conclusion you please about this example. So where does this leave our main
debate? In a way, we have shifted the locus of concern from ethics to metaethics. For I do not think the normative upshot has been affected at all: Kantianism still has the advantage simply because utilitarianism is impossible! In
other words, the utilitarian’s employment of the consequentialist critique has
backfired; for by demonstrating the practical equivalence and ethical adequacy, that is, conformity to the “ethical data,” of the two theories, it has
handed the theoretical laurels to the only one of the two that has not been defeated (and is otherwise plausible) on other grounds.
Put it this way: the experimentum crucis (i.e., the thought experiment),
even at its best, has pitted Kantianism proper against faux utilitarianism. Faux
utilitarianism is the theory that the right thing to do is that which has the best
foreseeable consequences. It is not genuinely consequentialist for the reason
given in Chapter 4, namely, that foreseeable consequences bear no knowable
relation to the actual, that is, relevant, that is, long-term relative net, consequences. Who has any idea whether saving the family in the attic will turn out
to be best for the universe in the long term? Then why does faux utilitarianism have the capacity to match our deepest ethical intuitions? Precisely because of the elusiveness of knowledge of ultimate outcomes, for this makes it
possible to find some kind of shorter-term/narrower-scope consequentialist
“justification” for any action (etc.) whatever.20 In other words, the great
power (and hence great appeal) of consequentialist reasoning is illusory.21 It
is an illusion because demonstrating the utility of something, which it is always possible to do (e.g., pulling the trigger is useful for making the gun fire),
does not ipso facto demonstrate that it is more useful (a fortiori in promoting
the greatest good) than any alternative option.
The Ethics of Ethics
77
KANTIAN CONSEQUENTIALISM
There is one other way that the consequentialist could challenge the nonconsequentialist credentials of Kantianism. If the Kantian really believed that respect and caring are of ultimate concern and hence the ultimate values, so this
argument goes, then would it not be self-contradictory for her to refuse to promote them on the basis of Kantian scruples?22 For example, if Saddam were
about to deploy weapons of moral destruction, which realistically threatened
to wipe out Kantian respect among the entire world’s population irresistibly
and irremediably, would it not be justified, on Kantian grounds, to eliminate
him by any means necessary . . . even though some means could well be manifestly non-Kantian because they would be treating Saddam merely as a
means to the preservation of Kantian regard in the world?
This prospect makes it seem that the consequentialism debate has at last
reached a genuine stalemate, a kind of “He says She says,” since of course
from the Kantian point of view it is this consequentialist proposal that would
be the real self-contradiction of Kantian values, precisely because it condones
(and could even command) treating somebody merely as a means.23 The way
to break the tie, however, is once again to point out the pointlessness of any
utilitarian appeal (and hence, by the argument of the first half of this book,
any consequentialist one). So what we are left with is the Kantian triumphing
in this way: “If it were possible, with conscious intention or any knowably reliable indirection, to maximize Kantian respect in the world, then perhaps our
duty would be to do so, or at least we might sit down and discuss whether it
is. But since it is not possible, i.e., humanly practicable, to so maximize any
good thing, or indeed even to forestall disaster, all that a Kantian could strive
to do is to be true to her values at all times.”24
CONSEQUENTIALIST KANTIANISM
There is one final echo of the consequentialist critique, which no stopping of
the ears can completely attenuate. It is that anybody in his or her right (and
moral) mind will pay attention to the (probably) foreseeable consequences of
their actions (or of whoever’s actions, etc., they are judging). But how can
Kantianism account for this, that is, justify this and endorse this? In reply I
will employ a strategy that was much in evidence in my earlier discussion of
consequentialism, namely, using one form of ethics to justify a different form
of motivation, for example, Adam Smith’s recommendation of egoistic motivation to assure a utilitarian outcome. In fact, as we have seen in this chapter,
78
Chapter Six
the consequentialist critique itself purports to be able to justify a kind of
Kantian motivation in terms of its utility. Let me just turn the tables, therefore, and suggest that Kantianism could justify (faux-) consequentialist motivation or intention, at least on some occasions. But how could it do this? For
Kantianism has one particular disadvantage with regard to this strategy,
namely, that it is itself normally conceived as a theory about motivation, or
will in Kantian terminology. Thus, for a Kantian to have a non-Kantian motivation seems to be ruled out ex hypothesi.
I think the answer must be that the full parsing of the Kantian’s motivation
when he or she does consider consequences would still give pride of place to
respect and/or caring rather than outcomes.25 Here is an example. I haven’t
the slightest idea whether my having started my stepson Sean on piano lessons will result in his having a happier life overall, not to mention the happiest life possible for him, and not to mention the best state of affairs for all sentient beings forever. I could see arguments on both (or all) sides: from “It was
just the discipline he needed to channel his prodigious talent” to “It stunted
his natural creativity and alienated him from the musical and social life of his
peer group.” But I love him and still think this suited his abilities and
prospects admirably; so my having the right attitude justified my being motivated to act with regard to the (merely) foreseeable consequences. But I am
not fooling myself that what I did will guarantee the best outcome in the long
run. It’s just that I’m hog-tied to figure out how else I could have shown my
concern than to have acted by my best lights, dim as they were. But the bottom line is the concern, not the consequences.
This is not game-playing, nor despair, nor the Noble Lie26; it is doing the
best one can. I admit, however, that there still seems to be something strange
about it. As Mitchell Silver has expressed it: “why should a temporary good
result believed to be uncorrelated with a final good result be a show of respect
[or caring]?” He suggests that the solution must be that “the former correlates
with something else that is a show of respect” and offers the following way
to think about it:
Shows of respect often are unconnected to consequences but are rather culturally/historically endorsed, for example, bowing in Asian cultures. Now it is universally true that, because people believe you can act to benefit them (and understandably so, since in the short run you usually can), trying to promote
someone’s well being is taken to be a sign of respect [and/or caring]. Hence, because of this contingent, perhaps false, but universal belief, our actions in any
human culture, in order to be in accord with the morality incumbent on all rational beings to show respect [and caring], must implement that imperative by
the universal human “bow,” viz., attempting to benefit others. Neither the bow
nor the attempt to benefit may actually [specifically, in the long term] benefit or
The Ethics of Ethics
79
even tend to, but if it is expected as a show of respect, it is!
Thus for example: I know that your paying for dinner Tuesday evening may not
be conducive to my long term good, and I know that you know that too. Nonetheless I can’t shake the feeling that it was helpful to me (it certainly saved me money,
which is a short-term benefit) and was intended as such by you, so I took it as a
sign that you wanted to benefit me, and that made me think it showed an appreciation of me. [Similarly] I say Gesundheit to sneezing people even if . . . I don’t believe my saying so will ward off disease.27
Thus, what I am proposing would be the Gesundheit Theory of Morality, or
more properly, faux-utilitarian Kantianism, since I am defending the Kantian
propriety of doing things one does believe will be helpful, albeit not in the
full-blown consequentialist (i.e., omniscient) way. I would only add the
caveat that the “show” of respect need not be recognized as such or even
known to the recipient of that respect; the duty to express respect and caring
with some kind of faux-consequentialist behavior would be ours in any case,
if only because it may be the only way to satisfy the agent’s own desire to
help.28 It is understandable that such a desire would have been deeply implanted within us by evolution.29
But the ultimate answer to the consequentialist critique is that it represents
a misreading of Kantianism. For nowhere does Kant proscribe thinking and
acting in terms of (foreseeable) consequences . . . no more than he rules out
forswearing the use of persons (or, on my view, sentient beings). Recall that
Kant’s categorical imperative prohibits treating anybody merely as a means;
but to avoid treating anybody as a means altogether would be impossible and
absurd. Indeed, it is a good thing when my students use me to learn. But they
are merely using me, that is, abusing me (and their classmates and even themselves), when they cheat; so that is wrong. Analogously, then, we could say
that Kant prohibits only acting merely or solely out of regard for consequences. Whenever considering consequences is consonant with treating
everyone with due concern, it would be perfectly kosher, and possibly under
certain circumstances even obligatory.
JUSTIFYING THE SEARCH FOR JUSTIFICATION
So Kantianism has won this battle; but has it thereby lost the war? For in the
process of showing the superiority of Kantianism to utilitarianism, have we not
demonstrated the uselessness of ethics itself? Consider the chameleonic nature
that both theories have displayed, for example, in analyzing the thought experiment; does this not make even the “winner” liable to sophistic, casuistic
manipulation in support of any good or evil purpose? Furthermore, why not
80
Chapter Six
just stick with our original “pre-theoretic” intuitions since they are what we
judge the theories by, and not vice versa? Indeed, if we can judge good and
evil, right and wrong, without the theories, have we not the makings of a refutation of the prioritizing of ethical theory, analogous to Socrates’ refutation of
a theistic basis of ethics in the Euthyphro?30 To put theory “first” in the order
of importance (recall the Preface) seems like the tail wagging the dog. Finally,
if both the “loser” and the “winner” will always (at least be able to) generate
the same prescriptions, then who cares which is the true theory? So, as I say,
the issue has shifted from ethics to meta-ethics since we have now done with
the justification of ethics in the sense of ethicality—one ought to do what
ethics prescribes because all beings merit concern—and graduated (or deteriorated?) to the justification of ethics in the sense of ethicizing, that is, why
does ethical theorizing matter?31
Before I reply at length, let me put one pseudo-reply to rest. It is that ethicizing doesn’t need to be provided with any justification because, like, say,
astrophysics,32 it is intrinsically justified and that is good enough: we simply
want to know what is true. But the objector could well retort that that is not
good enough for an ethicist, as opposed to a physicist,33 since the former has
to be concerned about practical upshots. It would be absurd for a student of
the most practical subject matter of all—the one which asks, “How shall one
live?”—to be indifferent to the relevance of that question to his or her own
life’s work, not to mention, taking up the reader’s time.
