. 1
( 10)


This page intentionally left blank
Goodness and Justice

In Goodness and Justice, Joseph Mendola develops a uni¬ed moral theory
that defends the hedonism of classical utilitarianism, while evading util-
itarianism™s familiar dif¬culties by adopting two modi¬cations. His the-
ory incorporates a new form of consequentialism. When, as is common,
someone is engaged in con¬‚icting group acts, it requires that one perform
one™s role in that group act that is most bene¬cent. The theory also holds
that overall value is distribution-sensitive, ceding maximum weight to the
well-being of the worst-off sections of sentient lives. It is properly con-
gruent with commonsense intuition and required by the true metaphysics
of value, by the unconstituted natural good found in our world.

Joseph Mendola is professor and chair in the Department of Philosophy at
the University of Nebraska“Lincoln. He is the author of Human Thought
and of articles on ethics, metaphysics, and philosophy of mind.
cambridge studies in philosophy
General Editor walter sinott-armstrong (Dartmouth College)
Advisory Editors:
jonathan dancy (University of Reading)
john haldane (University of St. Andrews)
gilbert harman (Princeton University)
frank jackson (Australian National University)
william g. lycan (University of North Carolina at Chapel Hill)
sydney shoemaker (Cornell University)
judith j. thomson (Massachusetts Institute of Technology)

Recent Titles:
mark lance and john o™leary-hawthorne The Grammar of Meaning
d. m. armstrong A World of States of Affairs
pierre jacob What Minds Can Do
andre gallois The World Without the Mind Within
fred feldman Utilitarianism, Hedonism, and Desert
laurence bonjour In Defense of Pure Reason
david lewis Papers in Philosophical Logic
wayne davis Implicature
david cockburn Other Times
david lewis Papers on Metaphysics and Epistemology
raymond martin Self-Concern
annette barnes Seeing Through Self-Deception
michael bratman Faces of Intention
amie thomasson Fiction and Metaphysics
david lewis Papers on Ethics and Social Philosophy
fred dretske Perception, Knowledge, and Belief
lynne rudder baker Persons and Bodies
john greco Putting Skeptics in Their Place
ruth garrett millikan On Clear and Confused Ideas
derk pereboom Living Without Free Will
brian ellis Scienti¬c Essentialism
alan h. goldman Practical Rules: When We Need Them and When We Don™t
christopher hill Thought and World
andrew newman The Correspondence Theory of Truth
ishtiyaque haji Deontic Morality and Control
wayne a. davis Meaning, Expression and Thought
peter railton Facts, Values, and Norms
jane heal Mind, Reason and Imagination
jonathan kvanvig The Value of Knowledge and the Pursuit of Understanding
andrew melnyk A Physicalist Manifesto
For my daughter,
Lily Grif¬n
Goodness and Justice

A Consequentialist Moral Theory

University of Nebraska“Lincoln
©¤§ µ®©© °
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press
The Edinburgh Building, Cambridge  µ, UK
Published in the United States of America by Cambridge University Press, New York
Information on this title: www.cambridge.org/9780521859530

© Joseph Mendola 2006

This publication is in copyright. Subject to statutory exception and to the provision of
relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2006

©®-± ·-°-µ±±-±±°- eBook (EBL)
©®-±° °-µ±±-±±°-± eBook (EBL)

©®-± ·-°-µ±-µµ-° hardback
©®-±° °-µ±-µµ-° hardback

Cambridge University Press has no responsibility for the persistence or accuracy of µ¬s
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.

Acknowledgments page ix

1 Introduction 1

Part One. A Better Consequentialism
2 Multiple-Act Consequentialism 23
3 Three Objections 64

Part Two. Hedonism
4 Intuitive Hedonism 105
5 Natural Good 139

Part Three. Maximin
6 Just Construction 187
7 Maximin, Risks, and Flecks 226

Part Four. Advice for Atomic Agents
8 A Code 273

Bibliography 315
Index 323


I owe special thanks for extensive and helpful criticism during the long
unfolding of this project from Tom Carson, Allan Gibbard, Jaegwon
Kim, Mark van Roojen, and several anonymous referees. I am also
grateful for help from Robert Audi, Bryan Belknap, Tim Black, Dick
Brandt, Mark Cullison, Dave Cummiskey, Steve Darwall, Mark Decker,
Bill Frankena, Jean Grif¬n, Russell Hahn, Jennifer Haley, Robert
Hanna, Thomas Hill, Jr., Leo Iacono, Clayton Littlejohn, Heidi Malm,
Sally Markowitz, Donette Petersen, Peter Railton, Beatrice Rehl, Guy
Rohrbaugh, Margaret Skean, Mike Tonderum, Rainer W. Trapp, J. D.
Trout, Sheldon Wein, and the students of Philosophy 920 in fall 2004.
Thanks to the publishers for permission to reuse material from the arti-
cles “Multiple-Act Consequentialism,” forthcoming in Nous; “Intuitive
Maximin,” Canadian Journal of Philosophy 35 (2005), 429“439; “Con-
sequences, Group Acts, and Trolleys,” Paci¬c Philosophical Quarterly 86
(2005), 64“87; “Justice within a Life,” American Philosophical Quarterly 41
(2004), 125“140; and “Objective Value and Subjective States,” Philoso-
phy and Phenomenological Research 50 (1990), 695“713. Modi¬ed forms
of “Intuitive Hedonism,” forthcoming in Philosophical Studies, and “An
Ordinal Modi¬cation of Classical Utilitarianism,” Erkenntnis 33 (1990),
73“88, are used with the kind permission of Springer Science and Busi-
ness Media.


Ethicists must deliver intuitive platitudes about lying, murder, theft, injury,
and the whole familiar bunch. If we can hope to surprise or advance, it is
only in neglected or undeveloped corners of normative consensus. True
conclusions in ethics must be mostly boring.
And yet ethicists must deliver a vindication of common moral claims.
We must give arguments that should convince all the reasonable that
our normative claims are correct. To put it grandly, we must deliver a
transcendental vindication of those claims, from the point of view of the
universe. To put it less grandly, we must provide a direct argument for the
truth of the claims, independent of appeal to common moral intuitions.
We need to provide considerations that ought to convince those who are
outside normative consensus that we are right and they are wrong. If we
merely monger common intuitions, unsupported by argument, evidence,
or fact beyond their mere familiarity or warm fuzziness, we will have little
to add to the standard wisdom of the street and the sea lanes. And worse,
to the immoral, disagreeable, skeptically minded, or just diverse, it may
not unreasonably seem that there is nothing to philosophical ethics but so
much talk, so much high-minded or sanctimonious but otherwise empty
And so we face two dragons. There may be no suf¬ciently transcen-
dental vindication of any ethical claims, so that ethics is just a bunch of
chatter. Or there may be a transcendental vindication of something, but
not of the common platitudes, perhaps rather of something shockingly
revisionary like the empty moral equivalence of all human action or the
moral insigni¬cance of human life or pain. In either case, philosophical
ethics fails. Ethical discourse is committed both to the normative plati-
tudes and to the objective correctness of such claims. Perhaps there are

limits to what should be required of an argument in ethics, and perhaps
there are limits to the probity of commonsense intuition. But if we can™t
provide a somewhat transcendental vindication of most moral platitudes,
then ethics sinks, with all hands.
Or so I believe. I believe that it is a constraint of the meaning of
“morally wrong” that intuitively heinous murder is morally wrong, and
yet that there is a robustly objective status to that claim, so that contrary
meaning constraints would be incorrect. But if you think I™m wrong
in one way or the other, if you think that intuitive moral implications
are alone enough to vindicate an ethical theory or that transcendental
vindication alone is enough, if you think that one of these constraints is a
mere idol of our marketplace, you should still be satis¬ed if a view meets
both tests. And since both tests must be met by any theory that has the
ghost of a chance of convincing all sober practicing ethicists, you should
still hope that both monsters can be slain by the true theory.
This book tries to develop a moral theory that meets both tests. This
chapter will sketch that theory, my arguments, and the structure of the
book. Sections I and II introduce the theory and provide some rough
motivation for its features. Section III introduces my arguments for it.


