00001.tif

PDF

Download

Open document

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may
be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedtbrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly
to order.
A Bell & Howell information Company
300 North Zeeb R oad. Ann Arbor. M l 48106-1346 USA
313/761-4700 800/521-0600

HOW ETHICAL INTERNALISM AFFECTS THE SCOPE OF MORAL CLAIMS
by
Deborah Jill Waldman
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Philosophy)
May 1995
Copyright 1995 Deborah Jill Waldman
UMI Number: 9621637
UMI Microform 9621637
Copyright 1996, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
....D e b o r a h .. _ J i l l . _ W a 1 d m a n ..................
under the direction of h.ex Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
Date .J£bn w x..l^ ..l9.95...
DISSERTATION COMMITTEE IERTATION COMMr
Chairperson
D eborah Jill W aldm an Barbara H erm an
S tu d en t's N am e C o m m ittee C h air's N am e
H ow Ethical Internalism Affects the Scope o f M oral Claim s
In its m ost distinctive form ulation, ethical internalism is defined by the claim that part o f w hat
it m eans to be b o u n d by a m oral obligation is that one has sufficient m otive for acting so as to fulfill
it. It is a response to some o f o u r m ost firm ly held com m onsense m oral intuitions; including, for
exam ple, that there is som ething odd about som eone asking for a reason to do w h at she know s she
m orally ought to do, and that being bound by a moral obligation entails som e constraint on how one
m ay choose to act. Internalism accords so well w ith these intuitions th at it is w idely held not only to
be an attractive view about th e relationship betw een obligation and m otivation, b u t to be a
conceptual tru th -d en iab le on pain o f incoherence or m oral skepticism .
Ethical internalism is m ost often criticized for requiring that one deny com m onsense m oral
intuitions that are at least as firm ly held as those that m otivate internalism . In particular, it is
charged that claim ing that obligation and m otivation are necessarily linked com m its one to som e
form o f moral relativism or skepticism .
Despite th e w ide awareness o f these apparently devastating criticism s, ethical internalism
continues to be considered an attractive position about the relationship between obligation and
m otivation, even by those w ho will not entertain th e possibility o f m oral relativism. A t the same
tim e, the charge against internalism has seldom received m ore than cursory attention from either
internalists or externalists. G iven that the charge is generally taken as obvious and devastating, the
absence is puzzling. It is this observation th at serves as the starting p o in t for this dissertation. T h e
aim o f this dissertation is to determ ine how it is possible to accept internalism and to sustain a n ti-
relativistic moral intuitions, given the radical challenge internalism is com m only supposed to present
to o u r intuitions ab o u t the scope o f moral constraints.
C o m m itte e Chair s Signature Date
Acknowledgmenta
Before all else, I must acknowledge my gratitude to Barbara
Herman. Her patient and generous understanding and her
encouragement have been essential to my efforts from the earliest
stages of this project, and to my philosophical development
generally. In addition, her insight has stimulated my own, and
provided it with guidance so gently and successfully that I have
often failed to notice as I have profited from it.
I am indebted also to Sharon Lloyd for her careful reading
and helpful comments on various parts and versions of this
dissertation.
The opportunity to do research at Hoose Library--which in my
experience is unrivaled among philosophy libraries--has
contributed inestimably to this and other projects. A Merit
Fellowship from the University of Southern California allowed me
to devote myself to study during my first three years in graduate
school. The department of philosophy at USC provided a much
needed web of social and intellectual relationships and resources.
I am, of course, greatly in debt to my friends and family for
making everything possible. I am especially grateful to Mark
Daley, Tom Mrowka, and Paul Romanelli for our friendship and
communion, and for providing the encouragements and
discouragements that are essential to living well.
Table of Contents
Chapter One: The Benefits and Burdens of Ethical Internalism 1
Chapter Two: Gilbert Harman's Argument for a Link between
Internalism and Moral Relativism 31
Chapter Three: How the Truth of Internalistic Judgments
Depends on Their Subjects 60
Chapter Four: The Relationship of Moral Relativism to
Internalistic Moral Theories 111
Chapter Five: Strategies Internalistic Moral Theories Might
Use to Avoid Moral Relativism 142
Chapter Six: Can an Internalistic Moral Theory Be Plausible? 193
Chapter Seven: A Related Challenge to Traditional
Internalistic Moral Theories 212
Works Cited 237
iii
Chaptar Ones
The Benefits and Burdens of Ethical Internalism
In its most distinctive formulation, ethical internalism is
defined by the claim that part of what it means to be bound by a
moral obligation is that one has sufficient motive for acting so
as to fulfill it.1 It is a response to some of our most firmly
held commonsense moral intuitions; including, for example, that
there is something odd about someone asking for a reason to do
what she knows she morally ought to do, and that being bound by a
moral obligation entails some constraint on how one may choose to
act. Internalism accords so well with these intuitions that it is
widely held not only to be an attractive view about the
relationship between obligation and motivation, but to be a
conceptual truth--deniable on pain of incoherence or moral
skepticism.
Ethical internalism is most often criticized for requiring
that one deny commonsense moral intuitions that are at least as
firmly held as those that motivate internalism. In particular, it
is charged that claiming that obligation and motivation are
necessarily linked commits one to some form of moral relativism or
skepticism. William Frankena characterizes this charge somewhat
more subtly as follows:
■'•There are actually a number of competing, but similar views that are
sometimes called "ethical internalism". Much of this chapter will be
devoted to clarifying the assumptions that define ethical internalism
1
[I]n the end the internalist must argue, as Falk
does, not only that externalism involves a gap
between obligation and motivation, but that such
a gap cannot be tolerated, given morality's task
of guiding human conduct autonomously. Then,
however, the externalist will counter by pointing
out that internalism also entails a danger to
morality. Externalism, he will say, in seeking
to keep the obligation to act in certain ways
independent of the vagaries of individual
motivation, runs the risk that motivation may not
always be present, let alone adequate, but
internalism, in insisting on building in
motivation, runs the corresponding risk of having
to trim obligation to the size of individual
motives.2
Despite the wide awareness of these apparently devastating
criticisms, ethical internalism continues to be considered an
attractive position about the relationship between obligation and
motivation, even by those who will not entertain the possibility
of moral relativism.3 At the same time, the charge against
internalism has seldom received more than cursory attention from
2"Obligation and Motivation in Recent Moral Philosophy," in Essays in
Moral Philosophy, A. I. Melden, ed. (Seattle: University of Washington
Press, 1958), pp. 79-80.
^Indeed, historically, the currency of ethical internalism seems to be
matched by the currency of such criticisms. Note, for example, that
Hume, whose advocation of a close association between motive and
obligation is explicit, has historically been criticized for his
implicit moral skepticism (and that Hume seems to have taken pains to
deny this) . In the nineteenth century, Sidgwick articulates this line
of criticism quite explicitly, apparently precisely because he takes the
identification of moral obligation with the existence of motive to be
the starting place for moral theorizing. At the same time, Sidgwick's
treatment of the criticism is so quick that it suggests that even in the
nineteenth century, the link between moral skepticism and the
identification of moral obligation with the existence of motive was
considered too obvious to require detailed argument. (See, for example,
The Methods of Ethics (Indianapolis: Hackett Publishing Company, 1981),
p. 12. A more explicit discussion can be found in "Some Fundamental
Ethical Controversies," Mind 14, no. 56 (1889), p. 485.)
It is also worth noticing that this criticism of internalism has
continued to be widely accepted. A recent articulation of it can be
found in Gilbert Harman's "Moral Relativism Defended", for example.
(Philosophical Review 84 (1975), pp. 3-22.)
2
either internalists or externalists. Given that the charge is
generally taken as obvious and devastating, the absence is
puzzling. It is this observation that serves as the starting
point for this project. What I want to consider in this project
is how it is possible to accept internalism and to sustain anti-
relativistic moral intuitions, given the radical challenge
internalism is commonly supposed to present to our intuitions
about the scope of moral constraints.
With few exceptions, internalists do in fact attempt to deny
moral relativism— and with good reason. If internalism is
attractive, it is because it allows us to maintain that some of
our most firmly held moral intuitions are true. It is thus quite
natural that internalists would also be concerned to deny moral
relativism, because moral relativism is so plainly at odds with
many of our most firmly held moral intuitions--including, for
example, that all human agents are bound to conform to morality,
and that moral judgments are objective. The motivation for
accepting internalism is, thus, closely allied with the motivation
for denying moral relativism; each is attractive precisely because
it allows us to accept our most firmly held moral intuitions as
veracious.
One consequence of this alliance is that internalist moral
theories tend to deny moral relativism from the outset. The
charge against internalism is thus forestalled, but in a very
3
unsatisfying way, because the charge is not addressed directly.4
When one looks at such a theory from the point of view of a
critic, it is hard to see why one should think it a good account
of our firmly held moral intuitions rather than simply incoherent
(in its conjunction of internalism with a denial of moral
relativism).
Internalists' explicit denial of moral relativism is not only
unhelpful in assessing the charge that they are committed to it,
but also makes assessing the charge more complicated. Internalist
moral theories that are constructed so as to support our anti-
relativistic intuitions may already compensate for any
relativistic consequences that supporting internalism might have.
In order to understand how internalist moral theories can be
plausible in light of the common criticism that internalism
imposes implausible limitations on the scope of moral constraints,
it will therefore be necessary to uncover which features of them
are required by or compensatory for their supporting internalism.
Because of these difficulties, I will try to assess the major
line of criticism against internalist moral theories by stripping
away the features of internalist moral theories that are not
immediately necessary to their stands on the relationship between
obligation and motivation. My strategy will be to focus on the
charge, and try to characterize it as a claim about the structure
4This seems to be why an argument as straightforward as Harman's can
still be devastating--what is there to prevent it?
of internalist moral theories. At the same time, I will develop a
spare, almost minimal model of an internalist moral theory. As I
elaborate the charge and the model, sketches of full-fledged moral
theories that follow the model emerge. The impact of their stands
on the relationship between obligation and motivation on their
structure and content will thus be made explicit, and the content
of the charge against internalist moral theories that they entail
moral relativism or moral skepticism clarified. This will enable
us to assess the prospects for plausible internalist moral
theories.
To pursue this project we will need to isolate the defining
assumptions of internalism. To this end, I will first challenge
the dichotomy between internalism and externalism as Frankena
describes it. In this chapter, I will argue that there is an
intuition more basic than the one on which Frankena's dichotomy is
built that warrants directing our attention to a wider category of
theories than those commonly considered to be internalist
theories. Then, following the strategy I have outlined, I will
consider the burdens Frankena and others assign to internalism to
determine whether a moral theory can respond to the intuitions
underlying internalism and remain plausible.
Is Ethical Internalism Reasonable?
In its traditional formulation, ethical internalism begins
with the assumption that "somehow the very fact of a duty entails
5
all the motive required for doing the act."5 Internalism is not
merely a position about whether moral theories need to be
concerned about motivation. Ethical externalists, who deny that
there is any necessary link between moral obligation and
motivation, may join internalists in taking some claims about
motivation as fundamental. One could, for instance, be either an
externalist or an internalist and assume that conscientious agents
can act morally because they are conscientious; so might one be
either and value moral theories according to how well they account
for that assumption. At issue between internalists and
externalists is whether motivation is “built into" moral
obligations,6 not whether motivation is to be taken care of in
some way or other. Internalists claim that being bound by a moral
obligation entails that one has sufficient motive for acting so as
to fulfill it.
Ethical internalism, as I have characterized it, thus
involves a very strong claim. Even if one is sympathetic to the
idea that obligation and motivation are linked, one might wonder
why would we want to make such a sweeping assumption. Why not
instead think that being bound by a moral obligation entails only
that one has some reason for acting so as to fulfill it? Or that
5w. D. Falk, "'Ought' and Motivation," in Ought, Reasons, and Morality:
the Collected Papers of W. D. Falk (Ithaca, New York: Cornell
University Press, 1986), p. 29.
6This way of putting it is from "Obligation and Motivation in Recent
Moral Philosophy," p. 41. Frankena attributes this way of putting the
point to C. L. Stevenson.
6
being bound by a moral obligation entails that if one is well-
disposed toward moral considerations, one has sufficient motive
for acting so as to fulfill it?
Internalism begins with the stronger claim because of its
aims. Internalism is supposed to enable us to account for certain
of our moral intuitions that deny that a gap between obligation
and motivation can be tolerated.7 The particular intuitions
usually summoned on behalf of internalism can be characterized,
more or less, by three sorts of claims: (i) that the question “Why
should I act morally?" is sometimes a philosophically interesting
question; (ii) that moral obligations imply real constraints on
our ability to act contrary to them; and (iii) that paradigmatic
moral behavior is autonomous. The same aims seem to motivate the
weaker positions suggested as more plausible alternatives. The
weaker positions cannot sustain them, however.
If we accept internalism, that is, if we claim that
obligation entails sufficient motive, then we assume that an agent
bound by a moral obligation can find acting to fulfill it
choiceworthy. The internalist will consequently find the question
"Why should I act morally?" problematic. If an agent asks this
question, then it is prima facie evidence that the agent is not
bound by the moral obligation in question. It is open to the
internalist to argue that the agent's question is meaningless,
7Frankena, "Obligation and Motivation in Recent Moral Philosophy, ” p.
79.
7
unanswerable, misleading, etc.; but in any case the internalist
must consider the case apparently paradoxical because we do not
seem to excuse people from their moral obligations on the grounds
that they do not find them compelling.
The externalist, who denies that there is a necessary link
between obligation and motivation, cannot consider any agent's
asking "Why should I act morally?" philosophically interesting.
The presence or absence of a motive for acting morally is for the
externalist merely a contingent psychological fact about an agent.
If we assume that being bound by a moral obligation entails
that one has a reason for acting so as to fulfill it, but not that
one has sufficient motive for so acting, then our account of an
agent's asking "Why should I act morally?" will be similar to the
externalist's. It is true that we would have to say that the
agent has a reason for acting morally. However, since reasons are
not always compelling, there would be nothing odd about the
agent's apparently not wanting to acting morally. One might think
that this account will have more to say about the case, because it
may be difficult to show the agent that she has a reason. Even if
it is difficult, however, whether the difficulty reveals a
philosophical or merely practical problem seems to depend on our
stance about internalism. Given that we are assuming that being
bound by an obligation entails that an agent has a reason for
acting to fulfill it, but not sufficient motive for doing so, we
8
must also assume that an agent may have a reason for acting in a
certain manner even if she cannot be motivated to do so. On this
account, if it is difficult or impossible to show an agent that
she has a reason, it does not necessarily mean that she does not
have a reason. Moreover, if we reject internalism, then the fact
that it is sometimes difficult to show the agent who asks, "Why
should I act morally?" that she has a reason to act morally is no
more interesting than it would be in any other case in which an
agent fails to recognize that she has a reason. The case presents
no special problems.
The other alternative account fares no better. If we claim
that being bound by a moral obligation entails only that if one is
well-disposed toward moral considerations, one has sufficient
motive for acting to fulfill it, then our account of "Why should I
act morally?" will also be like the externalist's. An agent who
asks is presumably not well-disposed toward moral considerations.
The case seems to present no philosophical problem.
Defending the second sort of intuition, that moral
obligations imply real constraints on our ability to act contrary
to them, also seems to require us to accept internalism. If
obligation entails sufficient motive, then when one is bound by a
moral obligation, there are conditions under which one will act so
as to fulfill it. Neither of the weaker alternatives suggests any
such constraints. Claiming that one has a reason for acting to
9
fulfill one's moral obligations in a sense that does not imply
that one has sufficient motive does not have any implications for
whether one may be motivated to so act. Claiming that agents who
are well-disposed toward moral considerations have sufficient
motive for acting to fulfill their moral obligations does not have
any implications for how other agents are capable of behaving.
The third sort of intuition, those that roughly amount to
some version of the claim that paradigmatic moral behavior is
autonomous, less clearly favor internalism. Although these
intuitions are usually characterized vaguely, and seem to resist
articulation, the underlying idea seems to be that H[t]he feature
which distinguishes moral obligations from all others is that they
are self-imposed."8 Mature moral agents are supposed to be
capable of recognizing and choosing what is morally obligatory
without nonmoral sanctions or external influences. Certainly
assuming externalism, i.e., denying that there is any necessary
link between obligation and motivation, provides no means for
supporting such intuitions.9 On the surface, internalism does not
seem to support these intuitions either. However, a particular
internalist view presumably will provide support for these
intuitions because it will have to give an account of how it is
8P. H. Nowell-Smith, Ethics (Baltimore: Penguin Books Inc., 1954), p.
210 .
9This is not to say that these intuitions are incompatible with
externalism. As Frankena suggests, there may be ways for an externalist
to defend these intuitions.
10
that all agents have sufficient motive for fulfilling their moral
obligations. Given the differences in agents' external
circumstances and the similarities of the moral obligations by
which they are bound, such accounts usually rest the existence of
sufficient motive for acting to fulfill their moral obligations on
the subjective accessibility of the authority of moral
obligations.
The first proposed alternative to internalism, to assume only
that being bound by an obligation entails that one has a reason
for acting so as to fulfill it, may be used in defending some of
these intuitions in much the same way. Proponents of such
accounts may explain the fact that agents have reasons for acting
as morality requires by the subjective accessibility of the
reason-givingness of moral obligations. Since they deny that
having such reasons entails that one has sufficient motive, these
accounts will not be able to claim that the overridingness of
moral obligations is universally accessible, however.
Consequently, they presumably will not be able to defend the
intuition that all mature moral agents are capable of
"autonomously" choosing to act morally, i.e., that the authority
of moral imperatives is self-imposed.
The second alternative to internalism, to assume only that
for agents who are well-disposed toward moral considerations,
obligation entails sufficient motive, provides less support for
11
the third sort of intuitions. Although it can defend moral
obligations as self-imposed for agents who are well-disposed
toward moral considerations, it has nothing to offer in defense of
the relationship between the constraints of morality and other
agents.10
Internalism, then, does appear to be well-suited to filling a
niche in moral theorizing that neither externalism nor the weaker
alternatives considered can fill. Despite the strength of
claiming that obligation entails sufficient motive, the
pervasiveness of internalism*s motivating intuitions demand that
we take internalism seriously. To do so we will need to get more
clear about what internalism is, and what filling its niche
requires.