I accept that argument.34 In fact, I believe that ethicizing has practical value
for the very enterprise of becoming more ethical. Therefore let me mention
some ways that ethical theorizing can be useful:
1. A theory can jump-start our thinking by telling us what we are aiming at
in trying to answer an ethical question. It does this by moving us from a
mere word, such as “right” (or “wrong”), to a precise explication of its
meaning. For example, suppose the president of the United States were
trying to decide whether to invade Iraq, that is, what is the right thing to
do? If she were what we called a nationist in Chapter 3—the kind of consequentialist who valued one’s country above all else—she would know
exactly what needed to be pondered or researched to answer the question,
namely, whether that action would be likely to secure America’s welfare
in the long run better than any alternative action.
2. Another way to think about what theory does for us is that it offers a reason for our decision. So, instead of the uninformative “We are going to invade Iraq because it’s the right thing to do,” we’ve got, “We are going to
invade Iraq in order to secure the welfare of our country.” Why is this
progress? Not simply because it gives us something to say (or think to our-
The Ethics of Ethics
81
selves); after all, that might just be window-dressing or propaganda or
disingenuousness or bad faith or public relations/marketing. The great usefulness of having a reason, I submit, is that it invites critical scrutiny,
whether by oneself or others. In other words: Having a reason for something raises the question of whether it is a good reason.
To return to our example: The reason given might seem already to be a
good one, namely, protecting our country. But note that this reason has two
distinct components, which we can set up as premises in a formal argument, thus:
1) One ought always to choose that course of action which promises the
greatest welfare to one’s own country. (ethical/theoretical premise)
2) Invading Iraq is more likely to secure our country’s welfare than any
alternative option. (factual premise elicited by the ethical/theoretical
premise)
Therefore we should invade Iraq.
It becomes clear that for this to constitute a good reason to invade Iraq,
both premises of the argument must be shown to be true (or probable).
Theory has therefore shown itself to be useful by placing this demand
upon us. For either or both of the premises might have seemed obvious to
the would-be invader, even enjoying the status of assumptions scarcely
worth awareness not to mention questioning. But now that they stand forth
in all their naked assertiveness, they demand attention and examination.
What good is that? To help us be sure we are making the right decision!35
3. Theory or theorizing also demonstrates its utility by offering us alternatives to consider. In the example, the given theory was a kind of ethical nationism, which even upon explicit inspection could seem a no-brainer to
the invasion proponent . . . unless she were cognizant of other theoretical
options. For example, utilitarianism holds that the welfare of all nations
(i.e., people), not just one’s own, carries moral weight. This would in turn
augment the set of facts that would be deemed relevant to deciding the
question, thus:
1) One ought always to choose that option which promises to maximize
the (aggregate) welfare of the world’s people (and, by extension, nations).
2) Invading Iraq is more likely to secure the world’s welfare than any
alternative option.
Therefore we should invade Iraq.
I dare say the second, factual premise of this argument is on even shakier
ground than that of the previous argument; but in any case, it suggests that
we ought to consider not only such things as the safety and prosperity of the
homeland but also the security of the Iraqi people from foreign invasion and
82
Chapter Six
occupation as well as domestic unrest, how the regional balance of power
might be altered (Iran ascendant) and with what consequences (e.g., for local women’s rights, peace in Palestine, the world economy, etc.), what regional conflicts might be exacerbated (Kurdish/Turkish, and Pakistani/Indian relations over Kashmir should fundamentalist forces be strengthened in
Pakistan or need at any rate to be appeased by the government), whether a
“pre-emptive” strike against Iraq provides a pretext for other nations to
wage wars with their own antagonists (Russia vs. Chechnya), or to acquire
nuclear weapons in their own defense (Iran, North Korea), and so forth. A
tall order, but then a war is not something to be undertaken lightly.
4. But suppose the multiplicity of theories converged on the same answer to
this (or any!) ethical question: what use would there be to considering
them? I suggest that they offer an exhaustive range of motivational
“hooks” to lure in people of every psychological “persuasion.” Thus, those
people who are disposed to “watching out for Number One” could be appealed to on the basis of egoism; those who care about people and animals
generally would respond to utilitarian claims; and those who stand up for
rights and doing one’s duty would hark to the call of Kantianism. All could
be given a convincing reason to do the right thing.36 In sum, while one
kind of reasons for ethical theorizing is to help us know what to do, a second kind is to help motivate us actually to do it.
5. Finally, another advantage of there being a multiplicity of ethical theories,
despite their apparent superfluity if they all yield the same answers, is that
they keep us on our ethical toes. For as we saw in our examination of the
thought experiment, the very fact that Kantianism and utilitarianism gave
different prima facie responses stimulated us to consider the problem more
carefully. And that ultimately may be the most important way that ethics and
ethical theorizing prove their utility: not necessarily by generating “right answers” to ethical questions, but by guiding our thinking about ethical issues.
So much for the utility of ethicizing. But what of its ethics? Let us consider
the question in light of the arguments of this book. If consequentialism (and
in particular, utilitarianism) were true, then for ethics (i.e., ethicizing) to be
ethical, it would have to be shown not merely useful but more useful than (or
at least as useful as) any alternative activity (not to mention, career) one
might choose instead. After all, the question before us is not just whether ethicizing is or might be useful, since, for example, it could be quite useful in providing ethicists with a living! Meanwhile there are clearly many occasions
when the last thing one should be doing is ethical theorizing! I have argued
that ethicizing can help us to become more ethical, and of course by consequentialist lights being ethical is a good thing by definition since it translates
The Ethics of Ethics
83
into doing one’s best to make this the best possible world. But might there be
other activities (etc.) that could make us even more ethical than (theoretical)
ethicizing would? Possibly. Some forms of meditation or yoga might do the
trick, or a life of selfless service to the sick or the poor, or working with
wounded animals . . . . I don’t doubt there are many candidates.37
But of course I do not accept consequentialism to begin with. It was a foregone conclusion that I would be able to concoct some happy (albeit necessarily faux) consequentialist story about ethicizing since, I have argued, that
can be done for anything at all. I do happen also to believe this story, but it
may be (indeed, is likely?) wishful thinking for all that. Thus, my having
demonstrated the utility of ethicizing would seem to be doubly removed from
establishing the ethics (i.e., ethicality) of ethicizing. It does not do so in terms
of utility since, as just argued, maximum utility of all options would be required, but I have demonstrated no such thing; and, for all we have said, it
does not do so in terms of Kantianism since, according to that theory, utility
is not decisive.
Now recall how this final discussion began: the question was whether ethicizing matters. I then proceeded immediately to a defense of the utility of
ethicizing because I accepted the challenge that ethicizing ought to reflect
what ethics is about, viz., doing the right thing, and doing the right thing
seems a practical affair. But that approach presumed that the only way something could matter or be ethical or even be practical was in terms of its utility; but that is precisely the position that has been rejected in this book. Therefore the real question before us is whether the value of ethicizing can be
characterized in Kantian terms.
Yes, like this: there are many ways to try to get people (including oneself)
to do the right thing, for example, habituation, conditioning, coercion, legal
sanctions, brainwashing, and so forth; but only ethical theorizing does so in a
way that fully respects us, namely, as rational and autonomous. This is the
kind of beings (or ends-in-themselves) we are; therefore it is imperative that
we respect ourselves as such. Since according to Kantianism, as we have
noted, it is perfectly all right to treat people as means (in this case, to the end
of promoting ethicality) so long as at the same time they are being treated as
ends-in-themselves, it follows that in the normal run of cases a Kantian would
find nothing wrong with doing ethical philosophy for the purpose of promoting ethicality, and might even conceive it as an ethical imperative. For it is
when one thinks, articulates, discusses, and debates one’s way through to the
oughts and ought-nots that they become truly binding on a rational and autonomous being. At that point what had been two—justification and motivation: dancers in a pas de deux throughout this book—become one.
Om
84
Chapter Six
NOTES
1. Strictly speaking I should add: “if ethics requires a foundation, that is, a justification in terms of an underlying principle or theory.” The alternative, I suppose, is
that ethics is simply a natural phenomenon. Cf. (Western) religion: it would seem to
have a foundation or justification only if there is a God; otherwise it is (only) subject
matter for sociology, etc. See also Appendix I.
2. Bok (1988) provides an interesting discussion of what Kant was actually getting at with his use of this motto. Cf. also Donagan (1979), pp. 206–7.
3. Where terrorism is defined as (a form of) non-state violence. Of course that
definition raises issues of its own.
4. Cf. Bufacchi and Arrigo (2006).
5. Kant himself supported capital punishment; see Kant (1797), 6:331ff. Cf. my
remarks in the sequel about possible Kantian support for torture.
6. See Appendix I for a discussion of meta-ethics.
7. See Kant (1785), pp. 414–15.
8. Cf. Cummiskey (1996) in speaking of his proposed ethical theory of Kantian
consequentialism: “The duty to promote the good is a categorical imperative” (p. 6).
9. That is, while itself a normative claim, Kant’s formulation of humanity must,
like any normative ethical theory, pass various meta-ethical tests (as per Appendix I),
including the requirement that any ethics consist in a supreme injunction or categorical imperative, in order to be in the running for the status of being (genuinely) ethical and true.
10. Mitchell Silver has suggested (in a personal communication) that “Be rational”
is a categorical imperative, which, furthermore, must be shared by all ethical theories.
My response is that it cannot be because it would lack any ethical force. To whom—
that is, to what sort of “nature”—would such an injunction be addressed? If a being
were not presumed rational, both by nature and under the circumstances, then what
ethical force could “Be rational,” or any other injunction, have? (Analogy: knowledge
of the law is presumed by law but is not itself mandated by law.) Yes, it would be categorical in the sense of being “brute”; but it would carry no more ethical weight than
“Shut the door!” if the addressee were not in a condition to respond to it rationally.
The same, by the way, could be said about freedom as a prerequisite of ethicality:
surely it would be absurd to command somebody to “Be free!” Therefore any ethical
theory is entitled simply to presume that human beings are always in a condition of
freedom and rationality, or else that ethics fails to apply to a person to the degree that
he or she is bereft thereof.
