(Routledge Studies in Ethics and Moral Theory 43) Benjamin Sachs - Explaining Right and Wrong - A New Moral Pluralism and Its Implications-Routledge (2017)

“This book contains several original arguments and thought-provoking
ideas. The primary argument about moral theorizing as an explanatory

endeavour binds moral theory, normative ethics, and applied ethics into a
unique argumentative field that paves the way for many important, distinct,
and novel insights that the author discusses. All this opens new ground for
future debates in the field.”
—Vojko Strahovnik, University of Ljubljana
Explaining Right and Wrong
Explaining Right and Wrong aims to shake the foundations of contemporary

ethics by showing that moral philosophers have been deploying a mistaken
methodology in their efforts to figure out the truth about what we morally
ought to do. Benjamin Sachs argues that moral theorizing makes sense only if
it is conceived of as an explanatory project and carried out accordingly. The
book goes on to show that the most prominent forms of moral monism—
consequentialism, Kantianism, and contractarianism/contractualism—as well
as Rossian pluralism, each face devastating explanatory objections. It offers
in place of these flawed options a brand-new family of normative ethical
theories, non-Rossian pluralism. It then argues that the best kind of non-
Rossian pluralism will be spare; in particular, it will deny that an action can
be wrong in virtue of constituting a failure to distribute welfare in a particular
way or that an action can be wrong in virtue of constituting a failure to rescue.
Furthermore, it also aims to show that a great deal of contemporary writing
on the distribution of health care resources in cases of scarcity is targeted
at questions that either have no answers at all or none that ordinary moral
theorizing can uncover.
Benjamin Sachs is a lecturer in philosophy at the University of St. Andrews.

He has written on topics in normative ethics, political philosophy, philosophy
of law, animal ethics, and coercion. He has papers in journals including
Philosophical Studies and The Australasian Journal of Philosophy.
Routledge Studies in Ethics and Moral Theory
For a full list of titles in this series, please visit www.routledge.com
34 The Intrinsic Value of Endangered Species

Ian A. Smith
35 Ethics and Social Survival

Milton Fisk
36 Love, Reason and Morality

Edited by Esther Engels Kroeker and Katrien Scaubroeck
37 Virtue’s Reasons
New Essays on Virtue, Character, and Reasons
Edited by Noell Birondo and S. Stewart Braun
38 In Defense of Moral Luck

Why Luck Often Affects Praiseworthiness and Blameworthiness
Robert J. Hartman
39 Risk, Technology, and Moral Emotions

Sabine Roeser
40 Wittgenstein’s Moral Theory

Edited by Rashef Agam-Segal and Edmund Dain
41 Welfare, Meaning and Worth

Aaron Smuts
42 Moral Skepticism
New Essays
Edited by Diego E. Machuca
43 Explaining Right and Wrong

A New Moral Pluralism and Its Implications
Benjamin Sachs
Explaining Right
and Wrong
A New Moral Pluralism
and Its Implications
Benjamin Sachs
First published 2018
by Routledge
711 Third Avenue, New York, NY 10017
and by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2018 Taylor & Francis
The right of Benjamin Sachs to be identified as author of this work
has been asserted by him in accordance with sections 77 and 78 of
the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or
reproduced or utilised in any form or by any electronic, mechanical,
or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or
retrieval system, without permission in writing from the publishers.
Trademark notice: Product or corporate names may be trademarks
or registered trademarks, and are used only for identification and
explanation without intent to infringe.
Library of Congress Cataloging-in-Publication Data
Names: Sachs, Benjamin, author.
Title: Explaining right and wrong : a new moral pluralism and its
implications / by Benjamin Sachs.
Description: 1 [edition]. | New York : Routledge, 2017. | Series:
Routledge studies in ethics and moral theory ; 43 | Includes
bibliographical references and index.
Identifiers: LCCN 2017044434 | ISBN 9781138307353
(hardback : alk. paper)
Subjects: LCSH: Ethics—Methodology. | Normativity (Ethics)
Classification: LCC BJ37 .S23 2017 | DDC 170/.44—dc23
LC record available at https://lccn.loc.gov/2017044434
ISBN: 978-1-138-30735-3 (hbk)
ISBN: 978-1-315-14230-2 (ebk)
Typeset in Sabon
by Apex CoVantage, LLC
To my parents, Marcia and Raymond, for their
unfailing support
Contents
Acknowledgments x
Introduction 1
1 Normative Ethical Theorizing as an Explanatory Project 5
2 How Should We Choose Between Competing Explanatory

Stories? 17
3 Against Monism 32
4 Against Rossian Pluralism 61
5 Non-Rossian Pluralism 83
6 The Question of Scope, Part 1: Distributive Moral Concerns 96
7 The Question of Scope, Part 2: Non-Distributive Moral

Concerns 115
8 Doing Harm and Failing to Rescue 126
9 The Distribution of Health Care Resources 143
10 Conclusion 158
Appendix A: Moral Meta-Explanation 164

Works Cited 168
Index 179
Acknowledgments
This book has been a long time in the making—about 13 years if you start
counting from when I began writing what turned out to be Chapter 9. In
that time I’ve received an immense amount of help in writing this book,
both from colleagues at the various places where I’ve worked and studied
and from friendly acquaintances in the broader philosophical community.
They’ve been incredibly generous to me and most of them have asked for
little to nothing in return.
My most sincere thanks goes to Volko Strahovnik and an anonymous
referee for Routledge, each of whom read and commented on the full draft
manuscript. Thanks as well to Andrew Weckenmann and Alexandra Sim-
mons at Routledge for shepherding me through the publication process.
To the people who initially helped me to work out what this whole
project was about and provided encouragement as I was getting started—
Nick Beckstead, Molly Gardner, Dale Jamieson, S. Matthew Liao, Japa
Pallikkathayil, Doug Portmore, Bill Ruddick, Samuel Scheffler, and Russ
Shafer-Landau—I’m grateful. My gratitude as well to Katherine Hawley,
who helped me with my prospectus.
Marcia Baron, John Broome, Jessica Brown, Jonathan Dancy, Kather-
ine Hawley, Dale Jamieson, Joseph Kisolo-Ssonko, Brian McElwee (twice),
Douglas Portmore, Mike Ridge, Toni Rønnow-Rasmussen, Russ Shafer-
Landau, Justin Snedegar, Christine Straehle, Saranga Sudarshan, Jens Tim-
merman, and Pekka Väyrynen each read and commented on a draft of a full
chapter of this book and for that I thank them sincerely.
Finally, I want to express my appreciation for the long list of people with
whom I’ve had helpful conversations about the ideas presented in this book—
a list that I suspect I’ve left incomplete (apologies to those I’ve forgotten to
mention): Chrisoula Andreou, Derek Ball, Greg Bognar, Anna Burkard, Ben
Colburn, Rowan Cruft, Miranda Fricker, Patrick Greenough, Dan Haus-
man, Katherine Hawley, Ulrike Heuer, Dale Jamieson, Hallie Liberto (many
times), Brian McElwee, Tristram McPherson, Uri Leibowitz, Collin O’Neil,
Theron Pummer, Wlodek Rabinowicz, Martin Rendall, Benedict Rumbold,
Ben Saunders, John Skorupski, Nick Treanor, Pekka Väyrynen, Leigh Vicens,
Daniel Viehoff, Kenny Walden, Jonathan Way, and Brynn Welch.
Acknowledgments xi
The ideas presented here were initially presented in conference/workshop
talks and seminars at the London School of Economics and Political Science,
the University of Glasgow, the Ethics and Explanation Workshop at Not-
tingham University, the University of Stirling, The Conference of the Inter-
national Society for Utilitarian Studies, The Joint Session of the Aristotelian
Society and the Mind Association, the University of Sheffield, the British
Society for Ethical Theory annual conference, Cal Poly San Luis Obispo,
King’s College London, the University of Connecticut, the University of
Massachusetts-Boston, the University of Utah, the Polish-Scottish Philoso-
phy Workshop, and my home institution, the University of St. Andrews.
Finally, I thank: Oxford University Press for permission to reprint, in
Chapter 3, material from my “Non-consequentialist Theories of Animal
Ethics”, Analysis Reviews 75 (2015): 638–54; Springer for permission to
reprint, in Chapters 4 and 5, material from my “Direct Moral Grounding
and the Legal Model of Moral Normativity”, Ethical Theory and Moral
Practice 18 (2015): 703–16; and John Wiley and Sons for permission to
reprint, in Chapter 7, material from my “The Status of Moral Status,” Pacific
Philosophical Quarterly 92 (2011): 87–104.
Introduction
Say what you will about act-consequentialism and Kantianism, at least they
hold out the promise of teaching us something that we didn’t already know.
Unapologetic act-consequentialists like Peter Singer and Shelly Kagan are
famous for telling us that widespread beliefs about infanticide and about the
moral obligatoriness of donating large portions of one’s wealth to charitable
causes are mistaken. And Kant offers us very surprising conclusions about
masturbation, dishonesty, and suicide.
By contrast, in the last 30 years or so we’ve been introduced to a lot of
normative ethical theories that not only don’t yield bold conclusions about
what we morally ought to do but in fact appear to be specifically designed
not to. Even act-consequentialists and Kantians have gotten in on the act.
Act-consequentialists have modified their theory by adding features such as
agent-relative value and by attributing value to integrity and the upholding
of rights, and these alterations serve to moderate what would otherwise be
act-consequentialism’s strange and demanding set of obligations. Kantians,
meanwhile, have been hard at work showing that we can stay within Kant’s
basic moral framework while arriving at more common-sense conclusions
about masturbation, deception, suicide, etc.
Contemporary act-consequentialists and Kantians are joined in this effort
by those philosophers who prefer to build a normative ethical theory around
the ideas of rights and duties. Although it may seem, superficially, that we
leave ourselves very little flexibility once we commit ourselves to the exis-
tence of rights and duties, given how those concepts have a bit of an abso-
lutist air about them, this turns out not to be so. Deontologists are quick to
point out that rights and duties can be pro tanto. This makes the question of
what one morally ought to do a matter of weighing considerations, at least in
those cases that aren’t entirely one-sided. Noticing this, an increasing number
of non-consequentialists these days dispense entirely with talk of rights and
duties and simply talk in terms of moral considerations or reasons.
My view on this trend is that if, at the end of the day, what we get out of
our moral theorizing is a theory that yields all the conclusions that we were
ex ante inclined to accept, what was the point of the whole endeavor? I had
hoped moral theorizing would be a way of learning something about what
I morally ought to do.
2 Introduction
This is not to deny that we should continue to talk of moral reasons. I have
my own theory about where moral reasons come from and what explanatory
work they do; I explain it in Chapters 4–5. What I do deny, however, is that
moral reasons talk is helpful in normative and applied ethics. Moreover, it’s
pernicious, because it’s readily available to be used as a tool for avoiding
ever having to admit the falsity of a substantive claim about right and wrong
that one was initially inclined to accept.
Avoiding this conservative bias in moral theorizing is easy if you’re
attracted to some very concrete all-encompassing generalization about what
one morally ought to do. For instance, act-consequentialists have the idea
that one always ought to maximize the good, and Kantians have the idea that
one always ought to obey the categorical imperative, and these commitments
serve as dialectical anchors that enable act-consequentialists and Kantians
to pull off the Ulysses-like feat of avoiding the temptation to accommo-
date every case-specific intuition that comes along, if they’re inclined to try
(which, as I noted already, most of them these days are not). But what if, like
me, you’re not attracted to any version of moral monism?
For moral pluralists, the answer has traditionally been that one should
use the method of reflective equilibrium, which involves sometimes adjust-
ing one’s theoretical beliefs in light of one’s case-specific (or what I will here-
after call ‘verdictive’) beliefs but also being willing to adjust one’s verdictive
beliefs in light of one’s theoretical beliefs. I will show, however, that we’ve
been too quick to accept this answer. We haven’t gotten clear on why one
would ever be justified in adjusting one’s verdictive beliefs in light of one’s
theoretical beliefs, and therefore as yet we can’t legitimately claim to know
why the method of reflective equilibrium is the correct moral methodology.
And I suspect that it’s our ignorance as to why the method is correct that
explains why we pluralists are always running afoul of it—why, that is, we
customarily adopt either the strategy of being biased toward holding on to
our verdictive intuitions when they conflict with our theoretical convictions
or the strategy of rigging things so that the theoretical claims we accept
deliver no verdictive implications at all because they’re merely pro tanto.
My view is that we can vindicate the classic understanding of moral theo-
rizing, on which it’s a way of correcting one’s mistaken beliefs about what one
morally ought to do and also (naturally) a way of learning which theoretical
claims are correct, if and only if we treat moral theorizing as an explanatory
project. Therefore, I begin the book by arguing, in Chapter 1, that the theo-
retical claims contained in normative ethical theories should be represented,
for the purposes of assessing the relative merits of those theories, as explana-
tory claims—claims about what explains the verdictive facts. I then dis-
cuss how we should go about assessing the plausibility of explanatory claims
and how we should understand the dialectical relationship between claims as
to what explains the verdictive facts (“explanatory claims”) and claims as to
what those verdictive facts are. I endorse the method of reflective equilibrium
and the widely held view of that method on which it requires two sets of
Introduction 3
intuitions as inputs, one of which should be verdictive intuitions. I simply add
that if I am right that moral theorizing should be understood as an explana-
tory project, then our explanatory intuitions should constitute that second set.
Chapter 2 offers a detailed picture of a moral methodology that includes an
appeal to our explanatory intuitions.
Beginning in Chapter 3 I deploy the method of reflective equilibrium,
understood the way I argue in Chapters 1–2 that it ought to be understood.
In that method it is standard practice to begin by identifying ‘fixed points’—
intuitions so firmly held that they constitute non-negotiable demands on
the resulting sets of theoretical and verdictive beliefs. Using a small num-
ber of such fixed points among explanatory intuitions, I argue for discard-
ing the three most popular versions of moral monism: consequentialism,
Kantianism, and contractarianism/contractualism. Then, in Chapter 4, I use
another such fixed point as a basis on which to reject the most popular
version of moral pluralism, namely Rossianism. These arguments are based
not on idiosyncratic intuitions, but rather on widespread intuitions that we
moral philosophers rarely use on account of our failure to recognize that
moral theorizing is an explanatory project.
The arguments of Chapters 3–4 leave us with a family of moral views I
call Non-Rossian Pluralism (NP), and in Chapter 5 I flesh out and defend
that family. Next, in Chapters 6–7, I get more specific about which kind of
NP we should accept. I argue in Chapter 6, again on explanatory grounds,
that NP should be welfaristically distribution-insensitive—i.e., that it should
deny that it matters morally in itself how welfare is distributed. And in
Chapter 7 I argue that NP should deny that facts about the moral status (or
moral standing or moral considerability or membership in the moral com-
munity) of a living thing can be part of the explanation of why treating it a
certain way is impermissible, permissible, or obligatory.
The last two chapters continue the project of giving further shape to NP
while also exploring its implications for real world cases. In Chapter 8 I
argue that there is no such thing as a duty of rescue, and based on that
claim I argue for an absolutist position with respect to the ethics of harmful
research on animal subjects. Finally, in Chapter 9 I focus my attention on
the distribution of health care resources, arguing that there is nothing that
could explain the truth or falsity of the proposed substantive answers to
the questions in this area and that we are therefore forced to conclude that
the questions either have no answer at all or no answer that ordinary moral
theorizing can reveal.
Chapters 8–9 constitute an admission that NP has drastically counter-
intuitive implications. Since I endorse the method of reflective equilibrium
and thus the use of verdictive intuitions in moral theorizing, I cannot take
this outcome lightly, and I make this problem the topic of Chapter 10, my
conclusion. There I propose an admittedly oversimplified overall assessment
of the prospects for NP and its competitors, on which NP has the advan-
tage of according well with our explanatory intuitions and the drawback of
4 Introduction
violating many of our verdictive intuitions, while many of its competitors
have the advantage of consistency with our verdictive intuitions and the
fault of doing violence to many of our explanatory intuitions. Based on an
argument to the effect that our explanatory intuitions are, for now, at least
as worthy of our trust as our verdictive intuitions, I conclude with the claim
that accepting NP is provisionally at least as justified as accepting any of its
competitors.
One caveat, however: If one does wish to simply dismiss NP, and with
it all the highly counterintuitive conclusions of Chapters 8–9, I admit that
there is a perfectly legitimate reason to do so even if one cannot identify a
flaw in my reasoning. This entire book is framed by its first chapter, which is
an effort to understand how we should approach moral theorizing; yet there
is no argument in this book for the claim that we should engage in moral
theorizing as our way of arriving at our substantive moral beliefs, though
I do see myself as undermining, in §3 of Chapter 2, the most influential
argument against the fruitfulness of moral theorizing. Still, this book’s lack
of an argument in favor of moral theorizing means that the controversial
conclusions contained in Chapters 8–9 are really disguised conditionals of
the form “If we should base our substantive moral beliefs on the output of
a sound process of moral theorizing then we should believe [some substan-
tive moral claim].” If, having read those chapters, the reader wants to reject
those controversial conclusions without identifying a flaw in my arguments
for them, I cannot quibble. The message of this book is that we can hang on
to all of our common-sense verdictive beliefs, but if we do then we shouldn’t
pretend that those beliefs are supported by the solid doxastic justification
that moral theorizing can provide. And if our beliefs aren’t justified in that
way, then the question becomes whether they’re justified at all.1
Note
1. We might have to fall back on an appeal to the existence of a reliable faculty of
verdictive judgment, as the particularist Jonathan Dancy does (2004, Chapter 8),
or assert that some verdictive moral claims are self-evident. I won’t pursue the
matter here.
1 Normative Ethical Theorizing
as an Explanatory Project
1. Introduction
My goal in this chapter is to establish that when we assess normative ethi-
cal theories we should represent their generalizations as explanatory claims
and assess those claims for intuitiveness. In §2 I argue that if we do this then
moral theorizing can be a way of gaining knowledge about what one mor-
ally ought to do. In §3 I take up an argument to the effect that doing this
constitutes bad methodology, while in §4–5 I consider and reject two argu-
ments to the effect that doing this is not necessary for moral theorizing to be
a way of gaining knowledge about what one morally ought to do. Finally,
in §6, I note the implications this proposed methodology has for the theory
and practice of the method of reflective equilibrium.
2. Representing Normative Ethical Theories as Being

Composed of Explanatory Claims
This chapter is about what kind of project moral theorizing is, where by ‘moral
theorizing’, I mean reasoning all at once about which normative ethical the-
ory to accept and which case-specific ethical claims to accept. Before we
can embark, however, we need an initial, uncontroversial picture of what
a normative ethical theory is. To this end, the first thing to note is that we
wouldn’t call a set of claims a ‘theory’ if none of those claims were general-
izations. Furthermore, we wouldn’t call a set of claims a theory of normative
ethics if it didn’t contain generalizations connecting verdictive facts—facts
as to the permissibility, obligatoriness, supererogatoriness, etc., of various
actions or narrow act-types—to other facts. (So, for instance, Benthamite
utilitarianism contains a generalization connecting moral permissibility and
obligatoriness to utility-maximization.) So for now I’ll insist only that a
normative ethical theory is that kind of set of generalizations.
With this settled, here’s our question: How should we decide which
normative ethical theory to accept? There’s one option that immediately
presents itself once we acknowledge that we each have intuitions as to the
verdictive facts. This is the method of deriving as complete a list as possible
6 An Explanatory Project
of the verdictive claims we find intuitively plausible and then constructing a
set of generalizations, X, that is consistent with that list and then accepting
X for that reason. In other words, we could work exclusively from the bot-
tom up. But almost no philosopher holds that we should proceed in this
way. Furthermore, if this methodology is correct then moral theorizing can-
not be a way of correcting mistaken beliefs as to what one morally ought to
do. We tend to think, however, that it can, and I certainly hope that it can, as
I emphasized in the Introduction. Therefore, I will not concede that it cannot
without thoroughly exploring the alternatives.
Consequently, the following question presents itself: Under what concep-
tion of moral theorizing does the just-mentioned methodology emerge as
mistaken? Or to put it more simply: Why not work exclusively from the
bottom up?
I submit that the just-mentioned methodology is clearly mistaken if moral
theorizing is (perhaps among other things) an explanatory project—
specifically, the project of trying to come up with good generalizations as to
how the verdictive facts are explained. If one thinks of the project this way
then one immediately sees an argument against working exclusively from
the bottom up. We can admit that working from the bottom up to a certain
extent is justifiable, for the obvious reason that we should want, at the end
of the day, to wind up with a set of generalizations that is consistent with
our verdictive intuitions. But we can point out in addition that, since gener-
alizations as to what explains the verdictive facts can themselves be intuitive
or counterintuitive, we should want, at the end, of the day, to wind up with
a set of generalizations the individual members of which are themselves
intuitively plausible. This gives us a reason to work from the top down to
a certain extent, deriving as complete a list as possible of the explanatory
generalizations that strike us as intuitively plausible and then shaping our
verdictive beliefs so as to make them consistent with that set. For instance,
we might have the intuition that the fact that some action is the breaking of
a promise can explain that action’s being impermissible. Working from the
top down means accepting this generalization because it’s intuitively plau-
sible and then accepting the verdictive claims that it implies.
I would add, as a final note, that I think most of us already in fact do this.
Certainly most of us work from the top down to a certain extent, and I
think when we’re working from the top down we’re usually doing the exact
thing I just described. So I conceive of my contribution in this chapter not
as the introducing of a new moral methodology but rather as introducing
as a topic of philosophical inquiry a methodology already in use and show-
ing why it is defensible (§3) and necessary (§4–5). Furthermore, if it can be
established that we should work from the top down for the reason I just
gave—which is the thesis of this chapter—and also work from the bottom
up—which I assume no one will dispute—then we have a vindication of the
moral methodology that just about everyone accepts: the method of reflec-
tive equilibrium. I end this chapter, in §6, by showing how conceiving of
An Explanatory Project 7
moral theorizing as an explanatory project vindicates the method of reflec-
tive equilibrium while at the same time explaining why we so often violate
it despite accepting it.
3. The Reliability of Our Explanatory Intuitions

If we follow my proposed method the choices we make between competing
normative ethical theories won’t be good ones if the intuitions on which I
am proposing we rely—our explanatory intuitions—are unreliable. I admit
that there is legitimate skepticism about the reliability of our moral intu-
itions in general. This is fair enough, but we have no choice but to rely on at
least some of them.1 One has to start with what one already believes,2 and if
one does not trust one’s intuitions at least some of the time then one cannot
participate in moral theorizing as one will have no pre-existing moral beliefs
to reflect upon and refine.3
It is worth noting, further, that the two most popular arguments against
the reliability of our moral intuitions do not apply, at least in any powerful
way, to our explanatory intuitions.
The first argument proceeds by way of inference from the mountain of
research that has been done showing that our moral intuitions are influ-
enced by morally objectionable biases and can be swayed by morally irrel-
evant factors concerning the agent’s environment and the framing of the
question (see, for instance, the work of Joshua Greene). The reason this
argument is, if effective at all, more effective in undermining the reliability
of verdictive intuitions than explanatory intuitions is, quite simply, that all
the research done to date has been on verdictive intuitions. Perhaps similarly
damaging evidence could be collected against explanatory intuitions. We’ll
have to wait and see.
The second argument is the evolutionary skepticism argument, a sub-
argument within Sharon Street’s ‘Darwinian Dilemma’ argument against
moral realism (Street 2006). When laying out the evolutionary skepticism
argument, Street says that our moral judgments are heavily influenced by
our basic evaluative tendencies (“an unreflective, non-linguistic, motiva-
tional tendency to experience something as ‘called for’ or ‘demanded’ in
itself” (119)), which are directly influenced by natural selection. Although
Street does not speak of moral intuitions, what is true of moral judgments
must be true of moral intuitions, since the latter are a subset of the former.
This gets us the claim that our moral intuitions are heavily influenced by
our basic evaluative tendencies. This claim can be used in a skeptical argu-
ment since our basic evaluative tendencies have been subject to the forces of
natural selection, and the forces of natural selection are blind to moral truth,
given moral realism. In other words, whether a basic evaluative tendency
could survive natural selection has nothing to do with its truth, given moral
realism. Putting all this together, we should expect there to be no correlation
between our moral intuitions and the moral truth, given moral realism.
The effectiveness of this argument depends on whether the moral intu-
itions mentioned in it are verdictive. Basic evaluative tendencies—those psy-
chological features of us and our ancestors that are directly influenced by
natural selection—must surely be tendencies to evaluate certain actions or
types of actions, since natural selection selects for and against behavioral
tendencies (or their genetic bases). In other words, basic evaluative tenden-
cies must be verdictive tendencies. So Street can safely claim that natural
selection must exert a strong influence on our verdictive intuitions; she can
point out, quite simply, that our verdictive tendencies directly influence our
behavior and natural selection promotes some behaviors while rejecting
others.
The parallel argument for skepticism about explanatory intuitions would
be much more tenuous, because in the case of explanation there is no ana-
logue of basic evaluative tendencies. We don’t have “unreflective, non-
linguistic, motivational” tendencies to experience certain things as explained
by others. We may have tendencies vis à vis moral explanation, but they’re
surely not motivational, which means they’re not directly connected to
behavior. They are, I admit, indirectly connected to behavior, since explana-
tory claims bear inferential relations to verdictive claims (as I will argue
in Chapter 2), which themselves are directly connected to behavior (as I
already conceded). Therefore, Street can safely claim that natural selection
exerts some influence on our explanatory intuitions; she can point out that
our explanatory tendencies are indirectly connected to our behavior and
natural selection promotes some behaviors while rejecting others.4 Again,
however, the connection is more tenuous in this case than in the case of
verdictive intuitions, so there are more grounds for suspicion in that case
than in this one.
I conclude, therefore, that it is reasonable for us to put weight on our
explanatory intuitions. This is because we have to use some intuitions and
our explanatory intuitions, at least for now, seem at least as reliable as our
other useable intuitions—our verdictive intuitions.
4. The Possibility of Appealing to Our Intuitions About

Conditionals or Categorical Claims
I’ve suggested that if we conceive of moral theorizing as an explanatory
project then we can vindicate the view that moral theorizing is a way of
gaining knowledge as to what one morally ought to do. But whether this is
the only way to achieve this vindication is another matter. In this section and
the next I explore two alternatives.
First, while I’ve suggested formulating moral generalizations as explana-
tory claims and then appealing to our explanatory intuitions to sort through
them, an alternative idea would be to formulate those generalizations as
conditionals or categorical claims and then sort through them by appealing
to our judgments as to their truth. For instance, we might represent a theory
as claiming
‘necessarily, if an action is an instance of killing then it is impermissible’

(conditional) or ‘necessarily, all killings are impermissible’ (categorical
claim)
and then reflect on our judgments as to the truth of these conditionals/

categorical claims.
But this proposal must answer the question of how we arrive at our judg-
ments about moral conditionals/categorical claims, and I doubt there’s a
good answer to be given. Or, more carefully, there’s no good answer to be
given if the moral conditionals/categorical claims aren’t analytically true or
false (in particular, if the action-descriptions they contain do not refer to any
thick or thin moral concepts).
One possibility—and this is what C. D. Broad (1930, 282) and possibly
W. D. Ross (1939, 170) believed—is that we arrive at candidate conditionals/
categorical claims by “intuitive induction” from our verdictive judgments.5
But this is working from the bottom up, whereas we are searching for a way
of working from the top down.
Another possibility is that we can arrive at our judgments regarding
conditionals/categorical claims via some sort of inference from our explana-
tory judgments. However, in this case the current proposal is parasitic on my
preferred solution of construing moral generalizations as explanatory claims.
The final possibility I want to examine is that some of our judgments
about the candidate conditionals/categorical claims might simply be intu-
itions about them. The problem with this proposal is that the justificatory
force of such intuitions is too provisional to be of use in this context. If I
have an intuition that necessarily all As are Bs, this may well constitute a
sufficient justification, or at least a reason, for believing that all As are Bs if I
haven’t yet examined all the As to verify that they’re Bs and examined all the
non-Bs to verify that they’re non-As. But in our moral theorizing we can and
should carry out precisely that examination with respect to any candidate
categorical or conditional, using our verdictive intuitions as reference points.
Having done that, it’s hard to see how our intuition as to the truth or falsity
of some candidate categorical or conditional could still count for anything.
Therefore, what might begin as an attempt to work from the top down will
inevitably and justifiably turn into a working from the bottom up.
Just consider: Suppose I have an intuition of the form ‘all As are Bs’. Sup-
pose, further, that I’ve checked all the As and found that in each case I have
a verdictive intuition that that A is indeed a B and checked all the non-Bs
and found that in each case I have a verdictive intuition that that non-B is
indeed a non-A. This obviously constitutes a reason for me to accept the
proposition ‘all As are Bs’. But if my intuition that all As are Bs also counted
in favor of accepting that proposition, then I’d have two reasons to accept
it. This, clearly, is one more reason than I have.6
I conclude from all this that conceiving of moral theorizing as the proj-
ect of (perhaps among other things) coming up with plausible moral cat-
egoricals or conditionals provides no justification for working from the top
down. Therefore, it doesn’t show how moral theorizing can be a way of
gaining knowledge about what one morally ought to do.
5. The Possibility of Appealing to Theoretical Criteria

In §2 I proposed a way of vindicating the idea that moral theorizing can
be a way of gaining knowledge as to what one morally ought to do and
we are now asking whether there is another way. The new proposal I want
to examine is that moral theorists should work from the top down for the
same reason that scientific theorists do. Scientific theorists will sometimes
(provisionally) accept a candidate theory partly on the grounds that its set of
generalizations has the theoretical virtues to a sufficient degree, and partly
on those grounds—i.e., because that theory possesses the theoretical virtues
to a sufficient degree—will sometimes dismiss an apparent observation that
is inconsistent with that theory. This, for scientists, is the analogue of reject-
ing some verdictive claim that one finds intuitively plausible.7
There’s no doubt that we can appeal to the theoretical virtues as a way
of working from the top down, and I will argue later that we should do this
when the only alternative is working exclusively from the bottom up. But we
should not, I insist, adopt this method as our main way of working from the
top down. Even if we allow that appealing to theoretical virtues is a good way
to get at the scientific truth of the matter, which is an unpopular view,8 we’re
still a long way from having a strong argument for making the parallel move
in the moral case, because the case for appealing to theoretical virtues in sci-
ence is stronger than the case for appealing to theoretical virtues in morality.
This is true for two reasons, which I explore in the next two sub-sections.
5.1 Scientific Prediction and Moral Prediction9

One of the theoretical virtues is power, which can be understood as explana-
tory power or predictive power. Before assessing the case for appealing to
the virtue of power in the moral case, we must first get a grip on the distinc-
tion between moral explanation and moral prediction.
One famous view about the distinction between explanation and predic-
tion in the scientific case is that they are symmetric across the dimension of
time, such that explanation is something we do with regard to past events
and prediction is the same activity with regard to future events (Hempel and
Oppenheim 1948). It would be awkward, however, to distinguish moral
explanation and moral prediction in this way. This is because the moral
domain is atemporal in the relevant sense. Whatever resources we have for
arriving at correct moral beliefs regarding actions, those resources are avail-
able regardless of whether the action is in the past, present, or future.
Perhaps though there is a more fundamental distinction to be made between
explanation and prediction. The reason why scientific thinking about past
events is so noticeably different from scientific thinking about future events
isn’t the brute fact that the former events are in the past and the latter are in
the future. Rather, it is that scientific thinking about future events must pro-
ceed in the absence of the relevant authoritative source of knowledge, whereas
scientific thinking about past events need not. The most authoritative source
of knowledge about events is, in most cases, observation. When we think
about past events we can often rely on our observations of those events in
order to figure out whatever we’re trying to figure out about them. There-
fore, scientific thinking about past events—that is, scientific explanation—is
authoritative unto itself; in other words, its results can be authoritative in
the most robust sense. Obviously, this is not so with scientific thinking about
future events. In that case, we have to try to figure things out based on prelimi-
nary, or prima facie, sources of knowledge, and only later will we have observa-
tions and thus our authoritative source of knowledge (this is when confirmation
or disconfirmation happens).
It seems to me, then, that the most fundamental truth about prediction
is that it is a mode of inquiry that proceeds in the absence of the relevantly
authoritative source of knowledge about the matter being investigated. This
being the case, making moral predictions is pointless for those engaged in
moral theorizing. Here’s why:
Engaging in moral prediction making with regard to moral question X
would mean bringing certain deliberative resources to bear on X while ini-
tially leaving aside whatever authoritative resource(s) there might be with
respect to X. But why do this? In the scientific case we have little choice.
Unless we want to rest content having no beliefs about what will happen
in the future, we must forge ahead and do the best we can in forming such
beliefs even though we don’t yet have the relevant observations. But since
the moral domain is atemporal in the relevant sense, as mentioned above,
there is no need to divide up our deliberative resources like this, and it would
be silly to do so.10 If we want to think about whether φ-ing will be permis-
sible if someone does it at future time t, the most reasonable way to proceed
would be to immediately bring all our available deliberative resources to
bear on the question.
What I conclude from this is that moral prediction cannot be pulled apart
from moral explanation. Therefore, choosing between candidate normative
ethical theories based on their predictive power counts as a way of work-
ing from the top down if and only if choosing between candidate norma-
tive ethical theories based on their explanatory power counts as a way of
working from the top down, which it doesn’t. The explanatory power of a
scientific theory is its consistency with our observations, and therefore the
‘explanatory power of a normative ethical theory’ has to be understood as
its consistency with our verdictive intuitions (the relevant analogue of obser-
vations, as I already mentioned). Consequently, choosing between compet-
ing normative ethical theories based on their explanatory power counts as
working from the bottom up.
The absence of a distinction between explanation and prediction in moral
theorizing not only undermines the claim that appealing to the theoretical
virtue of predictive power is a way of working from the top down, it also
undermines the claim that appealing to the theoretical virtue of fecundity is
a way of working from the top down. The fecundity of a normative ethical
theory—its suggesting of new lines of inquiry—would count as a reason to
accept it only if those new lines of inquiry themselves yielded predictions
which could then be confirmed. But since any line of inquiry suggested by
a normative ethical theory would itself be an ethical line of inquiry, the
argument just given applies once again.
This leaves other theoretical virtues such virtues as simplicity, elegance,
and coherence.11 Importantly, the most popular arguments in philosophy
of science as to why we should appeal to these three and other remaining
explanatory virtues deploy as a key premise the claim that making such an
appeal has in the past led to the adoption of theories that wound up mak-
ing true predictions (Psillos 1999, 172; Tulodziecki 2013).12 This being the
case, appealing to simplicity, elegance, and coherence as a way of choosing
between rival scientific theories is a way of dealing with the temporality
of scientific investigation. The idea is that since, necessarily, we can never
know of any scientific theory how predictively powerful theory it is, and
yet we care how predictively powerful it is, we choose between scientific
theories based on their possession (or lack thereof) of these three theoretical
virtues as an indirect way of choosing between them based on their predic-
tive power. Again, however, this wouldn’t make sense in moral theorizing,
because moral theorizing is atemporal in the relevant sense. The predictive
power of a normative ethical theory is, as I’ve just explained, equivalent to
its explanatory power, and its explanatory power is always knowable.13
5.2 Explanatory Intuitions in Science and in Moral Philosophy

Here’s the second reason why the case for appealing to theoretical virtues
in scientific theorizing is stronger than the case for appealing to theoretical
virtues in ethical theorizing: Scientists have little choice but to appeal to
theoretical virtues if they want to work from the top down. There don’t
seem to be any other criteria to which they might appeal. Of particular
importance, for our purposes, is that it’s very unlikely that scientists will
have helpful explanatory intuitions to which to appeal. Two disanalogies
between the scientific case and the moral case are relevant here. First, in
the moral case what needs to be intuited is the existence of an explanatory
relationship, whereas in the scientific case sometimes what would need
to be intuited are not only an explanatory relationship but also the very
existence of one of the relata. In the moral case we already know that there
are such things as harm, deceit, failures of respect, failures to maximize the
good, etc., and we also know that there are such things as impermissibility,
obligatoriness, etc. We have the relata in hand and we just need to figure
out the explanatory relationships. By contrast, some explanatory claims in
science posit entities, such as sub-atomic particles, whose existence is at first
just as much of a challenge to figure out as their explanatory role. Second,
and more importantly, given how surprising and bizarre the natural world
has turned out to be, the more that’s been discovered about it, the idea of
relying on explanatory intuitions in science is laughable. We now know that
the natural world contains all sorts of entities and relationships that the
most creative mind couldn’t have dreamt up.
So I don’t think that moral philosophers should be too impressed by the
fact that scientists work from the top down by appealing to the theoretical
virtues. Scientists do this out of necessity, not because it is an ideal solution.
In the moral case we have recourse to a criterion, explanatory intuitiveness,
that can be reasonably treated as a guide to the truth (§3) and favors some
theories over others (§2). This constitutes a strong case for working from the
top down in that way to the extent that we can.
6. An Interesting Corollary
Before drawing this chapter to a close I want to point out that once we
acknowledge the points made above, we open up the possibility of resolving
a tension between the theory and the practice of moral reasoning.
Most moral philosophers believe that the method of reflective equilibrium
is the right method to use when trying to determine which normative ethical
theory to accept. This method involves first arriving at various considered
judgments, some verdictive and some general, and then making adjustments
to those judgments so as to make them consistent with each other. The offi-
cial dogma regarding the method is that whether an initial considered judg-
ment is verdictive or general should make no difference to one’s willingness
to adjust or discard that judgment.14 However, in practice this rule is fre-
quently violated, as anyone familiar with the literature in normative and
applied ethics will have noticed. Philosophers generally demonstrate a much
greater reluctance to give up on a verdictive judgment than to give up on a
general judgment. What’s going on here?
My explanation is twofold. First, most moral philosophers seem to believe
that the generalizations contained in normative ethical theories are to be
represented as categoricals or conditionals. If that were true, then it would
make perfect sense to be more reluctant, when deploying the method of
reflective equilibrium, to be biased in favor of giving up general judgments
over giving up verdictive judgments. As I said above (§4), there’s nothing
in a moral categorical or conditional to remain attached to once one has
searched for counterexamples to it. So if one discovers that a candidate
categorical or conditional faces several apparent counterexamples, then of
course one should give it up.15
Second, I suspect that many moral philosophers experience an ever-so-subtle
temptation to make their moral reasoning parallel to good scientific reasoning,
and that this is because they’ve failed to notice the disanalogies between
moral and scientific reasoning that I pointed out in §5.16 The generalizations
contained in scientific theories aren’t apt for being judged on intuitiveness
and often posit entities and processes whose existence cannot be confirmed
via observation at least for the time being, whereas their particular judgments
(judgments about natural events) are based on observation. Because of this, it
makes sense in scientific theorizing to hold one’s initial particular judgments
more fixed than one’s initial general judgments. By contrast, we can, as I
argued in §2, represent the generalizations contained in normative ethical
theories as explanatory claims, thereby making them apt for evaluation
based on their intuitiveness or lack thereof. When we combine this insight
with the uncontroversial claim that verdictive claims in ethics, like morally
explanatory claims, are apt for evaluation based on their intuitiveness or lack
thereof, we get to the conclusion that proper moral inquiry is epistemically
balanced where proper scientific inquiry is epistemically imbalanced: there is
no sense in which moral philosophers have better access to the particular facts
than to the theoretical facts, whereas for scientific theorists this certainly is
the case. Because of this balance in moral inquiry there is no reason, as there
is in the scientific case, to be more hesitant to give up one’s initial particular
judgments. Thus, acknowledging the potential role of explanatory claims in
moral reasoning reinforces the correctness of the official dogma of the method
of reflective equilibrium while also explaining why we are so often tempted
into violating it.17
Notes
1. This point has been made before. See Huemer 2005, §5.1; Hooker 2002; Audi
2015, 76.
2. McDermott (2008, 12), Gaut (1994, 33).
3. One might suggest that one could arrive at one’s initial set of moral beliefs by
accepting someone’s testimony. This is true but of course just pushes back the
question, forcing us to ask how the testimony provider arrived at her initial set
of moral beliefs. If we follow the chain backward far enough, we will arrive at
someone trusting a moral intuition.
4. A helpful analogy, suggested to me by an anonymous referee, would be beliefs
about which things are poisonous. True, we don’t have basic tendencies to
believe that certain things are poisonous and others aren’t, but this doesn’t
mean that natural selection has exerted no influence on such beliefs.
5. Both Broad and Ross used this expression.
I say that this ‘possibly’ was Ross’s view since another reasonable interpretation
of what he’s saying on the cited page is that, as a matter of anthropology,
judgments about conditionals originally were arrived at by intuitive induction
from verdictive judgments. This wouldn’t commit Ross to any particular fact
about how you and I, today, arrive at our judgments about conditionals.
6. One might respond that things look differently in cases in which I check whether
all As are Bs only to find that a few of them, intuitively, are not. In such a case
it would be reasonable for me to feel torn, and it might be posited that the only
way to account for this is by supposing that my intuition that all As are Bs
remains a reason to believe ‘all As are Bs’. By way of rejoinder, I would insist
that we have another way of accounting for the reason I have to believe ‘all As
are Bs’, namely all those cases in which my verdictive intuition, with respect to
some A, is that it is indeed a B and all the cases in which my verdictive intuition,
with respect to some non-B, is that it is indeed a non-A plus the recognition that
my intuitions regarding those As that appear to not be Bs and those non-Bs that
appear to be non-As could be mistaken.
7. This analogy between observations in science and verdictive intuitions in moral-
ity is endorsed by List and Valentini (2016) and McDermott (2008, 14).
8. The classic criticisms are to be found in van Fraassen (1980; 1989).
9. The argument that I make in this sub-section is based on an argument Ladyman
(2012, 42–6) makes against appealing to the virtue of power as a way of choos-
ing between competing a priori metaphysical theories.
10. One might think that in order to test a normative ethical theory we have to
engage in a certain amount of artificial dividing. Suppose there’s some case, X,
about which we are confident in our verdictive intuitions. We might build up a
normative ethical theory while setting aside the question of what to say about
X, and then use X as a test case to determine whether we’ve built a good theory.
One might worry that if we don’t set aside some cases in this manner then it’s
difficult to imagine how else we might test our theory.
The mistake in this line of reasoning is its distinguishing between the theory-
building stage and the theory-testing stage. Because of the temporality of sci-
ence (i.e., the fact that at different points in time we have different observations
with which to work) there is no choice but to make this distinction in scientific
theorizing. In moral theorizing, on the other hand, there need not be any distinc-
tion; every intuition in which we are confident can be both a theory-building
intuition and a theory-testing intuition. Surely it makes the most sense to use
them this way. If we are confident in an intuition, then why would we want to
tie one hand behind our backs by refusing to take it into account when building
our theory? And when we reflect on what we have accomplished thus far in our
theorizing project, obviously we should want to check whether our theory does
in fact manage to accommodate that intuition.
One might respond that although moral theorizing isn’t temporal in principle,
it’s temporal in practice since some of our intuitions don’t occur to us until after
we’ve constructed a candidate theory. By way of rejoinder I say fair enough, but
this doesn’t undermine the point I am making that there is never any need to
reason about some moral question in the absence of the relevantly authorita-
tive source of knowledge about it. Although I may never have entertained the
question of what the right thing to do in situation X is, once I do entertain that
question I do so while in possession of the relevantly authoritative source of
knowledge about it if there is one.
11. DeGrazia endorses the use, in ethical theorizing, of appeals to coherence (or
what DeGrazia calls “compatibility”) and simplicity (1996, 16–17). Hooker
(2000, 19–23), likewise, advocates the simplicity criterion. Nagel (1979, x), on
the other hand, rejects appeals to simplicity and elegance.
12. Huemer (2009, 218) makes note of and formalizes this kind of argument for
appealing to simplicity (parsimony) in science, though he goes on to construct
three other arguments for the same conclusion (219–25).
13. There is a second extant argument for appealing to simplicity in scientific
theorizing, namely Roger White’s (2005) argument. However, his argument is
clearly inapplicable to moral theorizing as it relies on an asymmetry that holds
in the scientific case but not, as I will show in §6, in the moral case: that the
specific facts are verifiable in a way that the theoretical facts are not.
14. As Daniels says, with respect to the set of more particular judgments and the set
of more general judgments over which one reasons in the method of reflective
equilibrium, “There is no set of judgments that is held more or less fixed” (1979,
266).
15. And in that case, one might as well not bother using one’s intuitions about those
categoricals/conditionals as inputs to reflective equilibrium. This thought may
be what G. A. Cohen (2008, 4) was thinking when he said that there is just one
set of inputs to the process—one’s verdictive intuitions.
16. Ronald Dworkin, for instance, assumes that if the method of reflective equilib-
rium is understood as a way of discovering moral facts (what Dworkin calls the
‘natural model’ of reflective equilibrium), then it will be construed as analogous
to scientific inquiry and will include a bias toward the verdictive (1989, 27–9).
17. Here’s a second interesting corollary to the argument of this chapter:
Joshua Greene (2013) has argued, famously, that empirical studies show that
characteristically act-consequentialist moral judgments—judgments based on a
concern to promote the overall good—are driven by controlled cognitive pro-
cesses whereas characteristically deontological moral judgments are driven by
automatic emotional responses, and has gone on to argue, further, that con-
trolled cognitive processes are more likely to guide us to the truth than automatic
emotional responses in many of the cases over which act-consequentialists and
deontologists disagree and that therefore characteristically act-consequentialist
moral judgment is more likely to yield the moral truth of the matter in those
crucial cases than is characteristically deontological moral judgment. (Peter
Singer (2005) endorses this line of argument.)
Given the argument of this chapter, however, it’s not clear that Greene is right
about what’s characteristic of act-consequentialist and deontological moral judg-
ment. I’ve argued that we should choose between competing normative ethical
theories based on their respective explanatory claims, so there’s a sense in which
what’s characteristic of a normative ethical theory is its explanatory claims. By
contrast, if all one is asked to do is what Greene asked the subjects of those
empirical studies to do, namely to reach a verdictive judgment, it is enough (and
in fact most convenient) to appeal to moral conditionals or categorical claims.
(Appealing to explanatory claims is useful only when theorizing.) So it may be
fair to say that the kind of judgment that the subjects of Greene’s studies engaged
in was not characteristic of any normative ethical theory.
(Still, however, Greene could claim that the moral conditionals/categorical
claims instructing us to promote the overall good typically appear in act-
consequentialist theories. This is true, but we should not overstate its impor-
tance. As I’ve argued elsewhere (Sachs 2010a) moral categoricals/conditionals
that are strongly associated with act-consequentialism can be accommodated
within non-act-consequentialism, and vice-versa.).

2 How Should We Choose Between
Competing Explanatory Stories?
1. Introduction
I argued in the previous chapter that moral theorizing makes sense only if it’s
construed as an explanatory project, which means we should judge candi-
date normative ethical theories in part on how good of an explanatory story
they tell, and make use of our explanatory intuitions in so judging. This
chapter tackles the issue of exactly how to do that. I expect most or all of
what I say here to be obvious, perhaps to the point of being boring, but it
constitutes necessary groundwork for the rest of the book, in which I show
that if we follow the strictures laid out in this chapter—that is, if we actu-
ally approach moral theorizing as an explanatory project—this will have
surprising, radical, and important implications for the conclusions we will
reach in normative and applied ethics.
In §2 I get specific about what kind of explanatory claims normative eth-
ical theories should be construed as containing. In §3 I discuss how we
should deal with the apparent existence of counterexamples to all seemingly
plausible moral generalizations, having agreed that those generalizations
should be construed as explanatory claims. In §4, I offer criteria for decid-
ing which explanatory claims to accept (and thus criteria for deciding which
normative ethical theory to accept). Finally, §5 concludes.
2. What Kind of Explaining Do Normative Ethical

Theories Attempt to Do?
2.1 Atomistic Explaining and Holistic Explaining

Thinking of the generalizations contained in normative ethical theories as
explanatory claims raises an immediate worry: Can’t the things that are sup-
posed to be explained, the verdictive facts, be explained without the use of
generalizations? If I want to explain why it was wrong of Jones to train her
dog to attack people, can’t I just offer the (non-general) claim that Jones’s
action created a serious danger to peoples’ physical safety? Given the avail-
ability of explanations of this sort, it may seem mysterious why we engage
18 How Should We Choose Between Competing Explanatory Stories?
in the project of generalization instead of simply generating a list of each
verdictive fact and its corresponding explanation.
My conjecture is that generalization becomes a desideratum for moral
thinking only once we notice certain patterns and regularities in the verdic-
tive part of the moral domain. We might notice, for instance, that actions that
harm innocent children are more frequently impermissible than actions that
benefit them. This gives us a sense that there is an order or a shape to the
moral domain, and consequently we set ourselves the task of constructing a
normative ethical theory that represents that order or shape.
Therefore, generalization is desirable in a moral theory because it is desired.
The inference from desiredness to desirability is legitimate in this context,
flawed though it may seem. The task here is to identify criteria for moral
theorizing, and there is no way to do this without taking up the perspec-
tive of the people engaged in the search for the best theory. (As I said in the
Introduction, I will not try to persuade the reader that moral theorizing is a
worthwhile project, though later in this chapter I will undermine one argu-
ment that devalues it.) To a certain extent, their intentions set the nature of
the activity. Some of their aims might be peripheral, such that we could rep-
resent the activity as not having that aim. But this is not one of those cases.
Moral theorizing would be unrecognizable if it didn’t involve generalization.
We wouldn’t say I was theorizing if all I did was take each purported verdic-
tive fact as an isolated case and attempt to identify an explanation for it.
This tells us that what I have thus far been labeling ‘the’ explanatory proj-
ect is actually two projects. First, we want to explain each verdictive fact.
I’ll call this the project of atomistic explanation. Second, we want to explain
the set of all verdictive facts. I’ll call this the project of holistic explanation.
Offering a holistic explanation requires generalizing, but generalizations
do not feature in atomistic explanations.1 Therefore, normative ethical
theories, composed as they are of generalizations and nothing else, do not
contain atomistic explanations. However, even though our topic is how to
choose between normative ethical theories, it will still be important for us
to understand the nature of atomistic explanations. This is because holis-
tic explanations—our primary interest—are composed of generalizations
about atomistic explanations, in a sense to be explained in §2.3. Therefore,
in the remainder of this section I undertake to identify the features of atom-
istic explanations.
2.2 What Are the Relata of Atomistic Explanations?