Classical utilitarianism is one traditional ethical theory.1 As a form of
consequentialism, it claims that the rightness of acts depends on the value of
their consequences, that rightness depends on goodness in that way. And
it claims in particular that value is utility, the very earthy happiness of all,
pleasure and the absence of pain.
The maximization of everyone™s happiness and the minimization of
everyone™s unhappiness is one humane and reasonably rational grounds for
ethics. But there are forceful intuitive objections to such a theory, which
suggest that, even in the best case, it must be modi¬ed.2 We should not
simply ignore the obvious moral salience of general happiness. But nor
should we ignore standard objections to classical utilitarianism.
It may be natural to presume with classical utilitarianism that the
best outcome contains the greatest sum of well-being. But this aggrega-
tive conception of best outcomes seems improperly indifferent to how
well-being is distributed among individuals. If overall utility would be

1 Mill (1979).
2 But for a spirited defense of classical utilitarianism, see T¨ nnsj¨ (1998).
a o

maximized under some arrangement in which goods were distributed
to people whose circumstances were already comfortable, while other
people were allowed to starve, then that is the arrangement that classical
utilitarianism would recommend. And that would be wrong. It would
be unjust.
And consequentialism itself, at least in a direct form, which speci¬es
that a right act is one that will produce the best overall outcome, faces
serious objection, whatever conception of good outcomes we deploy.
Consequentialism seems to require that we perform whatever act will,
in a given situation, produce the best overall outcome. And this may
require doing something unjust and intuitively horrible, such as mur-
dering an innocent. There is a standard consequentialist response to this
dif¬culty, which is rule consequentialism or some other indirect form.
For instance, our duties may not be best acts, but rather acts required by
best systems of moral rules, where systems of moral rules are best if their
general acceptance would lead to best outcomes. But there is also a stan-
dard dif¬culty with this response. It is hard to understand how indirect
consequentialisms, when they differ from simple and direct forms, respect
the original intuitive motivation for consequentialism, which is that best
outcomes are what morally count.
This book develops a variant of classical utilitarianism that can evade its
traditional dif¬culties, while yet retaining its intuitive motivations. This
variant maintains the hedonism of the classical view, the traditional con-
ception that well-being or goodness is the presence of pleasure and the
absence of pain, and argues that familiar objections to hedonism are mis-
taken. But two key modi¬cations of classical utilitarianism will allow the
view proposed here to evade the other two familiar intuitive objections
to traditional forms of utilitarianism, objections rooted in two sorts of
concern about justice.
Classical utilitarianism is a consequentialist view. It holds that the moral
status of an act depends on the value of its consequences. And one compo-
nent of the true ethical theory is, I will argue, a new form of consequen-
tialism. It evades at once the traditional intuitive dif¬culties of act con-
sequentialism, and the unstable rationale that plagues indirect forms like
rule consequentialism. I call this conception of proper action Multiple-
Act Consequentialism, or MAC.
The key to this possibility is that there are group acts. Indeed, there is a
multiplicity of overlapping group acts. There is often no single fact of the
matter that a piece of your momentary behavior is in particular an act you
can perform in the moment, rather than a portion of a longer individual

act that will take time and require the cooperation of yourself tomorrow,
or a portion of a corporate act of some group of several individuals of
which you are part. There is, rather, often a multiplicity of things you
are doing through a single bit of behavior all at once. These facts allow
MAC to perform direct consequentialist evaluation of acts that yet yields
intuitive normative implications. This is roughly how it works: There can
be a group act of which you are part that requires particular activity by you,
which yet con¬‚icts with activity required if you take part in another group
act or perform certain individual acts. Direct consequentialist evaluation of
these overlapping acts can yield a con¬‚ict. And, I will argue, the central
consequentialist rationale “ which is that the value of consequences is
what matters “ supports resolving the con¬‚ict in favor of the alternative
available to you that is your part of the group or individual act with the
best consequences.
We are rowing together, in important pursuit, and I can secretly ease
up to create a little extra aesthetic utility, to sip my lemonade and rattle
the ice cubes. You will all keep rowing and get us where we need to go,
around that turbulence, and I can grab a little extra utility on the side,
by that slight injustice. But then I would be defecting from an act with a
stronger direct consequentialist rationale. I shouldn™t do it.
MAC involves four key tenets: (1) There are group agents of which
individuals are constituents, and such that an individual may be part of
more than one group agent, and their acts constituents of the acts of more
than one group agent. (2) Direct consequentialist evaluation of the options
of group agents is appropriate. (3) Sometimes, but only sometimes, one
should follow one™s role in a group act even at the cost of the overall
good one could achieve by defecting from that role. One should defect
from a group act with good consequences only if one can achieve better
consequences by the defecting act alone than the entire group act achieves.
(4) When different bene¬cent group agents of which one is part specify
roles that con¬‚ict for one, one should follow one™s role in the group act
with more valuable consequences.
How Multiple-Act Consequentialism works in detail, why it is true,
and how it helps deliver intuitive implications with a ¬rm consequentialist
rationale are the burdens of Part One. But there are other reasons to
worry about utilitarianism and its descendants. Classical utilitarianism
seems transparently unjust when it suggests that we should maximize the
good irrespective of its distribution. And it does not deploy a conception
of human good on which ethicists have reached consensus. These are
both worries about the basic normative principle that is applied by

consequentialism. The basic normative principle I favor involves the
second and third elements of my view, which correspond to these two
points of intuitive concern about traditional utilitarianism. This principle
amalgamates and reconciles one traditional conception of goodness and
one traditional conception of distributive justice. It unites a classical
utilitarian conception of goodness rooted in pleasure and pain, and an
egalitarian conception of distributive justice requiring special concern for
the worst off. I will call this principle the Hedonic Maximin Principle,
or HMP for short.
The hedonism of HMP is developed and defended in Part Two. Hedo-
nism is not fashionable, and we will need to examine this prejudice. I will
have quite a bit of arguing to do before I can convince most contemporary
ethicists that hedonism is viable. But hedonism is a venerable position. For
instance, its utilitarian roots are evident. It is the position of the classical
utilitarians, of Bentham, the Mills, and Sidgwick. We will see that this
venerable view deserves greater respect than it has lately received from
philosophers, and we will also see that it ¬ts well with the transcendental
vindication of ethical claims that we will eventually need to produce.
But there is a signi¬cant modi¬cation of the traditional utilitarian nor-
mative principle that is the third component of my view. It is a particular
distribution-sensitive assessment of the value of overall states of affairs.
The Hedonic Maximin Principle re¬‚ects another concern than the max-
imization of hedonic value. It is immediately responsive to the distributive
justice of situations, in a way that skews our concern toward the bene¬t of
the worst-off among us. Consequentialists shouldn™t just pile up happiness
in the world “ in hoards of ingots or cities of butter sticks “ but should
also make sure that it is properly and equitably distributed.
That second aspect of HMP, concerning justice, is developed and
defended in Part Three. It is characteristic of some of traditional utilitar-
ianism™s competitors, and helpful in disposing of some traditional objec-
tions to traditional utilitarianism. For instance, John Rawls™s in¬‚uential
theory of social justice, explicitly conceived in opposition to the utilitar-
ian tradition, requires, after certain basic liberties are assured and it is also
assured that differences in income are attached to positions open to all, that
differences in prospects for monetary income and wealth of representa-
tive members of basic social groups satisfy a “Difference Principle”, which
unites a concern with maximization and a concern with just distribution.3