11. I am using “hypothetical” to refer to something that applies only under certain
conditions; thus, “categorical” is synonymous with “unconditional,” that is, under any
and all conditions. For example, “Do not lie” would be hypothetical in the Kantian
scheme I am proposing because it is shorthand for “Do not lie whenever (or if ) doing
so would involve treating someone merely as a means,” whereas “Do not treat anybody
merely as a means” has no qualification at all (other than its being addressed to somebody who can cognize it and act on it). The Kantian hypothetical is sometimes conceived with the more restrictive meaning of appealing to some inclination or desire—
The Ethics of Ethics
85
for example, “Do not lie if you hope to stay out of jail”—but the kind I mean appeals
to the full-fledged Kantian agent, who can will to act contrary to inclination.
12. Sophie’s Choice by William Styron (New York: Random House, 1979).
13. A “real-life” example of the kind of short-shrifting of ethical reasoning I am
arguing against is Greene (2008). In my opinion Greene uses merely stick-figure
stand-ins or caricatures of utilitarianism and Kantianism to draw his conclusions.
14. Cf. Aristotle: “. . . moral virtue comes about as a result of habit, whence also
its name (ethike) is one that is formed by a slight variation from the word ethos
(habit)” (Nicomachean Ethics: bk. 2, ch. 1).
15. I credit Joel J. Kupperman (1983) with this idea. Also cf. Adams (1976), Hare
(1981), and Railton (1984).
16. That is, insofar as she or society can exercise control over her having it.
17. Cf. Donagan (1979)’s parsing of a different thought experiment (pp. 207–8).
18. See Kant (1785), p. 393.
19. For that matter, being discovered by the Nazis might not lead to everybody’s
death. I suppose an appropriate response to that would be “And I’m the Queen of
Sheba.” For this reason I do not (of course) emphasize the consequentialist aspect of
my refutation of the decisiveness of the Nazi-type thought experiment. As I explained
in Chapter 4, my refutation of utilitarianism does not rest on skepticism (“not . . . total ignorance of the future . . . but ignorance of the total future”).
20. See my “Showdown” in Philosophy Now no. 52 (August/September 2005), p. 51.
21. And hence its danger is real, since one can “justify” anything. (But this is true for
any theory, so consider my treatment of this problem in the final section of this chapter.)
22. Cummiskey (1996) articulates and defends a “Kantian consequentialism” along
these lines, according to which Kantian values take precedence over eudaemonistic
ones (i.e., those having to do with happiness) but remain, for all that, goals we ought
strive to achieve maximally, even at the cost of violating them, for example in the short
term, as the result of a means/end calculation. “The duty to strive as much as one can to
promote the flourishing of rational beings . . . is the very essence of treating humanity
as an end. . . . Kant’s moral theory does not provide a rationale for basic agent-centered
constraints that limit what we can do in the pursuit of this . . . moral goal” (pp. 158–59).
I must admit that parts of the Grundlegung (Kant 1785) lend themselves to this kind
of interpretation, including the very opening: “There is no possibility of thinking of
anything at all in the world, or even out of it, which can be regarded as good without
qualification, except a good will” (p. 393). That might seem to suggest that good-willing (which I would parse as exercising an attitude of respect and caring) is analogous
to pleasure in the hedonistic form of utilitarianism, as something aggregable, even
maximizable: the more of it, the better. The alternative I am defending is for good-willing to be understood as an obligation. While it is true that it would be a good thing for
a person to fulfill this obligation, that is not what grounds the obligation. (That would
be somewhat analogous to conflating good outcomes with rewards, as was discussed
in the Introduction in a note about karma.) The obligation is instead elicited and justified by the nature of ends-in-themselves, as explained in Chapter 5. Even if fulfilling
this obligation were to result in disaster, including the diminution of good-willing, it
would remain an obligation.
86
Chapter Six
23. However, I must acknowledge that Kant (1797) himself would seem to have
provided a general allowance of prima facie violations of his own categorical imperative, a kind of two wrongs make a right, thus:
. . . if a certain use of freedom is itself a hindrance to freedom in accordance with universal laws (i.e., wrong), coercion that is opposed to this (as a hindering of a hindrance to
freedom) is consistent with freedom in accordance with universal laws, that is, it is right.
Hence, there is connected with right by the principle of contradiction an authorization to
coerce someone who infringes upon it” (6:231).
24. By the way, Cummiskey’s (1996) term for the position I have just refuted,
“Kantian consequentialism,” could also be a name for the empirical claim that the
best world (possibly defined hedonistically or eudaimonistically) could be brought
about by promoting Kantian values, i.e., concern for all beings as ends-in-themselves;
in other words, this would be Kantianism as means versus Cummiskey’s Kantianism
as end. But again I say: if there were a way to predict optimal outcomes, we might sit
down and discuss this. But there isn’t.
25. Cf. Lenman (2000): “. . .we might prefer a theory that tells a different story
about what the point is of our concern with visible consequences. And such a story
would precisely not be consequentialist” (p. 365).
26. The notion in Plato’s Republic that the welfare of society could justify perpetrating a falsehood on the populace (414b).
27. Personal communication (29 Sept. 2007).
28. Would intercessory prayer therefore be ethical, that is, specifically, when there
is no empirical basis to believe it will have any kind of effect (for example, when
praying for complete strangers who know nothing about it and who will not otherwise
be affected by anything you do as a result of praying)? It is hard for me to see how
this could be a rational practice, a fortiori how it could be an ethical one. Kantianism
does not endorse mere sentimentalism, not to mention superstition. (Of course I am
open to anybody attempting to argue for the rationality of intercessory prayer. Nor am
I questioning prayer simpliciter since it can undoubtedly induce calm and reflection,
which can have further practical benefits.)
29. Keeping in mind the argument from Chapter 4, however, that nothing about
evolution guarantees the utilitarian good.
30. In Plato’s Euthyphro dialogue, Socrates argues that we are able to judge the
(goodness of the) gods by our pre-theological sense of goodness, rather than judging
goodness by our knowledge of the gods.
31. This is surely not the first time in the history of philosophy that the question
has been raised. Epictetus had something to say on the subject two millennia ago, to
wit:
The first and most necessary place (part) in philosophy is the use of theorems (precepts),
for instance, that we must not lie: the second part is that of demonstrations, for instance,
How is it proved that we ought not to lie: the third is that which is confirmatory of these
two and explanatory, for example, How is this a demonstration? For what is demonstration, what is consequence, what is contradiction, what is truth, what is falsehood? The
The Ethics of Ethics
87
third part (topic) is necessary on account of the second, and the second on account of the
first; but the most necessary and that on which we ought to rest is the first. But we do the
contrary. For we spend our time on the third topic, and all our earnestness is about it: but
we entirely neglect the first. Therefore we lie; but the demonstration that we ought not to
lie we have ready to hand. (Enchiridion, item 52, my emphasis).
And before him, Aristotle:
. . . the present inquiry does not aim at theoretical knowledge like the others (for we are
inquiring not in order to know what virtue is, but in order to become good, since otherwise our inquiry would have been of no use) . . . . (Nicomachean Ethics, bk. 2, ch. 2)
32. Or philosophy, according to de Sousa (1986) as previously quoted.
33. One might even question the value of physics if it truly did lead nowhere of
practical import; but in fact even those obscure quantum physics turn out to have exceedingly practical implications, as immediate as the computer I am writing these
words with (cf. Bindloss 2003).
34. Hence too I am partial to “phi beta kappa” (from the initials of the Greek
philosophia biou kybernetes) or “philosophy [as the] guide of life.”
35. Hence, to repeat Gibson’s and Lewin’s remark from the Preface: “There is
nothing so practical as a good theory.”
36. Of course by “the right thing” I mean whatever Kantianism would prescribe,
but see the following paragraph for my answer to the charge of question-begging.
37. I owe this point to Allan Saltzman (personal communication).
Appendix I
What Are We Talking About?
(What is ethics?)
This book is about ethics, but what is ethics? I take the topic to be partitioned
in the following way:
Ethics (proper): The study of right and wrong and good and bad in the senses
of morality and the good life. Sometimes these objects of study are themselves referred to as “ethics,” so ethics (the discipline) studies ethics (the
subject matter). When necessary to distinguish these, I refer to the former
as “ethicizing” and the latter as “ethicality.”
The following are all subsets of ethics as a discipline, i.e., forms of ethicizing:
Normative ethics: Theories about ethics (the subject matter). For example, the
theory of utilitarianism holds that the right thing to do is that which has the
best consequences of all available options. Normative ethics is the effort to
formulate and defend the true theory of ethics against all opposing candidates.1 This book presents a normative defense of Kantianism.
Applied or practical ethics: The application of theories of ethics to general or
particular acts, practices, problems, issues, policies, and so forth. For example, one could consider whether abortion is ever justified by considering it in the light of Kantianism, the theory that one ought never to treat any
person merely as a means. In Chapter 6 of this book, ethical theory is applied to ethical theorizing itself.
Meta-ethics: The study of ethics as such, including the question, “What is
ethics?” Now as we saw above, “ethics” (in the sense of ethicizing) can itself
mean the study of ethics (in the sense of ethicality), so how is meta-ethics
89
90
Appendix One
different from ethicizing? It isn’t; that is, meta-ethics is itself part of the study
of ethicality. It is the more rarefied part. Specifically it is whatever part of the
inquiry is “above” (“meta”) normative ethics. Since I have characterized normative ethics as the theoretical study of ethics, the most straightforward kind
of meta-normative ethics is what we could call meta-theory. This is the effort to set and defend the most general characteristics that any bona fide ethical theory must have, that is, even to be “in the running” as a candidate for
truth (at which point normative ethics would take over). For example, a commonly accepted meta-ethical principle is that ought implies can; this holds
that ethical theory mandates only actions we are incapable of performing.
Thus, “Thou shalt walk to the Moon” would not qualify. The principle follows from a general characterization of ethics as the guide to life, for how
could it be so if its guidance could not be utilized?