I have been and will continue to take it for granted that what is to be
explained in each case is an individual verdictive fact—the moral impermissi-
bility, obligatoriness, supererogatoriness, etc., of an action or narrow action-
type. The interesting question is what does the explaining. And it seems clear
that one thing many of us want out of a moral explanation is information
about the relationship between the verdictive and the non-normative. To see
How Should We Choose Between Competing Explanatory Stories? 19
why, consider a case in which we believe that I am morally required to φ.
No doubt it would be informative to be told that this is because someone
has a right that I φ, or to be told that this is because there is a very strong
moral reason in favor of my φ-ing. But it wouldn’t be informative enough.
We would then want to know why the right exists or why the moral reason
exists. My impression is that usually we’re not satisfied until the verdictive
fact has been related to the non-normative. We don’t merely want to be told
that an action that has a certain kind of to-be-doneness, like obligatoriness
or supererogatoriness, also has some other kind of to-be-doneness, like
being the object of a right or a moral reason. We want to know why it has
any to-be-doneness at all. Now the division between the normative and the
non-normative is controversial, and I don’t want to wade into the contro-
versy. I’ll commit myself only to the vague claim that the domain of the nor-
mative is the domain of to-be-doneness and to-be-avoidedness. At no point
in the rest of this book will it be important for us to have a firmer grasp on
the distinction.2
2.3 The Grounding Model of Atomistic Explanation

Having established this much we can finally, I submit, say something about
what makes for a good atomistic explanation. First, a good atomistic expla-
nation must meet a relevance constraint (Väyrynen 2009, 94–5). To see this,
consider a normative ethical theory on which each action that is obligatory
is obligatory because people approve of it and each action that is impermis-
sible is impermissible because people disapprove of it. The reason why we
don’t devote serious attention to this theory is that it has an irrelevance
problem. Most of us would say that what people approve or disapprove
of is in most cases irrelevant to the moral obligatoriness or impermissibil-
ity of the action in question. So the relevance constraint, as I understand
it, requires that the non-normative explanans must be relevant in a certain
(unspecified) way to the verdictive explanandum.
Identifying this first property of moral explanation helps us to identify
a second property of it. Consider again the approval/disapproval theory
of morality. One way to capture the root of its relevance problem is this:
for any impermissible action, φ, there is some fact about φ, the fact that
people disapprove of it, that is mentioned in the purported explanans but is
not explanatorily relevant. This tells us that relevance, or lack thereof, can
be a feature of facts, which in turn implies that the atomistic explanatory
relationship can be a relationship between the fact to be explained and the
fact or set of facts that does the explaining (which, in any event, would’ve
seemed a sensible thing to say).3
A third property of atomistic explanations is asymmetry (Väyrynen 2009,
95). If X morally explains Y then Y doesn’t morally explain X. So if the
fact that the death penalty involves killing explains its wrongness, then its
wrongness doesn’t explain that it involves killing.
We have thus far determined that the atomistic explanatory relationship
(1) is subject to a relevance constraint, (2) can holds between facts, and (3) is
asymmetric. What kind of explanation conforms to these three features? I
don’t think any kind of explanation with which scientists are concerned fits
the bill.4 But these three features are among the features that are supposed to
obtain in metaphysical grounding explanations. This is the kind of explana-
tion on offer when we say, for instance, that the fact that the glass is fragile
is grounded in the fact that it has the particular molecular structure it does
(Väyrynen 2013, 161–2). So without implying that the relation between a
verdictive fact and what explains it is precisely like the relation between
the fragility of the glass and its molecular structure, I will say that atomis-
tic explanations of verdictive moral facts are ‘grounding explanations’. A
‘grounding explanation’, stipulatively, is an explanation that conforms to
the three features just identified.
2.4 The Is-Ought Problem

There is a problem for the grounding model of atomistic moral explana-
tion.5 Grounding, in the moral case, can be understood as a relationship
between a verdictive fact and a non-normative fact/set of facts. Yet, some
(e.g., Maguire 2015) have thought it mysterious how a non-normative fact
or set of facts could ever be sufficient to ground any moral fact, of which
verdictive facts are one kind.6 This is a version of the “is/ought” concern that
Hume enunciated so forcefully, albeit a version that concerns grounding as
opposed to inference. So if verdictive facts are grounded and we are intent
on not committing the fallacy, we either have to say that no non-normative
fact or set of facts can on its own ground a moral fact or that grounds don’t
fully explain what they ground. Therefore, by proposing that the atomistic
explanatory project is the project of identifying, for each verdictive fact, the
non-normative fact or set of facts that explains it, it seems that in effect I
am proposing that the atomistic explanatory project is doomed to failure.
Like just about everyone else, I find myself at a loss as to how to get around
this worry. But if non-normative facts don’t on their own ground verdictive
facts, then at least for each verdictive fact there is some non-normative fact or
set of facts that plays an indispensible role in its grounding. So the atomistic
explanatory project would be the project of identifying those non-normative
facts. This still allows us to say that when we engage in atomistic explanation
we are in the business of identifying grounding relationships between non-
normative facts and verdictive facts, though we’ll have to allow that what we
are identifying may be relationships of partial grounding.7
2.5 Conclusion
We began this section already having reached the conclusion that the gen-
eralizations contained in normative ethical theories should be represented
as explanatory claims for the purpose of sifting through candidate theories.
The goal for this section was to identify as which kind of explanation they
should be construed. I began by arguing that they should be construed as
generalizations about atomistic explanations. The goal then became to illu-
minate the nature of atomistic explanation. I advocated a grounding model
of atomistic explanation and stipulated three features that an atomistic
explanation must contain.
3. Generalizations and Exceptions
3.1 Introducing Enablers and Disablers

In Chapter 1, I argued that when we’re sorting through candidate norma-
tive ethical theories we should represent the generalizations they contain as
explanatory claims, and then in §2 of this chapter I said, more specifically,
that they should be understood as generalizations about the existence of
atomistic explanatory relationships. It is worth asking whether going this
route is helpful in the effort to avoid one of the central problems of moral
theorizing: the problem of exceptions.
First, however, we need to be clear on what an exception is. Suppose that
Jones promises to Smith that she (Jones) will φ, and suppose that Jones is not
morally required to φ. There are two main reasons a situation like this might
arise, assuming that promise making is normally morally significant in some-
thing like the way most people think it is. First, it may be that this particular
promise is indeed morally significant in the usual way, but that there is some-
thing else of greater moral significance counting against Jones keeping her
promise. Second, it may be that this particular promise is not morally signifi-
cant in the way promises usually are. Only in the latter case can it really be
said that we have encountered an exception to the generalization connecting
promises to moral requirements, since in the former case the connection is
there but is simply overridden. I will focus, then, exclusively on the latter sort
of case, and will intend this sort of case by my use of the word ‘exception’.
It is often said that it is very hard to find non-trivial moral generalizations
that don’t seem to admit of exceptions. This isn’t quite precise enough, since
what would count as an exception to a generalization isn’t clear until it’s
been determined how that generalization is to be expressed. But it certainly
is true that it is very hard to find non-trivial moral conditionals or moral
categorical claims that don’t seem to admit of exceptions, under a standard
understanding of what moral conditionals and moral categorical claims are.
This, it seems to me, is what has moral philosophers worried, and what lies
at the heart of an influential argument for moral particularism (the argu-
ment in Dancy (2004)). And, indeed, there have been many attempts in the
last 15 years to propose new kinds of moral conditionals (e.g., Little 2001)
and moral categorical claims (e.g., Väyrynen 2009) that avoid the usual
problem of exceptions.
How does the problem of exceptions look if we interpret the moral gen-
eralizations normative ethical theories contain not as moral categoricals/
conditionals but as generalizations about atomistic explanatory relation-
ships? That depends on what we think a good atomistic explanation looks
like. Consider an example: Suppose I promise you that I’ll pick you up at the
airport next Wednesday. And let’s assume that I am now morally required to
do so. What should we say is the explanans for this explanandum? We could
say the explanans is: I promised you to do it, and you didn’t end up find-
ing a more convenient way of getting from the airport to your home, and
my car didn’t break down, and a tornado watch didn’t start just the time
when I would’ve needed to start driving to the airport, and . . . etc. In other
words, we could say that the explanans is every fact (except the explanan-
dum itself) such that if that fact hadn’t held then the explanandum wouldn’t
have held (setting aside any question of overriding). But if we employ this
strategy then the prospects for offering explanatory generalizations are dim.
This is because it seems that, for each moral requirement, there is a unique
set of facts such that if those facts hadn’t held then the moral requirement
wouldn’t have held. Even if this isn’t quite the truth—even if there are some
sets of moral requirements that are counterfactually dependent on the exact
same set of facts—we’re still in trouble as far as generalization goes. The
whole point of coming up with moral generalizations is, as I said earlier,
that they serve to capture the regularities and patterns we think we see in the
distribution of obligatoriness and impermissibility. To capture a regularity,
any generalization we include in our normative ethical theory is going to
have to be applicable to many verdictive facts, which means its explanans
will have to be a set of facts that is orders of magnitude smaller than the set
of all facts such that if any of those fact hadn’t held then some particular
verdictive explanandum wouldn’t have held.
What we have to say then is that some of these counterfactually necessary
facts do no explanatory work, and therefore we need a way of character-
izing what role they play. The only option, it seems, is to posit that they
determine whether other facts do explanatory work. Thus, we arrive at a
distinction, for which Dancy (2004, Chapter 3) has argued, between facts
that do explanatory work and facts that determine whether other facts do
explanatory work. Going forward, I’ll adopt Dancy’s terminology for facts
that determine whether other facts do explanatory work. The ones whose
presence is necessary for some fact to do explanatory work will be desig-
nated enablers of the explanatory ability of that fact, and the ones whose
absence is necessary will be designated disablers.
I’m going to wait until the outset of §4 to make a precise proposal as to
how we should construe the generalizations contained in normative ethical
theories, but the general idea is that they will be understood as claims to
the effect that X explains Y under certain circumstances having to do with
enablers and disablers. (To return to the example from earlier in this section,
I think a generalization that could plausibly be understood to explain why
I am morally required to pick you up at the airport this Wednesday is: That
an agent promised to φ explains, under certain circumstances having to do
with enablers and disablers, that agent’s being morally required to φ.) This
being the case, the problem of exceptions is already taken care of. Excep-
tions can simply be understood as cases in which a relevant enabler is absent
or a relevant disabler is present. Since the explanatory claims themselves are
going to be construed in such a way as to account for enablers and disablers,
what we have been calling exceptions aren’t actually going to be exceptions
to our explanatory claims.
3.2 A Worry About Context-Sensitivity

I have just argued that among the facts whose instantiation is counterfactu-
ally necessary for the instantiation of a verdictive fact we should make a
distinction between those that explain the verdictive fact and those whose
presence or absence merely enables the explanatory facts to do their explan-
atory work. One might worry, however, that this distinction is illusory, and
that what we really have here is a case of the context-sensitivity of good
explanation.
In philosophy of science the analogous worry about context-sensitivity
often gets treated as one aspect of a broader question about the aim of scien-
tific theorizing. Specifically, van Fraassen (1980) argues that the aim of sci-
ence is merely to produce empirically adequate theories—i.e., theories that
make true predictions—as opposed to uncovering truths about underlying
structures. If this pragmatic theory of scientific theorizing is accurate, then
worries about which fact ‘truly’ explains some datum are ill-conceived wastes
of time. However, we can safely discard the moral analogue of this worry—
i.e., the pragmatic theory of moral theorizing—because, as I argued in §5.1
of Chapter 1, it makes no sense, when theorizing about morality, to think of
normative ethical theories as issuing predictions.
The more relevant way to make out the context-sensitivity worry in the
moral domain is by constructing an analogue of van Fraassen’s (1980)
theory of scientific explanation. Van Fraassen argued that a scientific
explanation is an answer to a ‘why’ question and that ‘why’ questions are
implicitly contrastive—that when we ask why x is the case we are really ask-
ing why x-rather-than-y is the case. For instance, if someone asks why the
lion ate the gazelle, it would be appropriate to answer ‘because he was hun-
gry’ if what the questioner really wanted to know was why the lion ate the
gazelle as opposed to nothing; on the other hand, it would be appropriate to
answer ‘because there was no other prey to be found’ if what the questioner
wanted to know is why the lion ate-the-gazelle-as-opposed-to-something-else.
Similarly, if a questioner wants to know why Smith is obligated to pick up
Jones at the airport, the person arguing for the context-sensitivity of good
moral explanation might say: it’s appropriate to answer ‘because she prom-
ised to’ if the questioner wanted to know why Smith was obligated to pick
up Jones as opposed to all the other people who needed a ride from the
airport; whereas it’s appropriate to answer ‘because the weather is fine’ if
the questioner wanted to know why Smith was obligated to pick up Jones
from the airport this time as opposed to that other time (during the tornado
watch) when Smith wasn’t so obligated. What we see happening here is that
a fact—that the weather is fine—is an explainer of some other fact in one
context but not an explainer of that same fact in another context.
Although this worry about the distinction between facts that explainer on
the one hand and facts that enable/disable on the other is reasonable, I am
not going to respond to it.8 I argued in the previous sub-section that we have
no choice but to accept the distinction if we want to be able to formulate
moral generalizations that even attempt to do what moral generalizations
are supposed to do. My agenda here, as I said in the Introduction, is to offer
a picture of what shape moral theorizing should take on the assumption that
we should be engaging in moral theorizing.
For what it’s worth, however, what I’ve said so far in this chapter should,
I would think, take some of the wind out of the sails of those who are hos-
tile to the distinction on which I’m relying. Ever since Dancy laid out the
distinction, the impetus to reject it has sprung mainly from the widespread
desire among moral philosophers to rescue moral generalism from Dan-
cy’s argument for moral particularism—an argument that (on some ways
of making it out) rests on the distinction. However, if I’m right that when
theorizing we should represent the generalizations that normative ethical
theories contain as explanatory claims, then Dancy’s distinction is no threat
to moral generalism. To the contrary, Dancy’s distinction is necessary for the
viability of the generalist project—the project of moral theorizing. Everyone
in this debate—both Dancy and his opponents—has failed to see this on
account of the fact that they have all been laboring under the groundless and
mistaken assumption that the generalizations contained in normative ethical
theories should be made out as categoricals or conditionals.
4. How to Determine Which Normative Ethical Theories

Do Their Explanatory Job Well
I argued in Chapter 1 that normative ethical theories should be understood
as being composed of explanatory generalizations, and in §2 of this chapter
I clarified that we should understand those generalizations as being about
the existence of atomistic explanatory relationships. Now that we know
how to think of normative ethical theories when deliberating about them,
we can turn to the question of how to choose among them. In this section, I
propose a set of criteria for making that choice. The set is not intended to be
exhaustive; it is to be used alongside whatever the other legitimate criteria
are. My concern is simply to put forward the criteria that are appropriate
given the explanatory content that I have argued normative ethical theories
possess.
To begin with we need to distinguish two kinds of grounding claim and
the differing roles they play in explanation. As I said earlier, grounding is
a relationship between facts. Technically, therefore, we make a grounding
claim only when we claim that a token non-normative fact or a set of token
non-normative facts grounds a token verdictive fact. So we might say that
the fact that Jones’s action constituted a breaking of a promise grounded
its wrongness. Let’s call such claims token-grounding claims. Atomistic
grounding explanations take the form of token-grounding claims.
So when we generalize, we generalize about the existence of token-grounding
relationships; in other words, we claim that some tokens of non-normative
facts of kind N ground some tokens of verdictive facts of type M. Since
generalizations of this sort commit themselves to nothing regarding the con-
ditions under which a token of type N will actually ground a token of type
M, they should be understood as having an ‘under certain conditions’ clause
built into them. So the template for such a generalization is: Token non-
normative facts of type N ground token verdictive fact of type M under
certain conditions. The “certain conditions” qualifier allows for the possi-
bility that some enabler might be absent or some disabler present. Let’s call
generalizations that conform to this template type-grounding claims.
I argued in §2 that the generalizations contained in normative ethical the-
ories should be understood as purporting to holistically explain, as a unit,
the set of verdictive facts. We now have an expression for those generaliza-
tions: they are type-grounding claims.
Take Benthamite utilitarianism (BU) as an example. Suppose that Smith
kicks Jones, and that according to BU that action was impermissible. BU will
make this token-grounding claim, which I’ll call X1: an atomistic grounding
relationship holds between the non-normative fact that Smith’s action failed
to maximize utility and the verdictive fact that it was impermissible. BU
will also make other token-grounding claims of the same form: X2, X3, etc.
Additionally, BU will generalize from these token-grounding claims to the
following type-grounding claim, Y1: the fact that an action fails to maximize
utility atomistically grounds, under certain circumstances, the impermissi-
bility of that action. And Y1 will serve as BU’s holistic explanation of the set
of verdictive facts as a whole. (Things will work similarly for a pluralistic
theory, the difference being that a pluralistic theory will say that plural type-
grounding claims, Y1, Y2, Y3, etc., holistically explain the set of verdictive
facts as a whole.)
Having set out this framework, what remains to be done is to identify
criteria for accepting or rejecting candidate type-grounding claims.
4.1 Criteria for Accepting Token- and Type-Grounding Claims

Type-grounding claims both imply and are implied by token-grounding claims.
For instance, if I accept the type-grounding claim that the fact that an action
is a promise-breaking grounds the wrongness of that action under certain
conditions, then I must accept the token-grounding claim that the fact that
Jones’s action was a promise-breaking grounded its wrongness if I judge
that the relevant conditions are met in that case. Here I’ve moved from a
type-grounding claim to a token-grounding claim. And if I accept that the
fact that Jones’s action was a promise-breaking grounded its wrongness,
then I must accept that an action’s being a promise-breaking is a kind of
fact that grounds the wrongness of that action under certain circumstances.
In this way we can move from token-grounding claims to type-grounding
claims.
Given the existence of these inferential relationships, any criteria there
might be for accepting or rejecting token-grounding claims are going to be
very useful for sorting between candidate type-grounding claims. And there
is indeed at least one such criterion. Token-grounding claims can cohere or
fail to cohere with the verdictive claims one accepts. Token-grounding claims
imply verdictive claims, because grounding is factive.9 If the fact that Jones’s
action was a promise-breaking grounded its wrongness, then Jones’s action
was wrong.
I now want to add one final criterion that applies to both token- and
type-grounding claims: intuitiveness. I argued in §3 of Chapter 1 that we
should treat our explanatory intuitions as reasons for belief, but now I can
be more precise and decompose the class of “explanatory intuitions” into
two sub-classes: token-grounding intuitions and type-grounding intuitions.
Crucially, we actually do have both kinds of intuition. We have intuitions
about token-grounding claims;10 we find it intuitive, for instance, to say
that what explains the wrongness of the rapist’s action was the harm it
caused to the victim. Something similar, though not quite the same, is true of
type-grounding claims.11 Whereas type-grounding claims cannot themselves
be intuitive or counterintuitive, claims that bear an extremely close logical
relationship to type-grounding claims can be intuitive or counterintuitive,
which is just as helpful. Here is how I see it working: If I say that non-
normative facts of type X ground verdictive facts of type Y under certain
circumstances, one thing I have committed myself to (unless it is impossible
for the circumstances to be met) is the claim that it is possible for a non-
normative fact of type X to ground verdictive fact of type Y. And claims
of that sort can certainly be intuitive or counterintuitive. For instance, as
mentioned earlier, we probably find it rather counterintuitive to suppose
that the fact that people disapprove of an action can possibly explain the
impermissibility of that action.
In the foregoing discussion I moved freely between talk of ‘grounding’
and talk of ‘explaining’, thereby implicitly making the assumption that
having an intuition about a morally explanatory relationship just is hav-
ing an intuition about a moral grounding relationship. But one might hap-
pily allow that people have intuitions about explanatory relationships in
the moral domain while remaining skeptical of the idea that intuitions about
moral grounding relationships are widespread.12 A grounding relationship,
according to the analysis I offered in §2.3, is a relationship between a non-
normative fact and a verdictive fact such that the former is especially rel-
evant, in an asymmetrical way, to the latter. It is highly doubtful whether
people, if asked of two facts whether they bear this relationship to each
other, would often have any intuition about that.
I concede this, but insist nevertheless that we may rightly speak of peo-
ple having moral grounding intuitions. An analogy may help: If we went
around asking people whether courage is a virtue, whether greed is a virtue,
etc., surely we would be able to elicit intuitions about those matters. Those
intuitions would count as intuitions about virtue even if there is some true
analysis of virtue out there to be found (e.g., a virtue is a disposition that
constitutes a mean . . .) such that almost no one is familiar with it and no
one would have an intuition as to whether courage or virtue instantiates it.
It is enough if people have intuitions about the applicability of a concept,
virtue, to a range of character traits; the fact that most people lack any
particular analysis of that concept doesn’t undermine this fact. Similarly,
as long as people have intuitions that certain non-normative facts do or
don’t explain certain verdictive facts, we can say that people have explana-
tory intuitions. And if we help ourselves to the further claim that people
have grounding intuitions, we are not thereby claiming that people have
intuitions as to whether those cases instantiate the three features of my
particular analysis of explanation. Rather, it is a shorthand way of saying
that people have intuitions about the applicability of a certain concept,
moral explanation, and that the grounding analysis of that concept is the
correct one.
4.2 Criteria for Accepting Sets of Type-Grounding Claims

One of the things I did in the previous sub-section is to lay out criteria for
determining which individual type-grounding claims we should accept. I
now want to offer two criteria for determining which sets of type-grounding
claims we should accept, and thereby offer a standard for good holistic
explanation.
I mentioned in Chapter 1 that many philosophers of science advocate
averting to theoretical virtues as a way to sort between competing scientific
theories, and that we could, similarly, avert to such criteria to sort between
competing normative ethical theories. Now that we have distinguished
between atomistic explanation and holistic explanation, we can say, more
precisely, that theoretical virtues could be used to sort between competing
sets of type-grounding claims.
Of course, I argued in Chapter 1 that averting to theoretical virtues
should not be our main strategy, and furthermore I listed (in §4.1 of this
chapter) some other kinds of criteria that are available for sorting between
individual type-grounding claims, including, importantly appealing to our
intuitions about them. But there may be two sets of type-grounding claims
such that the members of one set do just as well on those criteria as do the
members of the other. Now one might suggest that in addition to the crite-
ria already suggested, we might also appeal to our intuitions about sets of
type-grounding claims. But I don’t think we have such intuitions over and
above the intuitions we have about individual members of such sets.13 So
we should revisit the proposal to appeal to theoretical virtues as a way of
sorting between competing sets of type-grounding claims that are otherwise
equally acceptable.
I think we should accept simplicity as a criterion for breaking ties between
competing sets of type-grounding claims. If two sets of type-grounding
claims are equally acceptable given the criteria laid out in §4.1 and any
other relevant criteria, then we have no choice but to appeal to theoretical
virtues. And that’s it—that’s my argument for appealing to them. The only
other thing to be said to motivate this proposal is to show why the pur-
ported ‘virtue’ of simplicity really is a virtue, since I already argued (§5 of
Chapter 1) that the most popular reason to count simplicity as a virtue in
the scientific case doesn’t transfer to the moral case.
In the scientific case one aspect of simplicity is number of fundamental
principles. Similarly, in the moral case I propose that a normative ethical
theory is simple to the extent that its set of type-grounding claims has fewer
members.
What’s the value of simplicity so understood? Although this isn’t the main
argument for appealing to simplicity in the scientific case, there is the thought
that one achieves a better understanding of an event insofar as one is able
to chalk it up to causes that are at work over a wide range of cases. And,
conversely, there is the thought that one understands a causal force better
insofar as one is able to see it at work over a wide range of cases. The general
idea seems to be that one understands a certain natural domain better to the
extent that one can unify that the events within that domain under a relatively
limited set of fundamental forces.14 Similarly, I hold that one understands the
moral domain better to the extent that one can unify moral phenomena under
a relatively limited set of non-normative grounds.15 What else could count as
gaining a better understanding of the moral domain if not coming to see the
various moral phenomena as resulting from a limited set of non-normative
phenomena? This is why simplicity is a theoretical virtue.
5. Conclusion
In this chapter I argued for various claims about how we should determine
which grounding explanations and generalizations about them to accept.
I want to show that if we take these lessons seriously when we engage in
moral theorizing, this will have implications for which conclusions we reach.
In the rest of this book I illustrate this point by doing some moral theorizing,
using the method of reflective equilibrium.
I will come to some very surprising conclusions—conclusions that we
don’t see other moral philosophers arriving at despite their using the same
method. I want to be clear, straight off, as to why this is going to happen. It
will happen because, in accordance with the argument I gave in Chapter 1,
the general judgments that I will use as inputs to the method of reflective
equilibrium will be explanatory judgments (which I argued in §2 of this
chapter are judgments about grounding relationships) as opposed to moral
categoricals or conditionals, which is what most moral philosophers use. On
the other hand, my arrival at very non-standard conclusions is not going to
be the result of my initial set of explanatory judgments being idiosyncratic.
My view is that we theorists should be using the method of reflective
equilibrium collectively as opposed to independently. Here I think the anal-
ogy with science is apt. Scientific theorizing is treated as a collective effort;
this is because we don’t think any scientist has special access to the scien-
tific facts, and so we consider it legitimate for the scientific community at
large to sit in judgment of the claims that each individual scientist makes.
Now although moral epistemology is probably less well understood than
scientific epistemology, I think it’s pretty safe to say that nobody has special
access to the moral facts (Driver 2006; Sliwa 2012; McGrath 2008, 2011);16
though there may be certain individuals, such as children and some mentally
ill people, who we can safely say have inferior access. So, analogously, I con-
sider it legitimate for the moral community at large to sit in judgment of the
claims that each individual moral theorist makes. Given this, it makes more
sense for us all to work together from the start, as scientists generally do,
than for each of us to work in isolation and then engage in critical scrutiny
of each other’s theories only after they’ve emerged fully formed. And part of
working together from the start, in the moral case, is pooling our intuitions:
We should be using the method of reflective equilibrium as an effort to build
a view that does as good a job as possible of accommodating the intuitions
that are widely held. Consequently, the explanatory intuitions from which I
start my arguments in subsequent chapters will be ones that, to my under-
standing, are widely held.17
Notes
1. It is important to see that the project of holistic explanation does not reduce to
the project of atomistic explanation. In this case an analogy with science should
be helpful. An atomistic explanatory project in science would be to explain, for
each case of a window shattering, why it did. But one also might try to explain
the patterns and regularities we observe over the domain of window-shattering
and non-shattering events; this would be a holistic explanatory project. Just as
identifying the cause of each window-shattering does not amount to explaining
why window-shatterings distribute the way they do, identifying (for instance)
the explanation of the wrongness of each case of innocent child-harming does
not amount to explaining why verdictive facts about child-harming actions dis-
tribute the way they do. Likewise, atomistic explanation does not reduce to
holistic explanation. Again, the scientific analogy is apt. I might know all the
true scientific generalizations, but knowing why some particular thing happened
involves (at least) knowing which generalization it falls under. In the moral case
I might have the correct set of explanatory generalizations, but knowing why
some particular action is (say) impermissible involves (at least) knowing which
member of the set is applicable in the given case.
2. Since the only atomistic explanatory claims in which I’m interested are the ones
that offer a non-normative explanans, I am correspondingly interested only in
the generalizations that generalize about that kind of atomistic explanation and,
likewise, interested only in normative ethical theories composed of generaliza-
tions of that sort. Thus, while I opened this chapter saying that I wanted to
establish how we should sort between candidate normative ethical theories, I
should now say more specifically that I want to establish how we should sort
between candidate normative ethical theories that purport to take us from the
non-normative to the verdictive.
Fortunately, there are plenty of normative ethical theories that purport to do
that; in fact the ones that don’t are in the minority. For example, Benthamite
utilitarianism relates non-normative facts to verdictive facts directly through its
generalization that the fact that an action maximizes utility explains its being
morally permissible (and obligatory). Rossian pluralism, meanwhile, relates
non-normative facts to verdictive facts indirectly through its generalizations
regarding certain action-types constituting pro tanto duties and its further gen-
eralization to the effect that some action’s being supported by the balance of
pro tanto duties explains that action’s obligatoriness. (For ease of discussion I
will use theories of the Benthamite sort—theories that make their connection
directly—as my case studies for the rest of this chapter, but everything I say
about them will hold as well for theories of the other sort.)
3. Having said that, I can now elucidate my earlier claim that holistic explanations
are composed of generalizations about atomistic explanation. In such a general-
ization what is asserted is that facts of a certain kind bear atomistic explanatory
relationships to facts of another kind (e.g., in general the fact that an action is
a killing explains the moral impermissibility of that action). I will give a more
precise template for such claims at the beginning of §4.
4. The candidates would be constitutive explanation, causal explanation, and nomic
explanation.
Constitutive explanation is a relationship between properties or between objects,
and so fails to meet the second criterion.
And we know, independently of anything I’ve said here, that in the moral
domain we’re not looking for causal explanations; we know this because if X
causally explains Y then there must be a time lag between X and Y, and in the
moral domain there is no time lag between an action’s being (say) impermissible
and its having whatever feature explains its impermissibility. (Thanks to Kath-
erine Hawley for pointing this out to me.)
As to nomic explanation, here’s the disanalogy: In moral explanation, the
explanans and explanandum are facts about the same thing. So, for instance,
the harmfulness of spreading rumors about someone grounds the wrongness of
spreading rumors about someone. This ‘same-thing’ constraint can be added to
the second feature of atomistic explanations mentioned in the main text. With this
feature added in, nomic explanations no longer have the second feature. The law
that explains why the water froze has as its explanandum a feature of the water
(that it froze) and as part of its explanans a feature of the air (its temperature).
5. A second problem, which I will not discuss here, is whether my claim that accept-
ing a normative ethical theory means accepting the existence of grounding rela-
tionships between non-normative facts and verdictive facts could possibly be
acceptable to a moral anti-realist. Suffice it to say I do not intend to rule out
anti-realism and that if there are quasi-realist positions that can allow for talk
of moral facts and the non-normative facts that ground them, then I’d be happy
to have such quasi-realists on my side.
6. Stratton-Lake (2011, 367) believes this is not problematic at all, but I’ll concede
for the sake of argument that it is.
7. For more on how partial grounding might work, see Leary (2017).
8. Raz (2000), Crisp (2000), Strandberg (2008), and Fogal (2016) each argue for
the validity of this worry. For responses, see Schroeder (2007, Chapter 2) and
Scanlon (2014, Chapter 5), both of whom accept the distinction; see also Bader
(2016).
9. To me this seems obvious, and I’ve never seen it disputed. Maguire (2015, 196,
203) assumes it’s true.
10. To my knowledge, this has never before been pointed out in the literature, obvi-
ous though it is.
11. Kagan (1988, 14–15); Parfit (2011, 415). Something close to this claim is made
by Hooker (2002). He says we have intuitions about relevance; that is, about
what kinds of fact about actions count morally in favor of or against our doing
them.
12. Thanks to an anonymous editor for alerting me to this objection.
13. Take W. D. Ross’s theory as an example. He held (or we can construe him as
having held), that five kinds of non-normative facts about an action can make
that action obligatory: facts about fidelity, righting previous wrongs, gratitude,
beneficence, and non-maleficence. For each such fact we can appeal to our intu-
ition as to whether it really can do that; one can generate an intuition about
whether an action can be obligatory in virtue of being what gratitude requires.
But I don’t think one would be able to generate an intuition about whether all
obligatoriness facts are accounted for by this five-item list.
14. For the sake of simplicity, my discussion of scientific explanation is meant to be
limited to the natural sciences, hence allusion above to the ‘natural domain’.
15. Mark Schroeder says this about reasons (Schroeder 2007, 2). See also Sigd-
wick’s talk of the philosopher seeking “unity of principle” (Sidgwick 1981, 6).
In addition, Rawls (1999a, 144) seems to endorse a simplicity criterion, as do
List and Valentini (2016).
16. Or, more carefully, we have no way of knowing of any particular individual that
she does.
17 Specifically, widely held among the people with which whose intuitions I am
familiar. Unfortunately, this is a tiny sliver of humanity.
3 Against Monism
1. Introduction
In this chapter and the next two I investigate the plausibility of various nor-
mative ethical theories. For organizational purposes I divide the discussion
in two: this chapter deals with monistic theories, while the next two concern
pluralistic theories.
Specifically, in this chapter I discuss consequentialism, Kantianism, and
contractarianism/contractualism, all of which are generally categorized as
instances of moral monism on account of the fact that they seem to be most
naturally representable as containing just one moral generalization. Since I
argued in Chapter 2 that we should represent the generalizations contained
in normative ethical theories as type-grounding claims connecting non-
normative facts to verdictive facts, I will represent each of these theories as
containing nothing more than one such type-grounding claim.
To depict act-consequentialism this way we have to classify evaluative facts as
non-normative facts. Only by doing this can we represent act-consequentialism
as being committed to the existence of just one type-grounding relationship
between verdictive facts and non-normative facts—namely the relationship
such that all the verdictive facts are grounded in facts about maximizing the
good.
I go in for this classification on the principle of charity, because all else
being equal, moral monism is superior to moral pluralism. In §4 of Chapter 2
I claimed that all else being equal a normative ethical theory is better to the
extent that it is simpler, and to that end the fewer non-normative facts one’s
theory countenances as grounding the various verdictive facts, the simpler it
is. Consequently when it comes to the simplicity criterion, consequentialism,
since it is a version of moral monism, can’t be beat.
But I will argue in this chapter that all else is not equal. If in order to be
monists we have to accept a type-grounding claim that is itself counterin-
tuitive or yields counterintuitive token-grounding claims or fails to deliver a
great many of the verdicts we find intuitively appealing, then we should con-
sider not being monists. This chapter will focus exclusively on the criterion
of token-grounding claims. The question will be whether the three candidate
versions of moral monism can deliver a plausible set of token-grounding
Against Monism 33
claims. I will begin by arguing that not one of them can do this. From these
three examples I will generalize to an overall conclusion to the effect that the
prospects are dim for discovering a version of moral monism that offers good
token-grounding explanations. This is an important critique, because each
normative ethical theory is nothing more than a set of explanatory claims.
2. What Do Monistic Theories Really Say About

How the Verdictive Facts Are Grounded?
If we’re persuaded that we should represent normative ethical theories, for
the purpose of theorizing, as being composed of type-grounding claims, then
there’s a quite natural suggestion, for each of consequentialism, Kantianism,
and contractarianism/contractualism, as to how they should be represented.
The suggestion, for each theory, is that the key non-normative fact around
which it is formulated is supposed to be the non-normative fact that grounds
all the verdictive facts. I represent the three monistic theories this way in the
middle column of Table 3.1 (taking act-consequentialism as a stand-in for
consequentialism per se).
Table 3.1 Standard and Alternative Ways of Representing Monistic Theories
Natural Way to Alternative Way to Represent the

Represent the Theory, Theory, on which the Key Non-
on which the Key Normative Fact Identifies the
Non-Normative Circumstances Under Which, and
Fact is a Ground of the Explanation as to Why, Other
Verdictive Facts Non-Normative Facts Ground the
Verdictive Facts
Act- All verdictive factsNon-normative facts X, Y, and Z

consequentialism are grounded in each can ground the fact that some
facts about the action is/isn’t permissible, and in fact
maximization of the do ground that fact just when, and
good. because, that action does/doesn’t
maximize the good.1
Kantianism All verdictive facts Non-normative facts X, Y, and Z
are grounded in each can ground the fact that some
facts about obeying action is/isn’t permissible, and in
or not obeying fact do ground that fact just when,
the Categorical and because, that action does/
Imperative. doesn’t accord with the Categorical
Imperative.2
Contractarianism/ All verdictive facts Non-normative facts X, Y, and Z
Contractualism are grounded in facts each can ground the fact that some
about what could action is/isn’t permissible, and in fact
be agreed to under do ground that fact just when, and
certain circumstances. because, that action is/isn’t in accord
with what could be agreed to under
certain circumstances.
34 Against Monism
However, this interpretation of these theories can be and has been challenged.
The general idea is to reinterpret monistic theories as identifying a key non-
normative fact whose presence or absence explains why other non-normative
facts can do their verdict-grounding work. Each of the three theories, inter-
preted along these lines, is given in the right-hand column of Table 3.1.
So as not to delay any further my discussion of the most popular forms of
moral monism, I relegate to Appendix A a discussion of whether it’s plau-
sible to reinterpret them in this way. My view is that it is not, and so in this
chapter I discuss those views on the more natural interpretation of them.
3. Consequentialism
There is an objection to consequentialism that’s been floating about for a
while: that sometimes consequentialism gets the right answer for the wrong
reason. The idea is that in some of the cases in which consequentialism
delivers the correct verdict about a case, its moral explanation—that is, its
explanation of why that verdict holds—is defective. The objection was origi-
nally directed specifically against utilitarianism; Bernard Williams’s (1973,
117) critique of how utilitarianism handles the Jim and the Indians case
might be the first instance of it, depending on how that critique is to be
understood. And it is often said (e.g., Arneson 2000a, 239; Hinman 2008,
227) that Rawls criticized utilitarianism this way. In any event, there are
several casual mentions of this objection, some of them directed narrowly
at utilitarianism and others more broadly at consequentialism, to be found
in the literature.3 On the other hand, there are just two philosophers, as
far as I can tell, who make the objection central to their case for rejecting
consequentialism.4
I aim to rectify this situation by laying out and defending a detailed ver-
sion of the explanatory objection to consequentialism. §3.1 describes what I
take to be the moral explanation template that consequentialism must offer.
In §3.2 I lay out the first objection to the token-grounding explanations
that follow from this template, which is that they are incompatible with the
directedness of wrongdoing—the idea that sometimes we wrong each other.
Then §3.3 contains my second objection to these explanations, which is
simply that they are counterintuitive.
3.1 Consequentialism’s Moral Explanations

As I argued in Chapter 2, normative ethical theories should be represented
as sets of type-grounding claims. For any monistic theory, of course, it will
be a set of one. So our first order of business is to determine what single
type-grounding claim consequentialism makes. Unfortunately, there would
be disagreement among consequentialists regarding precisely which type-
grounding claim to adopt. As I see it, the best strategy for coping with this
fact is to ascribe to consequentialism a type-grounding claim specific enough
Against Monism 35
to be paradigmatically consequentialist but vague enough to allow for most
of the details that particular consequentialists would want to insert.
A paradigmatic consequentialist type-grounding claim would ground
the verdictive facts about actions in facts about goodness. For act-
consequentialism, it would be specifically the goodness of all the conse-
quences of the action, while for rule-consequentialism it would be the
goodness of all the consequences of universal acceptance of or adherence
to various possible sets of rules.5 Many consequentialists have been happy
to include among the relevant consequences the fact of the action’s being
performed, so we should include that too. This means, for instance, that if
there is intrinsic badness to the telling of a lie, then when determining the
goodness of the consequences of an act of lying or of a set of rules permit-
ting lying we would have to include that disvalue.
To hone in further, a consequentialist type-grounding claim would ground
the verdictive fact about an action in the amount of goodness it yields.
Finally, it is a hallmark of consequentialism that it is comparative. For act-
consequentialism this is true in that what grounds the verdictive fact about an
action is the relative goodness of that action’s consequences compared to the
alternative actions. For rule-consequentialism this is true in that what grounds
the verdictive fact about an action is the relative goodness of the consequences
of universal acceptance of or adherence to various possible sets of rules.
There are dozens of other questions one would have to answer in order
to produce something specific enough to actually count as a type-grounding
claim. But if we began to offer answers to these questions, then we would
end up producing a type-grounding claim to which some consequentialists
would refuse to sign on. Since the goal here is to critique consequentialism’s
explanatory story per se, we ought to avoid that level of detail. Thus, I sub-
mit that our target should be the type-grounding claim at which we have
already arrived:
Consequentialism’s Type-Grounding Claim (CTGC): The fact that an
action is impermissible can be grounded, under certain circumstances, by
the fact that
a) the goodness of all the consequences of that action is less that the good-
ness of all the consequences of some alternative action, or
b) that action would be prohibited by the set of rules such that the good-
ness of all the consequences of universal acceptance or adherence to
that set of rules is better than the goodness of all the consequences of
universal acceptance or adherence to any other set of rules.
Just about all self-described consequentialists accept some version of this.

Act-consequentialists accept some specification of version a, while rule-
consequentialists accept some specification of version b.
Now that we have a general idea of how the set of verdictive facts is to be
explained given consequentialism, we can move on to the first objection to it.
36 Against Monism
3.2 The Directedness Objection
3.2.1 The Objection

My first objection to consequentialism is that its type-grounding claim,
CTGC, implies token-grounding claims that cannot make sense of the
directedness of some instances of wrongdoing. I’m relying here on the intui-
tive idea that some cases of wrongdoing have a direction or a vector in the
sense that they are something one person does to another. More formally,
for an instance of wrongdoing to be directed is for it to be the case that the
agent, in virtue of engaging in that act of wrongdoing, wrongs someone.
What I am contending, therefore, is that sometimes people wrong others
and in virtue of the fact that consequentialism contains CTGC as its sole
type-grounding claim, its token-grounding claims cannot allow for this.
One might ask what it is to wrong someone. My view is that “wrong-
ing” is a primitive moral concept, which is to say that it is inexplicable in
more basic moral terms. However, it is possible to illuminate the notion of
wronging someone by situating it in relation to other moral categories with
which we are familiar. Drawing on the work of R. Jay Wallace (2013), Ste-
phen Darwall (2006), David Owens (2012, Chapter 2), and Nicolas Cornell
(2015, 110–11), I suggest two such categories.
a) Obligation: Wronging bears a close connection to the idea of obliga-

tion. The zone of normative concerns that deals with obligation can
be referred to as relational (Wallace) normativity or the normativity
of accountability (Darwall). To wrong someone is always to violate an
obligation to her, and wronging a person means doing or failing to do
something to/for her that you were accountable to her for doing/not
doing. A failure of this sort is a failure to act on a particular kind of rea-
son, namely a deontic (Wallace) or second-personal (Darwall) reason.
b) Reactive attitudes: When A wrongs B, B is in a privileged position to
complain or object to A’s choice. Moreover, A’s wronging B puts B in
a privileged position to demand an apology from A. Finally, A’s action
gives B a special reason to reassess her attitudes toward A—a reason A’s
action does not give third parties. Particularly, A’s wronging B makes it
appropriate for A to resent and later (if amends are made) to forgive B.
Wallace and Darwall each argue that consequentialism cannot account for
the just-mentioned kinds of normativity and reactive attitudes. I believe that
this is true, and that correspondingly it is true that consequentialism cannot
account for wronging.6
Identifying the close links between the concept of wronging, the concept
of obligation, and the appropriateness of various reactive attitudes helps to
reinforce how important it is that a normative ethical theory allow that some
Against Monism 37
instances of wrongdoing are instances of wronging. A failure to do this is
also a failure to provide a place in one’s normative ethical theory for these
closely connected notions. Surely we won’t be satisfied with a theory that, for
instance, cannot make sense of one individual being accountable to another
or of one individual being appropriately resentful at the conduct of another.7
My criticism, that consequentialism cannot account for wronging, is based
on the observation that the token-grounding claims implied by CTGC do
not identify a wrong-sufferer. A sample token-grounding claim derived from
CTGC would be (for act-consequentialism) ‘Jones’s φ-ing was impermissible
because the goodness of all its consequences was less than the goodness of
all the consequences of some other action available to her’ or (for rule-conse-
quentialism) ‘Jones’s φ-ing was impermissible because it would be prohibited
by the set of rules such that the goodness of all the consequences of universal
acceptance or adherence to that set of rules is better than the goodness of all
the consequences of universal acceptance or adherence to any other set of
rules.’ Since wronging is a relationship between a wrongdoer and a wrong-
sufferer, those token-grounding claims do not account for wronging.
Not only does this fact about CTGC constitute a strong reason to reject
consequentialism, it also undermines what has typically been perceived as a
strong reason in favor of consequentialism—namely, that it delivers the right
result in the Peter Singer’s Pond Case and others like it in which one can,
at little cost to oneself, prevent a great tragedy befalling another individual.
The usual intuitive response to the Pond Case isn’t simply that one is mor-
ally obligated to save the child, but also that one would wrong the child
by failing to do so. But consequentialism cannot allow for this, because its
token-grounding explanation of the wrongness of failing to save the child
in the Pond Case will simply point out an impersonal fact about the case.
3.2.2 A First Response to the Objection

A consequentialist’s first step by way of undermining this objection might
be to stipulate that the kind of consequentialism she is concerned to defend
is person-affecting consequentialism—the version on which everything that
is good is good for someone. In addition, she might insist that each indi-
vidual’s personal good be conceived of as non-fungible, which would allow
us to say that oftentimes something irreplaceable is lost even when one does
the right thing.8
The second step would be to propose that the impersonal facts about
actions that are mentioned in the CTGC-derived type-grounding claims are
grounded in these personal facts. So, for instance, the fact that the goodness
of all the consequences of Jones’s φ-ing was less than the goodness of all
the consequences of some other action available to her would be grounded
in, among other facts, the fact that Jones’s φ-ing harmed Smith. And the
third step would be to propose that the grounding relationship is transitive,9
38 Against Monism
which would get the consequentialist conclusion that the wrongness of an
action is, contrary to what my objection said, grounded in personal facts.
However, the success of this response would not on its own undermine
the objection, since any CTGC-implied token-grounding claim will still fail
to identify a wrong-sufferer. It will identify individuals, no doubt, but will
unfortunately identify lots of them—all individuals affected by the action—
and will therefore not pick out any one (or ones) of them as having a special
status. The consequentialist might be tempted to bite the bullet and allow
that everyone mentioned in a token-grounding explanation derived from
CTGC is a wrong-sufferer, but this temptation will surely melt away once
we remind the consequentialist that such explanations will mention not only
those harmed by the wrongful action, but those benefited (if any) and the
wrongdoer herself. Surely the beneficiaries and the wrongdoer do not count
as wrong-sufferers.
But perhaps this is too quick. What if wrongdoing is properly con-
ceived of as an offense against the moral community? This would be the
moral analogue of the criminal law, where offenders are prosecuted on
behalf of the public at large. If wrongdoing is an offense against the moral
community, then we might plausibly say that every instance of wrongdo-
ing makes a wrong-sufferer out of the moral community, which includes
those who benefited from the wrongdoing and perhaps even the wrong-
doer herself.
Whether this move will do the trick depends on whether the consequen-
tialist, in saying that a wrongful act wrongs the moral community, is using
the verb ‘to wrong’ in the standard way, or whether instead she is using
that label to name something else. To make this determination we need to
appeal to the other moral concepts that we used to help us to get a grip on
wronging.
We first appealed to a set of concepts related to the notion of obligation.
Given the relation wronging bears to those other normative concepts, if
the consequentialist is right to say that a wrongful act wrongs the moral
community, then it should be the case that, e.g., each act of wrongdoing is
the violation of an obligation to the moral community, that we are account-
able to the moral community for each act of wrongdoing, that each act of
wrongdoing is the violation of a second-personal reason the agent had with
respect to the moral community, etc. But this certainly seems not to be the
case. Some acts that everyone, including consequentialists, will acknowledge
to be wrongful seem distinctly private—rape, for instance.
Second, we appealed to a set of reactive attitudes that are typically licensed
by wronging. So given the connection between wronging and reactive
attitudes, if the consequentialist is right to say that a wrongful act wrongs
the moral community, then it should be the case that, e.g., the moral com-
munity is in a privileged position to complain or object to each wrongful
act, the moral community is in a privileged position to demand an apol-
ogy from the wrongdoer for each wrongful act, the moral community may
Against Monism 39
appropriately feel resentment toward and (if amends are made) later forgive
each wrongdoer, etc. But this seems false. Take a case of theft, where a bur-
glar breaks into B’s home and takes B’s wallet. Intuitively, we would want
to apply the ‘wrong-sufferer’ label to B as a sort of honorific—as a way of
picking out someone whose complaints and demands deserve special atten-
tion. But we couldn’t do this if the moral community was wronged by that
act of burglary. Similarly, we would happily admit that the moral commu-
nity might reasonably experience some impersonal reactive attitudes, such
as indignation, toward the thief in light of her behavior. But resentment and
eventual forgiveness would be entirely out of line for anyone but B.10
Thus far, I have been granting for the sake of argument that wrongdoing
may properly be considered an offense against the moral community. How-
ever, a final problem for the consequentialist, insofar as she deploys this
idea as a way of showing that CTGC can identify a wrong-sufferer, is that
CTGC is incompatible with the idea that wrongdoing is an offense against
the moral community. In order for wrongdoing to be an offense against the
moral community, there would have to be a necessary connection between
moral requirements and what the moral community can demand. But there
is nothing in the consequentialist idea that suggests such a connection.
The best strategy for the consequentialist, in response to this, is to portray
consequentialism as being implied by some other ethical theory in which there
is a necessary connection between moral requirements and what the moral
community can demand. That other ethical theory might, for instance, be a
version of contractualism on which moral requirements are the rules of behav-
ior to which the members of a moral community could agree to hold them-
selves. Given this version of contractualism, it would make sense to say that a
failure to do what morality requires is an offense against the moral community
Darwall has recently defended a contractualist view like this and further-
more pointed out that the members of the moral community might endorse
as binding the set of rules whose general acceptance or adherence would
maximize the good. In other words, contractualism might imply a rule-
consequentialist criterion of permissibility.11 Derek Parfit (2011, Chapter 16)
has argued recently that this is in fact the case.
The problem with this strategy, however, is that it provides no defense of
rule-consequentialism. What the work of Darwall and Parfit shows is that
consequentialist criteria of permissibility are separable in principle from con-
sequentialist explanatory claims.12 Therefore, demonstrating the compatibil-
ity of the rule-consequentialist criterion of permissibility with the idea that
wrongdoing is an offense against the moral community is insufficient to estab-
lish that rule-consequentialism can explain how wrongdoing is an offense
against the moral community. Darwall and Parfit show us how the former
demonstration might go: We begin with an idea of wrongdoing as an offense
against a moral community and work our way up to a non-consequentialist
explanatory claim—something along the lines of an action is wrong in virtue
of violating a set of rules that all members of the moral community could agree
40 Against Monism
to—from which we then derive a rule-consequentialist criterion of permissibil-
ity. But their work does not show, nor can I imagine how it might be shown,
that we can work in the other direction, starting from a rule-consequential-
ist explanatory claim and arriving at the idea of wrongdoing as an offense
against a moral community.
3.2.3 A Second Response to the Objection

The first response to my directedness objection to consequentialism focused
on the possibility of using rule-consequentialism to argue that the moral
community is a wrong-sufferer with respect to each wrongful act. This
response got hung up on the apparent fact that the various reactive attitudes
and features of obligation that are closely associated with wronging simply
don’t seem to hold with respect to the moral community for each wrongful
act. The second response takes a more direct route, in a way: It proposes
that the various reactive attitudes and features of obligation that are closely
associated with wronging would be built into the optimal set of rules itself,
such that that set of rules would pick out one particular individual as the
wrong-sufferer for each wrongful act.
But this isn’t going to work, because a set of rules cannot license reactive
attitudes in the right way. What a set of rules can do, is to pick out, for each
instance of rule-breaking, some individual that has certain prerogatives with
respect to that instance of rule-breaking—e.g., some individual who is to be
apologized to or, alternatively, who can yield that prerogative by conferring
forgiveness on the offending party, etc. But while a set of rules can require
people to behave as if certain reactive attitudes were appropriate, a set of rules
cannot make any reactive attitude appropriate.13 So, for instance, a set of rules
could somehow institutionalize resentment on the part of rape victims toward
their attackers, but it cannot make it true that such a reactive attitude is appro-
priate. But it’s the fact that it’s appropriate for a certain person, and not others,
to resent the rapist that picks out that particular person as the wrong-sufferer.
So it remains true that rule-consequentialism cannot identify a wrong-sufferer.
3.2.4 A Third Response to the Objection

The first response to my directedness objection to consequentialism began by
pointing out that many individuals may well be mentioned in CTGC-derived
explanation of why some act was wrong. I now want to consider a third
response, which begins with the same thought as does the first response but
takes it in a different direction. Whereas the initial thought was that perhaps
everyone mentioned might qualify as a wrong-sufferer, this third response
suggests that everyone who is both mentioned and harmed qualifies as a
wrong-sufferer.14
Of course, depending on which version of consequentialism is to be defended,
this response might need to be refined. Given act-consequentialism or some
Against Monism 41
versions of rule-consequentialism, it is possible to do wrong without harming
anyone (e.g., by failing to augment the good by as much as one could have).
So we may want to stipulate that by ‘harm’ we mean ‘left worse-off relative to
the morally required action,’ such that an individual counts as harmed by some
wrongful action just in case it leaves her worse-off than she would have been if
the right action had been done.
Even with this amendment accounted for, the proposal is still going to
yield drastically counterintuitive implications. For instance, let’s reverse the
classic Transplant Case, so that instead of killing one healthy person to save
five who are terminally ill, the transplant surgeon kills five healthy people to
save one who is terminally ill. (The latter needs multiple transplants and the
five healthy people each possess only one organ healthy enough to be trans-
planted.) And let’s add a further fact, which is that the terminally ill person
has a mortal enemy who is terribly upset at his life being saved.
In this case we have six people who are harmed by the surgeon’s actions.
On the current suggestion, all of them would be wrong-sufferers, including
the person who is upset at a life being saved. This is very hard to accept.
To get around this problem, the consequentialist might want to say that
judgment-sensitive harms don’t count in the calculus. By a ‘judgment-sensitive
harm’ I mean the phenomenological unpleasantness of certain judgment-
sensitive mental states. For instance, in the reversed Transplant Case, the dis-
appointment of the mortal enemy is a judgment-sensitive harm because it’s
unpleasant and caused by her judgment that the terminally ill person deserved
to die. Or, to take another example, if one is offended at a white woman and
a black man getting married, this is a judgment-sensitive harm because it’s
unpleasant and caused by one’s judgment that racial intermarriage is immoral.
The problem with this response is that there is no consequentialist ratio-
nale for discounting judgment-sensitive harms. A non-consequentialist would
have less trouble here; she could simply say that an agent doesn’t count as
having caused x if another agent’s agency intervened between the original
agent doing what she did and x occurring, and could then point out that
making a judgment is an exercise of agency. But consequentialism always
takes into account all the downstream effects of the action being evaluated;
intervening agency makes no difference to the consequentialist.
Another way for the consequentialist to respond is by pointing out that
the objection in the reverse Transplant Case arises only for versions of con-
sequentialism on which well-being is what goes into the calculus. If instead
of this person-affecting consequentialism we were to put forward an imper-
sonal consequentialism on which only objective good and bad go into the
calculus, the problem would disappear, for then we could say that it’s not
objectively bad for the mortal enemy to be upset at the life being saved.
But any purely impersonal consequentialism (and this solution to the
reverse Transplant Case does require a purely impersonal consequential-
ism) will definitely succumb to the directedness objection, because on purely
impersonal consequentialism there are no victims at all! On purely impersonal
42 Against Monism
consequentialism all the explanations of token instances of wrongdoing are
(naturally) purely impersonal; no one gets mentioned in them.
The final way I could imagine a consequentialist responding to the reverse
Transplant Case problem is by adopting a theory of well-being on which
the mortal enemy’s anguish over the life being saved isn’t a harm. What we
would need here is a theory of well-being—perhaps a eudemonistic one—
that has moral content. That way, the moral objectionableness of the ene-
my’s anguish figures into whether it’s a harm.
The problem, however, is that if the mortal enemy’s anguish over the life
being saved isn’t a harm, then surely it’s a benefit. (If the moral character
of one’s phenomenological states helps to determine what effect they have
on one’s well-being, then surely the moral depravity of the mortal enemy’s
anguish not only makes it not-harmful but also beneficial—perhaps as a
case of getting what one deserves.) And if a theory of well-being says that
malevolence-caused anguish is a benefit, then with any reasonable semblance
of symmetry that theory will also say that malevolence-caused delight is a
harm. But if malevolence-caused delight is harm, then we can reconstruct the
Transplant Case once again so that it poses a problem for the consequential-
ist. In the new Transplant Case once again five are killed to save one, but
one of the healthy patients in this case has a mortal enemy who is delighted
at her demise. Regarding this case the new theory of well-being will say that
the mortal enemy is harmed by what the surgeon does (because malevolence-
caused delight is a harm), and so the version of consequentialism that contains
this theory of well-being and says that everyone who is both harmed by an
action and is mentioned in the token-grounding claim derived from CTGC vis
á vis that action is wronged will have to say that the mortal enemy is a victim.
3.2.5 A Fourth Response to the Objection

Act-consequentialism is often associated with Bentham’s slogan, “each to
count for one, and none for more than one.” So it is natural to think, as
is sometimes explicitly claimed, that on a version of act-consequentialism
that countenances only personal good the wrongfulness of an action is to
be explained by how it failed to take fully into account someone’s interests
or good, thus making that individual a wrong-sufferer.15 Imagine that in
deciding whether to φ or not-φ the agent made a list of all the personal
goods and bads that would befall each affected individual under the two
scenarios. If the list were thorough and all the personal goods and bads
were weighted proportionately and without heed to whose goods or bads
they are (except insofar as this version of consequentialism embraces agent-
relativity), then the agent who makes her decision based on this method will
never do wrong. So when wrong is done, this means someone’s personal
good was weighted incorrectly.
Of course, we don’t usually deliberate like this; we use shortcuts and heu-
ristics. So the proposal can’t literally be that when an agent does wrong she
wrongs everyone whose personal good she failed to account for correctly
Against Monism 43
in her deliberation, otherwise just about every time an agent does wrong it
will turn out she’s wronged many people. Rather, the proposal must be that
a failure to account sufficiently for someone’s personal good is a feature of
actions, as opposed to a feature of deliberations.
But this is simply false. There is no fact as to whether an action succeeds
or fails to appropriately take into account some individual’s person good.
Suppose an agent fails to give money to a homeless person she passes on the
street. In virtue of what could we say, truthfully, that this action failed to
appropriately take into account the homeless person’s good? What makes it
true that the action discounted the homeless person’s personal good and false
that (for instance) the action accounted accurately for the homeless person’s
personal good but drastically over-counted the agent’s personal good?16
3.2.6 Conclusion
In this section I have objected to consequentialism on the grounds that
CTGC cannot account for the directedness of some instances of wrongdoing.
I began by situating the idea of directed wrongdoing among other familiar
moral notions, each of which requires that in cases of directed wrongdoing
there be a wrong-sufferer. I argued, next, that the token-grounding claims
implied by CTGC never identify a wrong-sufferer, and that for that reason
cannot explain instances of directed wrongdoing.
3.3 The Counterintuitiveness Objection

All of consequentialism’s token-grounding explanations are to be derived
from CTGC, and many of those explanations are counterintuitive. This is
important, because I argued in Chapter 1 that our explanatory intuitions
ought to be given weight in determining which normative ethical theory to
accept. I cannot argue that you ought to find any token-grounding explana-
tion counterintuitive, for the obvious reason that arguments cannot provide
reasons for intuitions (intuitions, by definition, cannot be held on the basis
of reasons). But I might be able put them in such a light that you do in fact
find them counterintuitive. I’ll now try.
Suppose Jones really wants to buy a new bicycle but does not have enough
spending money to do so. So she grabs a knife from her kitchen, walks to
a dimly lit street, and waits for the next lone passerby. When that pass-
erby, Williams, comes along, Jones springs without warning from her hiding
place, stabs Williams in the stomach, grabs Williams’s purse and runs away
as Williams bleeds to death.
Let’s assume that consequentialism would condemn Jones’s action as
wrong. As to why the action is wrong, a token-grounding explanation
derived from the act-consequentialist version of CTGC will say that it is
wrong because the goodness of all of its consequences is less than the good-
ness of all of the consequences of some other action, φ, that Jones could
have performed, while a token-grounding explanation derived from the
44 Against Monism
rule-consequentialist version of CTGC will say that it is wrong because it
would not be allowed by the system of rules the universal acceptance of or
adherence to which would, among all the possible sets of rules, have the best
consequences. But each of these token-grounding claims is highly counter-
intuitive. What we want to say, I take it, is that the explanation is that Jones
killed Williams.
Of course, we would need to say more if Jones chose to defend her behavior
in response to this token-grounding claim, but that’s beside the point. We’re
talking about which facts explain the wrongness of wrongful actions; we’re
not talking about which facts must be brought forward to establish that an
action was wrong. (Remember that, as I said in Chapter 2, those two sets of
facts might not be identical, owing to the distinction between facts that do
grounding work and facts that enable other facts to do grounding work.)
Of course, this is merely to report how things seem to me. My goal is just
to elicit in you a certain reaction by telling a story and drawing a con-
trast between two different kinds of substantive moral reasoning that might
ensue, depending on which moral explanations the parties to that reasoning
accept. That reaction I intend to elicit is the coming to have, or the realiza-
tion that you already have, an intuition that many of the token-grounding
claims implied by CTGC are false.
3.4 Conclusion
I have argued in this section that consequentialism is vulnerable to two
objections: First, its type-grounding claim, CTGC, yields token-grounding
claims that fail to allow that some wrongdoing is directed. Second, many of
the token-grounding explanations that CTGC yields are deeply counterin-
tuitive. Along the way I discussed various minor modifications that could
be made to CTGC that might be thought to help its case. None of them,
however, did the trick.
I do not conclude from all this that CTGC is defective. All I have shown
is that CTGC is not up to certain explanatory tasks. If CTGC were merely
one type-grounding claim in a pluralistic moral theory, then maybe the other
type-grounding claims in that theory could be invoked to do the explana-
tory work CTGC can’t do. I’ll defer until Chapter 8 a discussion of whether
we should accept as part of a pluralistic theory the two ideas into which
CTGC can be decomposed: the idea that morality requires us not to harm
others and the idea that morality requires us to confer benefit on others.
While I don’t (yet) conclude that we should reject CTGC, I do conclude
that we should reject consequentialism, since consequentialism is composed
of nothing but CTGC and CTGC’s token-grounding claims don’t do a good
job of explaining many of the verdictive facts.
But perhaps I have adopted too narrow a conception of consequential-
ism. It is sometimes said that, for consequentialists, facts about goodness
do not ground the verdictive facts; rather facts about reasons ground the
verdictive facts. If a theory of this sort included a distinctly consequentialist
Against Monism 45
way of understanding of reasons, then it might deserve the ‘consequential-
ist’ label. I’m thinking, for instance, of a theory on which all moral rea-
sons are grounded in the good and the strength of our overall moral reason
to do something is proportional to the strength of the good in which it is
grounded, or, alternatively, of a theory on which for something to be good
just is for it to constitute or ground a reason of a certain sort.17
A theory of this sort would be a version of ‘Rossian Pluralism’, a family
of ethical theories that I discuss in the next chapter. So although I have no
interest in insisting that such a theory would not also qualify as a version of
consequentialism,18 I will not discuss it here.
4. Kantian Ethics and Cruelty to Animals
4.1 Kantian Ethics as Monistic

I now want to investigate Kantian ethics. The most straightforward way
to represent Kantian ethics as a form of moral monism is to convert the
Categorical Imperative into a type-grounding claim and let that claim stand
alone as a theory unto itself. To illustrate, here are the first two formulations
of the Categorical Imperative in Kant’s words (translated into English):
Act only according to that maxim through which you can at the same
time will that it become a universal law.
(Groundwork 4:421)19
So act that you use humanity, in your own person as well as in the per-
son of any another, always at the same time as an end and never merely
as a means.
(Groundwork 4:429)
If we turn these imperative sentences into declarative ones and bring for-
ward their implicit claims about moral grounding, we arrive at:
The fact that one acted on a maxim that one could not have willed
to become a universal law grounds the moral impermissibility of one’s
resulting action.
The fact that one’s action failed to treat one’s humanity or that of
another as an end in itself or treated it as a mere means grounds the
moral impermissibility of that action.
There are many ways of objecting to this way of representing Kantian eth-
ics. The most important objection, for our purposes, would be the idea
that it is a mistake to portray the Categorical Imperative as the entirety of
Kant’s normative ethical theory. Kant does, after all, put forward a Doc-
trine of Right and a Doctrine of Virtue in the Metaphysics of Morals, from
46 Against Monism
which he derives various rights and duties, and according to some interpre-
tations of Kant, these doctrines are supposed to have independent moral
force (specifically, their moral force is not supposed to stem from their being
derivable from the Categorical Imperative).20
I have no argument against interpreting Kantian ethics this way. However,
in this chapter I am interested in examining the prospects for moral monism,
and since it is possible to represent Kantian ethics as a version of moral
monism and many moral philosophers do think of it that way, it is worth
our trouble to investigate the viability of Kantian ethics so conceived. More-
over, even if Kantian ethics is not monistic in letter, it is monistic in spirit.
All of the elements of Kantian ethics, including the Categorical Imperative,
the Doctrine of Right and the Doctrine of Virtue, are derived from the same
basic idea: that we should always act in a way that is consistent with respect
for humanity.
4.2 Cruelty to Animals

This aspect of Kantian ethics is both its greatest strength and its Achilles
heel. It is a strength in that it provides a basis for, and makes Kantian ethics
consonant with, some very powerful moral ideals. One of these is the ideal of
showing respect for persons and treating them with dignity, which is embod-
ied in the second formulation of the Categorical Imperative. The other is the
“What if everyone felt free to act that way?” test of the moral permissibility
of a way of acting, which is embodied in the first formulation. So if we repre-
sent Kantian ethics monistically, with a rephrased version of the Categorical
Imperative as its sole type-grounding claim, Kantian ethics is going to score
very well on the parameter of offering a compelling type-grounding claim.
But there is also trouble to be found in representing morality as funda-
mentally concerned with whether our actions are consistent with respect for
humanity. The problem, of course, is that it leaves us at a loss when we try
to explain the apparent moral limits on our treatment of beings that are not
autonomous.
This problem arises most famously in Kant’s discussion of violence against
animals in The Metaphysics of Morals. He says there:
With regard to the animate but nonrational part of creation, violent and
cruel treatment of animals is far more intimately opposed to a human
being’s duty to himself, and he has a duty to refrain from this; for it
dulls his shared feeling of their suffering and so weakens and gradually
uproots a natural predisposition that is very serviceable to morality in
one’s relation with other men . . . Even gratitude for the long service
of an old horse or dog . . . belongs indirectly to a human being’s duty,
namely, his duty with regard to those animals; considered as a direct
duty, however, it is always only a duty of the human being to himself.
(6:443, emphasis original)21
Against Monism 47
Because animals are not autonomous—that is, they lack humanity—they
do not fall under the Categorical Imperative and thus killing them cannot
constitute a violation of any duty to them.22 Kant would have to say this not
just about shooting a dog, but also about the most unimaginable forms of
cruelty, as long as that cruelty is inflicted on an animal.
The problem here is not that the Categorical Imperative, as a type-grounding
claim, is inconsistent with certain verdicts we are inclined to accept. Kant
shows us how to get the right verdicts, at least some of the time. The prob-
lem, rather, is in the token-grounding explanations of those verdicts the
Kantian type-grounding claim implies. What grounds the wrongness of a
given act of cruelty to an animal, apparently, is how it is damaging to one’s
own humanity (O’Neill 1998). So such an act wouldn’t wrong the animal.
This is terribly counterintuitive.23,24
4.3 A Response and a Rejoinder

The Kantian needs to propose that in spite of initial appearances, her theory
can allow that the wrongness of a case of cruelty to an animal is grounded in
its harmfulness to the animal, whereas the objection I’ve been leveling proposes
that the Kantian explanation of its wrongness is grounded in its being damag-
ing to the agent’s own humanity. Now if the grounding relationship is transi-
tive, then the two proposals aren’t inconsistent; it could be that the although
the wrongness of a case of cruelty to animals is grounded in its being damag-
ing to the agent’s own humanity, it is also grounded in its harmfulness to the
animal, because its being damaging to the agent’s own humanity is grounded
in its harmfulness to the animal. But the transitivity of the grounding relation-
ship won’t help here (it is of even less help to the Kantian than it was shown, in
§3.2.2, to be to the consequentialist), because it’s not the case that an action’s
being damaging to one’s humanity can be grounded in its constituting cruelty
to an animal. If anything, the relationship between the two facts is causal.
4.4 Korsgaard’s Neo-Kantianism
4.4.1 Respect for Animal Nature

In a series of publications over the last decade Christine Korsgaard has
attempted to demonstrate that Kantian ethics is innocent of the charge of
not allowing for duties to animals. While the second formulation of the
Categorical Imperative tells us to refrain from treating humanity as a mere
means, and thus grounds obligations to all beings that possess humanity,
Korsgaard holds that the same argument that gets us to this formulation
also forces us to accept a separate formulation on which we are instructed
to refrain from treating animal nature as a mere means. This formulation,
according to Korsgaard, grounds duties to all beings that possess an animal
nature. In the rest of this section I explore what Korsgaard has to say in
48 Against Monism
defense of this claim, setting aside the question of whether Korsgaard’s inter-
pretation of Kant is accurate, since my goal here is simply to say whether
some monistic theory that is at least Kantian in spirit can hold that cruelty
to an animal wrongs that animal.
Korsgaard begins by asking: On what grounds are we obligated to resist
treating humanity as a mere means? Here Korsgaard makes the crucial point
that it is not on account of humanity possessing an out-there-in-the-world
absolute value. For the Kantian, value is constructed through the will. Cer-
tain ways of acting present themselves to us in a “normatively loaded” way
(2005, 83; 2009, 114), which is to say that they strike us as worth doing
or worth avoiding. For instance they whet an appetite, incite a passion or
arouse a curiosity (2011, 102), or stoke a desire or fear (2013, 637). Kors-
gaard calls these mental states incentives. She goes on to say that when an
individual with the capacity for reflective endorsement endorses an incen-
tive she confers value on the incentivized way of acting and makes it an end
in itself (2005, 92–3; 2011, 106; 2012, 8; 2013, 640). So if we are obligated
to resist treating humanity as a mere means, this must be because we reflec-
tively endorse humanity—the capacity for reflective endorsement itself. And
in reflectively endorsing humanity, we impose on ourselves a law requiring
that we respect humanity (2005, 93–5; 2012, 9–10).
What’s key to Korsgaard’s argument regarding animals is her further claim
that our humanity is not the only aspect of our nature that we reflectively
endorse. She contends that we reflectively endorse a range of incentives that
fall under our animal nature, such as seeking pleasure, sex, and play and avoid-
ing pain, mutilation, and loss of control (2005, 105; 2011, 108; 2013, 643).
This gets us to the endorsement of animal nature, but it still doesn’t get us
to duties to animals. This is because we haven’t yet established the scope of
our self-legislative ability. Isn’t it possible for a human to reflectively endorse
only her own animal nature, and thus to lay down for herself only a very
narrow law requiring that she not treat her own animality as a mere means
(2005, 100f)? Of course, by the same token one might wonder whether one
could avoid incurring duties to other humans all the while endorsing one’s
own humanity, which would of course be to prove too much.
Korsgaard responds that when one reflectively endorses an incentive one
does not merely endorse one’s own instantiation of it. Rather, one endorses it
in general (2005, 104; 2012, 13).25 So the laws one gives to oneself through
reflective endorsement include within their scope all the beings that possess
the incentive being endorsed (2005, 105; 2012, 13). Consequently, others’
ends bind each of us.
4.4.2 First Objection to Korsgaard: Having Standing

With Respect to a Law
I have two worries about Korsgaard’s argument here, the first of which I
discuss here and the second of which I set aside until §4.4.3. To understand
Against Monism 49
the first worry, we need to recall what our concern about Kantian ethics was
in the first place. The concern was that the token-grounding claims it offers
to explain the wrongness of instances of cruelty to animals, which avert to
how such an act of cruelty is an affront to the perpetrator’s own human-
ity, are deeply counterintuitive. Now with regard to Korsgaard’s Kantian-
ism, the worry is that although she has shown that Kantian ethics offers
token-grounding claims that are different than this, they aren’t much more
intuitive. This may seem like a surprising criticism, given that Korsgaard sets
herself the task of demonstrating that we have duties to animals. What bet-
ter token explanation could we have for the wrongness of a case of cruelty
to an animal than the fact that such cruelty constituted the violation of a
duty to the animal?
I admit that this would be a great explanation under a standard concep-
tion of what it means to have a duty to an individual. But Korsgaard has a
very weak notion of ‘duty to’. She contends that there are two ways in which
an entity can be the kind of entity (an end-in-itself) to whom one can have
duties. The weaker of those two ways, and the one under which animals
qualify, is by being a source of legitimate claims. And what it means to be a
source of legitimate claims is to be in a position to make a claim on a moral
agent in the name of a law the authority of which is recognized by that agent
(2005, 95–6; 2011, 104, 109). So this is what Korsgaard is saying is true of
animals when she says that we have duties to them.
The only example Korsgaard offers of how one gets to be a source of
claims in this sense, aside from the argument she gives to establish that ani-
mals qualify, is from the law: she says that individuals who are protected by
laws that they had no part and could have no part in creating can be sources
of claims in this sense (2011, 107). So if Country X has a law that bans rap-
ing people regardless of their nationality, this makes each foreign visitor to
country X a source of claims in the given sense. Each such visitor makes a
claim on the citizens of Country X—a claim to not be raped—in the name of
a law that is binding on those citizens all the while having played no part and
not being the sort of person who could play a part in the creating of that law.
Note, however, that the laws of Country X might make it quite clear that
violations of the prohibition on raping foreigners are to be treated as purely
criminal violations, thus affording foreigners who have been raped no abil-
ity to register a civil complaint or demand restitution. In other words, the
laws of Country X might make it quite clear that the raped party is to be
afforded no standing. So the mere fact of being able to make a claim on
some agent in the name of a law that that agent recognizes does not entail
having any standing with respect to that law.
Perhaps Korsgaard’s point, however, is that the mere fact that the for-
eigner had no part and could not possibly have had a part in creating the
law is insufficient to establish that she lacks standing with respect to vio-
lations of the law.26 After all, the country could afford her standing if she
were raped. This is enough, it seems, to open up the possibility that animals
50 Against Monism
could have standing with respect to the moral law despite having no part in
creating it.
Fair enough. Now taking it as given that animals could have standing
with respect to the moral law, how can we figure out whether they do?
Since according to Korsgaard the moral law arises out of certain mental
states (i.e., the reflective endorsement of incentives), one must look to those
mental states to settle the question of whether some animal has standing
with respect to the moral law it creates. And when we do this, we find no
answer. There is nothing in the attitude of endorsing an incentive or refusing
to endorse an incentive that could establish any fact one way or the other
whether, if at some point I engage in an action that is inconsistent with my
endorsement of some incentive, which of the individuals affected by that
action will have standing with respect to the law I thereby violate. Or, to
express this same idea by reference to the earlier explication of the notion
of wronging (§3.2.1): There is nothing in the idea of endorsing an incentive
that could establish that if at some point I engage in an action that is incon-
sistent with my endorsement of some incentive I thereby violate an obliga-
tion to some individual in virtue of engaging in that action, or that some
individual might be in a position to complain about the resulting action or
resent one for having done it or be in a position to demand an apology for it.
This brings us back to the earlier worry about Kantian ethics being too
self-regarding, in that all moral constraints on me arise from constraints I
impose on myself.27 The worry then is that in Kantian ethics, for all Kors-
gaard has said, there is no fact of the matter as to whether animals have
standing, and thus no fact of the matter as to whether they can be wronged.
4.4.3 Second Objection to Korsgaard: The Superfluousness of

Respecting Animal Nature
Supposing, however, that Korsgaard does establish that animals have mean-
ingful standing in a Kantian ethical framework, there is a second problem:
it’s unclear nevertheless whether we have any duties to them. To see whether
this is the case we have to examine how duties are extracted from the sec-
ond formulation of the categorical imperative, the formulation with which
Korsgaard is working. Under the classical understanding of the second for-
mulation, in which the concern is merely humanity, and animality is not
mentioned, there are four ways of violating it:
1. Treating one’s own humanity as a mere means.