3 Rawls (1971).

According to this principle, any inequalities in birth prospects of income
must be to the bene¬t of the worst-off. This is a kind of “maximin”
principle, which tells us to maximize the minimum level of some sort of
well-being, to make the worst-off as well-off as possible. HMP is also a
kind of maximin principle, but applied within the context of a classical
utilitarian value theory. It implies, roughly, that one outcome is better
than another when the worst-off are better off, and also that relative well-
being is, as the classical utilitarians suggested, a straightforward matter of
pleasure and pain.
More exactly, the Hedonic Maximin Principle incorporates two
clauses, which specify, respectively, a risk-averse treatment of chances and
a distribution-sensitive treatment of outcomes. The details may seem hard
to understand, unmotivated, or false until the detailed discussions of the
following chapters. But a brief full statement is this:
First, consider two lotteries (that is to say, alternative sets of chancy
outcomes where the probabilities of the members of each set sum to one)
over complete situations (that is to say, more or less, over possible worlds).
When those lotteries consist of the same number of equally probable
outcomes, the better lottery is the one that has the better worst-possible
outcome. If they have equally bad worst-possible outcomes, then the bet-
ter lottery is the one that has the better second-worst-possible outcome,
and so on. We can always represent a lottery whose possible outcomes
have rational probabilities as a lottery over equally probable outcomes, and
we can always compare two lotteries over equally probable outcomes by
comparing two equivalent lotteries that have the same number of equally
probable outcomes, so this method yields a complete ordering of lotteries
over complete situations, from worst to best. This ordering implies that
of any two lotteries over complete situations, the better lottery is the one
that has the better worst-possible outcome.
Second, consider two complete situations. Of two complete situations
that contain the same number of momentary bits of experience, the better
situation is the one that has the better worst momentary bit of experience.
The value of a momentary bit of experience is its level of ordinal hedonic
value or disvalue of a sort shortly to be characterized. If two situations have
equally bad worst momentary bits of experience, then the better situation
is the one that has the better second-worst momentary bit, and so on.
And any such situation is equivalent in value to another that has the same
number of momentary bits of experience at each level of ordinal value and
disvalue as the ¬rst plus any number of momentary bits with null value.
So this speci¬es a complete ordering from worst to best of complete

situations. It implies, more or less, that of any two such situations that
include painful experiences, the better situation is the one that has the least
severe most painful experience. It implies, more or less, that of any two
situations that contain no painful experiences, the better situation is the
one that contains the greater number of positively valuable experiences.
Despite what will seem at the moment the obscurity of some of the spe-
ci¬c details of this formulation, a normative principle of approximately this
sort is not unprecedented. The value theory incorporated in my account is
reminiscent of classical utilitarianism, and the utilitarian Henry Sidgwick
thought that, when two states of affairs contained the same overall quan-
tity of happiness, still one state might be preferable because its distribution
of happiness was just, and that just distributions were equal ones.4 There
are also nonutilitarian precedents. Of course, there is Rawls. But William
Frankena also proposed a deontological theory that incorporates two basic
duties, to maximize the good and to be just.5 Samuel Schef¬‚er has devel-
oped a hybrid ethical theory that not only allows agents sometimes to
desist from maximizing the good when pursuing individual projects, but
also requires that their maximization of the good be tempered with dis-
tribution sensitivity, so that the less the relative well-being of a person, the
greater the weight that should be given to bene¬tting him or her.6 David
Brink has built a kind of distribution sensitivity into his modi¬cation of
utilitarian theory, by specifying that certain sorts of basic well-being have
great weight in determining the value of a situation.7 Thomas Scanlon
has developed Rawls™s maximin principle in certain ways, and Thomas
Nagel has suggested that we need to give greater weight to the interests
of those who are less well-off.8
So ethical and political theories that mix a concern to maximize well-
being with a concern for just distribution, particularly in regard to the
worst-off, are not unknown. But HMP is somewhat different from and,
I will argue, superior to each of these competitors.
The ¬rst difference between my proposal and its close competitors is
that, of all these accounts, only Sidgwick™s deploys the kind of classical
hedonistic value theory, with pleasure the only positive value and pain
the only disvalue, that my account will develop. In the current climate,

4 Sidgwick (1907: 416“417).
5 Frankena (1973).
6 Schef¬‚er (1982).
7 Brink (1989: 268“273).
8 Scanlon (1973); Nagel (1991: 73).

that may hardly seem an advantage. The abandonment of classical hedonist
value theories by the recent competitors noted is motivated by widespread
worries that such traditional hedonist theories are inadequate, in partic-
ular because they have certain counterintuitive implications. But we will
eventually see that a hedonistic value theory, when properly developed
and deployed, is not unintuitive, and that indeed the ethical theory devel-
oped here has detailed normative implications that suitably match our
commonsense morality. And we will also see that a classical hedonistic
conception of value ¬ts far better than its competitors with one key sort
of argument that should underwrite ethical theories. It accords far better
with our need to provide a reasoned vindication and legitimation of our
normative principles and claims over possible competitors.
We will also see that HMP enjoys advantages over Sidgwick™s classical
alternative. According to Sidgwick™s account, justice as equality matters
merely in a quite secondary way “ for instance, when we are choosing
between two outcomes that involve the same overall quantity of happiness
distributed among the same number of people. The Hedonic Maximin
Principle marries a traditional utilitarian conception of value with the
greater distribution sensitivity characteristic of more recent competitors,
and in a way that we will see is at once more intuitive and capable of
reasoned vindication independent of appeal to ethical intuition.
Still, this greater distribution sensitivity may itself suggest grounds for
worry about HMP. That is because even some fans of maximin principles
have held that such straightforward application as I propose will yield
quite counterintuitive applications.9 And the novel form of my maximin
principle, which maximins over risks to momentary bits of experience,
and not over whole people, may suggest further grounds for intuitive
objection. But, as I™ve said, we will eventually see that in fact the theory
to be proposed here has plausible ethical implications, despite any apparent
threats of intuitive implausibility. In part, as we will see in Part Three,
this results from the way in which maximining over risks to momentary
bits differs from more familiar sorts. In part, as we will see in Part Two,
it is because hedonism is more intuitive than is often presumed. But
the intuitive ethical implications of my proposal also stem in part from
Multiple-Act Consequentialism as developed in Part One, which would
help in the defense of any plausible consequentialist principle, and also
from MAC™s speci¬c interaction with HMP.

9 For instance, Rawls and Scanlon.

Multiple-Act Consequentialism and the Hedonic Maximin Princi-
ple together imply the fourth component of my proposal. That fourth
component provides an intuitive set of speci¬c moral constraints, requir-
ing intuitively moral actions. I will call it the Proposed Code, or PC
for short. MAC and HMP are, I hope, of independent interest, but we
will see in Part Four that, in conjunction, they imply this intuitive set of
moral injunctions. It incorporates constraints reminiscent of the virtues
favored by Aristotelian virtue theories, which specify individual traits
whereby momentary individuals become parts of successful continuing
agents capable of unitary moral behavior over time. It also incorporates
constraints reminiscent of the obligations recognized by deontological
theories like Kant™s or Ross™s, which specify forms of action for individ-
uals whereby cooperative moral behaviors become possible and effective,
whereby individuals can constitute groups acting appropriately together.
And it also incorporates a properly chastened concern with the pur-
suit of justly distributed good. We will compare PC with the details
of commonsense morality, and see that they properly coincide. We will
also discuss two cases that common sense does not resolve, but that this
theory does resolve in a plausible way. They are our obligations to the
needy and our obligations to refrain from certain sorts of bad corporate
The Hedonic Maximin Principle incorporates one feature of tradi-
tional competitors of traditional utilitarianism, and so re¬‚ects one intuitive
concern for justice. And Multiple-Act Consequentialism and hence the
Proposed Code are reminiscent in a second way of some of utilitarianism™s
traditional competitors, and re¬‚ect another intuitive concern for justice.
So this moral theory is, in this dually chastened sense, a conception of a
just good. But still it is a conception that ultimately roots ethical valuation
in merely goodness, in just good, and indeed hedonic good of a classically
utilitarian form. Goodness retains that priority.