It is worth distinguishing internalism's claim that
obligation entails sufficient motive from the empirical
psychological claim that the thought that an action is morally
obligatory can lead one to do it. As both W. D. Falk and William
Frankena point out, one can be an externalist and still say the
latter.11 Indeed, as Falk points out, accepting the latter claim
10Such accounts might, however, offer an account of human nature that
permits counting what is self-imposed for morally superior agents as
self-imposed for other agents as well (perhaps following an Aristotelian
model).
1:1Frankena goes so far as to argue that an externalist can accept the
stronger claim "that motivation is built . . .into the process of
assenting to [a moral judgment]." ("Obligation and Motivation in Recent
Moral Philosophy," p. 66.)
12
will in some cases force one to reject internalism, as it does for
proponents of views like Pritchard's:12 if we claim that in
addition to an action's being a duty, an action has to be thought
a duty if an agent is to have sufficient motive for doing it, then
we immediately reject internalism because we deny that the fact of
an obligation by itself guarantees the existence of a sufficient
motive. If we claim with Pritchard that it is because the thought
of an action's being obligatory will function in many persons as a
motive that agents have sufficient motive to do as they ought,
then, as Falk points out, we grant that our motive for doing what
we ought
would not be a motive implicit in the very fact
of duty itself but a motive constituted by the
thought that we had the duty, and existing apart
from and additionally to the fact of duty.13
If the existence of a sufficient motive for doing what one ought
depends upon one's thinking one is so obligated, then it is an
open question in each case whether an agent has sufficient motive
for doing what she ought to do. In that case we grant that it
might turn out that one hasn't sufficient motive for doing what
one ought; for example, when one is unaware of one's obligation.
If so, we would have to deny that the existence of a moral
obligation entails the existence of a sufficient motive for
12See Falk, "'Ought' and Motivation," especially, pp. 28-29. See also
H. A. Pritchard, "Duty and Interest," in Readings in Contemporary
Ethical Theory, Wilfrid Sellars and John Hospers, eds. (New York:
Appleton-Century-Crofts, 1970), pp. 690-703.
13"'Ought' and Motivation," p. 27.
13
fulfilling it, i.e., we would have to reject internalism.
Consequently, even if thinking one lies under some obligation can
function as a sufficient motive for fulfilling it, that would not
support our acceptance of internalism.
One is not, of course, required to reject internalism to
accept that the thought that one has an obligation can motivate
one to fulfill it. If an internalist claims that the thought that
one has an obligation can motivate one to fulfill it, then she
must also claim either (i) that such thoughts are not the only
source of sufficient motivation for moral behavior (because one is
sometimes unaware of one's obligations), or (ii) that one is
always aware of one's moral obligations. Because the internalist
holds that obligation entails sufficient motive, she must also
claim that when one thinks one has an obligation to engage in some
behavior that is in fact impermissible, one has sufficient motive
for refraining from engaging in it, however (even if she also
holds that one has sufficient motive for engaging in it) .
I will, consequently, consider the claim
(I) the fact that an agent A ought to D entails A has
sufficient motive for doing D
to be the basic assumption of internalism.14
14It might be helpful to compare this interpretation of internalism's
basic assumption that motivation is built into obligation with the four
interpretations of it Frankena considers legitimately internalist:
(1) that the state of having an obligation includes or is
identical with that of being motivated in a certain way;
(2) that the statement, "I have an obligation to do B,“
means or logically entails the statement, "I have, actually
or potentially, some motivation for doing B"; (3) that the
14
(I) is mors restrictive than the intuition it responds to suggests
Internalism is supposed to be an attractive view about moral
obligation because it responds to some of our intuitions that deny
that a gap between obligation and motivation can be tolerated.
Thus, if internalism is an attractive view, it is attractive in
comparison with views about moral obligation that do not respond
to these intuitions, such as externalism (or the alternatives
considered above). Internalism as defined by (I), however, not
only involves defending this intuition, but defending the
reasons that justify a judgment of obligation include or
are identical with the reasons that prove the existence of
motivation to act accordingly; (4) that the reasons that
justify a judgment of obligation include or are identical
with those that bring about the existence of motivation to
act accordingly ("Obligation and Motivation in Recent
Moral Philosophy," pp. 58 - 59).
(1) is apparently stronger than (I) since "having sufficient motive" for
engaging in some behavior (at least on the surface) does not seem to
entail "being motivated in a certain way." For example, it is not
obvious that one couldn't have sufficient motive for engaging in some
behavior, but nonetheless not do so because one forgets and does
something that precludes it. (l) would therefore appear to yield a
version of internalism, though not the weakest one.
Although Frankena is not very clear on how (2) should be interpreted
so that it is an internalist claim, there seem to be primarily two
possibilities. If having "some motivation" entails having sufficient
motivation, then (2) appears to be a translation of (I) in terms of
moral language. In that case, (2) would be a more appropriate starting
point than (I) if the basic internalistic assumption were that
motivation is built into moral judgments rather than into moral
obligations themselves. If, on the other hand, having "some motivation"
entails only that the agent "feels a responsive beating in his heart"
(p. 65), then (2) is obviously significantly weaker than (I). As
Frankena notes, however, on such a reading (2) is compatible with
externalism (p. 79).
(3) and (4) are both weaker than (I). Both assert that motivation is
built into the conditions under which one would be justified in making a
moral judgment. Consequently, unless one is justified in making a moral
judgment precisely in those cases in which one is (relevantly) morally
obligated, neither (3) nor (4) will guarantee that motivation is built
into moral obligations. (Nor will either guarantee that motivation is
built into moral judgments, since one may be the subject of true moral
judgments one is in no position to see oneself).
15
strongest version of the intuition, according to which an agent
bound by a moral obligation always has sufficient motive to act so
as to fulfill it. Consequently, internalism will be attractive in
comparison with other views only if one thinks that a moral theory
is flawed if it fails to guarantee that an agent bound by a moral
obligation always has sufficient motive to act so as to fulfill
it. I would like to suggest that even if one is sympathetic to
internalism's aims, one may want to reject this strong version of
the intuition that motivates internalism because other pervasive
moral intuitions are apparently incompatible with it. I will then
argue that there is a claim that is weaker than (I) that avoids
these apparent incompatibilities and shares internalism's
advantage over externalism by responding to intuitions that deny
that a gap between obligation and motivation can be tolerated.
This weaker claim will therefore offer a better alternative to
externalism than internalism as defined by (I).
If, as I will argue, there is a claim that is weaker than (I)
that responds to intuitions that deny that a gap between
obligation and motivation can be tolerated, then to motivate
shifting our attention to the (merely) weaker claim it should be
sufficient to show that there are grounds for thinking (I) might
be too strong, rather than to provide a conclusive argument
against (I). Grounds for thinking (I) might be too strong can be
found by looking at some discussions of the relations between
16
moral obligation and motivation that are not directly related to
the debate between internalism and externalism. Although many of
the cases that are central to these discussions are typically
characterized so as to be counterexamples to (I), what they reveal
about the relations between moral obligation and motivation does
not depend on this. I will present two types of cases in which an
agent appears to be bound by a moral obligation but cannot
plausibly be described as having sufficient motive for acting to
fulfill it.15
Cases of the first kind are those in which acting morally
requires that a person consider doing what is for her unthinkable.
Consider, for example, a variant of one of Bernard Williams'
familiar scenarios. Jim, a thoroughgoing pacifist on a botanical
expedition,
finds himself in the central square of a small
South American town. Tied up against the wall
are a row of twenty Indians, most terrified, a
few defiant, in front of them several armed men
in uniform. . . .[T]he captain in
charge . . .explains that the Indians are a
random group of the inhabitants who, after recent
acts of protest against the government, are just
about to be killed to remind other possible
protestors of the advantages of not
protesting. . . .[The captain offers Jim] a
guest's privilege of killing one of the Indians
himself. If Jim accepts, then as a special mark
of the occasion, the other Indians will be let
off. Of course, if Jim refuses, then there is no
special occasion, and Pedro here will do what he
was about to do when Jim arrived, and kill them
■^Obviously moral 'oughts' that do not characterize a particular course
of action for a particular agent will not be internalistic 'oughts'.
(For example, "There ought to be less pollution.") But I assume there
is no disagreement about this.
17
all. . . . [A]ny attempt at . . .[rescuing all of
the prisoners] will mean that all of the Indians
will be killed, and himself [Jim].16
Arguably, Jim ought to kill one member of the group to save the
lives of the other members of the group. Jim will not even
consider doing so, however, because he will not so much as
entertain the possibility of killing. Consequently, Jim cannot
plausibly be described as having sufficient motive for killing a
member of the group.
Cases of the second kind are more common. These are cases in
which what morality demands cannot be brought about by conscious
decision alone. These would include cases in which what is
required is an emotional response, such as pity, gratitude,
respect, or love. A person overwrought with grief, for example,
may not be capable of experiencing owed gratitude. Also included
among these cases would be those in which acting morally requires
that one act with concentration, open-mindedness, detachment,
aggressiveness, courage, or calmness. Courageous behavior may be
morally required of one even if one lacks courage or confidence,
or is paralyzed by fear. These cases follow a pattern readily
illustrated by a case given by Nancy Davis that relies on
utilitarian moral intuitions.17 Suppose that to enjoy some work
16Bernard Williams, "A Critique of Utilitarianism", in J. J. C. Smart
and Bernard Williams, Utilitarianism: For and Against (New York:
Cambridge University Press, 1973), pp. 98-99.
17Nancy Davis, "Acting Utilitarians," Pacific Philosophical Quarterly 66
(1985), pp. 125-140. A more complex case of this type, which is also
more obviously a case involving moral obligation, is Derek Parfit's
18
of art, I need to not think of anything in particular but to
concentrate on the piece, and that I morally ought to dispose
myself such that I will best appreciate the piece. But I cannot
do this by thinking about what 1 morally ought to do, because to
do so would be distracting. In this case, it does not make sense
to say that I have sufficient motive for properly disposing myself
because I cannot choose to do this--I can only be drawn into it.
In cases of this type, agents cannot plausibly be described as
having sufficient motive for acting to fulfill their moral
obligations; even if they are physically capable of fulfilling
their obligations, such agents lack the requisite control over the
means of fulfilling them.
That cases like these play an important role in certain
discussions about moral obligation does not conclusively show that
there are such cases. Moreover, if there are such cases, the fact
that they are characterized in ways that apparently undermine (I)
does not conclusively show that such cases do undermine (I). What
the currency of such cases does show is that it is not generally
considered implausible that there are cases that undermine (I).
I believe that an internalist need not be seriously troubled
by apparent counterexamples to (I). Even if cases like those
described involve moral obligations that violate (I), they do not
jeopardize the intuition that ordinarily an agent’s having a moral
"Schelling's Answer to Armed Robbery." (Reasons and Persons (New York:
Oxford University Press, 1984), pp. 12-13.)
19
obligation entails that she has sufficient motive for fulfilling
it. Consequently, rather than committing to a version of
internalism, the realistic internalist may prefer to adopt a view
defined not by (I) but by the weaker claim
(I1) There are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for
doing D.
I will call these obligations internalistic obligations A 8
It is important to notice that (I1) cannot be taken as merely
asserting that in some cases an agent can be moved to fulfill a
moral obligation by the thought that it is morally required. (I1)
asserts that for some specified class of obligations being bound
by such an obligation implies one has sufficient motive. (I1)
thus sustains the direct contrast with views like Pritchard's. As
was remarked earlier, for views according to which one need think
one is morally obligated in addition to being morally obligated in
order to guarantee one has sufficient motive, being obligated does
not imply having sufficient motive.1920
18Although this project will focus on internalism about obligations, it
would be natural for an internalist about obligation to think that some
other moral relationships are internalistic as well. Presumably,
positing other internalistic relationships would present similar
concerns about limitations on the scope of claims about such
relationships.
19Pritchard‘s view is in fact weaker than this, since he does not even
claim that thinking one ought to behave in some way guarantees one has
sufficient motive for doing so.
20What if the class is defined by "obligations that an agent is aware
of"? Technically, this would be acceptable. But is an internalist
likely to accept that all internalistic obligations are such that the
agent is obligated if she is aware of it and not otherwise? This seems
unlikely given that there seem to be very few obligations which have
that quality. Really the problem is: internalism is appealing only as a
basic insight about moral relationships. If it cannot be understood as
20
(I1) also sustains the advantage internalism has over
externalism because it responds to intuitions that deny that a gap
between obligation and motivation can be tolerated. (I') merely
allows that some (but not all) moral obligations may lack this
property. The question "Why should I act morally?" will still be
philosophically interesting in some cases. If an agent is bound
by an internalistic obligation, then she has sufficient motive for
acting so as to fulfill it. If an agent bound by such an
obligation asks, “Why should I act morally?" then it is prima
facie evidence that she is not bound by the obligation. As for
internalism, asserting (I1) requires us to consider the case
apparently paradoxical because we do not seem to excuse people
from their moral obligations on the grounds that they do not find
them compelling.
It is worth noticing that (I1) does not permit the case to be
explained away by simply saying that an agent who asks must not be
bound by an internalistic obligation. This is because the class
of internalistic obligations is specified independently of the
agent's behavior.21 Regardless of how the class of internalistic
obligations is specified, it will remain a possibility that an
such an insight, then it ceases to be appealing. We have to be able to
argue that morality has the role of guiding conduct autonomously. If
(I') is true but its scope is insignificant, then it becomes irrelevant.
See below, pp. 26-30.
21These issues are discussed more formally below. See pp. 26-30.
21
agent bound by an internalistic obligation will ask, “Why should I
act morally?"
Similarly, (I') supports the other intuitions that motivate
our interest in internalism: that moral obligations imply real
constraints on our ability to act contrary to them, and that
paradigmatic moral behavior is autonomous. Since being bound by
an internalistic obligation entails that one has sufficient motive
for acting so as to fulfill it, if one is bound by an
internalistic obligation, then there are conditions under which
one will choose to act so as to fulfill it. Thus, internalistic
obligations imply real constraints on our ability to act contrary
to them. Moreover, those constraints are self-imposed in the
sense that no external sanction is required for one to choose to
act so as to fulfill an internalistic obligation.
The shift to (I1) is corroborated by Falk's discussion of
internalism. Although Falk tends to contrast internalism and
externalism by emphasizing differences in how they interpret moral
language rather than differences in their stands on (I), the
translation between the two perspectives is not difficult. Falk
begins with the claim that among the ’oughts' of ordinary moral
discourse is "the purely formal motivational sense" of 'ought',
which he characterizes as follows:
[TJhere is a habit of speech according to which,
when a person asks, "need I really, or have I, if
only dispositionally, a reason for doing this
act?" he might as well have said, "should 1 or
22
ought I really to do this?" the latter expression
being in fact the more colloquial.22
According to Falk, an agent has a dispositional reason for
some action when reflecting on the action (and on the situation)
“is capable of determining him to do it." This is in contrast
with occurrent reasons, the presence of which indicates that the
agent is already "being impelled" by the thought of the action to
do it.23 If we set aside for a moment questions about these two
conceptions of reasons and provisionally accept this
characterization as meaningful, we can get a provisional
characterization of the distinction between internalism and
externalism Falk is after. Internalists are those who claim that
(F) some moral 'oughts' are motivational 'oughts', i.e., some
moral obligations are such that reflecting on an action
that would fulfill it [and on the situation] is capable of
determining an agent who is obligated to do so.2425
Externalists deny that any moral 'oughts' have the purely formal
motivational sense.
"One has sufficient motive to do what one ought to" is
trivially true when its 'ought' is taken in the motivational
sense, but may be false if 'ought' is taken in some other sense.
22"'Ought' and Motivation,” p. 34.
23"'Ought' and Motivation," pp. 25-26. "The thought of the action" must
I think be taken as broadly as possible; it seems in the end to be
possible that it may be either more or less than a conscious mental
representation of the action.
24In fact, Falk argues that some moral oughts in ordinary moral
discourse are not motivational 'oughts’. (See "'Ought' and
Motivation.")
25Falk later argues that internalists should claim that all moral
'oughts' are motivational 'oughts', but he does so on independent
grounds. (See "'Ought' and Motivation," p. 41.)
23
Thus Falk's internalist will regard assertions of the form 'X has
sufficient motive to do what she ought to' as in some cases
trivially true and in others as a matter of empirical fact.
Whether one has a reason to do what one ought to is always an open
question if one is an externalist. Falk thus seems to see
internalism as arising from a premise like (I1) rather than (I).
It is convenient to think of internalism as asserting (I)
rather than (I') because (I) captures the intuitions behind
internalism more straightforwardly, and consequently is more
commonly taken to define internalism. I have argued that this is
a poor characterization of the basic assumption of internalism
because it seems to undermine the plausibility of internalism
unnecessarily. It may nonetheless be adequate to characterize
internalism by (I); it may be, as Falk suggests, that all
plausible moral theories asserting (I1) entail (I) as well. Since
the characterization of internalism by (I) is deeply embedded in
the literature, I will reserve the term "internalism" for moral
theories that satisfy (I). I will call moral theories that
satisfy (I1) internalistic, since such moral theories respond to
the intuition underlying internalism but may also consider some
obligations as noninternalistic. It will be convenient to refer
to (I1) as the Basic Internalistic Premise:
(BIP) There are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for
doing D.
24
Notice that there is an asymmetry in the definitions of
“internalistic moral theories" and "internalistic obligations".
To say that a moral theory is internalistic is to make a claim
about what kinds of obligations the theory posits; an
internalistic moral theory holds that some obligations are
internalistic obligations. To say that an obligation is
internalistic is to make a claim about how certain agents might be
motivated; an obligation is internalistic if an agent bound by it
has sufficient motive to act so as to fulfill it. Internalism
holds that all obligations are internalistic.