2. Treating the humanity of others as a mere means.
3. Failure to treat one’s humanity as an end in itself.
4. Failure to treat the humanity of another as an end in itself.
In a much earlier publication Korsgaard offers an example of each kind of

violation (1996, 125–28). We need not review the examples of Type 1 and
Against Monism 51
Type 3 violations, as for the purposes of this discussion we are concerned
exclusively with our treatment of others (animals). This leaves Type 2 and 4.
Let’s begin with the latter.
Korsgaard offers, as example of a Type 4 violation, failing to promote
the happiness of others. Happiness is achieved through the realization of
ends that one sets for oneself, so a failure to promote the happiness of
another is a failure to take seriously the value of her ends, which implies
a failure to take seriously her capacity to confer value through endorsing
something as an end. Consequently, we have a duty to promote the hap-
piness of others.
Now the question arises: When we amend the second formulation of the
Categorical Imperative so that it mentions animality in addition to human-
ity, are there then more actions that constitute Type 4 violations? Could I
fail to set the animality of others as an end? I might, but it’s hard to see what
the objection to it would be. If I fail to set the humanity of another as an
end, this is objectionable because it constitutes a denial of that other’s abil-
ity to confer value on something through the act of reflective endorsement.
Nothing analogous happens if I fail to set the animality of another as an end.
As to Type 2 violations, Korsgaard offers the example of false promising.
When people keep their promises this is an exercise of humanity—of good
willing. A false promise can work only when and because others keep their
promises. Therefore, false promises use the humanity of others as a mere
means, and so we have a duty not to make them.
Now if we reformulate the second formulation of the Categorical Impera-
tive so that it mentions both humanity and animality, are there then more
ways of committing Type 2 violations? Can I use the animality of others
as a mere means? Well, I might use a woman’s sex drive as a means to the
satisfaction of my own. But this would be wrong only if she didn’t consent.
But we already knew, from having consulted the original second formula-
tion, that it’s not permissible for me to have sex with a woman without her
consent. So it doesn’t appear that we learn anything important by noting
that the animality of another can be used as a mere means.
This is a surprising conclusion. I would guess that most people share
Korsgaard’s sense—and indeed even now I have to stop myself from getting
carried away by the intuitive pull of it—that bringing animals and animality
under the ambit of the second formulation of the Categorical Imperative
must make some difference to how we’re permitted to treat them. How
could it be, for instance, that hunting animals for sport is not ruled out by
this new formulation?
The answer, in short, is that animality was already covered by the sec-
ond formulation of the Categorical Imperative before Korsgaard amended
it. Some uses of another’s animality are ruled out by the unamended
formulation—namely, all those uses of another’s animality that are also
impermissible uses of another’s humanity. Getting someone drunk in order
to make her want to have sex with you may constitute such an example.
52 Against Monism
Meanwhile, the rest of the uses of another’s animality are ruled in by the
unamended formulation. Having sex with someone with that person’s unim-
paired consent would be one case.
But doesn’t that mean, one might ask, that for those individuals that
lack humanity, the amended second formulation tells us something that we
didn’t know already regarding how it’s permissible to treat them—namely,
that we shouldn’t use their animality as a mere means? The first thing to
note, by way of response, is that there is a risk of going too far here. If,
because I’m lazy, I train my dog to fetch my slippers, am I using an aspect
of my dog’s animality (namely its instinct to retrieve) as a mere means to
my end? Surely we don’t want to say that, since that would be to condemn
such dog training as immoral. The obvious escape is to point out that there
doesn’t seem to be anything objectionable, from the dog’s point of view, to
fetching my slippers. Now there are two ways to understand this. We could
simply understand this as a way of saying that such dog training doesn’t
make my dog unhappy, but this observation is entirely out of place when
considering whether such dog training is a Type 2 violation (it seems more
appropriate to the question of whether it’s a Type 4 violation, but we’ve
already seen why one cannot commit a Type 4 violation with respect to an
animal). Alternatively, we could understand this as a way of saying that my
dog would, if she could, consent to being trained this way. This is precisely
the route the Korsgaard elects to take. She says, “We may interact with the
other animals as long as we do so in ways to which we think it is plausible
to think they would consent if they could” (2011, 110).
The problem here is that I cannot literally imagine my dog conferring or
withholding consent. My dog wouldn’t be that particular dog any more if
it had the capacity to consent; it would be some other being. Now I can
imagine an animal just like my dog but with the added capacity for consent,
and I can imagine her dissenting from certain ways of being treated. But if
that’s the basis for my duty to refrain from doing some particular thing to
her, then it’s not her that constitutes the source of that duty; rather it’s the
hypothetical agent I constructed. And so it’s not a duty to her.
4.5 Conclusion
The argument I have presented in this section is not intended as a dismissal
of Kantian ethics. I have said nothing against the idea that we can derive
moral constraints on our behavior from the idea of respect for humanity.
Rather, I object to using this idea as a starting point when trying to explain
why it’s impermissible to treat animals cruelly. All I object to, then, is monis-
tic Kantian ethics. For all I’ve said the Categorical Imperative might be sal-
vaged as one part of a pluralistic normative ethical theory that has some
other element that offers a different explanation of the wrongness of cruelty
to animals.28
Against Monism 53
5. Contractarianism/Contractualism and Cruelty to Animals
5.1 Varieties of Contractarianism and Contractualism

The final kind of moral monism I want to explore is contractualism/
contractarianism. Like Kant, contractualists and contractarians attempt to
show how moral constraints arise from rational acts of will. The major dif-
ference is that for Kant, as we have seen, the relevant acts of will are, at least
in the first instance, inwardly directed. They set the standard for one’s own
behavior. By contrast, the contractualist/contractarian wants to show that
each individual’s will is directly a constraint on the behavior of others. Sim-
ply in virtue of my willing or failing to will, endorsing or failing to endorse,
consenting to or dissenting from, some course of action or rule, you are
obligated to do certain things and refrain from doing others.
This immediately eliminates one problem, which is that to the extent that
a theory is inwardly directed it will have a difficult time explaining how
any wrongdoing could constitute wronging. The question that remains is
whether when it comes to non-humans any of the wrongful things we might
do to them could count as wronging them. I want to argue that monistic
contractualism/contractarianism is incompatible with the idea that non-
humans can be wronged. Of course, this is a quite broad family of moral
theories, so it would be impossible to undertake the thorough survey of all of
its variations necessary in order to decisively establish that it is doomed by
the problem I’ve just mentioned. So I have no choice but to narrow my tar-
get, and it seems to me the most sensible way to do this is to focus on Mark
Rowlands’s version of contractarianism. There are three reasons for this.
First, of all the contractarians and contractualists, Rowlands is the one
who’s gone to the greatest efforts to demonstrate, from within that frame-
work, that we have obligations to non-humans.29 Second, contractarianism
seems, prima facie, more likely than contractualism to underwrite obliga-
tions to non-humans. Contractarians tend to see moral constraints as aris-
ing out of a covenant designed to protect a set of individuals whose interests
can be sensibly represented to others during a real or hypothetical process
of negotiation. We have no trouble thinking of certain non-humans as hav-
ing interests, and indeed as having some of the same interests as we do.
The problem of course is in understanding how they could gain a seat at
the table, lacking as they do any ability to negotiate for themselves. Still,
though, the problem doesn’t seem unmanageable. Contractualists, on the
other hand, see moral constraints as arising out of an exchange of claims
and counter-claims among morally motivated individuals. The problem for
contractualism is that it’s not clear what a non-human could bring to the
table, so to speak, for a process of negotiation so conceived. The bare pos-
session of interests wouldn’t seem to make any difference.
Third, contractualist thought is dominated by Scanlon’s theory, a version of
contractualism that seems almost certain not to be able to ground obligations
54 Against Monism
to non-humans. The most important reason for this is that Scanlon’s theory
isn’t supposed to identify the grounds of any of our obligations, much less
any obligations we might have to non-humans. Scanlon is in the business of
showing what it is for a certain action to be right or wrong, not what makes
an action right or wrong.30 Furthermore, Scanlon himself does not think that
his theory can underwrite obligations to non-humans (1998, 177–85). (Of
course, Scanlon insists that his theory is not supposed to describe the entirety
of morality, but rather just that aspect of it concerning what we owe to each
other. Yet, it is precisely the intuition that there are certain things we owe
to non-humans that’s driving our concern here. Furthermore, our question
is whether we should accept contractarianism/contractualism as a monistic
theory; that is, as capturing the entirety of morality.)
For all these reasons, then, I consider it most sensible to focus on Row-
lands’s contractarianism and see how far that gets us.
5.2 Rowlands’s Argument for Animal Rights

Rowlands’s goal is to develop Rawls’s contractarianism and demonstrate
that, contrary to what Rawls and most of his commentators have thought,
Rawls’s contractarianism provides the basis for ascribing rights to ani-
mals. In what follows I offer a very brief summary of Rowlands’s argument
(2009, Chapter 6). It doesn’t do justice to the nuance with which Rowlands
approaches his subject, especially his careful exegesis of Rawls, but it will
give us everything we need to assess Rowlands’s case for animal rights.
The key to that case is Rowlands’s view regarding the thickness of the
veil of ignorance. Rowlands emphasizes that the entire construction of the
original position, including the veil of ignorance, is, for Rawls, to be based
on prior moral commitments. In other words, the original position is not
supposed to be a morally neutral starting point. Rather, it is supposed to
embody the moral convictions of which we are most confident and that
play the most fundamental roles in our moral thought. From these con-
victions, via the original position thought experiment, we are supposed
to be able to derive further moral constraints. Rowlands then points out
that one of these central moral commitments is that if an individual is not
responsible for possessing property P, then that individual is not entitled to
whatever benefits accrue from her possession of P. This is our anti-moral
arbitrariness conviction. Rowlands then makes his key move in defending
animal rights: he points out being human is not something for which any of
us are responsible. Therefore, given that from behind the veil of ignorance
we are supposed to be ignorant of whatever properties of ours we are not
responsible for, the veil of ignorance must then deprive us of the knowledge
of our humanity. And if we consider what set of rules we would endorse
while not presupposing our humanity, we find that we would favor rules
that afford many protections for animals, such as protection against being
killed for food.31
Against Monism 55
5.3 An Objection to Rowlands
As far as this goes, I say “fair enough”. But does this get us as far as say-
ing that animals can be wronged? Rowlands thinks it gets us to the conclu-
sion that animals have direct rights, for instance against being killed for food,
where a direct right is a right the possession of which by X does not depend
on the possession of a right by any other individual, Y. (Direct rights can be
thought of as the correlates of direct duties.) If animals have this kind of right
then, certainly, they can be wronged, and therefore if Rowlands has success-
fully defended the idea that animals possess direct rights then he has shown
that they can be wronged.
The problem, however, is that we certainly wouldn’t want to say that
anything protected by the set of rules agreed upon in the original position
has direct rights. We might, for instance, adopt a rule protecting certain
inanimate objects like the Egyptian pyramids, and yet we wouldn’t want to
say that the pyramids have direct rights.
The closest Rowlands comes to responding to this objection is when he
says the following:
the scope of morality is restricted to things that an occupant of the

original position could rationally worry about being. I can, in the origi-
nal position, worry about being at least certain sort of non-human
animal since there is something that it is like to be them (at least some
of them). [. . .] But if I were told that I was going to be a plant, or a car,
then I couldn’t care less what happened to me (and rationally so). Plants
and cars are not sentient, hence do not suffer. [. . .] The contractarian
position, then, makes sentience the cut-off point for morality.
(2009, 160)
There is an important ambiguity in Rowlands’s use of the expression “the

scope of morality” in the first sentence of the block quote and “morality”
in the last. For Rowlands, these terms designate either the extent of moral
protection or the extent of the possession of direct rights. If the former, then
Rowlands is clearly mistaken. It is quite plausible that the set of rules that
would result from our deliberations in the original position would include
protection for inanimate objects like the pyramids. If the latter, then what’s
going on in this passage is Rowlands is changing his argument for animal
rights. It would seem Rowlands realizes that the scope of direct rights will
be much too broad if X’s being protected by the agreed-upon rules is suf-
ficient for X’s possession of such rights, so he suddenly adopts the further
claim that X’s being the sort of thing that there is something that it is to be
like is necessary for X’s possession of direct rights.
Let’s set aside the worry about Rowlands’s consistency and simply focus
on the idea that that X’s being the sort of thing that there is something that
it is to be like is necessary for X’s possession of direct rights. My worry is
that this isn’t a particularly contractarian position. One doesn’t arrive at a
56 Against Monism
conclusion like this by reasoning from a contractarian starting point—that
is, by thinking about which rules could be agreed upon by self-interested
contractors (perhaps from behind a veil of ignorance). Granted, from a
contractarian starting point one could, indeed would, reach the conclusion
that there is something very important about the distinction between beings
such that there is something that it is like to be them, and other beings:
namely, the existence of the former sort of beings would constrain the delib-
erations of a rational contractor in a way that the existence of the latter sort
of beings (or objects) would not. But there is nothing particularly contracta-
rian about saying that this difference grounds a difference in the possession
of direct rights. Indeed, it is hard to see what motivation for saying this that
there could be from within contractarianism.
A much more natural thing to say, from the contractarian perspective,
is that whether an individual is the kind of being that can possess direct
rights depends on whether he or she is, or could be, a party to the contract.
As Darwall emphasizes, we tend to think that when we violate someone’s
direct right we thereby become accountable to that person, just as we would
any time we wrong someone.32 It makes a great deal of sense to say that if
I violate a rule that was agreed upon, or could have been agreed upon, by
you and I in a contracting situation, then I am accountable to you for that.
This is a much more Hobbesian way of thinking about contractarianism
than that with which Rowlands would be comfortable.33 Rowlands is a
Rawlsian, and he explicitly rejects the understanding of Rawlsian contrac-
tarianism on which the outcome of the original position has force in virtue
of its modeling some hypothetical agreement. Rather, Rowlands (2009, 136)
insists, the original position simply models a certain method of moral rea-
soning, and what makes a moral consideration valid is that we each would
arrive at it by reasoning in that way.
I have no interest in arguing that this is a bad interpretation of Rawls or
a bad version of contractarianism. I simply want to point out that once we
strip contractarianism of its distinctly contract-based element, then there
doesn’t seem to be anything to say one way or the other regarding whether
animals have rights. What we’re left with is the idea that the true moral
constraints are those that we would want enshrined if we reasoned about
the matter in a particular way. As a solitary rational agent, thinking (from
behind a veil of ignorance) about what I would want, all that would mat-
ter to me is that individuals do certain things and refrain from doing other
things. Therefore, all we can get from this version of contractarianism, it
seems to me, is a sorting of actions into the categories of permissible, obliga-
tory, and impermissible. We cannot, I contend, extract from this version
of contractarianism any further facts about who is obligated to whom, or
about the appropriateness of resentment, the making of a complaint or the
demanding of an apology in response to certain actions. In other words, we
can’t get a directed morality out of it. This is roughly the same as the first
objection I raised against Korsgaard’s version of Kantianism.
Against Monism 57
6. Conclusion
In this chapter we’ve surveyed the three most popular versions of moral monism,
and we’ve found that consequentialism cannot allow that some wrongdoing
is wronging, while Kantianism and contractarianism/contractualism cannot
countenance the claim that some wrongdoing is wronging animals.
This does not amount to a refutation of moral monism, since obviously
there are many other possible versions of moral monism. Still, at the risk of
overgeneralizing, I want to close this chapter by saying something about the
prospects for moral monism in light of the problem for it that I raised here.
It seems that the various wrongful actions are wrong for different reasons.
There are two apparent sources of these differences: (1) different wrongful
actions affect different individuals; (2) for each such individual there are dif-
ferent ways of affecting that individual that are wrongful. The moral monist
has to show that this apparent diversity hides a deeper unity in the source of
moral concerns. Broadly speaking there appears to be two ways to do this.
First, one could, as the consequentialist does, say that each of these different
moral concerns, insofar as it can be placed under the rubric of goodness and
badness, is indeed a prima facie moral concern, but the only actual moral
duty is to do the best one can in balancing them. Alternatively, one could,
as the Kantians, contractualists, and contractarians do, say that these vari-
ous considerations are on their own nothing more than something for the
rational agent to deliberate about, and that it is the deliverances of a rational
agent’s will that determine our moral obligations. Neither response seems
to do justice to the apparent diversity in the sources of moral concern. The
former strategy is too all encompassing: Each time I do right or wrong, it’s
because of how I treat everyone. The latter strategy is too egoistic: Each time
I do right or wrong, it’s because of how I exercise or respond to my own
rational agency.
Pluralism, whatever might be said against it, can easily avoid these prob-
lems. Let’s turn our attention now to what can be said against its different
varieties and what might be offered by way of rejoinder.
Notes
1. In conversation with act-consequentialists I’ve found that they’re sometimes
inclined to reinterpret their theory along these lines when confronted with the
explanatory objection to their theory that I raise later in this chapter.
2. This interpretation of Kantianism and the subsequent interpretation of
contractarianism/contractualism are suggested by Väyrynen (2009, 101–2).
3. Shaw 2006, 215; Stewart 2009, 63; Schroeder 2007, 38–9; Wringe 2003, 85;
Lichtenberg 1983, 546; Ross 1988, 37–9. Lichtenberg and Ross actually con-
flate this objection with another objection, which is that consequentialism gets
the wrong verdict in some cases. The objections certainly are closely related, I
admit. But as I’ve argued elsewhere (Sachs 2010a), there could be a brand of
consequentialism that is vulnerable to the explanatory objection but gets all the
verdicts right.
58 Against Monism
Stratton-Lake (2011) raises something like the explanatory objection to con-
sequentialism, though he seems to mostly have rule-consequentialism in mind.
Interestingly, he claims that Rossian pluralism escapes this objection—a claim
I’ll deny in Chapter 4.
4. Kymlicka (2002, Chapter 2, esp. pp. 24–5 and 28–32) (see also Kymlicka 1989,
28–9); Stratton-Lake (2011).
Nick Zangwill (2011) offers something like this objection, though it is directed
exclusively at indirect consequentialism and it is not general in the sense of sug-
gesting that indirect consequentialism generally gets the right answer for the
wrong reason; rather his worry is that it does so in a particular kind of case.
Also, this criticism itself is indirect in that for Zangwill it does not on its own
count against indirect consequentialism, but rather does so in conjunction with
a theory of permissible partiality and morally acceptable deliberation.
5. For the sake of simplicity I set aside other versions of indirect consequentialism,
such as motive consequentialism. But the arguments I make here against act-
and rule-consequentialism could be used, mutatis mutandis, against the other
versions of indirect consequentialism as well.
6. Mark Nelson (2015) makes this argument specifically against utilitarianism.
7. I was prompted to point this out after attending a conference in which Samuel
Scheffler suggested that a consequentialist might take it as a badge of honor
that her theory cannot allow for wronging. (Note: Scheffler was not endors-
ing this line of thought.) A consequentialist might say, he pointed out, that the
idea of wronging is an outdated vestige of morality’s early roots as a system for
maintaining peace and order in small communities. Now that the most press-
ing moral problems seem to be more global and impersonal, a consequentialist
might say we have no need for such an interpersonal notion as wronging.
8. Chappell (2015).
9. Whether this is true is controversial. I will grant for the sake of argument that
it is.
10. The position I take in this paragraph follows Strawson (1962), Darwall (2006,
66–7) and Wallace (2007, 30–1).
11. Darwall (2006, 310–13). This is an aspect of a more general point Darwall makes,
which is that there is no in-principle reason why agent-neutral principles of conduct
couldn’t be derived from an agent-relative theory of moral obligation (such as Kan-
tianism or contractualism). See Darwall (2006, Chapter 1 n.16, Chapter 4 n.32).
12. Parfit (2011, 417–18) is explicit about this.
13. Darwall (2006, 15–17) makes the same point a broader context: He says that
to propose that the social utility of someone’s having a reactive attitude justi-
fies their having that attitudes is to make a “wrong kind of reason” error with
respect to reasons for attitudes.
14. Thanks to Claire Benn and Fiona Woollard for mentioning this possibility and
aiding me in thinking through it. I later discovered that a quite sophisticated
version of this view has been defended in print. See Roberts (2002).
15. Thanks to Pekka Väyrynen for pushing me to address this version of act-
consequentialism. That an act-consequentialist might make this move is noted
in Kymlicka 2002, Chapter 2. The best contemporary example of an act-
consequentialist who at least might believe this is the utilitarian Peter Singer, who
at times describes the principle of utility as falling out of the more basic moral
fact that everyone’s interests count equally. See, for instance, Singer (1993, Chap-
ter 1). Alternatively, it is open to us to say that the injunction to weigh everyone’s
(equally strong) interests equally just is the principle of utility (Hare 1984, 107)
or just is the act-consequentialist moral principle (Scheffler 1982).
16. The consequentialist could respond at this point by saying that correctly
accounting for someone’s personal good is a comparative matter—a matter of
Against Monism 59
assigning the correct relative weights to the personal good of the various indi-
viduals. But this can’t be right. When more than two people are affected by
some action, mere comparative facts to the effect that A’s personal good should
be weighted this much relative to B’s and B’s personal good should be weighted
that much relative to C’s, and so on, may leave it indeterminate what the right
thing to do is. There needs to be some way of calibrating the various relative
weightings so as to yield an overall fact as to which course of action will pro-
duce the most (properly weighted) personal good. But if there is an overall cali-
bration that covers everyone, then for any wrongful action involving more than
two individuals it’s going to be indeterminate which particular miscalibration
was involved, and thus indeterminate who was wronged.
17. See, for instance, Portmore (2011).
18. Urmson (1974, 114–15) does so insist.
19. The translation I’m relying on for this quotation and the next is Kant (2011).
20. Pogge (1997), for instance, argues that we can arrive at Kant’s universal prin-
ciple of justice, part of his Doctrine of Right, without starting at the Categorical
Imperative. But O’Neill (2012) contends that the principle is indeed supposed
to follow from the Categorical Imperative.
One might suggest, as a less extreme departure from classic monistic Kantian-
ism, that the various duties that Kant derives from the Categorical Imperative
are supposed to be grounds of verdictive facts and that the Categorical Impera-
tive is supposed to explain why those duties have their grounding power when
they do. But this is the alternative understanding of Kantianism I discuss and
discard in Appendix A.
21. The translation that I’m relying on here is Kant (1996).
22. Jens Timmermann helpfully distinguishes the question of whether Kantian eth-
ics allows for a duty to not do X to an animal from the question of whether
Kant’s ethics allows for a duty to an animal to not do X to it. Timmermann
(2005) argues for an affirmative answer to the former question. If his argument
works, it means Kantian ethics is consistent with the judgment that certain ways
of treating animals are in themselves to be avoided. In other words, Kantian
ethics wouldn’t have to fall back on the claim Kant himself makes that avoid-
ing treating animals in certain ways is a means to the end of avoiding treat-
ing human beings immorally. However, as Timmermann admits, his argument
doesn’t get us to the conclusion that Kantian ethics is consistent with the judg-
ment that certain ways of treating animals wrong them (or, in Kantian terms,
violate a duty to them). What I want to explore in this section is whether there
is some route to that conclusion.
23. Kant admits this and attempts to explain away the intuition. See his comments
on amphiboly in The Metaphysics of Morals 6:442.
24. One worry I have about Kantian ethics is that this problem may be more wide-
spread than generally recognized. Traditionally, the problem is supposed to be
that what appear to be violations of duties to animals are construed as violations
of a duty to oneself. So Kantian ethics appears a bit too self-regarding. My con-
cern is that Kantian ethics may be entirely self-regarding. In the Groundwork the
Categorical Imperative is derived from the idea of rationality. Kant says that act-
ing rationally means giving laws to oneself, and from this initial premise he goes
on to argue that rationality requires willing in accordance with the Categorical
Imperative. Thus the Categorical Imperative emerges as a self-imposed constraint
(see Korsgaard 2012, 18). Another aspect of Kantian ethics that should make us
suspicious is its insistence that it is only maxims, and not actions, that are to be
subject to the Categorical Imperative. One’s maxim is the principle on which one
decides what to do. So it appears that Kantian ethics, at least in the first instance,
judges our practical reasoning and not our actions (Herman 2016, 34).
60 Against Monism
Looked at this way, Kantian ethics seems to be a lot like virtue ethics, where
the basic idea is to offer a standard for being a good person and only deriva-
tively offer a standard for treating others well (or not offer one at all). (This
affinity between Kantian ethics and virtue ethics is noted by Baron (1997). The
version of Kantianism that seems most aptly described this way is the most
recent version of Korsgaard’s Kantianism (Korsgaard 2009), on which moral
constraints are a byproduct of one’s effort to constitute oneself as an agent.)
Whatever the merits of virtue ethics, a central weakness of that approach to
ethics is that, unless it is merely one element of a pluralistic normative ethical
theory, it isn’t consistent with the idea that some of the things we do wrong
others. According to Rosalind Hursthouse, virtue ethics is the view that wrong
actions are wrong because they instantiate a vice: “[W]hat is wrong with lying,
when it is wrong, is . . . that it is dishonest, and dishonesty is a vice. What is
wrong with killing, when it is wrong . . . [is] that it is callous and contrary to the
virtue of charity” (1999, 6, original emphasis removed). That an action instanti-
ates a vice is a fact about the agent, if it is a fact about anyone. It is certainly
not a fact about the individual to which the action was done. So if we cite the
viciousness of an action as explaining its wrongness, nothing we say picks out
anyone (except perhaps the agent) as the wrong-sufferer.
25. In an interesting footnote, Korsgaard offers the following analogy:
imagine a white male who claims that in valuing his own freedom he is
only valuing the freedom of white males: if, unknown to himself, he turned
out to be a black woman . . . then he would agree that his freedom doesn’t
matter. Our response would be that either he’s insincere or deceiving him-
self, that he’s suffering from a failure of reflective imagination.
(2005, 104)
26. Thanks to Marcia Baron for helping me to see this point.

27. In a similar vein, Martha Nussbaum worries that on Korsgaard’s Kantian account
there is a “strange indirectness about the route to animal welfare” (Nussbaum
2011, 241).
28. One might worry that Kantian ethics is by nature exclusive, due to its claim that
humanity (or the ‘good will,’ as it is referred to in certain key passages) is the
sole unqualified good. If this is our starting point, it seems we must reject any
moral constraint that doesn’t in some way reflect a concern for humanity. How-
ever, one might argue, as Onora O’Neill (1998, 220) has, that Kant isn’t really
committed to there being no other unqualified goods; he is simply committed
to humanity being one of them. This perhaps shows the way to constructing a
version of moral pluralism with Kantian ethics as one element.
29. Another effort along these lines is made by Cohen (2007; 2016). Space prevents
me from giving both Cohen and Rowlands a thorough examination, so I focus
exclusively on Rowlands here. Briefly, though, I think the problem with Cohen’s
argument is that it proceeds by way of the idea of moral status, a concept whose
usefulness I critique in Chapter 7.
30. Scanlon (1998, 12, 391 n.21). This aspect of Scanlon’s theory was brought to
my attention by Stratton-Lake (2003).
31. This strategy for incorporating animals into Rawls’s contractarianism was sug-
gested much earlier by VanDeVeer (1979) and Barry (1989, 204).
32. See §3.2 above.
33. Nicholas Southwood (2008, 185–6) argues that even on the Hobbesian version
of contractarianism there will be trouble explaining how one individual could
be accountable to another for moral violations; I will not discuss that possibility
here.
4 Against Rossian Pluralism
Moral monism is the view that there is just one kind of non-normative fact
that grounds all the verdictive moral facts. Moral pluralism, by contrast,
holds that there are multiple kinds. In the previous chapter I criticized
monism on the grounds that it seems unlikely to be able to offer a compel-
ling picture of how the verdictive facts are grounded. In this chapter my
main goal is to establish that the most popular version of pluralism is vul-
nerable to the same objection.
The most popular version of pluralism is the kind on which the verdic-
tive facts are directly grounded by other normative facts. Of course, like
nearly everyone, the people who accept this view also believe some version
of the claim that normative facts supervene on non-normative facts. There-
fore, assuming that the verdict-grounding normative facts don’t reduce to
non-normative facts, they must claim in addition that the verdict-grounding
normative facts are grounded in non-normative facts. What makes this
view pluralist is its further contention that there are different kinds of non-
normative fact that can ground the verdict-grounding normative facts.
Different versions of this view can be distinguished according to their
divergent views about which normative facts ground the verdictive facts.
One version is the view that the verdictive fact about an action is grounded
the moral reasons for and against performing it. This is the version on which
I will focus here. It is by far the most popular version of the view and mostly
likely other versions of the view can be made compatible with it.
I am going to reject this view on substantive grounds. This may seem odd,
as it seems to be a purely meta-ethical view. But this appearance is based on
a misunderstanding of normative ethics, which Chapters 1–2 were intended
to dissolve. I argued there that when theorizing we should understand nor-
mative ethical theories as offering competing views regarding which facts
ground the verdictive facts. Now the view I’ve been picking out doesn’t say
which facts ground the verdictive facts, and so I admit of course that it isn’t
a normative ethical theory. The view I’ve been picking out, however, does
purport to identify the type of facts that ground the verdictive facts, and
therefore marks out a family of normative ethical theories. In this chapter
I reject this family on the grounds that each member of it will exhibit a
serious shortcoming. I’ll label this family of views Rossian Pluralism (RP)
since Ross’s (1988) normative ethical theory, on which non-normative facts
ground prima facie duties and prima facie duties ground the verdictive facts,
is the most well-known member of the family.
In this chapter I offer a two-part argument for rejecting RP: RP is flawed
as a family of normative ethical theories (§1) and there are no meta-ethical
drawbacks to rejecting it (§2). The chapter concludes with §3.
1. RP as a Defective Family of Normative Ethical Theories
1.1 Introduction
H. A. Prichard and W. D. Ross put Rossian Pluralism on the map. They argued
that certain kinds of action (Ross said there were seven kinds1) are obliga-
tory. If a single action is of more than one kind, such that it constitutes both
the discharging of at least one obligation and the violation of at least one
obligation, the rightness or wrongness of the action is determined by the
combined degree of obligatoriness of the individual obligations (Prichard
1912, 29 n. 1; Ross 1988, 46, 123).
For a long time, Prichard and Ross were in a lonely minority. RP was sim-
ply not very popular. This is because pluralism per se was not very popular.
All the most prominent moral philosophers before Prichard and Ross were
moral monists. This includes Kant as well as the consequentialists Bentham,
Mill, Sidgwick, and Moore. Prichard and Ross’s pluralism—or what went
by the moniker ‘intuitionism’—seemed to be what you were left with if you
wanted to be a non-consequentialist but not a Kantian. Even after Ross’s
The Right and the Good was published, in 1930, monism remained the
dominant view throughout the mid-twentieth century. This was the era in
which consequentialism, and particularly utilitarianism, ruled the landscape.
So for a while, whether you were a Rossian pluralist depended mostly
on whether you were a pluralist, which in turn depended a lot on whether
you found consequentialism or Kantianism attractive. So pluralism, and
by extension RP, was motivated by certain substantive views about right
and wrong. And we know those substantive views were rather unpopular,
because there were very few Rossian pluralists.
Then something changed. A new version of RP—one in which the verdict-
grounding normative facts are reasons—was introduced to the literature
by Kurt Baier (1958, Chapter 3) and J. O. Urmson (1974). By 1989 Shelly
Kagan, in his book The Limits of Morality, was pronouncing RP (though
he didn’t use that label, of course) to be a “familiar picture of how moral
requirements are generated” (1989, 65). And even Kagan himself, a strict
consequentialist, endorsed it (1989, 65; 1988, 17–18)! Nowadays RP is
pretty much taken for granted, as I’m sure the reader has noticed. Act-
consequentialists accept it (e.g., Portmore (2011)); rule-consequentialists
accept it (e.g., Hooker (2002)); non-consequentialists accept it (e.g., Nagel
Against Rossian Pluralism 63
(1986, 203), Rachels (1999, 196)); even people who reject both consequen-
tialism and non-consequentialism—particularists—accept it (e.g., Dancy
(2004)). What happened?
The short answer, I think, is that philosophers started to get very inter-
ested in the nature of reasons. Indeed, the most challenging task for anyone
who wants to make a case against RP is showing that we can, while reject-
ing it, hold on to some very intuitive claims about reasons. I am going to
postpone that meta-ethical discussion until §2. For the rest of this section
I take on the comparatively easy task of showing that RP is defective as a
normative ethical theory.
1.2 The Too-Many-Grounding-Claims Problem

One of the jobs that a normative ethical theory is supposed to do is to holis-
tically explain (that is, explain as a set) the apparent verdictive facts, and
holistic explanations can be graded on how simple they are. This much is
familiar from Chapter 2. I also argued there that all else being equal we
should use the simplicity criterion to determine which normative ethical
theory to accept. Now I haven’t argued that all else is equal, and in fact I
think it isn’t; by the end of Chapter 5 I will have argued that RP is inferior to
one of its competitors. But in case those arguments fail to convince, I present
this argument against RP as a fallback.
In Chapter 2 I asserted that a normative ethical theory is simpler to the
extent that it contains a smaller set of type-grounding claims—i.e., claims
about how the various kinds of verdictive facts are grounded—and I argued
for holding normative ethical theories to a simplicity criterion (§4.2). Now
any version of RP, as I’ve defined it, will contain two sets of type-grounding
claims (since RP holds that there are two layers of grounding underlying
each verdictive fact): a set of claims regarding the grounding relationships
that non-normative facts bear to moral reasons, and a set of claims regard-
ing the grounding relationships that moral reasons bear to verdictive facts.
So we need to determine whether the simplicity criterion applies to just one
of the sets of type-grounding claims or to the two combined.
The key to making this determination is to recall what the motivation
for the simplicity criterion is. The motivation, I maintained in Chapter 2,
was the idea that one thing we want out of a normative ethical theory (and
I emphasize again, this desideratum should play a role only when all else
fails) is to help us to see the vast array of verdictive facts as stemming from
a relatively limited set of non-normative phenomena. If this is something
a normative ethical theory is supposed to help us to do, then I submit that
whether a version of RP manages to do it depends on how many type-
grounding claims are in its two sets of type-grounding claims combined.
This is because it takes both steps—the step from non-normative facts to
moral reasons and the step from moral reasons to verdictive facts—for any
version of RP to get us from the non-normative to the verdictive.
Let’s set aside the moral reasons verdictive facts grounding claims and
focus instead on the non-normative facts moral reasons grounding claims.
For those who maintain that non-normative facts ground moral reasons,
what’s the most reasonable thing for them to say regarding how many ways
there are for non-normative facts to ground moral reasons? I deny that non-
normative facts ground moral reasons (see §2.2), but I can certainly get
myself in the mode of thinking that way (if I couldn’t, I wouldn’t be able to
understand most contemporary published works on normative and applied
ethics!). And when I do, I find myself wanting to be quite generous about
what it takes for moral reasons to arise. I want to say, for instance, that
for each individual I have a moral reason to augment that individual’s wel-
fare. Furthermore, the advocates of RP themselves seem to hold that moral
reasons come cheap—that lots of non-normative facts are moral reason-
grounding (though many of the resulting moral reasons are very weak).2
Ross, for instance, held that we have a moral reason to augment others’
welfare (he called this the pro tanto duty of beneficence). They also tend
to accept, as well they should, that extremely small differences in the non-
normative facts can ground differences in moral reasons. This latter fact is
held to be true both for individual moral reasons and for sets of moral rea-
sons: Small differences in the non-normative facts can, presumably, ground
small differences in the strengths of individual moral reasons. And small
differences in the non-normative facts can ground differences in how sets of
moral reasons interact together to produce an overall reasons-based case for
doing something—this is one aspect of the popular ‘reasons holism’ thesis.3
(This means that insisting, as Ross does, that only a small number of kinds
of non-normative fact can be reason-grounding is only half the battle when
it comes to achieving simplicity.)
Given that reasons come cheap and that all manner of small non-normative
differences can affect the strengths of reasons and affect how they combine
to produce an overall reasons-based case for doing something, any version
of RP will have to contain an unimaginably large set of non-normative
facts  moral reasons grounding claims. So versions of RP are about as non-
simple as normative ethical theories can be,4 and thus fare miserably on the
simplicity criterion.5 The only way to avoid this defect is by being strict about
which non-normative facts can ground moral reasons and about the variation
in how moral reasons can combine to produce verdictive facts, but I wonder
what the argument could be for such strictness other than that it is necessary
for avoiding the objection I’m now raising.
1.3 The Directedness Objection

My second objection to RP concerns the directedness of morality—that is,
the fact that some wrongdoing is wronging. The question is whether RP can
allow for this feature of morality. Stratton-Lake (2011, 374–6) argues that
it can and puts this purported fact forward as a selling point of the theory.
I submit that it cannot. As I said in the previous sub-section, any ver-
sion of RP is going to have to include a set of claims regarding the moral
reasons  verdictive facts grounding relation. The important question for
our purposes is what these claims will say in cases—of which I argued
just now that there will be a lot—in which there are multiple moral rea-
sons relevant to the decision. What they will say, I submit, is something of
the form ‘Jones’s φ-ing was impermissible because the reasons supporting
Jones’s φ-ing were, collectively, weaker than the reasons supporting some
other course of action available to Jones’. But this is an entirely impersonal
explanation, and consequently it fails to identify a wrong-sufferer.
Now a Rossian Pluralist might make use of the gambit to which I earlier
imagined consequentialists and Kantians appealing—that of proposing that
the grounding relationship is transitive. If the relationship is transitive and if
any fact to the effect that the set of reasons supporting one action is stronger
than the set of reasons supporting some other action is always grounded in
the various reasons supporting the two actions and how they interact with
each other, then it turns out to be true that RP implies that the impermis-
sibility of Jones’s φ-ing is grounded in the various reasons for and against
that action.
However, a moral reasons  verdictive facts grounding claim of this sort
fails to identify a wrong-sufferer, for two reasons. First, it identifies too many
things as being relevant to the verdictive fact of the matter. The facts about
an action that, by transitivity, ground the verdictive fact about it are, on this
picture, all the moral reasons for (if it’s obligatory) or against (if it’s imper-
missible) performing it.6 But lots of individuals, aside from the one who we
might intuitively pick out as the victim of wrongdoing, might be referred to
by those reasons. There’s nothing in RP that picks out any particular one of
the individuals mentioned in the catalog of relevant moral reasons as being
the victim. Since wronging is a relationship between a wrongdoer and a
victim, an explanation of the sort I’ve described cannot be an explanation
of wronging.
Suppose, for instance, that φ-ing is impermissible, that one of the moral
reasons opposing it is that it harms Jones, and that another of the moral rea-
sons opposing it is the sadistic benefit it brings to Williams, who enjoys see-
ing Jones suffer. So Jones and Williams have equivalent ‘status,’ you might
say, in RP’s grounding explanation of the impermissibility of φ-ing: they are
both mentioned in the moral reasons that ground its impermissibility. Con-
sequently there is nothing in RP to pick out Jones, as opposed to Williams,
as the victim of the agent’s φ-ing.
Of course, maybe I was wrong to assume that there is a moral reason oppos-
ing the conferring of sadistic benefit. But we can construct a sort of mirror
image thought experiment, with just as troubling a result, on the assumption
that there is a moral reason favoring the conferring of sadistic benefit.7 Sup-
pose φ-ing is morally obligatory, that it’s the only way of keeping a promise to
Jones, and that it would confer sadistic benefit on Williams because it would
come at a painful cost to the agent herself, whom Williams despises. So, again,
Jones and Williams have equivalent status in RP’s grounding explanation of
the obligatoriness of φ-ing: they are both mentioned in the moral reasons that
ground the obligation. Consequently, there is nothing in RP to pick out Jones,
as opposed to Williams, as the victim of the agent’s failure to φ.
I have been objecting to the idea that a moral reasons  verdictive facts
grounding claim can identify a wrong-sufferer on the grounds that the left-
hand element of such a grounding claim, namely the set of reasons relevant
to the verdictive fact, will often mention many people and therefore fail to
pick out any one of them as having a special status that could reasonably
be construed as victimhood. I also have a second objection to that idea,
namely that the set of reasons will not mention people in the right way.
Suppose what I have a reason to do is ‘sing Jones’s favorite song’ and that
I’m obligated to do this as well. Is Jones a candidate for victimhood should
I wrongly fail to sing Jones’s favorite song? Not necessarily; it depends on
to whom the obligation was made. Maybe the obligation was to Williams,
because I promised Williams that I would sing Jones’s favorite song. So in
the case of an obligation, identifying a possible victim requires discovering
whom the obligation was to. But reasons aren’t to anyone (Liberman and
Schroeder 2016, 108), so if reasons ground obligations, as RP insists, then
there’s simply nothing to appeal to in an attempt to identify a possible vic-
tim of one’s failure to discharge an obligation.
1.3.1 A Response to the Objection

From conversations I’ve learned that most RP supporters want to escape this
objection by presenting a different theory about what makes some action
a case of wronging. I have been assuming that what makes φ-ing a case of
wronging is that φ-ing is wrong and its wrongness is grounded by the fact
that doing it would involve the agent coming to bear some objectionable
relationship (deceiving, harming, breaking a promise to) another individual.
An alternative is that what makes φ-ing a case of wronging is that by φ-ing
the agent would act contrary to a pro tanto duty or a patient-specific moral
reason (a moral reason to do or not do something to some particular indi-
vidual). Basically, the proposal here is to take my just-mentioned assump-
tion about wronging and subtract the part about wrongdoing; to say, that
is, that the fact that φ-ing constitutes wrongdoing is not part of what makes
φ-ing a case of wronging.
One possible problem with the new theory is that it leaves open the possi-
bility that there could be wronging without wrongdoing. But I am not going
to make that the grounds of my rejection of the new theory.8 First, in my
experience many RP supporters are happy to bite this bullet or don’t even
consider it a bullet to bite. Second, the RP supporter can avoid this implica-
tion by saying that while acting contrary to a pro tanto duty or a certain
kind of moral reason is what makes an action a case of wronging, it is not
all that needs to be the case for an action to be a case of wronging. It also
needs to be the case that the action is wrong. In other words, wrongdoing is
an enabling condition for wronging. (This might seem like an ad hoc move,
but I’m willing to let it go.)
1.3.2 A Rejoinder to the Response

Instead, I reject the new theory for the following reason. If an action’s being
wrong is not part of the explanation of its being a case of wronging, then
it should also not be part of the explanation of the various phenomena
that wronging brings into play. I said in the previous chapter that when an
individual is wronged a certain set of reactive attitudes become appropri-
ate. For instance, it becomes appropriate for the wronged party to resent
and demand an apology from the perpetrator. I contend that there is no
adequate explanation for these phenomena that doesn’t appeal to a wrong
having been done.
Certainly, however, many will dispute this contention; we can be sure of
this because there are several philosophers who hold that moral residue can
be explained without appealing to a wrong having been done, and the phe-
nomena we have been talking about—e.g., the appropriateness of resent-
ment and the demand for an apology—are aspects of moral residue. Moral
residue is a term meant to capture a set of phenomena whereby sometimes
even when one does as morality requires, one’s action leaves a morally sig-
nificant remainder. It seems that, sometimes, even when one does as one is
morally required to do it is appropriate to feel guilt or regret over what one
has done, and it also seems that sometimes one may be obligated to make
amends to someone for what one has done. These are two of the relevant
phenomena and, as mentioned already, the appropriateness of resentment
and the demand for an apology are two more.
What, then, are the available strategies for explaining moral residue with-
out appeal to wrongdoing? Since RP is the view that a moral requirement
to φ is grounded in the moral reasons for and against φ-ing, the obvious
move for the supporter of RP to make is to suggest that moral residue is
explained by the fact that sometimes when the balance of moral reasons
favors φ-ing there are also moral reasons favoring not-φ-ing, and conse-
quently one often acts contrary to moral reasons even when doing what one
is morally required to do. Oberdiek (2004, 333), for one, makes this move.
The problem, however, is that unfulfilled moral reasons don’t seem to be
per se the kind of thing that can give rise to moral residue. If it’s appropriate
for me to feel guilty over what I’ve done, or if I owe an apology to someone
for what I’ve done, it seems it would have to be the case that somehow I failed
to do what I should have done. It’s not enough that my action wasn’t perfect,
that it left something to be desired. Instead, it should be that something was
legitimately asked of me, and I failed to do it. But reasons (and this includes
moral reasons) do not per se make demands of us; they’re not deontic.9 It’s
merely in their nature per se to favor an action. To bolster this point, remem-
ber that if we are operating under the belief that moral reasons are grounded
in non-normative facts, as RP says, then it seems we should be quite liberal
about them. We should admit, for instance, that the fact that φ-ing would help
someone grounds a moral reason to do it. But surely a failure to act on such a
reason isn’t necessarily something that it’s appropriate to feel guilty about or
even something to regret, for that matter. If I pass up the opportunity to flash
a friendly smile at someone (as I do many times a day, every day), and doing
so would’ve brightened that person’s day, it would be odd if I came to regret
that (barring the existence of some other consideration in favor of doing it).
The response will have to be that only some moral reasons can, when
unfulfilled, give rise to moral residue. This suggestion is similar to what has
been, historically, the most popular way to explain moral residue: positing
that sometimes when we do as we are morally required to do we never-
theless violate a pro tanto obligation or duty.10 What the two suggestions
have in common is the idea that within the class of morally relevant non-
normative facts, only some are such as to ground morally normative entities
(moral reasons or pro tanto duties/obligations) that when unfulfilled can
give rise to moral residue.
To set up the discussion, let’s first ask whether there are such things as
pro tanto obligations or duties. It bears mentioning that not even Prichard
or Ross thought there were. Prichard spoke of “obligatoriness” but not of
“obligations”. Meanwhile, Ross said that a pro tanto duty, or a ‘prima facie
duty’ as he called it, isn’t a thing but rather
the characteristic . . . which an act has, in virtue of being of a certain

kind (e.g. the keeping of a promise), of being an act which would be
a duty proper if it were not at the same time of another kind which is
morally significant.
(1988, 19)
In other words, if an action has a characteristic that would make that action
morally required (or ‘duty proper’, as Ross would say) in the absence of any
other relevant considerations, we call that characteristic a pro tanto duty. But
this sounds exactly like what a present-day adherent of RP would say a moral
reason is!11 So the strategy of explaining moral residue by reference to pro
tanto duties, which Ross himself adopted (1988, 28), is at risk of being indis-
tinguishable from the proposal to explain it by reference to moral reasons.
To avoid this fate we need to part ways with Prichard and Ross and
say that a pro tanto obligation/duty really is a thing. This on its own is a
troublesome move to make, since the way we speak of obligations/duties
strongly suggests, as Liberman and Schroeder (2016, 107–8) have pointed
out, that there are no such things and that uses of the word “obligation”
are “just nominalizations of the is obligated to relation between agents and
actions”.12 But even allowing that it’s possible that a pro tanto obligation/
duty is a thing, the RP theorist still must tell us what thing it is and why we
should believe such things exist. She doesn’t need pro tanto obligations or
duties to fill the role of grounding the verdictive facts; she already has moral
reasons at hand to fill that role. To posit that, in addition to moral reasons,
there are pro tanto obligations or duties is to postulate a new entity simply
for the sake of explaining moral residue.13 So if we asked the RP supporter
what a pro tanto duty or obligation is, the only way she could say some-
thing helpful would be by saying that pro tanto duties/obligations are what
explain moral residue. Consequently, when the adherent of RP tells us that
moral residue arises in a certain case due to the presence of an unfulfilled
pro tanto obligation or duty, she simply tells us that moral residue arises in
that case due to the presence of something whose lack of fulfillment can give
rise to moral residue. And this is no explanation at all.14
This same objection applies to the other proposal we are considering—the
proposal that there is a sub-class of moral reasons such that leaving them
unfulfilled gives rise to moral reasons. The problem, again, is that we have
no independent way of getting a grip on which moral reasons those are. The
only possible identifying description of them would be: the ones that, if not
fulfilled, give rise to moral residue.
I conclude from all this that the RP supporter cannot show how to explain
moral residue, including the appropriateness of resentment and the demand
for an apology, without appeal to wrongdoing. But this was the core of the
RP supporter’s strategy for escaping the directedness objection to RP. The
idea was to show that even though RP explains wrongdoing in a way that
doesn’t pick out anyone as having been wronged, RP can still pick out a
wronged party (that is, the party who may appropriately feel resentful and
demand an apology) since the facts about who has been wronged are not
explained even in part by the facts about what wrongdoing has been done.
1.4 Summary
In this section I have identified two ways in which RP does a bad job of
explaining the verdictive facts. First, it fails miserably on the simplicity
criterion, because it must contain an unimaginably large number of type-
grounding claims covering the relation between non-normative facts and
facts about moral reasons. Second, given its picture of the grounding rela-
tion between moral reasons and verdicts, many cases of wronging will be
mischaracterized by RP as non-directed wrongdoing.15
2. What Meta-Ethical Bullets Must We Bite in Rejecting RP?