The theory developed here is a modi¬cation of classical utilitarianism to
evade pressing intuitive objections. Yet there are other dominant strands
in common sense, not just a few intuitive objections to a single dominant
utilitarian strand. Mere modi¬cation may not seem enough.
Our complex social world is heir to many different ethical traditions.
If we are to deliver intuitive normative implications, we need a normative
theory that is in some sense a reconciliation of various competing strands in

our normative inheritance. Even if we achieve such a reconciliation, it may
be hard to see how there can be an argument that should independently
convince everyone that such a conciliatory normative view is true, and
not just an arbitrary political compromise among competing dominant
normative factions. Yet if there isn™t such an argument, then there will be
no transcendental vindication of such a view. But focus for the moment
on the point about reconciliation.
Each of the traditional competing ethical systems “ for instance,
Aristotle™s, Mill™s, and Kant™s “ emphasizes certain strands in our rich and
confusing normative heritage, toward which each of us feels some intu-
itive pull. While such strands can be developed in many ways, and some
developments leave them inconsistent and competing like the famous his-
torical systems, still the strands should, it seems, be woven together in a
consistent and uni¬ed manner, if they can be. Since the various intel-
ligent and conscientious individuals who have prominently represented
each of the various strands of our tradition were surely all onto something
that is re¬‚ected in our commonsense ethical consciousness, and since that
common sense should bear at least some argumentative weight, it would
be better, other things equal, if an ethical theory drew all the strands
together, if it found some way to interpret the various apparently com-
peting elements in our tradition of ethical discussion so that they came
out consistent and uni¬ed, if it provided some sort of reconciliation of
these various elements of our diverse tradition, and not just a patched-up
and barely acceptable version of one strand “ say, the utilitarian strand. Of
course, it is also important that this reconciliation eventually be capable of
a proper independent rationale, of a transcendental vindication. But even
bare reconciliation may be hard enough.
There are two deep con¬‚icts in our tradition that my proposal can
reconcile in a well-founded way, despite its roots in utilitarianism. One
re¬‚ects the standard teaching division of moral theories. There are con-
sequentialist theories, like utilitarianism, which hold that right action is
that which leads to the maximization of value. There are deontological
theories, like Kant™s and Ross™s,10 which hold that right action is not right
solely “ nor perhaps even in part “ because of the value of its consequences,
but rather at least in part because of its intrinsic nature. There are virtue
theories, like Aristotle™s, which focus ¬rst not on the rightness of action
but on the goodness of lives. And there are also rights-based theories.

10 Kant (1996a); Ross (1930).

While my account is within the utilitarian tradition, this standard
typography isolates four strands that it weaves together, though with a
certain spin and emphasis re¬‚ecting its utilitarian origin. It is a reconcili-
ation from the perspective of one of the key strands. No doubt there are
other reconciliations possible, but this is one. The third component of my
proposal, the distribution-sensitive form of HMP, demands a certain kind
of equity, and hence respects Kant™s central belief that reason demands that
we treat like as like. It also constitutes one sort of distributional right. The
¬rst component of this theory, Multiple-Act Consequentialism, naturally
encompasses basic concerns of deontological and virtue-based theories,
as we will see in Part One. This proposal is one legitimate heir to and
development of the central concerns of each of these traditional classes of
alternative ethical theories, and hence partakes of some of the intuitive
rationale of each. We will gradually see this in detail.
A second familiar typography of ethical theories whose threads can be
drawn together is Thomas Scanlon™s triad of “philosophical explanations
of the nature of morality”.11 These are general theories about the nature
of the grounds of truth for moral claims, the proper moral epistemology,
and the nature of the reasons that morality provides for us. Scanlon distin-
guishes intuitionism, philosophical utilitarianism, and contractarianism as
theories of this sort.
The primary underlying concern for intuitionists is to preserve the full
range of intuitive ¬rst-order normative judgments about the right, good,
and just, which do of course more or less directly engage our motivation.
The Proposed Code will seem, I hope, quite hospitable to intuitionists.
But consider the other two philosophical explanations of morality.
Philosophical utilitarianism, as distinguished from utilitarianism as a
¬rst-order normative doctrine, is the thesis that the only fundamental
moral facts are facts about individual well-being. Such facts have obvious
motivational signi¬cance, because of our at least loosely sympathetic and
benevolent inclinations, and hence sympathy is, according to philosoph-
ical utilitarianism, the primary moral motivation.
Contractarianism holds that an “act is wrong if its performance under
the circumstances would be disallowed by any system of rules for the
general regulation of behaviour which no one could reasonably reject as a
basis for informed, unforced general agreement.”12 The basic idea is that
morality is a scheme of cooperation that should command the informed

11 Scanlon (1982).
12 Ibid., 110.

and uncoerced allegiance of all the reasonable, and that it is supported
by the desire to be able to justify one™s actions to others on grounds that
they could not reasonably reject.13 It is rooted not in benevolence but in
a kind of reciprocity.
While philosophical utilitarianism is not a ¬rst-order normative doc-
trine, it seems to have normative implications. If all that matters morally
is individual well-being, that seems to imply that we should maximize the
sum of individual well-being. Scanlon once argued that the contractarian
conception may support more distribution-sensitive normative principles,
since it naturally directs our attention to those who would do worst under
a possible system of normative rules. If anyone has reasonable grounds for
objecting to that system, it is likely that they will.14 And there are also
other ways in which reciprocity supports forms of moral constraint that
seem to con¬‚ict with the maximization of well-being, which have become
more prominent in Scanlon™s later work,15 and to which we will return.
If both these philosophical conceptions of the nature of morality are
present in our normative tradition, and if both are rooted in motivations “
benevolence and reciprocity “ to which most of us by training or nature
are susceptible, then we seem to face a strain in our ethical inheritance.
It is a strain not unlike the strain between self-interest and morality that
Sidgwick discerned in our practical reason.16 It may seem that the issue
between, on the one hand, philosophical utilitarianism and a concern
to maximize the sum of well-being, and, on the other, contractarianism
and a concern with distribution and other manifestations of reciprocity,
cannot be rationally adjudicated. So some reconciliation of these disparate
philosophical conceptions of morality would seem, if possible, desirable. If
they could be developed together into one consistent account, that would
eliminate the threat of deep incoherence in our normative tradition.
But we will see that they are consistent, and reconciled in the the-
ory proposed here. And so this view can claim the heritage of both the
contractarian and the philosophical utilitarian traditions, once they are
clari¬ed and rationalized. And as I said, we will also see that my proposal
is quite friendly to intuitionists.
Of course, more than loose-limbed motivation is required. The next
section sketches and locates my arguments for this theory.

13 Ibid., 116.
14 Ibid., 123.
15 Scanlon (1998).
16 Sidgwick (1907: 496“509).


There are three converging arguments for this ethical theory, all of which
I believe are necessary. This re¬‚ects that fact that two monsters threaten
ethics, and one has two heads.
The ¬rst argument does not depend on an appeal to ethical intuitions,
and provides the suitably transcendental vindication we require, the nec-
essary argument more or less from the point of view of the universe,
which I will call a “direct” argument. It has ¬ve broad steps. First, I will
argue that a crucial feature of ethical discussion is that it is a form of
reason giving in which appeals for a reasoned justi¬cation of legitimate
normative principles and claims must always be capable of being met.
Second, I will argue that this feature of ethical discussion can be plausibly
underwritten only by a particular metaphysics of the ethical according to
which, ¬rst, there are natural and objective properties of pleasantness and
painfulness that are yet normative properties and, second, there are certain
principles constraining how the value of a whole is related to the values
of its parts. Third, I will argue that this entails the Hedonic Maximin
Principle. Fourth, I will argue for the truth of Multiple-Act Consequen-
tialism, whatever our moral intuitions. Fifth, I will argue that the Hedonic
Maximin Principle and Multiple-Act Consequentialism jointly imply the
Proposed Code in the situation in which we ¬nd ourselves.
Let me expand the ¬rst three steps of this argument, those leading to
HMP, so that I can dispose of necessary background, so that I can explain
better what I mean by a direct argument, and so that I can properly
introduce the ¬nal doctrinal element of the book. Here is the expanded
form of the ¬rst three steps:
Premise One: Normative practice, by which I mean our practice of
ethical and political evaluation, has as a central and indefeasible commit-
ment something we might call “justi¬catory reason giving”.
Premise Two: Justi¬catory reason giving crucially involves

(a) normative claims that express justi¬catory reasons for or against things, which
(b) are governed by consistency and other logical constraints, and
(c) are capable of something we might call “deep justi¬cation”.

Premise Three: To give a deep justi¬cation of a normative claim is to
show that no con¬‚icting claim is appropriate, that there is an objective
asymmetry that vindicates a practice of reason giving deploying the ¬rst
against a practice deploying the second.