The burdens of choosing between an internalistic moral theory
and an externalist one will differ somewhat from those attached to
the choice that Frankena describes. By assuming only the Basic
Internalistic Premise we allow that a gap between obligation and
motivation is sometimes permissible. Thus, one consequence of the
shift to the Basic Internalistic Premise is that we may no longer
consider externalist moral theories inadequate on the grounds that
we cannot tolerate a gap between obligation and motivation.
Externalist moral theories instead bear the burden of insisting on
a gap between obligation and motivation in cases in which such a
gap cannot be tolerated: cases in which agents are bound by
internalistic obligations. Correspondingly, the burdens of
holding an internalistic moral theory will differ from those of
25
holding a version of internalism. Whereas holding a version of
internalism requires us to face the possibility that obligations
may be trimmed to individual motives (which perhaps leads to moral
relativism or skepticism), holding an internalistic moral theory
seems to have two troublesome implications: (i) that the class of
obligations that are claimed to imply the existence of sufficient
motive may be trimmed to individual motives--giving the part of
morality that has the role of guiding morality autonomously less
scope than one wants for it, in addition to possibly leading to
moral relativism or skepticism; and (ii) that internalistic
obligations may conflict with noninternalistic requirements. The
former burden is roughly the same as that of internalism.
My aim in this project will be to investigate some of the
burdens of holding internalistic moral theories.
Methodology
Internalistic moral theories are supposed to be attractive in
comparison with externalist moral theories because they differ
from externalist moral theories by not leaving a gap between
obligation and motivation in the cases in which we cannot tolerate
such a gap. Consequently, unless an internalistic moral theory
can give sufficient scope to the class of obligations that
guarantee sufficient motive, it will not have the desired
advantage over a competing externalist moral theory. To guarantee
that the class of internalistic obligations specified by an
26
internalistic moral theory has sufficient scope to make the theory
attractive in comparison with a competing externalist moral
theory, it must be the case that whenever we (intuitively) think
that an agent lies under an obligation that entails she has
sufficient motive to fulfill it, that obligation is included in
the class of internalistic obligations specified by the theory.
(For versions of internalism, according to which all moral
obligations are internalistic, this requirement is trivially met.)
I will call this property completeness.26
We should also assume that any plausible internalistic moral
theory will meet a second constraint. It is generally assumed
that the choice between internalism and externalism does not
depend on a choice about what agents are morally obligated to do.
Instead, the plausibility of claims about the relations between
obligation and motivation depends (in part) on their compatibility
with what we know about the moral obligations by which agents are
bound. Thus we will not find an internalistic theory plausible
unless its claim that there is a certain class of internalistic
moral obligations is compatible with what we know about the moral
obligations by which agents are bound. If a theory implied that
26There will of course be a symmetrical constraint. An internalistic
view should not be too wide in scope; i.e., it should not include any
obligations for which a gap between obligation and motivation is
permissible. That an internalistic view meets this constraint is not
essential to find it appealing in comparison to externalism, however,
since even if an internalistic view had this flaw, it could be viewed as
superior to externalism in its treatment of (truly) internalistic
obligations. For this reason, we will not use this constraint to limit
the field of potentially plausible internalistic theories.
27
in some case an agent internalistically ought to do what she
noninternalistically ought not to do, it would imply that the
choice between an internalistic and an externalist moral theory
depends on a choice about what agents are morally obligated to do.
We should therefore assume as a minimal condition for the
plausibility of an internalistic moral theory that judgments about
obligations in the class it specifies as internalistic should not
directly contradict any judgment about noninternalistic moral
obligations; more precisely, it should not turn out that for some
agent A and course of action D, A internalistically ought to D and
ought not to omit D-ing, but noninternalistically ought not to
D.27
Care should be taken at this point not to allow the
difficulties of describing and individuating actions to obscure
the content of this assumption. We allow that some internalistic
moral theories may allow moral conflicts in the following mold:
One ought always to do as one has promised, and
ought never to omit what one has promised.
One ought not to act so as to cause unnecessary
harm.
X has promised Y that she will have Y's dog
killed after he dies (out of pity for the
surviving dog). Y dies, but contrary to Y's
prediction, the dog's disposition is not
drastically affected.
27We are not warranted in just assuming that internalistic obligations
do not conflict with other obligations, because some theories permit
conflicts of obligations.
28
X is in a situation such that keeping a promise
involves causing unnecessary harm; therefore X is
bound by two conflicting obligations.
To describe X's predicament as one of conflicting obligations,
what is relevant is that however we describe what she does, X will
fail to do at least one of the two things she ought to do.
Whether or not there is a single description of one of her choices
such that it is something which X both ought and ought not to do
is (presumably) not relevant to the question of whether X is bound
by conflicting obligations (according to a given moral theory that
allows conflicts of obligations). Similarly, regardless of
whether we can describe a course of action D such that A
internalistically ought to D and ought not to omit D-ing, but
noninternalistically ought not to D, what we cannot permit is the
possibility of A's behavior being viewed as a decision between the
internalistic obligation and the noninternalistic obligation. I
will call this restriction congruence.28
The aim of this project is to investigate the burdens of
holding a moral theory that is internalistic. I will use these
two criteria as a sieve to remove from consideration moral
theories that it would not be reasonable to hold.
^Notice that productivity is not a property required to make an
internalistic view plausible. Internalistic obligations might be
superfluous from the point of view of generating all moral obligations.
All that is wanted to support the acceptance of an internalistic view is
that one of the roles of morality is the "autonomous governance of
conduct". Even if the internalistic part of a moral theory is
superfluous in the obligation-generating sense, it would not be
superfluous if it explained how morality can have the latter role.
29
To have a moral theory that has congruence takes away the
conflict problem. To have a moral theory that has completeness is
to have a response to the worry that internalistic obligations
will have to be trimmed to individual motives. In this way, the
worries about inadequate scope and conflict are translated into a
worry about existence: Are there any internalistic moral theories
that pass the completeness and congruence tests?
This is the strategy I will be following in this project. I
will try to characterize moral theories that are internalistic and
plausible by investigating the consequences of assuming the Basic
Internalistic Premise, completeness, and congruence. The charge
that internalistic moral theories imply that moral relativism
holds for internalistic obligations appears to be the most serious
obstacle to maintaining that a moral theory satisfies all three;
since moral relativism seems to be incompatible with ordinary
moral judgments, an internalistic moral theory that implied that
an extreme form of moral relativism holds for internalistic
obligations would seem to fail to meet either the completeness
restriction or the congruence restriction. Consequently, I begin
by considering a well-known argument linking an internalistic
intuition to moral relativism, one given by Gilbert Harman.
30
Chapter Two s
Gilbert Harman's Argument for a Link between Internalism and Moral
Relativism
Possibly the most influential argument in favor of a link
between internalism and moral relativism is that given by Gilbert
Harman in “Moral Relativism Defended." Harman does not explicitly
offer his argument as one linking ethical internalism to moral
relativism. He describes his argument as a defense of a theory of
truth evaluation for a certain class of moral judgments which he
calls "inner judgments". The theory of truth evaluation is then
supposed to require us to accept moral relativism. Nonetheless
Harman does end up pitching his argument to internalists because,
as I will argue, positing inner judgments in the first place seems
to require that we assume that:
(JIM) some moral judgments imply that their subjects have a
motive for acting in accordance with them.
I will call this claim the "Judgments Imply Motives Premise".
Since this claim is a consequence of the Basic Internalistic
Premise (i.e., that there are some moral obligations such that the
fact that an agent ought to D entails she has sufficient motive
for doing D), if Harman's argument is successful, it will show
that at least some internalistic moral theories sanction moral
relativism.1 In this chapter, I will argue that Harman's argument
does not succeed in showing that assuming the Judgments Imply
Motives Premise commits one to moral relativism.
1I.e., those that posit inner judgments.
31
I will not challenge Harman's claim that the theory he
proposes has moral relativism as a consequence. What I want to
consider is whether the arguments Harman offers in favor of the
theory are sufficient to warrant its adoption.
To understand Harman's proposal, we must first recognize that
it concerns only a particular kind of moral judgment: inner
judgments. He describes inner judgments as follows:
We make inner judgments about a person only if we
suppose that he is capable of being motivated by
the relevant moral considerations. We make other
sorts of judgments about those who we suppose are
not susceptible of such motivation.2
Inner judgments have two important
characteristics. First, they imply that the
agent [the subject] has reasons to do something.
Second, the speaker [the utterer of the judgment]
in some sense endorses these reasons and supposes
that the audience [those to whom the speaker's
judgment is addressed] also endorses them. Other
moral judgments about an agent, on the other
hand, do not have such implications; they do not
imply that the agent has reasons for acting that
are endorsed by the speaker.3
Inner judgments are thus distinguished from other judgments by
their implications. Unlike other moral judgments, inner judgments
imply that (1) the subject is capable of being motivated by
relevant moral considerations (in the situation the judgment is
about), (2) the subject has certain reasons for acting in the
prescribed manner, (3) the speaker endorses the reasons in (2),
2"Moral Relativism Defended," p. 4.
3"Moral Relativism Defended," p. 8.
32
and (4) the speaker supposes the audience endorses the reasons in
(2) .
Explanations of these four characteristics
(1) and (2) of our list of defining characteristics of inner
judgments imply that the subject is capable of being motivated by
relevant moral considerations, and that she has reasons for acting
in the prescribed manner. Both (1) and (2) are ambiguous as
stated, but Harman does offer more specific accounts of them.
(1), which requires that the subject is capable of being
motivated by relevant moral considerations, is ambiguous primarily
because one might think that the capacity to be motivated by moral
considerations is something one has in virtue of being human.
That capacity is not the capacity that Harman intends, however.
Harman intends an agent's “capacity to be motivated by relevant
moral considerations" (in particular situations) to be dependent
upon her particular dispositions to act in certain ways. This can
be readily observed by considering Harman's treatment of his
cannibal case. In Harman's example, a group of cannibals have
eaten the sole survivor of a shipwreck. Harman claims that the
cannibals cannot be the subjects of a true inner judgment that
“They ought not to have eaten their captive" because of their
primitive morality; presumably he supposes that their primitive
morality prevents them from being able to feel the force of
whatever moral considerations make eating their captive morally
33
wrong. For a judgment to be an inner judgment, its subject must
be, as Harman puts it, “susceptible to certain moral
considerations"; she must not be “beyond the motivational reach of
the relevant moral considerations."4
(2), which requires that the subject has certain reasons for
acting in the prescribed manner, is ambiguous because “reasons"
is. When talking about reasons, Harman uses "having a reason" in
a motivational, not explanatory, sense. Even if “the captive is a
human being" can be understood abstractly as a reason for
refraining from killing the captive, that does not mean that every
agent will have a reason in the motivational sense to refrain from
killing the captive: for example, some agents may be completely
indifferent to such actions. According to Harman, for some
cannibals there may be no reason that they could have for
refraining from killing and eating the captive. I will follow
Harman in using "having a reason" to mean "having a motivation-
affecting reason".5
Claiming that some moral judgments satisfy defining
characteristics of inner judgments (1) and (2) thus immediately
commits one to the Judgments Imply Motives Premise, i.e., to the
claim that some moral judgments imply that their subjects have a
4"Moral Relativism Defended," p. 8.
5There remain other questions about reasons, of course; notably, What
are they? and How do they guide action? I will not deal with these
difficult questions here. What I want to assume is only that there are
reasons that affect motivation involved in reason-guided action.
34
motive for acting in accordance with them. In fact, claiming that
some moral judgments satisfy (1) and (2) seems to commit one to a
fairly strong version of internalism. In particular, if we assume
that when one is moved by relevant moral considerations, one is
engaging in reason-guided action, then a true inner judgment that
"A ought to D" implies not only that A has reasons for pursuing D,
but that she has motivationally sufficient reasons for pursuing D.
Positing inner judgments, then, seems to commit one to assuming
that there are some judgments that imply that their subjects have
motivationally sufficient reasons for acting in accordance with
them.
Characteristics of inner judgments (3) and (4), which require
that the speaker endorses and supposes her audience to endorse
certain reasons, are meant to depend upon an Aristotelian or
Humean assumption that an agent's having reasons for acting (in
any instance) “have their source in goals, desires, or
intentions."6 I will follow Harman in using the expression
"motivating attitudes" to mean (all of) the sources of an agent's
reasons for acting. According to Harman, the utterer of an inner
judgment endorses the reasons of the subject of the judgment, and
shares that endorsement with an audience, in the sense that the
speaker, subject, and audience all share some motivating attitudes
that are the source of some of the subject's reasons for acting as
6"Moral Relativism Defended," p.9.
35
prescribed by the judgment. Characteristics (3) and (4) thus
require that inner judgments can only be made by a speaker who
believes that she shares certain motivating attitudes with other
agents; if, in particular, the subject of her moral judgment has
relevant motivating attitudes in common with her, and so do those
in the audience of the judgment, then the speaker will be making
an inner judgment.
(3) and (4) do not yield additional information about the
ways in which the subject might be motivated to act, but instead
provide information about whether her reasons for acting are
shared by others. (3) and (4) thus do not provide considerations
that will affect the question of whether or not the subject has
reasons or sufficient reasons for doing anything; if the fact that
someone's asserting “A ought to D" is a true inner judgment
implies that A has sufficient reasons for pursuing D, it does so
regardless of who the speaker and audience of the judgment are.
The form of Harrpan's argument, then, is that certain
internalist assumptions about moral judgments (including, at
least, the Judgments Imply Motives Premise) combined with extra
assumptions about inner judgments entail moral relativism.
Harman's proposal is thus best understood as a proposal about how
to evaluate judgments that imply that their subjects have a motive
to act in accordance with them. If we are externalists, and thus
deny that any moral judgments imply that their subjects have a
36
motive to act in accordance with them, then we should deny that
there are inner judgments. Harman's proposal will presumably not
have any force for externalists. Moreover, I will argue that
despite Harman's special interest in inner judgments, his argument
primarily exploits the properties of internalistic obligations,
and that where it does not it fails to be persuasive.
The role of restrictions on the model of motivation in Harman's
argument
Harman's aim in restricting motivating attitudes to goals,
desires, and intentions is to exclude a Kantian account according
to which reason by itself may be the source of some reasons for
acting. Although Harman does not try to offer an argument for why
we should reject this Kantian assumption, he does give his purpose
in doing so: to preserve certain intuitions which he intends to
assume as data in his arguments. These intuitions are that it is
not unreasonable to assume that there may be nothing that would
serve as a reason for beings from outer space not to hurt humans,
for Hitler not to have ordered the slaughter of millions of Jews,
for a person raised within an orthodox tradition that requires and
condones certain acts which we would call murder not to murder
within that tradition, or for cannibals not to engage in a
particular cannibalistic act. If reason itself could provide
reasons for acting, and the agents in these four cases were
capable of reason, then it would seem to be unreasonable to assume
37
that nothing could serve as a reason for each to act as described.
It is as a consequence of this observation that Harman confines
his argument by assuming that reason by itself cannot provide
reasons for acting.
If we join Harman in accepting the intuitions, we are
committed to rejecting claims according to which it is
unreasonable to assume that nothing could have served as a reason
for the four agents to act as described. In that case, we should
probably also join Harman in denying that reason by itself can
provide reasons for acting.7 But in that case we should actually
go further, because Harman's restriction is not strong enough to
preserve these intuitions. In general, if there is some
motivating attitude common to all four agents of Harman's examples
that can by itself produce reasons for acting, then it may be
unreasonable to assume that there is nothing that might serve as a
reason for acting for the agents to act as described. If Harman's
intuitions compel us to deny that reason may be a source of
reasons for acting, then it would seem that they should also
compel us to deny that any motivating attitude that is shared by
all agents capable of reason-guided action may be a source of
reasons for acting.
7Probably only, because there might of course be other factors which
prevent reason from producing the appropriate reasons for acting in the
cases described. This part of the argument is not airtight, as Harman
apparently recognizes. We are, not unreasonably I think, supposed to
grant that some intuitions similar to those mentioned can be constructed
so that their reasonableness contradicts the assumption that reason can
produce reasons for acting in any agent.
38
Harman's intuitions should not compel us to go quite so far,
however. Since it is only morally interesting behavior that we
are concerned with, we need only be concerned about whether
motivating attitudes common to all agents can produce in those
agents reasons for acting that might guide action that is
apparently in accordance with a moral principle.8 If, for
example, the sole motivating attitude shared by all agents is a
desire to eat when they are hungry, and if this desire can produce
in an arbitrary agent a reason for acting that motivates food-
procuring behavior (when they are hungry), we would not have any
difficulty maintaining Harman's intuitions--unless food-procuring
behavior is equivalent in all agents to some behavior apparently
in accordance with a moral principle. Similarly, Harman's
examples illustrate no obvious contradiction with the assumption
of a faculty of reason that produces reasons for acting in all
agents if reason is supposed not to be sufficient to produce in
all agents behavior apparently in accordance with a moral
principle.
It does not seem reasonable to suppose that there is no
motivating attitude that can produce a reason for acting in every
agent, since, for example, hunger can produce reasons for acting
8The reason why it is "action apparently in accordance with ..."
rather than "morally good actions" in which we are interested is that
only the former is required to support Harman's intuitions. To reject
Harman's intuition about Hitler, for instance, we do not need to say
that Hitler could have been motivated by duty not to commit genocide,
but only that Hitler had a reason which could have prevented him from
committing genocide.
39
in all agents that we know of. But perhaps it is less implausible
to assume that no shared motivating attitude can produce in an
arbitrary agent reasons for acting that motivate behavior
apparently in accordance with a moral principle. No more than the
latter will be required by Harman for his examples to be
persuasive.9 Harman does not argue for this assumption (or for
the less general one concerning reason alone) , so we should
consider this as a limit on the conditions under which we will
find his arguments convincing: we must assume that
(H) No motivating attitude shared by all agents can produce in
an arbitrary agent reasons for acting that motivate
behavior apparently in accordance with some moral
principle.