In this section I argue that there are no major meta-ethical drawbacks to
rejecting RP. I do this by discussing, separately, the two major apparent
advantages of RP and arguing in each case that we can hang on the advan-
tage while rejecting RP.
2.1 Reasons Ground Oughts and the First Apparent

Advantage of RP
I endorse
Reasons Ground Oughts: Overall (that is, all-things-considered) oughts,
requirements, prohibitions, etc. are grounded by reasons.16
There’s a very important reason why I start this section with an endorsement
of Reasons Ground Oughts. There are some, such as Joshua Gert (2005) and
John Broome (2013), who want to define reasons by reference to oughts. The
rejection of RP removes the most obvious way of defining ‘moral reasons’
by reference to moral oughts, namely by defining ‘moral reasons’ as the enti-
ties that ground moral oughts. So for those, such as Gert and Broome, who
deny that the concept of a reason is primitive and can be understood on its
own terms, there may arise a worry that rejecting RP causes us to become
unmoored—that it causes us to lose our grip on what moral reasons are. To
those theorists I offer Reasons Ground Oughts, the truth of which raises the
possibility of (but doesn’t require) defining moral reasons as the kind of things
that (can) contribute to overall oughts. Of course, this could not be a complete
definition of moral reasons, since there might be other kinds of reason that
contribute to overall oughts. Something, then, will have to be added to allow
us to distinguish moral reasons from other reasons. Here we can just add in our
favored theory of the distinctiveness of the moral point of view, for instance
a) that moral thinking, and only moral thinking, is subject to a universaliz-

ability constraint (R. M. Hare, Marcus Singer), or
b) that moral reasons arise from a distinctive kind of second-personal address,
which presupposes a certain sort of authority (Stephen Darwall), or
c) that moral considerations are categorical—that they apply to you
regardless of whether you set yourself the goal of behaving morally.
The other reason why my endorsement of Reasons Ground Oughts is very

important is that if I can show this endorsement to be consistent with the
rejection of RP, I will thereby undermine one of the strongest arguments for
RP. Let me explain.
Reasons Ground Oughts is an extremely popular thesis, and my impres-
sion is that much of the popularity of RP arises from how it nicely harmo-
nizes with Reasons Ground Oughts. If one accepts Reasons Ground Oughts,
as I do (Sachs 2008a), while rejecting RP, then one is in the uncomfortable
position of endorsing one key plank of the reasons-first program in moral
philosophy while rejecting another. You might want to say, “Look, either
reasons come first, or they don’t. How could it be that reasons come first
when it comes to overall oughts, but not when it comes to moral oughts?”
That’s a fair question. My answer proceeds in two steps. First, I want to
argue that it’s possible that Reasons Ground Oughts is true and RP is false.
Second, I want to show why we should think it is the case that Reasons
Ground Oughts is true and RP is false.
First, how is it even possible? Because, I respond, the conceptual con-
nection between reasons and overall oughts is tighter than that between
moral reasons and moral requirements. We cannot, I submit, understand
a question about what one ought to do overall other than as a question
about what one’s reasons support (McKeever and Ridge 2012). Acting as
one ought to overall just is acting on the balance of one’s reasons. By con-
trast, there is no such barrier to denying that a question about what one
morally ought to do is a question about what one’s moral reasons support.
Scanlon makes no conceptual mistake when he says, “judgments of right
and wrong . . . are judgments about what would be permitted by principles
that could not be reasonably rejected” (1998, 4) and “thinking about right
and wrong is, at the most basic level, thinking about what could be justified
to others on grounds that they, if appropriately motivated, could not reason-
ably reject” (1998, 5).17
Now, why think that it actually is the case that Reasons Ground Oughts
is true and RP is false? The key point, in answering this question, is that the
objections I raised against RP in §1 are explanatory objections. That is, they
are objections to the effect that RP tells a problematic story about how the
verdictive facts are grounded. The force of these objections is contingent on
our accepting what I tried to establish in Chapter 1—that one of the key
criteria on which we should judge a normative ethical theory is the quality
of its explanatory story. So the objections I raised against RP in §1 threaten
our ability to hang on to Reasons Ground Oughts only if (1) the objections
apply against Reasons Ground Oughts and (2) it matters whether they apply
(that is, it matters whether a theory about what one ought to do overall tells
a good explanatory story).
The more important of the two objections I raised against RP certainly does
not apply to Reasons Ground Oughts. I argued that RP is incompatible with the
directedness of wrongdoing. Similarly, Reasons Ground Oughts is incompatible
with the directedness of failing-to-do-what-one-overall-ought-to-do. This is
no problem, however, as failing-to-do-what-one-overall-ought-to-do is never
directed; it’s not something one does to another (as perhaps revealed by the fact
that there is no transitive verb in English for such a failure, while (by contrast)
in the moral case you can ‘wrong’ someone.)
As to whether it matters that a theory of the overall ought tell a good
explanatory story, I haven’t argued that it does. Of course, that’s not an
argument to the effect that it doesn’t; we’ll have to wait to see whether such
an argument could be made.
I have argued just now that not only is it possible that Reasons Ground
Oughts is true while RP is false, but that we have good reasons for thinking
that this is not only possible but actual. Of course, my argument is premised on
an assumption that the moral domain and the overall domain are distinct—an
assumption some reject (e.g., Tännsjö (2010)). I do believe they’re distinct,
though it is beyond the scope of my discussion here to give an argument to
that effect. I should say, however, that the conceptual distinctness of morality
and rationality seems obvious; it seems to follow directly from what I said
above regarding how we should understand questions about overall oughts
compared to how we should understand questions about moral oughts. Of
course, a single thing can be sorted under distinct concepts, so much more
needs to be said here. I’ve said more elsewhere (Sachs 2006, Chapter 2), and
having done that I’ll take my assumption as granted for the rest of this chapter.
2.2 The Legal Model of Normativity and the Second

Apparent Advantage of RP
2.2.1 Introducing Moral Rationalism

It is extremely plausible to suppose that there is a tight connection between
how one’s moral reasons stack up and what one is morally required to do.
If we accepted RP, this would be really easy to explain. RP includes the
claim that verdictive facts—of which moral requirements are one kind—are
grounded by moral reasons facts, so if we accepted RP we could simply say
that moral requirements are grounded in how the moral reasons stack up.
But if we reject RP we are thereby challenged to explain why there should
be a tight connection between moral reasons and moral requirements. How
serious of a challenge this is depends on just how tight the connection is
supposed to be.
I maintain that we can, while rejecting RP, explain this much: if one is
morally required to φ then the balance of one’s moral reasons does not favor
not-φ-ing. In other words morality cannot demand that we act contrary to
the balance of our moral reasons. I label this claim Moral Rationalism.
In §2.2.2–4 I show how Moral Rationalism can be explained in a way
that is consistent with the rejection of RP. I then ask whether we should have
wanted to be able to explain an even tighter connection between moral rea-
sons and moral requirements than the one posited by Moral Rationalism.
2.2.2 Introducing the Legal Model of Moral Normativity

I’ll begin by introducing a model of legal normativity. I should acknowledge
straight off, however, that it is controversial whether there is such a thing as
legal normativity; i.e., that there are legal oughts and legal reasons that are
not a species of moral oughts and moral reasons. (For the argument against
the existence of this domain, see Hershovitz 2015.) I want to suppose for
the sake of argument, however, that legal oughts and legal reasons do indeed
exist. If they do, surely one’s legal requirements ground one’s legal reasons,
not the other way around.18 In other words, to find out what one has a legal
reason to do, one looks to the laws to discover what one is legally required
to do. Granted, there are often non-legal reasons to do the things that there
are legal reasons to do. So, for instance, there is a prudential reason for
me to wear my seatbelt, as the law requires. But there are no legal reasons
aside from the ones grounded in legal requirements. In other words, actions
obtain their legal significance by playing a certain role in the legal system.
In addition to pointing out the model’s intuitive plausibility, I can say two
things in its support. First, it has not, to my knowledge, ever been disputed.19
Second, there don’t seem to be any reasonable alternatives to it, assuming
that there is indeed a separate domain of legal normativity. There cannot be a
domain of normativity without reasons. And if legal reasons aren’t grounded
in legal requirements, where do they come from? Could laws ground legal
reasons? This proposal would deliver nearly all the same implications regard-
ing which legal reasons we actually have as does the model I have been advo-
cating, on which laws ground legal requirements which in turn ground legal
reasons. The two theories would disagree on this issue only when there are
two laws that issue contradictory imperatives with regard to some particular
action, φ. In such a case, the normal course of action for the theoretician
or the practitioner (e.g., the judge) would be to determine which law takes
precedence (averting to a principle such as, perhaps, that federal law trumps
state law (to use an example from the U.S. legal system)), after which it
would then be taken as settled that the agent legally ought to do which-
ever action is required by the law that takes precedence. This squares nicely
with the view that to figure out which legal reasons we have we look to our
legal requirements. The alternative model of legal normativity, on the other
hand, suggests a somewhat bizarre and unfamiliar model of legal reasoning
in conflict cases. On the alternative model one would begin by inferring that
one has a legal reason to φ and a legal reason to not- φ, and then one would
weigh those reasons in order to determine what one legally ought to do. It’s
a mystery on what basis one would weigh those reasons.
The model of legal normativity, as I’ve been defending it so far, has
these two elements: (1) Legal requirements ground legal reasons, and (2)
there are no legal reasons other than those grounded in legal requirements.
I now wish to add a third element to the model: the claim that all legal
requirement-grounded reasons are of the same strength. The basic, and to
my ear plausible-sounding, idea of this third element is that the fact that “it’s
the law” counts equally strongly in favor of each action it favors.20
I now want to propose a Legal Model of Moral Normativity, which is
basically the claim that what is true of legal reasons, on the proposal just
made, is true of moral reasons as well. So the Legal Model of Moral Nor-
mativity is the claim that
a) moral requirements ground moral reasons, and

b) there are no moral reasons other than those grounded in moral require-
ments (though there are often reasons of other kinds to do the things
that one is morally required to do), and
c) all moral reasons are of the same strength.
If we accept the Legal Model of Moral Normativity, then Moral Rational-
ism follows. Whenever one is morally required to do something, one either
has only a moral reason to do that thing or one has a moral reason to do
something else as well but the two reasons are equally strong.
Should we accept it?21 Let’s begin with the first element.
2.2.3 The Argument for the First Element of the Legal

Model of Moral Normativity
One aspect of the purportedly tight connection between moral requirements
and moral reasons is the fact that if one is morally required to do something
then one has a moral reason to do it. I argued in §1 that we should not
explain this fact by claiming that moral reasons ground moral requirements.
So if we are not to write off this fact as an unexplained coincidence, we have
to say that moral requirements ground moral reasons.
That’s it—that’s my argument for the first element. I have nothing more
elaborate to offer. Scanlon is probably the most famous defender of the first
element, but I don’t want to commit myself to the contractualist outlook
that underpins his argument for it.22 Stephen Darwall has tried to argue for
it, but I don’t think he succeeds.23 The only thing else I can do at this point
is play defense. This is what I now attempt. Having done that, I will return
to the second and third elements in the next sub-section.
There are three common objections to the claim that moral requirements
ground moral reasons. The first objection is that if it is plausible to think
that something’s being valuable doesn’t ground reasons (e.g., to want it or
to try to obtain it) and that instead it’s the features of the thing that make
it valuable that ground the reasons,24 then by analogy it must be plausible
to think that something’s being morally required doesn’t ground the reason
to do it and that instead it’s the features of the action that make it mor-
ally required that ground the reason. I’m going to set aside that objection
here, for three reasons. First, quite frankly, I have nothing interesting to say
in response to it. Second, I have not committed myself to the just-stated
view about value. Third, Darwall (2010) has already offered a persuasive
response.
The second objection is that if we accept that moral requirements ground
moral reasons, then we end up with a disjointed picture of the morally good
person. On the one hand, we want to say that the morally good person acts
on her moral reasons. On the other hand, as Michael Smith (1994, 75–6)
emphasizes, it seems like moral fetishism to resolve to φ on account of the
fact that φ-ing is morally required as opposed to resolving to φ on account
of the fact that φ-ing constitutes, for instance, keeping one’s promise, the
taking care of one’s child, making amends for one’s earlier wrongdoing, etc.
We can’t say both of these things if we say that moral requirements ground
reasons.
To set up my response, I think we first need to pull apart two worries that
can easily get conflated. Is Smith saying that it’s moral fetishism to be moti-
vated by the thought that something is morally required, or that it’s moral
fetishism to be motivated by that fact while not being motivated by the fact
about the action that makes it morally required?25 It must be the latter. Kant
famously said that there is nothing unqualifiedly good except the good will,
by which he meant the will to do one’s moral duty for the sake of duty itself.
This of course is a very strong claim, but the weaker claim that there is
something commendable about the person who seeks to do right because it’s
right is not unreasonable (see Sliwa 2017 for a defense of this idea). I think
our worry would be in regard to the person who has this concern but not
the corresponding concern to, e.g., keep one’s promises, take care of one’s
child, make amends, etc.
Now consider the person who is motivated to avoid φ-ing both because
φ-ing is morally forbidden and because φ-ing would harm someone.26 Does this
person display a fetishistic attitude toward morality? Would we hesitate to
describe this person as morally good? Surely the answer to both questions is no.
The third objection is a worry about double counting.27 Suppose I am under
a moral requirement to φ in virtue of the fact that I promised to do so. Pre-
sumably promises generate moral reasons. But if being morally required to do
something also generates moral reasons, and I am morally required to keep
this promise, then somehow I have a second moral reason to keep my promise.
If we accept the second element of The Legal Model—that there are no
moral reasons other than the ones grounded in moral requirements—then
this objection is disarmed, because the promise itself would not directly
ground a reason. Let’s turn now to whether we should accept that second
element, as well as the third.
2.2.4 The Arguments for the Second and Third Elements

of the Legal Model of Moral Normativity
I propose to turn the double-counting worry on its head. If, as I noted in
§2.2.3, there’s an argument to be made for the claim that moral require-
ments ground moral reasons and there are no other valid objections to it
other than the double-counting worry, then we can use the double-counting
worry and the claim that moral requirements ground moral reasons as
premises in an argument for the second element of The Legal Model. That
is my argument for the second element.28 Admittedly, most people probably
have the intuition that certain non-normative facts, such as the fact that an
action would harm someone, themselves ground moral reasons. The second
element of The Legal Model renders it incompatible with that intuition,
and that certainly counts against The Legal Model. But the sting of not
being able to accommodate this intuition can be mitigated somewhat by
the fact that The Legal Model combined with NP implies that the morally
significant non-normative facts each ground moral requirements, which
themselves ground moral reasons. It can be mitigated further by conceding
that reasons come in different kinds.29 If we make this latter concession,
there is nothing preventing us from saying that non-normative facts, such
as the fact that an action would harm someone, ground reasons; it’s just that
the reasons in question won’t be moral reasons.
As to the third element, I also have a very simple argument. Whether
a reason of a certain kind can vary in strength depends on whether the
fact that grounds the reasons of that kind can itself vary in some way. For
instance, the fact that some action would serve my self-interest grounds a
reason to do it, and the reasons I have to do those various self-interested
things may well vary in strength since some of those actions serve my self-
interest more than others. By contrast, there are no dimensions along which
the fact that something is morally required can vary.
Of course, the fact that something is morally required is always grounded
in some other fact, and that other fact might have dimensions. (For instance,
the fact that φ-ing would harm someone might ground a moral requirement
not to do it, and the fact of causing harm has the dimension of causing a
certain amount of harm.) But that makes no difference to the strength of any
moral reason, since (given the second aspect of The Legal Model) that other
fact cannot ground a moral reason.
Furthermore, as I already noted, we can help ourselves to the claim that
the fact that φ-ing would harm someone can ground a non-moral reason
not to φ, and as long as we’re doing this we should help ourselves further to
the sensible claim that this reason will vary in strength depending on how
harmful the action in question is. Since some of those actions will also be
morally prohibited, we have a perfectly good explanation as to why there
might be stronger overall reasons to abide by some moral requirements than
there are to abide by others. So we need not reject the third element of The
Legal Model in order to save that bit of data.
Finally, we need not reject the third element in order to make sense of the
idea that the value of abiding by moral requirements can be explained. Scan-
lon, various rule-consequentialists and various contractarians have offered
explanations of this sort. All I have said in defense of the third element is
that the fact that something is morally required is a dimension-less fact. I
have not said that it is a fact that cannot be expressed in other terms. So for
instance we are free to say, if we want, that the fact that something is mor-
ally required is the fact that it would be demanded by any system of rules to
which no one could reasonably reject, and we could then give an account of
the value of acting in accordance with such rules.
2.2.5 Moral Rationalism Revisited

We began §2.2 with a challenge to explain the apparent tight connection
between moral requirements and moral reasons while rejecting RP. I’ve just
argued that this can be done if we interpret the tight connection thesis as
being equivalent to Moral Rationalism, which says: If one is morally required
to φ then the balance of one’s moral reasons does not favor not-φ-ing. The
question I want to address now is whether Moral Rationalism suggests a
tight enough connection between moral requirements and moral reasons.
Many theorists believe that if one is morally required to φ then the balance
of one’s reasons favors φ-ing. (Notice that this is a claim about the balance
of one’s reasons per se, not just one’s moral reasons.) The extant arguments
for this are based on the purported existence of a strong connection (in fact,
what is usually held to be a conceptual connection) between moral require-
ments and blameworthiness.30 The arguments generally take this form:
P1. If X is morally required to φ, then X would be blameworthy for

failing to φ.
P2. If X would be blameworthy for failing to φ, then the balance of X’s
reasons favors X’s φ-ing.
C. Therefore, if X is morally required to φ, then the balance of X’s rea-
sons favors X’s φ-ing.
If we accept the Legal Model of Moral Normativity, then, as I pointed

out in §2.2.2, whenever one is morally required to do something one either
has only a moral reason to do that thing or one has a moral reason to do
something else as well, but the two reasons are equally strong. The first sort
of case is compatible with the conclusion of the above argument, but the
second sort of case is not. To allow for the second sort of case, I propose
substituting for P1
P1*: If X is morally required to φ, then X will be blameworthy if she

fails to φ.
The difference between P1 and P1* is that P1 does, while P1* does not,
insist that if X is morally required to φ then if she fails to φ she must be
blameworthy for failing to φ. P1* merely insists that X must be blamewor-
thy. This allows her to be blameworthy for getting herself into a situation in
which φ-ing was morally required of her, as opposed to being blameworthy
for what she did once she found herself in it. This, I think, allows us to say
what we want to say about cases of the second sort, just discussed (assum-
ing there are such cases)—cases in which one has two moral requirement-
generated reasons of the same strength such that one cannot act on them
both: we want to say you can’t be blamed for abiding by one requirement
and not the other. But we want to say that you can be blamed for getting
yourself into that situation. (Then again, maybe it’s possible to blamelessly
end up in such a situation. If so, we would want to reject P1*. But we would
also be robbed of our grounds for supporting P1. Obviously, I would be
content with this turn of events.)
If we replace P1 with P1* then the above argument becomes invalid.
Yet, P1* still gets us a strong connection between moral requirements and
blameworthiness, and if we accept Moral Rationalism then we still get a
strong connection between moral requirements and moral reasons. I don’t
see what the drawback would be.
2.3 Summary
In this section I focused my attention on two claims about reasons that strike
most moral philosophers as true and seem to constitute much of the appeal of
RP, Reasons Ground Oughts and Moral Rationalism. I argued that we can hang
on to both even while rejecting RP. Importantly, however, in order to explain
Moral Rationalism while rejecting RP, I had to endorse The Legal Model of
Moral Normativity. So in order to have a fair discussion of the options, I am
obligated to fold The Legal Model of Moral Normativity into the view I favor
and defend the package as a whole. This is what I will do in Chapter 5.
3. Conclusion
In this chapter I argued, first, that there are two valid explanatory objec-
tions to RP. The first is that RP runs afoul of the simplicity desideratum by
containing an enormously large number of type-grounding claims and the
second is that RP provides bad atomistic explanation by being inconsistent
with the existence of directed wrongdoing. I then argued that we can hang
on to both of the claims that seem to underlie the widespread support that
RP enjoys—namely Reasons Ground Oughts and Moral Rationalism—even
while rejecting RP.
Notes
1. Though he ended up whittling the list down to five.
2. For a defense of the ‘reasons come cheap’ thesis, see Schroeder (2007, Chapter 5).
3. For a defense of reasons holism, see Dancy (2004, Chapter 6). I call the rea-
sons holism feature popular because although Dancy uses its truth as a central
premise in his argument for particularism, many non-particularists (known as
‘generalists’) accept it too (e.g., McKeever and Ridge 2006).
4. Indeed, isn’t that their whole point? What Rossian pluralists take themselves to
be getting right, I gather, is that just about any non-normative fact can make a
moral difference.
5. Russ Shafer-Landau has reminded me that even a theory that contains an unimag-
inably large set of non-normative facts  moral reasons grounding claims could
do well on the simplicity criterion, provided that it has a unifying explanation of
why the various non-normative facts ground moral reasons.
I concede that this is true. Whether this is damaging to my objection to RP
depends on whether there is a convincing unifying explanation to be found.
Shafer-Landau has suggested the one provided by Mark Schroeder (2007,
Chapter 1–2). Schroeder, who agrees that reasons come cheap (2007, Chapter 5),
holds that for each fact that grounds a moral reason (or, for Schroeder, is a
moral reason), r, there is the same explanation of why it does: that fact helps to
explain why the action that r supports is a way of promoting the object of one
of the agent’s desires (29). The problem for Schroeder is the following corollary
of this meta-explanatory claim: if φ-ing would promote the object of one of my
desires, then I have a reason to φ. This causes trouble because Schroeder has a
very expansive theory about what we desire. The problem isn’t that Schroeder’s
theory yields too many reasons, but rather that its yielding a lot of reasons
makes it inconsistent with many of our verdictive intuitions—specifically the
cases where we intuit that one person has a stronger overall reason to φ than
another person does. (This is exactly the sort of verdictive intuition that Schro-
eder designed his theory to accommodate; see the first two pages of Schroeder
2007). If reasons come as cheaply as Schroeder believes, then the reasons each
of us have are likely to coincide to a great degree. (I borrow this objection from
Brian McElwee.)
6. Except (1) when some of the moral reasons for or against the action are exclu-
sionary reasons—reasons whose presence prevents certain other reasons from
counting in favor of or against the action (thanks to Russ Shafer-Landau for
reminding me of this)—and (2) if Gert (2005) is right and some against an
action reasons don’t to contribute to making that action impermissible while
some reasons for an action don’t contribute to making that action permissible.
7. I won’t worry about constructing a thought experiment on the assumption that
there is neither a moral reason favoring nor a moral reason opposing the con-
ferring of sadistic benefit, since I doubt any RP supporter would make this
assumption.
8. For what it’s worth, Owens (2012, 44) says that all wrongings are wrongdoings,
though he doesn’t argue for this.
9. Liberman and Schroeder (2016, 107) and Shpall (2014, 158–61) agree.
10. Hare (1981, Chapter 2), Conee (1987, 241), McConnell (1996, 39–44), Brink
(1994, 220–3), and Dahl (1996, 94).
One strategy that used to be popular is to suggest that moral residue can be
explained by the fact that sometimes when we do what we are morally required
to do we nevertheless harm someone (McConnell (1996, 161), Conee (1987, 242),
Hill (1996, 193–4)). But this, to me, seems like a sub-species of the broader pro-
posal that sometimes when we do what we are morally required to do we never-
theless act contrary to moral reasons, since RP supporters will say that we have
a moral reason to avoid harming people but also moral reasons stemming from
other sources. So this proposal doesn’t merit a separate discussion, in my judgment.
11. David Brink (1994) quite explicitly translates Rossian claims about pro tanto
duty into the language of reasons.
12. We should say, similarly, that uses of the word ‘duty’ are nominalizations of the
is duty-bound to relation between agents and actions.
13. Russ Shafer-Landau suggested to me that a pro tanto duty is an invariably rel-
evant moral reason, while Joseph Raz is known for cashing out the supposed
duty to obey the law in terms of an exclusionary reason. If pro tanto duties are
invariably relevant moral reasons or exclusionary reasons, then postulating pro
tanto duties doesn’t amount to postulating a new entity. But by the same token
it fails to provide a good explanation of moral residue, because, as I already
argued, reasons can’t explain moral residue.
14. A different strategy, proposed by Shpall (2014), is to explain moral residue by
positing that sometimes when we do as we are morally required to do we never-
theless violate a commitment. This is in one way better than the pro tanto duty/
obligation strategy, since we have an independent grip on what a commitment
is (everyday normative talk includes talk of ‘commitments’). But it is in another
way worse, because it doesn’t hold out the promise of explaining the whole
range of cases in which moral residue arises. As Shpall more or less admits,
becoming morally committed to something requires an act of will; I must make
a promise, sign my name, take a vow, etc. Yet, there are cases in which moral
residue arises for some agent in spite of his not having engaged in any relevant
act of will. Suppose, for instance, that I face a situation in which I must either
harm A or harm B. If I harm A, I might owe an apology to her, despite the fact
that I never engaged in any act of will that committed me to refraining from
harming her.
15. I’d be remiss if I didn’t mention what I take to be a third compelling objection
to RP. (I omit a full discussion of it here since my goal is to demonstrate the
implications for normative ethics of the methodology laid out in Chapter 2. The
objection I lay out in this footnote has nothing to do with the points I made in
Chapter 2, since it has nothing to do with moral explanation.)
RP, like consequentialism, will have difficulty accounting for the (apparent)
fact that in some situations there is nothing that the agent is morally required to
do. Presumably versions of RP will say that an action is morally required just in
case it is better supported by the moral reasons than all the other actions. Any
version of RP that seeks to avoid implying that there are very few situations in
which there is nothing that one is morally required to do will have to be quite
strict about which non-moral facts ground moral reasons, thus yielding the con-
clusion that there are lots of situations in which the agent has no moral reason
to do anything. I have already expressed skepticism about the plausibility of
this move (which also came up in connection with my first objection to RP).
The other way is to tell a more complex story about how moral reasons come
together to generate moral requirements. This is the route Douglas Portmore
takes (2011, Chapter 5). Similarly, Joshua Gert (2005) tells a complex story
about how reasons per se come together to generate requirements per se.
16. Pekka Väyrynen, in conversation, raised the worry that once I commit myself
to the claim that reasons aren’t deontic, as I did when discussing moral residue
(§1.3.2), it’s no longer available to me to say that reasons can ground oughts,
since oughts are deontic.
In response, I want to admit that, yes, I am committing myself to the idea that
something deontic (an ought) can be grounded by something non-deontic (a
reason). But this should be entirely uncontroversial; everyone who thinks that
moral facts have non-moral explanations accepts this. And my admission
that moral reasons can ground deontic phenomena is no threat to my earlier
assertion that moral reasons cannot ground moral residue. That latter assertion
wasn’t based on any claim to the effect that moral residue is a deontic phenom-
enon. Rather, it was based on a claim that moral residue clearly results from
some deontic phenomenon; that is, moral residue is clearly (what we might call)
a secondary deontic phenomenon.
17. Darwall points out that if the concept of moral obligation were the same as
the concept of being best supported by the moral reasons, then supererogation
would be a conceptual impossibility. Clearly, however, it is not. Darwall (2010,
147–8).
18. Although I do not intend to defend this theory of legal normativity here, I want
to take this opportunity to defuse one likely objection to it. This theory of legal
normativity, like RP, is a claim about grounding. It is not a conditional. If it were
a conditional, there would be serious trouble for it. It would be the claim: if one
is legally required to φ then one has a legal reason to φ. This claim is subject to
all sorts of good counterexamples having to do with unjust laws. This theory of
legal normativity, since it is a type-grounding claim, merely says that under certain
conditions something’s being legally required can ground a reason to do it.
19. Of course, it would be discarded by anyone who rejects the existence of a sepa-
rate domain of legal normativity.
20. One might wonder why, then, it is appropriate for some violations of the law to
be punished more severely than others. (Thanks to John Broome for alerting me
to this question.) It wouldn’t be appropriate, I admit, if the sole legitimate pur-
pose of criminal punishment were a legal purpose—namely upholding respect
for the law as such. And I concede that punishment requires a purely legal justi-
fication (i.e., punishment is justified only in cases of law-breaking). But justifica-
tion is merely about when it’s OK to punish; it leaves aside the separate question
of how to punish when it’s OK to punish. To answer that latter question we
need to inquire as to the legitimate purpose(s) of punishment, and we may find
that punishment has a legitimate non-legal purpose—for instance, promoting
the social good, in which case it we would have an argument for the legitimacy
of using punishment for deterrence and prevention. And the deterrence and
prevention motives could explain why it’s appropriate for some violations of the
law to be punished more severely than others—viz, some violations of the law
are more important to deter and prevent than others.
21. One might worry, I suppose, that if legal and moral normativity work in the
same way, then law and morality won’t be distinguishable at all. But there are
several ways to distinguish moral and legal norms; see Hart (2012, 167–80).
22. Scanlon (1998, Chapter 4). Shafer-Landau (2003, Chapter 8) also endorses this
claim. But he doesn’t argue for it; he simply argues for the weaker claim that
if you are morally required to do something then you have a reason to do it.
Broome (2004, 53), meanwhile, says that moral requirements ground reasons,
but he doesn’t say specifically that they ground moral reasons, and I think he
probably would resist that claim. Finally, Temkin argues (2012, §7.4), in an
interesting discussion of intransitivity, that rightness is a good-making feature of
actions and I think endorses, toward the end of that discussion (211), the appar-
ent corollary that rightness is a reason-grounding feature of actions.
23. The ideal way to test its truth would be to deploy the method of difference:
construct two possible worlds, both containing some individual, X, such that
the two worlds are alike in all respects except that in only one of them X is
morally required to φ. Then one would examine those two worlds in order to
see whether X contains a moral reason to φ in the one world that she lacks in
the other. Unfortunately, we can’t do this. Given the supervenience of the moral
on the non-moral, two worlds that differ in terms of what moral requirements
they contain must also differ in some non-moral respect. But we might hope
to deploy a less pure but still effective version of the method of difference, as
Darwall does (2010, 155–6).
This is Darwall’s argument: In one case, I “intentionally tromp on your foot
to gain an advantage in getting to the table with designer underwear on sale
when I could easily have gone around and risked doing without a pair of Cal-
vin Kleins.” In the other case, I intentionally tromp on your foot, but this time
because it “is the only way I can avoid significant injury to myself.” Let’s grant
that I violate a moral requirement in the first case but not the second. So the
desired difference is there. And presumably we all have the intuition that I had a
stronger reason to do otherwise in the first case than I did in the second. So far
so good. But this is not a pure case since there is a second difference: the bur-
densomeness of doing otherwise. This allows us, if we are committed to denying
that moral requirements ground reasons, to insist that the burdensomeness in
the second case attenuated the strength of the reason I had in that case not to
step on your foot, and that it did so directly—that is, not by undermining what
otherwise would have been a moral requirement.
24. Many theorists believe that a thing’s being valuable just is its having some fea-
ture (other than its value) that provides reasons. The locus classicus for this view
is Scanlon (1998, 97).
25. Both readings are left open by what Smith actually says.
26. Remember, The Legal Model of Moral Normativity allows that we can have
non-moral reasons to do what we are morally required to do. So it can allow
that being motivated by the fact that φ-ing would harm someone counts as
being motivated by a reason; we just have to concede that facts about harm
ground (or are) non-moral reasons.
27. See Dancy (2004, 16–17) and Stratton-Lake (2011, 370–1).
28. Another, probably less convincing to most people, consideration in favor of
accepting the second element is this: We tend to think that we all have a moral
reason to prevent someone’s φ-ing if her φ-ing would be wrong. Suppose that
what grounds that reason isn’t the fact that the agent’s φ-ing would be wrong
but rather that it would constitute [inflicting harm, or breaking a promise, or
telling a lie, or . . .]. The problem with adopting such a view is that it has the
implication that we have moral reasons to tame nature, since animals in the wild
are always harming each other. For instance, predators kill and parents abandon
their young. If, on the other hand, what grounds the prevention reason is the
fact that φ-ing would be wrong, then there are no implications for intervening in
nature, since when animals do these things they do no wrong. So we should say
that the fact that an action constitutes [inflicting harm, or breaking a promise,
or telling a lie, or . . .] grounds no moral reason for a bystander to prevent it.
Given this, it seems a bit implausible to assert that such a fact does ground a
moral reason for the agent herself not to do it. It would be odd if moral reasons
for agents and moral reasons for bystanders failed to run in tandem.
29. I argue for this claim in Sachs (2006).
30. See, for instance, Darwall (2006, 28) and Portmore (2011, 43–4) for this kind of
blameworthiness-based argument and for citations of other similar arguments.
5 Non-Rossian Pluralism
1. Introducing Non-Rossian Pluralism

Having set aside monism and Rossian pluralism, we are left with non-Rossian
pluralism (NP), which is the view that all the verdictive facts are grounded
by non-normative facts and there are multiple kinds of non-normative fact
that can do such grounding work.1 Informally, NP is the view that when
an action is impermissible, this is because it is a lie, or harms someone, or
something like that; and when an action is obligatory, this is because it’s
the keeping a promise, or the fulfilling of some role-based obligation, or
something like that.
This is a resoundingly commonplace way of thinking about what makes
things impermissible, obligatory, etc. In general, when conscientious people
think about moral matters or discuss them with others, they advert to non-
normative facts like these, and typically don’t advert to facts about how the
moral reasons stack up. NP allows that when an agent makes her choices
based on her awareness of such non-normative facts, she often thereby suc-
ceeds in doing the right thing in light of the fact about it that makes it
right. Thus, NP allows that a certain deliberative activity—an activity that is
central to our conception of what it takes to be a morally good person—is
not the exclusive province of philosophers. Furthermore, surely even we phi-
losophers still find token-grounding claims along the lines of what made the
rapist’s action wrong was the harm it caused the victim intuitively compel-
ling. Some versions of NP will contain token-grounding claims like this, and
will thus offer intuitively plausible atomistic explanations, thereby avoiding
one of the two problems that, I argued in Chapter 3, beset consequentialism.
Moreover, NP avoids the other problem for consequentialism, which was
shown in Chapter 4 to also be a problem for RP, namely incompatibility
with the directedness of wrongdoing. There is no reason that a version of
NP couldn’t contain token-grounding claims that mention particular indi-
viduals. For instance, the one just given mentions the rape victim, and thus
implies that she was wronged.
Finally, NP avoids the other problem for RP, namely doing hopelessly
poorly on the holistic explanatory criterion of simplicity. No version of NP
earns top marks on this score, no doubt—only moral monism could do that.
But it’s possible for a version of NP to at least do well. This isn’t possible
for any version of RP, on account of the fact that reasons seem to come
cheap. By contrast, if we put ourselves in the mode of thinking that non-
normative facts can directly ground verdictive facts, we’re not likely to be
tempted to endorse any similar ‘cheapness’ thesis—that is, a claim that there
are innumerable kinds of non-normative fact that can ground verdictive
facts. To drive home the contrast, consider that some versions of NP might
include the type-grounding claim the fact that an action is the telling of a
lie can make it wrong, whereas versions of RP would have to give a differ-
ent explanation of nearly every case of wrongful lying, since such cases will
differ in terms of precisely which moral reasons count for and against lying
in those cases.
Given that NP escapes all the explanatory objections that plague RP and
consequentialism, on what grounds could it be argued that we should not
accept the view? First, there is the question of whether NP, unlike Kantian-
ism and contractualism/contractarianism, can say something sensible about
cruelty to animals. I will argue in Chapter 7 that it can. Second, there is
the question of content. Particular versions of NP might be objectionable
because of the grounding claims they contain. But in this chapter I make
no commitments regarding what grounding claims the best version of NP
would contain; I aim to defend NP only qua family of normative ethical
theories.
Third, there is the matter of structure. In the rest of this chapter I want
to focus on whatever objections there might be to the structure of NP as
opposed to its substance and content. NP is defined structurally; it is the
family of normative ethical theories that holds that verdictive facts are
grounded by different kinds of non-normative facts. This on its own leaves a
number of other structural questions unanswered. However, I have already
taken a stand on three of them—namely the three elements of The Legal
Model of Moral Normativity.
From here forward when I refer to ‘NP’ I refer to the version of that
theory that has The Legal Model built into it. I can think of four things that
might seem troubling about NP so defined.
The first is that NP seems to provide a drastically oversimplified picture
of how various actions come to be impermissible, obligatory, etc. Could it
really just be that, say, if an action constitutes the telling of a lie, this estab-
lishes that it’s impermissible? Or if an action constitutes the only way to
keep a promise, this establishes that it’s obligatory? NP would be extraor-
dinarily dubious if it had such consequences; however, it does not. NP is a
theory about the grounds of verdictive facts; it is not a theory about the suf-
ficient conditions for the obtaining of verdictive facts. NP tells us that when
an action is impermissible (for instance), this is because it possesses some
non-normative property, such as being the telling of a lie. It does not tell us
that the instantiation of some single non-normative property is ever suffi-
cient for it to be impermissible. The presence of an enabler or the absence of
a disabler might also be necessary (see Chapter 2).
Non-Rossian Pluralism 85
The other three natural objections to NP are more complex and each will
be discussed in turn over the course of the next three sections.
2. The Genuine Moral Dilemmas Objection

Because NP says that there are multiple kinds of non-normative fact that
can ground verdictive facts and also says (in accordance with the Legal
Model of Moral Normativity) that all moral requirements ground moral
reasons, that there are no other moral reasons aside from the requirement-
grounded ones, and that all moral reasons are of equal strength, NP seems
to imply that there are situations that take the following form: one is mor-
ally required to φ and has a corresponding reason of strength x to φ, and
one is also morally required to not-φ and has a corresponding reason of
strength x to not-φ, and there are no other moral reasons either to φ or to
not-φ, and therefore the overall moral reasons in favor of these two courses
of action are equally strong. So it seems morality can give us two incompat-
ible moral requirements of equal strength. Such a situation is known as a
genuine moral dilemma.
This is troubling. And in fact, avoiding a picture of the moral domain
on which moral conflicts have this shape constitutes a strong motivation—
perhaps the quintessential motivation—for RP. As Broome (1999, 412) puts
it, “Weighing is just what reasons are made for.”
There are well-known problems for theories that allow genuine moral
dilemmas, having to do with their apparent violation of ought-implies-can and
other intuitively plausible tenets of deontic logic (see Brink 1994). But I am
going to set aside these concerns, for two reasons. First, they’re not particular
to NP. Some versions of RP will allow for genuine moral dilemmas—namely,
those that allow for the existence of trumping reasons. (For instance, if there is
(as an RP supporter might plausibly think there is) a trumping reason against
choosing to end the life of one’s child, then in Sophie’s Choice Sophie faces a
genuine moral dilemma.) Second, it would be an enormous undertaking to try
to refute all the arguments against the possibility of genuine moral dilemmas,
and I don’t consider myself up to the task.2
So I’m going to proceed as if the problem for NP is simply that it’s incon-
sistent with the widespread belief that genuine moral dilemmas are rare.
Many would allow, I suspect, that some of what we would, in ordinary
discourse, refer to as ‘tragic’ dilemmas are genuine moral dilemmas, but at
the same time deny that there are genuine moral dilemmas outside tragic
contexts. NP doesn’t seem to allow for this.
I have two things to say by way of response to this objection to NP.
2.1 The First Response to the Objection

First, this counterintuitive implication of NP comes hand-in-hand with a
theoretical virtue: it allows us to say what we want to say about moral resi-
due. If, for instance, I break a trivial promise, and I do so in order to avoid
telling a big lie, we would think that I would owe something to the person to
whom I broke the promise. NP has a wonderful explanation for this: some-
times even when one does as morality requires one also in so doing violates
a moral requirement, because the situation one faced was a genuine moral
dilemma. Surely, it is appropriate to feel regret or guilt any time one violates
a moral requirement, and surely one owes an apology to any individual
whom one has wronged in violating that requirement, and so the presence
of a genuine moral dilemma is a good explanation for moral residue.3
The ability of NP to explain moral residue wouldn’t count for much if
moral residue could be accounted for given RP.4 But I already argued, in
§1.3.2 of Chapter 4, that it cannot. Given the inability of RP to offer a good
explanation of moral residue, it really does matter that NP can do so by
positing genuine moral dilemmas. The problem for RP, recall, is that its most
plausible strategy for explaining why moral residue might be generated in a
case in which an agent does as she is morally required to do is that in some
such cases the agent at the same time violates a pro tanto duty, and yet we
have no independent grip on what a pro tanto duty is. By contrast, we do
have an independent grip on what a moral requirement is. Moral require-
ment is one of the fundamental moral categories; it constitutes part of the
conceptual core by reference to which all other moral concepts are situated.
So it is quite meaningful to say that some piece of moral residue obtains
because of the violation of a moral requirement.
This, of course, doesn’t mean that there’s nothing to be said against posit-
ing genuine moral dilemmas. It simply shows us that each normative ethical
theory must find some way of navigating the twin threats of countenancing
(a lot of) genuine moral dilemmas and failing to explain moral residue. My
theory goes to the extreme of entirely avoiding the latter problem while
possibly leaving itself vulnerable to the former (more on this in §2.2). But
every normative ethical theory must take some position or other along this
spectrum, and there’s no position to be taken that doesn’t have a cost.
2.2 The Second Response to the Objection

Second, strictly speaking NP says nothing about how frequent moral dilem-
mas are. NP is merely the claim that there are plural non-moral facts that
can directly ground verdictive facts. It is not a claim about which non-moral
facts can directly ground verdictive facts, and consequently it implies noth-
ing about how frequent moral conflict is. As for moral dilemmas, all it says
is that however many moral conflicts there are, every one of them is a genu-
ine moral dilemma.
Of course, this worry about countenancing too many genuine moral
dilemmas should rightly play a role in one’s ultimate decision about whether
to accept NP. For all I say in this chapter it may yet turn out that NP suc-
cumbs to this objection. We will know this to be the case if, in our effort
to avoid countenancing a lot of moral dilemmas, we are forced to adopt
a version of NP that is highly implausible in some other way. To lay my
cards on the table, I think we should endorse a version of NP that says
that there are rather few facts about an action that can make that action
impermissible or required. This is something I actually want to say indepen-
dently of the fact that it helps NP to avoid the worry about genuine moral
dilemmas. For instance, I hope to be able to persuade the reader that it is
a strength in its own right for a normative ethical theory to deny that an
action can be required in virtue of its constituting the distributing of welfare
in a certain way (Chapter 6) or in virtue of its constituting a failure to rescue
(Chapter 8).
3. Deliberative Rationalism
The third objection to NP concerns moral deliberation. NP holds that facts
about moral reasons don’t ground verdictive facts. If this were true, then it
would be hard to imagine that thinking about one’s moral reasons could
be a way of deliberating about what one morally ought to do. I expect this
apparent implication of NP will strike many people as highly counterintui-
tive. Many will want to say, I suspect, that if you want to figure out what
you morally ought to do then thinking about your moral reasons is precisely
what you should do. I’ll call this view Deliberative Rationalism. In this sec-
tion I aim to offer an independent argument (an argument that doesn’t use
NP as a premise) that Deliberative Rationalism is false.
3.1 The Initial Argument and a Concession

The problem for Deliberative Rationalism is that when it comes to deter-
mining the strengths of moral reasons, we have no clue what we’re doing.5
And this, informally, is my objection to it. (I’ll give my official version of the
objection later.)
I should begin by qualifying what I just said. Urmson (1974, 119) suggests
we can appeal to intuition to weigh up reasons, and I admit that this is true.
In really simple cases, such as a case where the choice is between killing an
innocent person and breaking a trivial promise, we can generate intuitions
about the comparative strengths of the reasons at hand. But in that case
it would be pointless to use them. In such a case one will have a verdic-
tive intuition in which one is confident on reflection, and this intuition will
inevitably yield the same verdict as one would have reached by consulting
one’s intuition about the comparative strengths of the reasons and inferring
a verdict on the basis of them. So one might as well just rely on one’s verdic-
tive intuition in determining what one is morally required to do.
By contrast, I submit that in hard cases (i.e., cases about which one does
not have a verdictive intuition in which one is confident on reflection),
where it would be quite helpful to have intuitions about the comparative
strengths of the reasons at hand, we cannot generate such intuitions. Since
this is a psychological claim and I’m not aware of any empirical studies on
the matter, all I can do to support the claim is invite the reader to introspect:
Imagine you faced a situation in which you had promised a friend you
would pick her up in the airport. However, on your way there you encounter
some road construction, which you knew about but had forgotten to factor
in to your decision about when to set off for the airport. If you follow the
detour, you’ll be very late. Your other option is to drive through the blocked-
off zone, thus endangering and frightening the construction workers on-site.
Now ask yourself, which moral reason is stronger: the moral reason favor-
ing following the detour, grounded in the well-being of the construction
workers, or the moral reason favoring ignoring the detour, grounded in your
promissory obligation to your friend?
My conjecture (based on my own introspection) is that when we ask our-
selves questions like this we answer them inferentially. First, we ask ourselves
what would be the right thing to do in the situation, and then we extrapolate
from that an answer to the reason-strengths question. (Of course, a procedure
as simple as this wouldn’t be available for any case in which more than two
moral reasons were involved. More on this later.) I presume the reader’s own
introspection yielded the same result.
So here’s the situation: In simple cases, having an intuition about the rela-
tive strengths of the reasons would be of no help because one’s intuition on
this matter will always mirror one’s verdictive intuition, and so one might
as well just appeal directly to one’s verdictive intuition. And in the hard
cases we don’t have intuitions about the relative strengths of the compet-
ing reasons. To put this colloquially: In hard cases, if the right thing to do
comes down to how the reasons on either side tot up, then there had better
be quantities to tot. But there never are, for all our intuitions can tell us.67
3.2 The Uselessness of Inferential Beliefs About

Reason-Strength
In order to show that thinking about one’s moral reasons is a way of think-
ing about what one morally ought to do, one needs to show that one of the
two allegations I made above—i.e., that thinking about one’s moral reasons
is useless in easy cases and that thinking about one’s moral reasons is useless
in hard cases—is false. I assume that the better strategy is to challenge the
latter, since the former seems obviously true. What, then, is the best thing for
a supporter of Deliberative Rationalism to say regarding deliberation about
hard cases? What she would say, I assume, is that we can work in both direc-
tions. Specifically, when trying to determine what one morally ought to do
in some hard case, A, we can
a) identify the moral reasons relevant to A: N1, N2, N3, . . ., NX, then
b) identify other situations—B, C, D, . . ., Z—in which the same reasons
apply, then
c) appeal to verdictive intuition to determine what one ought to do in B,
C, D, . . ., Z, then
d) use that information to determine the relative strengths of N1, N2,
N3, . . ., NX, then
e) use that information to determine what one morally ought to do in A.
Sometimes we know a verdictive fact and can use that knowledge to infer
facts about the relative strengths of certain reasons. (This is precisely how
Joshua Gert (2005) suggests we can discover the relative strengths of rea-
sons.) Then we can take that knowledge and work in the other direction,
inferring further verdictive facts from the just-discovered facts about the
strengths of various reasons.
This is sound methodology, I admit. (We might call it ‘lateral’ moral rea-
soning.) But it doesn’t get us what we want. Moral reasoning isn’t supposed
to merely fill in the blank spots in our verdictive belief-set; it’s supposed to
be a way of correcting one’s mistaken verdictive beliefs if one has any. And
this is exactly what we deliberators lack if Deliberative Rationalism is true
and we don’t have intuitions about reason-strengths in hard cases. Without
those intuitions, we don’t have a second set of moral inputs in hard cases
(as opposed to easy cases, where in addition to verdictive intuitions we have
reason-strength intuitions). Moral reasoning becomes a matter of figuring
out how much information one can extract from one’s verdictive intuitions
plus the non-normative facts. But if one reasons in this way, then one will
never be pressured to alter or change any of one’s pre-existing verdictive
beliefs. And that, officially, is my objection to Deliberative Rationalism. In
other words, I reject Deliberative Rationalism because it is incompatible
with the idea that forms the basis of this book—the idea on which the Intro-
duction and Chapter 1 were premised: that moral theorizing can be a way
of discovering that one was mistaken about what one morally ought to do.
3.3 Turning the Tables

To recap: If NP is true and in morality the direction of deliberation is the
same as the direction of grounding, then one picture of moral deliberation
that cannot be true is Deliberative Rationalism. This was supposed to be a
reductio of NP, because Deliberative Rationalism is so appealing initially.
But I’ve argued that Deliberative Rationalism is false.
Now we should ask the natural follow-up question. Never mind what
picture of moral deliberation NP rules out, what picture of moral delib-
eration does NP rule in? Does it fall prey to the same problem that beset
Deliberative Rationalism, namely that when using it one will never be pres-
sured to discard a previously held verdictive belief? I criticized Deliberative
Rationalism on the grounds that since we do not have beliefs about reason-
strength in hard cases that are supportable independently of our verdictive
beliefs, we will never be forced to correct any of our verdictive beliefs if we
deliberate in the way that Deliberative Rationalism suggests. Similarly, it
may appear that the non-Rossian pluralist has an easy way to escape having
to accept any verdicts she finds counterintuitive. If she accepts that non-
normative facts of kind N can ground verdictive facts of kind V, and there is
a situation in which N is instantiated but according to the direct pluralist’s
verdictive intuitions V is not instantiated, she can just claim that some other
fact, Z, which is also instantiated, is a disabler of N.
This escape route, however, will often not be available to the non-Rossian
pluralist. We can’t say everything we want to say about disablers. Some
of the things we might want to say would be drastically counterintuitive.
This is because we have intuitions about which kinds of fact can disable
which explanations. This is an inevitable concomitant of having explana-
tions about what explains what. Suppose, for instance, that I lie to you
about where I’ve been for the past three hours. If I lied to you because
of the embarrassment that telling the truth would have caused me—say,
for instance, I was supposed to be taking our kids to the park but instead
was playing video games—I suspect we’ll all have the intuition that there
is nothing here to undermine that impermissibility-grounding power of the
fact that my utterance was a lie. If, on the other hand, I lied to you so as
to not spoil the surprise birthday party that I’m throwing for you—say, for
instance, I spent the last three hours checking out potential venues—I sus-
pect we’ll all have the intuition that the impermissibility-grounding power
of the fact that my utterance was a lie is indeed undermined in this case.
NP makes it possible to work from the top down. This is not to say that
people frequently deliberate in the top down way NP would suggest we
should nor that people frequently try nor that people would find it easy
to do so if they tried. My claim is merely that it can be done, which is
more than can be said for Deliberative Rationalism. If NP is true, what
we should do if we want to work from the top down is, first, to determine
what the true type-grounding claims are. I argued in Chapter 2 that we
have at our disposal a methodology for accomplishing this. Once we found
the true type-grounding claims, our task is to determine which actions are
of the specified non-normative types—i.e., which actions possess the non-
normative features picked out as important by the true type-grounding
claims. This is a purely empirical task, and it shouldn’t present any insur-
mountable difficulty most of the time. The final step is to figure out, for each
action that possesses one of the relevant non-normative features, whether it
possesses any relevant enabling or disabling features, such that the relevant
non-normative facts might fail to do their grounding work. Fortunately, we
have intuitions about enabling and disabling, and we can appeal to them in
order to complete this final task. Having done so, we will thereby arrive at
the verdictive facts, some of which we would have initially been inclined to
reject.
The moral epistemology I sketch here occupies an attractive intermediate
point between moral-epistemic absolutism and moral-epistemic minimal-
ism. Moral-epistemic absolutism, I’ll stipulate, is the view that normative
ethics should be aimed at the discovery of general moral truths (however
highly specified and complex) from which one can infer verdictive facts
without the use of judgment. I am not an absolutist, as I hold we should be
aiming to discover type-grounding claims, which cannot be applied to cases
without the use of judgment. Moral-epistemic minimalism, I’ll stipulate, is
the view that it’s no mark against a normative ethical theory if it leaves
such a wide scope for judgment that the theory will never deliver verdictive
conclusions that we find counterintuitive. (I have in mind here RP and also
the view endorsed by McNaughton (2002).) I am not a minimalist, as I hold
that we should appeal to our intuitions about explanation and also about
enabling and disabling and thereby sometimes resign ourselves to accepting
verdictive conclusions that we find counterintuitive.
4. The Phenomenology Objection

I’ve argued that thinking about one’s moral reasons is not a good method of
moral deliberation. But many people report that when they think their way
through tough moral cases it seems as though their deliberation is delibera-
tion about how the moral reasons on each side balance out (e.g., Kagan
1988, 19).8 So I have some explaining to do.
I want to offer a three-part explanation. First, I should point out that noth-
ing I’ve said here commits me to the view that it’s pointless to deliberate about
one’s moral reasons. For while I’ve said that such deliberation can’t help one
to figure out what one morally ought to do, I can allow that it can help one to
figure out what one ought to do all-things-considered. After all, I’ve endorsed
Reasons Ground Oughts, which allows that what one ought to do all things
considered can depend on how the moral reasons relevant to the case balance
out against the relevant non-moral reasons. Now this explanation wouldn’t
get us very far if the phenomenology in question was the phenomenology
of distinctly moral, as opposed to all-things-considered, deliberation. But
I doubt that all-things-considered deliberation where moral matters are at
stake would feel much different from purely moral deliberation. Importantly,
in both cases one might feel anticipatory versions of guilt, regret, etc., as one
considers the option of failing to do what one is morally required to do.
Admittedly, there’s one way that all-things-considered deliberation where
moral matters are at stake could come to feel quite different from purely
moral deliberation—namely if the person doing the all-things-considered
deliberating had the belief that the all the considerations at stake were
amoral. This brings me to the second part of my explanation of the phenom-
enology: To the extent that we sometimes feel as though we are weighing
moral reasons as a way of deliberating about what we morally ought to do,
this may simply be because of false normative beliefs. There could be cases in
which we balance multiple reasons against each other, where some of them
are moral reasons and others are non-moral reasons, but we falsely believe
the non-moral reasons to be moral reasons. To take an actually controversial
case: some people believe (as W. D. Ross did) that if φ-ing would contribute to
my self-perfection then I have a moral reason to do it. But maybe it’s actually
the case that this particular non-normative fact grounds a non-moral reason.
And this point about false beliefs can be redeployed to explain other cases
in which we deliberate about our moral reasons in order to determine what
we morally ought to do. For instance, there could also be cases in which we
balance multiple reasons against each other, where they are both moral rea-
sons and have the same weight, yet we falsely believe that they have differing
weights. After all, not everyone accepts that part of The Legal Model that says
that all reasons grounded in moral requirements are of the same strength. False
beliefs like this make it in fact possible to deliberate about one’s moral reasons
as a way of arriving at a conclusion about what one morally ought to do.
The third part of my explanation involves an appeal to the distinction
between practical deliberation and theoretical deliberation. Practical delib-
eration is deliberation about what to do, where the conclusion of such delib-
eration is intending, or willing, or acting, or something like that. Moral
deliberation is a kind of theoretical deliberation, since it is reasoning toward
the fact about what one morally ought to do. (Though of course it’s usually
theoretical deliberation toward a practical end—the end of doing the right
thing.) The key point is that all of the non-normative facts that would be rel-
evant to moral deliberation are also relevant to practical deliberation. So if
the phenomenology of moral deliberation in hard cases is the phenomenol-
ogy of weighing non-normative facts thought of simply as facts (as opposed
to non-normative facts thought of as reasons), then there might not be much
difference in feel between moral deliberation about situation X and prac-
tical deliberation about situation X. In both cases, one’s thought process
could take the form: “On the one hand φ-ing would hurt someone, but on
the other hand not-φ-ing would be deceptive.” There’s nothing in NP that
implies that practical deliberation of this sort is pointless, even in a genuine
moral dilemma. The mere fact that all the moral reasons are in perfect bal-
ance does not imply that practical deliberation about the facts that ground
those reasons cannot be productive. In practical deliberation we might legit-
imately take those facts into account in an entirely different way from how
it would be appropriate to account for them in moral deliberation.
5. Conclusion
In this chapter I defended NP against what I take to be the most obvious
objections that would be raised against its structure. I cannot in this section
summarize all the arguments just mentioned. But in the interest of having a
fair comparison of our options in normative ethics I want to make a final
note of an important debt we incur by adopting NP, namely putting our-
selves in a tenuous spot with respect to genuine moral dilemmas. NP does
not, strictly speaking, imply their existence, but it does imply that every
moral conflict is a genuine moral dilemma.
How big this problem is for NP we cannot know until we begin to hone
in on the best version of it, and thus gain some idea of which non-normative
facts it holds out as morally significant and (as a consequence) how frequent
moral conflicts will be on the theory. Having gone some way toward doing
this, I will return to the question of moral conflict in Chapter 10.
***
Before turning to Chapter 6, I want to take the opportunity to explain how
the contents of the second half of this book will be related to what has trans-
pired in the first half. In Chapter 6 I argue that no generalization having to
do with the distribution of welfare or of any object whose moral impor-
tance is dependent on its relation to welfare can ground any verdictive facts.
Since, as I argued in Chapter 2, we should treat normative ethical theories
of being composed of nothing more than generalizations that do explana-
tory work, I therefore conclude that we should exclude such generalizations
from our normative ethical theory. In Chapter 7 I argue, once again drawing
on the methodology of Chapter 2, that we can give very sensible answers
to moral questions about our treatment of animals (and other individuals
less psychologically sophisticated than you and I) without adverting to talk
of moral status, and that furthermore there is no independent explanatory
work that facts about moral status could possibly do.
It will be only in Chapters 8–9 that I begin to bring out the verdictive
implications of accepting NP, especially as by contrast to RP. The big pic-
ture, which I have already hinted at, is that if we put ourselves in the mode
of thinking, as NP says we should that non-normative facts, when they’re
explanatorily relevant, are relevant by virtue of grounding verdictive facts,
we’re likely to wind up with a much shorter list of non-normative facts that
are explanatorily relevant than we would if we were in RP mode—i.e., if we
were thinking that explanatorily relevant non-normative facts are relevant
in virtue of grounding moral reasons. So we’re likely to end up with a more
spare normative ethical theory. Specifically, in Chapter 8 I argue against
duties of beneficence, and using this conclusion combined with the conclu-
sion of Chapter 7 I argue that we should stop conducting harmful research
on animals. And finally, in Chapter 9, I draw on Chapters 6 and 8 to argue
that our best normative ethical theory will have no answers to the questions
that philosophers seem to be trying to answer when engaging in thought
experiments regarding the distribution of scarce health care resources.
Notes
1. If the grounding relation is transitive, which I have thus far allowed for the sake
of argument, then NP encompasses RP; RP would be the view that the verdic-
tive facts are grounded by certain normative facts and (indirectly) grounded by
certain non-normative facts. Likewise, if moral reasons reduce to non-normative
facts, then NP is RP plus the reduction claim. However, there would still be
room for pluralists to disagree about how the non-normative facts/moral reasons
ground the verdictive facts. In particular, most (I take it) would believe that the
non-normative facts/moral reasons jointly ground the verdictive facts. In other
words, the non-normative facts/moral reasons relative to an action are in some
sense combined (perhaps by having their reason-giving strength aggregated) and
it’s this combination that grounds the verdictive fact about that action. This pic-
ture of how the grounding works is the sort of picture that’s quite natural only if
one thinks that the non-normative facts/moral reasons behave in more the way
that non-reductivists would say moral reasons behave and less in the way non-
reductivists would say that non-normative facts behave. It’s moral reasons, not
non-normative facts, that can easily be thought of as having a weight and as being
combinable into a sort of supra-entity (i.e., a supra-reason). So if the ground-
ing relation were transitive or reductivism were true, then although in principle
all pluralists would accept NP most of them would believe that the things that
ground the verdictive facts are things that behave in a reason-ish way.
This version of NP—the picture that most people would accept were they con-
vinced that grounding is transitive or reductivism is true—I consider just as objec-
tionable as RP, and for just the same reasons. Both models of moral grounding
embrace the general idea that the verdictive fact about an action is grounded in all
the considerations relevant to it, an idea to which I objected in §1.3 of Chapter 4.
So neither the truth of the transitivity of grounding nor of reductivism would be
of any help to me. The same is true, obviously, of the falsity of these theses. This
is why I don’t discuss them.
2. But I would like to point out that I’ve already put a dent in the ought-implies-can
argument against the possibility of moral dilemmas. Insofar as there is an argu-
ment that ought implies can, it’s likely to proceed by way of blameworthiness.
The first premise would be that ought-but-failed-to implies blamable-for, and the
second would be that blamable-for implies could have. But I’ve already argued,
in §2.2.5 of Chapter 4, against the first premise.
3. Sinnott-Armstrong (1988) may have been the first philosopher to use moral resi-
due as an argument for genuine moral dilemmas.
4. It also wouldn’t count for much if moral residue could be accounted for given
moral monism. But moral monism can’t allow for moral conflict, and therefore
obviously cannot account for moral residue.
5. Scanlon (2014, Chapter 5) agrees.
6. I first raised this worry specifically against Douglas Portmore’s consequentialism,
which is a version of RP. See Sachs (2013). A similar point has been made by
Richardson (1990, 287–9). McMahan (2000, 105) confronts the same problem
and simply bites the bullet: he says that the direction of discovery is the reverse
of the direction of grounding. What this means, I think, is that he accepts RP but
rejects Deliberative Rationalism.
7. So far I have mentioned just two ways of arriving at a reason-strength belief:
(1) having a reason-strength intuition, and (2) inferring such a belief from one’s
verdictive beliefs. I don’t want to deny that there might be other ways of arriving
at a reason-strength belief; I simply maintain that the other methods don’t solve
the problem for Deliberative Rationalism. Here are three case studies:
Kearns and Star believe that a reason to φ is evidence that one ought to φ,
and the strength of any reason is the strength of the evidence it constitutes,
where the strength of a piece of evidence is the degree to which it raises
the probability of whatever proposition it is evidence for (Kearns and Star
2009, 230–2; see Thomson 2008 for a similar view). For our purposes
the question is: If Kearns and Star were right about what a reason is and
what determines its strength, would this make available a new defense of
Deliberative Rationalism? The answer, unfortunately, is no. One will have
no greater ease in directly discerning the degree to which some fact makes
it more likely to be the case that one ought to φ than one would have in
directly discerning the strength of some reason to φ.
A similar result emerges if we adopt thoroughgoing Humeanism, where
‘Humeanism’ is the view that each reason is grounded in a desire, ‘thorough-
going Humeanism’ is a view on which the strength of a reason is determined
partly or wholly by the strength of the desire that grounds it. Here the ques-
tion is: If thoroughgoing Humeanism were right about what a reason is and
what determines its strength, would this make available a new answer as
to how one can determine how strong one’s reasons are? Again, the answer
is no. For some Humeans (e.g., Smith 1996, 167) the relevant desires are
hypothetical, meaning we would have a good method for discerning the
strengths of reasons if and only if we had a good method for discerning
the strengths of hypothetical desires—which, of course, we don’t. For other
Humeans (e.g., Manne 2016) the relevant desires are actual but the weighti-
ness of the various reasons is based on agents’ hypothetical prioritizations
of the satisfactions of their various desires, meaning that we would have
a good method for discerning the strengths of reasons if and only if we had a
good method for discerning agents’ hypothetical prioritizations—which, of
course, we don’t (Manne (2016, 134) admits as much).
The final case study is Mark Schroeder’s (2007) theory of reason-strengths.
First, for clarification, his view is that reasons don’t have weight, but rather
sets of reasons do (126). And one of the ways for one set of reasons, A, to
be weightier than another, B, “is for the set of all the (right kind of) reasons
to place more weight on A to be weightier than the set of all (the right kind
of) reasons to place more weight on B (138).” So if we’re to have a reli-
able way of figuring out where the balance of reasons to act lay, we need a
reliable way of figuring out where the balance of reasons to place weight
on reasons lay. Do we? Schroeder says some helpful things (141–44), but
they do not, nor do I think they’re intended to, actually constitute a use-
able method for figuring out where the balance of reasons to place weight
on reasons lay. Providing such a method isn’t really necessary given the
agenda of Schroeder’s book. He’s interested in showing how we can explain
why there are weightier reasons for certain people in certain situations to
do certain things—and, in particular, that a thesis that connects reasons
to desires provides a good explanation—as opposed to showing how we
can figure out which reasons for certain people in certain situations to do
certain things are weightier than which.
8. Thanks to Miranda Fricker for pushing me to address this objection.