Premise Four: One form of the necessary objective asymmetry is an
objective normative fact to which a normative claim and not its possible
competitor is true. It requires a kind of normative realism. Another form
supports principles for constructing the value of wholes from the value of
their parts that can be vindicated in a properly objective and asymmetrical
fashion against possible competitors. But those are the only operative
Premise Five: There are certain sorts of objective normative facts, and
certain principles of construction that can be given the relevant deep
justi¬cation with properly objective asymmetry.
Premise Six: These facts and principles entail the Hedonic Maximin
Two points about this argument require immediate attention. First,
it invokes justi¬catory reason giving. The peculiarly normative feature of
normative discourse, of ethical and political discourse, is its capacity to
provide an articulation of reasons of a certain sort, and not merely reasons
as explanations such as we can hope to give of any natural phenomenon.
It is justi¬catory reason-giving utterance. It purports to give reasons that
justify or condemn, reasons genuinely for or against. Justi¬catory reason
giving is a deep enough commitment of our normative practice that
skepticism about its legitimate possibility can generate a kind of corrosive
skepticism about ethics and the normative in general. It is arguably the
most central and crucial feature of our normative practice. But whether
it is most central or not, it is surely central enough that at least many of
us would conclude, if this commitment of ethics cannot be vindicated,
that ethics is a kind of scam, just a lot of empty talk. This is in fact what
many people, including many philosophers, believe, and on something
recognizably like these grounds. We can see in this way that justi¬catory
reason giving is quite central to our practice of normative evaluation, that
its loss would be enormous and shattering. This fact underwrites the need
for our ¬rst argument, but also supplies its ¬rst premise. An important
and central thing about normative evaluation is that it purports to be
justi¬catory reason giving, the giving of genuine justi¬catory reasons for
or against things.
My second point is that it will turn out that legitimate justi¬catory
reason giving has a certain cost, which involves the ¬nal doctrinal element
of my proposal. This is a metaethical position that, like the normative
position it supports, is a con¬‚uence of traditional competitors.
Parts Two and Three will sketch an account of normative judgment
and language that adopts insights regarding normative discourse of both

cognitivist and noncognitivist authors, both authors who hold that normative
sentences have genuine truth conditions and those who deny this. And this
position can indeed claim to reconcile standard metaethical contenders in
a second way also, within cognitivism.
Some traditional cognitivists postulate non-natural normative proper-
ties, properties in some sense outside the natural order. Some other cog-
nitivists claim that normative features are constituted by natural properties
of a non-normative sort. But my proposal is that certain obviously natural
and quite concrete properties “ for instance, the property of being painful “
are in fact normative properties. My view is in that sense midway between
naturalist and non-naturalist forms of cognitivism. It has some af¬nity with
“sensibility” theories, like those of Wiggins and McDowell, which claim
that normative properties are analogous to allegedly “secondary” proper-
ties like color, and are constituted by our sensibilities. But my alternative
implies that normative properties are fully mind-independent and objec-
tive properties of objects. It is a form of full-blown normative realism.
The normative properties it deploys are much as our pre-Galilean ances-
tors conceived color to be, not as Boyle and Locke conceived color to be.
Only such normative properties can make proper sense of justi¬catory
reason giving; they will turn out to be necessary to that necessary end.
This component of my metaethical proposal is developed in Chapter 5.
But there is also a second component of this metaethical proposal,
developed in Chapter 6. This is a set of construction principles that allow
us to generate an ordering of wholes from better to worse given informa-
tion about the value contained in them, and that help provide HMP with
its distribution-sensitive form. These elements of my view are under-
written by features of our reason-giving normative discourse that some
noncognitivists have developed, and that also ¬nd expression in current
“practical-reasoning” and “constructivist” conceptions. So the position
to be developed here exhibits some af¬nities with current noncognitivist
projects, as well as cognitivist af¬nities. It is a con¬‚uence of familiar com-
petitors. While partisanship has greatly sharpened our understanding of
the theoretical alternatives in metaethics as well as in normative ethics, it
is implausible that opponents on these matters are simply and straightfor-
wardly mistaken.
My insistence on the need for a direct argument, for a transcendental
vindication of normative claims, may itself seem quite unreasonably parti-
san. If there is now a dominant methodological view in normative ethics, it
is that we should seek re¬‚ective equilibrium among our ¬rm intuitive eth-
ical judgments, without worrying much about any intuition-independent

vindication of those claims, or other metaethical and metaphysical niceties.
This methodology shapes much of the best and most interesting current
work in ethics. It seems to many contemporary ethicists that it really
doesn™t matter how or even whether there are robustly objective norma-
tive facts that underwrite our intuitive judgments, while I am insisting
on their signi¬cance. What™s more, it is hard to deny that we should have
more con¬dence in the particular commonsense moral judgments that
I aim ultimately to vindicate here than in the robust and controversial
metaphysical premises of my attempted vindication of those judgments.
But there are three reasons why I think that we need to pursue the
transcendental vindication of normative claims, why we need to provide
a direct argument for the truth of those claims, independent of appeal to
normative intuitions.
First, the meaning of our moral words invokes such an intuition-
transcendent basis, I believe. We need to see how the correctness of even
such normative judgments as we are all inclined to make is in fact delivered
by the world in which we live. There is inevitable controversy about the
nature of the world, but there must be some true story about what makes
our common intuitive normative judgments correct, or we are mistaken
in those judgments. While the metaphysical and metaethical details of
the particular story I tell are surely more controversial than the normative
claims that I rest on them, still the details are plausible, and would suitably
underwrite those normative claims. And if no such true story is available,
then those normative claims are in fact inappropriate, whether we like it
or not. It surely isn™t enough that the words of those claims echo vividly
through our hearts and minds. Some horrendous normative sentences
have echoed vividly through the hearts and minds of human beings.
Of course, my claim about the meaning of our moral words is con-
troversial, and some will think that this ¬rst reason should be granted no
weight at all. But my next two reasons are more ecumenical. The second
reason to pursue a direct argument for a moral theory is that there are
reasonable and rational people who fall outside of our normative con-
sensus, and we need something true to say to those people that should
convince them to join the consensus. Objective facts that root a true nor-
mative theory would provide it, as would other forms of transcendental
The third reason to pursue a direct argument is that even those of us
inside the relevant intuitive consensus still coherently differ about many
practically signi¬cant moral matters, and will continue to differ as we
re¬ne the coherent equilibrium of our individual normative intuitions.

We need some appropriate means to resolve these disputes that are beyond
our re¬‚ective consensus, and it seems that only transcendental vindication,
only direct argument independent of appeal to normative intuitions, can
provide it.
Depending on who “we” are “ depending on whether all minimally
intelligent and reasonable people in all times and places, or merely the
currently dominant community of academic normative ethicists, consti-
tute the relevant community “ these two reasons will vary in weight. But
they cannot be evaded at once without real cost. The broader the commu-
nity of normative intuition we seek, the more signi¬cant the normative
disputes it will leave unresolved. Of course, we may not be able to get
everything we want. Maybe we cannot resolve signi¬cant moral disputes.
But it is surely worth trying.
Some will accept none of these reasons even to attempt the transcen-
dental vindication of normative claims and theories. But I will also provide
two more arguments for my theory. The ¬rst argument for my proposal,
which I have just sketched, does not depend on appeals to ethical intu-
ition. But the second and third arguments do depend on appeals to ethical
intuition. As I said, I myself think that all these arguments are necessary.
Even if there is a transcendental vindication of some particular normative
theory, even if there is a proper normative argument from the point of
view of the universe, still ethics and ethical discourse might be bankrupt,
since the theory for which there is such a proper direct argument might
have morally abhorrent implications. Indeed, I think that in that situation
it would be only in a strained sense that there are genuine practical rea-
sons at all, even genuinely normative reasons of self-interest. But if you
disagree, you may still think that my second and third arguments, from
commonsense moral intuition, are important. Perhaps you think they
trump all other forms of argument, even a direct argument. Or perhaps
you think that they alone are suf¬cient in the absence of such a transcen-
dental vindication, or at least in the absence of a unique transcendental
vindication of one single moral theory. Or perhaps you think that they are
all that could matter in support of such a theory. And even if you think
these appeals to intuition should bear no weight, at the very least you
should admit that they are important in the eyes of many contemporary
My second argument focuses on the ¬rst three basic components of
my proposal “ its hedonism, maximin distribution over risks to bits of
experience, and Multiple-Act Consequentialism “ which together entail,
in our circumstances, its fourth component, the Proposed Code. This