I will consider the plausibility of this assumption below.
Hannan's proposal for evaluating the truth of inner judgments
We are now in a position to describe Harman's proposal.
Harman's proposal for evaluating inner judgments begins with a
general proposal for evaluating judgments involving the moral
sense of 'ought'. Such judgments are to be understood in terms of
a four-place predicate:
"Ought(A,D,M,C)" means roughly that, given that
[agent] A has motivating attitudes AT and given
90ne might argue that we need to assume that no motivating attitude
shared by all agents can produce in an arbitrary agent reasons for
acting that when "present" always motivate behavior apparently in
accordance with some moral principle. (See footnote 8 of this chapter,
for an illustration of why the more encompassing "behavior apparently in
accordance with a moral principle" is required rather than "morally good
action".) Rather than try to justify the adequacy of this weaker claim,
I will use the stronger claim. My argument will not depend upon this
choice. See footnote 25.
40
[that A is in conditions] C, D is the course of
action for A that is supported by the best
reasons. In judgments using this sense of
“ought," C and M are often not explicitly
mentioned [and] are indicated by the context of
utterance. Normally, when that happens, C will
be “all things considered" and M will be
attitudes that are shared by the speaker and the
audience.10
Suppose that a speaker makes an inner judgment that is represented
by "Ought(A,D,M,C)". Harman proposes that the inner judgment is
true if and only if (a) Ought(A,D,M,C) holds, and (b) the agent A,
the speaker, and those in the speaker's audience have motivating
attitudes M.
Harman's proposal has some initial plausibility.11 Condition
(a) implies that the agent has reasons to follow course of action
D, so the judgment will satisfy condition (2) of the list of
defining characteristics of inner judgments (i.e., that the
subject has certain reasons for acting in the prescribed manner).
If the speaker believes that (b) holds, then conditions (a) and
(b) together imply that the judgment will also satisfy conditions
(3) and (4) of the list (i.e., that the speaker endorses and
supposes her audience to endorse certain reasons for acting in the
prescribed manner). That condition (b) holds for the judgment
10"Moral Relativism Defended," p. 11.
1:1By the time Harman reaches this point in his proposal, he is
distinguishing inner judgments from internalistic judgments; he supposes
that the moral "ought" is the internalistic "ought". One should wonder
what is supposed to justify this move. But that is not our concern
here. We are interested in the proposal as it applies to inner
judgments. As we have seen, it is plausible that inner judgments report
internalistic obligations. By restricting our attention to inner
judgments, we thus avoid having to be particularly concerned about
Harman's jump from moral "ought" to internalistic "ought".
41
follows from conditions (3) and (4) of inner judgments provided
that the speaker rightly supposes the audience to share motivating
attitudes M.
It is difficult to see that condition (a) should hold for the
judgment we are considering, however. Since the judgment is an
arbitrary inner judgment, the most we may legitimately claim about
the agent's reasons is that the agent has sufficient reasons in
support of course of action D (and that only if we assume that
when one is moved by relevant moral considerations, one is
engaging in reason-guided action).12 To say that the agent has
sufficient reasons in support of D would not necessarily be
grounds for claiming that her best reasons support D, however;
this would presumably depend on what we mean when we say that some
course of action is supported by the agent's “best" reasons. How,
then, might the authority of D as the available action which is
supported by the agent's best reasons be established?
To make a claim about what the agent's best reasons support
we will need to use some standard to rank either the agent's
reasons for acting or sets of her reasons for acting that support
particular courses of action she is choosing between. Such
standards are rarely explicitly offered, except by those who
regard the class of morally relevant reasons as being very
12This follows from defining characteristic (1).
42
limited, such as Kant.13 Nonetheless the implicit use of such
standards is common; any formalist version of ethical internalism
will need to provide one. The plausibility of accounts like
Harman's that rely upon ranking courses of action according to
reasons that support them can be no greater than the plausibility
of the standards upon which they depend.
There are many complications that arise in specifying such
standards. To evaluate Harman's proposal, it will be useful to
focus on how such standards are related to speaker-relativity. I
discuss the complications that arise in ranking courses of action
using a standard for ranking reasons for acting in greater detail
in the chapter three.
It seems, first of all, that on any plausible account of best
reasons that is compatible with Harman's arguments, the course of
action supported by the agent's best reasons depends upon all of
her reasons for acting, and hence on all of her motivating
attitudes, including but not merely constituents of M, the
motivating attitudes that she shares with the speaker. One
cannot, for example, tell that the best of the cannibal's reasons
supports her refraining from eating her captive without knowing
13James Martineau offers one of the few serious accounts of a standard
for ranking reasons. (See Types of Ethical Theories, 2 vols. (Oxford:
Oxford University Press, 1885).) Martineau's effort to produce such an
account was ridiculed by his contemporaries, especially by Sidgwick, who
seems to have regarded the need for such an account as the chief defect
of the Butlerian approach to ethics.
43
what kinds of reasons she has for and against eating her captive.
If the cannibal's choice is between eating the captive in order to
insure a good harvest and letting the captive live out of
ambivalence toward her community's needs, then we might think that
eating the captive is supported by her best reasons (regardless of
whether we share the cannibal's interest in her community's
welfare). If her choice is between acting to insure a good
harvest and acting to respect a human being's right to life, our
appraisal might be different.
One could deny that in general determining which course of
action is supported by an agent's best reasons requires us to
consider all of her reasons for acting. One might think that
there is a privileged class of best reasons; for instance, one
might think that there are “moral" reasons which are always the
best reasons. In that case, one might deny that nonmoral reasons
(and the motivating attitudes from which they spring) matter to
the evaluation of what an agent has best reasons for doing when
she has some moral reasons for acting. On accounts such as these
the course of action supported by an agent's best reasons does not
depend upon her nonmoral reasons, but it does depend upon all of
her moral reasons. Consequently, the course of action supported
by an agent's best reasons would depend on all of those of her
motivating attitudes that may be the source of moral reasons.14
14If, in addition, moral reasons are thought to derive from a single
motivating attitude, or fixed group of motivating attitudes, and these
44
Let Ma be the set of motivating attitudes that determine
which course of action is supported by agent A's best reasons.
(The idea is that Ma is the set of all of A's morally relevant
motivating attitudes.) Whether Ma includes all of the agent's
motivating attitudes or just those of her motivating attitudes
which may produce moral reasons, it does not seem that Ma need be
identical to M, the set of motivating attitudes the agent shares
with the speaker. This is because the truth of the judgment does
not depend upon what the speaker knows about Ma - Consequently, it
is not necessary that the speaker knows (or can reasonably expect
to know) whether, for example, the agent's motivating attitudes
are such as to preclude the possibility of having the best reasons
for doing D (regardless of whether M and C are appropriately
applicable to her) . The agent might have reasons of which the
constitute the "M" of all inner judgments, then the evaluation of inner
judgments about an agent A will not depend upon others of A's motivating
attitudes. Harman certainly does not have an account like this,
however. On his account, even if there are moral reasons, and these are
always the best reasons, the determination of which motivating attitudes
they have as their source is still a matter of convention. Moreover, if
"M" were fixed and had the above properties, then moral relativism would
not result from Harman's arguments, only that some agents are not proper
subjects of any inner judgments.
If we had such an account, we could apparently evade Harman's
argument for moral relativism for inner judgments even granting (H) .
Such an account would also be badly in need of reconciliation with
common sense. The most obvious flaw in such an account is that it would
deny that one could have moral reasons that support conflicting courses
of action. For example, if the keeping of a promise were to conflict
with carrying out a familial obligation, the account could not allow
that one had moral reasons in support of both. Some reasonable moral
theories may claim that there are no moral conflicts in cases such as
these--that one's duty is simply hard to determine. If we added this to
our account, then in cases in which there are apparent moral conflicts,
we would have to claim either that the agent lacks moral reasons
altogether, or that she has a moral reason but it is not clear what it
supports. In either case, how to determine what courses of action are
supported by an agent's best reasons will remain obscure.
45
speaker is unaware for doing some action incompatible with D, and
for which the agent has reasons that are superior to any reason
that she could have for doing anything else. An example of such a
case is suggested by Harman's example of an 'Ought(A,D,M,C)'
judgment which is not an inner judgment: "As a Christian, you
ought to turn the other cheek; I, however, propose to strike
back."15 Presumably, in virtue of being a Christian, the subject
of the judgment is supposed to have motivating attitudes
additional to those she shares with the speaker. These motivating
attitudes are supposed to provide reasons for acting such that the
course of action supported by her best reasons is different from
that supported by the speaker's. If the speaker in Harman's
example was unaware that the person she was addressing was a
Christian, then she might erroneously make the inner judgment:
"You ought to strike back." In such cases, the speaker will
merely be wrong in her judgment, even though she may be correct in
her assessment of the agent’s circumstances and of the motivating
attitudes they share. The judgment would be false precisely
because in such cases is not identical to the set of motivating
attitudes they share (even if contains all of the motivating
attitudes they share).16
15"Moral Relativism Defended," p. 10.
16It might be thought that this last result might be avoided if it were
guaranteed that the motivating attitudes M shared by the agent and the
speaker produce reasons with a special kind of motivational force, in
which case we could claim that there is a reliable causal link between
46
Harman's proposal consequently amounts to claiming not only
that the truth of inner moral judgments is determined relative to
the motivating attitudes which the speaker believes to be shared
by the agent and herself, but also that it is determined relative
to all of the agent's morally significant motivating attitudes.
Harman's inclusion of the motivating attitudes the agent shares
with the speaker in his proposal therefore does not serve to make
the evaluation of the judgment depend on the agent's motivating
attitudes, but on the speaker's, which would otherwise not
necessarily find their way into the evaluation.17
having M and having the best reasons for doing D. This seems to be
Harman's purpose when he shifts the discussion to moral bargaining.
According to Harman's moral bargaining proposal, the source of any
justification for the speaker's assertion of the judgment J (i.e., for
the speaker's belief that she and the agent have some special motivating
attitudes in common) is the speaker's justified belief in the agreement
of certain intentions (understood very widely) of the agent and herself.
Their having these particular intentions is supposed to arise and be
"agreed" to by implicit convention. Since the speaker has the
motivating attitudes presumed to be shared, we are to assume that she
knows what reasons for acting they would produce in an agent who has
them under circumstances C. Limitations on the utility of this
assumption for supporting an inner judgment arise as they did earlier.
If the agent does not happen to possess the intentions the speaker
expects her to, J may be false, even if in fact the agent's having those
intentions would make J true. Also, the agent may pretend to have
intentions that she does not in fact have, or the speaker may be a poor
judge of the agent's intentions. Hence, the agent's having particular
intentions is independent of the speaker's beliefs about the agent's
intentions. Although the force of motivating attitudes generated by
certain kinds of intentions may be ascertainable, the presence of the
intentions in another agent cannot be. Thus, even granting the moral
bargaining proposal, the truth of J remains relative both to the
speaker's beliefs and to facts about the agent alone (the agent's
motivating attitudes) independently. With or without Harman's agreement
proposal then, the truth of J is relative to both.
17Since the truth value of the judgment depends on the agent's
motivating attitudes and those she shares with the speaker, a somewhat
more general, but also more illuminating way of understanding the
judgment might be as "Oughti(the agent's motivating attitudes, the
speaker's motivating attitudes, C, D)", to be evaluated as true just in
case D is the course of action for the agent supported by the best
47
There turn out, then, to be two distinct kinds of evidence
needed to establish the truth of inner judgments: (i) facts about
the agent needed to determine whether the behavior prescribed
really is supported by the agent's best reasons, and (ii) facts
about the relationship between the agent and the speaker needed to
determine whether the speaker is justified in making an inner
judgment. This is as expected given the definition of inner
judgments. Defining conditions (1) and (2) suggest the need for
the first kind of evidence, and conditions (3) and (4) suggest the
need for the second kind of evidence.
Is Harman's proposal plausible?
Why would we find this a plausible account of how any moral
judgment, inner or not, is evaluated?
If we accept the Judgments Imply Motives Premise, and hence
we think that some moral judgments imply that their subjects have
a motive for acting in accordance with them, then it is natural to
think that inner judgments are such judgments, and hence that
facts about an agent relating to how she is capable of being
motivated might be relevant to whether an inner judgment about her
is true. To the extent that claims about how the subject is
capable of being motivated would affect claims about what the
subject has a motive for doing, such claims will be useful in
reasons, C describes the agent's situation, and the speaker and the
agent share some motivating attitudes.
48
determining whether the judgment is true, just as they would be
for any moral judgment that implies that its subject has a motive
for acting in accordance with it; evidence relevant in evaluating
one is evidence relevant in evaluating the other.
It is much more difficult to see why an inner judgment should
be evaluated relative to the speaker. Neither the Judgments Imply
Motives Premise nor the Basic Internalistic Premise makes
reference to the speaker, and hence, neither suggests speaker-
relativity. Neither does externalism suggest speaker-relativity.
The speaker-relativity of Harman's proposal reveals a
difficulty in his overall strategy. Rather than arguing that the
defining characteristics of inner judgments characterize the
significant features of some class of judgments, Harman merely
stipulates the defining characteristics of inner judgments without
argument. Consequently, even if we accept Harman's claim that
there are judgments which have characteristics (l)-(4), we do not
necessarily have reason to think that (1)—(4) are significant
features of the judgments which satisfy them rather than merely
accidental characteristics; we may grant that some judgments
satisfy (1)—(4), but think they are evaluated in precisely the
same fashion as other judgments.
The question of whether a judgment is an inner judgment will,
of course, depend upon the speaker. But that is no reason to
think that the truth of a particular inner judgment is evaluated
49
relative to the speaker, or, for that matter, relative to the
subject. At minimum, what is needed to make the evaluative
proposal plausible is a principled reason for thinking that inner
judgments are evaluated differently than are other judgments.
If we accept the Judgments Imply Motives Premise, i.e., that
some moral judgments imply that their subjects have a motive for
acting in accordance with them, then we will have reason to think
that the truth of such judgments will depend upon their subjects'
motives and how they are capable of being motivated. As was
discussed above, the agent-relative part of Harman's proposal has
some plausibility if we assume the Judgments Imply Motives
Premise.
It is difficult to find principled support for the speaker-
relative part of the proposal, however. One place that one might
think to look for such support is in R. M. Hare, whose
prescriptivist account of moral judgments might be thought to
suggest it.18 On Hare's account, moral judgments entail
imperatives directed at all agents. If anyone is not subject to
the moral judgment, then the judgment is false. It might be
thought that Hare's account entails some form of speaker-
relativism because he defines evaluative moral judgments so that
they can only be made by agents who subscribe to the imperatives
that follow from them. Like Harman, however. Hare does not use
18See R. M. Hare, The Language of Morals (London: Oxford University
Press, 1952).
50
the latter claim to characterize the distinctive features of
evaluative judgments, but only to define them. Thus, even Hare's
account does not help to render the speaker-relativity of Harman's
proposal plausible.
Does speaker-relativity provide an explanatory advantage?
In the absence of an obvious reason for assuming that some
judgments are evaluated relative to the speaker, it seems natural
to consider the significance of including the speaker in
evaluating the truth of an inner judgment according to Harman's
proposal.
We have already observed that whether a course of action is
supported by the agent's best reasons will not generally be
accessible to the speaker. We have also observed that the
speaker, in making her judgment, is expressing both a belief in
the similarity of the motivating attitudes of the agent to her
own, and some belief about the agent's motivating attitudes by
themselves.19 Furthermore, we observed that the judgment could be
true even if the motivating attitudes shared by the agent and the
speaker were not the motivating attitudes responsible for
producing the best reasons for acting in accordance with the
judgment. What these observations suggest is that, whatever the
relation between the truth value of the judgment and the speaker's
19This is because the judgment could not be true unless there were no
incompatible actions for which the agent had better reasons. See
pp. 45-47.
51
beliefs about the agent's and her own dispositions to act, that
relation can most efficiently and intuitively be captured in a
discussion of the speaker's action of judging. It seems that on
Harman's account we would lose nothing and gain in clarity by
evaluating the truth of the judgment "A ought to do D” separately
from the truth of the claim “The speaker has reason to make that
judgment."
Would we really lose nothing?
Recall that by definition the speaker can only make inner
judgments about an agent if she believes that the agent shares
with her some motivating attitudes M. Even if these shared
motivating attitudes are sufficient to guarantee that reasons
generated by them have a high probability of turning out to be
best, there is no clear reason to think that others of the agent's
motivating attitudes are not equally capable of producing reasons
that are just as likely to be best. In such a case, any reference
to M or, more generally, to whatever the agent has in common with
the speaker, is apparently irrelevant to moral appraisals of the
agent's behavior. It seems equally irrelevant in cases in which
the agent does have the best reasons to act in accordance with the
judgment, but not on account of anything which she shares with the
speaker. Inner judgments, then, appear not to mark off a class of
moral judgments that are evaluated differently from others.
52
The point is borne out by ordinary practice. What we are
interested in when talking about moral judgments as such is not
usually who made them but whether they are correct. If someone in
a crowd calls out, "Abortion is wrong!" who it was who called it
out seems to have very little to do with our assessment of its
truth. That the judgment we are considering is an inner judgment
does not seem to induce us to shift our attention to the identity
of the speaker. Suppose, for example, that the heckler calls out
instead, “Everyone agrees that murder is wrong. No one ought to
perform abortions!" The verbose heckler certainly seems to be
making an inner judgment. Yet just as before, it seems natural to
think that her identity is irrelevant in evaluating her claim that
no one ought to perform abortions.20 Without further argument
there seems to be no reason to think that inner judgments are
evaluated relative to the speaker.
The speaker's judgment conveys information about the speaker
regardless of whether the truth of the judgment is evaluated
relative to the speaker. That a person makes a particular
judgment by itself conveys much information. The speaker is aware
of the agent's behavior, and takes an interest in it. The speaker
believes she knows and understands the agent to the extent that
she believes that the agent is capable of certain kinds of
20Nor does an intuition that moral judgments are relational support
speaker-relativity; even if moral judgments are thought to be
relational, they need not be thought to be speaker-relative rather than
agent-relative.