6 The Question of Scope, Part 1
Distributive Moral Concerns
The agenda for the rest of the book is to get a better idea of how the most plau-
sible version of non-Rossian pluralism (NP) will look. This means identifying
some of the generalizations it will include and others it will exclude. Now we
already know, based on the argument of Chapter 2, that the generalizations
to be included in a normative ethical theory are type-grounding claims. Hav-
ing said this much, I am in a position to state the thesis of this chapter: the
best version of NP will exclude grounding claims concerning the distribution
of welfare; i.e., it will deny that any non-normative fact about the distribution
of welfare can ground the obligatoriness or impermissibility of any action. In
fact I will argue, more broadly, that this is true of all generalizations regarding
the distribution of welfare and all generalizations regarding the distribution
of anything whose moral importance is dependent on its relation to welfare.
Hereafter I’ll refer to this combined class of claims as ‘welfaristic distributive
generalizations’ (WDGs). I am not, however, going to argue that there are no
true moral generalizations regarding the distribution of welfare; I’m actually
going to offer, at the end of this chapter, an optimistic roadmap for identifying
some. What I am going to deny, however, is that there are any generalizations
about the distribution of welfare that do grounding work.
Before beginning the argument, I want to say something about why my
thesis is important given the concession I just made. If no WDG is itself a
true grounding claim, then any true WDG would have to be true in virtue
of being derivable from some true grounding claim. Consequently, the only
thing we should be debating, with respect to any WDG, is whether it is
derivable from something else. This methodological implication is this chap-
ter’s key take-away.
Importantly, this is not the sort of debate that contributors to the literature
on distributive justice/ethics are having with each other. Contributors to that
debate put forward candidate generalizations, which are judged for whether
they are intuitively plausible on their own, deliver plausible case-specific
verdicts, and are simple enough so as to seem non-gerrymandered (which is
probably a way of examining whether they can plausibly be thought to do
explanatory work). I trust this description rings true to anyone who is familiar
with that literature.
The Question of Scope, Part 1 97
This not a sensible debate to have over generalizations that one knows are
not grounding claims. Imagine I asserted the generalization, ‘on every Wednes-
day afternoon, Bobby should give Sue an apple’. You would recognize instantly
that I wasn’t attempting to identify a ground of obligatoriness, and therefore
you wouldn’t worry about its intuitive plausibility, or its implications for cases,
or whether it seems gerrymandered. You would simply ask, “Why?” (I might
respond, for instance, that Bobby promised to give Sue an apple on every
Wednesday afternoon, thus implicitly adverting to some generalization about
promise keeping grounding moral obligatoriness). Odd as it may seem, I’m sug-
gesting that we should treat WDGs in precisely that way.
As to why no WDG could possibly be a true grounding claim, the answer
has to do with the issue of scope. No distribution of a good can be described
without setting its scope. Simply identifying the good to be distributed and
the designated rule or pattern according to which one ought to distribute
it isn’t good enough. One must also specify among which individuals one
ought to distribute the designate good in the designated way. My claim is
that once we specify the scope of any WDG we will find, no matter what
scope we chose, that the resulting WDG fails to constitute a true grounding
claim. We will know this because we will see that it fails to identify some
non-normative fact that is morally important in itself, which is something
a generalization must do in order to qualify as true type-grounding claim.
So, for instance, if the generalization ‘the fact that an action constitutes
the keeping of a promise can ground the obligatoriness of that action’ is a
true type-grounding claim, this means that it is morally important in itself
whether promises are kept.
Now most theorists working in this area defend their favored candidate
WDGs without specifying their scope.1 I do not intend here to pick apart any
of these defenses. They are not a target for me, since they fail to even attempt
the project that I am saying must be done—namely, defending a WDG while
specifying its scope. Instead of attempting to establish a scope, they implicitly
assume one. So my intention here is simply to demonstrate that if they were
to attempt the project, they would fail. Indirectly this will constitute a refuta-
tion of many of the arguments that have been given for certain WDGs.
Before doing all this, however, I want to be more specific about my topic.
I said I’m targeting all WDGs, but theorists differ as to why we’re obli-
gated to distribute welfare (or another good whose moral importance is
dependent on its relation to welfare) in a certain way. Parfit (2002), in a
famous paper, notes that among those who believe that we’re obligated to
bring about a certain distribution of welfare, some of them believe that this
is because that distribution is good, whereas others accept the obligation
on non-evaluative grounds. Furthermore, theorists differ not only in their
explanation as to why we’re obligated to bring about a certain distribution,
but also, of course, in which distribution they would have us bring about.
These distinctions need not delay us, however, since I take as my target all
WDGs. When I need examples of such claims, I will simply abstract away
from the question of why we might be supposed to be obligated to bring
about that distribution. And as to the question of what that distribution is
supposed to be, I will use egalitarian, prioritarian, and sufficientarian distri-
butions as my case studies. Again, however, I intend my thesis to apply no
matter what the distribution.
I begin, in §1, by arguing that there is a problem for WDGs regarding the
distribution of welfare; this argument forms the basis for the overall argument
of the chapter. The problem is this: As I said, every WDG must be assigned a
scope. The widest-possible scope, for generalizations about the distribution
of welfare, would include all individuals with a welfare. However, such a
WDG would have extremely counterintuitive implications regarding what
is morally important in itself due to the fact that among individuals with a
welfare—a class that includes both human beings and snakes—there are rad-
ical natural differences in welfare-potential. To solve this problem we could
go to the opposite extreme of setting the scope of our favored generalization
such that it applies only within groups of individuals who are precisely equal
in terms of welfare-potential. This, however, would render the generalization
practically irrelevant, since it is probably very rare for two individuals to
be precisely equal in this way. The alternative, of course, is to find a middle
ground between the extremely wide scope and the extremely narrow scope,
but the problem is that all the moderate stopping points are arbitrary, and
arbitrary facts can do no explanatory work. The conclusion of §1 will be
that no WDG regarding the distribution of welfare identifies something that
is morally important in itself.
Having made this argument in §1, I explore in §2 the idea, proposed by
Jeff McMahan and Peter Vallentyne, that the focus on welfare is what’s
causing trouble for our distributive generalizations, and that the problem
might be solved by focusing instead on fortune. This seems an appealing
move because unlike welfare there are not drastic natural differences in
potential to be fortunate. However, I argue that the success of some gener-
alization concerning the distribution of fortune rests on the success of some
generalization regarding the distribution of welfare, and consequently the
turn toward fortune is a red herring. In §3 I generalize the lesson of §1–2
and show that if there is a problem for WDGs concerning the distribution
of welfare, then there is a problem for all WDGs (that is, even the WDGs
concerning goods besides welfare whose moral importance is explained by
their connection to welfare). Section §4 is a response to an objection, while
§5 is where I show how we might construct a plausible derivation of a WDG
from a true grounding claim, which amounts to conceding that for all I’ve
said some WDG might be true. Finally, §6 concludes.
1. An Objection to Distributive Welfarism

I intend in this section to argue against the view that it matters non-derivatively
how welfare is distributed. For ease, I propose a label, distributive welfarism,
to cover all WDGs regarding the distribution of welfare. I use ‘welfare’ here
as a synonym for ‘well-being’.
I said in the introduction to this chapter that the basic problem for dis-
tributive welfarism arises because of the existence of extreme natural differ-
ences in welfare-potential. However, whether such differences exist depends
on the truth about personal identity and on the relationship between an
individual’s psychology and her welfare. In order to be precise about this,
it’s important to get certain stipulations on the table. One assumption I am
making is that some elements of welfare are achievable for an individual
only when and because she possesses some psychological capacity.2 I am
going to assume, further, that for each domain of welfare there is a limit to
how much welfare one can have if one cannot access that domain of wel-
fare. (For instance, if knowledge is a domain of welfare, then there is a limit
to how much welfare one can have without having any knowledge.) Or, to
put it another way, each domain of welfare is irreplaceable. This being the
case, the fewer of the welfare-conferring psychological capacities an indi-
vidual possesses, the less welfare that individual can have. I will use capac-
ity for welfare to refer to the greatest amount of welfare that an individual
could enjoy without any change in the individual’s psychological capacities.
Meanwhile, welfare-potential will designate the greatest amount of welfare
that an individual could enjoy, period.
On some theories of personal identity—in particular, those that hold that
an individual’s identity through time depends on some kind of continuity
or lack of change in her psychological states or faculties3—each individual’s
welfare-potential is limited necessarily: change an individual’s psychological
capacities too much, and you end up destroying that individual and bringing
into existence a different one.4 Because individuals vary widely in their psy-
chological capacities, these necessary limits in welfare-potential constitute
necessary inequalities in welfare-potential. My suspicion is that some such
theory of personal identity is accurate, and so my initial argument against
distributive welfarism will proceed against the background of such a theory.
There are, however, theories of personal identity that imply that between
distinct individuals there are no necessary differences in welfare-potential,
and I will assume the truth of some such theory while making a second argu-
ment against distributive welfarism (§1.5).
On we go, then, to the first argument. I begin by reviewing the various
ways of assigning the scope of distributive welfarism.
1.1 Scope #1: Sentient Individuals

I want to start with a very broad scope and get narrower from there. The
broadest-possible scope would include all individuals with a welfare. There
is still a debate about what it takes to have a welfare, but we all agree that
sentience is sufficient. Consequently, I will take as the first candidate the
scope that includes all and only sentient beings, and I’ll call it then Sentient
Individuals Scope. This might not be the broadest-possible scope, but any
broader scope would face the same problem that I am about to present for
the Sentient Individuals Scope.
Here is the problem: In virtue of possessing far more numerous and more
sophisticated psychological capacities, most humans have a far greater
capacity for welfare than any non-human (Frey 2003, 178–9; 2011, 185;
McMahan 2002, 195; Vallentyne 2005, 405–6). And if, as we are assum-
ing, there is a necessary limit to each individual’s welfare-potential, most
humans must also have a far greater welfare-potential than any non-human.
With these basic facts in hand, it is possible to advance a critique of dis-
tributive welfarism with the Sentient Individuals Scope, no matter which
pattern of distribution any particular distributive welfarist advocates. Let’s
start with egalitarianism, which says that we ought to distribute welfare
equally among sentient beings, and thereby implies that inequalities in wel-
fare between sentient individuals matter morally in themselves. Well, snakes
are sentient. However, there is simply no moral appeal to the idea that it
matters morally in itself whether a human’s welfare is equal to that of a
snake.
And sufficientarianism fares no better. It says that we ought to secure a
level of welfare for each sentient being that meets or exceeds a threshold,
and thereby it matters morally in itself whether individuals’ welfares are
above or below a certain threshold. This is highly problematic regardless
of where the threshold is. Any threshold that’s relevant to a typical human
will be obviously much too high to be relevant to a snake, and any thresh-
old that’s low enough to be relevant to a snake will be clearly irrelevant to
typical humans.
Prioritarianism is a more difficult case, since most prioritarians follow
Parfit (2002, 104–5) in insisting that the urgency of an individual’s claim to
welfare is due is determined entirely by her status quo ante level of welfare,
not by how her status quo ante level welfare compares to that of others. But
this is a very misleading distinction. Prioritarianism has purchase only in
cases in which morality allows an individual to not do all she can to improve
the welfare of everyone, whether because of the existence of a personal pre-
rogative or because of resource scarcity. This is because prioritarianism (nat-
urally) tells us how to prioritize improving one individual’s welfare relative
to improving another individual’s welfare. When prioritarianism delivers
such a verdict, that verdict will be grounded in the conjunction of the fact
as to the status quo ante welfare of the one individual and the fact as to the
status quo ante welfare of the other individual(s) (and perhaps other facts
as well, such as how much the welfare of each of them can be augmented).
It’s hard to see what substantive difference there is between a theory’s say-
ing that this conjunction is morally important in itself and a theory’s saying
that what’s morally important in itself is how the status quo ante welfare
of the one compares to the status quo ante welfare of the others. Therefore,
in the end, we are licensed to conclude regarding prioritarianism the same
thing we concluded with regard to egalitarianism: It has the implication that
it matters morally in itself that some human has more welfare than some
snake.
Holtug (2007) recognizes the just-mentioned problem for prioritarianism
and says he is prepared to bite the bullet. Similarly, Crisp (2003, 760–1) notes
the parallel problem for sufficientarianism and says he is prepared to accept
the verdictive implications of setting the bar relatively high—namely that we
have urgent distributive obligations with respect to sentient animals with a
low welfare-potential. One has to admire Holtug and Crisp here for being so
bold, but nevertheless they’ve each missed the bigger picture. Yes, welfaristic
sufficientarianism and prioritarianism (and egalitarianism) with the Sentient
Individuals Scope each yield counterintuitive case-specific implications, but
that’s not their fundamental problem. Their fundamental problem is that
they identify as morally important in themselves facts that clearly aren’t.
1.2 Scope #2: Sentient Humans

The seemingly obvious way to solve the foregoing problems is to exclude
non-humans, leaving sentient humans as the only individuals covered by
the scope. Call this the Sentient Humans Scope. But this move cannot be
defended, because it requires placing moral weight on species membership.
It implies that the mere matter of an individual’s species membership helps
to determine what we owe to her by way of augmenting her welfare, even
among species all of whose members have a welfare. It’s hard to see how
this could be the case.
1.3 Scope #3: Psychologically Normal Humans

We could avoid the charge of speciesism if we narrowed the scope even fur-
ther such that it excluded some humans. There are many ways of doing this,
but if we also want to avoid the problem that doomed the Sentient Individu-
als Scope—that the individuals that fall within the scope differ significantly
in welfare-potential—our best strategy would be to narrow the scope to psy-
chologically normal humans. In fact we can just stipulate that by “psycho-
logically normal humans” we mean those humans whose welfare-potential
is near the top of the range of human welfare-potentials, thus guaranteeing
only relatively small differences in welfare-potential among the individuals
in the scope. And if we wanted our normative ethical theory to accommodate
the idea that we have welfaristic distribution obligations to other beings
besides psychologically normal humans, we could designate other scopes
too. So, for instance, “severely psychologically disabled humans” could be a
scope, “snakes” could be a scope, and so on.
The problem with the scope just introduced, which I’ll call the Psycholog-
ically Normal Humans Scope, is that there are no discontinuities in welfare-
potential, by which I mean that on the scale from zero welfare-potential
to the highest-possible welfare-potential, there are no positions that as a
matter of necessity are unoccupied. There is a continuous range of levels of
psychological sophistication and corresponding to it a continuous range of
welfare-potentials, and it is a merely contingent fact that any points on that
range are unoccupied or less-heavily occupied than others. Thus, there is an
unavoidable arbitrariness to using differences in welfare-potential to delimit
the scope of distributive welfarism.5
This is not a mere line-drawing problem. A line-drawing problem arises,
I take it, when we have two categories, A and B, and there are some items
such that we are unsure whether to put them in category A or category B.
Our problem here certainly starts out that way. We have two categories,
the humans who should be in the scope of distributive welfarism and the
humans who should not (or who should be in their own scope), and we
find ourselves unsure how to categorize certain humans (perhaps the mildly
mentally handicapped ones). But our problem then gets much deeper: When
we reflect on why we are unsure, we realize that the very idea of there being
an A and a B is based on nothing. As soon as we fix our attention on it, it
melts away. And this spells the end of distributive welfarism with psycho-
logically normal humans as its scope.
1.4 Scope #4: Equal Psychological Sophistication

To avoid the arbitrariness problem, we might decide that any particular
scope should include only beings that are equally psychologically sophis-
ticated and therefore have equal welfare-potential. Call this the Equal Psy-
chological Sophistication Scope. But this scope would yield an incredibly
anemic WDG. On the assumption, stated above, that there are many small
differences in psychological sophistication, it would tell us little; it would be
able to set priorities for our treatment of, or condemn inequalities among,
only very few pairs of individuals.
1.5 Discharging the Assumption

If, as we’ve been assuming, each individual’s welfare-potential is limited as
a matter of necessity, then individuals differ necessarily in their welfare-
potential. This latter proposition played a role in grounding the objections I
made against the Sentient Individuals Scope and the Psychologically Normal
Humans Scope. But the assumption might be false (it was, after all, based
on a conjecture about personal identity for which I have not argued), and I
now want to reverse it and demonstrate that nevertheless I can offer strong
arguments against distributive welfarism with either of these two scopes.
If each individual’s welfare-potential weren’t limited necessarily, then
there would be no inequalities in welfare-potential, thus undermining the
given objections to the Sentient Individuals Scope and the Psychologically
Normal Humans Scope, but there would still be necessary differences in
capacity for welfare6—a fact that gives rise to a new problem. To see how,
note first that distributive welfarism would still need to be assigned a
scope, and it would emerge as terribly counterintuitive unless the scope was
quite narrow such that the individuals who fell within it did not exhibit
large, necessary differences in capacity for welfare. (Even holding fixed the
assumption that in principle we could render a snake as psychologically
sophisticated as you, there is no intuitive appeal to the idea that it mat-
ters morally how your welfare compares to a snake’s. This is because your
capacity for welfare is far greater than the snake’s as a matter of neces-
sity.) But narrowing the scope would be an arbitrary move. The resulting
version of distributive welfarism would take it as given that no individual
undergoes psychological enhancement (this is built into the definition of
capacity for welfare). But why treat this as given if psychological enhance-
ment is possible in principle (as it surely is) and identity preserving (as it
is according to the theory of personal identity that we are now treating as
given)? Why would we think that inequalities in welfare matter morally in
themselves but inequalities in capacity for welfare do not, given that, as a
matter of principle, an individual’s capacity for welfare is just as subject to
our control as her welfare itself?
1.6 Interim Conclusion

This section was intended to show that no version of distributive welfarism
manages to identify something that matters morally in itself. But to paint
with this broad of a brush is to obscure the fact that there are different les-
sons to be learned from the distinct problems facing distributive welfarism
depending on how its scope is set. With the Equal Psychological Sophistica-
tion Scope distributive welfarism emerges as more or less practically irrele-
vant, so I will address it no further in this chapter, on the assumption that no
one would consider it worth rescuing. With the Sentient Individuals Scope
distributive welfarism delivers drastically counterintuitive implications
about what is morally important, and so again I will assume that no one
would bother trying to rescue it. Meanwhile, distributive welfarism with the
Psychologically Normal Humans Scope is arbitrary, but I acknowledge that
its arbitrary aspect would emerge as defensibly non-arbitrary if there were
a foundational theory justifying it. I explain this concession in §5. Finally,
distributive welfarism with the Sentient Humans Scope is arbitrary as well,
and in a way that I assume no foundational theory could possibly justify. So
I will discuss it no more.
2. From Welfare to Fortune

The lesson to be learned from §1 is that due to large necessary differences
in capacity for welfare, and possibly also in welfare-potential, any welfaris-
tic distributive generalization, or ‘WDG’ as I have been calling them, that
implies that it matters morally how individuals compare welfare-wise is
going to face trouble. To avoid this problem we need to design a WDG that
is sensitive to differences in capacity for welfare or in welfare-potential.
(I’ll assume differences in welfare-potential are the important matter here,
but what I say will be applicable mutatis mutandis if it turns out that there
are no such necessary differences and instead the key issue is differences in
capacity for welfare.)
Broadly speaking, there are two ways to design a WDG that is sensitive
in this way. One way—the way we have been exploring, without success—
is by narrowing its scope so that it doesn’t include individuals who differ
drastically in their welfare-potential. The other way is to allow for a broad
scope but stop concerning ourselves with how much welfare individuals
have and instead concern ourselves with the ratio, for each individual,
between her welfare and her welfare-potential. This would mean abandon-
ing the idea that, strictly speaking, morality requires us to attend to the
distribution of welfare. But in effect it might be just as good. If we return
to our earlier assumption that each individual’s welfare-potential is limited
as a matter of necessity, a WDG that implies that we’re obligated to attend
to certain individuals’ welfare/welfare-potential ratios will in effect require
that we attend to those individuals’ welfares. And even if we don’t assume
that each individual’s welfare-potential is limited as a matter of necessity,
it will remain the case that given the state of the medical and psychologi-
cal sciences it is often not possible to confer the psychological enhance-
ments that would raise an individual’s welfare-potential, and so a WDG
that instructs us to attend to individuals’ welfare/welfare-potential ratios
will again usually require in effect that we make alterations to individuals’
welfares.
The proposal, again, is to abandon distributive welfarism—that is, to
abandon all generalizations that imply that the distribution of welfare mat-
ters morally in itself—and instead endorse the idea that the distribution of
welfare/welfare-potential ratios matters morally in itself. I’ll refer to this
latter idea as distributive fortunism, because, as Jeff McMahan has argued,
it is plausible to think that the extent to which an individual is fortunate
or unfortunate is determined by the ratio between her welfare and her
welfare-potential.7
Distributive fortunism has just two supporters as far as I know: McMa-
han (1996; 2009) and Peter Vallentyne (2005).8 Both of them use the Sen-
tient Individuals Scope in their version of distributive fortunism, so I’ll use it
here. McMahan arrived at distributive fortunism by considering some of the
troubles we encounter when thinking about whether the lower welfare of
those humans who are by their nature severely psychologically limited gives
them a special claim on society’s resources. He noted that it is extremely
difficult to construct a WDG that would bring such humans into its scope
and yet not deliver drastically counterintuitive conclusions (particularly
regarding what we owe to non-humans). He then formulated a conception
of misfortune and built a distributive generalization around it as a way of
showing that we can have a theory of distributive ethics that avoids those
implications. Vallentyne and I, meanwhile, arrive at distributive fortunism
by recombining McMahan’s insights such that they yield a general critique
of distributive welfarism, and then looking for a way to save the idea that
we are required to attend to the distribution of welfare in practice, all the
while without endorsing distributive welfarism.
The argument that McMahan and Vallentyne give for distributive fortun-
ism with the Sentient Individuals Scope, and the argument I have given thus
far, is thus quite simple: distributive fortunism with the Sentient Individuals
Scope avoids some of the pitfalls of distributive welfarism with the Sentient
Individuals Scope. Fair enough, but we should want to know whether it can
be motivated independently.9 That’s the question I propose to examine in
the rest of this section.
To make my argument I first need to engage in a bit of evaluative taxonomy.
First, there is a distinction between personal goods—things that are good
for someone—and impersonal goods. Distributive generalizations always
take personal goods as their distribuend. This much should be uncontrover-
sial, since distributive questions are questions about who should get how
much of what. Second, there is a distinction between instrumental personal
goods and final personal goods. Distributive generalizations, I submit, must
take final personal goods as their distribuend if they are to have moral force
of their own and thus qualify as grounding claims. The basic idea here is
that while it might matter derivatively how instrumental personal goods are
distributed, the only distribution that matters in itself is the distribution of
the final personal goods.
While I concede that welfare is a final personal good, the ratio between
one’s welfare and welfare-potential strikes me as a merely derivative good.
This becomes clear if we consider a case where there is a change in one with-
out a change in the other. Imagine, then, that Jones gains a new psychologi-
cal capacity. (Remember, not even the continuity theory of personal identity
implies that it’s impossible for an individual to gain new psychological capaci-
ties; it simply implies that it’s impossible for an individual to gain a lot of new
psychological capacities.) And imagine, further, that Jones exercises it from
time to time. Does this seem like an attractive possibility for Jones? Would
you envy Jones if this happened to her? Imagine, for instance, she were to
be endowed with much better memory—a better memory, in fact, than any
human has ever had. If you think that this should appeal to her, then I suspect
you think that one’s level of fortune or misfortune isn’t really what matters.
Remember, the details of the case don’t settle whether Jones become more or
less fortunate; in fact it’s most natural to think of a turn of events like this—
where one acquires a new psychological capacity and gets to exercise it from
time to time—as being mostly a wash in terms of fortune or misfortune. By
contrast, it’s most definitely not a wash when it comes to welfare. It’s a wel-
fare-improvement. That, I propose, is why you think it should appeal to her.
Given that fortune is a merely instrumental good, a generalization govern-
ing its distribution could at best have derivative force. With this settled, it
becomes clear how we are to go about determining whether to accept dis-
tributive fortunism: We need to determine whether it’s possible to ground
distributive fortunism in some generalization that has moral force of its own.
But now the problem for distributive fortunism becomes apparent. The
obvious candidate for such a grounding generalization is some version of dis-
tributive welfarism; this is certainly where McMahan and Vallentyne begin
their argument for distributive fortunism. But the very reason they endorsed
a version of distributive fortunism was for the sake of avoiding having to
endorse any version of distributive welfarism. And now we see that making
the former endorsement actually requires making the latter endorsement! In
other words, if some version of distributive fortunism has moral force, this
must be because some version of distributive welfarism has moral force. Yet,
as I tried to show in §1, no version of distributive welfarism has moral force of
its own. (It may have derivative moral force; that is for us to investigate in §5.)
I conclude, therefore, that the turn to distributive fortunism is of no
help.10 The problem remains, which is that we have been unable to formu-
late a version of distributive welfarism with its scope specified that seems to
have moral force of its own.
3. Broadening the Target

I now want to argue that the just-mentioned problem for distributive for-
tunism reappears for every distributive generalization that contains as its
distribuend an item whose moral importance is entirely dependent on its
relation to welfare.
I said earlier that welfare is a final personal good. We should, however,
help ourselves to a stronger claim: Welfare is the only final personal good.
Or, alternatively, the constituents of welfare are the only final personal
goods. I also said that distributive generalizations must take final personal
goods as their distribuend if they are to have moral force of their own. From
these two premises we may conclude that distributive generalizations must
take welfare as their distribuend if they are to have moral force of their own.
This means that the best we can hope for, with respect to generalizations
that take something other than welfare as their distribuend, is to show that
they have derivative moral force—specifically, moral force in virtue of bear-
ing the right relation to some generalization regarding the distribution of
welfare. And, indeed, there are plenty of distributive generalizations that
have actually been defended that fit this characterization. Versions of dis-
tributive fortunism are, as we saw in §2, instances of this kind of generaliza-
tion. But there are other examples, too.
a) Many theorists believe that if we concern ourselves with the distribu-

tion of welfare then we absolve people of the responsibility they bear
for their own welfare. In response, many theorists have formulated
responsibility-sensitive theories of distributive justice, of which luck
egalitarianism is the most well-known type.11 In these theories it is
suggested either that the proper distributive generalization would take
opportunity for welfare as its distribuend12 or that it would take welfare
as its distribuend but be concerned only with certain alterations to it
(namely, the ones for which the individual cannot be held responsible).
b) Other theorists worry that if we concern ourselves with the distribution
of welfare then we commit ourselves to the impossible task of actually
assessing how well off different individuals are. This was one of Rawls’s
arguments for taking primary goods as the currency of justice instead of
welfare.
c) As a further example I’ll mention Ronald Dworkin (2000), who wor-
ries that if we concern ourselves with the distribution of welfare then
we commit ourselves to devoting excessive social resources to providing
welfare for those whose have expensive tastes. His solution is to pro-
pose that we focus on the distribution of resources instead of welfare.
d) Finally, some theorists (e.g., Sen 1992) share Dworkin’s worry about
concerning ourselves with the distribution of welfare but are concerned
that if we try to fix it by focusing on the distribution of resources we
disadvantage those who by no fault of their own are less efficient at
converting resources to welfare. They propose, consequently, that we
should focus on the distribution of capabilities.
It is legitimate to make the move from distributive welfarism to one of

these generalizations only on two conditions: First, that distributive wel-
farism has moral force; and second, that the alternative generalization fixes
some problem inherent in focusing on welfare and instead instruct us to
focus on the distribution of something that is related to welfare in the right
way. I’m happy to admit for the sake of argument that in the case of the
four just-mentioned generalizations the second condition is met. But as I
said in arguing against distributive fortunism, the problem is that the first
condition might, for all we have so far discovered, not be met; we have
found that no version of distributive welfarism has moral force of its own,
and we have not yet investigated whether any of them have derivative moral
force. Consequently, the argument of §1 not only puts distributive welfarism
under threat but also threatens the idea that there is any moral importance
to the distribution of anything whose significance is dependent on its rela-
tion to welfare.
4. An Objection
I have argued that all versions of distributive welfarism have no moral force
of their own. One might wonder, however, how much this shows. Admitting
that no one has yet found a way of fully spelling out a version of distributive
welfarism—where ‘fully spelling out’ means specifying a scope—while
avoiding serious difficulties does not commit us to concluding that no ver-
sion of distributive welfarism has moral force of its own. To make this infer-
ence, one might worry, is to impose too high of a standard on those who
are convinced that the distribution of welfare matters morally. We don’t,
for instance, require ourselves to have a fully satisfactory theory of what
constitutes lying before we can consider ourselves justified in believing that
the generalization prohibiting lying has moral force of its own. In the case
of lying, we consider it enough if we are justified in believing that the uncon-
troversial cases of lying are wrong, and wrong because they’re lying. Why
not say that it’s enough if we are justified in believing that the central cases
of failure to distribute welfare appropriately are wrong, and wrong because
they constitute such failures?13
My response is that it would be enough, but we are not in fact justified
even with regard to the central cases. To illustrate the point, I will borrow an
example from Thomas Nagel’s article “Equality.” This is a classic example
from a classic article that sought to motivate a concern for the distribution
of welfare. If any case should do the trick, it’s this one.
Suppose I have two children, one of which is normal and quite happy,
and the other of which suffers from a painful handicap. Call them
respectively the first child and the second child. I am about to change
jobs. Suppose I must decide between moving to an expensive city where
the second child can receive special medical treatment and schooling,
but where the family’s standard of living will be lower and the neigh-
borhood will be unpleasant and dangerous for the first child—or else
moving to a pleasant semi-rural suburb where the first child, who has a
special interest in sports and nature, can have a free and agreeable life.
[. . .] the gain to the first child of moving to the suburb is substantially
greater than the gain to the second child of moving to the city.
(Nagel 2002, 75)
Nagel expects the reader to have the intuition that there is a special moral
pull in favor of moving to the expensive city, and takes this to count in favor
of a moral theory that incorporates distributive concerns as opposed to a
theory that is composed of nothing more than the principle of utility.
But this inference is too quick. There are other ways to explain why the
parent should move to the expensive city (assuming that he should). For
instance, we might think that one of the duties of a parent is a duty to
ensure that one’s child does not suffer unnecessarily. Given the facts of the
case as Nagel presents them, it’s reasonable to read into the case the fur-
ther fact that the only way neither child ends up suffering is if the family is
moved to the city. The broader point, however, is simply that a verdictive
intuition is not, in itself, an intuition in favor any particular generaliza-
tion or family of generalizations. What generalization to endorse in light
of a verdictive intuition is something we have to reason out. And part of
that process involves coming up with a list of candidate generalizations. We
can’t take just any generalizations and examine the extent to which they are
supported by our case-specific intuitions; that would take forever. Rather,
the best strategy is to take a list of generalizations such that each of them
seems sensible on its own—that is, in advance of considering the extent
to which it accords with our verdictive intuitions—and then examine each
member of that list to determine how well it accords with our verdictive
intuitions. (Remember, I argued in Chapter 1 that the generalizations out
of which normative ethical theories should be composed are themselves apt
to be evaluated for intuitive plausibility.) An analogy: We all have the case-
specific intuition that it was wrong of Germany to invade Poland on Friday,
September 1, 1939. We could use this intuition in support of the following
generalization: That a foreign invasion takes place on a Friday can ground
the wrongness of that action. But we wouldn’t actually do this, because that
generalization doesn’t strike us as plausible on its own. Similarly, we have
yet to manage to formulate a complete (that is, with scope included) version
of distributive welfarism that seems to us to have moral force of its own.
So we shouldn’t count our intuition in Nagel’s case in favor of some as-yet-
unspecified version of distributive welfarism.
5. A Provisional Conclusion and a Qualified Concession

to the Distributive Welfarist
In §1 we examined four candidate scopes for distributive welfarism, and
in each case we found that the resulting versions of distributive welfarism
would have no moral force of their own. As I argued there, distributive
welfarism with the Psychologically Normal Humans Scope emerges as
arbitrary because the scope itself is arbitrary. Admittedly, however, aside
from the arbitrariness problem distributive welfarism with this scope looks
pretty appealing. When a principle seems arbitrary but otherwise possesses
appeal, the natural thought would be to keep it as is while identifying a
non-arbitrary foundation for it. The challenge, then, is to eliminate the arbi-
trariness of limiting the scope of distributive welfarism to psychologically
normal humans.
The first step would be to identify a way in which human psychological
normality is morally important. We have already, in §1, eliminated one pos-
sibility: It is not important in virtue of having some important implication
for one’s welfare. As far as one’s welfare is concerned, human psychological
normality is an arbitrary classification that simply obscures the underlying
fact that many small differences in psychological functioning ground many
small differences in welfare-potential. If human psychological normality
were to map on to some morally important property, that property would
have to be a threshold property (unlike welfare-potential), since psychologi-
cal normality itself is a threshold.
To my mind, there are at least two promising avenues to pursue if we want
to argue that psychological normality is morally important. First, depend-
ing on where the threshold of human psychological normality is set, it may
turn out that all and only psychologically normal humans have the ability
to offer justifications to others for their actions. Since that ability is mor-
ally important according to contractualists, the corresponding threshold of
psychological sophistication will be morally meaningful for them. Second,
there is some way of setting the threshold of human psychological normality
such that it maps onto the abilities to rationally pursue one’s welfare and
to give morally transformative consent to contracts. Since these abilities are
morally important according to contractarians, the corresponding threshold
of psychological sophistication will be morally meaningful for them.14
The second step would be to establish a connection between distributive
welfarism and the morally important abilities identified in the first step. So,
for instance, a contractualist might argue for the supplemental claim that
any system of rules that might be agreed to by a group of agents concerned
to be able to justify their behavior to each other would include a version of
distributive welfarism. A contractarian might argue for the supplemental
claim that a group of rationally self-interested agents living in the state of
nature would agree to form a society unified under a sovereign authority
only on the condition that the benefits of social cooperation be divided
according to some version of distributive welfarism.15
What has moral importance in itself, according to the contractualist, is
that we act in ways that are justifiable to others. For the contractarian, it
is morally important in itself to regulate one’s behavior according to rules
that have been or could be agreed upon by rationally self-interested agents.
If distributive welfarism (with the Psychologically Normal Humans Scope)
can be shown to follow from the conjunction of some true generalization,
such as the contractualist generalization or the contractarian generalization
just mentioned, and supplemental non-philosophical premises, then dis-
tributive welfarism has derivative moral force.16 This is the only defense of
distributive welfarism that is possible.17 Importantly, it is not the kind of
defense of distributive welfarism usually found in the literature. As I said
earlier, versions of distributive welfarism are assessed as if they at least could
have moral force of their own. I admit, of course, that it seems as though
they can. They can strike us as intuitive, for instance. But insofar as they do,
this is because we are presupposing something like the Psychologically Nor-
mal Humans Scope. That is a mistake. It takes a prior moral commitment,
perhaps a commitment to contractualism or contractarianism, to validate
that scope, or any scope, for that matter. So if a certain answer to a scope
question is going to be attached to a candidate generalization regarding the
distribution of welfare, it is no longer legitimate to assess that generalization
on its own. The only question to ask is whether that generalization can be
derived from the same prior moral commitment that gave us our answer to
the scope question.
This brings us to the important conclusion about moral methodology
for which I promised in the introduction to this chapter that I would be
arguing. A great deal of literature on distributive ethics/justice consists
in philosophers coming up with novel welfaristic distributive generaliza-
tions (WDGs) and then holding them up for examination as if they at least
could have moral force of their own. (There is a good motivation for this—
namely, to see if we can come to agreement on matters of distribution while
(presumably) still disagreeing on more foundational moral matters, such
as whether contractualism or contractarianism is true.) But given what
I’ve argued here, candidate WDGs should be assessed the same way we
would assess the claim, ‘Bobby should give Sue an apple on every Wednes-
day afternoon’—i.e., we should simply ask how the claim can be derived
from some generalization that has moral force of its own. Of course,
many philosophers who work in ethics have the (again, very understand-
able) ambition to make progress on distributive questions without getting
bogged down in foundational moral matters. These philosophers, many of
whom are interested in real-world moral questions about the distribution
of scarce medical resources, often advert to thought experiments involving,
e.g., people stranded on deserted islands, a single dose of a drug, a crowded
lifeboat, etc. (Again, I trust that this description of the discourse rings true
to those who have read some of the literature.) And they tend to suggest
narrowly tailored WDGs, as if those generalizations could have force on
their own. If what I have said here is correct, both of these projects need to
be fundamentally rethought, and in Chapter 9 I will drive home this point
by examining more closely the moral questions surrounding the distribu-
tion of scarce medical resources.
6. Conclusions
We began this chapter by exploring the idea that some generalization regard-
ing the distribution of welfare has moral force of its own. I argued that none
of them do. Thus, we have made some initial progress in identifying the best
version of NP: We know now that it will not contain any welfaristic distrib-
utive generalizations (WDGs). That is the main conclusion of this chapter.
This conclusion makes for a sobering prospect for those who start from
the idea that distribution matters morally, such as Amartya Sen (1992, 3–4)
and G. A. Cohen (1989, 906), who simply assume that we are in the busi-
ness of determining what it is that shall be distributed equally. If they want
to defend a WDG, they will no longer be able to do so by identifying one
that has intuitive force, gets the cases right, seems non-gerrymandered, etc.
It may, however, be the case that they will manage to identify a grounding
claim the truth of which, taken together with supplemental non-philosophical
premises, implies some WDG. Consequently, philosophers who work on dis-
tribution must not shy away from foundational moral theory. This is the
second part of my conclusion.
Notes
1. Here are some examples of this methodology in practice.
• In defense of welfaristic egalitarianism: Sen 1992; Temkin 2011; Persson
2007; McKerlie 1989; Dworkin 2000.
• In defense of welfaristic prioritarianism: Arneson (2000b). Likewise, Schef-
fler (1982, esp. pp. 31–2) and Feldman (1997, Chapter 8, esp. pp. 161–2)
each deploy this methodology in defending the prioritarian element in their
respective versions of consequentialism.
2. So, for instance, certain emotional capacities are necessary for achieving various
relationship goods, and certain cognitive capacities are necessary for achieving
various intellectual goods.
3. For a defense of such a view, see Parfit (1987, Chapters 10–11).
4. This is Savulescu’s (2011) view.
5. We should beware of allowing ourselves to get fooled by the ease with which we
sort human beings into the categories of psychologically normal and psycho-
logically abnormal. This is easy for us to do because we have a set of familiar
categories of mental disability and, thanks to psychologists and psychiatrists,
relatively well-defined criteria for sorting individuals into those categories.
Let’s not forget, however, that these categories and criteria exist to serve a
set of practical agendas. For instance, we have to decide which individuals
can legitimately plead insanity as a criminal defense, which individuals should
be given welfare benefits on account of their inability to do socially useful
work, and which individuals should be allowed to take controlled substances.
(These, anyway, are some of our practical agendas. We also might have more
sinister agendas when it comes to diagnosing mental disability, but I won’t go
into that here.) These neat distinctions do not map onto any underlying neat
distinctions in levels of mental functioning. Rather, for any domain of mental
functioning there is a continuum from high levels of that functioning to the
complete absence of it. We have to draw a line somewhere, and we can do so
non-arbitrarily if we’re careful, but this doesn’t mean that we’re responding to
some underlying normative fact of the matter as to which levels of functioning
are defective and which are not.
6. Here’s why: Whatever the true theory of welfare is, it’s necessarily true. (All
high-level evaluative truths are necessary truths.) This theory includes some
facts about how the exercise of psychological capacities constitutes welfare and
about how each domain of welfare is irreplaceable. So if we hold fixed the
actual facts about an individual’s psychological capacities (which is what we
should do, given that capacity for welfare is defined in terms of those facts),
then we can derive a necessary fact as to the most welfare that that individual
can have. That latter fact will be, by the definition of ‘capacity for welfare,’ that
individual’s capacity for welfare.
Admittedly, I’ve encountered in discussion a fair bit of resistance to the idea that
there are any differences (never mind necessary differences) in capacity for wel-
fare. My reason for thinking that there are is, as I just explained, simply that many
elements of welfare are realized through the exercise of psychological capacities
that some individuals lack. But this argument rests on an unstated assumption:
that there is a universally applicable true theory of welfare. Some people believe
instead that there is a distinct true theory of welfare for each living thing.
I admit that there is some intuitive plausibility to this. It does seem a bit chau-
vinistic to infer, from the fact that having loving relationships is an element of
welfare for me, that it’s also an element of welfare for the snake, even though
the snake has no capacity for love. It may seem more natural to construct a
theory of welfare specific to that snake, based on our observations and our
theorizing about its nature.
If that’s the agenda, however, then we confront a challenge: How do we make
welfare comparisons between individuals? Suppose that Smith’s welfare scale
runs from a (the least welfare she could have) to z (the most welfare she could
have), while Jones’s welfare scale runs from α to ϖ, and that we know what
point in Smith’s scale Smith occupies and what point in Jones’s scale Jones occu-
pies. From this information, how can we reach a conclusion regarding which
of them has more welfare? It depends on the details of the case, but at the very
least we would have to determine that a particular point in Jones’s scale cor-
responds to a particular point in Smith’s. But this is exactly what we’re now
assuming can’t be done. We’re assuming that Smith has her way of flourishing
based on her nature, and Jones has her way of flourishing based on her nature,
and that the one has nothing to do with the other.
If we cannot make welfare comparisons between individuals, then the egal-
itarian and prioritarian versions of distributive welfarism are at best practi-
cally irrelevant and at worst incoherent, because the inequalities they tell us to
address turn out to necessarily not exist. And as to the sufficientarian version,
it too would be incoherent, as it would be unable to pick out, as its sufficiency
threshold, a level of welfare that for each being with a welfare would map onto
some particular location on that being’s scale of potential welfare. So the claim
that there are no differences in capacity for welfare is no threat to my agenda
here.
7. This is actually a much-simplified version of the theory of fortune that McMa-
han (2002, Chapter 2) proposes.
The major worry about the ratio-based conception of fortune is that it might
imply that having congenital psychological limitations is not a misfortune. Let’s
suppose that some version of the continuity theory of personal identity is true,
and correspondingly that one’s genetically determined psychological make-up is
identity fixing. This being the case, congenital psychological limitations, while no
doubt reducing one’s welfare, will also reduce one’s welfare-potential to a cor-
responding degree, thus leaving the afflicted individual no worse-off ratio-wise.
Since the turn from welfare to fortune is being put forward on my opponent’s
behalf, I’m going to simply assume that this objection to the given account of
fortune can be answered. (For what it’s worth, I think that McMahan (2009)
has done a nice job of undermining the objection.)
8. This claim, as applied to Vallentyne, is a bit of a fudge. Distributive fortun-
ism implies that it matters morally in itself how different individuals’ welfare/
welfare-potential ratios compare, and this isn’t quite what Vallentyne believes.
He believes something closer to the view that it matters morally in itself how
different individuals’ welfare/capacity for welfare ratios compare. This fudge
will make no difference moving forward. What I say against distributive fortun-
ism applies just as well to Vallentyne’s view.
9. In fairness to McMahan, his concern isn’t the prospects for distributive general-
izations but rather the question of what we owe to the severely psychologically
disabled.
10. For an argument to the effect that distributive fortunism is simply false, see
Arneson (1999) and Holtug (2007).
11. See, e.g., Temkin (2011).
12. Arneson (1989) used to endorse equality of opportunity, and for the reason just
mentioned.
13. Thanks to Wlodek Rabinowicz for pushing me to address this objection.
14. This kind of inference—from the moral importance of a particular psychologi-
cal ability to the claim that for an individual to fall within the scope of the the-
ory she must meet some threshold of psychological sophistication—is explicitly
made by Rawls (1999a, 441–3).
15. It might seem disingenuous of me to take seriously, as I have been here, the possi-
bility of grounding welfaristic distributive generalizations in contractarianism or
contractualism, given my earlier hostility to those theories (see Chapter 3). Two
things, however: First, I never said we shouldn’t be contractarians/contractualists;
rather, I said we shouldn’t be monistic contractarians/contractualists, which is
to say that we shouldn’t accept only contractarianism/contractualism. Second,
this book is about normative ethics, but since normative ethics doesn’t exhaust
the normative domain there are other things to be contractarians/contractualists
about. Normative ethics is, I take it, the study of the moral constraints incumbent
on individual moral agents as such. This would mean we need a separate theory
to discuss the special moral constraints, of which there surely are some, that
apply only to collective moral agents, such as states, and only because they’re
such agents. Call this subject matter ‘political morality’. I believe, as I will explain
in Chapter 9, that contractarianism/contractualism is much more plausible as a
political morality than as a normative ethical theory. On the possibility of differ-
ent theories being true in different moral domains, see Bräanmark (2016).
16. Why derivative moral importance but not non-derivative moral importance?
Because of what hypothetical contractors, or individuals trying to agree to a set
of rules that would allow them to justify their behavior to each other, can and
cannot accomplish. What they can accomplish, if contractarianism or contrac-
tualism is true, is to make it the case that we are morally obligated to abide by
some generalization regarding the distribution of welfare. What they cannot do
is make it the case that some such generalization has moral force of its own. No
one can make something have moral importance of its own. The fundamental
moral truths are necessarily true.
17. This conclusion is in the same spirit as some of Parfit’s (2002) comments in his
“Equality or Priority?” In the first four sections of that paper Parfit discusses
a scope objection to welfaristic egalitarianism. Whereas I have been concerned
with the question of scope along the dimension of psychological sophistication,
Parfit is concerned with the question of scope along the dimensions of space
and time. His objection to welfaristic egalitarianism is that it makes no sense to
worry about inequalities in welfare between people who live in different periods
of time or different parts of the world or universe. Parfit then points out that
there is a version of welfaristic egalitarianism that avoids this problem: deontic
egalitarianism. Of course, the deontic egalitarian must explain why some wel-
fare inequalities are unjust and others aren’t. This is broadly in line with what I
argue here; in fact Parfit even suggests, like I do, that an appeal to contractarian-
ism might do the trick.
Non-Distributive Moral Concerns
1. A Distinctly Pluralistic Way of Answering Scope Questions

Just as any candidate distributive generalization raises a scope question, so
does any candidate non-distributive generalization. If I say that the fact that
an action is harmful can ground the impermissibility of that action, I can
fairly be asked: Harmful to whom? Likewise, I can fairly be asked: Breaks
a promise to whom? Deceives whom? Constitutes ingratitude to whom?
These are among the non-comparative scope questions.
For the various versions of moral monism it’s generally immediately
obvious how these questions ought to be answered. If all moral concerns
stem from the same source, one need only look to that source to determine
which individuals are covered by the various moral concerns that derive it.
Suppose, for instance, that monistic Kantianism, along the lines of what I
described in Chapter 3, is the right way to go. There may be a great many
duties, but they all stem from the Categorical Imperative. And looking at
the argument for the Categorical Imperative it becomes clear—or at least it
seemed clear before Korsgaard and others muddied the waters for us—that
only autonomous individuals are objects of the correlative duties. Similarly,
for contractarianism; whatever the agreed-upon rules it’s clear—or at least
it seemed clear before Rowlands came along—that they apply to all and
only those individuals who can rationally pursue their self-interest and give
morally binding consent to agreements. Meanwhile, act-consequentialism
has the potential to do even better by avoiding scope questions altogether!
If we adopt a purely impersonal act-consequentialism, then it doesn’t mat-
ter to whom we do the things we do. This entirely removes the possibility
of chauvinism; in particular, the feature—so often attributed to Kantian-
ism and contractarianism—of holding that what we do to non-humans
doesn’t matter in the same way as does what we do to humans. Indeed, act-
consequentialism ever since Jeremy Bentham has had the virtue of seeming
to say something sensible about how we morally ought to treat non-humans.
But for moral pluralists it’ll never be simple or obvious how the various
scope questions should be answered. And the unfortunate fact of the mat-
ter is that moral pluralists have gotten themselves tied up in knots trying
to answer scope questions. I believe that this is avoidable, and that moral
pluralists—including non-Rossian pluralists such as myself—can handle
scope questions quite well. This is not to say that for such theorists scope
questions shouldn’t be difficult to answer. They often will be, and for good
reasons that I will point out. But I do think such theorists have made things
harder on themselves than they need to be. There are two reasons for this,
in my judgment.
First, moral pluralists have operated under the belief that morality is
distribution-sensitive at the grounding level. I of course argued in Chapter 6
that this isn’t true. But if it were true, then taking non-humans into consider-
ation would raise real problems.1 It would mean that in various ways humans
and non-humans would be in moral competition with each other: there would
be situations in which morality demands that we respond appropriately to the
relative moral importance of the concerns of some human and the concerns
of some non-human. Our verdictive intuitions would often tell us that the
human’s moral concern should win out, while our more theoretical convic-
tions would tell us that mere species membership cannot make a moral dif-
ference. A satisfying answer to the scope questions corresponding to these
generalizations would somehow have to answer to both sets of judgments.
Second, moral pluralists, when grappling with scope questions, have often
made use of expressions like moral status, moral standing, moral consider-
ability, moral personhood, and membership in the moral community. As I
will argue later in this chapter, putting scope questions in these terms cannot
help us to answer them, but it can hinder us.
In a way, these two problems for moral pluralists are symptoms of a deeper
problem besetting both pluralistic and non-pluralistic non-consequentialist
moral theorizing: non-consequentialists are, I think, in the grip of the idea
that every sentient individual is morally important to some extent. This is
the supposed fact about each of us that is often intended to be captured by
talk of ‘moral status’ or its cognates, and also sometimes by talk of ‘dig-
nity’. And the belief in such facts is why pluralistic non-consequentialists are
prone to conceiving of moral questions, especially moral questions involv-
ing human relations with non-humans, under a distributive guise. If every
individual is morally important to some extent, then surely different individ-
uals are morally important to different extents, and this undoubtedly must
make a difference to what we morally ought to do in various situations. But
this would be true only if some moral concerns were distributive; otherwise
there would be no place in moral reasoning for bringing in facts about the
relative moral importance of different individuals.
I think there is a much more sensible way for a pluralistic non-
consequentialist to answer those ‘to whom’ questions with which we began
this chapter;2 a way that doesn’t appeal to anything like an idea of the moral
importance of the individual. I borrow it from James Rachels (2004). For
each ‘to whom’ question, we ask ourselves what would have to be true of
an individual for that individual to be able to be wronged that way. In some
cases the answer will be obvious. For instance, if the question is, “Harmful
to whom?” the obvious answer is, “any individual that can be harmed.” And
so we thereby bring into the moral domain, with respect to this particular
moral concern, a set of individuals whom we have identified by description.
(Of course, identifying them more robustly would require a lot more work.)
In other cases the answer won’t be obvious. Suppose the question is,
“Deceives whom?” We might jump to the answer, “any individual that can
be deceived,” but on reflection we will probably find that answer unsatis-
factory. Playing peek-a-boo with a baby seems to involve repeatedly deceiv-
ing the infant into thinking that one has disappeared. Yet, it doesn’t seem
wrong. Now a normative ethical theory that has the type-grounding claim,
“That an action constitutes deception can make it impermissible,” doesn’t
strictly speaking force us to say that it’s wrong. Perhaps in the given case
some relevant disabler is present. But upon reflection we might become
convinced that that’s not what’s going on here. We may become convinced
instead that ‘constitutes deception’ is merely convenient shorthand for some
more complicated non-normative feature that an action can have that can
ground the wrongness of that action—something about inducing belief and
reliance, perhaps. This will press us to engage, as many moral philosophers
have, in a substantive investigation as to what’s really morally problematic
about the utterances we call deceptive. Finding the right answer here might
be quite difficult, which is why, as I said earlier, answering scope questions
can be quite difficult even if we don’t make either of the two mistakes (also
mentioned earlier) that pluralistic non-consequentialists usually make when
they try to answer them. The good news, however, is that when we finally
get to the heart of the matter—that is, when we finally find out what it is
about some actions that initially made us want to object to them on grounds
of deceptiveness—the rest of the task will be easy. Whatever it is about indi-
viduals that makes them liable to be affected in the way that’s picked out as
crucial by the new moral analysis of deception at which we arrive, all and
only those individuals that have that feature will be in the scope with respect
to this particular moral concern.
This way of answering non-comparative scope questions has two virtues,
aside from being relatively straightforward. First, and most crucially within
the context of this book, it leads us to a sensible position regarding our
treatment of non-humans. If we think that the fact that an action harms
someone can ground the wrongness of that action—a claim for which I will
argue in Chapter 8—and that being sentient is sufficient for being that sort
of ‘someone’, then, following the Rachels method, we’ll arrive at the conclu-
sion that morality forbids cruelty to non-humans. Moreover, we get to the
conclusion that cruelty to a non-human wrongs that individual. Our token-
grounding explanation of the wrongness of an instance of harming will pick
out two individuals only: the agent and the individual who is harmed. So
we’ll be able to read off quite easily from that explanation who the wrong-
doer and wrong-sufferer are. This is an important point, because we earlier
rejected the most popular forms of moral monism on account of their inabil-
ity to do this. And RP couldn’t do this either, which was the main grounds
for our rejection of it as well.
Second, the Rachels method of answering non-comparative scope ques-
tions preserves more of the appeal of moral pluralism than does the method
of averting to moral status facts. What’s appealing about moral pluralism is
that it answers to our intuition that wrongdoing comes in differing kinds,
which are to be given differing explanations. If we answer scope questions
by appeal to moral status facts, we thereby homogenize our explanations
of the various instances of wrongdoing. Suddenly, there’s a fact common to
all token-explanations of wrongdoing: that the individual who was acted
upon had (some amount of) moral status. This makes it difficult to explain
how it could be the case that, as both Rachels and Michael Thompson
(2004, 354–8) emphasize, for each kind of wrongdoing there is a different
class of individuals such that it would be wrong to treat them that way.
By way of response one might insist that the possession of moral status
is merely a necessary, but not a sufficient condition, of being the kind of
individual that it can be wrong to treat in a certain way. This would allow
that other facts about individuals, such as their sentience or lack thereof,
helps to explain why it’s wrong or not wrong to treat them in certain ways.
But now it looks as though moral status facts have become an explanatory
third wheel. Whether this appearance is veridical is the topic for the rest of
this chapter.
2. What Roles Might Moral Status Facts Play?