argument appeals to commonsense moral intuitions of the same level of
generality and abstraction as those ¬rst three components. Whose com-
monsense intuitions? The suitably re¬‚ective intuitions of contemporary
academic ethicists, since they will provide a quite restrictive test, and since
they also generate the common intuitive objections we must consider.
I will argue that hedonism, maximin, and Multiple-Act Consequen-
tialism are each independently at least generally intuitive. That such an
argument is possible re¬‚ects the fact that my proposal is a con¬‚uence of
traditional competitors, that it is a legitimate heir of each of the chief
and apparently competing normative strands present within our complex
and not obviously coherent ethical tradition, the tradition inherited by all
contemporary moral philosophers. It also re¬‚ects the fact that this pro-
posal is a modi¬cation of classical utilitarianism to evade standard intuitive
My argument by appeal to these general intuitions has a positive aspect.
But it also involves discussion of what will seem in the current philosoph-
ical climate pressing general intuitive objections to such a view, and espe-
cially to its apparently controversial hedonism, its extreme distribution
sensitivity regarding momentary bits of lives, and its consequentialism. It
will also involve detailed discussion of the traditional intuitive objections
to more familiar forms of utilitarianism from which my proposal descends.
We will see with con¬dence that the three basic components of the
theory developed here are acceptably intuitive at the somewhat general
level in question, that my proposal is not known intuitively to be false
in that way. We will also see, though with less certainty, that these three
components are, despite current fashion, dominantly suggested by re¬‚ec-
tive intuition focused at the same level of abstraction, and hence that the
theory seems intuitively true at that level of detail.
Our discussion of the general intuitions that in fact support but may
seem to undercut hedonism will come in Chapter 4. The intuitive nature
of the maximin structure of HMP will be the concern of Chapter 7. The
mechanism of Multiple-Act Consequentialism, though founded in fact
in Chapter 2, will also ¬‚ow directly out of standard intuitive objections
to consequentialism, which are answered for MAC in Chapter 3.
It also matters, of course, how these parts ¬t together. The third argu-
ment for my proposal is that it yields properly intuitive detailed appli-
cations, that it yields properly intuitive judgments about particular cases.
This argument depends on the intuitive nature of the Proposed Code,
which is implied, in our situation, by the conjunction of Multiple-Act
Consequentialism and the Hedonic Maximin Principle. This code is

developed, in Chapter 8, with an eye to comparison with reconstruc-
tions of commonsense morality by Aquinas, Aristotle, Donagan, Fried,
Gert, Kant, Ross, Scanlon, Sidgwick, and four religious traditions, though
my dominant focus again will be contemporary academic intuition. We
will see that the Proposed Code can withstand the scrutiny of our re¬‚ec-
tive moral consensus. We will see that it can withstand the very speci¬c
objections in application that are the stock-in-trade of intuitionists. But,
more positively, we also will see that it provides a plausible and suggestive
framework for understanding the detailed content of commonsense moral
intuition, once our normative intuitions about cases are drawn together
and corrected by re¬‚ection for idiosyncracy, inconsistency, incoherence,
and vagueness.
Here™s a summary map: Part One develops Multiple-Act Consequen-
tialism. In particular, Chapter 2 argues that it is required by the facts,
independent of moral intuition, while Chapter 3 shows how it evades
familiar intuitive normative objections to act consequentialism.
Part Two concerns the hedonism of the Hedonic Maximin Principle.
Chapter 4 argues that hedonism is properly intuitive, while Chapter 5
attempts a direct and intuition-independent argument for hedonism.
Part Three concerns the maximin structure of HMP. Chapter 6 pur-
sues an intuition-independent vindication of that maximin structure, and
Chapter 7 argues that it is properly intuitive.
Part Four draws everything together. Chapter 8 develops the Proposed
Code as a joint implication of HMP and MAC, and shows that it is
properly intuitive in detail.

Part One
A Better Consequentialism
Multiple-Act Consequentialism

Act consequentialism “ the view that right acts are those individual acts
with best consequences available in the circumstances “ has an obvious
and intuitive rationale. To make the world as good as possible is a plausible
moral goal. But indirect forms of consequentialism promise more intuitive
normative implications, though at evident cost of intuitive rationale. This
chapter will introduce a new form of consequentialism, Multiple-Act
Consequentialism or MAC, which combines the intuitive rationale of
act consequentialism and the intuitive normative implications of the best
indirect forms.
MAC has four key tenets: (1) There are group agents of which we
are constituents. (2) Direct consequentialist evaluation of the options of
group agents is appropriate. (3) Sometimes we should follow our roles
in a group act even at the cost of the overall good we could achieve by
defection from those roles. In particular, one should defect from a group
act with good consequences only if one can achieve better consequences
by the defecting act alone than the entire group act achieves. (4) When
different bene¬cent group agents of which one is part specify roles that
con¬‚ict for one, one should follow the role in the group act with more
valuable consequences.
MAC is a natural response to three standard objections to familiar forms
of act consequentialism. Section I sketches these three objections, the
indirect consequentialism that is the standard consequentialist response,
and standard objections to indirect consequentialism. We need another
approach. The rest of the chapter develops MAC. It evades the inherent
dif¬culties of indirect forms largely because it is a new direct form. The
next chapter will apply MAC in response to the three standard objections
to familiar direct forms. But we will see in this chapter that if a basic

consequentialist moral principle is true “ for instance, the Hedonic Max-
imin Principle or a standard utilitarian valuation principle “ then MAC is
true independent of normative intuitions like those that root these three


Consider three standard objections to act consequentialism. First, act
consequentialism seems to be “an excessively demanding moral theory,
. . . [which] require[s] that one neglect or abandon one™s own pursuits
whenever one could produce even slightly more good in some other
way.”1 But it seems morally permissible to spend money on an old sail-
boat, or to buy your daughter a nice toy knight, even when that money
might be put to better use alleviating the suffering of the starving.
Second is the standard deontological objection that act consequential-
ism seems insuf¬ciently demanding. It doesn™t forbid “ indeed, it some-
times requires “ lying, or injuring or even murdering the innocent, when
those things will generate an overall increase in the good. Some of the
force of this objection is directed solely against act consequentialisms that
deploy a traditional utilitarian value theory. We might evade this element
of the objection by adopting a theory of value that ceded weight to death,
or even to murder per se. But there is another element of the objection
that cannot be evaded by this ¬x, a ¬x that in any case seems in question-
able accord with some traditional motivations for consequentialism, and
won™t ¬t HMP. That second element is that it seems wrong to murder
even to prevent several other murders. Even this problem can be evaded
by a modi¬cation of basic consequentialist normative principles in which
the value of states of affairs is relativized to particular agents. The world is
worse for me if I murder, but worse for you if you murder. But this may
seem a kind of trick that evades the letter of the objection in question only
by abandoning the spirit of act consequentialism. And it requires a theory
of value that cannot be independently motivated. Consequentialists need
another response.
The third standard objection to act consequentialism is that it is in some
ways at once too permissive and too demanding. It directs that we ignore
our individual special obligations “ our obligations to our own children or
friends and our obligations of gratitude and reparation “ and pursue the