53
motivations. The speaker expects her moral judgments to convey
information about their subjects of interest to whomever she makes
them known. Those who hear the speaker's moral judgments can
assess them based on what they know about both their subjects and
the speaker. If they believe that the speaker is insensitive to
the strength of an agent's personal commitments, they may not
agree with her judgments. They may think that the agent ought to
do otherwise. And they may press the speaker to reconsider her
judgment. Since we can discuss the question of whether someone is
justified in making a judgment about an agent independently of the
question of whether the judgment is true, it is possible to
discuss the question of whether a speaker is justified in making a
judgment regardless of whether the truth of the judgment is
evaluated relative to the speaker.
Without some further argument, it is arbitrary and
counterintuitive to understand some moral judgments so that their
truth value depends upon the justification of the speaker's
beliefs.21 We seem capable of and interested in interpersonal
discussion about moral judgments, without regard for the agreement
or disagreement of the speakers with the agents about whom they
speak. It also appears that the justification of the speaker in
21There may, of course, be other reasons for thinking that the truth
value of some moral judgments depends upon the justification of the
speaker’s beliefs. The point being made here is that Harman's argument
is insufficient by itself to establish that moral relativism is entailed
by assuming the existence of inner judgments.
54
making particular moral judgments can be assessed and understood
on its own.
One might nonetheless insist that inner judgments are
intended to convey information about both speaker and subject, and
thus that inner judgments must be evaluated relative to the
speaker in order to preserve the information speakers intend to
convey. What I have argued is that Harman has given us no
compelling reasons to think that facts about either the speaker or
her motivating attitudes play a role in the evaluation of the
truth of an inner judgment that do not arbitrarily assume that the
truth of some moral judgments depends upon who makes them.
Is moral relativism suggested If we do not assume speaker-
relativity?
If we abandon the speaker-relative part of Harman's proposal,
we are left with a proposal that says, in effect, that judgments
satisfying (1) and (2) of the defining characteristics of inner
judgments (i.e., judgments that imply that their subjects are
capable of being motivated by relevant moral considerations and
have reasons for acting in the prescribed manner) are evaluated
relative to the motivating attitudes of their subjects. Without
the speaker-relativity, is moral relativism suggested by
relativity to motivating attitudes?22
22Recall that Harman does not propose that inner judgments are evaluated
relative to the audience; according to Hannan's proposal, only the
speaker's beliefs about the audience's motivating attitudes affect the
truth of the speaker's inner judgment (and not the motivating attitudes
members of the audience in fact have).
55
That judgments satisfying (1) and (2) are evaluated relative
to the motivating attitudes of their subjects implies that, for a
given course of action D, two judgments 'A ought to D' and 'B
ought to D', both satisfying (1) and (2), may have different truth
values. Thus, (H), the claim that no motivating attitude shared
by all agents can produce in an arbitrary agent reasons for acting
that motivate behavior apparently in accordance with some moral
principle, suggests that a form of individual moral relativism
might be true, at least formally; in particular, (H) suggests23
that if, for some agent A and course of action D, the judgment 'A
ought to D' is true and satisfies (1) and (2), then there is an
agent B such that the judgment 'B ought to D' satisfies (1) and
(2) and is false.
Claim (H) is not supported very convincingly, however. The
only support Harman provides for (H) comes from hypothetical,
unverified, and apparently unverifiable cases. We probably should
assume that (H) is true if we include Martians and Harman's
potential protected children of crime in our lists of kinds of
agents, as Harman has said. But Harman has not given us any
reason to suppose that (H) holds for the domain of real human
agents. That Harman has given such obscure examples is itself an
indication that even if (H) is true in some global sense, it is
23"Suggests" rather than "implies" because the implication relies on the
assumption that there are not two distinct motivating attitudes that
produce reasons that support course of action D in A and B respectively
that are each best relative to A's and B's motivating attitudes.
56
not true when its domain is the population of agents ordinary
people actually meet. Harman's examples and arguments therefore
do not suggest that ordinary agents differ in their motivating
attitudes in ways that would affect what they are moral obligated
to do.
Still less is it suggested that ordinary agents might ever
meaningfully make judgments that satisfy (1) and (2) about anyone
as different from them as the agents in Harman's examples. Even
if there were such agents, encounters with them are apparently
sufficiently infrequent that it might be difficult for an ordinary
person who encounters such an agent to realize that she has met
someone who shared with her nothing that produces reasons for
acting that motivates apparently moral behavior. Recognition of
such differences would have to precede informed judgments (that
satisfy (1) and (2)) about unusually motivated agents.24
Consequently, there seems to be little reason to think that a form
moral relativism holds for judgments that satisfy (1) and (2) made
by ordinary human agents.
It is important to notice that to argue that there is
insufficient evidence to support (H) in a way that suggests that a
24It is my suspicion that the need for prior motivational assessment is
the reason some people find Harman's Hitler and Stalin examples so
troubling. Whether we wish to call Hitler a bad man or an evil man, or
a man who "did the wrong thing" in ordering the slaughter of millions,
seems to depend on whether we are trying to understand the man, history,
human psychology, or moral linguistics, and to what extent we are
capable of understanding. Our beliefs about the motivating attitudes of
Hitler will depend upon what kind of entity we are trying to make him
out to be.
57
form of moral relativism holds for a class of judgments made by
ordinary human agents does not require us to deny that agents
sometimes find themselves unable to understand apparently immoral
behavior in other agents. One often finds oneself unable to
imagine how violent crimes can be committed by people who are not
insane. We might consider ourselves incapable of even imagining
how a serial killer could have planned and carried out her
confessed crimes. But even if this is true it does not entail
that we think that she cannot be motivated to act in apparently
morally permissible ways by anything which might motivate us. We
might, for example, think that the fear of certain and severe
punishment would have prevented her acts. And if we do think so,
we are denying that the serial killer is an example that supports
(H).
Without examples that are a little closer to home, the
generalized argument only seems to suggest that a form of moral
relativism holds if the domain of agents includes very unusual
agents. In the absence of evidence that such agents exist or are
likely to exist, such a claim is unlikely to seem disturbing.25
25In footnote 9 I suggested that a weaker alternative to claim (H) might
be adequate to sustain the intuitions Harman takes as data: the claim
that no motivating attitude shared by all agents can produce in an
arbitrary agent reasons for acting that when "present" always motivate
behavior apparently in accordance with some moral principle. I have
argued that moral relativism is not suggested even if Harman's examples
warrant the assertion of (H), i.e., even if they warrant assuming that
no motivating attitude shared by all agents can produce in an arbitrary
agent reasons for acting that motivate behavior apparently in accordance
with some moral principle. Consequently, if I am right, then neither
will moral relativism be suggested if Harman's examples only warrant the
weaker claim.
58
There is reason to suspect that Harman undercuts his own
argument, however. His overall aims seem to undermine the force
of his arguments linking the efficacy of moral considerations in
motivation and moral relativism. It is important for Harman's
larger argument in support of his moral bargaining theory that (H)
be understood to impose a weak constraint. In particular, Harman
hopes to motivate his moral bargaining theory with universalist
intuitions about killing and letting die. He uses the fact that
letting die is intuitively preferable to the vast majority of
people to give grounds for preferring his moral bargaining theory.
In doing so, Harman apparently concedes that even if his
relativistic thesis is correct, there will be some moral judgments
(satisfying (1) and (2)) that will be nearly universally
applicable. Ultimately, Harman seems to suppose that the form of
moral relativism he argues for is false when the domain of agents
is restricted to the vast majority of us. The argument in support
of Harman's moral bargaining theory is thus at odds with the
argument for his claim that the truth of moral judgments
(satisfying (1) and (2)) must be evaluated relative to the
intentions or motivating attitudes of their subjects. The latter
requires greater diversity in ordinary cases than is permitted by
his reliance on shared intuitions in the former.
59
Chapter Three:
How the Truth of Internalistic Judgments Depends on Their Subjects
Although the dependence of the truth of inner judgments on
the motivating attitudes of their subjects when combined with
Harman's defense of claim (H), i.e., that no motivating attitude
shared by all agents can produce in an arbitrary agent reasons for
acting that motivate behavior apparently in accordance with some
moral principle, does not suggest that a form of moral relativism
holds for inner judgments, I believe that an argument similar to
Harman's suggests that a form of moral relativism holds for
internalistic judgments. The aim of this chapter is to develop
this argument.
Harman's strategy is to claim that certain assumptions we
make about moral judgments affect the kind of model we can use for
evaluating the truth of those judgments. The part of his argument
that is successful seems to show that if we assume the Basic
Internalistic Premise (i.e., that there are some moral obligations
such that the fact that an agent ought to D entails she has
sufficient motive for doing D), then we are committed to a model
according to which moral judgments are formally evaluated relative
to the motivating attitudes of their subjects.1 I have argued
1That is, a model according to which moral judgments only hold of agents
who have certain motivating attitudes. The relativity is merely formal
in the sense that if all agents have these motivating attitudes, then
evaluating a moral judgment will not involve verifying that an agent has
the right motivating attitudes. In that case particular agents'
motivating attitudes will not affect the moral judgments by which they
are bound.
60
that Harman has offered us no reason to think that the formal
relativity to motivating attitudes implied entails any kind of
moral relativism. Defending the Basic Internalistic Premise
commits one to far more than relativity to motivating attitudes,
however. Resting the possibility of moral obligation (for
internalistic obligations) on the presence of sufficient motive in
particular agents commits one to a model for evaluating the truth
of moral judgments according to which moral judgments are formally
evaluated relative to everything that affects what agents have
sufficient motive for doing. Even though, by definition, all of
agents' motives have their sources in motivating attitudes,
whether an agent has sufficient motive for behaving in a
particular way depends on other things too; including, for
example, her roles, means, capacities, and particular interests.
Suppose, for example, that two people are considering undergoing
the same high risk medical procedure. Each has equal concern for
her own health, comparable knowledge of the risks involved, and
has been advised to have the procedure by the same doctors. One
trusts the doctors' advice, and has sufficient motive to have the
procedure; the other does not trust the doctors, and, rather than
have sufficient motive to have it, has sufficient motive not to.
The motivating attitude that is the source of both agents'
relevant motives is presumably the same, however: a desire to live
61
and be healthy. Agreement of motivating attitudes is not enough
to guarantee agreement of motives.
Though it was difficult for Harman to argue that human agents
differ in their motivating attitudes in ways that affect whether
they have motives for acting in accordance with moral principles,
it is not difficult to see that agents differ in their roles,
means, capacities, and particular interests in ways that might
affect whether they have sufficient motive for acting in
accordance with moral principles. Rather than seeking after moral
monsters, we have only to look at each other. Suppose, for
example, the two people in the case just described were making
decisions about whether someone else--an incapacitated relative,
perhaps--should undergo the high-risk medical procedure. These
decisions certainly would have moral content. Regardless of how
we would judge the behavior of these agents, their behavior seems
to be of a sort that is likely to be considered morally
permissible or impermissible. It is also clear that the two
agents imagined differ in ways that affect what they have
sufficient motive for doing.
According to internalistic theories, there are moral
judgments that may only be true if their subjects have sufficient
motive for acting in accordance with them. Consequently, an
internalistic theory must evaluate the truth of moral judgments
according to a model that is responsive to differences among
62
agents that affect whether they have sufficient motive for acting
in accordance with moral principles— differences such as in their
capacities to trust others, as in the above example, or in their
religious beliefs, or in their conceptions of their own
limitations, etc. Such differences among ordinary human agents
are genuine and common. Holding an internalistic theory thus
seems to commit one to a model for evaluating the truth of moral
judgments that is not only formally agent-relative, but really
agent-relative; real human agents differ in ways that affect
whether the same moral judgments can hold of them.
It is thus natural to think that internalistic moral theories
are committed to a stronger form of agent-relativity than mere
relativity to motivating attitudes. I will argue that plausible
moral theories that posit internalistic obligations must sanction
a form of agent-relativity that is stronger than mere relativity
to motivating attitudes. To this end, I construct a general model
of internalistic theories in the next section. Once I have
examined how agent-relativity arises from the Basic Internalistic
Premise, I will return to the subject of moral relativism and
constraints like (H)-
A model of plausible internalistic moral theories
We will be concerned here with moral judgments that report
internalistic obligations (i.e., obligations that imply that
agents bound by them have sufficient motive for acting so as to
63
fulfill them), rather than with the subclass of inner judgments or
judgments that merely imply that their subjects have motives for
acting in accordance with them. In response to the doubts raised
in chapter two about the justification for assuming speaker-
relativity, I will assume that judgments that report internalistic
obligations (hereafter, internalistic judgments) are not speaker-
relative. In order to facilitate discussion, I will begin by
accepting a version of the first part of Harman's proposal for
internalistic judgments: that an internalistic judgment should be
characterized as a judgment that an agent's best reasons for
acting support her following some course of action which is
available to her. This amounts to making certain minimal
assumptions about motivation, including: (i) that there is some
way of ascribing to agents some kind of (motivation-affecting2)
reasons for acting, and (ii) that reasons for acting guide or
produce actions when we say actions are the result of
deliberation.3 Since it is beyond the scope of this project to
engage in the difficult and independent project of proposing and
defending an account of motivation, this project simply assumes
2See above, pp. 33-34 (chapter two).
3Claim (ii) does not require that reasons for acting have a causal role
in action guided by them. For example, suppose A buys gold because A
believes it will benefit her financially to do so. We might have an
account of reasons according to which A's behavior, which is the result
of deliberation, is guided by a reason for acting such as "an
inclination to buy gold (at time t) out of an interest in maximizing
financial gain," even if we believe that at t only confidence in her
earlier reasoning and excitement might reasonably be said to cause A's
behavior.
64
these limited claims.4 I will avoid taking stands about how
actions are described and individuated for the same reason. For
the purposes of this project, it will not be necessary to have
such accounts. When it is said that an agent’s best reasons
support following course of action D, it will suffice to interpret
this as claiming that if the agent were to act “on her best
reasons" (that is, if she were to be guided by her best reasons5),
then she would engage in behavior which would fall under the
description of D. It will not be necessary to describe what an
agent is trying to do, or her representation of her choices,
beyond in this limited sense.
Before proceeding, it will be helpful to notice that we
usually make moral judgments when it is in virtue of some features
of an agent (including, for example, circumstances and motivating
attitudes) that we think she ought to follow some course of
action, that is, in cases in which we would think that anyone
similar to her in having certain features would have best reasons
to follow that course of action. Moral judgments, including
internalistic judgments, may in this way be thought to depend upon
universal claims. If so, then there are two parts to the
4It may be useful to notice that the proposal does not rely upon the
assumption that to act morally an agent must act on the best of her
reasons. We do assume that if she did act on her best reasons she would
do what she ought to do, but that does not preclude the possibility that
she might do what she ought to do by acting on other reasons.
5As was remarked above, what exactly it means to say that an agent's
action is guided by some particular reasons will be specified by a
particular theory. Note also, however, that we are assuming that
theories of the sort we are considering will provide some such account.
65
evaluation of an internalistic judgment: the first, whether agents
who have certain features ought to follow some course of action
(where circumstances, including the deliberative situation itself,
are taken to be among the features of an agent), the second,
whether the agent in question has these features. This section
will focus on how particular internalistic judgments must be
evaluated by assuming that we know enough about agents to judge
their behavior accurately. While we do this, however, we should
remember that if it is in virtue of particular features of the
agent that an internalistic judgment about her is true, then it
should make sense to think of an internalistic judgment as
following from an instance of a universally applicable schema 'If
x has certain relevant features Fx, then the best of x's reasons
support her following D'. The significance of how internalistic
judgments are evaluated to the relationship between internalistic
moral theories and moral relativism may lie mostly in these
schemata (hereafter, internalistic schemata). I will put off
discussing internalistic schemata until chapter four.
We begin, then, with the non-speaker-relative proposal that
an internalistic judgment that an agent ought to follow some
course of action should be characterized as asserting that the
behavior specified is supported by the best of the agent's reasons
for acting. Following the methodology discussed in chapter one,
our strategy will be to first build a model of how internalistic
66
judgments are evaluated on plausible internalistic theories, and
then to isolate features of an agent upon which the evaluation of
an internalistic judgment about her depends.
The Basic Internalistic Premise,
(BIP) there are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for
doing D,
by itself provides little insight into how internalistic judgments
are evaluated, and hence, little insight into whether or how
evaluating them depends upon individual agents. The Basic
Internalistic Premise is also a very limited claim, because it
says nothing about how it is that the existence of a moral
obligation entails the existence of sufficient motive for acting
to satisfy it. For this reason, the Basic Internalistic Premise
is also not especially compelling when taken by itself. For an
internalist moral theory to be at all plausible, it must provide
an explanation of how the Basic Internalistic Premise can be true
and a defense of it.6 A defense of the Basic Internalistic
Premise will also yield some information about what capacities are
required for an agent to be bound by an internalistic obligation.
The most obvious explanation of why the Basic Internalistic
Premise is true we have already seen is blocked: since an agent's
being bound by an internalistic obligation does not require that
she think she is obligated, we cannot support the Basic
6The basic premise of internal ism obviously fares no better when taken
independently.
67
Internalistic Premise merely by arguing that thinking one is
obligated functions as a sufficient motive for acting to satisfy
such obligations.7
W. D. Falk's version of the internalist's assumption differs
from the Basic Internalistic Premise by yielding a
characterization of internalistic moral theories that gives us a
start at an answer to how we can support the Basic Internalistic
Premise.8 Recall that Falk considers the defining assumption to
be
(F) some moral 'oughts' are motivational 'oughts’, i.e., some
moral obligations are such that reflecting on an action
that would fulfill it [and on the situation] is capable of
determining an agent who is obligated to do so.
Given Falk's account of dispositional reasons, (F) tells us that
if an agent motivational-ought to engage in some behavior, then
she has sufficient motive for doing so such that further
7See chapter one, especially footnote 11.