I said just now that moral status facts are at risk of becoming explanatory
third wheels once we’ve given a Rachels-style answer to whatever scope
question we’re considering. Consider, for instance, how this will work when
discussing the ethics of killing. (This seems like the right case study for us
since, as I said, appeals to moral status are ubiquitous in philosophical dis-
cussions of abortion and infanticide.) If we take a Rachels-style approach to
the corresponding scope question we’ll begin by asking the question, “What
is it about an individual that can make her vulnerable to the wrong of kill-
ing?” We’ll then settle on what strikes us as the most defensible answer—
something about having a valuable future, or something about valuing one’s
future, perhaps. Finally, we’ll determine which individuals have the just-
identified characteristic, and we’ll conclude that all and only those individu-
als fall within the scope of this moral concern—the concern about killing.
As I said already, I don’t see what explanatory work would be left for a
moral status fact, having gone through this reasoning process. Yet, it’s quite
common for people to say that some individual’s possession of (a certain
amount of) moral status ‘makes’ it wrong to treat them a certain way, or that
a certain way of treating an individual is wrong ‘because’ that individual has
(a certain amount of) moral status. For instance, Elizabeth Harman, David
DeGrazia, and Jeff McMahan have each said things like this in discussing
the ethics of killing,3 and other philosophers have said things like this when
discussing other moral concerns.4 Assertions like these certainly seem to
attribute explanatory work to moral status facts.
Yet, they might not be intended to attribute independent explanatory
work to moral status. If moral status facts reduce to the facts already picked
out as important in the Rachels-style answer to the scope question—e.g.,
having a valuable future, or valuing one’s future—then saying that they do
explanatory work is simply expressing in different words something we
were already inclined to accept, namely that the just-mentioned facts do
explanatory work.
Alternatively, moral status facts might not be intended to do any explana-
tory work at all. Perhaps moral status facts are the explananda. Of course,
we otherwise would have thought that the explananda are the verdictive
facts, but perhaps moral status facts are supposed to reduce to verdictive
facts. Maybe, in other words, my having (a certain amount of) moral status
is nothing over and above it being the case that there are certain specific
ways of wronging me, such as killing me, harming me, etc.5
And there is another kind of non-explanatory fact that moral status facts
might be supposed to be. Mary Ann Warren (2000, 5) claims that to have
moral status “is to be an entity toward which moral agents have, or can
have, moral obligations”. All Warren means to claim, I take it, is that any
entity that has moral status can be wronged in some way(s) or other. In
other words, the fact that I have moral status isn’t the fact that killing me
would wrong me or the fact that harming would wrong me; rather it is that
there is a non-empty set of actions that would wrong me.
In this section I have discussed four roles that moral status facts might
play. They are:
a) The Independent Explanatory Role: Moral status facts ground verdic-

tive facts and do not reduce to other facts.
b) The Reductive Explanans Role: Moral status facts reduce to non-
normative facts that ground verdictive facts.
c) The Reductive Explanandum Role: Moral status facts reduce to specific
(sets of) verdictive facts that are to be explained.
d) The Existential Explanandum Role: The fact that some individual has
moral status facts is the fact, which is to be explained, that there is at
least one way of wronging her.
Since my concern is with whether we need to talk about moral status when
inquiring into how we ought to treat fetuses, disabled humans, animals,
etc., I am not going to discuss whether moral status facts might plausibly be
thought to play the second or third role listed above. If moral status facts
do play one or the other of these roles, it remains true that there is no need
to talk about moral status, since we could instead just talk about the facts
to which moral status facts reduce. In what follows, then, I will discuss the
other two possibilities.
3. Moral Status Facts in an Independent Explanatory Role

When we give a Rachels-style answer to a scope question, we commit our-
selves to the existence of certain verdictive facts as well as to the existence
of certain non-normative facts (e.g., an individual’s possession of sentience,
or self-awareness, or rationality, etc.) that are supposed to ground the ver-
dictive facts. If moral status facts are supposed to play an independent role
in the explanation of the verdictive facts, the question is: What role is there
left to be played?
Typically, those who employ the notion of moral status make inferences
from non-normative facts to facts about moral status, and from facts about
moral status to verdictive facts.6 So, for instance, they infer from the fact
that pigs have are sentient that they have moral status, and from the fact
that they pigs have moral status to the fact that pigs have a claim not to be
caused unnecessary pain.
This pattern of inference suggests that the non-normative facts don’t
ground the verdictive facts, but rather ground the facts that ground the verdic-
tive facts. What’s appealing about this explanatory picture?7 My hypothesis is
that its appeal lies in its apparent ability to steer clear of the naturalistic fallacy.
When we commit ourselves to the view that some set of non-normative facts
on its own grounds some moral fact (i.e., some verdictive fact), we seem to
commit the cardinal sin of jumping from an ‘is’ to an ‘ought’ (as I admitted
in Chapter 2, §3.3). Positing the existence of some further fact mediating the
inference from a non-normative to a moral state of affairs helps us avoid this
appearance.
What are we to say about assigning such an explanatory role to moral
status facts? One thing we might say is that it is admirable to be cautious
about the naturalistic fallacy.8 But it also bears pointing out that such uses
of moral status do not make the problem go away; they do not explain how
we get from non-normative states of affairs to moral states of affairs. We
don’t need to be told that there is some further fact, besides facts about,
e.g., harming, lying, and promise-breaking, helping to ground the verdictive
facts; if we take the naturalistic fallacy seriously we believe that already. We
need to be told what that further fact is. Labeling that further fact a moral
status fact serves only to give the false impression that one knows what that
fact is. And this is for the simple reason that claims of the form ‘X has (some
amount of) moral status’ sound nothing like mere placeholders for some
state of affairs we-know-not-which; rather, they sound like ascriptions of
properties.
The naturalistic fallacy might be a real problem, but the insertion of moral
status into an intermediate explanatory role seems to make it disappear. This
is an illusion, of course. The problem hasn’t gone away; it’s been obscured
by a verbal finesse.
4. Moral Status Facts in an Existential Explanandum Role

It is often said that to say that some beings have moral status is to say that
they are ‘within the moral domain’ or can ‘be an object of moral concern,’ or
‘count from a moral point of view.’9 I’ve suggested that we can understand
such claims as asserting that the being can be wronged in some way(s). But
would such assertions serve a useful purpose?
That depends on exactly what we mean by ‘can be wronged in some
way(s)’. I assumed earlier that it means that for the individual in ques-
tion the set of ways in which she can be wronged is non-empty. (I return
to this assumption later in this section.) But if having moral status means
being able to be wronged in at least one way, then I cannot see why one
would ever need to know which entities possess moral status. The difference
between there being zero ways that X can be wronged and there being at
least one way that X can be wronged is no more important than the differ-
ence between there being 120 ways that X can be wronged and there being
at least 121 ways that X can be wronged. Crucially, the claim that an entity
can be wronged in at least one way cannot be deployed as a premise in an
argument regarding how some scope question should be answered.
One might concede all this and yet still insist that the claim that an entity
cannot be wronged in any way—and thus has no moral status—can indeed
be used as a premise in an argument about how we ought to answer a scope
question.10 The basic idea here is that once I know that X is not the sort of
individual that can be wronged, then I can know, with respect to each kind
of action, that doing that action to X won’t wrong X. Such an argument
would proceed this way:
X cannot be wronged in any way.

If an entity falls within the scope of moral concern Δ then there is at
least one way that that entity can be wronged.
Therefore, X does not fall within the scope of moral concern Δ.
Unfortunately, such an argument, despite its innocent appearance, would

be question begging. This is because one could be justified in believing the
first premise only if one were already justified in believing the conclusion.
There is no way to be justified in believing that an entity cannot be wronged
in any way without going through the entire Rachels-style thought process:
thinking about all the various kinds of wrongdoing, reasoning about which
characteristics underwrite the possibility of being wronged that way, and
then finding out whether the individual possesses those characteristics. This
is true even if the ‘individual’ in question is a mere object, such as a chair.
And so one must know that a chair does not fall within the scope of moral
concern Δ in order to know that it cannot be wronged in any way.
At this point I wish to withdraw the assumption that some entity’s being
within the moral domain is just her being able to be wronged in at least one
way, for there is another way to interpret this locution and others like it.
Consider Warren’s claim, quoted earlier, that to have moral status “is to be
an entity toward which moral agents have, or can have, moral obligations”,
or Kamm’s (2007, 229) contention that “an entity has moral status when, it
its own right and for its own sake, it can give us reason to do things such as
not destroy it or help it.”11 It is reasonable to interpret Warren and Kamm
as saying that to have moral status is to be eligible to be able to be wronged.
I take ‘eligibility’ in this case to be a modal notion; it is about what is pos-
sible. So an individual’s having moral status, then, would amount to there
possibly being at least one way in which that individual could be wronged.
If this is what moral status is, then, again, ascriptions of moral status will
serve no purpose. There is just nothing one can do with the information
that it is possible that there is at least one way that X can be wronged. The
problem here is the same as the problem just discussed.
“On the contrary!”, one might object, “If I know that either X can be
wronged in at least one way or it’s possible that X can be wronged in at least
one way, then I know that I should take X into account when deliberating
about what to do”.12 This objection is invited by a reading of Tom Regan
(1981, 19), who contends,
X has moral standing if and only if X is a being such that we morally

ought to determine how X will be affected in the course of determining
whether we ought to perform a given act or adopt a given policy.
The objection could be sustained if there were such things as (what I will
call) deliberative obligations—obligations to take certain individuals into
account in deciding what one ought to do.13 One might think, on the con-
trary, that we are really only answerable for our intentions and actions; that
we are morally innocent so long as we do the right thing, regardless of how
we arrived at the conclusion that that was the right thing to do.
Nevertheless, there is some intuitive pull to the idea of deliberative obliga-
tions. Unfortunately, no one has ever offered a theory of the scope and basis
of such obligations. This is troublesome, as there are numerous questions that
arise immediately when considering the idea of deliberative obligations. If the
suggestion is that one is always obligated to take into account, when deter-
mining what one ought to do, every individual that can be or at least possibly
can be wronged, the requirement is much too demanding. My high school
English teacher can be wronged, obviously, but I need not take her into con-
sideration when deciding whether I ought to have a sandwich for lunch.14 If,
on the other hand, the suggestion is that one is obligated to take into account
just those individuals whose interests might be affected by what one does,
other problems arise. First, an epistemic one: How can I identify all the people
whose interests could be affected by my contemplated courses of action, given
how easy it is for someone to have an interest in something? (One can have an
interest in something merely in virtue of caring deeply about it and thus pos-
sibly being happy or sad about how it turns out.) Second, even if those cases
in which I have all the relevant information, some problems remain:
a) Other-regarding interests. If it would make X happy if I burned a cross

on an African-American’s lawn, should I take that interest into account?
As a consideration for, or a consideration against?
b) Exclusionary reasons. We sometimes have moral reasons not to take
other reasons into account (which, when those reasons are grounded in
interests, means not taking all interests into account).15
c) One thought too many. In some cases it might be disrespectful, or dis-
play bad character, to engage in full deliberation about some possible
course of action.16
The bottom line is that, on the assumption that X’s having moral status is
the fact that X can, or possibly can, be wronged, the knowledge that X has
moral status does not on its own tell us whether one should take X into
account when deciding what one ought to do.
So it is simply not true that when we deliberate about what we ought
to do we are required to take into account all beings that have moral sta-
tus. The actual deliberative obligation, if there is one, must be much more
lenient. And so deliberating in a morally permissible way does not require
knowing which individuals have moral status.
In this section we have found that if ascriptions of moral status merely
tell us that the being in question can be wronged in some way or other, or
possibly can be wronged in some way or other, then such ascriptions tell us
nothing useful. This should not come as a surprise. What is important to
know is in what ways different individuals can actually be wronged.
5. Conclusion
My main goal in this chapter has been to put forward a method of answer-
ing scope questions for non-comparative moral concerns—a method that
is suitable for moral pluralists. I endorsed James Rachels’s method. For
each non-comparative moral concern our task is to think about what non-
normative characteristics an individual would need to have in order to be
wronged that way. Sometimes the answer will be immediately obvious,
while in other cases the answer will become clear only after engaging in a
more substantive investigation of the ultimate source of the moral concern.
In any event, having identified the crucial non-normative characteristic, the
rest of the job is empirical: identify which individuals have that character-
istic. Whichever ones do, they are within the scope of that moral concern.
Secondarily, I set out to resist the idea, prevalent in most non-
consequentialists’ writings on scope questions, that there is a broader or
more unitary question to answer in this area—i.e., some question about
which individuals are in The Scope of Morality per se, or about which indi-
viduals have (how much) Moral Importance. I stressed that these questions
make sense only from a morally monistic point of view, and that insofar as
we moral pluralists attend to them we risk discarding what was attractive
in the first place about our moral pluralism. In particular, I tried to demon-
strate the uselessness of engaging in talk of moral status. Moral status facts
play no independent role in moral explanation and there is nothing we can
say about scope questions using moral status talk that we can’t say, and usu-
ally more perspicuously, without it.
Notes
1. One might think that the issue is whether morality includes distributive con-
cerns at all. But this is not the case. If there is a derivatively true distributive
generalization, then the type-grounding claim that yields it should also yield a
clear delineation of its scope, thus obviating the need for any discussion specifi-
cally of the scope of that distributive generalization. See §5 of Chapter 6.
2. E.g., Harmful to whom?, Breaks a promise to whom?, Deceives whom?, Consti-
tutes ingratitude to whom?, etc.
3. Harman (2011, 728; 2004, 104; 1999, 312, 316; 2003, 176, 179, 186, 193–4);
DeGrazia (2012, 28); McMahan (2013a, 276).
4. E.g., Buchanan (2009, 346, 360); Douglas (2013, 473); Agar (2013a, 72, 73;
2013b, 82, 83); Cohen (2007, 188).
5. This, I think, is how Cohen (2016) understands what he calls “moral standing”.
6. See, e.g., Sumner (1981), Beauchamp and Childress (2009, 64), Cohen (1986),
Fox (1978), DeGrazia (2012), Buchanan (2009), Vallentyne (2005, 423–32).
7. Part of the answer may lie in the fact that it is a version of Rossian Pluralism
(RP)—a view that, as we explored in Chapter 4, has a lot to be said for it in vir-
tue of its compatibility with Reasons Ground Oughts and Moral Rationalism.
In particular, just as moral reasons can be weighed against each other, thereby
providing a way of accounting for how there can be moral conflicts that are
not genuine moral dilemmas, degrees of moral status can perhaps be weighed
against each other. In any event, insofar as this explanatory picture is supposed
to be appealing for any of the reasons that RP was supposed to be appealing, I
refer the reader to my arguments against RP in Chapter 4.
8. That is, on the assumption that there really is a distinction between non-
normative and moral facts and that no amount of the former can ever ground
the latter. These assumptions gave been questioned, of course, but here I grant
them to my opponent for the sake of argument.
9. Goodpaster (1978), Steinbock (1992, 5), Rollin (2006, 34, 44); Bernstein (1998,
114).
10. I thank Collin O’Neill for alerting me to this objection.
11. For a similar assertion, see Harman (2003, 174)
12. I thank Lori Gruen for pushing me to consider this objection.
13. This objection would be made more defensible by substituting ‘when deciding
what to do’ for ‘when deciding what one ought to do’. Other individuals have
reasons to care about my practical reasoning that they lack when it comes to
my moral reasoning, since only the former kind of reasoning necessarily results
in my acting, intending, willing, etc. To avoid confusion, however, I will keep
Regan’s wording. My later objections apply whatever the wording.
14. We would need a very expansive conception of ‘taking X into consideration’ in
order to avoid this problem. In particular, the needed conception of ‘taking X
into consideration’ will not require that I actually entertain the thought of X
when I deliberate. We would have to say that there are some acts of deliberation
we actually engage in that get us credit for acts of deliberation we don’t actually
engage in. For instance, perhaps at one point I thought in general terms about
what I may permissibly do with respect to my teacher, and perhaps this means
I get credit for thinking about her with respect to my sandwich ordering before
ordering my sandwich despite the fact that I don’t actually think of her before
ordering my sandwich. And what would ‘getting credit’ amount to in this case?
It could mean nothing other than that I am off the hook, morally speaking.
But why would I be off the hook? Obviously, I am off the hook because there
is no moral requirement to think of my teacher with respect to my sandwich
ordering before ordering my sandwich. And why would that be? Again, there
is only one available answer. There is no requirement because it is not possible
that this act of sandwich ordering will wrong my teacher. Thus, if we have a
duty to take everyone with moral status into consideration before making all of
our decisions, there is no way to determine exactly what goes into discharging
this obligation without first engaging in some first-order moral reasoning (e.g.,
reasoning about who might be wronged by my act of sandwich ordering). In
other words, we’d need to discharge our deliberative obligations in the course of
figuring out what they are! That being the case, we might as well stop worrying
about deliberative obligations.
15. See Raz (1990).
16. See Williams (1981).
8 Doing Harm and Failing
to Rescue
1. Introduction
My objective in this chapter is to determine whether two purported ground-
ing facts really are grounding facts. First, can the fact that an action harms
some individual make that action morally impermissible? I will argue in §2
that it can. Second, can the fact that an action constitutes a rescue make that
action morally obligatory? In §3 I conclude that it cannot. Finally, in §4, I
build off the conclusions of §2–3 to argue for an abolitionist position on
performing harmful research on animals.
2. Doing Harm
2.1 Introduction
Most cases of harm doing are intuitively wrong, but there is also a non-
negligibly large set of cases of harm doing that intuitively are not. This forces
us to consider the possibility that harm doing is not wrongness grounding.
The best way to motivate that idea would be to find something else, besides
their harmfulness, that most of the intuitively wrongful cases of harm doing
have in common—something not found in the intuitively permissible cases
of harm doing—and identify that something else as the wrongness ground-
ing feature of the actions. The alternative is to insist that the harmfulness of
an action can indeed make it wrong, and that the cases of harm doing that
are not wrong are cases in which enablers are absent or disablers are pres-
ent. I argue in this section for taking this second, more conservative, tack.
2.2 The Diversity of Cases of Intuitively Permissible Harming

To stimulate our thinking about this, I begin by listing categories of harmful
action about which many people seem to have the intuition that actions of
that kind are not morally impermissible:
a) Consented-to harm (e.g., masochism as a sexual fetish)

Doing Harm and Failing to Rescue 127
b) Self-harm (e.g., making sacrifices for friends, charitable causes, or
humanitarian efforts)
c) Compensated harm (e.g., taking part in a professional boxing match)
d) Refusing to protect others at a cost to oneself (e.g., ducking out of the
way of a projectile that then hits the person standing behind you)
e) Self-defense (e.g., killing in war)
f) Punishment
g) Harm in zero-sum cases (e.g., applying for a job and thereby preventing
someone else from getting it)1
Intuitively, there are many differing factors to be cited, for the different
categories, as to why the harmful actions of that type aren’t wrong. In the
first case we’ll cite the presence of consent, in the second case it’s the fact
that the harm was visited upon oneself, while in the third case it’s the pres-
ence of compensation. In the fourth case we’ll avert to the fact that one
did not initiate the harmful sequence. The fifth case is more complicated;
some may think that the fact that it’s self-defense is enough explanation
on its own, while others would hold that the reason goes deeper—perhaps
something about how the harmed party has ceded her Hohfeldian immunity
to harm by making herself a threat to others. The sixth case is also (notori-
ously) controversial, as there is widespread disagreement over what justifies
inflicting harmful punishment, with deterrence, prevention, and retribution
constituting the most popular options.
The seventh case is complicated as well. The question here is why it is
morally permissible to harm someone by taking the job that he otherwise
would have gotten, buying the house for which he put in the next-highest
bid, getting married to the woman who would have taken him as a second
choice, etc. Obviously sometimes this is not permissible, for instance when
one succeeds in the competition by cheating or by doing something that’s
independently morally wrong, or when the competition itself is set up in
an unfair way. (These exceptions are noted by Feinberg (1984, 220) and
Haslett (2003, 640–1).) But when this is permissible, the best explanation as
to why is probably that in zero-sum games we have no choice but to inflict
harm (Haslett 2003, 640–1). There is no competing without at least one of
the competitors being harmed, and if we refrain from competing then we
collectively end up even worse-off.
The apparent diversity of explanations is quite important for our pur-
poses. This is because we have to determine the proper way to conceptualize
what we’ve learned when we discover that there are harmful actions that are
morally permissible because the harm was consented to, and also harmful
actions that are morally permissible because the harm was visited upon the
agent herself, and so on for each of these categories of action. Because of
the diversity of explanations it appears that we’re not going to be able to
pull off the trick I mentioned earlier: finding some feature, x, that’s lack-
ing in all or most of the purported cases of morally permissible harm, which
128 Doing Harm and Failing to Rescue
would then make it plausible to propose that it’s not actually the harmful-
ness of harmful actions that can ground their wrongness but rather their
possessing feature x.
Admittedly, pulling off this trick would be a legitimate prospect if con-
sequentialism were a possibility. If invoking rule-consequentialism, for
instance, were an option here, then one might argue that all seven categories
of harmful action are such that, plausibly, the value-maximizing set of rules
would permit them.
But in Chapter 3 we rejected consequentialism.2 Of course, the rejec-
tion was not based on a complete examination of all the implications of
accepting consequentialism as compared to all the implications of reject-
ing consequentialism, which means that the rejection should be consid-
ered provisional. So if we were simply at a loss here—if we didn’t know
what to say about these varying categories of apparently permissible harm-
infliction—then it would be entirely appropriate to put consequentialism
back on the table.
2.3 The Exceptions as Cases in which a Disabler Is Present

But there is another good option on the table, namely the conservative
option I mentioned earlier. We should insist that the harmfulness of an
action can ground its wrongness, and chalk up the exceptions to the pres-
ence of disablers or absence of enablers. The case for going the conservative
route is not just our inability to identify attractive alternatives; rather there
is the all-important further point that the conservative option allows us to
make intuitively plausible type- and token-grounding claims. It allows us
to make the type-grounding claim that the fact that an action is harmful can
ground its impermissibility; surely this is quite an intuitively plausible thing
to say. And with regard to particular cases of wrongful harming it allows
us to say that their wrongness is grounded in their harmfulness. As noted
in Chapter 3, this is exactly what we want to say about a great many cases
of harming.
If we go this route, then, again, we are in the position of positing the exis-
tence of certain enablers or disablers of the wrongness of inflicting harm.
Based on the seven categories identified earlier, here are seven disablers we
could posit:
a) That a harmful action is consented to can disable the wrongness of

inflicting that harm
b) That a harmful action is visited upon the agent him/herself can disable . . .
c) That a harmful action is compensated can disable . . .
d) That a harmful action simply passed on a harm from a sequence not
initiated by the agent him/herself can disable . . .
e) That a harmful action is undertaken in self-defense can disable . . .
f) That a harm is inflicted as justified punishment can disable . . .
g) That a harm is inflicted in the course of competing in a zero-sum game
can disable . . .
Positing the existence of these disablers jibes well with our intuitions. It
does strike us as intuitively plausible to suppose, for instance, that the fact
that a harmful action is executed in self-defense can disable what otherwise
would have been the wrongness of that action, and similarly for the other
posited disablers. So that is another mark in favor of taking the conserva-
tive route.
The only apparent drawback that I can identify is that one might want an
explanation of why each of the disablers can do its disabling work. But as
I argue in Appendix A, we should be willing to do without explanations of
this sort despite the reasonableness of the desire for them. Explanation has
to run out somewhere, and it’s no less plausible to chalk up as unexplain-
able the ability of some factor to do disabling work than it is to chalk up as
unexplainable any other work that gets done in the moral sphere.
With that objection disabled (no pun intended), my argument for identify-
ing harm doing as a wrongness grounding non-normative feature of actions
is complete. I now want to investigate the moral relevance of rescuing.
3. Rescuing
3.1 Introduction
Consider Peter Singer’s (1972) famous Pond Case: Jones is walking by a pond
when he hears a distressed shout. He turns his attention to the pond and
discovers that a small child has fallen in and is drowning. Saving the child
would be easy, and also cost-free except for the fact that Jones’s nice shoes
would get muddy and ruined by wading in to the pond.
Now suppose Jones declines to save the child, and the child dies. Our
intuitive belief about this case will be that Jones has violated an obligation.
So ideally the normative ethical theory on which we eventually settle will
deliver that verdict. The question I want to begin with is what grounding
explanation of that verdict we should hope our theory can offer.
One candidate explanation is that saving the child was obligatory on grounds
of its constituting the maximization of the good. This, however, is an act-
consequentialist explanation, and we already rejected act-consequentialism.
Of course, rejecting act-consequentialism doesn’t involve rejecting the idea
that for some wrong actions the explanation of their wrongness is their
failure to maximize the good. But using act-consequentialism’s explanatory
story even just for the purpose of the Pond Case and others like it is a very
unattractive move, as I pointed out in Chapter 3. Surely it’s counterintuitive
to suggest that the wrongness of Jones’s action is grounded in how it was
a failure to maximize the good. Similarly, the rule-consequentialist is put
in the uncomfortable position of suggesting that the wrongness of Jones’s
action is grounded in how it was a failure to abide by the optimal set of
rules.
We should look, then, to non-consequentialist explanations. And when
we look to the non-consequentialists for their view on this matter, what
they tend to do is sort this kind of case under the purported duty of rescue.
Assuming that they mean to invoke the duty of rescue not just as a valid
conditional or categorical connecting moral obligatoriness and rescue but
rather (or in addition) as a grounding claim, their proposal is that in the
Pond Case Jones has an obligation to save the child on grounds of that
action constituting a rescue. This proposal deserves close scrutiny.
3.2 The Purported Duty of Rescue

To set up a discussion of the quality of this explanation we first need to
fix precisely on what is meant by ‘rescue’. One question is whether ‘rescue’
is supposed to denote the saving of a life. On reflection, it seems rather
uncharitable to interpret ‘rescue’ this way. If we think that an action can
be obligatory on grounds of its constituting the saving of a life, we most
likely also think that losing one’s life constitutes suffering a great harm and
actions that don’t save lives can be obligatory on grounds of preventing a
great harm from befalling someone. It seems, then, that we’re inclined to
sort the duty to save a life under the broader duty to prevent great harm.
This brings us to the other natural interpretation of ‘rescue’, which is
that rescuing someone is preventing a severe harm from befalling her. Now
severity would seem to denote something about how bad the harm is for its
recipient, and how bad a harm has to be for its recipient in order to qualify
as severe would seem to be contextually determined. This gives rise to a
problem. If one is categorizing a harm as ‘severe’ or ‘not severe’ in the con-
text of a discussion about the moral significance of the harm, then naturally
the moral purpose of that discussion is going to set the parameters for what
counts as a severe harm. It’s going to be correct to call a harm severe just
in case the harm is morally significant. That being the case, calling a harm
‘severe’ cannot count as giving a non-circular explanation as to why it’s
obligatory to prevent it, except that it has something to do with how bad the
harm is. Rather, calling a harm ‘severe’ could at best be a way of asserting
that it’s obligatory to prevent it.
One might object that we have an independent—that is, non-moral—way
of fixing the difference between severe and non-severe harms. The problem,
however, is that we have too many of them. In the context of a boxing
match, it would be incorrect to say that A causes severe harm to B by batter-
ing B’s face until it’s bruised and swollen. (I take it that coming away from a
boxing match with a bruised and swollen face counts as getting away easy.)
On the other hand, if A and B are two strangers who encounter each other
on the street, it would be entirely apt to label that kind of harm ‘severe’. This
multiplicity of standards for severity explains why the moral context needs
its own standard appropriate to it.
From this I conclude that by labeling a harm ‘severe’ one hasn’t really
explained, or even tried to explain, why that harm is such that an action can
be required on grounds of its preventing a harm of that kind.
Where does this leave us with respect to the duty of rescue? We have been
asking what feature of an action is meant to be picked out by calling that
action a rescue, but neither of the two features we have identified thus far—
that it was the preventing of a death or that it was the preventing of a severe
harm—seems up to the task of grounding the obligatoriness of an action. So
I propose that we simply discard the duty of rescue as a possible explanation
of Jones’s duty to rescue the child in the Pond Case.
We need, then, another candidate non-consequentialist explanation for it. I
can think of no better candidate than the obvious one: It’s obligatory because
it constituted the preventing of a harm, period.
3.3 Prevent Harm
3.3.1 What Is Preventing Harm?

Before we go on to assess whether this is a good explanation, we should get
clear first on what harm prevention is. Harm can be understood either com-
paratively or non-comparatively. If we understand it comparatively, the two
bases of comparison most likely to be suggested are the individual’s status
quo ante well-being and the individual’s counterfactual well-being—i.e., the
level of well-being at which she would have otherwise landed.3 On the first
account, a harm to an individual is an event, φ, which causes a decrease in her
well-being compared to her status quo ante. On the second account, a harm
to an individual is an event, φ, which causes a decrease in her well-bring
compared to the level of well-being at which she would have landed absent φ.
For the sake of clarity I’ll assign a variable to four quantities: A represents
Williams’s well-being status quo ante; B represents the level of well-being at
which Williams would have landed absent φ; C represents the level of well-
being at which Williams would have landed given φ; and D represents the
well-being Williams actually did end up with. (φ is being treated here as an
event that will happen unless someone intervenes.)
The first account of harm suggests that harm prevention occurs just when
one intervenes so as to mitigate the well-being-diminishing effect of an event
that otherwise would have left someone worse-off than her status quo ante.
So my intervention in Williams’s life counts as preventing harm to her just
in case A > C and D > C.
The second account of harm suggests that harm prevention occurs just
when one intervenes so as to mitigate the well-being-diminishing effect of
an event that otherwise would have left someone worse-off than where she
would have landed without that event. So my intervention in Williams’s life
counts as preventing harm to her just in case B > C and D > C.
But harm prevention on the first account seems to take account of a
moral irrelevancy by attending to the status quo ante, and harm prevention
on the second account seems to have the same flaw in virtue of attending
to where the individual would have landed absent some event that was
going to happen without the agent’s intervention. If there is some moral
significance to what an agent does to/for Williams, surely the fact that
D > C accounts for it; in other words, it’s that the agent left Williams better
off than she would have landed without the agent’s intervention. Whether
A > C is irrelevant, because A is irrelevant; likewise, whether B > C is irrel-
evant, because B is irrelevant. Not only is it intuitive to deny the relevance
of A and B in cases in which D > C, but in addition the denial is required
as a matter of consistency, because we’re already inclined to deny the rel-
evance of A and B in cases in which C > D. We are happy to say that if Jones
intervenes so as to make Williams worse-off than she would have ended up
without Jones’s intervention, then that’s all we need to know to conclude
that Jones has done something morally suspect—i.e., harmed Williams; it
doesn’t matter to us whether Jones’s intervention took the form of making
Williams worse-off (than her status quo ante or than where she would have
landed absent the intervention) or merely depriving Williams of a benefit
she had coming to her.4
What I take from the discussion so far is that there is nothing morally
significant per se about preventing harm, comparatively understood. There
may well be something morally significant about intervening in an indi-
vidual’s life such that she ends up better off than she would have ended up
without the intervention, but there are more ways of doing this than by
preventing harm to her.
We get the same result if we understand harm non-comparatively. On this
interpretation of harm, suffering harm is just being in a welfaristically bad
state. So if I have the flu, or am heartbroken, or am terrified of what’s under
my bed, I am suffering harm.
Is there anything morally significant about preventing harm, so under-
stood? Suppose Williams is about to meet the love of her life, but I
intervene so that instead of encountering that person Williams instead
encounters someone who will end up a causal friend of hers. Surely I do
something morally objectionable here. But since Williams suffers no non-
comparative harm, the only way to capture the moral relevance of what
I’ve done is by pointing out that I’ve intervened to make Williams worse-
off than she would have ended up without my intervention. This means
that adopting just a non-comparative account of harm isn’t an option if
we want to capture the moral relevance of harming; we need the com-
parative account. But adopting both isn’t a viable option either; doing
so would commit us to double counting the moral wrongness of many
actions.5
3.3.2 Failing to Prevent Harm as an Instance of Failing

to Confer Benefit
I conclude from all this that the best way to spell out our intuitive idea that
preventing harm matters morally is to say that it matters morally when one
intervenes so as to render someone better-off than she otherwise would have
ended up. We need a name for what exactly one does in such situations; I
will choose ‘confer benefit’. This isn’t an ideal expression, since it connotes
a situation in which one intervenes so as to make someone better off than
her status quo ante, whereas by stipulative definition it actually covers all
situations in which one intervenes so as to leave someone better off than she
would have ended up without the intervention, some of which are cases in
which she ends up worse-off than her status quo ante. Still, as long as we are
clear on the stipulative definition, no confusion should ensue. Furthermore,
in everyday talk it’s actually not inapt to use the ‘confer benefit’ expression
to describe situations in which one does not end up better off than one’s sta-
tus quo ante. For instance, it would not be at all odd for me to say that the
vaccination benefited me by preventing me from getting the flu.
With this all clarified, we finally have a new proposal on the table as to
what grounds the obligatoriness of saving the child in the Pond Case: it is
that saving the child confers a benefit on him. Now this proposal immedi-
ately forces us to confront the fact that failing to confer benefit is something
we’re all doing all the time. For instance, I know a few good jokes, and it
certainly brightens one’s day to have a brief laugh. But I regularly fail to
bestow that benefit on a great many people. The apparent implication here
is that I violate an obligation in so failing.
I say “apparent” implication because the mere presence of an obligation-
grounding property is insufficient to establish the obligatoriness of the
action in question. This is because of enablers and disablers. Yet, an appeal
to enablers and disablers is not going to get us out of our fix in this case,
because the plausible ones don’t seem widespread enough. If we get our-
selves into the mode of thinking that an action available to X can be obliga-
tory in virtue of its constituting an intervention that benefits Y, and then ask
what features could serve as disablers in those cases, we will likely come up
with a list like this: Y deserved, perhaps because of her own negligence or
carelessness, to be no better off than she would have ended up being without
the intervention; X had earlier been mistreated by Y; X’s bestowing a benefit
on Y would harm Z, etc. If we clear away the cases of failure-to-prevent-harm
that have one or more of these features, we will still be left with a whole
heap of cases.
Perhaps then we need to activate the nuclear option: averting to the sacri-
fice required to confer the benefit. Maybe the fact that conferring a certain
benefit would itself constitute a harm to the agent can disable the obligatori-
ness of conferring the benefit. This is the ‘nuclear option’ because activating
it would have incredibly widespread effects, and this of course is because
conferring benefit on others usually involves some harm to oneself, if only
in the form of inconvenience or opportunity cost. Taking this option means
not getting much at all, verdict-wise, out of adopting a normative ethical
theory that says that the fact that an action constitutes the bestowing of a
benefit can ground the obligatoriness of that action; For instance, we cer-
tainly would no longer get the intuitive result in the Pond Case. We could
try to take a more moderate course by saying that only the prospect of
the agent’s suffering severe harm can do the disabling work, but then we
once again put ourselves in the position of making the idea of severity do
explanatory work.
3.4 Reaching a Verdict on the Pond Case

To review: We began this section with the Pond Case, about which the wide-
spread intuition is that Jones is obligated to save the boy, and we have since
then been searching for a good grounding explanation of this purported
fact. We discarded the idea that it’s the fact that saving the boy is preventing
his death and also the idea that it’s the fact that saving the boy is the avert-
ing of a severe harm to him. We then determined that preventing harm is
not different in a morally significant way from other ways of intervening in
someone’s life so as to render her better-off than she otherwise would have
ended up, so we imported preventing harm into the broader category of
conferring benefit. Finally, we found that conferring benefit is not a likely
obligatoriness-grounding feature of actions.
That being the case, we seem to have exhausted the obvious options for
identifying what grounds the obligatoriness of saving the boy. The conclu-
sion to which we are forced is that it is morally permissible to not save the
boy. As if it needs to be said, this is a highly counterintuitive conclusion, and
of course is merely representative of a potentially infinite number of similar
cases about which we are doomed to endorse a shocking conclusion. As to
what we should do about this counterintuitiveness problem, I’ll postpone
comment until §5. For now, my agenda is to show that the problem is even
worse!
4. Implications for Animal Research

I now want to explore the implications of what’s been said already for the
ethics of painful research on sentient laboratory animals for human benefit,
a widespread activity that I will refer to for convenience by the shorthand
‘harmful animal research’.
4.1 The Moral Objection to Harmful Animal Research

Given what’s been said already, it is quite easy for us to express precisely
what is objectionable about harmful animal research. The moral concern
over harmful animal research arises because of the harm such research
causes, naturally, and as argued in §2 causing harm is an impermissibility-
grounding feature of actions. Now to determine the moral relevance of this
feature to the case at hand, we simply need to inquire as to the scope of the
moral concern over the causing of harm. As I argued in Chapter 7, no dis-
cussion of moral status is necessary (or even helpful) despite the fact that the
individuals who we are considering classifying as victims of this harm are
animals. And determining the scope of the moral concern regarding causing
harm is quite straightforward. One simply identifies the kind of harm in
question and asks what properties an individual needs to possess in order to
be apt to be harmed that way. In our case, we’re talking about pain, and sen-
tience is the property an individual needs to be harmed by pain. Therefore,
since by ‘harmful animal research’ we mean painful research conducted on
sentient animals, we can conclude directly that the harmfulness of animal
research constitutes an impermissibility-grounding feature of it.
4.2 The Moderate Stance on Animal Research

But that might not be all there is to say about the ethics of animal research.
The majority opinion, among philosophers who have discussed the issue, is
that harmful animal research is permissible some of the time. Whereas abso-
lutism on either side of the issue was commonplace before the idea of ethics
as a matter of reason weighing really took hold,6 such a position is now
passé. And not only is it passé in act-consequentialist circles—which isn’t
surprising, since act-consequentialists aren’t absolutists about anything7—
but also among non-consequentialists, such as Sapontzis (1992), Clune
(1996), DeGrazia (1996), and LaFollette (2011).
For these theorists, the moral permissibility of a piece of harmful animal
research depends on the amount and likelihood of the prospective benefit for
those who have the condition being researched and the amount of harm for
the animals serving as research subject. Consequentialists tend to express this
in terms of utility or goodness and badness, while non-consequentialists tend
to characterize these factors in terms of interests. There may be additional
factors as well. For instance, on McMahan’s view if the animal is going to be
killed then we have to take account of the degree of psychological unity that
the animal, at the time it is killed, bears to the future time-slices of itself that
would have existed in the future. Furthermore, consequentialists and non-
consequentialists are likely to disagree on the moral mathematics we should
engage in once we know how much harm and benefit is likely to result from
both courses of action; non-consequentialists are might attach weights to
certain harms or benefits, while consequentialists will likely not.
For our purposes, though, what these theorists agree about is more impor-
tant than what they disagree about: they agree that both the harm and the
benefit matter. We’ve discussed the moral significance of the harm already,
so we should examine now the purported moral significance of the benefit.
What these theorists say is that if the expected benefit to third parties of
conducting the research is large enough, then the study in question will be
morally permissible in spite of the harm it causes to the animal subjects.
There are four ways of understanding just how the prospect of benefit to
third parties is supposed to make this moral difference.
4.3 Four Understandings of the Moral Significance of

Prospective Benefit to Third Parties
4.3.1 Prospective Benefit as an Overrider of the Moral

Significance of Harm
The proposal may be that the moral significance of benefit to third parties
overrides the moral significance of harm to the animals. Consideration A
overrides consideration B just in case A has greater weight than B, and in
the opposite direction, but the presence of B does not eliminate the presence
of A. The unmistakable signal that overriding is occurring is residue. For
instance, if A overrides B and one acts on A, there should be some cause for
regret at not having been able to do B.
We have been thus far interpreting these theorists as agreeing with the
claim made in §2, namely that the fact that an action harms someone can
ground the impermissibility of that action. But if these theorists are posit-
ing that the harmfulness of animal research can be overridden, then they do
not in fact agree with the conclusion of §2. Impermissibility does not have
weight, so there is no sense to be made of the idea that the impermissibility
of an action is overridden. What these theorists must believe, instead, is that
the harmfulness of an action can ground a moral reason not to engage in it.
Moral reasons have strength and can therefore be overridden—in this case
by a moral reason to confer benefit on third parties.
But now the problem with the current interpretation of these theorists, on
which they posit that the benefit to third parties can override the harm to
the animal subjects, should be clear: It is inconsistent with what was argued
in Chapter 4, namely that the verdictive facts about actions do not hold in
virtue of any sort of balancing of considerations.
4.3.2 Prospective Benefit as an Underminer of the Moral

Significance of Harm
We should consider, then, undermining, which occurs when some disabler is
present or enabler is absent (see Chapter 2). We know already how enablers
and disablers work, and in this particular case the idea would be that while
harm to the animal subjects is the kind of fact that can ground the impermis-
sibility of the action that causes it, that fact does no grounding work when
the research is expected to confer a large enough benefit on third parties.
In cases of disabling there is no residue; if the fact of harmfulness doesn’t
do its grounding work then it does no work at all, and so it is simply mor-
ally irrelevant. But this is clearly not what these theorists believe about
the harm that harmful animal research causes to its animal subjects; they
clearly accept that there is something regrettable about the harm caused
by such research even when it is permissible. Nor would it be reasonable
of them to believe that the harm caused to animals is morally irrelevant, at
least if we take our intuitions about disabling seriously, as I’ve argued we
should. Intuitively, the fact that an action benefits X does not disable the
impermissibility-grounding ability of the fact that that action harms Y. The
prospective benefit seems to be a separate issue as opposed to something
that changes the significance of the prospective harm.
We have been asking what, according to those who think that harmful
animal research is permissible depending on how much harm and benefit it
causes, is the moral significance of the benefit. So far we have examined the
possibility that the benefit is supposed to be able to override or undermine
the moral significance of the harm, but these turned out to be dead ends.
There are, however, two other interpretations we should explore.
4.3.3 Prospective Benefit as a Ground of Permissibility

One is that the prospective benefit of the research is a permissibility-
grounding fact about it. The obvious problem here is that if this were true
then, assuming that what I said in §2 is true, any piece of research that both
harmed the study animals and benefited third parties would be both permis-
sible and impermissible, which is a contradiction. Of course, these theorists
could deny what I said in §2, that the harmfulness of an action is a kind
of fact that can ground the impermissibility of it, but then they would be
left unable to explain why some harmful animal research is impermissible,
which is part of their view.
4.3.4 Prospective Benefit as a Ground of Obligatoriness

The final possibility is that these theorists mean to suggest that an action
can be obligatory on grounds of its constituting the conferring of a ben-
efit. If this is the case, then moral dilemmas abound in the arena of harm-
ful animal research. The studies are always (by definition) harmful to the
animal subjects, making it impermissible to conduct them, but many of
them are prospectively beneficial to third parties, making it obligatory
to conduct them and hence impermissible not to. There is nothing con-
tradictory about such a view, as tragic a description of our situation as it
may be.8 The problem with it, rather, is that it is inconsistent with what
I argued in §3, namely that the conferring benefit is never a ground of
obligatoriness.
4.4 Conclusion
We have exhausted all the possible ways in which the prospective benefit
that harmful animal research promises for third parties might be morally
significant. We are forced to conclude that it has no moral significance at
all. That being the case, the morality of the situation is quite simple: The
harmfulness of the research grounds its impermissibility, end of story. Again,
it goes without saying that this leaves us in a position of having to sign off
on some terribly counterintuitive case-specific conclusions. If conducting
somewhat unpleasant research on a handful of mice could cure a terrible
disease infecting millions of human beings, it would be impermissible to do
so according to what I’ve just argued. Most people will find this impossible
to believe.
5. Conclusion
In §3–4 we arrived at highly counterintuitive conclusions regarding the
Pond Case and harmful animal research, and by extension committed our-
selves to highly counterintuitive conclusions regarding innumerable other
cases that are similar in the relevant respects. At the end of the last chapter I
want to say something about how to proceed in light of this. For now, how-
ever, I simply want to argue that things might not be as bad as they seem. I
will argue with regard both to cases like the Pond Case (§5.1) and cases like
harmful animal research (§5.2) there are sometimes other ways to get to the
verdictive conclusions we want.
5.1 An Alternative Strategy in Cases Like the Pond Case

The Pond Case is one in which an innocent person stands to suffer a great
loss that is easily preventable. One potentially morally troublesome feature
of some such cases is, however, absent in the Pond Case: institutional com-
plicity. There is no institution that causes the child to fall into the pond,
and certainly no institution that you or I play a role in upholding. When
that feature is present there is a possibility of saying that there is an obliga-
tion to intervene, though not because it is the bestowing of a benefit but
rather because failure to do so would constitute inflicting harm by sup-
porting harmful institutions. Thomas Pogge, famously, has long advocated
making this kind of move with respect to the poverty-related harms that
people in the developing world are suffering.9 He argues that much of the
poverty in the developing world is caused by global financial institutions
(e.g., the World Bank and the IMF), tariffs and other barriers to trade, and
the propensity of developed countries to treat de facto rulers, no matter
how cruel and unconcerned with the needs of their people, as legitimate by
purchasing from them the natural resources of the countries they control.
And he argues, further, that individual citizens of developed nations play a
role in all this by supporting the governments that create and sustain these
arrangements.
Perhaps it’s no great improvement in a normative ethical theory if it
yields the conclusion that our providing aid to impoverished people in dis-
tant corners of the earth is obligatory, since intuitions on this case are far
from univocal in favor of saying that such action is obligatory. But the basic
framework of Pogge’s argument can be reapplied to yield quite intuitive
conclusions as well. In particular, one’s institutional complicity is rather
clear in at least most cases of poverty among one’s own compatriots. So if
we have the intuition, as surely most of us do, that we should not simply
turn a blind eye to the poverty-related suffering of people we encounter in
our daily lives, Pogge’s argument is of some use to us.10
Of course, Pogge’s argument is controversial; indeed it’s probably fair to
say that the majority of what’s been written on it is skeptical of its sound-
ness, at least when applied to global as opposed to domestic poverty. I think
it’s sound, but it’s far beyond the scope of reasonable ambition to try to
establish that here. My point is simply to get across the point that nega-
tive moral conclusions—that some kind of action does not possess some
verdictive feature—are always provisional. So although the line I pushed
in §3—that conferring benefit is a not an obligation-grounding feature of
an action—certainly has counterintuitive implications, it might not have as
many as it would initially seem.
5.2 An Alternative Strategy in Cases Like Harmful Animal Research

I argued in §4 that harmful animal research is simply impermissible and
then admitted that this conclusion has terribly counterintuitive implications
for cases. Again, however, as with the case of poverty relief, this austere
conclusion may not have as many counterintuitive verdictive implications
as one would think. If it is morally impermissible to do harmful animal
research, then (it might be thought) that’s tantamount to saying that it’s
morally impermissible to do the only thing that can be done to save the lives
of people who have currently incurable deadly diseases.
I intend to take some of the sting out of this conclusion by arguing that
there is a morally permissible alternative way of conducting some of the
research that would otherwise be done on animals. Two caveats, however:
First, I don’t want to come off as cavalier, so let me admit straight off
that the implications of condemning harmful animal research are indeed
sobering. It’s tragic when otherwise promising human lives are cut short
due to disease, and if the conclusions of this chapter were to be taken seri-
ously then we might see many more tragedies of that sort. Second, some
experts on medical research, including the authors of a recent article in
BMJ (Pound and Bracken 2014), would argue that the losses from ceas-
ing harmful research on animals would not be that great and might be
outweighed by the benefits that would accrue from redirecting the funding
for such research to other forms of medical research, on the grounds that
animal research has certain inevitable scientific flaws. They may be correct,
but I will proceed as if they are mistaken. Although it goes without saying,
I want to admit that I am not an expert on medical research. Importantly,
I am unable to offer any sort of broad overview of the field of medical
research and separate that sphere of it that would be crippled by a refusal
to conduct harmful research on animals from the sphere of it that either has
no use for animal research or does have a use for it but could instead make
use of the alternative I am about to propose. I am simply going to propose
an alternative that surely would be useful for rescuing some research proj-
ects, and then leave it to the experts to tell us how much medical progress
would nevertheless be stalled if the argument given in §4 were taken seri-
ously. Now for my rescue effort:
Among those who take a strong position against harmful animal research
and want to minimize the effect their position would have on medical prog-
ress, the usual strategy is to argue is that computer models constitute a good
substitute for animal subjects for some of the research that could be done.
Without implying any reservations about the effectiveness of this strategy, I
want to propose an additional one: Use humans as research subjects.
This proposal raises an immediate question: If the research in question
would be objectionable if carried out on animal subjects, and this would
be on account of its harmfulness, wouldn’t its harmfulness make it wrong
in the case of humans as well? The answer is that in the case of humans we
can make an appeal to disablers. As I argued in §2, there are a variety of
conditions such that their presence can disable what would otherwise be
the impermissibility-grounding power of the fact of harmfulness. Of par-
ticular relevance for our purposes here is the fact of consent and the fact
of compensation. Humans can consent to have harmful medical research
conducted on them, and they can also be compensated for the harm.
This is not the case, or at least not in any straightforward way, in the case
of animals. Obviously, animals cannot offer consent in the morally relevant
sense—that is, consent that disables. They can have desires, no doubt, but
desiring something is insufficient for consenting to it; this is why we think
young children cannot consent to medical research. An adult human, on the
other hand, can be made to understand what the researcher is proposing to
do to her and can project her own existence into the future in such a way as
to develop a robust appreciation of what it will be like to suffer the possible
consequences, and these mental faculties seems to be what we take as neces-
sary for morally relevant consent.
The case of compensation is more complicated, but still there seems to be
a strong disanalogy between humans and animals even in this case. Humans
can be offered money, which is an all-purpose means for us to obtain many
things we might want. There is no analogue of money for animals. Of course,
this is not to say that there is nothing we can do to bestow benefit on them.
Certainly, we can benefit them by giving them food, shelter, ways to amuse
themselves, etc. But since these are laboratory animals, we’re morally obli-
gated to do these things for them anyway. Providing food, shelter, and means
of amusement for a living thing is simply obligatory when one has caused
that being to be in a position of dependence on oneself. (This certainly seems
to be what we want to say about the case of bringing a child into the world,
anyway.) And beyond these basic kinds of provision, it’s difficult to see what
more can be done to improve the life of a laboratory animal.
In any event, as I said, my proposal is to offer money to people for subject-
ing themselves to medical experimentation that would otherwise be done on
animals, when possible. The money would serve as compensation, which is
one disabler of the wrongness of harming them, and it would induce con-
sent, which is a second. In effect, then, I am proposing that we play it safer
than our best ethical theory would suggest we need.
Still, though, I acknowledge that my proposal will be seen as absolute
nonsense within research ethics circles. This is not to say that the idea of
using money to induce participation is novel or particularly controversial;
using money in this way is already common practice in later phases of the
research process, though there is a lot of hand-wringing even about this.
But I am proposing that we expand this strategy to cover earlier phases of
the research project—phases in which so little is known about the safety of
what is being tested that it is generally considered unthinkable to conduct
the test on humans. Admittedly, some of the substances that are currently
tested on animals are so likely to be hazardous and unlikely to be beneficial
that no sane, well-informed human would agree to be subjected to them, no
matter how much money were offered. This is just another way of making
the concession I already made earlier: Refusing to perform harmful animal
research will lead to substantial amounts of medical research simply ceas-
ing. (Though one would hope that, and it would make some sense if, the
research for which humans would in principle be suitable research subjects
but for which no amount of money can induce them to sign up will turn out
to be the least scientifically promising projects.)
Anyway, as I said, the position I’m taking here is about as far outside
the mainstream of theorizing on the ethics of human subjects research as
it is possible to be. The nearly universal belief is that potentially harmful
research on human subjects needs to meet many more conditions besides
compensation and consent in order to be morally permissible. But I have
already argued elsewhere, with respect to all the most popular of those con-
ditions, that they constitute overkill (Sachs 2010b; 2011). I won’t rehearse
those arguments here.
Notes
1. This last item requires a bit of explanation. The idea here is that there are cer-
tain goods the supply of which is limited, and so if one has an interest in obtain-
ing such a good that interest qualifies as a “competitive interest”—an interest
that can be advanced only at the expense of setting back another individual’s
similar interest (Feinberg 1984, 218–21).
2. Of course, what we rejected there was consequentialism as a monistic theory,
whereas all that’s being proposed here is to deploy a consequentialist explana-
tion of the wrongness of some wrongful actions, thus leaving open the possibil-
ity of deploying non-consequentialist explanations of the wrongness of other
wrongful actions. Note, however, that in Chapter 3 it was these very kinds of
case—harmful actions—for which we determined that consequentialism cannot
give a plausible explanation of their wrongness.
3. If the reader prefers a basis of comparison other than the two I entertain here,
that’s fine. The objection I’m going to raise against the supposed moral impor-
tance of harm prevention given a comparative account of harm holds regardless
of what the basis of comparison is.
4. This approach strikes me as so obviously right that it cannot be argued for. For
what it’s worth, Feinberg (1986, 149) and Parfit (1987, 374) agree with me.
5. One option would be to construct a theory on which the comparative and non-
comparative accounts of harm are each applicable to, and only to, distinct domains
of action. It’s worth recalling that the non-comparative account of harm was
developed originally by Shiffrin (1999) and Harman (2004; 2009) as a way of
accounting for the apparent possibility of harming an individual in causing her
to exist; it seems clear that the comparative account of harm can’t allow for this
possibility.
6. See, for instance, the absolutist positions on animal research adopted by Regan
(1983), Cohen (1986) and Frey (1980).
7. See, for instance, the non-absolutist positions on animal research adopted by
the consequentialists Singer (1995) and Frey (2011), and by McMahan (2005,
534–5), who is a consequentialist when it comes to our treatment of any indi-
vidual who falls below the ‘threshold of respect’, which includes all or almost
all animals.
8. This seems to be Nussbaum’s (2006, 403–5) and Slicer’s (1991, 121) view, and
they both do indeed use the word ‘tragedy’ to describe the situation.
9. See, for instance, Pogge (2005).
10. Of course, as I emphasized in Chapters 3–5, we should be concerned not just
with getting the intuitive verdict but with getting an intuitive explanation of
that verdict. Admittedly, Pogge’s argument doesn’t yield the conclusion that
acting so as to mitigate the poverty of those around is obligatory-on-grounds-
of-constituting-the-alleviation-of-suffering. So it doesn’t deliver a claim that we
would want to make. However, its explanation of the obligatoriness of acting
in this case—that it constitutes rectifying an institutional harm—is, at least, not
counterintuitive and is also not incompatible with the explanation we would
most want to offer (since it’s possible for an action to be obligatory on more
than one grounds).
9 The Distribution of Health
Care Resources
1. Introduction
In the previous chapter I illustrated how approaching moral theorizing as
an explanatory project would lead us to surprising substantive answers to
general and case-specific ethical questions. In this chapter my aim is to show
how approaching ethical theorizing as an explanatory project will lead us to
surprising conclusions regarding which case-specific questions have answers
that moral theorizing can reveal.
My case study will be the ethics of health care resource distribution, and
my conclusion will be that the ethical questions philosophers have been
asking and arguing about for decades now have either have no substantive
answer at all or no substantive answer of the sort that the tools of moral
theorizing are useful for discovering.1 An implication of this is that a good
chunk of what’s been published in applied ethics in the last 30 or so years
is a red herring.
To give the reader a feel for what my target is here, let me get specific. I
am talking about the literature that tries to work out the importance, for the
question of how health care resources should be distributed, of (among other
things) the number of people a particular resource is used to benefit, the age
of the individuals who need the resource, the extent to which the individuals
who need the resource can benefit from it, the current level of destitution of
the needy people, and the moral importance of giving the needy individu-
als at least a chance to receiving a needed resource. Among the prominent
contributors to this literature are Frances Kamm, Peter Singer, John Broome,
Dan Brock, Alex Voorhoeve, and Michael Otsuka.
Again, as was the case with my discussion of animal research in the pre-
vious chapter, I don’t want to seem as if I’m treating this matter casually. I
don’t take lightly the prospect of discarding as a red herring a large swath of
philosophical writings on any subject, never mind the distribution of health
care resources. For the record, I consider the public provision of health care
to be morally required (Sachs 2008b) and I’ve been interested in health
care resource distribution for longer than I’ve been interested in any other
philosophical issue. Furthermore, the philosophers working on this issue
144 The Distribution of Health Care Resources
are, needless to say, some of the smartest around and the answers they’ve
developed are some of the most interesting and nuanced to be found in any
field of philosophy. As interesting as the questions and proposed answers
are, however, I’ve arrived at the conclusion that there is nothing that could
ground the correctness or incorrectness of these answers. So once again the
problem is explanatory.
That there is nothing that could explain the truth of the proposed answers
follows in a somewhat straightforward fashion from the conclusions I
reached in Chapters 6 and 8. My aim in §2 is to show that this is the case.
Next, in §3, I concede for the sake of argument that I might have been
wrong in my conclusions from Chapter 6 or those from Chapter 8, but I
argue that, even so, there is another argument available to show that the
questions of health care resource distribution have either no answer at all or
no answer of the sort that can be derived through moral theorizing. Finally,
I review, in §4, the various twists and turns (of which there will be many) of
the arguments in §2–3.
2. Health Care Resource Distribution, Distributive Principles,

and Failing to Confer Benefit
2.1 Introduction: The Classic Thought Experiments in the

Literature on Health Care Resource Distribution
As I said in the introduction, my agenda for this chapter is to dispute the
idea that moral theorizing can be used for making progress on the ethics of
health care resource distribution. In the course of doing this I will focus my
attention heavily on a certain kind of thought experiment found in the liter-
ature on this issue. Although the thought experiments themselves aren’t my
direct concern here—my direct concern, rather, is the issue that the thought
experiments are supposed to illuminate—it’s important for me to attend to
them. Thought experiments of the sort I’ll identify serve to make people
believe that the real-world issues that the thought experiments are supposed
to mirror are of a sort on which moral theorizing can be usefully brought
to bear. So given that I want to argue that the real-world issues are not of
that sort, I need to undermine the work that the thought experiments are
doing. Now, to get a feel for this genre of thought experiment, here are a few
representative examples I’ve extracted from the literature:
Suppose two people each have a fatal disease. Each can be treated and
each would gain twenty years of life if she were treated. But one of
them is already disabled from another cause, so that treating her would
restore her to a life with disability. The other is healthy apart from
the particular disease, and treating her would restore her to a healthy
life. . . . It would be unfair to deny the first person her life on the
grounds that she is already suffering from disability. . . . If resources are
The Distribution of Health Care Resources 145
limited, the state would do more good treating the able-bodied person
than the disabled one. . . . How should we resolve this conflict between
fairness and doing the most good?
(Broome 1994, 37)
Suppose that we must choose between using scarce medicine to save

to patients [sic] lives or instead to save one patient who is a surgeon
and will save five other patients [sic] lives if she is saved. The five addi-
tional lives the surgeon will save are an indirect benefit of our saving
her . . . The Indirect Benefits problem is whether the additional five lives
that the surgeon would save justify giving him priority over the other
two patients who need the medicine.
(Brock 2003, 2)
Suppose, for example, that we have a choice between saving A’s life and
saving B’s, and alongside B is C who has a sore throat. Our drug that
can save B’s life can also in addition cure C’s sore throat. [N.B.: The
question posed here is whether we should give the drug straightaway to
B & C or instead toss a coin.]
(Kamm 1996, 101)
Thought experiments such as these are ubiquitous in the literature on the

distribution of health care resources.2 On the assumption that the writings
in which these thought experiments appear are supposed to address a real-
world moral problem, namely how we should distribute scarce health care
resources,3 the philosophers who employ these thought experiments can be
justifiably accused of making a significant methodological error in virtue of
creating flawed thought experiments to advance our thinking on that issue.
This is the thesis for which I will argue in the remainder of this section.
Specifically, those who think the thought experiments are helpful can be
confronted with a dilemma. The thought experiments propose that “we”
must choose how to distribute some health care resource, and the question
around which the dilemma is situated is the question of who is the ‘we’ that
faces the decision.
2.2 First Horn of the Dilemma