Another example of a meta-ethical constraint on theories comes from the
application of sociobiology to morality. According to this characterization of
ethics, our deepest intuitions about right and wrong are rooted in our evolutionary development. So for instance, we are capable of and approve altruism
because beings like us who exhibited this quality have been more likely to
survive under the conditions of existence in our world (and hence pass on
their “altruism genes”) than ones who lacked it.2 There would presumably
then be a corresponding meta-ethical criterion that ethics must be biologically
(i.e., evolutionarily) sustainable for ethical agents as such.3 This is similar to
but interestingly different from the first principle (“ought implies can”) that
ethics must be practicable in everyday terms; so not only must any ethics refrain from prescribing impossible feats, but also from being self-defeating.4
For example, if an ethics commanded twiddling your thumbs no matter what,
even supposing this do-able it would presumably vanish without a trace in a
generation—hardly a viable guide to life.5
The first answer above to the question “What is ethics?” viz., “the guide to
life,” is widely accepted (although subject to interpretation). The second, that
ethics is a biological adaptation, has been resisted but is now all the rage among
philosophical ethicists, including yours truly, although its complete significance
remains obscure. A third is as controversial as ever, namely, that ethics is a cultural phenomenon. An implication would be that ethics is “relative” (to cultures) as opposed to being uniformly compulsory across all cultures. The most
common conception of the latter alternative has been as a set of absolute obligations laid down by God; but even 2,400 years ago Socrates could see6 that
God is somewhat beside the point in figuring out the source of ethics’ claims on
us and whether ethics is fundamentally the same across all times and places.
What Are We Talking About?
91
The truth of ethical relativism would tend to shift the study of ethics into
the realm of description (e.g., “For this outcast and propertyless subculture
stealing is [considered] an honorable act”) rather than prescription (e.g.,
“Stealing is wrong”), hence making it more amenable to the methodology
of history and science (sociology or cultural anthropology) than to philosophy. However, it would also be at cross-purposes to biology insofar as biology would tend to predict a uniform human nature7 (e.g., “A prohibition
on stealing has become ingrained since stealing is disruptive to the social
cohesion required for survival of the group on whom depends in turn the
survival of the individual gene-carrier”). Hence also a true theory of ethics
could seem a will-o’-the-wisp, and so normative ethics a somewhat pointless enterprise.
In this book I assume the falsity of ethical relativism. In other words, I accept as a meta-ethical principle that an ethical theory is, in the first instance,
about justification rather than explanation; for example, I want to know why
stealing is wrong (when it is wrong) rather than why certain people think that
stealing is wrong or are motivated not to steal (and others not).
Other meta-ethical principles cited in this book are:
Moral implies must: Any ethical theory, such as ethical egoism, to provide a
justification for the “command quality” of ethics. For example: Why must 8
I generally tell the truth, according to ethical egoism?
Universal agency: Any ethical theory must confer (ethical) agency status on
all persons of sound mind.
Nonmateriality: Any ethical theory must pick out the object of ethical concern
formally and not materially; that is, the specified object must be general,
for example oneself or the citizens of one’s country or persons of one’s sex,
and not a particular individual, for example, Joel Marks, or a particular
group, such as Americans or men.
Categorical imperative: Any genuine ethics or ethical theory consists in a
supreme injunction or principle of morality. This is a special instance of the
general requirement I would place on any theory, not just an ethical theory,
that it must be expressible in a simple statement (see Appendix II).
Intuitional adequacy: Any ethical theory must be able to satisfy our most basic
pretheoretic ethical intuitions.
92
Appendix One
NOTES
1. Some ethicists argue that theory or theorizing is inessential or even counterproductive to the understanding of ethicality. (Recall the discussion in the Preface.) Hinman
(1998) has made a response to this development that is analogous to my line of argument
in this book (and especially the final part of my Chapter 6, “Justifying the Search for Justification”), to wit: “these objections [to theorizing as such] help to sharpen the theories
themselves, for they must be refined in order to meet the objections” (ch. 8 introduction).
2. But see, e.g., Dawkins (2006), chs. 5 and 6, for admonitions against the too
straightforward application of evolutionary theory to social and moral phenomena.
3. Cf. this statement by the primatologist Frans de Waal (2006): “A viable moral
system rarely lets its rules get out of touch with the biological imperatives of survival
and reproduction” (p. 163).
4. I consider the implication of this additional requirement for the Kantian injunction to “let justice be done though the world perish” (Fiat justitia et ruant coeli) in
Chapter 6.
5. Its practitioners could survive and procreate if they depended on non-practitioners for aid; cf. observant Jews who depend on others to do necessary work on the
Sabbath. Thus, the relevant test must be along the lines of Kant’s formula of universal law: an ethical theory must be (evolutionarily) viable under the condition of everyone’s being a practicing ethical agent. Although I reject that formula as the true theory of ethics (see Chapter 5), it could well serve as a meta-ethical constraint; in other
words, it could be a necessary if not a sufficient condition of being the true theory of
ethics. I note also that finger-twiddling might be viable if nonsentient robots did the
necessary work that finger-twiddling precluded.
6. Cf. Plato’s Euthyphro dialogue.
7. Cf. Pinker (2003).
8. That is, ethically not empirically speaking, of course. Ethical egoism is not
claiming that you “can’t help but” be a truth-teller!
Appendix II
A Simple Theory
(What is theory?)
What is hateful to you, do not do to your neighbor. This is the whole Torah;
all the rest is commentary. Go and learn it.
—Hillel, Talmud, Shabbath 31a
For all the law is fulfilled in one word: Thou shalt love thy neighbour as
thyself.
—Paul, Galatians 5:14
This book is about ethical theory, but what is a theory? I won’t attempt a complete definition, but I would like to offer a certain characterization, to wit:
Theories are expressible by a single, simple statement. Note then that I have
just illustrated my thesis by offering a single, simple statement—my theory
of theoretical simplicity, if you will. What I wish to deny is that a theory is a
collection of statements about a given subject, even if it purports to explain
or elucidate that subject completely. A theory about any subject of substance
will indeed give rise to countless further statements, but somehow they will
all be “contained” within—that is, derivable from—a single, simple statement
that expresses its pith.
Let me give an example. The ethical theory of utilitarianism can be
summed up by the expression, “Maximize pleasure” (or, the declarative
equivalent of that injunction, “One ought always do whatever promises to
bring the most pleasure”). Now this motto immediately invites comments
from the uninitiated, such as the following:
“What a self-centered theory. It says everybody should just seek their own
happiness.” (No it doesn’t; it doesn’t say “Maximize your own pleasure.”
93
94
Appendix Two
It’s concerned with the sum total of pleasure in the universe.)
“What a short-sighted theory. How would anything great ever be achieved if
our highest concern were just getting the most pleasure?” (Not at all; “maximize” here has no time limit.)
“What a crass theory. Surely there are far grander things than mere pleasure.”
(That depends on what you mean by pleasure; “utility” can include creating and enjoying art, theorizing, even being a do-gooder.)
“What a rose-colored theory. Sometimes there is no pleasure to be had from
any available option.” (True enough; but the theory implies that under such
circumstances one should choose the least evil.)
“What an impossible theory. You can’t make everybody happy. Sometimes
you even have to sacrifice your own pleasure for the greater good.” (Right
you are; but there is nothing in the formulation of the theory that contradicts that.)
“What a ridiculous theory. People can’t always be performing these immense
calculations of what will have the best results.” (Of course not; but that’s
only because maximum pleasure could never be achieved in that way.)
Etc. ad inf. Now, does that mean the theory of utilitarianism is highly complex (and hence my theory about the essential simplicity of theories is false)?
My answer is “No”; I would say the theory is not at all complex . . . it’s just
that its simple and intuitive essence has infinite implications for the things
that concern us . . . which is exactly why it’s such a powerful theory. The “slogan” “Maximize pleasure” is in fact perfect; it’s just that it invites reflection
on its full meaning. But the meaning is right there, and all those commentators above were simply failing to pay thoughtful attention to those two little
words in combination: “Maximize pleasure.” Their comments were based on
assumptions that were blinding them to what the words straightforwardly
mean. For example (again), “Maximize pleasure” does not say/mean “Maximize your pleasure,” although it could be used to mean that and may sound
as if it does mean that to someone who is unaccustomed to considering the
pleasure of anyone other than him/herself.
An alternative account of theoretical simplicity could be that there are different slogans for different folks; and hence I have been mistaking the part for
the whole. Thus, “Maximize pleasure” might be just one of many (potentially
an infinite number of?) ways the (supposed) essence of utilitarianism could
be put, depending on the interests, biases, knowledge, or intellectual capacity
of the speaker’s audience or interlocutor. For example, to a person obsessed
with rights, the theory could be summed up as, “Rights are not fundamental
but derivative from welfare.” Compare: We may be used to thinking of the
essence of Copernicanism as: The Sun is at the center of the universe. But if
A Simple Theory
95
we were addressing the “Stable Earth Society,” might we more usefully put it
as, “The Earth moves”?
Kant might seem to provide a perfect example of the latter interpretation of
theoretical simplicity. He is notorious for having provided multiple formulations of his ethical theory, known as the categorical imperative, which sound
utterly unlike one another; for example, “Never treat anyone simply as a
means” versus “Always act according to that maxim which you could will to
be a universal law of nature.” So is there a single slogan that sums up the essential Kantian ethical intuition (as I maintain there ought to be) or only any
number of ways of expressing the theory, which is inherently complex, to
various audiences having varying concerns, predilections, personalities, etc.?
And so any given slogan would be just one part of a complex whole, not the
simple essence of the whole shebang.
I am going to stick to my original position: Any theory worth its salt ought
to have a central intuition that is expressible simply. In the case of Kant, my
solution to the puzzle of the multiple formulations is that the first formulation
captures the essence and the second isn’t even correct.1 In the case of Copernicus, I’d say the motion/s of the Earth is/are an implication of the central
conception. And ditto for the utilitarian take on rights.