8There seems also to be an obvious difference between Falk's
characterization and that given by the Basic Internalistic Premise: if
an agent (motivational-)ought to D, then according to the Basic
Internalistic Premise she has sufficient reasons to D, whereas according
to (F) she merely has some kind of reason to D. This difference is only
apparent, however, because of Falk's definitions of dispositional and
occurrent reasons. If according to Falk's terminology the agent has an
occurrent reason to D, then she has reasons sufficient to move her to D,
because her having an occurrent reason to D means that she is being
impelled toward it. If the agent has a dispositional reason to D, then
she has reasons sufficient to move her to D, because her having a
dispositional reason to D means that she is capable of being moved to D
by reflecting on D. Falk understands by both kinds of reasons
sufficient reasons.
The acceptance of the idea of dispositional reasons obviously may
complicate our understanding of 'sufficient reasons'. Apparently, one
has sufficient reasons for doing anything it is possible one will choose
to do. If I choose to eat your lunch because it's available and I am
hungry, then I have sufficient reasons for doing so. I also have
sufficient reasons for not doing so, if further reflection on the
behavior (and situation) would lead me to refrain from it.
68
reflection on her situation and on the behavior would lead her to
engage in it. It will be useful to consider Falk's account
briefly as a case study of how one might defend the Basic
Internalistic Premise.
On Falk's account, morally sanctioned courses of action are
those that an agent would find most choiceworthy after honestly
and conscientiously reflecting on them. The process of reflection
has two parts, which need not be thought of as occurring
independently: ascertaining facts about one's situation, and
reflecting on these facts.9 The first is a process of discovery.
Its aim is an adequate characterization of one's situation. The
second process is a "fact-supervening cognitive step"10 that is
more like tasting than discovering. Falk explains this in
interpreting Hume:
[I]n evaluation one reviews the nature of the
case not in order to clarify further either its
logical or empirical properties; but to extract,
from all that one knows it to be, one's response
to a proper view of it. One passes from the
facts to the merit by obtaining perceptual
evidence for the merit from the effect on one of
reviewing them.11
Ascertaining moral facts involves a third kind of activity, consisting
of ascertaining facts about a reflective person's reactions to the
situation at hand.
10"On Learning about Reasons," in Ought, Reasons, and Morality: the
Collected Papers of W. D. Falk (Ithaca, New York: Cornell University
Press, 1986), p. 68.
11The passage is from "Hume on Is and Ought," P- 135. This position is
expounded with much greater detail in "Fact, Value, and Nonnatural
Predication." Both are in Ought, Reasons, and Morality: the Collected
Papers of W. D. Falk.
69
To ascertain whether one has a moral obligation, one needs to
reflect on one's situation and the courses of action between which
one must choose. But this is still not enough. One must then ask
whether one's reflective decision is what Falk calls "formally
complete", meaning that further reflection will not change one's
appreciation of its choiceworthiness. Only those choices that are
formally complete will indicate moral obligations. If one is a
mature and intelligent person, then one is capable of engaging in
both kinds of reflection adequately, and hence of finding morally
sanctioned courses of action most choiceworthy. One may of course
fail to do what one ought to do, despite having this capacity.
But a mature and intelligent person is only capable of failing to
act morally if she is not reflecting adequately on her choices
(relative to the standard of formal completeness).12
The morally sanctioned courses of action are the only courses
of action that would result from formally complete reflection.
Consequently, acting to advance the interests of oneself or of
those one loves, when conceived as something contrary to morality,
is not an alternative outcome of formally complete reflection.12
12That one can try to accomplish a task, and fail to do so because of
external influences would not present an objection to this position. An
internalist who follows Falk's account would presumably count a
conscientious attempt to accomplish the task (as opposed to, say, one
involving self-deception) as the morally required behavior, rather than
the task's successful completion.
12There is room to criticize Falk on this point. One motivation for
moving to a position more general than Falk's is that we might side with
Butler and Sidgwick against Falk on this point; that is, we might claim
that even after an arbitrary amount of reflection, one might choose to
70
On Falk's account, there will only be one class of courses of
actions each of which would be found choiceworthy on reflection.14
Falk's account defends the Basic Internalistic Premise on the
grounds that (a) recognizing (or knowing, thinking, or feeling)
that something is good can be motivationally efficacious, and (b)
there are some good things such that we are able to recognize them
as good.
Other moral theories also assert versions of claims (a) and
(b), and hence, can support the Basic Internalistic Premise.
Perhaps the most straightforward apparent example would be a
Kantian moral theory. Although it is not clear that Kant's own
view is internalistic, an internalistic Kantian moral theory that
would use versions of claims (a) and (b) to support the Basic
Internalistic Premise might claim that if a rational agent is
act to advance one's own or another's interests, even if it conflicts
with moral duty.
14This is not to require that there is exactly one course of action that
is choiceworthy in any situation. However, in every situation there is
presumably supposed to be a single class of noncontradictory actions
that are choiceworthy, in the sense that it is not supposed to be the
case that after formally complete reflection an agent might either (i)
consider it best to D and not to omit D-ing, and (ii) consider it best
to omit D-ing and not to D. The problem with allowing both (i) and (ii)
to be outcomes of formally complete reflection is that it requires us to
allow that an agent could come to contradictory conclusions after
reflecting an arbitrary amount. There is no difficulty in assuming that
an agent could come to find choiceworthy after formally complete
reflection either of two courses of action that are incompatible in the
sense that they preclude each other. For example, if my rent is due
tomorrow, no amount of reflection will decide the question of whether to
pay it in the morning or the afternoon--each is equally reasonable, even
though I cannot both pay it in the morning and in the afternoon.
However, there would be something wrong with the way I was reflecting on
my situation if no amount of further reflection could alter my judgment
that it was best to pay my rent at precisely 10 AM, but I could have
reflected in such a way so that I would judge that I should pay it at 4
PM such that no amount of further reflection would change my judgment.
71
morally obligated to engage in some behavior, then she is capable
of seeing that it is right and experiencing respect for the moral
law. Given that respect for the moral law is at least as powerful
as any other motivating attitude, these abilities together would
ensure that the agent has sufficient motive for engaging in the
specified behavior, i.e., that she is capable of choosing to
engage in it. Less contentiously, though also less
straightforwardly, Butler, Hume, and Mill can be viewed as
defending the Basic Internalistic Premise on the basis of claims
(a) and (b).
There are other ways one might defend the Basic Internalistic
Premise.15 I will focus on moral theories that use versions of
claims (a) and (b) to defend the Basic Internalistic Premise
because these include the most commonly advanced internalistic
moral theories, and because the challenge presented by Harman's
argument for moral relativism has the most force against such
theories. Moral theories that use versions of claims (a) and (b)
to defend the Basic Internalistic Premise claim that being bound
by an internalistic obligation requires that those of one's
15One way is to argue that some good things are such that agents have
sufficient motive for acting so as to produce them independently of
whether they perceive their goodness. For example, consider a theory
according to which the satisfaction of desires is always good. Eating
when one is hungry is, consequently, a good thing on this view. One
might use such a theory to support the Basic Internalistic Premise by
claiming that whenever an agent is hungry, she has sufficient motive to
eat anything palatable that is available, simply in virtue of being
hungry, and independently of her being aware of the goodness of the
satisfaction of her desire for the food. The only well-known moral
theory that seems to offer this kind of account is Sidgwick's. This
account is discussed in chapter seven.
72
capacities that are relevant to how one can be motivated to act
satisfy certain conditions. Consequently, these theories seem to
be open to the possibility that some agents (those who lack the
specified capacity) will not be bound by internalistic
obligations. Since we do not commonly suppose there to be many
agents so excluded, a moral theory that had this consequence would
probably fail to meet the completeness restriction (i.e., whenever
we (intuitively) think that an agent lies under an obligation that
entails she has sufficient motive to fulfill it, that obligation
must be included in the class of internalistic obligations
specified by the theory16) . The theory would, thus, probably not
be plausible.
The rest of this chapter will be aimed at characterizing the
types of capacities that plausible internalistic moral theories
might ascribe to agents in order to support the claim that
recognizing that something is good can be motivationally
efficacious in a way that provides a defense of the Basic
Internalistic Premise.
For concreteness, let us consider a particular case.
Consider the following situation. An agent, X, plans to see a
movie one evening. A few minutes before she leaves, someone comes
to the door trying to raise money for the local soup kitchen. It
occurs to X that for the eight dollars the movie costs, the local
16See chapter one.
73
soup kitchen could feed five hungry homeless people. X only has
eight dollars and is on a limited budget, so donating and going to
the movies are mutually exclusive options. X takes a few moments
to consider the virtues of each of her options. She decides to go
to the movies. Let us now consider whether X ought to have done
so.
For internalistic theories of the sort we are considering,
knowing what an agent ought to do will require not only that we
know various things about the courses of action available to her
(for example, which of them manifest or produce what goods), but
also that we know various things about how she deliberates. We
will need to know, for instance, whether the agent is capable of
recognizing some relevant moral considerations as such. This, in
turn, will involve knowing about what motivates and can motivate
the agent (and under what conditions), as well as about her
capacities for detecting morally significant aspects of her
behavior, which may require a detailed understanding of her moral
beliefs. To determine whether the agent in our example, X, has
acted as she ought to have, will therefore require that we have
access to various facts about X that X herself may not even
have.17 For our purposes, it will be helpful to assume that we
17 There are several ways in which an agent might lack such information
when acting. For example, depending upon what view of motivation we
accept, an agent's reasons for acting might not necessarily be
completely "in view" at the time of acting. Also, the motivating
attitude producing a reason is not necessarily in view at the time of
acting; it may be that one can acquire a reason for doing something
solely in virtue of having appropriately disposed one's self at a prior
74
have such access. In making this assumption we must, of course,
keep in mind that this assumption is appropriate only as a tool
for understanding what aspects of an agent an internalistic theory
can rely upon to guarantee that she has sufficient motive for
acting to fulfill her internalistic moral obligations. Beyond
such uses, the assumption has no place in internalistic moral
theorizing; if agents have sufficient motive for acting to fulfill
the moral obligations that are supposed to be internalistic, then
such motives must be accessible from the agents' points of view.
In our example, the agent, X, considers her options and
chooses one, aware to some extent while deliberating of some of
her reasons for pursuing each course. To enable us to decide
which course of action was supported by the best reasons, our
moral theory presumably must provide us with a standard of the
goodness of agents' reasons or of an ideal deliberative procedure
that we can apply to X's reasons.18 Of course, moral theorizing
(especially by internalists) normally involves far more than this
assumption. Normally, such standards are embedded in a larger
time at which the motivating attitude from which the reason springs was
in one's view. (See footnote 3 of chapter three.) In addition, an
agent's understanding of her own moral beliefs or motivating attitudes
may be vague and unsystematic.
18Such a standard might but need not be given extensionally; for
example, given a particular list of n reasons for acting labeled
ri, r2 , .. ., rj, . .., rn, by "If an agent has reasons for acting ri,...,rj,
then r7 is her best reason for acting, and if an agent has reasons for
acting ri,...,rn, then rg and rj are her best reasons for acting (both
are equally good)." A nonextensional example can be lifted from a
cursory reading of Kant: "For any reasons for acting xi,...,xm, if an
agent has (exactly) those reasons for acting, then her best reason(s)
for acting is any x^ such that action on x* could be willed as a
universal law."
75
account of moral concepts such as moral value and personhood.
Nothing beyond the existence of some kind of standard for
assessing an agent's reasons (yet) seems to be required by the
Basic Internalistic Premise, however. Suppose, then, (following
the methodology proposed in chapter one) that we assume only that
we use some standard to rank our agent X's reasons for acting.
Suppose further that we decide that she ought to have given the
money to the soup kitchen. Of course, all of the reasons we may
use in our analysis are somehow present in X when she deliberates,
and she decides to go to the movies instead. What then can our
judgment indicate about how X might have chosen to act? More
generally, what does an internalistic judgment (i.e., one that
implies that an agent has sufficient motive to behave in
accordance with the judgment) imply about its subject's capacities
to be motivated?
That we are considering internalistic theories will force us
to reject answers that attempt to reduce capacities to be
motivated to mere capacities to act. For example, we can rule out
extreme accounts according to which the capacity to choose to act
would allow that if the chemistry of an agent's body could be
externally altered so that she would choose to act as she ought to
in a particular instance, the agent would be counted as being
capable of choosing to act as she ought to. The internalist
claims that internalistic obligations define behavior that is
76
special because the (mere) fact of obligation entails that the
agent is capable of choosing to act so as to satisfy the
obligation. On the extreme account proposed, that capacity would
not render actions that satisfy moral obligations special, because
altering the chemical balance of an agent1s body can cause the
agent to choose to act immorally as well.19 More generally, since
agents are typically able both to succeed and to fail in
satisfying their obligations, having the capacities implied by
being bound by an internalistic obligation must require more of an
agent than that she is physically able to do what she ought to.
It is also insufficient to claim that having the capacity to
choose to act so as to fulfill an obligation implies merely that
the agent would so act if she accepted or deliberated in
accordance with our standard. For suppose that while X is still
deliberating, she is told by her friend Y in the next room that
she has the best reasons for giving the eight dollars to the soup
kitchen. Let us suppose that Y does not explain how she came to
this conclusion, but simply tells her after the manner of some
moralists what she ought to do, that is, what the behavior
supported by the best reasons is. Assuming that X believes that Y
understands her reasons for both choices, could Y's assertion give
X any information which would directly affect her deliberative
assessment? Y's assertion might act as a kind of goading of X,
19As , for example, in the impairment of judgment associated with alcohol
or a stroke.
77
if, for example, if X likes to agree with, please, or impress Y,
but in that case it would add to X's reasons rather than make her
deliberation more informed. Y's assertion might cause X to wonder
whether she missed something, and therefore to reconsider. But
again, assuming that X considers the same reasons, her
deliberation need not be affected. Even if Y were to explain why-
each of X's reasons for acting should be taken into account
exactly to the measure assigned by the standard, that ranking
cannot directly affect X's evaluation of her reasons for acting.
X could come to accept the standard (or one similar to it), and
could therefore come to assess her reasons for acting in
accordance with it. Until X came to see her reasons for acting in
accordance with the standard's ranking of them, however, she could
always ask why she should order her reasons for acting the way the
standard does, and she could not productively be given an answer
using "obligation" or "best reasons" language. X's ability to
choose to act so as to fulfill her obligation would, consequently,
not be guaranteed by the fact of her obligation, but only by the
fact that X's way of choosing or deliberating is reflected by the
standard. Whether an agent chooses or deliberates in accordance
with any particular standard is thus not settled by the fact that
she is bound by a moral obligation. Since for the moral theories
we are considering, the fact of an agent's being bound by an
obligation by it self ensures that the agent has sufficient motive,
78
an agent's having sufficient motive cannot depend upon whether she
deliberates in accordance with a particular standard--that would
add an additional constraint.20
Before proceeding, it will be useful to explicitly set up our
model of plausible internalistic moral theories in order to be
clear about its limitations as we build it. The only positive
common feature of plausible internalistic moral theories that we
have so far been able to specify is that they will provide some
kind of standard which, in particular cases, can be applied to an
agent's reasons for acting so that we can say of her that her best
reasons support engaging in some specified behavior. This
suggests that a good first pass at building a model of these
theories would be provided by characterizing these standards.
Let us, then, characterize such a standard as a (partial)
function mapping sets of reasons for acting to sets of best
reasons for acting. An example of a standard (call this the
Gourmand Standard) might model the behavior of agents who eat
whenever they have an opportunity and want to do so; it might be
characterized by the set of ordered pairs of sets of reasons for
acting R such that, for some a, R includes an inclination to eat a
out of desire to eat a, and the singleton of an inclination to eat
a out of desire to eat a, i.e., {(set of reasons for acting R,
{inclination to eat x out of desire to eat x))I inclination to eat
20See chapter one.
79
x out of desire to eat x e R). Another standard (call this the
Simple Self-indulgence Standard) might model the behavior of
agents who eat whenever they have an opportunity and want to do so
and drink whenever they have an opportunity and want to do so; it
might be characterized by {(set of reasons for acting R,
{inclination to eat x out of desire to eat x, inclination to drink
y out of desire to drink y)) | inclination to eat x out of desire
to eat x e R, or inclination to drink y out of desire to drink y
e R). Let us say that a reason for acting ri is one of A‘s best
reasons relative to a standard S whenever S maps A's reasons for
acting to a set containing r^.21
We can use this model of standards for judging reason-guided
action to characterize what it might mean to say that an agent
deliberates in accordance with a standard. An agent deliberates
in accordance with a particular standard S just in case whenever
her behavior is reason-guided, she chooses to act in a way that is
supported by some reason that is best according to the standard S.
In terms of the model, an agent deliberates in accordance with a
particular standard S if and only if whenever her behavior is
reason-guided, some reason in the set to which S maps the set of
her reasons for acting appears to support her behavior. An agent
who always tries to eat when she is aware of something she wants
21Recall that we have not assumed that reasons for acting are limited to
an agent's desires for a particular thing. (See, for example, pp. 64-65
(chapter three).)
80
to eat can be said to deliberate in accordance with the Gourmand
Standard (even if we would not call her a gourmand) . She might
also, as it turns out, try to act in such ways as would warrant
saying that she deliberates in accordance with the Simple Self-
indulgence Standard. Nothing excludes the attribution of multiple
standards to an agent in this way.
There are traditionally two senses in which it is said that
an agent accepts or conforms to action-guiding principles: it is
sometimes supposed to signify that an agent believes a group of
such principles are true, and sometimes that an agent tries to
regulate her behavior in accordance with such principles. The
first corresponds to an agent's having a certain moral outlook,
and the second, either to an agent's trying to live up to her
moral outlook in practice, or, if the agent does not accept the
outlook she is trying to conform to, to her trying to behave in
ways that might appear to be signs of her having that moral
outlook. Our characterization of deliberating in accordance with
a standard may at first appear to be of the second kind. In fact
it is not of either kind, because standards in the model are not
action-guiding, but are rather functions that we interpret as
action-describing. To say, for example, that an agent deliberates
in accordance with the Gourmand Standard, is to say that she eats
whenever she has an opportunity and wants to, and not that she is
always motivated in such cases by a desire to eat something (she
81
may sometimes be motivated primarily by a curious palate, a desire
to be polite, compulsion, etc.). I will reserve the expressions
"standard" and "deliberates in accordance with a standard" as
technical terms here, and use other expressions, like "action on a
principle" and "follows rules", when more elaborate, action-
guiding ideas are meant.