One answer is that each of us, as an individual, does or might face such
a moral quandary. If this is the answer, however, then the apparent quan-
dary may well constitute no moral problem at all. This is because of what I
argued in Chapters 6 and 8, namely that an action cannot be impermissible
on grounds of its violating some distributive principle nor obligatory on
grounds of its constituting the bestowing of a benefit on others. So there is
nothing to explain why it could be morally impermissible for an individual
moral agent to distribute her limited health care resources in one way rather
than another (barring further stipulations such as that the agent bears a
special relationship with one of the needy people).
Of course, what I said in Chapters 6 and 8 might be mistaken. But even
if I’ve been mistaken, there is still a problem, which is that in the real world
individual moral agents hardly ever find themselves in possession of surplus
health care resources. In general, we bother to obtain health care resources
only on an as-needed basis. The rest of the health care resources are con-
trolled by governments, insurance companies, medical charities, hospitals,
and (in the form of knowledge and skill) medical professionals. So on the
assumption that the ‘we’ in question is you or I—i.e., just some moral
agent—then it seems as though there isn’t a real-world moral problem to be
illuminated by these thought experiments.4
2.3 Second Horn of the Dilemma

The other horn of the dilemma proceeds on the assumption that the ‘we’
who confront the problem of how to distribute scarce health care resources
is not you or I as individual moral agents but rather collective moral agents
such as insurance companies, charitable hospital trusts, or states. This
changes everything. It may well be that such collective agents are obligated
to certain individuals to confer benefit on them and/or that such entities
are morally constrained by certain distributive generalizations whose truth
would be derivable from the truth of contractarianism/contractualism, as I
admitted at the end of Chapter 6.
The important point here is that contractarianism/contractualism has its
greatest plausibility when applied to collective entities such as corporations,
charities, and states. With respect to those institutions it is plausible, so I
argue (Sachs 2016), to think that some of the moral constraints to which
they are subject stem from a kind of contractual arrangement first put in
place by their creators in the form of a charter or constitution and bind-
ing on subsequent participants in the collective endeavor. It is less plausible
to suppose that the moral constraints governing interactions between indi-
vidual moral agents have a contractual source. (Even Scanlon could agree
to this; his claim, as I mentioned in Chapter 3, is that contractualism is the
correct story about what wrongdoing is, as opposed to a story about the
grounds of wrongdoing.) This is why I say that it may be the case that,
notwithstanding what I argued in Chapters 6 and 8, collective entities have
distributive obligations and/or obligations to confer benefit.
But this admission does not amount to the slightest vindication of the
thought experiments we have been discussing. For if the issue is whether
certain collective entities are obligated to use the resources at their disposal
to confer health-related benefit on certain individuals and how they should
distribute those resources when they’re limited, then the thought experiments
are very poorly designed. To see why, consider how things can go wrong with
regard to a different genre of philosophical thought experiment—the “tick-
ing time bomb” (TTB) case. Such thought experiments, like most thought
experiments, abstract away from real-world details to a certain extent. So,
for instance, in the TTB case, the authorities know that lives are in danger
from a ticking time bomb, they know which person has the relevant infor-
mation about the bomb, they know this information will be procured if and
only if that person is tortured, and they know that there is no way to prevent
death and destruction aside from finding and dismantling the bomb. These
assumptions are counterfactual; in the real world there is always some uncer-
tainty in emergency situations involving terrorist bombs. Now counterfactu-
ality, as Marcia Baron notes, is not always a bad thing. Often it’s very helpful
to have a certain “distance between the example and the point in support of
which the example is put forward”, as she puts it (2013, 631). But in TTB
cases the scenario in the thought experiment and the scenario about which a
philosophical point is supposed to be made are just the same: they are cases
in which there is something important to be gained from torturing someone.
And this makes the counterfactuality of TTB cases a problem, because when
we answer the question the thought experiments pose we are liable to be
misled into thinking we’ve discovered something relevant to the real world,
even though we haven’t (Baron 2013, 630–1).
Similarly, the typical health care resource distribution thought experiments,
such as the ones excerpted above, describe a situation that is of the same
kind as the real-world problem about which they are trying to make a point:
cases in which multiple people need some health care resource but there is
not enough of it to give all of the people all they need. This being the case,
it is very important that the thought experiments get the pertinent details
right. And this they fail to do. For when you are invited to consider a situa-
tion in which ‘we’ possesses some number of pills, and there are X number
of people who need a pill, and they differ in such-and-such ways, etc., it is
quite natural to assume that ‘we’ are individual moral agents who happen
to find ourselves in possession of a resource that others need for their health
and who are willing to part with it for the sake of those others. But such
stories falsely represent how we get our health care—i.e., almost never from
concerned individuals—and thereby commit the bigger sin of misrepresent-
ing the moral constraints relevant to the decision.5 If you are making an
allocation decision on behalf of some collective entity, then, assuming that
the contractarian/contractual case can be made for your collective entity
being morally constrained in its health care resource allocation decisions,
your job is to discern the distributive implications of the contract upon
which your collective entity is founded. The typical thought experiment,
by contrast, encourages us to think about health care resource distribution
decisions as if they wear their moral importance on their sleeve—that is,
as if they matter morally on their own, which the argument of Chapters 6
shows that they cannot.
The thought experiments I have been discussing here treat health care
resource distribution as a matter to be resolved through moral theorizing—
that is, through working out the general and particular truths about how
individuals morally ought to act. But in the overwhelming majority of cases
in which health care resource distribution decisions must be made, the answer
as to how to make them is a matter of political morality in the sense defined
at the end of Chapter 6. Therefore, one of the major inferences to be drawn
from the argument of Chapter 6 is that many moral philosophers have an
overblown view concerning which moral problems the tools of moral theo-
rizing can be marshaled for solving.
3. An Independent Argument for a Similar Conclusion

Supposing, however, that the conclusion of Chapter 6 or that of Chapter 8
was mistaken, it is still unclear what could explain the truth of the claims
philosophers make regarding how we ought to distribute limited health care
resources.
3.1 What if Failure to Confer Benefit Is a Morally Significant

Feature of Actions?
Let’s assume, first, that the conclusion of §3 of Chapter 8 was wrong, and
that failing to confer benefit is morally significant. That being the case, the
conclusion of Chapter 5 becomes quite relevant to the issue of the distri-
bution of scarce health care resources. In Chapter 5 we learned that non-
normative features of actions have their morally explanatory significance
by way of grounding verdictive features of actions, such as being impermis-
sible, obligatory, supererogatory, etc.6 Non-normative features of actions do
not, we learned, ground moral reasons for or against actions. So if an agent
has a pill and Jones and Williams each need it in order to avoid death, there
is no option to say that the prospect of conferring benefit on Jones and the
prospect of conferring benefit on Williams ground moral reasons in favor
of the agent’s doing each of those two actions. And what seems the most
sensible thing to say, having eliminated that option, is that the prospect of
conferring benefit on Jones makes it obligatory to do so and the prospect
of conferring benefit on Williams makes it obligatory to do so. Thus, the
agent faces a genuine moral dilemma, and hence there is nothing that could
explain the truth of the claim that the pill should be given to Jones or the
claim that the pill should be given to Williams. And all the variables on
which those who engage in the typical thought experiments tend to con-
cern themselves with—how many years of life Jones and Williams have to
gain, which one of them is worse off, whether either of them has children
who depend on them for sustenance, etc.—don’t change that fact. So those
thought experiments, on the assumption under which we are now operating,
constitute a waste of time.
3.2 What if Individual Moral Agents Are Constrained
by Distributive Principles?
Now let’s assume, instead, that the conclusion of Chapter 6 was in error,
meaning that individual moral agents can indeed be bound by distributive
principles and that the agent faces a situation in which she possesses a lim-
ited quantity of a certain health care resource relative to the number of
individuals who need some of that resource.7 I am going to argue in this sub-
section that even given this assumption there is nothing that could explain
the truth of some claim to the effect that the individual is obligated to dis-
tribute that quantity of that resource in one way or another.
For there to be a truth of this sort that has an explanation, the true dis-
tributive principle, whatever it is, would have to do the explaining. If one
believes that it does, one must say whether that principle applies directly or
indirectly to the distribution of that quantity of the resource that the agent
possesses. This constitutes a dilemma.
3.2.1 First Horn of the Dilemma: The True Distributive

Principle Applies Directly to the Distribution of the
Quantity in Question
Clearly, one should not take the former route. A distributive principle that
took as its distribuend the particular quantity of the health care resource
that the agent possesses would be much too narrow. It may be plausible to
say, for instance, that there is a true distributive principle that takes health
care resources as its distribuend. But it’s preposterous to think that there is
a true distributive principle that takes as its distribuend the particular pills I
have in my possession. The particular pills that I have in my possession are
not different enough from certain other goods—namely the other pills of
that type and the other therapeutic agents that serve purposes similar to the
purpose the pills are supposed to serve—to be such that the distribution of
them could be governed by its own principle.
To me this seems like an obvious point, which raises the question of why
anyone would ever have thought otherwise. I have a conjecture: Suppose that
some very broad distributive principle is true—say, that resources should be
distributed equally, or that welfare should be distributed equally. It’s all too easy
to allow oneself to slip into the mode of thinking that if egalitarianism is the
basic truth in distributive justice, then each one of us should be egalitarian. That
is, we should approach life with an egalitarian outlook. In other words, it’s in a
way natural to subtly convert principles of distributive justice from principles
distinguishing right action from wrong action into principles of virtue.
With other kinds of moral concerns it might be entirely appropriate to
make that conversion. Perhaps it’s right, for instance, to make an inference
from the fact that morality obligates us not to lie to the claim that morality
requires us to be disposed not to lie. But inferences of this sort are invalid
in the case of distributive principles, because there is only the most tenuous
of connections between having a certain distribution-related disposition and
promoting the realization of that distribution. Supposing, for instance, that
an egalitarian distribution (of some kind of good) is what we’re obligated
to bring about, it may or may not be helpful for me to have an egalitar-
ian disposition—a disposition to distribute equally the goods I control. It
depends on the status quo distribution and on what distributive disposition
each of the other agents has.
What we can conclude so far is that if there is a true distributive prin-
ciple that applies to the distribution of some kind of health care resource,
it doesn’t apply directly to the quantity of that resource that any particu-
lar agent possesses, neither in the form of obligating her to distribute that
quantity in a certain way nor in the form of requiring her to have a certain
disposition to distribute goods per se in a certain way.
3.2.2 Second Horn of the Dilemma: The True Distributive

Principle Applies Indirectly to the Distribution of the
Quantity in Question
It must be, then, that whatever the true distributive principle, it applies indi-
rectly to the case of an individual distributing some quantity of a health care
resource she possesses. What that means is that there is some larger class
of goods, of which the quantity of health care resource in question is one
instance, such that the distribution of that class is governed by some true
principle. As to what that larger class of goods is, there are broadly speak-
ing two possibilities, and trouble lies ahead no matter which possibility we
have in mind, thus constituting a sub-dilemma within the second horn of the
original dilemma.
3.2.2.1 FIRST HORN OF THE SUB-DILEMMA: THE LARGER CLASS OF GOODS

TO WHICH THE TRUE DISTRIBUTIVE PRINCIPLE APPLIES DIRECTLY
IS SOME VERY LARGE CLASS OF GOODS
The first horn is the claim that the class of goods to which the true distribu-
tive principle applies is some very large class of goods encompassing health
care resources and other goods besides. Suppose, for instance, that societal
resources per se should be distributed in an egalitarian manner. What can
we learn from this that would be relevant to the situation of an individual in
possession of a quantity of a certain kind of health care resource? Certainly
not that that quantity should be distributed in an egalitarian manner. To
make such an inference is to commit the fallacy of division—the fallacy of
making an inference from the fact that a class of things has a certain property
to the claim that each sub-class within that class has that property as well.8
Now one might respond that although the pursuit of a certain kind of
distribution of some large class of goods does not entail the pursuit of that
kind of distribution of any sub-class of goods within that class, the pursuit
of a certain kind of distribution of some large class of goods will necessarily
have some implications or other for how an individual should distribute the
limited sub-class of those goods at her disposal. So the idea would be that
the individual should do her bit. This of course is true, but totally unhelpful,
for we are in no position to judge what “her bit” is, when it comes to such
a monumental task as ensuring that the distribution of some large class of
goods, of which health care resources are just one sub-class, conforms to a
certain pattern.
3.2.2.2 SECOND HORN OF THE SUB-DILEMMA: THE LARGER CLASS OF GOODS

TO WHICH THE TRUE DISTRIBUTIVE PRINCIPLE APPLIES DIRECTLY IS
HEALTH CARE RESOURCES
The second possibility is that a true distributive principle applies to health

care resources per se. This has become known in the literature as the ‘sepa-
rate spheres’ thesis or ‘isolationist’ approach; connoting the idea that justice
or injustice can be done purely in virtue of how health care resources are dis-
tributed. The thesis is controversial (some skeptics: Segall 2010, 89–92; Mar-
mot 2004, 45–6; Sen 2004, 23–4; Peter 2004, 101ff; Gosseries 2017, 140–2;
Albertsen and Knight 2015, 166), but I’ll grant it for the sake of argument.
If there are separate spheres of distributive justice, this must be because
different kinds of good have their own internal distributive logic, so to
speak; i.e., differing patterns of distribution are appropriate to different
goods. Our question is what pattern of distribution would be appropri-
ate to the kind of good that health care is. The answer seems obvious: We
should distribute health care so that everyone has enough of it to be healthy9
(though they may not have enough of something else, such as luck, and thus
nevertheless wind up unhealthy). This talk of having ‘enough’ should lead
us to a sufficientarian principle, where the level of sufficiency is the level at
which the individual is able to be healthy.
Supposing the distribution of health care resources is governed by a simple
version of sufficientarianism-in-health-care, containing just one principle,
which says something along the lines of “everyone ought to have enough
health care to be healthy”, then a situation of health care resource scarcity
is, by definition, a situation in which it is impossible to fulfill the demands
of justice-in-health-care!10 Justice-in-health-care, under our current assump-
tions, requires that everyone get enough health care resources so as to be
healthy, but the very problem we’re assuming we must confront is that there
aren’t enough resources for that. So once again we would face a situation in
which there is a true principle of justice, given our assumptions, and even so
it is not clear what could explain the truth of the claim that the agent is obli-
gated distribute the resources at her disposal in one way rather than another.
The reality, however, is that contemporary versions of sufficientarian-
ism don’t resemble the simple kind I just sketched out. Some contemporary
versions begin with a principle along the lines of the one I sketched above
but add to it other principles that tell us how to prioritize people who are
below sufficiency in case not everyone below the level can be brought above
it (Huseby 2017). Other versions have no outcome-oriented principle at
all—i.e., no principle telling us to bring people up to sufficiency—and just
are a priority rule or set of priority rules prioritizing the needs of those who
are below the level of sufficiency (e.g., Ram-Tiktin 2012; 2017).11
The kind of sufficientarianism-in-health-care that is nothing more than
a priority rule or set of priority rules is implausible as a theory of justice-
in-health-care because it cannot condemn wasting health care resources.
However few health care resources we have at our disposal, we can always
abide by a priority rule governing their allocation. So if health care justice is
nothing more than abiding by such a priority rule, then wasting health care
resources isn’t a health care injustice.
We ought, therefore, to turn back to the version of sufficientarianism-
in-health-care that tells us to give everyone all the health care resources
they need so as to achieve sufficient health and also tells us what to do
when we don’t have enough resources for that. The challenge for this kind
of sufficientarianism-in-health-care is to explain the relation between the
unfulfillable principle and the principle that takes hold in the non-ideal
circumstance of scarcity.12 To see why this is going to be such a difficult
challenge we first need to distinguish two kinds of distributive principle.
First, there are comparative principles—so named because the basic content
of such a principle is an account of what makes one distribution morally
favored over its alternatives.13 Comparative principles can identify a par-
ticular distribution as ideal, but the ideality of such a distribution is merely a
matter of its being morally favored over its alternatives. Maximin is a classic
example of a comparative principle. This principle says that one distribu-
tion morally favored over another if and only if it is better for the worst-off
person. If maximin picks out an ideal distribution in a particular case, this
is because all the other possible distributions are worse for the worst-off
person. Non-comparative principles, on the other hand, identify a certain
distribution as morally favored in virtue of its intrinsic nature as opposed
to its relation to other possible distributions. The principle telling us to give
everyone enough health care so as to enable them to be healthy is a non-
comparative principle.
Only non-comparative principles can be unfulfillable. (For comparative prin-
ciples the agent’s task is just to do the best she can, given the circumstances.)
And the problem is that a non-comparative principle will not itself designate its
replacement for non-ideal circumstances. When a certain distribution cannot
be achieved, there will always be multiple possible ways of falling short of it. If
the original principle is to tell us anything helpful in these cases, it must tell us
in which way we ought to fall short. But a non-comparative principle cannot
provide a ranking of the relative moral desirability of the various departures
from its demands.14 The one distribution that is to-be-achieved, according to
a non-comparative principle, is to-be-achieved in virtue of its intrinsic nature,
not in virtue of any way in which it is an improvement over alternative dis-
tributions. So a non-comparative principle cannot countenance the idea that
those alternatives have any moral desirability at all. Again, take the example
of the principle of sufficiency. Suppose that you have eight apples to distribute
among four people, where a sufficient quantity of apples is three. Should you
give three apples to each of two people and split the remaining two apples
between the other two people? Or should you give everyone two apples each?
There is nothing in the principle of sufficiency that points you one way or the
other. Insofar as we are able to make a judgment as to the relative goodness of
those two alternatives, we are relying on something other than the principle of
sufficiency.
What this tells us is that in the version of sufficientarianism-in-health-care
under consideration, the original principle—the one telling us give everyone
enough health care so as to enable them to be healthy—isn’t going to entail
the auxiliary principle—the one that tells us what to do when we don’t have
enough resources to give everyone all the health care they need in order to
be healthy. Rather, the two principles must each be justified independently.
We might say that the two principles must be robustly independent.
Consequently the sufficientarian-in-health-care owes us an explanation
of what happens to the original principle when we’re condemned, because
of scarcity, to violate it and we turn our attention instead to fulfilling the
auxiliary principle. She shouldn’t simply say that the auxiliary principle sup-
plants the original principle. While we have been concerning ourselves with
the question of how one should distribute the resources at one’s disposal in
a situation of scarcity, this is not the only crucial moral question to which
scarcity gives rise. Clearly, one is also obligated to strive to eliminate the
scarcity.15 In order to explain the existence of such an obligation, we must
say that the original principle retains its normative force.16 There would be
no moral demand to eliminate the scarcity if the unachievable distribution
were not required by a still-valid principle.
The fact that the original principle still has force raises the troubling
possibility that doing what the original principle still requires in cases of
scarcity—i.e., eliminating the scarcity—could be incompatible with doing
what the auxiliary principle requires in those cases.17 This forces us to con-
front yet another dilemma, centered on the question of whether in cases like
this there is some sort of assurance that there will be no such conflict.
If there is no such assurance, then we have to confront the possibility that
a second auxiliary principle will in certain cases need to be introduced so as
to arbitrate the conflict between the original principle and the first auxiliary
principle. And then we have to admit that the introduction of the original
auxiliary principle didn’t resolve so much as perpetuate the non-ideality it
was supposed to address.
Yet, on the other hand, it is hard to explain how there could be a guaran-
teed absence of incompatibility between doing what the original principle
still requires in cases of scarcity (i.e., eliminating the scarcity) and doing
what the auxiliary principle requires. This would be explicable if the aux-
iliary principle was derived from the original principle, but we’ve already
determined that that’s not the case. Failing that, a necessary compatibility
would seem like too much of a happy coincidence.
I raise these questions mostly to give a feel for how complicated things get
once we really start thinking about the structure of the health care resource
scarcity problem under the assumption that health care is a separate sphere
of distributive justice. Having identified these complications, it seems myo-
pic to continue to think that we can make progress on the ethics of health
care resource distribution, on this assumption, through further moral theo-
rizing about the relative plausibility of different ways of distributing health
care resources under differing kinds of scarcity featuring people in differing
circumstances. What seems necessary, now that we have examined the mat-
ter more closely, is more theorizing about non-ideal justice.18
4. Summary of the Argument of This Chapter

The focus of this chapter was questions of health care resource distribution.
The thesis I advanced regarding these questions is that they have either no
answer at all or no answer of the sort that moral theorizing can produce.
I began by confronting my opponent with a dilemma: Either the ‘we’
referred to in the setting-up of the problem as to ‘how we should distribute
health care resource X’ is you and I as individual moral agents or it’s you
and I together as a collective agent such as a state. If my opponent takes the
former horn of the dilemma, then the conclusions of Chapters 6 and 8 pres-
ent a formidable challenge by implying that there simply wouldn’t be any
moral constraints applicable to an individual moral agent if she found her-
self in the kind of situation described above. On the other hand, if the state
or another collective agent found itself in such a situation, the arguments of
Chapters 6 and 8 allow that it could be morally constrained in its choice of
how to distribute the scarce resource; however, in this case the moral ques-
tion is one for political morality as opposed to normative ethics.
Second, I withdrew the claims argued for in Chapter 8 and allowed for
the sake of argument that individuals are morally obligated to confer ben-
efit on others. I then argued that, this being the case, cases of health care
resource scarcity are genuine moral dilemmas. This once again brings us to
the conclusion that there is nothing moral theorizing can do to point us to a
resolution of such situations.
Third, I withdrew the arguments of Chapter 6 and allowed for the sake of
argument that individuals are constrained by distributive principles. I then
confronted my opponent with a dilemma: Either the relevant distributive
principle applies directly or indirectly to the distribution of the quantity
of health care resources in the individual’s possession. Grabbing the first
horn of the dilemma is a very unattractive proposition, I argued, as it puts
one in the position of arguing that some particular small quantity of health
care resources that some individual possesses is the object of a distributive
principle.
The second horn of the dilemma is more complex. If my opponent takes
that route, she confronts a further dilemma. Either the distributive principle
applies indirectly by way of applying to some broad index of goods or indi-
rectly by way of applying to health care resources per se. If it’s broad, by
which I mean it takes as its distribuend some broad index of goods such as
resources per se, then we can extract no conclusions regarding what should
be done by the individual moral agent who faces the decision of how to dis-
tribute the limited health care resources in her possession. If the principle is
narrow, meaning that it applies specifically to health care resources, then it
must be a sufficientarian principle, since that’s clearly the most sensible prin-
ciple to apply to the distribution of a good, such as health care resources,
whose value disappears entirely once one has a certain amount of them. But
if it’s a sufficientarian principle then it’s unfulfillable, by definition, in cases
of scarcity. Admittedly, this leaves an interesting moral question of what
one should do in such a non-ideal circumstance. However, what’s needed to
answer that question is an investigation into the nature of non-ideal moral-
ity, not ordinary moral theorizing.
Notes
1. Recall that in Chapter 1 I defined moral theorizing quite broadly as the project
of reasoning about theoretical and applied ethical questions all at once. So the
thesis, just mentioned, that I seek to advance in this chapter doesn’t follow trivi-
ally from some narrow understanding of what moral theorizing is.
2. There are many more thought experiments like this in Kamm (1996). There are
also thought experiments like this in Broome (1995; 1988), Parfit (2011, 353),
Rakowski (1994), and innumerable other articles and books.
3. There is another possibility, which is that these thought experiments are being
used not to answer some question in applied ethics but rather to advance our
thinking about distributive ethics—for instance the question of the relative
plausibility of egalitarian, prioritarian, and sufficientarian principles. This being
the case, there is a flawed presupposition behind these thought experiments,
namely that distributive generalizations can have moral force of their own. This
cannot be the case, as I argued in Chapter 6.
4. Of course, if we broaden our understanding of what a health care resource is,
then the current problem vanishes. In particular, if money counts as a health
care resource, on account of the fact that it can be used to purchase things that
qualify as health care resources on a narrower construal of that term, then indi-
vidual moral agents do in fact frequently find themselves in possession of surplus
health care resources. But the real-world circumstances that fit this description
are drastically different from the stipulated circumstances of the typical thought
experiments. For one thing, there is often great distance between the agent who
possesses the resource and the individuals who could benefit from it. This is
the case, for instance, for denizens of developed western nations and needy
people in sub-Saharan Africa. For denizens of developed western nations, the
choice they face is which charitable organization to give their money to, not to
which individual to give their money. Already this is a big difference between
the real world and the stipulations of the thought experiment. In any event, if
these thought experiments really are supposed to advance our thinking about
the ethics of poverty alleviation, then they are terribly misleading because it is
never said that that’s what the thought experiments are for, nor do the authors
of the writings in which these thought experiments appear generally make any
attempt to situate their thoughts within, or respond to, what has already been
said in the philosophical literature on the ethics of poverty alleviation.
5. In fairness to one of the major users of this kind of thought experiment, Frances
Kamm, it should be mentioned that she often notes that it makes a moral dif-
ference who is doing the distributing of the resource (1996, 144, 185; 1994).
But she nonetheless draws conclusions from such thought experiments without
specifying who is doing the distributing.
6. Non-moral features of actions can also, as initially mentioned in Chapter 2
and reinforced several times since then, have moral significance by constituting
enabling or disabling conditions. But we can set this possibility aside for now.
7. The reader might have noticed that I’m now speaking of ‘distributive principles’
as opposed to continuing to speak, as I did in Chapter 6, of ‘distributive gener-
alizations’. I think of principles as claims that state verdictive facts about broad
act-types and are derivable from generalizations. So, for instance, from the gen-
eralization, ‘under certain circumstances an act’s being a failure to distribute
welfare equally can ground its impermissibility’, one can derive the principle,
‘under certain circumstances failing to distribute welfare equally is impermis-
sible’. I’ll continue to focus my attention on distributive principles, as opposed
to the generalizations from which they’re derived, for ease of speaking.
8. The mistaken inference is made by Culyer (2001, 276) and Culyer and Wagstaff
(1993, 452). They begin by accepting the principle of equality of opportunity
to flourish and pointing out that health is a necessary condition for flourishing.
From this they infer that health should be distributed equally. But what actually
follows from their assumptions is that the necessary conditions for flourishing—
the set of them as a whole—should be distributed equally. Anand (2004) makes
the same mistake, except that Anand accepts a different version of equality of
opportunity.
I also suspect, but am not certain, that Ronald Dworkin makes this infer-
ence. Dworkin accepts the principle of equality of resources, and conceives of
health care as a resource. He offers an account of what it is for resources to be
divided equally (the hypothetical auction and the envy test), and then applies
this account directly to the distribution of health care (Dworkin 2000, esp.
Chapter 9) Thus, from his claim that resources should be distributed equally,
Dworkin appears to be inferring that health care should be distributed equally. But
of course there are other resources besides health care. (For further discussion
of this issue, see Jacobs (2004, 178–92).)
Finally, Shlomi Segall appears to make this inference. Segall believes that it
is unfair for one person to be made worse-off (opportunity-wise) than another
due to reasons beyond her control (Segall 2010, Chapter 1), and he infers that
this principle should regulate the distribution of some health care (ibid., 93).
But of course there are other opportunity-affecting goods aside from health
care. Segall seems to rescind this inference immediately after making it, so it is
difficult to determine exactly what his position is.
Larry Temkin (2011, 73–4) pointedly refuses to make such an inference. John
McKie et al. (1996) are also careful not to make this mistake.
9. This basic thought seems to be endorsed in the Universal Declaration of Human
Rights, Article 25 of which says that “everyone has the right to a standard
of living adequate for the health and well-being of himself and of his family,
including food, clothing, housing and medical care and necessary social ser-
vices”. It’s also endorsed by Powers and Faden (2006, esp. pp. 23, 61), Ram-
Tiktin (2012; 2017), Alvarez (2007), Ruger (2010), Daniels (2008), and Axelsen
and Nielsen (2017, 116–17). There’s now an entire edited volume on sufficien-
tarianism in health care, edited by Carina Fourie and Annette Rid, who them-
selves are sympathetic to the approach (Fourie and Rid 2017, 8).
10. Alvarez (2007), a sufficientarian when it comes to health care, concedes this
point.
11. Hirose (2017) constitutes another example of a version of sufficientarianism
that just is a priority rule. However, it’s a priority rule for ranking states of
affairs, not for ranking actions. Consequently, it is not a target for me since it
doesn’t attempt to pull off the feat that I am arguing a normative ethical theory
cannot accomplish—the feat of saying something helpful about what we ought
to do in cases of health resource scarcity.
12. Powers and Faden (2006) describe the problem of distributive justice in health
care as one of non-ideal justice and are concerned with problems of resource
scarcity, but they do not examine this question.
13. I use the ‘morally favored’ language as a way of keeping the discussion non-
committal enough to cover both kinds of distributive principles—those that tell
us which distribution(s) are (how) morally good and those that tell us which
distribution(s) we morally ought to bring about.
14. Sen (2006, 219–20) makes basically this same point, though he makes the mis-
take of not explicitly restricting his claim to non-comparative principles.
15. Fourie (2017, 196). This is just an instance of the more general truth that part
of what it means to achieve justice in non-ideal circumstances is to work toward
eliminating the non-ideal circumstances. I am not aware of anyone who denies
this claim, and several theorists explicitly affirm it. See, for instance Rawls
(1999b, 89), Simmons (2010), Sreenivasan (2007), Robeyns (2008), Dunn (1990),
and Shields (2017, 92).
16. Efrat Ram-Tiktin, a sufficientarian in health care who appeals to an auxiliary
principle, fails to acknowledge this. See Ram-Tiktin (2012).
17. For instance, suppose that some of the scarce resource one does possess could
be invested so as to yield a greater quantity of that kind of resource at a later
time, thereby guaranteeing that there will be no scarcity at that later time. In
this case the imperative of eliminating the scarcity would require investing some
of the resource while the auxiliary principle, whatever it is, may well require
immediately using all of it.
18. Granted, a great deal of theorizing about non-ideal justice has been done already,
but the two kinds of non-ideality with which that literature is concerned—
non-compliance and the absence of political institutions—are different from the
problem of scarcity, which is our concern here. (These two kinds of non-ideality
became the focus of theorizing on non-ideal justice because Rawls (1999a, 216)
specifically defined non-ideal justice as justice in the circumstance in which
either there is a lack of well-ordered institutions or there is widespread non-
compliance with the demands of those institutions.)
10 Conclusion
Having laid out a moral methodology in Chapters 1–2 and argued against
competing normative ethical theories in Chapters 3–4, I’ve spent the rest of
this book exploring a new family of theories called non-Rossian pluralism
(NP). I’ve been developing it piece-by-piece, and consequently at no point
have we had the opportunity to assess its plausibility as a whole. Of course,
even now we’re not really in a good position to do so, since there are many
topics on which a normative ethical theory has to take a stand that we as
yet haven’t inquired as to what NP will say about. What we can do, and
what I will do in this brief concluding chapter, is to examine how it’s doing
so far, in light of the topics on which we know NP’s stand. Admittedly, when
I say we “know” NP’s stand on a certain issue I mean only that I’ve made
an argument in this book that it should take some particular stand, and of
course any of those arguments could be flawed. For now, though, assessing
my admittedly incomplete version of NP is all that can be done by way of
assessing the prospects for the theory. So that’s what I’ll do.
I’ll start with the bad news for NP, which is that the verdictive implica-
tions I extracted from NP in Chapters 8 (except §2) and 9 were extremely
counterintuitive.1 To begin with Chapter 9: Some of the thought experi-
ments commonly used in the literature on health care resource distribution
have apparently obvious answers. For instance, the lifeboat case in which
one can bring one’s lifeboat to an island that contains a 20-year-old in good
health who will live another 60 good years if saved or an island that con-
tains an 80-year-old in permanently poor health who will live just one more
year if saved, but one cannot do both, yields the apparently clear conclu-
sion that one is morally required to save the 20-year-old. Other thought
experiments are more difficult: For instance, if you have two doses of a pill,
should you give them to the person who needs both doses in order to not
die, or should you give one dose to each of two people who will lose the use
of their legs if they don’t receive a dose? Although people would probably
have contrasting intuitions about this case, it’s a safe bet that nearly every-
one would have what we might call the ‘meta-intuition’ that we can expect
the correct normative ethical theory to instruct us to do one or the other.
Yet, I showed in Chapter 9 that NP implies that all of these intuitions are
Conclusion 159
mistaken. Regarding the scenarios like the ones just given, NP will either tell
us that there’s nothing at all that the agent should do or that the situation
is a genuine moral dilemma, or it will tell us to look somewhere else for an
answer.
This simply compounded the problem we confronted in Chapter 8, where,
again, I argued from NP to some highly counterintuitive conclusions. In par-
ticular, I argued that if we could develop a medical intervention that would
save many, many people from a serious disease, but only at the cost of causing
harm to a few animals who would have to be used as research subjects in
the course of developing the intervention, it would be wrong to do so. This is
just representative of a potentially infinite set of actions, of the type prevent-
something-very-bad-from-happening-to-some-individuals-at-the-cost-of-
causing-mild-harm-to-some-other-individual(s), that will intuitively strike us
as permissible (maybe even obligatory, especially if the other individual is one-
self) but that NP will condemn as impermissible.
This double blow from Chapters 8–9 can be softened a bit, as I tried to
show earlier. In Chapter 9 I argued that insofar as health care distribution
thought experiments are constructed in a realistic way, it is likely that the true
political morality will yield answers to them. And in §5.1–2 of Chapter 8, I
tried to show that accepting the main conclusion of that Chapter—i.e., that
there is no duty of rescue—is compatible with it being morally required to
make significant sacrifices for the sake of impoverished people elsewhere in
the world. Still, in both cases highly counterintuitive results remain.
Actually, “highly counterintuitive” is an understatement. I assume we would
want to take it as a fixed point, for purposes of reflective equilibrium, that
saving the child in the Pond Case is obligatory, that if we can save a million
humans by doing mildly painful research on a few mice that we ought to do
it, and that morality requires us to save the 20-year-old in the lifeboat case
mentioned earlier. If we had started from these fixed points in our effort to
identify the most plausible family of normative ethical theories, we wouldn’t
have given NP a second thought.
So it would be helpful to recall how we got to this point. Perhaps the most
obvious way for a non-consequentialist to get what she wants in the Pond
Case, the case of harmful animal research, and lifeboat cases is to avert to
the balance of moral reasons. If we had this tool at our disposal, then it
would be easy to vindicate our verdictive intuitions about those cases. But
in Chapter 4 we decided to deprive ourselves of this tool. Of course, we had
a good justification for doing so: Grounding explanations that invoke moral
reasons are poor explanations. So what’s become apparent now is that the
price we pay for insisting on good explanations is being able to explain less
of what we want to explain.
Still, however, it’s not clear that we should pay that price. What I call
‘insisting on good explanations’ is just another case of relying on intuitions—
explanatory intuitions, specifically. Furthermore, as I noted at the end of
Chapter 5, NP puts us in a tight spot with regard to moral dilemmas: Since
160 Conclusion
it implies that every moral conflict is a genuine moral dilemma, it implies,
on the assumption that there are a lot of moral conflicts, that there are lots
of moral dilemmas.2
On the other side of the ledger, it bears mentioning that the explanatory
intuitions that guided us toward NP were themselves fixed points; the judg-
ments on which I drew in discarding balancing versions of moral pluralism,
as well as the intuitions I used in Chapter 3 in arguing against the most
popular versions of moral monism, were very strong ones. Moreover, NP
sets us a deliberative task that we have the tools to carry out, while Rossian
Pluralism fails to do so. Furthermore, NP can allow that acts of cruelty to
animals wrong those animals; no other normative ethical theory surveyed
here gets that result.
There is no way to make a serious attempt at totting up the total counter-
intuitiveness of NP and comparing it to the total counterintuitiveness of its
competitors; that would require figuring out how to individuate the (pos-
sibly infinite) cases on which we already know what the various theories
will say and coming up with an interval measure of counterintuitiveness. So
instead I will oversimplify drastically and assess the relative acceptability of
NP and it competitors in their oversimplified situations.
First, some of the competitors should simply be discarded. For instance,
I argued in Chapter 3 that act-consequentialism, Kantianism, and contrac-
tarianism each deliver terribly counterintuitive explanatory implications as
stand-alone normative ethical theories. Furthermore, in their original forms
they also each deliver terribly counterintuitive verdictive implications. Act-
consequentialism, for instance, delivers terribly counterintuitive implications
in the famous Transplant Case (and cases like it) and in cases concerning
giving to charity. Kantianism, meanwhile, delivers terribly counterintuitive
implications about masturbation, dishonesty, and suicide. And contracta-
rianism delivers drastically counterintuitive implications about our moral
obligations to disabled humans. These theories have little to recommend
them. However, there are reconstructed versions of each these theories that
can deliver intuitively plausible verdictive implications; these should stay in
the picture. Rossian Pluralism must be given consideration as well, because
it too can deliver intuitively plausible verdictive implications.
My oversimplification of the respective situations of the various theories
is this: NP has intuitively plausible explanatory implications, while some
of its competitors, including the ones mentioned just now, have equally
intuitively plausible verdictive implications. And NP has terribly counter-
intuitive verdictive implications while some of its competitors, including
the ones mentioned just now, have equally counterintuitive explanatory
implications.
Broadly speaking, then, my impressionistic representation of the conflict
between NP and its competitors is that they each have one very important
virtue and one very important defect.3 This, one might think, leaves us at a
stalemate.
Conclusion 161
One thing I expect that my readers will want to say about this is that this
is no stalemate because it is much more appealing to give up the explana-
tory intuitions with which the competitors to NP were shown, in Chapters
3–4, to be inconsistent than to give up the verdictive intuitions with which
NP was shown, in Chapters 8–9, to be inconsistent. This seems to tilt the
balance against NP. In response, I suggest we take care not to confuse the
phenomenology of finding a claim intuitively plausible with the phenom-
enology of wanting to believe a claim. For instance, I find myself strongly
drawn toward believing that 9/11 was not an inside job. On reflection, I
can distinguish at least two sources of this feeling. One is just the intuitive
plausibility of the claim. But another is that I just want to believe it; I would
feel like a crazy person if I didn’t. But the latter is the wrong kind of reason
for a belief. Similarly, with regard to the Pond Case and others like it, for
many of us there are probably two ways in which we are drawn to believe
that failing to save the child is wrong: The claim is intuitively plausible and
we would feel like bad people if we denied it.4 Again, however, the latter is
the wrong kind of reason to believe it. Likewise, we should be careful not to
confuse the phenomenology of finding a claim intuitively plausible with the
phenomenology of wanting a claim to be true. If we think that some people
are sometimes motivated to do the right thing because it’s right, and if we
think that true moral claims are (or will in the future be) more likely to
be believed than their denials, then we must conclude that a moral claim’s
being true makes it (or will in the future) make it more likely that the action
it recommends will be done or that the action it forbids will not be done.
This makes it rational to care which moral claims are true. It makes it ratio-
nal, for instance, to want it to be true that it would be permissible to save
a million innocent people from death by means of breaking the leg of some
other innocent person. (One is much less likely to be the one than to be one
of the million.) But, once again, wanting something to be true is the wrong
kind of reason to believe it. So while we might feel more reluctant to give
up our verdictive intuitions in the Pond Case, the save-a-million lives case,
and similar scenarios than we are to give up our strongly held explanatory
intuitions, reluctance brings in too much that’s irrelevant; it brings in the
wrong kinds of reason for belief.5 If we concentrate our attention exclu-
sively on how plausible the relevant judgments seem, I contend that we
won’t find a difference.
Thus, we return to the idea that we are at a stalemate. I contend, however,
that we are not. Although we cannot, based on my earlier oversimplified
summary of their respective situations, choose between NP and its com-
petitors based on which does the least violence to our intuitions, we can
choose between them based on the extent to which it is rational for us to
trust the intuitions to which they do violence. There are, as I discussed in
§3 of Chapter 1, two arguments in the philosophical literature purporting
to show that our moral intuitions are not particularly reliable guides to the
moral facts, assuming there are such facts: Greene’s empirical argument and
162 Conclusion
Street’s evolutionary skepticism argument. And I argued that both of those
arguments are at their strongest when targeted specifically against our ver-
dictive intuitions. This makes it rational for us to trust those intuitions less
than our explanatory intuitions.
That conclusion, however, comes along with some enormously important
caveats. First, it’s far from a settled matter whether either of those arguments,
never mind both, have any force at all. Second, the fact that Greene’s empiri-
cal argument has less force against explanatory intuitions may be entirely
down to the fact that no one has as yet attempted to conduct the relevant
kinds of studies on such intuitions; it may be that such studies, were they
to be conducted, would demonstrate that explanatory intuitions are suspect
in just the same way as Greene argues our verdictive intuitions are. Third,
it’s certainly possible that there are other arguments out there to be made—
arguments targeted specifically against the claim that our explanatory intu-
itions are reliable. It might be that the only reason that philosophers have
developed two arguments targeted against verdictive intuitions and no argu-
ments targeted against explanatory intuitions is that explanatory intuitions
have been flying under the radar in virtue of the fact that it hasn’t been
widely acknowledged that they constitute an indispensible input to moral
theorizing.
My conclusion, therefore, must be the following extremely weak one.
Provisionally—for want of a more rigorous appraisal of the strengths and
weaknesses of NP and its competitors as opposed to my incomplete and
impressionistic assessment, and failing a resolution of the debate over the
validity of Greene’s and Street’s arguments, and in the absence of any serious
investigation of the reliability of our explanatory intuitions—accepting NP
is at least as rational as, and maybe more rational than, accepting any one
of its competitors.
Notes
1. I would add, though, that I chose the topics of Chapters 8 and 9 because NP deliv-
ers counterintuitive implications in those domains. (It would have been a much
less interesting task exploring what NP has to say on questions about which it
delivers the verdicts we were already inclined to accept!) So NP might do better
in many domains than it does in the ones I explored.
2. This worry, however, turns out to be less serious than it seemed at the end of
Chapter 5. In Chapters 6–9 I argued for removing certain non-normative facts,
including some very widespread ones—especially the fact that some action would
confer a benefit on someone—from our list of requirement-grounding facts. Hav-
ing done this, we can conclude that moral conflict is a much less frequent phe-
nomenon than we would have initially been inclined to believe.
3. Jeff McMahan (2013b, 34) has argued, on entirely separate grounds, for this same
kind of conclusion—i.e., that it’s unlikely we’ll be able to come up with a norma-
tive ethical theory that doesn’t have drastically counterintuitive implications.
4. This worry is not reasonable when the concern is that one will adopt a mistaken
belief in the opposite direction. No one could plausibly be labeled a bad person
for mistakenly believing that she is morally required to φ. (Cases in which φ-ing
Conclusion 163
is evil do not constitute counterexamples to this generalization; if φ-ing is evil
then what might make you a bad person isn’t believing that φ-ing is morally
required but rather believing that φ-ing is morally permitted.) So insofar as we
allow these considerations to affect our belief-forming practices, we have a bias
toward adopting an overly strict morality.
5. This, I contend, is another good explanation of the phenomenon for which I
already offered one explanation (Chapter 1, §6), namely the phenomenon of
moral philosophers accepting the validity of the method of reflective equilibrium
but frequently violating it anyway (specifically, by being more attached to their
verdictive intuitions than their explanatory intuitions).
Appendix A
Moral Meta-Explanation
I begin by reproducing the table that appears in §2 of Chapter 3:
Table A.1
Natural Way to Alternative Way to Represent the

Represent the Theory, Theory, on which the Key Non-
on which the Key Normative Fact Identifies the
Non-Normative Circumstances Under Which, and
Fact is a Ground of the Explanation as to Why, Other
Verdictive Facts Non-Normative Facts Ground the
Verdictive Facts
Act- All verdictive factsNon-normative facts X, Y, and Z

consequentialism are grounded in each can ground the fact that some
facts about the action is/isn’t permissible, and in fact
maximization of the do ground that fact just when, and
good. because, that action does/doesn’t
maximize the good.
Kantianism All verdictive facts Non-normative facts X, Y, and Z
are grounded in each can ground the fact that some
facts about obeying action is/isn’t permissible, and in fact
or not obeying do ground that fact just when, and
the Categorical because, that action does/doesn’t
Imperative. obey the Categorical Imperative.
Contractarianism/ All verdictive facts Non-normative facts X, Y, and Z
Contractualism are grounded in facts each can ground the fact that some
about what could action is/isn’t permissible, and in fact
be agreed to under do ground that fact just when, and
certain circumstances. because, that action is/isn’t in accord
with what could be agreed to under
certain circumstances.
This table lays out a common sense and an alternative interpretation of

each of three main varieties of moral monism. Under the alternative inter-
pretations, the key non-normative fact around which each variety of moral
monism is constructed is construed not as an explainer of verdictive facts
but rather as a meta-explainer—that is, as a fact whose instantiation or lack
thereof explains why other facts can explain the verdictive facts.
Appendix A 165
The main attraction of having a theory of meta-explainers is that it gives us
the ability to explain why enabling and disabling happen, when they happen,
which are phenomena that cry out for explanation. Suppose, for instance,
that Jones tells a lie, and that we think that the fact that an utterance is a lie
can explain the impermissibility of that utterance, but in this case Jones’s lie
actually wasn’t impermissible, and there was no other factor in the case that
could have overridden the moral force of the fact that Jones lied. Here we
have a case in which something must have disabled the explanatory power
of the fact that Jones’s action was the telling of a lie. A meta-explanatory
theory would not only identify the disabler but also explain why it can do
its disabling work. For instance, if Jones’s lie was about the strength of the
poker hand he was holding, the act-consequentialist can say, plausibly, that
the telling of that lie did not lead to lesser overall good as compared to Jones’s
alternative courses of action. This fact is a disabler for act-consequentialism
as we are now understanding it, and the act-consequentialist can explain
why it does its disabling work, namely because the explanatory force of
the fact of lie-telling is explained by the explanatory force of reducing the
overall good.
I admit that we should want an explanation of why certain facts can do
enabling and disabling work and that this speaks in favor of adopting a
theory of moral meta-explanation (i.e., a theory of the type that appears in
the right-hand column of Table A.1), but I insist nevertheless that we should
not adopt any such theory. The trouble is this: Any plausible moral meta-
explanation is likely to strike us as an even more plausible moral explana-
tion. To take up the poker example again, suppose that telling a lie in the
context of a poker game doesn’t reduce the overall good as compared to
other possible courses of action. Now the question becomes, what explana-
tory role, if any, does such a fact play? One possibility is to assign it the
meta-explanatory role. But another possibility is to assign it the explanatory
role, by which I mean saying that the fact that an action doesn’t reduce the
overall good as compared to other possible actions can ground the permis-
sibility of that action, and then pointing out that in the poker case Jones’s lie
has that feature. I doubt anyone will want to take former course once they
recognize the availability of the latter course.
In other words, if our moral reflection leads us to the conclusion that
whether a lie reduces the overall good as compared to other possible actions
can make a difference as to whether that lie is impermissible, we’re likely
to think we’ve made a discovery about what explains the impermissibility
of lying when it is impermissible. We’ll be inclined to admit that we were
wrong to think that this grounding work could be done by the fact that the
utterance is a lie, and we’ll be inclined to say instead that the grounding
work can be done by the fact that the utterance reduces the overall good as
compared to other possible actions.
Pekka Väyrynen, who believes that there is some explanation for each ver-
dict-explaining fact,1 of why it does its explanatory work,2 discusses a posi-
tion on moral meta-explanation similar to the one I just took. He realizes
166 Appendix A
someone might say that the notion of a fact that explains why some other
fact can do morally explanatory work is superfluous, because any fact that
we might want to assign this meta-explanatory role we could instead assign
the explanatory role (2009, 101). But this is not my worry; I have no objec-
tion to the notion of a meta-explanatory fact. I just think that in each case
in which we recognize that we have an option to deploy it, we’ll prefer on
substantive moral grounds not to do so once we’ve considered the alterna-
tive of instead deploying the explanatory role option.
It’s likely out of meta-ethical concerns that someone would make the con-
trary choice, I suspect—in particular the meta-ethical concern with which
we started this discussion: the concern to explain why certain facts can do
enabling or disabling work. Meta-explanatory facts are built to play that
explanatory role. This is a valid meta-ethical concern; the idea of leaving
enablers and disablers without a normative explanation is quite unappeal-
ing (at least to me). But, then again, leaving any moral phenomenon without
a normative explanation is unappealing, isn’t it? And yet normative explana-
tion must run out; there is normative bedrock somewhere.3 I think instances
of enabling and disabling constitute some of that bedrock.
Is there any other reason to refuse to say that facts about what can enable
what and what can disable what are normative bedrock and to leave the
explanatory power of explainers unexplained? Väyrynen, in passing, seems
to suggest a second reason (aside from the one we just examined): that in
positing that some fact does some enabling or disabling work we don’t want
to be “just guessing” (2009, 95).4 My response is that fortunately we have
intuitions to appeal to, so we needn’t resort to guessing. We have intuitions
about which facts can do enabling or disabling work.5 So we’re in no worse
of an epistemic position with respect to identifying enablers and disablers
than we are with respect to figuring anything else of moral substance. We
start with our intuitions, and work our way forward from there.
I began this appendix by noting that there are alternative formulations
of act-consequentialism, Kantianism and contractarianism/contractualism
available—the formulations in the right-hand column of Table A.1 on which
those theories
a) represent distinctive views regarding which facts explain why other

facts explain the verdictive facts, but
b) don’t represent distinctive views on which facts explain the verdictive
facts.
The arguments I subsequently gave lead us to the conclusion that anyone