One final purported counterexample to my thesis should be considered
since it may seem the most obvious: the ethical theory of David Ross. Usually referred to as the doctrine of prima facie duties, although that label is unsatisfactory for a number of reasons,2 it holds that there is not one but five basic reasons that recommend an action as moral, namely, fidelity, reparation,
gratitude, promotion of the good, and non-maleficence.3 Thus, Ross’s view
seems on its face to be multiple. And of course it is in a sense. But I do not
see a contradiction of my thesis. First off, did I not just express his theory in
a single, simple statement? More significantly, the essence of Ross’s theory is
that the right thing to do is whatever action satisfies a basic set of fundamental considerations (however they turn how to be quantified, characterized, and
prioritized). In a word, his theory is distinctive in being pluralist. But that is
itself a single and simple assertion, is it not?
NOTES
1. See Chapter 5.
2. See Stratton-Lake (2002), pp. xxxiii–xxxviii.
3. Ibid., p. xxxv.
Appendix III
Animal Ethics
At the very beginning of his book, Defending Animal Rights, Tom Regan
(2001) refers to bioethics as “applied philosophy” (p. 1).1 In his first chapter,
which recapitulates an encyclopedia entry published some years earlier, Regan proceeds to review major normative theories of ethics to illustrate how
each in turn is compatible with and/or illuminates the moral status of nonhuman animals. Thus, what we might call animal ethics turns out also to be a
sub-branch of applied ethics. For example, the ethical theory of contractarianism (or, more properly, the set of theories of this type) stipulates that morality arises from (or, more precisely, can be thought of as arising from) an
agreement made by self-interested parties, who discover that mutual non-aggression, cooperation, etc., make sense as promoting their best interests.
From this it would follow that other animals do not deserve any special moral
consideration other than what would tend to further the welfare of the contractors; for example, it might be determined that it is wrong to treat animals
cruelly, but only because such behavior could have adverse effects on the human members of the community, by, say, tending to coarsen their treatment
of one another.2
Contrast the upshot of contractarianism to that of the ethical theory of utilitarianism, according to which one ought always do whatever promises to maximize the pleasure or welfare of all sentient beings. Such a theory accords moral
consideration to other animals directly and intrinsically since all animals (human included) are, presumably, sentient, that is, capable of experiencing pleasure and pain.3 Thus, here again, cruelty to nonhuman animals would be proscribed, but now for a very different reason, namely, that the animals
themselves deserve not to be treated in this way, and not merely, as with the previous theory, that cruelty to animals could have negative effects on humans.
97
98
Appendix Three
Furthermore, utilitarianism could on other occasions prescribe very different
behavior from contractarianism; thus, whereas contractarianism might condone
or even encourage factory farming as conducive to the contractors’, that is, human, thriving, since it makes meat readily and cheaply available, utilitarianism
could conceivably outlaw it as conducive to unnecessary suffering in the world
as a whole, since if less meat (or none) were produced, billions of animals
would experience far less pain from being spared the inhumane conditions of
mass production.
In animal ethics, then, the various normative ethical theories are taken as
givens, and the question is asked which is best suited to tell us about the
proper treatment of other animals. Hence also the exercise serves as a way to
evaluate the theories themselves; as Regan puts it, “the moral status of nonhuman animals plays an increasingly vital role in the assessment of competing ethical theories” (p. 2). Thus, if, when confronted with the reality, we experienced an unfailing moral intuition that the conditions of factory farming
were wrong, then a moral theory that condoned this method of meat production would be thereby shown inadequate. In our example, contractarianism itself would need to be rejected. Utilitarianism might also have to be rejected
if it in turn failed some further test relative to yet another moral theory, such
as Regan’s favored rights theory. For example, utilitarianism might approve
meat production as such as bringing the best overall balance of good into the
world, weighing animal pain against human pleasure, yet a duly sensitized
moral observer might find any use of animals for food to be as unacceptable
as such use of humans, and on the same grounds, namely, violation of an intrinsic right to life.
While I am in basic agreement with Regan’s views on animal ethics, I would
like to suggest a refinement. Actually, where I differ with him, I think, is on a
point of meta-theory, that is, the nature of theory itself. Moral theories—like
any theories—are not carved in stone. Not only can they be assessed as correct
or incorrect, true or false, and hence accepted or rejected, but they can also be
reformulated. In particular, I consider Kantianism to be a moral theory which
so essentially captures the essence of ethics that to reject it outright simply because it might seem to go against some pre-theoretic or extra-theoretic moral
intuition would be an over-reaction. Regan does reject Kantianism, albeit according it respect as componential to his own rights theory. In its original version, Kantianism failed to recognize the duties that are owed to animals directly.
Like contractarianism, Kant’s own Kantianism gave consideration to other animals only insofar as our treatment of them redounded upon us as humans. But
Regan’s intuition, which I share, is that animals have basic rights on their own
account, simply in virtue of what they are (and what we are, since we too are
animals).
Animal Ethics
99
But this does not lead me to discard Kantianism. Rather I have reinterpreted it, or tinkered with it if you will, to enable it to bring other animals into
the fold of the directly morally considerable.4 This leads me to conclude that
animal ethics is not applied ethics but rather normative ethic.5 For the consideration of nonhuman animals obliges us not only to decide among existing
moral theories but also to alter what we may consider to be the correct one in
a fundamental way. Otherwise I might have had to dub my proposal “Marksism”!
NOTES
1. Cf. my Appendix I on the relation between applied ethics and normative ethics.
2. This recapitulates the discussion in my Chapter 5.
3. Cf. Darwall (2003):
This structure enables consequentialism to take account, in principle, of a wider range of
considerations than can contractarianism or contractualism. Most notably, there is nothing
in the consequentialist conception of morality that ties it specially to the condition of other
members of the moral community, or that restricts its consideration to human beings. (p. 4)
4. In this I am not alone, as other Kantians have done likewise, such as Christine
M. Korsgaard, as previously noted, and Allen W. Wood (1998), who writes:
. . . honoring rational nature as an end in itself sometimes requires us to behave with respect toward nonrational beings if they bear the right relations to rational nature. Such relations, I will argue, include having rational nature only potentially, or virtually, or having
had it in the past, or having parts of it or necessary conditions of it. . . . although nonhuman animals may not possess rational nature itself, they do possess recognizable fragments of it.
I acknowledge that Wood’s account is more Kantian than mine, just as he acknowledges that “Her [Korsgaard’s] argument is more Kantian than mine” (note 11).
The tradeoff is that their efforts must be more convoluted than mine to achieve the
same theoretical end of a Kantianism that is not speciesist. At what point a theory
ceases to be “truly Kantian” at all is of course a judgment call. Meanwhile Julian H.
Franklin (2005) makes a straightforward effort to ground Regan’s notion of animal
rights on Kant’s ethics.
5. Cf. this from Martha Nussbaum (2006): “Correcting the oversight of previous theories is therefore not a matter of simply applying the same old theories to a new problem; it is a matter of getting the theoretical structure right” (p. 1)—albeit Nussbaum
chooses to craft “an alternative theory” (p. 95), which she calls the capabilities approach. On the other hand, Nussbaum is also content to have her theory viewed as “an
extension of or complement to Rawls’s theory” (p. 69), much as I view mine as a version of Kantianism. As she puts it: “Nothing would be less in the spirit of this project
100
Appendix Three
than the wholesale rejection of theories that have illuminated so much about core issues
. . . ” (p. 69). Indeed, Nussbaum finds it necessary to “extend” (p. 351) her own theory
to accommodate other animals; “Doing justice to the claims of nonhuman animals requires major further development of the approach” (p. 93).
Glossary
To find further explication of these terms in the text, please consult the Index.
Animal Golden Rule: Do unto others as you would have others (replacing
“them” in the standard formulation) do unto you. In other words, the others unto whom you are “doing” need not be the same others who might
“do” to you . . . not even the same kind of others.
Being: see End-in-itself.
Calvinism: see Ersatz egoism.
Camel’s Back Argument: Utilitarianism is false because it would make the
morality of any act, and, a fortiori, our possible knowledge of it, dependent
on even the most minute differential net value of the consequences of all
possible options until the end of time.
Candidate (ethical theory): strictly speaking, a theory of ethics that satisfies
the meta-ethical criteria (q.v.) of an ethical theory (q.v.) but may or may not
be the true one (analogous to a declarative sentence that is grammatical but
might be false). The notion is relative, however, in that at a given point in
one’s philosophical investigation, one may not realize that a particular theory fails a meta-ethical test. Thus, in the present investigation “Calvinism”
(a.k.a. ersatz egoism, q.v.) was early recognized to be an ersatz theory (q.v.)
and hence not even a candidate for the correct theory of ethics; but it took
half the book and more to make the case that utilitarianism too fails a metaethical test (viz., the criterion of ought implies can).1
Categorical imperative: the supreme injunction of an ethical theory; e.g.,
“maximize utility” in utilitarianism and “never treat anyone merely as a
means” in Kantianism.
101
102
Glossary
Consequentialism: the view that the basis of ethicality (q.v.) is the maximization of utility (q.v.).
Consequentialist (or Ethical) Continuum: an ordering of all possible (non-ersatz) consequentialist theories according to whose consequences count or
matter.
Consequentialist Critique (or Challenge): the claim that any plausible-seeming nonconsequentialist theory must involve, however hiddenly, a consequentialist appeal.
Consequentialist Illusion: see Illusion Argument.
Consequentialist Kantianism (a.k.a. faux-utilitarian Kantianism or the
Gesundheit theory of morality): the claim that Kantian regard can sometimes (or must always?) be expressed via (faux-) consequentialist behavior
(an answer to the Consequentialist Critique, q.v.).
End-in-itself (a.k.a. being): an entity that values things.
Epistemic Objection (a.k.a. Knowledge Argument, q.v.): Utilitarianism is false
because it would require omniscience to be practicable, and ethics must be
practicable.
Ersatz egoism (a.k.a. Calvinism [after the comic strip character in “Calvin and
Hobbes”] or insert-anyone’s-name-here-ism): the theory that everyone’s
primary obligation in life is the promotion of your (or some named individual’s) utility to the max.