1 Schef¬‚er (1988: 3).

good of all indiscriminately. Or at least it does that when it is allied with
a characteristic nonrelative value theory.
There is a standard consequentialist reply to all these standard objec-
tions. It is indirect consequentialism. It does not succeed, but why it fails
is important.
Act consequentialism prescribes that each individual agent in each sit-
uation act in direct accord with the proper ordering of options from worst
to best “ indeed, in the simplest form of act consequentialism, that each
always choose the best.2 A basic normative principle like HMP or a classi-
cal utilitarian principle is applied directly to evaluate individual acts from
among individual options. But consequentialism might alternatively assess
things other than individual acts by reference to valuable consequences,
and then assess individual acts indirectly, by reference to those other things.
This may evade the three intuitive objections to direct maximization of
the good.
There are a variety of possible forms of indirect consequentialism. Indi-
viduals have relatively stable motives and characters, which constrain indi-
vidual acts over relatively long stretches of time and across many situations
of choice. So perhaps we are to assess what motives are best on conse-
quentialist grounds, and allow best motives to determine proper actions.3
Or perhaps we are to focus on the best relatively stable characters.4 Other
forms of indirect consequentialism focus on the actions of many distinct
individuals at once. We might focus on relatively universal acceptance of
sets of moral rules, and claim that an act is morally permissible if and only
if it is allowed by rules whose acceptance by the overwhelming major-
ity of everyone everywhere could reasonably be expected to result in as
good consequences as would result from any other code identi¬able at the
time.5 Alternatively, we might focus on ideal rules for particular societies
in particular local conditions. Or we might focus on act-types, and adopt
consequentialist generalization, claiming that “an act is right if and only if
the consequences of its being performed by the agent and all other agents
similarly situated are at least as good as the consequences of any other

2 There is a distinction between subjective and objective consequentialist theories, which in
their simplest direct forms suggest respectively that the right act is that which is best according
to the basic normative principle applied upon our conception of our options or the fact of
our options. I presume the latter.
3 Adams (1976).
4 Railton (1984). Railton does not endorse an indirect form of consequentialism.
5 Hooker (1995: 20). See also the somewhat different formulation in Hooker (1996) and the
most recent version in Hooker (2000: 32).

available act™s being performed by the agent and all other agents similarly
situated.”6 Donald Regan has proposed yet another sort of indirect alter-
native, cooperative consequentialism.7 On this view, each agent ought to
cooperate, with whoever else is thus cooperating, in the production of
the best consequences possible given the behavior of noncooperators.8
We focus on the group present in any particular situation that is willing to
cooperate in pursuit of the proper consequentialist end in that situation,
and act in effective concert with that group.
While indirect forms of consequentialism promise a way around the
three standard objections to act consequentialism, they in turn face three
crucial objections. Not all of the objections succeed, but their nature and
the nature of the replies that they necessitate are important to us.
One classic objection to indirect consequentialism is that the vari-
ous indirect forms “ for instance, rule consequentialism “ have the same
implications as, are extensionally equivalent to, act consequentialism. This
would imply that those indirect forms are no more properly intuitive than
the act consequentialism on which they are supposed to be an intuitive
improvement.9 The argument for extensional equivalence seems straight-
forward. The best and most bene¬cent rules seem of necessity to allow
for exceptions to any general restrictions on behavior that they proffer,
exceptions that allow the local maximization of the good.
But in fact this objection is mistaken. None of the standard forms of
indirect consequentialism is extensionally equivalent to standard act conse-
quentialism. There are perhaps many reasons for this, but one is important
to us. One way in which extensional equivalence fails is instructive, since
it suggests that the intuitive normative advantages of indirect forms result
from the effects of cooperative behaviors.
This key objection to extensional equivalence is Allan Gibbard™s.10
Here is his case:

Smith and Jones . . . [are] placed in separate isolation booths, so that the actions
of one can have no in¬‚uence at all on the actions of the other. . . . [A] red push-
button [is] installed in each . . . booth. The only action of moral signi¬cance open
to either man will be to hold his push-button down at 10:00 a.m., or to refrain

6 Regan (1980: 94). But perhaps the classic statement is Harrod (1936). Murphy (2000)
suggests a new and somewhat analogous form to which we will return.
7 Actually, Regan calls it “co-operative utilitarianism”.
8 Regan (1980: 124).
9 For classic forms of this argument, see Lyons (1965) and Brandt (1963).
10 Gibbard (1965).

from doing so. . . . If at 10:00 a.m. both are holding down their push-buttons, they
receive cake and ice cream. . . . If only one of them is holding his push-button
down, however, they both receive electric shocks. . . . If neither is holding his
button down, nothing happens.11

This is how the case works: Notice that if at 10 a.m. Smith is not
holding his button down, then it is best for Jones not to hold his down.
But then the best act consequentialist act for Jones is not that which would
be best for each to engage in, nor that which would be best for rules to
prescribe to both, nor that which would be best for both to do in similar
circumstances.12 The coordinating effects of joint action block the alleged
extensional equivalence of act consequentialism and indirect forms.
Of course, some indirect forms of consequentialism “ for instance,
character consequentialism “ focus not on the acts of more than one agent
but on the acts of a single individual over many temporally distinct choice
situations. But notice that the phenomenon Gibbard notes has a temporal
analogue. It is simply Casta˜ eda™s “paradox” of act consequentialism seen
from a different angle: If among your options are conjunctive acts “
for instance, acts that take some time and require a temporal conjunction
of two shorter acts “ it may be that the conjunctive act has the best
consequences of all temporally extended acts available, but that neither
conjunct has good consequences on its own. The conjunctive act may be
in effect a cooperative action of two periods of your life.
So the ¬rst objection to indirect forms of consequentialism fails. But it
fails in an instructive way. Almost everyone grants that indirect forms have
more intuitive normative implications as long as they are not extension-
ally equivalent to act consequentialism. But Gibbard™s counterexample to
extensional equivalence suggests that it is in particular the coordination
of action in cooperative behavior that generates those intuitive advantages
for indirect forms. This perhaps re¬‚ects the traditional Kantian and con-
tractarian insight that cooperative activity and the respect and reciprocity
that support it undergird norms that forbid lies, murder, and injury.
Still, there is an important complication. In Gibbard™s counterexam-
ple, one of the two individuals does not perform their component of the
joint act that would have best consequences. If they had performed their
component, act consequentialism and indirect forms would require the

11 Ibid., 214.
12 There is a similar example developed in Gibbard (1990a: 6“7), a reprint of his 1971 disser-
13 Casta˜ eda (1973). For extensive discussion, see Feldman (1975, 1986).

same act of the second individual. Act consequentialism and indirect forms
are apparently extensionally equivalent in their recommendations for that
second individual in that situation. This is relevant because we will even-
tually see that the best form of consequentialism cedes normative salience
to actual forms of cooperative behavior on the part of others. In other
words, in the important cases of extensional equivalence or inequivalence
for the view to be developed here, the relevant analog of Gibbard™s ¬rst
individual does his part in the cooperative scheme, while in Gibbard™s case
he does not.
But there are also other sorts of counterexamples, which re¬‚ect other
aspects of cooperative activity, and which show a failure of extensional
equivalence even in the cases most relevant to the view to be developed
here. In the familiar Prisoner™s Dilemma, there are two individuals who
are so positioned that, if each acts directly to pursue their own self-interest,
then each will do better in that regard whatever the other does, and yet
both will lose relative to an outcome that was available by their joint action.
Act sel¬shness and more indirect forms of sel¬shness are not extension-
ally equivalent even when the other prisoner in fact acts in a cooperative
manner. This has the structure of the case we need. And there are also
moralized versions of the Prisoner™s Dilemma, in which two consequen-
tialists are positioned in such a way that, if each acts directly to pursue
best consequences, then each will do better whatever the other does, and
yet both will lose relative to an outcome that was available by joint action.
It might seem that this could not be, since the two consequentialists,
unlike the two sel¬sh prisoners, share a goal. But it can be “ for instance,
because the options available to someone can depend on how they choose
among their options. If someone chooses among options as an individual
act consequentialist, a wizard may torture all humans. If they choose as a
deontologist or otherwise, the wizard may promote the general welfare of
all. Hence all the options of our consequentialist might be worse on con-
sequentialist grounds if they choose as an act consequentialist rather than
otherwise. Such a wizard can also assure the characteristic payoff matrix
of a moralized Prisoner™s Dilemma.14 Note that my point isn™t merely
that there can be indirect negative consequentialist effects that result from
self-consciously deploying consequentialist decision procedures or having
consequentialist motives. In the case at hand, all the options open to an
agent will in fact be much worse according to consequentialism if the