who is enticed by one of the formulations in the right-hand column is going
to be even more enticed by the original view from which it is a departure—
the formulation in the center column. (At least, that is, unless they endorse
the alternative formulation specifically for the sake of getting around an
objection to which the original view is vulnerable, in which case the alterna-
Appendix A 167
tive formulation is ad hoc.) So it makes sense to do what I do in Chapter 3:
assess the prospects for the original versions of these views.
Notes
1. Actually, Väyrynen speaks of reason-explaining facts, but he certainly would be
willing to extend what he says about reason-explaining facts, mutatis mutandis,
to verdict-explaining facts.
2. Kagan (1988, 20–1) believes this too. He calls the explainers of the explainers the
“foundations” of normative ethics.
3. This doesn’t amount to an endorsement of normative foundationalism. Norma-
tive foundationalism is a claim in epistemology, whereas I’m talking about nor-
mative metaphysics.
4. In another work (2006, 719–20), Väyrynen offers a third argument—an argument
by analogy. He notes that in the case of epistemic verdicts—like the verdict that
one ought to believe X—we think that explainers aren’t bedrock. We tend to
think that there’s always an explanation as to why some fact explains such a
verdictive fact. So, for instance, if the fact that the dog is a whippet explains the
fact that I ought to believe that the dog is fast, this is presumably in virtue of how
the fact that the dog is a whippet raises the probability of the dog’s being fast, or
something like that.
As Väyrynen admits, the analogy is worrisome for someone like me only if
epistemic reasons and moral reasons share a nature. The idea that they do is, as
Väyrynen notes, widely accepted. But I’m skeptical. I don’t even think moral rea-
sons share a nature with other reasons for action. And my position on this isn’t
motivated by a desire to resist Väyrynen’s argument here. It’s motivated, instead,
by a desire to be able to give good explanations of both the moral verdicts about
actions and the other practical verdicts about actions; I give this argument in
Chapter 5.
5. Suppose, for instance, that I lie to you about where I’ve been for the past three
hours. If I lied to you because of the embarrassment that telling the truth would
have caused me—say, for instance, I was supposed to be taking our kids to the
park but instead was playing video games—I suspect we’ll all have the intuition
that there is nothing here to undermine the impermissibility-grounding power of
the fact that my utterance was a lie. If, on the other hand, I lied to you so as to
not spoil the surprise birthday party that I’m throwing for you—say, for instance,
I spent the last three hours checking out potential venues—I suspect we’ll all have
the intuition that the impermissibility-grounding power of the fact that my utter-
ance was a lie is indeed undermined in this case.
Works Cited
Agar, Nicholas. ‘Why Is It Possible to Enhance Moral Status and Why Doing So Is
Wrong?’. Journal of Medical Ethics 39 (2013a): 67–74.
———. ‘Still Afraid of Needy Post-Persons’. Journal of Medical Ethics 39 (2013b):
81–3.
Albertsen, Andreas and Knight, Carl. ‘A Framework for Luck Egalitarianism in
Health and Health Care’. Journal of Medical Ethics 14 (2015): 165–9.
Alvarez, Allen Andrew A. ‘Threshold Considerations in Fair Allocation of Health
Resources: Justice Beyond Scarcity’. Bioethics 21 (2007): 426–38.
Anand, Sudhir. ‘The Concern for Equity in Health’. In Public Health, Ethics, and
Equity, edited by Sudhir Anand, Fabienne Peter and Amartya Sen, 15–20. New
York: Oxford University Press, 2004.
Arneson, Richard. ‘Equality and Equal Opportunity for Welfare’. Philosophical Stud-
ies 56 (1989): 77–93.
———. ‘What, If Anything, Renders All Humans Morally Equal?’. In Singer and His
Critics, edited by Dale Jamieson, 103–27. Oxford: Blackwell, 1999.
———. ‘Rawls Versus Utilitarianism in the Light of Political Liberalism’. In The
Idea of a Political Liberalism, edited by Victoria Davion and Clark Wolf, 231–52.
Lanham, MD: Rowman and Littlefield, 2000a.
———. ‘Luck Egalitarianism and Prioritarianism’. Ethics 110 (2000b): 339–49.
Audi, Robert. ‘Intuition and Its Place in Ethics’. Journal of the American Philosophi-
cal Association 1 (2015): 57–77.
Axelsen, David V. and Nielsen, Lasse. ‘Essentially Enough: Elements of a Plausi-
ble Account of Sufficientarianism’. In What Is Enough? Sufficiency, Justice, and
Health, edited by Carina Fourie and Annette Rid, 101–18. New York: Oxford
University Press, 2017.
Bader, Ralf. ‘Conditions, Modifiers, and Holism’. In Weighing Reasons, edited by
Errol Lord and Barry Maguire, 27–55. New York: Oxford University Press, 2016.
Baier, Kurt. The Moral Point of View. Ithaca, NY: Cornell University Press, 1958.
Baron, Marcia. ‘Kantian Ethics’. In Three Methods of Ethics, edited by Marcia W.
Baron, Philip Pettit and Michael Slote, 3–91. Malden, MA: Blackwell, 1997.
———. ‘The Ticking Bomb Hypothetical’. In Disputed Moral Issues, 3rd ed., edited
by Mark Timmons, 628–38. New York: Oxford University Press, 2013.
Barry, Brian. Theories of Justice. Hemel Hempstead: Harvester-Wheatsheaf, 1989.
Beauchamp, Tom L. and Childress, James F. Principles of Biomedical Ethics, 6th ed.
New York: Oxford University Press, 2009.
Works Cited 169
Bernstein, Mark H. On Moral Considerability. New York: Oxford University Press,
1998.
Bräanmark, Johan. ‘Moral Disunitarianism’. Philosophical Quarterly 66 (2016):
481–99.
Brink, David. ‘Moral Conflict and Its Structure’. Philosophical Review 103 (1994):
215–24.
Broad, C.D. Five Types of Ethical Theory. London: Kegan Paul, Trench, Trubner &
Co, 1930.
Brock, Dan. ‘Separate Spheres and Indirect Benefits’. Cost-Effectiveness and Resource
Allocation 1 (2003): 4.
Broome, John. ‘Good, Fairness and QALYs’. In Philosophy and Medical Welfare,
edited by JM. Bell and Susan Mendus, 57–73. Cambridge, MA: Cambridge Uni-
versity Press, 1988.
———. ‘Fairness Versus Doing the Most Good’. Hastings Center Report 2 (1994):
36–9.
———. ‘Fairness in the Rationing of Health Care’. In Market Capitalism and Moral
Values, edited by Samuel Brittan and Alan Hamlin, 79–87. n.p.: Edward Elgar, 1995.
———. ‘Normative Requirements’. Ratio 12 (1999): 398–419.
———. ‘Reasons’. In Reason and Value: Themes From the Moral Philosophy of
Joseph Raz, edited by R. Jay Wallace, Philip Pettit, Samuel Scheffler and Michael
Smith, 28–55. Oxford: Clarendon Press, 2004.
———. Rationality Through Reasoning. Malden, MA: Wiley Blackwell, 2013.
Buchanan, Allen. ‘Moral Status and Enhancement’. Philosophy & Public Affairs 37
(2009): 346–81.
Chappell, Richard. ‘Value Receptacles’. Noûs 49 (2015): 322–32.
Clune, Alan C. ‘Biomedical Testing on Animals: An Attempt at a “Rapprochement”
Between Utilitarianism and Theories of Inherent Value’. The Monist 79 (1996):
230–46.
Cohen, Andrew. ‘Contractarianism, Other-regarding Attitudes, and the Moral Stand-
ing of Nonhuman Animals’. Journal of Applied Philosophy 24 (2007): 188–201.
———. ‘Contractarianism and Moral Standing Inegalitarianism’. Dialogue 55
(2016): 639–58.
Cohen, Carl. ‘The Case for the Use of Animals in Biomedical Research’. New Eng-
land Journal of Medicine 315 (1986): 865–70.
Cohen, G.A. ‘On the Currency of Egalitarian Justice’. Ethics 99 (1989): 906–44.
———. Rescuing Justice and Equality. Cambridge, MA: Harvard University Press,
2008.
Conee, Earl. ‘Against Moral Dilemmas’. In Moral Dilemmas, edited by Christopher
W. Gowans, 239–49. New York: Oxford University Press, 1987.
Cornell, Nicolas. ‘Wrongs, Rights, and Third Parties’. Philosophy & Public Affairs
43 (2015): 109–43.
Crisp, Roger. ‘Particularizing Particularism’. In Moral Particularism, edited by Brad
Hooker and Margaret Little, 23–47. New York: Oxford University Press, 2000.
———. ‘Equality, Priority, and Compassion’. Ethics 113 (2003): 745–63.
Culyer, Anthony J. ‘Equity—Some Theory and Its Policy Implications’. Journal of
Medical Ethics 27 (2001): 275–83.
Culyer, Anthony J. and Wagstaff, Adam. ‘Equity and Equality in Health and Health
Care’. Journal of Health Economics 12 (1993): 431–57.
170 Works Cited
Dahl, Norman O. ‘Morality, Moral Dilemmas, and Moral Requirements’. In Moral
Dilemmas and Moral Theory, edited by H.E. Mason, 86–101. New York: Oxford
Dancy, Jonathan. Ethics Without Principles. Oxford: Clarendon Press, 2004.
Daniels, Norman. ‘Wide Reflective Equilibrium and Theory Acceptance in Ethics’.
The Journal of Philosophy 76 (1979): 256–82.
———. Just Health: Meeting Health Needs Fairly. Cambridge: Cambridge Univer-
sity Press, 2008.
Darwall, Stephen. The Second-Person Standpoint. Cambridge, MA: The Harvard
———. ‘But It Would Be Wrong’. Social Philosophy and Policy 27 (2010): 135–57.
DeGrazia, David. Taking Animals Seriously. Cambridge: Cambridge University Press,
1996.
———. Creation Ethics. New York: Oxford University Press, 2012.
Douglas, Tom. ‘Human Enhancement and Supra-Personal Moral Status’. Philosophi-
cal Studies 162 (2013): 473–97.
Driver, Julia. ‘Autonomy and the Asymmetry Problem for Moral Expertise’. Philo-
sophical Studies 128 (2006): 619–44.
Dunn, J. Interpreting Political Responsibility. Cambridge: Cambridge University
Press, 1990.
Dworkin, Ronald. ‘The Original Position’. In Reading Rawls, edited by Norman
Daniels, 16–53. Stanford, CA: Stanford University Press, 1989.
———. Sovereign Virtue. Cambridge, MA: Harvard University Press, 2000.
Feinberg, Joel. Harm to Others. New York: Oxford University Press, 1984.
———. ‘Wrongful Life and the Counterfactual Element in Harming’. Social Philoso-
phy & Policy 4 (1986): 145–78.
Feldman, Fred. Utilitarianism, Hedonism, and Desert. Cambridge: Cambridge Uni-
Fogal, Daniel. ‘Reasons, Reason, and Context’. In Weighing Reasons, edited by Errol
Lord and Barry Maguire, 74–103. New York: Oxford University Press, 2016.
Fourie, Carina. ‘Sufficiency of Capabilites, Social Equality, and Two-Tiered Health
Care Systems’. In What Is Enough? Sufficiency, Justice, and Health, edited by
Carina Fourie and Annette Rid, 185–204. New York: Oxford University Press,
2017.
Fourie, Carina and Rid, Annette. ‘Introduction’. In What Is Enough? Sufficiency,
Justice, and Health, edited by Carina Fourie and Annette Rid, 1–8. New York:
Oxford University Press, 2017.
Fox, Michael. ‘ “Animal Liberation”: A Critique’. Ethics 88 (1978): 106–18.
Frey, R.G. Interests and Rights: The Case Against Animals. New York: Oxford Uni-
———. ‘Animals’. In The Oxford Handbook of Practical Ethics, edited by Hugh
LaFollette, 161–87. Oxford: Oxford University Press, 2003.
———. ‘Utilitarianism and Animals’. In The Oxford Handbook of Animal Ethics,
edited by Tom L. Beauchamp and R.G. Frey, 172–97. New York: Oxford Univer-
sity Press, 2011.
Gaut, Berys. ‘Moral Pluralism’. Philosophical Papers XXIII (1994): 17–40.
Gert, Joshua. ‘A Functional Role Analysis of Reasons’. Philosophical Studies 124
(2005): 353–78.
Works Cited 171
Goodpaster, Kenneth. ‘On Being Morally Considerable’. Journal of Philosophy 75
(1978): 308–35.
Gosseries, Axel. ‘Intergenerational Justice, Sufficiency, and Health’. In What Is
Enough? Sufficiency, Justice, and Health, edited by Carina Fourie and Annette
Rid, 121–43. New York: Oxford University Press, 2017.
Greene, Joshua. ‘Beyond Point-and-Shoot Morality: Why Cognitive (Neuro)Science
Matters for Ethics’. Ethics 124 (2013): 695–726.
Hare, R.M. Moral Thinking. Oxford: Clarendon Press, 1981.
———. ‘Rights, Utility, and Universalization: Reply to J.L. Mackie’. In Utility and
Rights, edited by R.G. Frey, 106–20. Minneapolis: University of Minnesota Press,
1984.
Harman, Elizabeth. ‘Creation Ethics: The Moral Status of Early Fetuses and the Eth-
ics of Abortion’. Philosophy and Public Affairs 28 (1999): 310–24.
———. ‘The Potentiality Problem’. Philosophical Studies 114 (2003): 173–98.
———. ‘Can We Harm and Benefit in Creating?’. Philosophical Perspectives 18 (2004):
89–113.
———. ‘Harming as Causing Harm’. In Harming Future Persons: Ethics, Genetics,
and the Nonidentity Problem, edited by Melinda A. Roberts and David T. Wasser-
man, 137–54. Dordrecht: Springer, 2009.
———. ‘The Moral Significance of Animal Pain and Animal Death’. In The Oxford
Handbook of Animal Ethics, edited by Tom L. Beauchamp and R.G. Frey, 726–37.
New York: Oxford University Press, 2011.
Hart, H.L.A. The Concept of Law, 3rd ed. Oxford: Oxford University Press, 2012.
Haslett, D.W. ‘Murder and the Exception for Fair Competition’. Social Theory and
Practice 29 (2003): 631–54.
Hempel, Carl G. and Oppenheim, Paul. ‘Studies in the Logic of Explanation’. Phi-
losophy of Science 15 (1948): 135–75.
Herman, Barbara. ‘Kantian Commitments’. Studi Kantiani XXIX (2016): 131–43.
Hershovitz, Scott. ‘The End of Jurisprudence’. The Yale Law Journal 124 (2015):
1160–204.
Hill, Thomas E. ‘Moral Dilemmas, Gaps, and Residues’. In Moral Dilemmas and
Moral Theory, edited by H.E. Mason, 167–98. New York: Oxford University
Press, 1996.
Hinman, Lawrence M. Ethics: A Pluralistic Approach to Moral Theory. Belmont,
CA: Thomson Wadsworth, 2008.
Hirose, Iwao. ‘Axiological Sufficientarianism’. In What Is Enough? Sufficiency, Jus-
tice, and Health, edited by Carina Fourie and Annette Rid, 51–68. New York:
Holtug, Nils. ‘Equality for Animals’. In New Waves in Applied Ethics, edited by Jes-
per Ryberg, Thomas S. Petersen and Claerk Wolf, 1–24. Basingstoke, Hampshire:
Palgrave Macmillan, 2007.
Hooker, Brad. Ideal Code, Real World. New York: Oxford University Press, 2000.
———. ‘Intuitions and Moral Theorizing’. In Ethical Intuitionism: Re-evaluations,
edited by Philip Stratton-Lake, 161–83. Oxford: Clarendon Press, 2002.
Huemer, Michael. Ethical Intuitionism. Basingstoke, Hampshire: Palgrave Macmil-
lan, 2005.
———. ‘When Is Parsimony a Virtue?’. The Philosophical Quarterly 59 (2009): 216–36.
Hursthouse, Rosalind. On Virtue Ethics. New York: Oxford University Press, 1999.
172 Works Cited
Huseby, Robert. ‘Sufficiency, Priority, and Aggregation’. In What Is Enough? Suffi-
ciency, Justice, and Health, edited by Carina Fourie and Annette Rid, 69–84. New
Jacobs, Lesley A. Pursuing Equal Opportunities. Cambridge: Cambridge University
Press, 2004.
Kagan, Shelly. Normative Ethics. Boulder, CO: Westview Press, 1988.
———. The Limits of Morality. New York: Oxford University Press, 1989.
Kamm, Frances. ‘To Whom?’. Hasings Center Report 24 (1994): 29–32.
———. Morality, Mortality. Vol. 1. Oxford: Oxford University Press, 1996.
———. Intricate Ethics. New York: Oxford University Press, 2007.
Kant, Immanuel. The Metaphysics of Morals, translated by Mary Gregor. Cam-
bridge: Cambridge University Press, 1996.
———. Groundwork of the Metaphysics of Morals, translated by Mary Gregor and
edited by Jens Timmermann. Cambridge: Cambridge University Press, 2011.
Kearns, Stephen and Star, Daniel. ‘Reasons as Evidence’. In Oxford Studies in Meta-
ethics. Vol. 4, edited by Russ Shafer-Landau, 215–42. New York: Oxford Univer-
sity Press, 2009.
Korsgaard, Christine. ‘Kant’s Formula of Humanity’. In Creating the Kingdom of
Ends, edited by Christine Korsgaard, 106–32. Cambridge: Cambridge University
Press, 1996.
———. ‘Fellow Creatures: Kantian Ethics and Our Duties to Animals’. In The Tan-
ner Lectures on Human Values. Vol. 25, edited by Grethe B. Peterson, 77–110. Salt
Lake City: The University of Utah Press, 2005.
———. Self-Constitution: Agency, Identity, and Integrity. New York: Oxford Uni-
———. ‘Interacting With Animals: A Kantian Account’. In The Oxford Handbook
of Animal Ethics, edited by Tom L. Beauchamp and R.G. Frey, 91–118. New York:
———. ‘A Kantian Case for Animal Rights’. In Animal Law: Tier und Recht, edited
by Julia Hänni, Margot Michel and Daniela Kühne, 3–27. Zurich: Dike Verlag,
2012.
———. ‘Kantian Ethics, Animals, and the Law’. Oxford Journal of Legal Studies 33
(2013): 629–48.
Kymlicka, Will. Liberalism, Community and Culture. New York: Oxford University
Press, 1989.
———. Contemporary Political Philosophy: An Introduction. Oxford: Oxford Uni-
Ladyman, James. ‘Science, Metaphysics, and Method’. Philosophical Studies 160
(2012): 31–51.
LaFollette, Hugh. ‘Animal Experimentation in Biomedical Research’. In The Oxford
Handbook of Animal Ethics, edited by Tom L. Beauchamp and R.G. Frey, 796–
825. New York: Oxford University Press, 2011.
Leary, Stephanie. ‘Non-naturalism and Normative Necessities’. In Oxford Studies in
Metaethics. Vol. 12, edited by Russ Shafer-Landau, 76–105. New York: Oxford
Liberman, Alida and Schroeder, Mark. ‘Commitment: Worth the Weight’. In Weigh-
ing Reasons, edited by Errol Lord and Barry Maguire, 104–20. New York: Oxford
Works Cited 173
Lichtenberg, Judith. ‘The Right, the All Right, and the Good’. Yale Law Journal 92
(1983): 544–63.
List, Christian and Valentini, Laura. ‘The Methodology of Political Theory’. In The
Oxford Handbook of Philosophical Methodology, edited by Herman Cappelen,
Tamar Szabó Gendler and John Hawthorne, 525–53. Oxford: Oxford University
Press, 2016.
Little, Margaret Olivia. ‘On Knowing the “Why”: Particularism and Moral Theory’.
Hastings Center Report 31 (2001): 32–40.
Maguire, Barry. ‘Grounding the Autonomy of Ethics’. In Oxford Studies in Metaeth-
ics. Vol. 10, edited by Russ Shafer-Landau, 188–215. New York: Oxford Univer-
sity Press, 2015.
Manne, Kate. ‘Democratizing Humeanism’. In Weighing Reasons, edited by Errol
Lord and Barry Maguire, 123–40. New York: Oxford University Press, 2016.
Marmot, Michael. ‘Social Causes of Social Inequalities in Health’. In Public Health,
Ethics, and Equity, edited by Sudhir Anand, Fabienne Peter and Amartya Sen,
37–61. New York: Oxford University Press, 2004.
McConnell, Terrance. ‘Moral Residue and Dilemmas’. In Moral Dilemmas and Moral
Theory, edited by H.E. Mason, 36–47. New York: Oxford University Press, 1996.
McDermott, Daniel. ‘Analytic Political Philosophy’. In Political Theory: Methods
and Approaches, edited by David Leopold and Marc Stears, 11–28. New York:
McGrath, Sarah. ‘Moral Disagreement and Moral Expertise’. In Oxford Studies in
Metaethics. Vol. 3, edited by Russ Shafer-Landau, 87–107. New York: Oxford
———. ‘Skepticism About Moral Expertise as a Puzzle for Moral Realism’. Journal
of Philosophy 108 (2011): 111–37.
McKeever, Sean and Ridge, Michael. Principled Ethics. New York: Oxford Univer-
sity Press, 2006.
———. ‘Elusive Reasons’. In Oxford Studies in Metaethics. Vol. 7, edited by Russ
Shafer-Landau, 110–37. New York: Oxford University Press, 2012.
McKerlie, Dennis. ‘Equality and Time’. Ethics 99 (1989): 475–91.
McKie, John, Kuhse, Helga, Richardson, Jeff and Singer, Peter. ‘Double Jeopardy, the
Equal Value of Lives and the Veil of Ignorance: A Rejoinder to Harris’. Journal of
Medical Ethics 22 (1996): 204–8.
McMahan, Jeff. ‘Cognitive Disability, Misfortune, and Justice’. Philosophy and Pub-
lic Affairs 25 (1996): 3–34.
———. ‘Moral Intuitions’. In The Blackwell Guide to Ethical Theory, edited by
Hugh Lafollette, 92–110. Malden, MA: Blackwell Publishers, 2000.
———. The Ethics of Killing: Problems at the Margins of Life. New York: Oxford
———. ‘Animals’. In A Companion to Applied Ethics, edited by R.G. Frey and
Christopher Heath Wellman, 525–36. Malden, MA: Blackwell Publishing, 2005.
———. ‘Radical Cognitive Limitation’. In Disability and Disadvantage, edited by Kim-
berley Brownlee and Adam Cureton, 240–59. Oxford: Oxford University Press, 2009.
———. ‘Infanticide and Moral Consistency’. Journal of Medical Ethics 39 (2013a):
273–80.
———. ‘Causing People to Exist and Saving People’s Lives’. The Journal of Ethics
17 (2013b): 5–35.
174 Works Cited
McNaughton, David. ‘An Unconnected Heap of Duties?’. In Ethical Intuitionism,
edited by Philip Stratton-Lake, 76–91. Oxford: Oxford University Press, 2002.
Nagel, Thomas. ‘Preface’. In Mortal Questions, edited by Thomas Nagel, ix–xiii.
Cambridge: Cambridge University Press, 1979.
———. The View From Nowhere. New York: Oxford University Press, 1986.
———. ‘Equality’. In The Ideal of Equality, edited by Matthew Clayton and Andrew
Williams, 60–80. Basingstoke, Hampshire: Palgrave Macmillan, 2002.
Nelson, Mark. ‘What the Utilitarian Cannot Think’. Ethical Theory and Moral Prac-
tice 18 (2015): 717–29.
Nussbaum, Martha. Frontiers of Justice. Cambridge, MA: The Harvard University
Press, 2006.
———. ‘The Capabilities Approach and Animal Entitlements’. In The Oxford Hand-
book of Animal Ethics, edited by Tom L. Beauchamp and R.G. Frey, 228–51. New
Oberdiek, John. ‘Lost in Moral Space: On the Infringing/Violating Distinction and
Its Place in the Theory of Rights’. Law and Philosophy 23 (2004): 325–436.
O’Neill, Onora. ‘Necessary Anthropocentrism and Contingent Speciesism’. Proceed-
ings of the Aristotelian Society Supplementary Edition 72 (1998): 211–28.
———. ‘Kant and the Social Contract Tradition’. In Kant’s Political Theory: Inter-
pretations and Applications, edited by Elisabeth Ellis, 25–41. University Park, PA:
Pennsylvania State University Press, 2012.
Owens, David. Shaping the Normative Landscape. New York: Oxford University
Press, 2012.
Parfit, Derek. Reasons and Persons. New York: Oxford University Press, 1987.
———. ‘Equality or Priority?’. In The Ideal of Equality, edited by Matthew Clay-
ton and Andrew Williams, 81–125. Basingstoke, Hampshire: Palgrave Macmillan,
2002.
———. On What Matters. Vol. 1. New York: Oxford University Press, 2011.
Persson, Ingmar. ‘A Defence of Extreme Egalitarianism’. In Egalitarianism: New
Essays on the Nature and Value of Equality, edited by Nils Holtug and Kasper
Lippert-Rasmussen, 83–97. Oxford: Clarendon Press, 2007.
Peter, Fabienne. ‘Health Equity and Social Justice’. In Public Health, Ethics, and
Equity, edited by Sudhir Anand, Fabienne Peter and Amartya Sen, 93–106. New
Pogge, Thomas. ‘Is Kant’s Rechtslehre Comprehensive?’. The Southern Journal of
Philosophy XXXVI, Supplement (1997): 161–87.
———. ‘Severe Poverty as a Violation of Negative Duties’. Ethics and International
Affairs 19 (2005): 55–83.
Portmore, Douglas. Commonsense Consequentialism. New York: Oxford University
Press, 2011.
Pound, Pandora and Bracken, Michael. ‘Is Animal Research Sufficiently Evidence
Based to Be a Cornerstone of Biomedical Research?’. British Medical Journal 348
(2014): g3387.
Powers, Madison and Faden, Ruth. Social Justice: The Moral Foundations of Public
Health and Health Policy. New York: Oxford University Press, 2006.
Prichard, H.A. ‘Does Moral Philosophy Rest on a Mistake?’. Mind 21 (1912): 21–37.
Psillos, Stathos. Scientific Realism: How Science Tracks Truth. Oxford: Oxford Uni-
Rachels, James. The Elements of Moral Philosophy. Boston: McGraw-Hill, 1999.
Works Cited 175
———. ‘Drawing Lines’. In Animal Rights: Current Debates and New Directions,
edited by Cass R. Sunstein and Martha Nussbaum, 162–74. Oxford: Oxford Uni-
Rakowski, Erik. ‘The Aggregation Problem’. Hastings Center Report 24 (1994): 33–6.
Ram-Tiktin, Efrat. ‘The Right to Health Care as a Right to Basic Human Functional
Capabilities’. Ethical Theory and Moral Practice 15 (2012): 337–51.
———. ‘Basic Human Functional Capabilities as the Currency of Sufficientarian
Distribution in Health Care’. In What Is Enough? Sufficiency, Justice, and Health,
edited by Carina Fourie and Annette Rid, 144–63. New York: Oxford University
Press, 2017.
Rawls, John. A Theory of Justice, rev. ed. Cambridge, MA: The Belknap Press, 1999a.
———. The Law of Peoples. Cambridge, MA: The Harvard University Press, 1999b.
Raz, Joseph. Practical Reason and Norms. Princeton, NJ: Princeton University Press,
1990.
———. ‘The Truth in Particularism’. In Moral Particularism, edited by Brad Hooker
and Margaret Little, 48–78. New York: Oxford University Press, 2000.
Regan, Tom. ‘The Nature and Possibility of an Environmental Ethic’. Environmental
Ethics 3 (1981): 173–98.
———. The Case for Animal Rights. Berkeley, CA: The University of California
Press, 1983.
Richardson, Henry S. ‘Specifying Norms as a Way to Resolve Concrete Ethical Prob-
lems’. Philosophy & Public Affairs 19 (1990): 279–310.
Roberts, M.A. ‘A New Way of Doing the Best That We Can: Person-Based Conse-
quentialism and the Equality Problem’. Ethics 112 (2002): 315–50.
Robeyns, Ingrid. ‘Ideal Theory in Theory and Practice’. Social Theory and Practice
34 (2008): 341–62.
Rollin, Bernard E. Animal Rights & Human Morality. Amherst, NY: Prometheus
Books, 2006.
Ross William David. Foundations of Ethics. Oxford: Clarendon Press, 1939.
———. The Right and the Good. Indianapolis: Hackett, 1988.
Rowlands, Mark. Animal Rights: Moral Theory and Practice. Basingstoke, Hamp-
shire: Palgrave Macmillan, 2009.
Ruger, Jennifer Prah. Health and Social Justice. New York: Oxford University Press,
2010.
Sachs, Benjamin. ‘Does Rationality Ever Require Immorality?’ PhD diss., University
of Wisconsin-Madison, 2006.
———. ‘Reasons and Requirements’. Ethical Theory and Moral Practice 11 (2008a):
73–83.
———. ‘The Liberty Principle and Universal Health Care’. The Kennedy Institute of
Ethics Journal 18 (2008b): 149–72.
———. ‘Consequentialism’s Double-Edged Sword’. Utilitas 22 (2010a): 258–71.
———. ‘The Exceptional Ethics of the Investigator-Subject Relationship’. The Jour-
nal of Medicine & Philosophy 35 (2010b): 64–80.
———. ‘Going From Principles to Rules in Research Ethics’. Bioethics 25 (2011): 9–20.
———. ‘Reasons Consequentialism’. Journal of Moral Philosophy 10 (2013): 671–82.
———. ‘Contractarianism as a Political Morality’. Proceedings of the Aristotelian
Society CXVI (2016): 49–67.
Sapontzis, Steve. ‘Animal Rights and Biomedical Research’. The Journal of Value
Inquiry 26 (1992): 73–86.
176 Works Cited
Savulescu, Julian. ‘Genetically Modified Animals: Should There Be Limits to Engi-
neering the Animal Kingdom?’. In The Oxford Handbook of Animal Ethics,
edited by Tom L. Beauchamp and R.G. Frey, 641–70. New York: Oxford Univer-
sity Press, 2011.
Scanlon, T.M. What We Owe to Each Other. Cambridge, MA: The Belknap Press, 1998.
———. Being Realistic About Reasons. New York: Oxford University Press, 2014.
Scheffler, Samuel. The Rejection of Consequentialism. New York: Oxford University
Press, 1982.
Schroeder, Mark. Slaves of the Passions. New York: Oxford University Press, 2007.
Segall, Shlomi. Health, Luck, and Justice. Princeton, NJ: Princeton University Press,
2010.
Sen, Amartya. Inequality Reexamined. Cambridge, MA: Harvard University Press,
1992.
———. ‘Why Health Equity?’. In Public Health, Ethics, and Equity, edited by Sudhir
Anand, Fabienne Peter and Amartya Sen, 21–33. New York: Oxford University
Press, 2004.
———. ‘What Do We Want From a Theory of Justice?’. The Journal of Philosophy
CIII (2006): 215–38.
Shafer-Landau, Russ. Moral Realism. New York: Oxford University Press, 2003.
Shaw, William H. ‘Contemporary Criticisms of Utilitarianism: A Response’. In The
Blackwell Guide to Mill’s Utilitarianism, edited by Henry R. West, 201–16. Mal-
den, MA: Blackwell, 2006.
Shields, Liam. ‘Some Questions (and Answers) for Sufficientarians’. In What Is
Enough? Sufficiency, Justice, and Health, edited by Carina Fourie and Annette
Rid, 85–100. New York: Oxford University Press, 2017.
Shiffrin, Seana Valentine. ‘Wrongful Life, Procreative Responsibility, and the Signifi-
cance of Harm’. Legal Theory 5 (1999): 117–48.
Shpall, Sam. ‘Moral and Rational Commitment’. Philosophy and Phenomenological
Research 88 (2014): 146–72.
Sidgwick, Henry. The Methods of Ethics. Indianapolis: Hackett, 1981.
Simmons, A. John. ‘Ideal and Nonideal Theory’. Philosophy & Public Affairs 38
(2010): 5–36.
Singer, Peter. ‘Famine, Affluence, and Morality’. Philosophy and Public Affairs 1 (1972):
229–43.
———. Practical Ethics. Cambridge: Cambridge University Press, 1993.
———. Animal Liberation. London: Random House, 1995.
———. ‘Ethics and Intuitions’. Journal of Ethics 9 (2005): 331–52.
Sinnott-Armstrong, Walter. Moral Dilemmas. Oxford: Basil Blackwell, 1988.
Slicer, Deborah. ‘My Daughter or My Dog? A Feminist Assessment of the Animal
Research Issue’. Hypatia 6 (1991): 108–24.
Sliwa, Paulina. ‘In Defense of Moral Testimony’. Philosophical Studies 158 (2012):
175–95.
———. ‘On Knowing What’s Right and Being Responsible for It’. In Responsibility—
The Epistemic Condition, edited by Philip Robichaud and Jan Willem Weiland,
127–45. New York: Oxford University Press, 2017.
Smith, Michael. The Moral Problem. Malden, MA: Blackwell Press, 1994.
———. ‘Normative Reasons and Full Rationality: Reply to Swanton’. Analysis 156
(1996): 160–8.
Works Cited 177
Southwood, Nicholas. ‘Deliberative Models of Contractualism’. Politics, Philoso-
phy & Economics 7 (2008): 183–208.
Sreenivasan, Gopal. ‘Health and Justice in Our Non-Ideal World’. Politics, Philoso-
phy & Economics 6 (2007): 218–36.
Steinbock, Bonnie. Life Before Birth: The Moral and Legal Status of Embryos and
Fetuses. New York: Oxford University Press, 1992.
Stewart, Noel. Ethics: An Introduction to Moral Philosophy. Cambridge: Polity
Press, 2009.
Strandberg, Caj. ‘Particularism and Supervenience’. In Oxford Studies in Metaeth-
ics. Vol. 3, edited by Russ Shafer-Landau, 129–58. New York: Oxford University
Press, 2008.
Stratton-Lake, Philip. ‘Scanlon’s Contractualism and the Redundancy Objection’.
Analysis 63 (2003): 70–6.
———. ‘Recalcitrant Pluralism’. Ratio (New Series) XXIV (2011): 364–83.
Strawson, P.F. ‘Freedom and Resentment’. Proceedings of the British Academy 48
(1962): 1–25.
Street, Sharon. ‘A Darwinian Dilemma for Realist Theories of Value’. Philosophical
Studies 127 (2006): 109–66.
Sumner, L.W. Abortion and Moral Theory. Princeton, NJ: Princeton University Press,
1981.
Tännsjö, Torbjörn. From Reasons to Norms: On the Basic Questions in Ethics. Dor-
drecht: Springer, 2010.
Temkin, Larry. ‘Justice, Equality, Fairness, Desert, Rights, Free Will, Responsibility,
and Luck’. In Responsibility and Distributive Justice, edited by Carl Knight and
Zofia Stemplowska, 51–76. New York: Oxford University Press, 2011.
———. Rethinking the Good. New York: Oxford University Press, 2012.
Thompson, Michael. ‘What Is It to Wrong Someone: A Puzzle About Justice’. In Rea-
son and Value: Themes From the Moral Philosophy of Joseph Raz, edited by R.
Jay Wallace, Philip Pettit, Samuel Scheffler and Michael Smith, 333–84. Oxford:
Clarendon Press, 2004.
Thomson, Judith Jarvis. Normativity. Chicago: Open Court Publishing, 2008.
Timmermann, Jens. ‘When the Tail Wags the Dog: Animal Welfare and Indirect Duty
in Kantian Ethics’. Kantian Review 10 (2005): 128–49.
Tulodziecki, Dana. ‘Underdetermination, Methodological Practices, and Realism’.
Synthese 190 (2013): 3731–50.
Urmson, J.O. ‘A Defence of Intuitionism’. Proceedings of the Aristotelian Society 75
(1974): 111–19.
Vallentyne, Peter. ‘Of Mice and Men: Equality and Animals’. Journal of Ethics 9
(2005): 403–33.
VanDeVeer, Donald. ‘Of Beasts, Persons, and the Original Position’. The Monist 62
(1979): 368–77.
van Fraassen, Bas. The Scientific Image. Oxford: Oxford University Press, 1980.
———. Laws and Symmetry. Oxford: Oxford University Press, 1989.
Väyrynen, Pekka. ‘Moral Generalism: Enjoy in Moderation’. Ethics 116 (2006):
707–41.
———. ‘A Theory of Hedged Moral Principles’. In Oxford Studies in Metaethics.
Vol. 1, edited by Russ Shafer-Landau, 91–132. New York: Oxford University
Press, 2009.
178 Works Cited
———. ‘Grounding and Normative Explanation’. Proceedings of the Aristotelian
Society Supplementary Edition 87 (2013): 155–78.
Wallace, R. Jay. ‘Reasons, Relations, and Commands: Reflections on Darwall’. Eth-
ics 118 (2007): 24–36.
———. ‘The Deontic Structure of Morality’. In Thinking About Reasons: Themes
From the Philosophy of Jonathan Dancy, edited by David Bakhurst, Brad Hooker
and Margaret Olivia Little, 137–67. Oxford: Oxford University Press, 2013.
Warren, Mary Ann. Moral Status. New York: Oxford University Press, 2000.
White, Roger. ‘Why Favour Simplicity?’. Analysis 65 (2005): 205–10.
Williams, Bernard. ‘A Critique of Utilitarianism’. In Utilitarianism: For & Against,
edited by J.J.C. Smart and Bernard Williams, 77–150. Cambridge: Cambridge
———. ‘Persons, Character, and Morality’. In Moral Luck, edited by Bernard Wil-
liams, 1–19. Cambridge: Cambridge University Press, 1981.
Wringe, Bill. ‘International Justice, Intervention, and the Prevention of Evil’. In Evil:
An Interdisciplinary Approach, edited by Margaret Sönser Breen, 75–94. Amster-
dam: Rodopi, 2003.
Zangwill, Nick. ‘Cordelia’s Bond and Indirect Consequentialism’. In Oxford Stud-
ies in Normative Ethics. Vol. 1, edited by Mark Timmons, 143–65. New York:
Index
act-consequentialism 1, 2, 62, 160; to holistic explanation 30; relevance

and animal research 135; and rescue criterion for 19; taking the form of
129; as a kind of moral monism token-grounding claims 25; transitivity
32; avoids scope questions 115; of 37, 47, 65, 93n1; see also exceptions
brain process behind its judgments to moral generalizations, holistic
16n17; counterintuitiveness of its explanation, moral generalizations,
token-grounding claims 43; how to non-Rossian pluralism
represent it 33–5, 164–7; its moral autonomy 46–7
generalization 2; see also animals,
directedness objection, interests Baier, Kurt 62
agent-relative value 1 Baron, Marcia 59n24, 147
Albertsen 151 basic evaluative tendencies 7, 8
animal nature 48, 50–2 benefit-conferring: and animal research
animal research 3, 93, 143; alternatives 135–8, 140–1; and harm-preventing
to 139–41; impermissibility of 134–8, 133; obligatoriness of 133–4, 137, 139,
139, 159; see also act-consequentialism, 145; see also enablers and disablers,
benefit-conferring; genuine moral health care resource distribution
dilemmas, harm, non-consequentialism Bentham 42, 62, 115
animality see animal nature Benthamite utilitarianism 5, 25, 30n2
animals 93, 119; act-consequentialism’s blameworthiness 77–8
approach to 115; capacity for welfare bottom-up moral theorizing see moral
100; contractarianism’s approach to theorizing
53–7, 84; contractualism’s approach Bräanmark 114n15
to 53–4, 57, 84; Kant’s approach to Brink 85
46–7; Korsgaard’s approach to 47–52; Broad, C.D. 9
moral pluralism’s approach to 116–17; Brock, Dan 143, 145
non-Rossian pluralism’s approach to Broome, John 70, 85, 143, 145
160; Rowlands’s contractarianisms’s
approach to 54–6; welfare of 112n6; capabilities: as distribuend 107
welfare-potential of 100; welfaristic capacity for welfare: defined in terms of
obligations to 100–1, 103; see also psychological capacities 99; necessary
animal nature, animal research, duties, differences in 102–4, 112n6; of
harm, interests humans and non-humans 100; see also
applied ethics 2, 13, 17, 143 animals
Arneson 34 case-specific intuitions see verdictive
atomistic explanations: as factive 26; intuitions
asymmetry criterion for 19; defined 18; Categorical Imperative 33, 45–7, 50, 51,
generalizing about 18, 21, 22, 24, 25; 59n24, 115, 164
having facts as relata 19, 25; is-ought causal explanation 30n4, 47
problem for 20, 120; not contained claims 49, 53
in ethical theories 18; not reducible Clune 135
180 Index
Cohen, G.A. 111 Darwinian Dilemma argument 7
coherence 12 DeGrazia, David 118–19, 135
compensation 127; in medical research deliberation: Deliberative Rationalism
140–1 87–90, 94n7, 160; moral 11, 42–3;
confirmation 11 moral deliberation given non-Rossian
consent 51, 52, 126, 127, 128; in medical pluralism 90–1; phenomenology of 91;
research 140–1 point of deliberating about one’s moral
consequentialism 3, 34, 62; and harm reasons 91–2; practical 92; see also
37–8, 128; how to represent it 33–4; deliberative obligations, enablers and
impersonal variant 41–2; person- disablers, moral generalizations, non-
affecting variant 37, 42–3; possible normative facts, verdictive intuitions
derivation from contractualism deliberative obligations 122–3
39; rejection of 44; relation to Deliberative Rationalism see deliberation
Rossian pluralism 44–5; see also act- deontic logic 85
consequentialism, Consequentialism’s deontic reasons 36
Type-Grounding Claim, directedness desires 79n5, 94n7
objection; goodness, Rossian dignity 46, 116
pluralism, rule-consequentialism directedness objection: against act-
Consequentialism’s Type-Grounding consequentialism 37; against
Claim: and counterintuitiveness of Consequentialism’s Type-Grounding
its token-grounding claims 43–4; and Claim 36–43, 44; against
picking out a wrong-sufferer/victim contractarianism 56; against Rossian
37–43; and the reactive attitudes pluralism 64–9, 71; inapplicable to
38–40; derivation of 35; separability non-Rossian pluralism
from a consequentialist criterion of directedness: defined in terms of wronging
permissibility 39; statement of 35; 36; not a feature of overall oughts 71;
see also directedness objection, harm see also reasons
constitutive explanation 30n4 disablers see enablers and disablers
context-sensitivity 23–4, 130 disconfirmation 11
contractarianism 3, 53, 114n15, 160; and distribuend 105, 106–7, 149; see also
human psychological normality 110; capabilities, primary goods, resources,
and scope questions 115; and the value welfare
of abiding by moral requirements distribution 87, 93, 96, 145, 149–51,
76; as a basis for a distributive 154–5; auxiliary principles of 153–4;
generalization 110–11, 146; how to comparative and non-comparative
represent it 33–4, 164–7; see also principles of 152; why morally
animals, directedness objection, important 97; see also goods, health
Hobbesian contractarianism, Rawlsian care resource distribution, non-Rossian
contractarianism, Rowlands’s pluralism, welfaristic distributive
contractarianism generalizations
contractualism 3, 53–4, 76, 114n15; and distributive fortunism: defined 104;
human psychological normality 110; as problem for 105–6, 107; see also
a basis for a distributive generalization distributive welfarism, Sentient
110–13, 146; how to represent Individuals Scope
it 33–4, 164–7; see also animals, distributive welfarism 107–9; abandoning
consequentialism, moral monism, rule- in favor of distributive fortunism
consequentialism 104–5; arguing for derivatively 110;
contrastivity of ‘why’ questions see arguments for abandoning 106–7;
context-sensitivity as ground for distributive fortunism
Cornell, Nicholas 36 106; defined 98–9; problem for when
Crisp 101 assuming no necessary differences in
welfare-potential 102–3; problem for
Dancy 21, 22, 24, 63 with Equal Psychological Sophistication
Darwall, Stephen 36, 39, 56, 70, 74, Scope 102, 103; problem for with the
81n23 Psychologically Normal Humans Scope
Index 181
101–2, 103, 109–10; problem for explanatory intuitions 3, 4, 6, 13, 14, 28,
with the Sentient Humans Scope 101, 43, 90, 160, 162; in science 12–13;
103; problem for with the Sentient my use of 28; reliability of 7–8;
Individuals Scope 99–101, 102, 103; token-grounding and type-grounding
relationship to welfaristic distributive variants distinguished 26; usefulness
generalizations 98–9 of in choosing between competing
double counting 75 grounding claims 26
Driver 29 explanatory judgments see explanatory
duties 1, 46, 49, 55, 69, 108; to animals beliefs
47, 49–52; in Kantianism 75; in Rossian explanatory power 11–12
pluralism 62; pro tanto duties 1, 30n2,
66–7, 69; the purported existence and fallacy of division 150
nature of pro tanto duties 68–9, 86; fecundity 12
see also moral reasons, moral residue Feinberg 127
duty of rescue see rescuing fortune: defined 104; not a final good
Dworkin, Ronald 107 105; see also distributive fortunism,
psychologically limited/disabled
egalitarianism 98, 111, 112n6, 149–50; humans
criticism of egalitarian version of Frey 100
distributive welfarism 100
elegance 12 genuine moral dilemmas 92; and moral
enablers and disablers 44, 117; and requirements 86; as objection to non-
benefit-conferring 133–4, 136–7; and Rossian pluralism 85–7, 159–60,
context-sensitivity 24; defined 22; and 162n2; in animal research 137; in
harm 126, 128–9; in medical research health care resource distribution 148,
140–1; intuitions about 90–1, 137, 154; see also moral requirements,
166; role of in moral deliberation 90; moral residue, Rossian pluralism
role of in moral generalizations 22, Gert, Joshua 70, 89
25; role of in non-Rossian pluralism goodness: role of in consequentialism
84; whether they can be explained 2, 28n17, 32, 33, 35, 37–8, 43–5, 57,
165–7; see also moral generalizations, 58n16, 135, 164–7
moral residue goods: distribution of 97; final vs.
enablers see enablers and disablers instrumental 105; personal vs.
end-in-itself 48, 50 impersonal 37, 41, 42–3, 105;
Equal Psychological Sophistication Scope see also fortune
see distributive welfarism Gosseries 151
evolutionary skepticism argument 7, 162 Greene, Joshua 7, 161–2
exceptions to moral generalizations:
problem of 21–3 Hare, R.M. 70
expensive tastes 107 harm 13, 18, 26, 75, 76, 117; comparative
explanation 2; authoritativeness of in the and non-comparative accounts of
scientific case 11; how different from 131–2; in animal research 135; not
prediction in the moral case 10–11, 12; always wrong 126–9; obligatoriness
how different from prediction in the of preventing 130–4; relevance to
scientific case 10–11; see also atomistic Consequentialism’s Type-Grounding
explanation, causal explanations, Claim 40–2; severity of 130–1, 134; to
enabling and disabling, explanatory animals 47; to self 127, 134; see also
beliefs, explanatory claims, benefit-conferring, consequentialism,
explanatory intuitions, explanatory enablers and disablers, judgment-
power, holistic explanation, moral sensitive harm/benefit, non-normative
meta-explanation, nomic explanation facts, rule-consequentialism, scope,
explanatory beliefs 9 token-grounding claims, type-
explanatory claims 6; relation to grounding claims
verdictive claims 8; representing Harman, Elizabeth 118
moral generalizations as 2, 5, 8 Haslett 127
182 Index
health care resource distribution 3, 93, version of 47–52, 56, 59n24; see also
111, 143–55, 158–9; and benefit- duties, Korsgaard, moral monism,
conferring 148; and distributive scope, type-grounding claims
principles 149–54; separate spheres/ Kearns 94n7
isolationist approach to 151–4; Knight 151
see also genuine moral dilemmas, knowledge: gaining about what one
sufficientarianism morally ought to do 5, 8, 10, 15n10;
Hempel 10 see also scientific knowledge
Herman 59n24 Korsgaard, Christine 47–52, 56, 59n24,
Hershovitz 72 115; see also animals, Kantianism,
Hinman 34 wronging
Hobbesian contractarianism 56
holistic explanation 63; criteria for good LaFollette 135
27–8; defined 18; not reducible to law 49, 79n13; and legal oughts 72–3;
atomistic explanation 29n1; type- and legal reasons 72–3; criminal law
grounding claims as providing 25 49; standing in 49–50; see also Legal
Holtug 101 Model of Moral Normativity
Hooker 62 Legal Model of Moral Normativity
human psychological normality: a 72–8, 92; as a part of non-Rossian
threshold 109; moral importance pluralism 84–5; how it relates moral
of 109–10, 112n5; see also requirements to moral reasons 74–6
contractarianism, contractualism, Liberman 66, 68
distributive welfarism Life saving see rescuing
humanity 45, 47–8, 50–2, 54; respect Little 21
for 46 luck egalitarianism 107
Humeanism 94n7
Hurtshouse, Rosalind 59n24 Maguire 20
Huseby 152 Manne 94n7
Marmot 151
institutions 138–9 maximin 152
integrity 1 McElwee, Brian 79n5
interests 135; and act-consequentialism McGrath 29
42; of animals 53 McKeever 71
intuitionism 62 McMahan, Jeff 98, 100, 104–6, 113n7,
intuitions: phenomenology of 161; 119, 135
see also enablers and disablers, McNaughton 91
explanatory intuitions, verdictive Medical research 139–41; humans as
intuitions subjects of 140–1; see also animal
intuitive induction 9 research, enablers and disablers
mere means 47–8, 50–2
judgment-sensitive harm/benefit 41–2, meta-ethics 61, 62, 63, 69
65–6 metaphysical grounding 20
method of reflective equilibrium 2, 3, 5,
Kagan, Shelly 1, 62 6–7, 13–14; collective vs. individualistic
Kamm 122, 143, 145 use of 28–9; my use of 28–9; see also
Kant 1, 45, 48, 53, 62, 75; see also verdictive beliefs
Kantianism Mill 62
Kantianism 1, 2, 3, 62, 160; affinity with Moore 62
virtue ethics 59n24; and cruelty to moral anti-realism 30n5
animals 84; and scope questions 115; moral arbitrariness 54, 98, 102, 103, 109
how to represent it 33–4, 45–6, 52, moral community 38–40, 116; see also
164–7; its moral generalization 2; its moral status
self-regardingness 50, 59n24; its token- moral conditionals/categorical claims 29;
grounding claims 47, 49; Korsgaard’s arriving at judgments about 14–15n5;
Index 183
formulating moral generalizations as moral reasons 1, 2, 19, 61, 72, 83, 93,
8–10, 13, 24; problem of exceptions 148; and invariable relevance 79n13;
to 21 cheapness of 64, 67–8, 78–9n5;
moral considerability see moral status discerning weight of 87–9, 94n7;
moral domain: atemporality of 10–11, explanatory/grounding role of in
12, 15n10; order/shape within 18 Rossian pluralism 63–9, 71–2; how
moral epistemology 29; moral-epistemic grounded given non-Rossian pluralism
absolutism 90–1; moral-epistemic 85; intuition and 87–9; no grounding
minimalism 90–1 role given non-Rossian pluralism 87;
moral fetishism 75 not deontic 67, 80n16; possibility of
moral generalism 24 defining by reference to moral oughts
moral generalizations 2; and enablers/ 70; reducibility 93n1; relation to pro
disablers 22; beliefs/judgments about tanto duties 68; weighing/balancing
3, 13–14; how to represent 2, 5, 8, 14, of 65, 72, 77, 85, 91, 92, 93n1, 136;
20–1, 24, 32; inferring from verdictive see also deliberation, Legal Model of
intuitions 6; need for 17–18; pro tanto Moral Normativity, moral residue,
2; relation to atomistic explanations self-perfection, welfare
18, 25; relation to holistic explanations moral residue 79n13, 80n16, 136; and
18; role of in deliberation 90–1; role commitments 79n14; and disabling
of in normative ethical theories 5, 6; 137; and genuine moral dilemmas
type-grounding claim template for 25; 85–6, 148; and non-Rossian pluralism
see also act-consequentialism, enablers 85–7; and Rossian pluralism 65–9;
and disablers, exceptions to moral inability of moral reasons to explain
generalizations, explanatory claims 67–9; inability of pro tanto duties
moral grounding explanations see to explain 68–9; relation to moral
atomistic explanations requirements 76–8
moral importance of individuals 116 moral standing see moral status
moral intuitions 3, 6, 9, 13, 29; see also moral status 3, 93, 116, 124, 135; appeal
enablers and disablers; explanatory to as a way of avoiding the naturalistic
intuitions; theoretical moral intuitions; fallacy 120–1; drawbacks of appealing
verdictive intuitions to facts about 118; in an existential
moral meta-explanation 164–7 explanandum role 121–3; possible
moral methodology 3, 5, 6; bias in 2 explanatory roles of 118–20; see also
moral monism 2, 3, 32–3, 34, 57, 61, wronging
62, 83, 115, 124, 160, 164; and moral theorizing: ability to yield answers
type-grounding claims 32; classifying to moral questions 3, 89, 143–4, 147–8,
contractualism as a version of 54; 154; as an explanatory project 2–3,
classifying Kantianism as a version of 6, 8, 17, 143; bottom-up version of
45–6; see also act-consequentialism, 5–6, 9, 10, 12; defined 5; doxastic
Benthamite utilitarianism justification provided by 4; epistemic
consequentialism, contractarianism, balance of 14; point of 1, 4, 89; top-
contractualism, Kantianism, rule- down version of 6, 9, 10, 11, 12, 13,
consequentialism, token-grounding 90; why it involves generalization
claims 18–19; see also theoretical virtues
moral motivation 75 moral truth 7
moral particularism 21, 24, 63
moral personhood see moral status Nagel, Thomas 62, 108–9
moral pluralism 2, 3, 32, 52, 57, 61, natural selection 7, 8
115–17, 124; see also animals, non- naturalistic fallacy see moral status
Rossian pluralism, Rossian pluralism nomic explanation 30n4
Moral Rationalism: consistency with non-consequentialism 1, 62; and animal
rejection of Rossian Pluralism 72–6; research 135; see also contractarianism,
defined 72 contractualism, Kantianism, non-
moral realism 7 Rossian pluralism, Rossian pluralism
184 Index
non-humans see animals political morality 114n15, 148, 154, 159
non-ideality: principles for dealing with Portmore 62
153–4; scarcity as a kind of 152–4; poverty 138–9
theorizing about 154, 155 prediction 11–12, 23; see also explanation
non-normative facts 90, 120; role of in Prichard, H.A. 62, 68
practical deliberation 92; role of in primary goods: as distribuend 107
moral explanation/grounding 18–22, principles 156n7
25, 27, 30n2, 75, 93, 97, 148, 164–7; prioritarianism 98, 112n6; as aspect of
role of in non-Rossian pluralism 83–5; sufficientarianism 152; criticism of
role of in Rossian pluralism 61–6; role prioritarian version of distributive
of harm as a 129 welfarism 100–1; whether a
non-moral reasons 76, 91; see also self- comparative view 100; see also scarcity
perfection Psillos 12
non-Rossian pluralism 3, 75–6, 92–3, 96; psychological capacities: and fortune
and non-distributive scope questions 105; and welfare 99, 112n6; see also
116; and rescue 87; counterintuitiveness capacity for welfare
of 158–9; defined 83; distribution- psychological enhancement 103, 104
insensitivity of 3, 111; its atomistic psychologically limited/disabled humans
explanations 76; its simplicity 83–4; 119; welfaristic obligations to 101–2,
its token-grounding claims 76, 117; 104; whether unfortunate 113n7
oversimplification objection against Psychologically Normal Humans Scope
84; relation to Rossian pluralism see distributive welfarism
93n1; see also animals, deliberation, punishment 127, 128
directedness objection, enablers and
disablers, genuine moral dilemmas, Rachels 63, 116, 118; see also Rachels
Legal Model of Moral Normativity, method for answering scope questions
moral reasons, moral residue, non- Rachels method for answering scope
normative facts, Rossian pluralism, questions 116–18, 120, 121, 123
verdictive beliefs, wronging Ram-Tiktin 153
normative domain 19 rationality: distinctness of from
normative ethical theory, definition of 5 morality 72
normative ethics 2, 13, 17, 61, 63, 71, Rawls 34, 54, 107
114n15 Rawlsian contractarianism 56
normative facts 61 Raz, Joseph 79n13
normativity of accountability 36 reactive attitudes 36, 38–9, 40, 50, 56,
67–9, 86
O’Neill 47 reasons 62–4; exclusionary 79n13, 123;
Oberdiek 67 non-directedness of 66; not deontic
obligation 36, 50; see also duties 67; their strength 76; see also moral
Oppenheim 10 reasons, non-moral reasons, Reasons
opportunity for welfare 107 Ground Oughts, reasons holism
original position 54–6 Reasons Ground Oughts 70–2, 78, 91
Otsuka, Michael 143 reasons holism 64
ought, overall 70–2, 91 reflective endorsement 48, 50, 51
ought-implies-can 85 Regan, Tom 122
overriding 21, 22, 136, 165 relational normativity 36
Owens, David 36 relevance see atomistic explanations
rescuing 139; what it is 130; whether
Parfit, Derek 39, 97, 100 obligatory 3, 126, 129–31, 134,
partial grounding 20 159; see also act-consequentialism,
personal identity 102; psychological non-Rossian pluralism, rule-
continuity view of 99, 105, 113n7 consequentialism
Peter 151 research on animal subjects see animal
Pogge, Thomas 138 research
Index 185
resources: as distribuend 107, 149, concerns 115–18, 123; see also act-
155; see also health care resource consequentialism, contractarianism,
distribution distributive welfarism, Kantianism,
responsibility 54, 106–7 non-Rossian pluralism, Rachels
Ridge 71 method for answering scope questions,
rights 1, 19, 46, 54–5; see also Rowlands’s Sentient Individuals Scope, welfaristic
contractarianism distributive generalizations
Ross, W.D. 9, 62, 64, 68, 91; see also second-personal reasons 36
Rossian pluralism Segall 151
Rossian pluralism 3, 30n2, 63, 83–4, 91, self-defense 127, 128, 129
93, 160; and genuine moral dilemmas self-evidence of moral claims 4n1
85; and Reasons Ground Oughts self-perfection: moral reasons for 92;
70–2, 78, 91; and trumping reasons non-moral reasons for 92
85; defined 61–2; motivation for 85; Sen, Amartya 107, 111, 151
relation to consequentialism 44–5; sentience 55, 118, 134–5
relation to non-Rossian pluralism 93n1; Sentient Humans Scope see distributive
too-many-grounding-claims problem welfarism
for 63–4; see also consequentialism, Sentient Individuals Scope: of distributive
directedness objection, duties, Moral fortunism 104–5; see also distributive
Rationalism, moral reasons, moral welfarism
residue, non-normative facts, non- Shafer-Landau, Russ 78n5, 79n13
Rossian pluralism, verdictive facts, Shpall 79n14
victim, wronging Sidgwick 62
Rossianism see Rossian pluralism simplicity 12, 28, 32, 63–4; see also
Rowlands, Mark 53–6, 115; see also non-Rossian pluralism
Rowlands’s contractarianism Singer, Marcus 62
Rowlands’s contractarianism 53–4 Singer, Peter 1, 37, 143; his Pond Case
rule-consequentialism 62; and harm 37, 129, 133–4, 138, 159, 161
128; and rescue 129–30; and the Sliwa 29, 75
reactive attitudes 40; and the value Smith, Michael 74–5, 94n7
of abiding by moral requirements Species membership: moral relevance of
76; counterintuitiveness of its token- 101, 116
grounding claims 43–4; directedness Star 94n7
objection as applied to 37; how to Stratton-Lake 64
represent it 35; possible derivation of Street, Sharon 7–8, 162
from contractualism 39 sufficientarianism 98, 112n6; as a non-
comparative principle 153; criticism of
Sapontzis 135 sufficientarian version of distributive
Scanlon 53–4, 71, 74, 76, 146 welfarism 100; in health care resource
scarcity: and prioritarianism 100 distribution 151–4, 155; see also
Schroeder 66, 68, 78–9n5, 94n7 prioritarianism
scientific knowledge 11 supererogatoriness 5, 18, 19
scientific theorizing: as a collective supervenience 61, 81n23
effort 29; as a way to access the truth
10, 23; as a way to make predictions Tännsjö 71
23; epistemic imbalance of 14; role theoretical criteria see theoretical virtues
of intuitions in 12–13, 14; role of theoretical moral claims see moral
observations in 10, 11, 14; role of generalizations
theoretical criteria in 12; temporality theoretical moral intuitions 2
of 11, 12, 15n10; see also theoretical theoretical virtues: appeal to in moral
virtues theorizing 10, 12, 27–8; appeal to
scope: and harming 135; in Kantianism in scientific theorizing 10, 12–13,
48; in Rowlands’s contractarianism 55; 27; (predictive) power 10; see also
questions of for non-distributive moral act-consequentialism, atomistic
186 Index
explanations, Consequentialism’s 93; how grounded given non-Rossian
Type-Grounding Claim, Kantianism, pluralism 83–4; how to explain 17–18,
non-Rossian pluralism, type- 18–20, 25; inferring from general
grounding claims beliefs 6, 90
Thompson, Michael 118 verdictive intuitions 4, 14, 15n10, 159,
Thomson 94n1 162; abandoning 2; accommodating
thought experiments 144–5, 146–8 2; as input to moral deliberation/
token-grounding claims: about harm theorizing 3, 5–6, 87–9, 108–9;
129; are implied by type-grounding moral analogue of observations 12,
claims 25; criteria for accepting 26–7; 15n7; relation to the explanatory
existence and usefulness of intuitions power of a normative ethical theory
about 26–7; of moral monism 32–3; 11–12; relation to moral categoricals/
imply type-grounding claims 25; imply conditional claims 9, 15n6; reliability
verdictive claims 26; see also atomistic of 7–8; see also moral generalizations
explanations, Consequentialism’s verdictive judgments see verdictive beliefs
Type-Grounding Claim, Kantianism, victim 26; inability of consequentialism
non-Rossian pluralism to identify a 37–9, 41–3; inability of
top-down moral theorizing see moral Rossian pluralism to identify a 65–6;
theorizing see also Consequentialism’s Type
tragic moral contexts 85 Grounding Claim
Transplant Case 41–2, 160 virtue ethics 59n24, 149; see also
trumping reasons see Rossian pluralism Kantianism
Tulodziecki 12 Voorhoeve, Alex 143
type-grounding claims 90–1, 96, 117;
about harm 129; are implied by Wallace, R. Jay 36
token-grounding claims 25; existence Warren, Mary Ann 119, 122
and usefulness of intuitions about welfare 3, 87, 112n6; a final personal
26–8; imply token-grounding good 105–6; and well-being 99; as
claims; of Kantianism 45; see also distribuend 105–8; comparability of
Consequentialism’s Type-Grounding across individuals 112n6; contingent
Claim, holistic explanation; moral limits to 104; its relation to moral
generalizations, moral monism, reasons 64, 98; limits to 99; purported
token-grounding claims universality of true theory of 112n6;
status quo ante level of 100, 131–2,
understanding 28 133; see also animals, capacity for
Urmson, J.O. 62, 87 welfare, opportunity for welfare,
utilitarianism 34, 108 psychological capacities, welfare-
potential, welfaristic distributive
Vallentyne, Peter 98, 100, 104–6 generalizations
van Fraassen 23 welfare-potential 98, 101; defined 99;
Väyrynen 21, 80n16, 165–6 necessary inequalities in 99, 102,
veil of ignorance 54, 56 103–4; necessary limits to 99, 102,
verdictive beliefs 9, 10, 14; how to 104; of humans vs. of non-humans
arrive at given non-Rossian pluralism 100; range of 101–2, 109; see also
89–91; implied by token-grounding animals, distributive welfarism
beliefs 26; implying reason-strengths welfaristic distributive generalizations
89; relation to explanatory beliefs (WDGs): do no grounding work 96–7;
8; role of in the method of reflective fixing the scope problem for 103–5;
equilibrium 13–14 how to derive 97, 98, 111; statement
verdictive facts 2, 5, 120; as explananda of scope problem for 97–8; see also
119; beliefs about 3, 4; how grounded distributive welfarism
93n1, 136, 148, 164–7; how grounded well-being: see welfare
given Rossian pluralism 61–9, 72, willing 51, 53, 57, 92
Index 187
Williams, Bernard 34 neo-Kantianism to account for 50;
wrong-sufferer see victim inability of Rossian pluralism to
wronging 53; ability of non-Rossian account for 65–9, 118; what it is 36;
pluralism to account for 117; and see also directedness, Rowlands’s
moral status 121–3; inability of contractarianism
consequentialism to account for
36, 57; inability of Korsgaard’s zero-sum cases 127, 129

(Routledge Studies in Ethics and Moral Theory 43) Benjamin Sachs - Explaining Right and Wrong - A New Moral Pluralism and Its Implications-Routledge (2017)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Routledge Studies in Ethics and Moral Theory 43) Benjamin Sachs - Explaining Right and Wrong - A New Moral Pluralism and Its Implications-Routledge (2017)

Uploaded by

Copyright:

Available Formats

“This book contains several original arguments and thought-provoking

ideas. The primary argument about moral theorizing as an explanatory

Explaining Right and Wrong aims to shake the foundations of contemporary

Benjamin Sachs is a lecturer in philosophy at the University of St. Andrews.

34 The Intrinsic Value of Endangered Species

35 Ethics and Social Survival

36 Love, Reason and Morality

38 In Defense of Moral Luck

39 Risk, Technology, and Moral Emotions

40 Wittgenstein’s Moral Theory

41 Welfare, Meaning and Worth

43 Explaining Right and Wrong

1 Normative Ethical Theorizing as an Explanatory Project 5

2 How Should We Choose Between Competing Explanatory

4 Against Rossian Pluralism 61

6 The Question of Scope, Part 1: Distributive Moral Concerns 96

7 The Question of Scope, Part 2: Non-Distributive Moral

8 Doing Harm and Failing to Rescue 126

9 The Distribution of Health Care Resources 143

Appendix A: Moral Meta-Explanation 164

2. Representing Normative Ethical Theories as Being

3. The Reliability of Our Explanatory Intuitions

4. The Possibility of Appealing to Our Intuitions About

‘necessarily, if an action is an instance of killing then it is impermissible’

and then reflect on our judgments as to the truth of these conditionals/

5. The Possibility of Appealing to Theoretical Criteria

5.1 Scientific Prediction and Moral Prediction9

5.2 Explanatory Intuitions in Science and in Moral Philosophy

2. What Kind of Explaining Do Normative Ethical

2.1 Atomistic Explaining and Holistic Explaining

2.2 What Are the Relata of Atomistic Explanations?

2.3 The Grounding Model of Atomistic Explanation

2.4 The Is-Ought Problem

3. Generalizations and Exceptions

3.1 Introducing Enablers and Disablers

3.2 A Worry About Context-Sensitivity

4. How to Determine Which Normative Ethical Theories

4.1 Criteria for Accepting Token- and Type-Grounding Claims

4.2 Criteria for Accepting Sets of Type-Grounding Claims

2. What Do Monistic Theories Really Say About

Table 3.1 Standard and Alternative Ways of Representing Monistic Theories

Natural Way to Alternative Way to Represent the

Act- All verdictive factsNon-normative facts X, Y, and Z

3.1 Consequentialism’s Moral Explanations

Just about all self-described consequentialists accept some version of this.

3.2.1 The Objection

a) Obligation: Wronging bears a close connection to the idea of obliga-

3.2.2 A First Response to the Objection

3.2.3 A Second Response to the Objection

3.2.4 A Third Response to the Objection

3.2.5 A Fourth Response to the Objection

3.3 The Counterintuitiveness Objection

4. Kantian Ethics and Cruelty to Animals

4.1 Kantian Ethics as Monistic

4.2 Cruelty to Animals

4.3 A Response and a Rejoinder

4.4 Korsgaard’s Neo-Kantianism

4.4.1 Respect for Animal Nature

4.4.2 First Objection to Korsgaard: Having Standing

4.4.3 Second Objection to Korsgaard: The Superfluousness of

1. Treating one’s own humanity as a mere means.

In a much earlier publication Korsgaard offers an example of each kind of

5.1 Varieties of Contractarianism and Contractualism