On the surface, the model might seem to have some obvious
problems. Perhaps the most straightforward involve time. When is
it the case that an agent deliberates in accordance with a
standard? If a standard accurately describes her (trying)
behavior up to now, is that enough to say that she deliberates in
accordance with it? Our need for the idea of deliberating in
accordance with a standard is sufficiently meager that we will not
need to give decisive answers to these questions. If we must
insist on attributing a particular well-defined standard to a
person then it is clear that Saul Kripke's well-known
reconstruction of Wittgenstein's critique of rule-following poses
troubling questions that may undermine this project.22 The model
allows us to avoid this at the cost of not being able to specify
any particular standard as the one that an agent deliberates in
accordance with--there will always be a (large) infinite class of
standards that will be appropriate. If there is any need for more
precision, it may perhaps be adequate to stipulate that an agent's
22See Saul A. Kripke, Wittgenstein on Rules and Private Language
(Cambridge: Harvard University Press, 1982) .
recent previous evaluations of her reasons for acting (at least in
cases in which there is no reason to think a substantive
psychological or physical discontinuity has occurred) are
manifestations of one vaguely defined standard. The model allows
us to refrain from taking a position on whether one's having or
thinking one has a particular set of moral beliefs has any impact
on what standard it is reasonable to say that one deliberates in
accordance with according to our limited descriptive, motivational
usage, or, in other words, on how the standard one deliberates in
accordance with reflects the action-guiding principles one thinks
reasonable or correct. The model also does not require us to take
a position on whether those who believe they share moral beliefs
deliberate in accordance with the same standards, although we
should expect them to deliberate in accordance with recognizably
similar standards. Must an agent always behave in accordance with
a standard to be said to deliberate in accordance with it? (Such
behavior will, of course, always be internal.) This question we
can answer: yes. The model requires it. But the idea of a
standard is extremely fuzzy, and it need not be the case that a
particular standard selects a single course of (internal) action
in any instance. Nor, as was remarked above, need there be an
answer to the question of whether two descriptions of standards
refer to the same standard.
83
Another aspect of the model that might at first seem puzzling
is its exclusive focus on successfully (internal) action-
motivating instances of orderings of reasons and ends. It may
seem illegitimate to focus on these if one thinks that no clear
difference in disposition can be discerned between the case of
one's thinking certain reasons or ends better than others, and
that of one's having acted so as to prefer acting in accordance
with some reasons for acting rather than others. The model does
not require us to take a position on this issue, however. It is
true that according to our characterization of deliberating in
accordance with a standard we should not be concerned with
nonmotivational orderings (including orderings of reasons and ends
which are not reasons for acting) when we consider what kind of
standard an agent deliberates in accordance with. This is because
we are analyzing a claim about reason-guided action, i.e., about
motivational orderings of reasons and ends. We should try to do
so with minimal assumptions about how agents order their reasons
and ends in general. As a consequence, the model focuses on a
motivational notion of deliberating in accordance with a standard
that may well turn out to be subsumed by a broader notion in
theories that follow the model.
One may also wonder why the model takes deliberating in
accordance with a standard as an action-describing notion rather
than an action-guiding one. It might be thought that we should
84
respond to the stronger intuition that only actions which are
motivated (in part) by the thought that something is good can be
good actions.23 One might think that by approaching standards
descriptively we will not make enough room for a significant role
for thinking something good. Our model does not, however, leave
too little room for such a role, because it does not require us to
suppose that the Basic Internalistic Premise is the only claim
upon which particular internalistic theories are built. An
internalistic theory might have much more to say about actions
that are caused by agents thinking them good. Consequently, a
significant role for thinking something good is not incompatible
with our model of internalistic theories.
Still, given our narrow, evaluative model, standards that
correspond to a moral theory that assigns a significant role to
thinking something good are unlikely to reflect the significance
of this role. Since standards are mappings of sets of reasons for
acting to sets of reasons for acting, they cannot
straightforwardly manifest a role for thinking something good
unless some reasons for acting only arise from such thoughts.
Thus, it may be thought that by failing to distinguish between
actions on the basis of whether they result in part from an
agent's thinking something good, we fail to delineate the role of
23As was discussed in chapter one, this claim is independent of the
Basic Internalistic Premise.
85
thinking something good in appraising actions enough to ensure its
significance.
There are two reasons to think that it is better to refuse to
assume chat thinking something good must play a role in motivating
actions that have moral significance. The first is that even if
an action must be motivated in part by the thought that it will
produce (or is) good in order for it to be a good action, doing
what one ought to do is rarely thought to require such
motivation.24 Since some moral judgments are about whether an
agent dees what she ought to, rather than whether her behavior is
good, we should not limit our attention to theories about only the
latter jcinds of judgments. The second reason to prefer that the
model be more inclusive is that there are received moral theories
that warrant it. For some Humean moral theories, for example,
actions that arise through unreflective but consistent habit are
as good as those arising directly from a conscious effort to act
morally or to produce good. Even for a Kantian moral theory,
according to which actions arising from the thought of their being
right always have special significance, an agent can satisfy an
obligation by acting according to duty even if she is not guided
by the tfr°ught of duty.25
24For an argument that no such motivation can be required, see W. D.
Ross, The Right and the Good (Indianapolis: Hackett Publishing Company,
1988), p p . 4 -5 .
25I do not wish to suggest that Kant's view is internalistic. Nor do I
wish to suggest that if his view is internalistic that on it an agent
can act a.s she ought to if she is unaware that the course of action she
pursues is her duty. My point is rather that Kant's ethics might
86
Now let us return to our example. If the observer's judgment
that the agent, X, ought to donate eight dollars to the soup
kitchen is a true internalistic judgment, then the agent's best
reasons support donating the money in a sense that implies that
she has sufficient motive for doing so. Since the agent's having
sufficient motive does not depend solely on her motivating
attitudes, it might seem that, on the accounts we are considering,
the agent cannot be obligated to donate the money unless she
already has sufficient motive for doing so— or, at least, unless
she seems, typically, to deliberate in accordance with our
standard. In either case it seems that
(A) X can only be obligated if she deliberates in accordance
with a standard similar to our standard.
This claim would be an unpalatable result for any moral theory.
For internalistic theories, it is especially difficult, however,
because it undermines the possibility that the Basic Internalistic
Premise states a deep fact about certain moral relationships.
Consider, for example, the two internalistic judgments "Miss
Manners ought to tell you to accept the black coral bracelet" and
"Miss Manners ought not to encourage commerce in endangered
species such as black coral by telling you to accept the
bracelet." Knowing that Miss Manners has in fact encouraged
sanction an internalistic moral theory according to which the reasons
which actually guide agents to discharge their obligations are
independent of what makes courses of action right. Internalistic
judgments (in the subtheory) would not be about agents acting
virtuously, but permissibly.
87
people to accept gifts of black coral and others made at the
expense of other organisms she knows are endangered, we should
expect that the first is very likely to be what Miss Manners will
do, hence, on this reading, the only one of the two likely to be a
true internalistic judgment. Of the two, the second judgment
might be thought to be more likely to be a true internalistic
judgment,26 but it cannot be true if claim (A) is. Given claim
(A), the motivational implications of (internalistic) moral
obligations do not reflect any special features of moral
obligations as such, but merely that in particular cases the
agent's behavior coincides with her moral obligations.
Internalistic judgments could have no normative force (in virtue
of being internalistic judgments) if claim (A) is true. Since we
generally do expect our internalistic judgments to be normative, a
theory that had claim (A) as a consequence would, therefore,
certainly fail the completeness test, and thus would fail to be a
plausible internalistic theory.
Fortunately for the prospects of internalistic theories, a
common and plausible assumption provides a basis for a strategy
for avoiding claim (A). This assumption is that moral judgments
make claims about agents' dispositional reasons. Recall (from
chapter one) that W. D. Falk characterizes dispositional reasons
26Particularly if the means of producing the gift is considered
extremely repugnant. Consider the mutilation of an elephant for ivory,
or the murder of an intruder for a shrunken head.
88
as follows: an agent has a dispositional reason for engaging in
some behavior if reflecting on the action (and on the situation)
"is capable of determining him to do it." This is in contrast
with occurrent reasons, the presence of which indicates that the
agent is already "being impelled" by the thought of the action to
do it.27 Dispositional moral theories claim that an agent who is
bound by a moral obligation would have an occurrent reason for
acting so as to fulfill that obligation if she assumed a moral
point of view.2829 On such theories, agents bound by moral
obligations need not have occurrent reasons for acting so as to
fulfill them. As a consequence, claim (A) (i.e., the claim that X
can only be obligated if she deliberates in accordance with a
standard similar to our standard) is no longer suggested. More
precisely, dispositional internalistic moral theories can avoid
claim (A) by making three claims:
(i) An agent's best reasons relative to a standard S (i.e.,
reasons in the set to which S maps her reasons for acting)
support engaging in some behavior only if the agent is
capable of deliberating in accordance with S, and one of
her best reasons relative to S supports engaging in that
behavior (regardless of whether she deliberates in
accordance with S);
(ii) An agent's [unconditionally] best reasons support engaging
in some behavior if for any member S of a specified class
27Falk, "'Ought' and Motivation," pp. 25-26. See also p. 18 (chapter
one) .
28See, for example, David Lewis, "Dispositional Theories of Value",
Proceedings of Che Aristotelian Society Supplement 63 (1989), pp. 113-
137 .
29It might be helpful to notice that dispositional moral theories need
not be internalistic because having an occurrent reason does not
necessarily guarantee that one has sufficient motive for acting on it.
89
of standards the agent's best reasons relative to S support
engaging in that behavior; and
(iii) Standards in the class of standards in (ii) include all
standards an agent would deliberate in accordance with if
she adopted "a moral point of view" (where "a moral point
of view" is specified by each particular theory).
According to this strategy, the standards specified do not
necessarily reflect how agents in fact deliberate. Instead, the
standards reflect how agents would deliberate if they adopted a
moral point of view. Thus, the fact that someone does not
deliberate in accordance with our standard does not excuse her
from being morally bound to conform to it. Moreover, the
motivational content of internalistic judgments derives from one
of the distinguishing characteristics of morally obligatory
behavior--its choiceworthiness under certain conditions: those
under which an agent will deliberate in accordance with some
members of a certain class of standards.30
For illustration, consider an internalistic theory I that is
based on this strategy. Suppose that according to I, an agent's
best reasons support engaging in some behavior B just in case
under conditions Cj (which are given in general, agent-independent
terms) she would deliberate in accordance with some standard S
such that doing so would mean that she chooses to engage in that
30whether applying such a view to an agent is likely to offer guidance
to the agent or to provide insight into her behavior would depend upon
the agent's control over her coming to deliberate in accordance with a
standard similar to one specified by the theory. This may involve the
capacity to adopt "a moral point of view". It will certainly involve
capacities related to habituating oneself.
90
behavior.31 The agent has sufficient motive for engaging in
behavior B because she has reasons that support doing so, and
could engage in reason-guided action resulting in her doing so.
Since the agent may also have sufficient motive for acting in a
manner incompatible with engaging in behavior B,32 she might not
do so. If the agent does not engage in behavior B, then that may
suggest that she does not deliberate in accordance with a standard
similar to one specified by I. It certainly would imply that at
the moment of acting, she was not taking "a moral point of view",
i.e., she was not acting under conditions Cj. Neither consequence
would be incompatible with the claim that the agent had an
obligation to engage in behavior B, however.
The conditions that define the class of standards in (ii)
(i.e., what a “moral point of view" consists in) will vary widely,
depending upon the particular internalistic theory proposed. An
intuitionist might propose that an agent can discover the
choiceworthiness of (and therefore typically will do) what she
ought to do by intuiting the comparative worth of her options; a
Kantian, by assuming a purely rational stance; a consequentialist,
by being concerned only with the outcomes; a utilitarian, by being
only concerned with the outcome in terms of the general good; an
egoist, by not being directly concerned with the general good (and
31In general, there may be several courses of action supported by an
agent's best reasons. For simplicity, I assume that in this particular
case behavior B is the only one.
32E.g., for inaction.
91
by being concerned directly only with oneself). In general, one
can give a description of the required conditions by constraining
either the substance or source of the reasons for acting that the
agent may be motivated by, or by constraining the faculties that
an agent may use in deliberating.
It would of course require a demonstration of the
intelligibility of each of these descriptions to ensure that they
legitimately follow the model. For example, the intuitionist
would have to show that her view is both well-defined and
internalistic; she would need to explain the faculty that must be
used to be in the proper state for intuiting moral worth (or how
we may know when we are intuiting moral worth), and to demonstrate
that we really do what is right when we (truly) perceive it. The
appropriate conditions must be noncircularly specifiable; in
particular, the theory cannot define the conditions as those such
that agents who adopt them act as they morally ought to, because
the conditions are supposed to be those that enable us to
determine what we morally ought to do.
In none of the examples of moral theories that I have listed
do the conditions that define moral points of view directly
specify the particular standards that agents will come to
deliberate in accordance with if they assume such a perspective.
The form of the standard is constrained in some cases; for
example, in the utilitarian case, if deliberating in accordance
92
with a particular standard quite clearly does not promote the
general good, the standard will certainly not be one in accordance
with which an agent who adopts a (utilitarian) moral point of view
would deliberate. It is conceivable that a moral theory might
depend upon restricting the particular standards agents may
deliberate in accordance with directly, however. For example, a
moral theory that characterizes morally obligatory behavior as
conformity with the doctrine "follow the dictates of our church".
This view should be distinguished from those characterized by
doctrines such as "faith will show you what to do“ and "pray with
us and you will come to know the way” in which the conditions for
discovery define the views. The type of theory of interest here
is based on an extreme version of the doctrine "follow the
dictates of our church": "what everyone has best reasons for doing
is what our church administration says they have best reasons for
doing”. Moral theories like this one, although they appear to
correspond to a very narrow class of standards, pose no new
problems, though perhaps more of the same problems. For an
adherent of such a theory, to make an internalistic judgment about
an agent is to judge that she is capable of deliberating in
accordance with a standard similar to one that corresponds to the
theory. That, in turn, is to say that there are conditions under
which the agent would deliberate in accordance with an appropriate
standard. Of course, specifying these conditions might be much
93
more difficult because of the specificity of being "an appropriate
standard".
General observations about how evaluating judgments according to
this model depends upon agents
As sketchy as it is, our model of plausible internalistic
theories is sufficient to provide some insight into the agent-
relativistic consequences of internalistic theories. Before
considering how it does, it is worth noting that even Harman's
argument takes notice of a kind of agent-relativity besides
relativity to motivating attitudes: relativity to circumstances.
Although the moral force of an agent's circumstances are not
always made explicit in internalistic judgments, circumstances are
usually implicitly restricted. For example, the judgment “You
ought to keep your promises," can be paraphrased as the more
obviously circumstance-relative judgment "Since you made a promise
(and it is possible to fulfill it, etc.), you ought to do what you
promised." Sometimes the moral content of circumstances that
warrant a moral judgment is more difficult to isolate even when
the circumstance-relativity is clear: "Since you are the only
child of ailing and single Z and you live in a society in which
children are expected to care for their ailing parents, you ought
to take responsibility for the care of Z", "Since there is no one
else on the platform, you ought to help the child off the tracks",
"Since the homeless man has asked you for food and you have some,
94
you ought to give him some", "Since you have benefited from the
hospitality of your host, you ought not to deride the food".
Besides that the sources of the circumstance-relativity in
internalistic judgments vary, what the latter examples demonstrate
is that the relativity of internalistic judgments to circumstances
is not dependent upon those circumstances having been wanted or
chosen by the subject of the judgment.
Circumstance-relativity may seem to require only an
insignificant kind of agent-relativism, since any moral theory
that recognizes conventionally-derived moral requirements (e.g.,
promising) will have to be circumstance-relative. It is
ordinarily assumed that circumstance-relativity can be
accommodated within universally applicable moral judgments. It is
nonetheless a kind of agent-relativity, albeit a kind that is
generally recognized even by canonically nonrelativistic moral
theories. We should take notice of it here because what we are
trying to isolate here are the formal requirements for evaluating
internalistic judgments--that is, those features of agents that
could cause internalistic judgments about some agents to be
evaluated differently from internalistic judgments about other
agents (whether or not these features in fact have this
consequence). Once we have isolated the ways in which
internalistic judgments are formally dependent upon their
subjects, we will be in a better position to see whether
95
internalistic judgments must be evaluated relative to individual
agents in practice. (I will consider the significance of
internalistic theories' relativistic consequences (whether, for
example, they sanction "moral relativism") beginning in chapter
four.)
In the last section, I argued that internalistic theories
that follow the model being developed can push back the question
of which agents are excluded from being bound by internalistic
obligations so that the obviously problematic exclusions (those
that were suggested if an agent can only be obligated if she
(already) deliberates in accordance with a particular standard)
are avoided. The model does not, however, allow us to push back
the question so far that we can be sure that theories that follow
the model deal with it plausibly. In each of the theories
described above, in order for an internalistic judgment to be
true, the subject must be capable of viewing the world (in
particular, her reasons for acting) in some way. She must, for
example, be capable of adopting a frame of mind or a moral stance
in which she would choose to behave in accordance with the
judgment (i.e., some "moral point of view" as given by a
particular theory). What exactly that says about an agent depends
upon which theory one accepts. Regardless of the details,
however, requiring that an agent be capable of adopting such a
stance requires that she have some particular characteristics for
96
the judgment to be a true internalistic judgment. Whether an
agent is bound by an internalistic obligation will depend upon her
having certain motivation-affecting capacities. On an
intuitionistic account, for example, an agent who lacks the
ability to perceive goodness will not obviously have sufficient
motive for acting to keep a promise. On its face, this would
suggest that such agents could not be bound by internalistic
obligations on such an account. On a Kantian account, certain
habitually weak-willed agents might be excluded; if there is no
way that an agent can view her situation and choices such that she
would typically act in accordance with the account's conception of
rationality, then we have no obvious grounds for thinking that she
has sufficient motive to act in accordance with rationality.