Ersatz theory: any theory that postulates as everyone’s primary obligation in
life the utility-maximization of one particular individual, e.g., Joel Marks,
or group, e.g., Americans or men or Homo sapiens.
Ethical egoism: the theory that everyone’s primary obligation in life is the
promotion of his or her own utility to the max. (See also ethical familism
et al. in Chapters 2 and 3.)
Ethical theory: a cohesive conception of ethicality (q.v.) that satisfies all
meta-ethical criteria (q.v.), including expressibility as a categorical imperative (q.v.). Cf. candidate.
Ethicality (a.k.a. ethics): morality and prudence (or the good life); the right
thing to do; the proper and best way to live.
Ethicizing (a.k.a. ethics): the study of ethicality (q.v.); especially the theoretical study thereof.
Ethics: see Ethicality and Ethicizing.
Faux Kantianism: the claim that one ought to live in accordance with exceptionless laws of behavior that are based on respect of persons and regardless of apparent outcomes.
Faux theory: a commonly mistaken form of a theory as strictly conceived.
Faux utilitarianism: the claim that one ought always to do and be such that
foreseeable utility is maximized.
Glossary
103
Gesundheit theory of morality: see Consequentialist Kantianism.
Hitler’s Humanity: the argument that every living person by approximately
the year 2100 and thereafter will owe his or her existence to Adolf Hitler
(because of Hitler’s pervasive influence and the extreme contingency of
personal identity via fertilization).
Humanism: ersatz speciesism for humans, or the theory that the interests of
all nonhuman species are subsidiary to those of the human species. (Cf. ersatz theory.) Versus ethical speciesism: the theory that every individual is
obligated (or at least entitled) to privilege the interests of the members of
its own species above those of all other species.
Illusion Argument: The appeal of consequentialism is illusory because its explanatory and justificatory accounts are of necessity partial (due to our limited knowledge), and so can always be tailored to the situation.
Inversion Principle: One ought to do something if it is the only thing one can
do—a fallacious inversion and extension of the meta-ethical criterion,
ought implies can.
Kantianism (a.k.a. Kant’s categorical imperative or the Formula of Humanity): Never treat any rational being (i.e., person), including oneself, merely
as a means but always at the same time as an end-in-itself (q.v.).
Kantian utilitarianism: (1) The ethical claim that everyone’s ultimate obligation is to promote the most Kantian world possible, i.e., a world in which
maximal respect is shown to all beings (Kantianism as end). (2) The empirical claim that the best world (possibly defined hedonistically or eudaimonistically) would be brought about by promoting Kantian values, i.e.,
concern for all beings as ends-in-themselves (Kantianism as means).
Knowledge Argument (a.k.a. Epistemic Objection, q.v.): Utilitarianism is false
because the kind of knowledge required by the theory in order to make an
ethical judgment—namely, long-term relative net utility—is beyond human powers.
Leviticus Objection: Utilitarianism must be false because otherwise absolutely everything one did or did not do would be ethically significant.
Marksism (a.k.a. Marks’s Kantianism): Treat all beings (or ends-in-themselves) with due respect and caring.
Meta-ethical criteria: the conditions any theory must satisfy to count as a
candidate ethical theory (q.v.).
Meta-ethics: the study of ethics as such, including the question, “What is
ethics?” (See also Appendix I.)
Normative ethics: theories of ethicality (q.v.). (See also Appendix I.)
Psychological egoism: the empirical theory that all human beings naturally
and irresistibly (consciously or unconsciously) desire their personal welfare above everything and everyone else.
104
Glossary
Theory: a cohesive conception of a subject matter that can be expressed in a
simple statement. (See also Appendix II.)
Utilitarianism: One ought always to act and be (insofar as this is under one’s
control directly or indirectly) such that the utility (q.v.) of all sentient beings is maximized.
Utility: good or welfare or happiness or pleasure or desire—(or preference- or
interest-) satisfaction or rationalized-desire-satisfaction, etc.
NOTE
1. Does this mean that (according to my argument) utilitarianism is not, strictly
speaking, false, since it is not even a genuine candidate ethical theory (analogous to
a sentence fragment that is not false—not to mention, not true—because it is not even
an assertion)? In Quinean fashion (Quine 1951) I do not worry about an analytic/synthetic matter such as this.
Bibliography
Ackerman, Bruce, and James S. Fishkin. 2004. Deliberation Day. New Haven: Yale
University Press.
Adams, Robert. 1976. “Motive Utilitarianism.” The Journal of Philosophy 73: 467–81.
Reprinted in Darwall (2003).
“Animal Abuse and Human Abuse: Partners in Crime.” People for the Ethical Treatment of Animals fact sheet. http://www.peta.org/mc/factsheet_display.asp?ID=132
(1 August 2008).
Appiah, Kwame Anthony. 1990. “Racisms.” Pp. 3–17 in Anatomy of Racism, edited
by David Theo Goldberg. Minneapolis: University of Minnesota Press.
Aristotle. Nicomachean Ethics. Trans. W. D. Ross.
Armstrong, Karen. 1994. A History of God. New York: Ballantine Books.
Asimov, Isaac. 1972. The Gods Themselves. New York: Doubleday.
Banfield, Edward C. 1967. The Moral Basis of a Backward Society. New York: Free
Press.
Baumgardt, David. 1946. “Legendary Quotations and Lack of References.” Journal
of the History of Ideas 7, no. 1: 99–102.
Bentham, Jeremy. 1789. The Principles of Morals and Legislation.
The Bhagavad-Gita: Krishna’s Counsel in Time of War. 1986. Trans. Barbara Stoler
Miller. Toronto: Bantam Books.
Bindloss, Ian P. 2003. “Contributions of Physics to the Information Age.”
http://www.physics.ucla.edu/~ianb/history/ (1 August 2008).
Bok, Sissela. 1988. “Kant’s Arguments in Support of the Maxim ‘Do What Is Right
Though the World Should Perish.’” Argumentation 2: 7–25.
Brandt, Richard B. 1979. A Theory of the Good and the Right. Oxford: Oxford University Press.
Broad, C. D. 1952. “Egoism as a Theory of Human Motives.” Pp. 218–31 in Ethics
and the History of Philosophy. London: Routledge and Kegan Paul.
105
106
Bibliography
Bufacchi, Vittorio, and Jean Maria Arrigo. 2006. “Torture, Terrorism and the State: a
Refutation of the Ticking-Bomb Argument.” Journal of Applied Philosophy 23, no. 3:
355–73.
Butler, (Bishop) Joseph. 1726. Fifteen Sermons Preached at the Rolls Chapel.
Byron, Michael, ed. 2004. Satisficing and Maximizing: Moral Theorists on Practical
Reason. Cambridge, U.K.: Cambridge University Press.
Callicott, J. Baird. 1989. In Defense of the Land Ethic: Essays in Environmental Philosophy. Albany, N.Y.: State University of New York Press.
Camus, Albert. 1955. The Myth of Sisyphus and Other Essays. Trans. by Justin
O’Brien. New York: Vintage Books. “The Myth of Sisyphus” was originally published as Le Mythe de Sisyphe. 1942.
Crisp, Roger. 1992. “Utilitarianism and the Life of Virtue.” The Philosophical Quarterly 42, no. 167: 139–60.
Cummiskey, David. 1996. Kantian Consequentialism. New York: Oxford University
Press.
Darwall, Stephen, ed. 2003. Consequentialism. Malden, Mass.: Blackwell Publishing.
Dawkins, Richard. 2006. The God Delusion. Boston: Houghton Mifflin.
de Sousa, Ronald B. 1986. “Desire and Time.” Pp. 83–100 in The Ways of Desire:
New Essays in Philosophical Psychology on the Concept of Wanting, edited by Joel
Marks. Chicago: Precedent Publishing.
de Waal, Frans. 2006. Primates and Philosophers: How Morality Evolved. Princeton,
N.J.: Princeton University Press.
DeLong, Howard. 1991. A Refutation of Arrow’s Theorem. Lanham, Md.: University
Press of America.
Donagan, Alan. 1979. The Theory of Morality. Chicago: University of Chicago Press.
Ellis, Albert. 1962. Reason and Emotion in Psychotherapy. New York: Lyle Stuart.
Epictetus. Enchiridion. Trans. George Long.
Feinberg, Joel. 1974. “The Rights of Animals and Future Generations.” In Philosophy
and Environmental Crisis, edited by William Blackstone. Athens, Georgia: University of Georgia Press. Expanded version at http://www.animal-rightslibrary.com/texts-m/feinberg01.htm (1 August 2008).
Feldman, Fred. 2006. “Actual Utility, the Objection from Impracticality, and the
Move to Expected Utility.” Philosophical Studies 129: 49–79.
Francione, Gary L. 2007. “Reflections on Animals, Property, and the Law and Rain
Without Thunder.” Law and Contemporary Problems 70, no. 1: 9–57.
Frankena, William K. 1973. Ethics. Englewood Cliffs, NJ: Prentice-Hall.
Franklin, Julian H. 2005. Animal Rights and Moral Philosophy. New York: Columbia
University Press.
Frazier, Robert L. 1994. “Act Utilitarianism and Decision Procedures.” Utilitas 6, no. 1:
43–53.
Fromm, Erich. 1956. The Art of Loving. New York: Harper & Row.
Gott III, J. Richard. 1997. “A Grim Reckoning.” New Scientist no. 2108: 36–39.
Greene, Joshua. 2008. “The Secret Joke of Kant’s Soul.” Pp. 35–80 in The Neuroscience of Morality: Emotion, Brain Disorders, and Development (Moral Psychology, Vol. 3), edited by Walter Sinnott-Armstrong. Cambridge, MA: The MIT Press.
Bibliography
107
Hall, Lee. 2006. Capers in the Churchyard: Animal Rights Advocacy in the Age of
Terror. Darien, Conn.: Nectar Bat Press.
———. 2008. “Speaking of Justice and Care.” Abolitionist-Online no. 7 http://www
.abolitionist-online.com/07l_hall2.shtml (1 August 2008).
Hare, R. M. 1981. Moral Thinking. Oxford: Oxford University Press.