14 This case is a slight modi¬cation of a case presented in Mendola (1986).

agent in fact acts in accord with consequentialism, whatever their motives
or decision procedures. If they take the best option available to them,
then all their options will be much worse than they otherwise would
have been. Of course, there are no wizards with nasty schemes like that
one. But a more realistic counterexample is available whenever accep-
tance of certain traditional constraints against lying and murder makes
better options available than would be available in a world of individual
act consequentialists. The consequent reduction in anxiety about the pos-
sibility of being murdered or injured, even if only for the common good,
would alone sometimes be suf¬cient. And truth telling allows a kind of
cooperative planning in the face of our ordinary ignorance of what others
believe and will do under various conditions that is crucial to success-
ful forms of cooperative behavior. And of course in the real world not
everyone is an act consequentialist. Others may refuse to admit someone
who is an act consequentialist into groups with bene¬cent goals, or may
not be able to suf¬ciently trust an act consequentialist to allow for the
successful cooperative pursuit of some important consequentialist goal.
Effective joint action seems also sometimes to require that some of the
cooperators abandon independent pursuit of the goal, even where trust
is not in question.15
Gibbard introduced another class of counterexamples to extensional
equivalence of the sort we need, which indeed introduce other phe-
nomena that are crucial for MAC. Gibbard calls these cases of “surplus
cooperation”.16 Imagine that we all cooperate in a very bene¬cial prac-
tice of truth telling. It would be very bad if many of us often lied, which
would undercut the practice. But given the fact that almost no one ever
lies, it may be bene¬cial for me to tell a few lies. I won™t tell enough
lies to undercut the practice with its signi¬cant general bene¬ts, but I
will grab a little extra utility on the side. There are some indirect forms
of consequentialism that require that I not lie in this circumstance, while
familiar forms of act consequentialism suggest that I should. What are

15 Gibbard (1990a: 157“237) argues that in a society of act utilitarians an established convention
of agreement keeping might under certain conditions create a coordination point suf¬cient
to allow a practice of agreement keeping among act utilitarians, and that such a convention
could be established among such utilitarians by a teaching practice of a certain sort described
there. He also argues more generally that in an act utilitarian society, “the agreements
that would be kept include almost all the agreements to which act-utilitarians want to
bind themselves” (159). But note the quali¬cations on pp. 201“205. He suggests that the
argument extends to truth-telling on pp. 158“159, but does not explain the extension. Still
this does not undercut the extensional inequivalences noted in the text.
16 Ibid., 20.

probably the most intuitive indirect forms of consequentialism fall within
the relevant space of this counterexample. In the instances most relevant
to our eventual concerns, act consequentialism and indirect forms are not
extensionally equivalent for this reason also.
So the ¬rst objection to indirect consequentialism fails in an instructive
way. Indirect forms of consequentialism are not extensionally equivalent
to standard forms of act consequentialism, because of the effects of coop-
eration. Therein lie the intuitive advantages of the indirect forms.
Let™s turn now to the second standard objection to indirect forms of
consequentialism. That second objection will play a rather speci¬c role
here, since it can be evaded by nonstandard indirect forms. It applies to
the standard and paradigmatic forms of indirect consequentialism, and
reveals a characteristic dif¬culty with these forms that suggests that they
should be developed in a particular direction.
For instance, both standard rule consequentialism and consequentialist
generalization imply that one should do what would generate best con-
sequences if at least most people did it, even though in fact most people
aren™t going to do it. Since in fact they aren™t, this may lead one to neglect
great goods that one could in fact actually achieve and great harms that
one will in fact cause by acting in accord with the ideal rules.17 Fafner
the dragon is asleep on his ingots over there, and we can get them if we
act together, which is a great good. But it would be foolish for you to do
your part alone.
Cooperative consequentialism does somewhat better in this regard. It
realistically focuses on the actual group of consequentialist cooperators.
But18 there are still structural problems even with cooperative conse-
quentialism. In many morally problematic situations, there may not be
any other proper consequentialists around with whom to cooperate. Cer-
tainly many humanly important forms of cooperation include individuals
who are motivated by other than explicit consequentialist concerns.
This suggests that the best form of indirect consequentialism would
focus on all forms of actual but normatively positive cooperation, and
not just on cooperating consequentialists. It is relevant that only some
people in fact cooperate, but it is also relevant that there are norma-
tively signi¬cant forms of cooperation that do not rest on any shared

17 There are possible rule consequentialist responses. See, for instance, Hooker (2000: 98“99).
18 At least if I have understood Regan (1980) properly. See, for instance, the key argument on
page 138. But in any case, the indirect consequentialism that I vaguely develop is a variant
of cooperative utilitarianism in Regan™s sense.

and explicitly consequentialist understanding. Some cooperation is clearly
toward evil ends, and should bear little normative weight. But still, coop-
erative activity can be in accord with basic consequentialist strictures “
productive of the good and even more productive than any alternative
form of cooperation “ even if consequentialist motives aren™t explicit in
the cooperators.
So the second objection has pushed us in a certain direction, toward
a speci¬c form of indirect consequentialism. According to such a view,
actual cooperative activity toward good consequences has a kind of nor-
mative weight that may properly trump individual maximization of the
good. Of course, that is a vague conception. For instance, it doesn™t tell us
when that cooperative activity trumps individual maximization. But it is
a rough start. And it ¬ts nicely with our reply to the ¬rst objection, which
suggested that the intuitive implications of indirect consequentialism stem
from cooperation.
Still, there is a third objection to all indirect forms of consequential-
ism, and I think that it succeeds. Indirect forms lack a coherent normative
rationale. For instance, if maximization of the good is the ultimate ratio-
nale of rule consequentialism, then it is ad hoc or incoherent to suggest
at the same time that individual acts that maximize the good are yet
sometimes wrong because they violate rules whose merely hypotheti-
cal general acceptance would have good consequences. Even forms of
indirect consequentialism that focus on actual cooperation, such as coop-
erative consequentialism or the vague improvement I just suggested, seem
to face a similar incoherence. It is a fact that consequences are sometimes
better served if an individual consequentialist defects from a cooperative
consequentialist scheme and leaves its success for others to assure. And
an indirect form of consequentialism that assesses such a defecting act
only by its ¬t with that best cooperative scheme seems to lack a coherent
consequentialist rationale.
It may seem that there can be no middle ground. Nevertheless, there is
a form of consequentialism that enjoys at once the intuitive implications
of indirect forms “ indeed, that closely tracks the intuitively ideal form
of indirect consequentialism suggested by my vague improvement of co-
operative consequentialism “ and that also has a coherent consequentialist
rationale. It can hence evade all the objections to indirect forms.
That form is Multiple-Act Consequentialism, or MAC. It will be
developed in the next two sections. The key point is that much coop-
erative activity is in fact group action. And so the ideal indirect form of
consequentialism is extensionally equivalent to an unusual direct form.

The normative advantages of the best indirect forms of consequential-
ism turn on the effects of actual cooperative activity. But many forms of
cooperative activity are group acts. Indeed, we will see that the forms of
cooperative activity that are normatively signi¬cant in the most crucial
ways are group acts. And group acts, like individual acts, can have a direct
consequentialist rationale.
MAC is a direct form of consequentialism that focuses on all acts,
including group acts. A focus on something like group acts has been pro-
posed before, by Postow, Jackson, and T¨ nnsj¨ .19 But MAC™s new and
a o
most characteristic feature is that it propounds a speci¬c way to balance
these sometimes con¬‚icting forms of agency. It will take us a little while
to understand the rationale and detail of this novelty that principally dis-
tinguishes MAC from familiar forms of act consequentialism. But its gen-
eral structural advantages over indirect forms are already evident. Group
actions can be conjunctions of individual acts, and hence can approximate
the cooperative activities invoked by the ideal but somewhat vague form
of indirect consequentialism previously sketched. And yet group action in
accord with a consequentialist principle can have a direct consequentialist
Some charge that indirect forms of consequentialism, which are appar-
ently intuitive in their normative implications, in fact collapse into unin-

. 1
( 10)