Hersh, Seymour M. 2004. “Torture at Abu Ghraib.” The New Yorker 80, no. 11:
42–47.
Hinman, Lawrence M. 1998. Ethics: A Pluralistic Approach to Moral Theory. 2d ed.
Fort Worth, Tex.: Harcourt Brace.
Hooker, Brad. 2008. “Rule Consequentialism.” The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. http://plato.stanford.edu/archives/spr2008/entries/consequentialism-rule/ (1 August 2008).
Howard-Snyder, Frances. 1997. “The Rejection of Objective Consequentialism.” Utilitas 9, no. 2: 241–48.
Hudson, James L. 1989. “Subjectivization in Ethics.” American Philosophical Quarterly 26, no. 3: 221–29.
Kagan, Shelly. 1998. Normative Ethics. Boulder, Colo.: Westview Press.
Kant, Immanuel. 1785. Grounding for the Metaphysics of Morals. Trans. James W.
Ellington. Indianapolis: Hackett Publishing, 1993. Originally published as
Grundlegung zur Metaphysik der Sitten.
———. 1788. Critique of Practical Reason. Trans. Lewis White Beck. Indianapolis:
Bobbs-Merrill, 1956. Originally published as Kritik der praktischen Vernunft.
———. 1797. The Metaphysics of Morals. Trans. Mary Gregor. Cambridge, UK: Cambridge University Press, 1996. Originally published as Metaphysik der Sitten.
———. 1799. “On the Supposed Right to Lie because of Philanthropic Concerns.” Pp.
63–67 in Kant/Ellington (1993). Originally published as “Über ein vermeintes
Recht aus Menschenliebe zu lügen.”
———. 1993. Lectures on Ethics. Trans. Louis Infield. Indianapolis: Hackett Publishing.
Kierkegaard, Søren. 1843. Fear and Trembling.
Korsgaard, Christine M. 2005. “Fellow Creatures: Kantian Ethics and Our Duties to
Animals.” In Tanner Lectures on Human Values, vol. 25, edited by Grethe B. Peterson. Salt Lake City: University of Utah Press.
———. Forthcoming. “Facing the Animal You See in the Mirror.” The Harvard Review of Philosophy.
Kupperman, Joel J. 1983. The Foundations of Morality. London: George Allen and
Unwin.
Langenfus, William L. 1990. “Consequentialism in Search of a Conscience.” American Philosophical Quarterly 27, no. 2: 131–41.
Lenman, James. 2000. “Consequentialism and Cluelessness.” Philosophy and Public
Affairs 29, no. 4: 342–70.
Leopold, Aldo. 1949. A Sand County Almanac. Oxford: Oxford University Press.
Lorenz, Edward. 1972. “Predictability: Does the Flap of a Butterfly’s Wings in Brazil
Set off a Tornado in Texas?” Paper presented at the annual meeting of the American Association for the Advancement of Science.
108
Bibliography
Marks, Joel. 2000. Moral Moments: Very Short Essays on Ethics. Lanham, Md.: University Press of America.
———. Moral and Other Moments (column) in Philosophy Now, bimonthly since June
2000.
Mason, Elinor. 2004. “Consequentialism and the Principle of Indifference.” Utilitas
16, no. 3: 316–21.
Mencius. 1970. Mencius. Trans. by D. C. Lau. Middlesex, U.K.: Penguin Books.
Mill, John Stuart. 1863. Utilitarianism.
Moore, G. E. 1903. Principia Ethica. Cambridge, U.K.: Cambridge University Press.
Norton, Bryan G. 1995. “Applied Philosophy versus Practical Philosophy: Toward an
Environmental Policy Integrated according to Scale.” Pp. 125–48 in Environmental Philosophy and Environmental Activism, edited by Don E. Marietta Jr. and
Lester Embree. Lanham, Md.: Rowman & Littlefield.
Nussbaum, Martha C. 2006. Frontiers of Justice: Disability, Nationality, Species
Membership. Cambridge, Mass.: Harvard University Press.
Pinker, Steven. 2003. The Blank Slate. New York: Penguin.
Plato. 2002. Euthyphro. Trans. by G. M. A. Grube. Indianapolis: Hackett Publishing.
Quine, Willard van Orman. 1951. “Two Dogmas of Empiricism.” Philosophical Review 60, no. 1: 20–43.
———. 1969. “Epistemology Naturalized.” Pp. 69–90 in Ontological Relativity and
Other Essays. New York: Columbia University Press.
Quinn, Daniel. 1995. Ishmael: An Adventure of the Mind and Spirit. New York: Bantam Books.
Railton, Peter. 1984. “Alienation, Consequentialism, and the Demands of Morality.”
Philosophy and Public Affairs 13, no. 2: 134–71. Reprinted in Darwall (2003) and
Scheffler (1988).
Regan, Tom. 1983. The Case for Animal Rights. Berkeley and Los Angeles: University of California Press.
———. 2001. Defending Animal Rights. Urbana and Chicago: University of Illinois.
Reps, Paul, ed. 1989. Zen Flesh, Zen Bones. New York: Anchor Books.
Russell, Bertrand. 1912. The Problems of Philosophy.
Scheffler, Samuel. 1982. The Rejection of Consequentialism. Oxford: Clarendon
Press.
———, ed. 1988. Consequentialism and Its Critics. New York: Oxford University
Press.
Shaw, William H. 2000. “Between Act and Rule: The Consequentialism of G. E.
Moore.” Pp. 6–26 in Morality, Rules, and Consequences: A Critical Reader, edited
by Brad Hooker, Elinor Mason, and Dale E. Miller. Lanham, Md.: Rowman & Littlefield.
Sider, Theodore. 1995. “Sorenson on Unknowable Obligations.” Utilitas 7: 273–79.
Silver, Mitchell. 2006. A Plausible God. New York: Fordham University Press.
Singer, Marcus G. 1977. “Actual Consequence Utilitarianism.” Mind 86: 67–77.
Singer, Peter. 2004. “Taking Humanism beyond Speciesism.” Free Inquiry 24, no. 6:
19–21.
Bibliography
109
Singer, Peter, and Jim Mason. 2006. The Way We Eat: Why Our Food Choices Matter.
Emmans, Pa. Rodale.
Smart, J. J. C., and Bernard Williams. 1973. Utilitarianism For and Against. London:
Cambridge University Press.
Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations.
Stolberg, Sheryl Gay. 2001. “Skin Deep: Shouldn’t a Pill Be Colorblind?” New York
Times, 13 May 2001.
Stratton-Lake, Philip. 2002. “Introduction.” Pp. ix–l in The Right and the Good, by
David Ross. Oxford, U.K.: Clarendon Press.
Taylor, Paul W. 1986. Respect for Nature: A Theory of Environmental Ethics. Princeton, N.J.: Princeton University Press.
Thich Nhat Hanh. 1988. The Heart of Understanding. Berkeley: Parallax Press.
Whewell, William. 1864. The Elements of Morality, including Polity. 4th ed. Cambridge, U.K.: Deighton Bell.
Wilson, Edward O. 1975. Sociobiology: The New Synthesis. Cambridge, Mass.: Harvard University Press.
———. 2002. The Future of Life. New York: Alfred A. Knopf.
Wood, Allen W. 1998. “Kant on Duties Regarding Nonrational Nature.” Proceedings
of the Aristotelian Society Supplement 72, no. 1: 189–210.
Wright, Robert. 2000. Nonzero: The Logic of Human Destiny. New York: Vintage
Books.
Zamir, Tzachi. 2007. Ethics and the Beast: A Speciesist Argument for Animal Liberation. Princeton, N.J.: Princeton University Press.
Index
Abu Ghraib, 65
Adams, Douglas, 54n42
altruism, 22
analytic/synthetic distinction, 104n1
animals, 35, 62–64
the animal condition, 43
animal ethics, 97–99
Animal Golden Rule, 66n3, 67n21
animalism, 35
applied ethics, 89, 97
Aristotle, 65, 85n14, 86n31
Asimov, Isaac, 38n29
assumptions, 81
autonomy. See freedom
beings. See ends-in-themselves
Bentham, Jeremy, 35, 39
blameworthiness, 48
Borge, Victor, 37n20
Broad, C. D., 16n17
Butler, Joseph, 16n17
butterfly effect, 51n6
Callicott, J. Baird, 19, 69n27
Calvin and Hobbes (comic strip), 20
Camel’s Back Argument, 42, 44, 101
candidate (ethical theory), 90, 101,
104n1
capitalism, 35
capital punishment, 72
caring. See concern
casuistry, 79
categorical imperative (Kantian supreme
injunction), 74, 79; ends-inthemselves formulation of, 61–66;
humanity formulation of, 61–62;
universal law formulation of, 61,
92n5
categorical imperative (meta-ethical
principle), 8, 73–74, 91
concern, 64–65, 78–79
consequentialism, 1, 17, 57–58
Consequentialist Continuum, 18–23
Consequentialist Critique, 3, 58, 77; of
divine command theory, 59–61; of
Kantianism, 65–66, 71–73
Consequentialist Illusion. See Illusion
Argument
Consequentialist Kantianism, 78–79
contingency, 41
contract theory, 68n23, 97–98
Copernican formula, 50n4
Cummiskey, David, 84n8, 85n22
Darwall, Stephen, 99n3
Dawkins, Richard, 52n15
de Sousa, Ronald B., 68n24, 87n32
de Waal, Frans, 92n3
111
112
Index
DeLong, Howard, 53n30
dignity, 72
divine command theory, 59–61, 62
Donagan, Alan, 2, 4n5, 85n17
due concern. See concern
dyadism, 21–24
egoism: ambiguity of, 7; vs. ersatz
egoism, 20; vs. selfishness, 9; vs.
utilitarianism, 18. See also ethical
egoism and psychological egoism
empiricist program, 55n50
ends-in-themselves (value-makers),
63–65, 69n27; human beings as, 64,
83. See also animals and sentient
beings
Epictetus, 86n31
Epistemic Objection, 39–41, 50; neglect
of, 2; replies to, 41–49
ersatz egoism (Calvinism), 20
ersatz theory, 22–23, 33; defined, 102.
See also ersatz egoism
ethical egoism, 7–14, 49; and divine
command theory, 59
ethical reasoning, x
ethical theory, ix–x. See also theory
ethicality, 89; and knowledge of future
outcomes, 45; motivation for, 82;
possible nullity of, 14; preconditions
of, 62–64, 84n10; the two
components of, 8
ethicizing, 89; ethicality of, 82–83;
utility of, ix–x, 3, 11, 17–21, 29,
79–82. See also theory
ethics: ambiguity of, 89; comparisons to
physics, 75, 80; practicability of. See
ought implies can; and religion,
24n12, 30–35, 59–61. See also
ethicality and ethicizing
ethics-makers (ethical subjects, ethical
agents), 62–64; as ends-inthemselves, 83. See also universal
agency
Euthyphro, 59–60, 80, 90
evolution, 66n1, 90
familism, 18–19
faux Kantianism, 71–74; defined, 102
faux theory, 102
faux utilitarianism, 46, 76, 78; defined,
102
faux-utilitarian Kantianism. See
Consequentialist Kantianism
Feinberg, Joel, 68n22
Feldman, Fred, 2, 53n24
fiat justitia ruat caelum (“Let justice be
done, though the heavens fall”), 72,
92n4
foreseeable outcomes, 45–46
Francione, Gary L., 69n27
Franklin, Julian H., 67n14, 70n33,
99n4
Frazier, Robert L., 51n7, 52n19
freedom: as a condition of ethics,
62–64, 84n10; as a human trait, 83
Gaia, 19
Gestalt shift, 37n20
Gesundheit theory of morality. See
Consequentialist Kantianism
Gibson, James J., xn2, 87n35
good, theory of the, 15n7, 32. See also
welfare
good will, 85n22
Gott III, J. Richard, 50n4
Green, O. H., 53n27
Greene, Joshua, 85n13
guide to life, 41, 48, 50n3, 87n34, 90
habit, 75
Hall, Lee, xn2, 70n33
happiness, 49, 55n48
Hillel, 93
Hinman, Lawrence M., 91n1
Hitler’s Humanity (argument), 42, 103
Hooker, Brad, 54n37, 54n45
Hudson, James L., 50n3, 51n5, 51n10,
53n26, 70n35
humanism, 35, 67n7, 67n8, 103
hypothetical imperative, 8, 14, 74,
84n11
Index
Illusion Argument, 76, 103
intrinsic value (valuable for itself),
63–64; vs. inherent value (valuable
in itself, end-in-itself), 63–65,
69n27; vs. instrumental value, 63–64
intuitional adequacy (meta-ethical
principle), 74, 76, 91; and the
reformulation of theory, 98; and the
superfluity of theory, 79–80, 82
Inversion Principle (fallacy), 14, 103
justification, 59–62, 84n1; vs.
explanation, 12, 91; vs. motivation,
20, 30, 37n14, 46, 77–78, 83
Kagan, Shelly, 52n19
Kant, Immanuel, 49, 54n34, 85n22,
86n23. See also Kantianism
Kantian consequentialism. See Kantian
utilitarianism
Kantianism, 1, 58, 61–66, 73, 79, 95,
98–99; and motivation, 75, 77–79.
See also categorical imperative
(Kantian) and faux Kantianism
Kantian utilitarianism (Kantian
consequentialism), 71–73, 77; two
types of, 86n24, 103
Karamazov, Ivan, 66n4
karma, 4n4
karma yoga, 66n2
kingdom of ends, 62
knowledge, 39–40
Korsgaard, Christine M., 67n18, 67n21,
68n25, 99n4
Kupperman, Joel J., 85n15
Langenfus, William L., 51n11
Lenman, James, 2, 4n5, 52n18, 52n19,
52n21, 53n28, 53n29, 54n45, 86n25
Leviticus Objection, 51n11
Lorenz, Edward, 51n6
love, 69n28, 70n33
marriage. See dyadism
martyrdom, 31–32
113
Mason, Elinor, 53n22
mattering, 18–20, 22, 35, 62, 64–65,
68n25, 69n27, 83
meditation, 83
Mencius, 68n22
meta-ethics, 14, 80, 89–91. See also
meta-theory
meta-theory, x, 90, 98
Mill, John Stuart, 35, 39
Moore, G. E., 50n4, 52n19, 53n31,
54n39, 54n46
moral agents. See ethics-makers
moral implies must (meta-ethical
principle), 14, 91. See also obligation
morality, 8
moral patients (ethical objects), 62. See
also ends-in-themselves and
nonmateriality
motivation, 75, 82. See also
justification
mythology, 61
nationism (ethical nationalism), 27–30,
33–35
Noble Lie, 78
nonconsequentialism, 57–58
nonmateriality (meta-ethical principle):
20, 91
nonsentient entities: as moral patients,
69n26; and moral agency, 92n5
normative, 12
normative ethics, 89, 91
Norton, Bryan G., xn2
Nussbaum, Martha C., 68n23, 99n5
obligation: and animals, 38n24; and
ethical theory, 14, 91; vs.
prescription, 8; vs. value, 85n22
ought implies can (meta-ethical
principle), 3–4, 13, 50n3, 90; vs. can
implies ought, 49; vs. cannot-but
implies ought (the Inversion
Principle), 14, 103; vs. ought implies
appear-to-be-able-to, 46. See also
guide to life
114
Index
Paul (the apostle), 93
philosophy, 15n3, 25n17, 51n12, 61, 66.
See also science
plants. See nonsentient entities
practical reason, 44–45
prayer, 86n28
proselytizing, 30–32
prudence, 8
psychological egoism, 7, 10, 13–14
Sisyphus, 64
skepticism. See knowledge
Smart, J. J. C., 52n14, 52n19
sociobiology, 90
Socrates. See Euthyphro
Sophie’s Choice, 75
speciesism, 35
sports teams, 29–30
Star Trek, 35–36
suicide, 36n4
Quinn, Daniel, 38n25
racism, 33–35
Railton, Peter, 54n44
rationality: as a condition of ethics,
62–64, 84n10; as a human trait, 83
Rawls, John, 99n5
reasons, 80–81
Regan, Tom, 69n27, 97–99
relativism, 90–91
religionism, 30–35
respect, 83. See also concern
rights theory, 98
Ross, David, 95
rules: vs. laws, 54n33, 71–73; of
prudence, 49. See also ruleutilitarianism
rule-utilitarianism, 47–48
Russell, Bertrand, 55n50
Saltzman, Allan, 87n37
satisficing, 55n47
Sayers, Dorothy L., 67n9
science: difference from philosophy, 7,
10–11
sentient beings, 19–20, 28, 35, 37n12,
67n7, 67n14
Shaffer, Jerome A., 53n23
Shaw, William H., 54n35
Sidgwick, Henry, 50n4
Silver, Mitchell, 78–79, 84n10
Singer, Marcus G., 50n4, 52n18, 54n40,
54n41
Singer, Peter, 52n19, 67n7
Taylor, Paul W., 69n26
tendencies. See rule-utilitarianism
terrorism, 72–73
theory, 21, 37n20, 91, 91n1, 104;
multiplicity of, 82; simplicity of,
93–95. See also ethical theory
Thich Nhat Hanh, 51n12
thought experiment, 25n17, 42, 74–76,
77
torture, 72–73
universal agency (meta-ethical
principle), 21, 91
universal law: as Kantian formulation,
61; as meta-ethical principle, 92n5
utilitarianism, 2, 18–19, 27, 29–30, 36,
39–41, 50n4, 67n7, 93–95, 97–98,
104n1; and divine command theory,
59–60; and motivation, 75;
refutations of, 39
utility. See welfare
value. See intrinsic value
value-makers. See ends-in-themselves
values, 77
valuing, 64, 69n26
welfare, 19–20, 31–32, 104
Whewell, William, 2, 4n6, 54n38
will, 75, 78
Wood, Allen W., 99n4
Zamir, Tzachi, xn2
About the Author
Joel Marks is professor emeritus of philosophy at the University of New
Haven (“UNH”). He received the B.A. in psychology from Cornell University and the M.A. and Ph.D. in philosophy from the University of Connecticut (“UConn”). Between his stints at UConn and UNH, Marks also taught at
the Portland (Maine) School of Art, St. John Fisher College, and the University of Rochester. Marks edited or authored three books of philosophy prior
to the present one, namely, The Ways of Desire: New Essays in Philosophical
Psychology on the Concept of Wanting (Chicago: Precedent Publishing,
1986), Emotions in Asian Thought: A Dialogue in Comparative Philosophy,
co-edited with Roger T. Ames (Albany: State University of New York Press,
1995), and Moral Moments: Very Short Essays on Ethics (Lanham, MD: University Press of America, 2000), as well as a scholarly Festschrift of the musical humorist Gerard Hoffnung, co-edited with David E. E. Sloane (Essays
in Arts and Sciences, vol. 21, October, 1992). In addition to numerous articles
in professional journals, Marks has written scores of op-eds and columns for
newspapers and magazines on both ethics and astronomy (sharing, as he does,
Immanuel Kant’s “ever new and increasing admiration and awe [of] the starry
heavens above . . . and the moral law within . . . .).” Since 2000 Marks has
been a regular columnist for Philosophy Now magazine. In addition to his
own speaking engagements, Marks has organized and chaired scores of symposia, panels, and colloquia. For many years Marks also hosted an interview
program on radio station WNHU, West Haven. Marks’s main areas of scholarly interest are theoretical and applied ethics, and both have come together
recently in his thinking about animal ethics, as witness this book. Marks is
currently the convener of an animal ethics study group at Yale University’s
Interdisciplinary Center for Bioethics.
115
Документ
Категория
Без категории
Просмотров
6
Размер файла
1 145 Кб
Теги
kant, 2009, joel, consequentialist, implies, reply, critique, mark
1/--страниц
Пожаловаться на содержимое документа