Regardless of the particular motivational capacity the
account relies on, all of these accounts are committed to some
form of agent-relativity from the outset. This is because for any
internalistic theory, the possibility of an agent’s deliberating
in accordance with a particular standard depends on the agent's
ability to deliberate in accordance with a standard at all, i.e.,
on her being able to order her reasons for acting in some way.
Consequently, on any internalistic theory the truth of an
internalistic judgment will depend on its subject having this
ability. For example, someone constantly paralyzed by indecision
is not a possible subject of an internalistic judgment because
97
although she may have the power to act on a decision were she to
make one, she does not decide, and therefore does nothing
(including standing still) that is the outcome of deliberation; if
an agent is incapable of deciding what to do (or not to do), then
she is incapable of deliberating in accordance with a standard.
At the same time, it seems that we need not require that an agent
have the power to successfully act according to the outcome of
deliberation to reasonably say that she deliberates in accordance
with a standard. This would depend upon what we understand to be
the outcome of deliberation.33 Furthermore, it is not necessary
that an agent have many choices. Sisyphus, although he apparently
hasn't much to deliberate about except whether to push the rock,
can deliberate (about pushing the rock) and decide (to push the
rock). It might be said of Sisyphus that he deliberates in
accordance with some standard of ordering his reasons for acting--
though, of course, to be appropriate the standard would have to be
very limited in scope.
An agent's choices will affect what standard she can be said
to deliberate in accordance with, however. Whether it can
reasonably be said of an agent that she deliberates in accordance
with a particular standard will depend upon whether the agent
finds herself in circumstances to which the standard is relevant.
33In particular, it depends on whether we think that some unsuccessful
trying behavior is supported by the same reasons as corresponding
successful trying behavior.
98
In the case of Sisyphus, for example, if we suppose him to be
sustained eternally by the gods without any desire for, or
possibility of his eating, we cannot reasonably say of him that he
deliberates in accordance with a standard similar to the Gourmand
Standard. However Sisyphus orders his reasons for acting, any
standard in accordance with which he deliberates will not map
reasons for acting in the domain of the Gourmand Standard to sets
of reasons for acting other than those to which they are mapped by
the Gourmand Standard, but neither will it so map reasons for
acting in a standard complementary to the Gourmand Standard (i.e.,
one that ranks reasons for acting opposite to the first with the
same domain); Sisyphus' behavior is compatible with both the
Gourmand Standard and its complement. There are, consequently, no
grounds upon which to claim that Sisyphus deliberates in
accordance with a standard similar to one and not the other.
Whatever standard we might say Sisyphus deliberates in accordance
with, we cannot reasonably say that it is similar to the Gourmand
Standard. In general, whatever standard we might say that an
agent deliberates in accordance with, we cannot reasonably say
that it is similar to a standard whose domain includes no set of
reasons for acting which are ever jointly present in her.
For similar reasons, in the equally extreme case of an agent
who, in all of her deliberations in which a particular standard S
has something to say, never gets to finish deliberating before
99
action must occur, we should not say that such an agent
deliberates in accordance with a standard similar to S. Since
such an agent would never have an opportunity to order her reasons
for acting in any situations that are morally significant
according to S, her deliberative behavior would be compatible both
with S and a standard complementary to S, just as it was in the
Sisyphus example. Thus, (even if her ideology is regarded as
settled) we should not say that such an agent deliberates in
accordance with a standard similar to S. Some sets of reasons in
the domain of S will in some instances describe reasons for acting
jointly present in the agent, but in none of those instances does
she have the opportunity to try to act. Consequently, in none of
those instances can the agent1s behavior be used to determine
which of S and its complement is similar to a standard she
deliberates in accordance with. As we have already remarked, this
is not to say that an agent must have the power to successfully
complete her action following her deliberation. The point here is
that she must have the opportunity to try to do something, whether
or not she will succeed.34
34This point depends upon our motivational characterization of
deliberating in accordance with a standard. A broader view of
deliberating in accordance with a standard would require a slightly
different claim, depending upon how the broader view is cast. If
deliberating in accordance with a standard involved making certain
nonmotivational judgments, the point would be that to say of an agent
that she deliberates in accordance with a particular standard, we should
require that an agent have an opportunity to order her reasons and ends
to produce such judgments in some situations described by the standard.
100
It might be helpful to look at an example of how these
restrictions could affect ascriptions of standards to agents. Let
us consider, then, the case of a very odd, hypothetical agent: Z
is a harried mother of ten who hardly has a moment's peace to
think of anything but the immediate needs of her young children.
In particular, Z never engages in successful moral decision-making
at all, not even internally. We are to imagine Z as someone who
has become something of an automaton, who responds to her
children's needs without commentary on them--without, for example,
approval, disapproval, decision, or consideration of their
importance. Z's very condition, we are to suppose, to have been
foisted on her without her having had an opportunity to reflect or
pass judgment on whether it was a good thing. Let us suppose her
life to be composed largely of the most mundane sorts of events
(morally speaking) mixed with frustrated apparently morally
significant decisions along the following lines. The harried
mother of ten is in the house making peanut butter sandwiches
while nursing one of her infant twins. From the window she sees a
small child running toward a blind intersection with a busy
street, apparently intending to retrieve a ball which is rolling
into the intersection. Being in the corner house, she would have
the opportunity to stop the child if she were only to go outside.
She does not because at the moment she sees this, three of her
101
children come into the kitchen whining for milk, cookies, and
sandwiches, thus distracting her at the crucial moment.
This is admittedly an unlikely life, if not an unusual type
of event. This should not concern us, however. Our aim here is
merely to take note of some limiting conditions, and not to defend
moral relativism. Considering cases that are necessarily
hypothetical should, therefore, not trouble us unless we misuse
our intuitions about them.
One might imagine that agents of this sort are not
particularly interesting since ordinary agents are frequently in
situations in which they must act quickly, despite the fact that
acting so precludes their considered judgment of what they have
best reasons for doing. The predicament of the harried mother of
ten, Z, is not one of hurried decision-making, however. Z, we are
supposing, is prevented from engaging in anything that could be
considered moral decision-making at all. She does not quickly and
erroneously decide to ignore the child outside her window, she is
distracted from it. She sees that the child is very probably in
danger, but never has the opportunity to realize that she could by
going outside spare the child. We might say that her deliberative
situation in this case has no moral significance because she fails
even to acquire reasons to act so as to spare the child. Given
the composition of Z's life, this would be to claim that the first
sort of restriction is relevant to whether Z can reasonably be
102
said to deliberate in accordance with a standard (i.e., that an
agent must have reasons for acting in the domain of a particular
standard to reasonably be said to deliberate in accordance with
it). Alternatively, we might say that Z's deliberative situation
has moral significance (i.e., in this instance she has reasons for
acting in the domain of some standard corresponding to a moral
theory) but that she fails to successfully order her reasons for
acting. Given the composition of Z's life, this would be to claim
that the second sort of restriction is relevant to whether Z can
reasonably be said to deliberate in accordance with a standard
(i.e., that an agent must act in situations that have moral
significance with respect to a particular standard to reasonably
be said to deliberate in accordance with it). On either analysis
of the situations that compose Z's life, Z fails to order reasons
for acting in situations that have moral significance.
Z's momentary predicament, of distraction, when taken by
itself is a common one. But her experiences taken as a whole life
make for a very uncommon story. Although Z may be said to
deliberate in accordance with some standard, because of her
unusual fortune she cannot be said to deliberate in accordance
with a standard which corresponds to anything like a set of moral
beliefs, even if she believes that a particular moral theory is
correct. Unlike someone eternally paralyzed by indecision, Z
103
makes decisions, just not morally significant ones (with respect
to standards that correspond to moral theories).
Restrictions on the standards Z can reasonably be said to
deliberate in accordance with, given her unusual life, reveal a
further aspect of agent-relativity in internalistic judgments. An
internalistic judgment can only be true if its subject can
reasonably be said to deliberate in accordance with a standard
similar to one that corresponds to a moral theory (and can
deliberate in accordance with it). To be a possible subject of a
true internalistic judgment, not only must an agent be capable of
assuming some stance with respect to her reasons for acting, but
she must sometimes have a morally significant set of reasons for
acting, and in those cases some opportunity to order her reasons
for acting.
The case of Z is an extreme one. It is helpful to compare it
with that of an agent who finds herself in a more ordinary
condition: distracted in some but not all morally significant
situations. Suppose that U is an agent who considers herself a
hedonistic act utilitarian. U has habituated herself to being
motivated by the aim of promoting the greatest sum of pleasures as
computed by a particular calculus of pleasures. Consequently,
most of her behavior indicates that she deliberates in accordance
with some standard similar to a (particular sort of) hedonistic
act utilitarian one. Occasionally, however, U finds herself in
104
situations in which she is distracted before trying to act, even
though had she not been distracted she would have tried to act.
Although it is difficult to know how exactly we should regard
situations of distraction with respect to the attribution of
conformity to standards, our characterization of deliberating in
accordance with a standard requires that situations of distraction
be viewed as situations not in the domain of the standard the
agent (actually) deliberates in accordance with. When U is
distracted in situations that are morally significant from a
hedonistic act utilitarian standpoint, that may, therefore,
constitute evidence that she does not deliberate in accordance
with a hedonistic act utilitarian standard; for example, if U is
distracted repeatedly, and fails to do what she can to prevent
herself from being distracted in similar cases. However, U's
being distracted in such cases also cannot provide evidence that
she deliberates in accordance with a standard that corresponds to
some other moral theory according to which the situation is
morally significant. If situations of distraction provide any
evidence at all about what standard U deliberates in accordance
with, they can only provide evidence that she deliberates in
accordance with a standard relative to which the situation is not
morally significant.
For the present purpose, that is, to contrast the relevance
of distraction to the attribution of conformity to standards to U
105
and Z, it suffices to notice that however we deal with distraction
ultimately, it might be reasonable to say that U deliberates in
accordance with a standard similar to one corresponding to a moral
theory, but it would not be reasonable to say this of Z. For Z,
these situations (which turned out to end in distraction) are the
only ones in her life which might have been grounds for claiming
that she deliberates in accordance with a standard that
corresponds to a moral theory, because they are the only ones in
which a morally significant set of reasons for acting might occur.
Internalistic Theories are thus committed to a form of agent-
relativity that is stronger than relativity to motivating
attitudes. Like Harman's argument that agents differ in their
morally significant motivating attitudes, however, this argument
depends upon positing very unusual agents. Consequently, like
Harman's argument, this argument fails to support its conclusion
for real human agents; that is, it does not show that
internalistic moral theories are committed to claiming that real
human agents differ in ways that would affect whether they can be
said to take on a moral point of view. This argument shows only
that internalistic moral theories are committed to claiming that
being bound by a moral obligation is formally dependent upon an
agent’s capacities, circumstances, and reasons. Nonetheless there
is reason to think that plausible internalistic moral theories
106
will in fact be committed to claiming that real human agents
differ in ways that would affect whether they can be said to take
on a moral point of view.
The usual stock of examples held up in support of moral
relativism are helpful here: anthropological reports, mass
murderers, children raised in hostile conditions, etc. Real
people do differ in ways that affect what they are capable of
choosing in morally significant situations; for example, in their
reactions to violence.35 The point can be made even without
resort to the extreme or unusual, however. Two people who differ
in nothing more than their capacities to trust others may
reflectively and conscientiously choose to act differently in
morally significant situations. An example I used to motivate the
strategy pursued in this chapter demonstrates this. Consider,
again, the case of a trustful person, T, and a mistrustful one, M,
each deciding whether to follow her doctors' advice in allowing a
high-risk medical procedure to be performed on an incapacitated
family member. Let us suppose that T and M both value the same
sorts of things, and that we do as well. Nonetheless T and M
35Notice the contrast with Harman: his argument has trouble using these
cases because, as those who deny that such cases suggest moral
relativism claim, there may not be any reason to assume that the
motivating attitudes of apparent moral monsters are different from other
agents. There is reason to think that agents' reasons and circumstances
differ in comprehensive ways, however.
Observing ideological differences among agents also illustrates the
point. Is there even any question whether any amount of reflection or
deliberation will bring the staunchest pro-life Christian
fundamentalists and pro-choice feminists together on, say, a decision
about whether to allow an abortion to be performed?
107
confidently choose differently. T trusts the doctors' advice, and
authorizes the procedure, and M does not trust the doctors, and
hence does not authorize the procedure. They choose differently
because they perceive their situations differently. An
internalistic theory will allow us to evaluate the case in various
ways. We may, of course, claim that one or both of T and M has
not assumed a moral point of view; in any given case, it may be
that someone has not. But to say that this is the only
possibility is to make a very strong claim, for it seems to assume
that moral obligation rests on an agent-independent norm of
trustfulness. Such an assumption may be independently attractive,
but it is difficult to maintain given an internalistic theory.
Suppose, for example, that M falls short of such a standard, so
that she is always mistaken when a decision turns on trustfulness.
Since M has an internalistic moral obligation to act otherwise,
then according to any plausible internalistic theory there must be
some moral point of view that M could adopt such that if she
adopted it she would choose to so act. But how are we to maintain
that there is such a perspective that M could adopt? It is just
such an incapacity that constitutes her mistrustfulness. We
cannot, of course, rule out the possibility that some particular
internalistic theory will be able to accommodate an agent-
independent standard of trust, but it does seem unlikely. Agent-
independent standards are precisely what internalistic theories
108
have so much difficulty accommodating. Whether an agent can see
her situation in a particular way, and hence, whether an agent can
be bound by an internalistic obligation, will depend upon various
of her capacities like the capacity to trust.
The case reveals still more about the structure of plausible
internalistic moral theories. Suppose that we accept some
internalistic moral theory. If we assume that both the trustful
person and the mistrustful one are adopting a moral point of view
when they decide, then there are two promising strategies for
evaluating the case, depending upon our account of the agents'
reasons. If we say that T and M have the same reasons for acting
in this case, but choose differently,36 then we must claim that,
despite their efforts to act morally, the standards that T and M
deliberate in accordance with differ in morally significant cases.
Since, by hypothesis, both T and M are adopting a moral point of
view when they decide, both standards may correspond to our moral
theory (and they will correspond to our moral theory if T and M
are assumed to be generally reflective and conscientious) . Thus,
our moral theory would consider agents (with certain differences
in motivation-affecting capacities) in the same situation with the
same reasons for acting who choose to engage in incompatible
36 Since, by hypothesis, they choose reflectively, conscientiously, and
confidently, we are presumably also supposing that in other similar
cases, T and M make similarly different choices.
109
behavior each to be adopting a “moral point of view" with respect
to their situation and reasons for acting.
If we say that T and M have different reasons for acting,
then, presumably (since the case is an arbitrary one), we assume
that they have different reasons for acting in all situations in
which their decision turns in part on how trustful they are. In
this case, we are claiming that the standards that T and M
deliberate in accordance with have significantly different
domains. (Relative to the standard in accordance with which the
other deliberates, each is like Sisyphus to the Gourmand
Standard.) Thus, our moral theory would consider agents whose
opportunities for acting in morally significant cases are
significantly different and disjoint each to be adopting a “moral
point of view” with respect to their situation and reasons for
acting, because of certain differences in motivation-affecting
capacities.
Both solutions require us to accept that significantly
different standards legitimately correspond to our moral theory.
Thus, plausible internalistic moral theories would seem to need to
depend upon a conception of a “moral point of view" that includes
different ways of looking at the world.
110
Chapter Four:
The Relationship of Moral Relativism to Internalistic
Moral Theories
In chapter three, I argued that internalistic moral theories
that defend the Basic Internalistic Premise using the claim that
recognizing that something is good can be motivationally
efficacious will be bound to claim also that whether an agent is
bound by an internalistic obligation will depend upon some of her
capacities and her circumstances. For a given moral theory, there
will be an array of characteristics of an agent that might affect
the evaluation of internalistic judgments. We should not expect
there to be a single moral point of view that is appropriate for
all agents. In this chapter, I will consider the relationship
between this kind of agent-relativity and moral relativism. I
will try to characterize the nature of the problem that
internalistic moral theories have with moral relativism, and the
impact of denying moral relativism on the structure of
internalistic moral theories.
Even if for a given population there is no single moral point
of view such that all agents in the population are capable of
adopting it (or morally ought to adopt it), it would not
necessarily follow that moral judgments must be evaluated relative
to features of their subjects. Locke, in his argument against the
possibility of innate moral principles, has provided us with a
clear explanation of why our relativistic thesis does not
111
necessarily entail this kind of moral relativism, even for
internalistic judgments:1
That men should keep their compacts is certainly a
great and undeniable rule in morality. But yet,
if a Christian, who has the view of happiness and
misery in another life, be asked why a man must
keep his word, he will give this as a reason:--
Because God who has the power of eternal life and
death, requires it of us. But if a Hobbist be
asked why? he will answer:--Because the public
requires it, and the Leviathan will punish you if
you do not. And if one of the old philosophers
had been asked, he would have answered:--Because
it was dishonest, below the dignity of a man, and
opposite to virtue, the highest perfection of
human nature, to do otherwise.2
Locke’s remarks illustrate how, even if agents differ in their
circumstances, abilities, and commitments, they may still come to
order their reasons and ends similarly in certain kinds of
situations--in particular, in those which some of us would think
had moral significance. Agents may order their reasons similarly
in that they may prefer the reasons in favor of similar actions,
even though their reasons and ends differ greatly. For example,
suppose that my friend and I each owe the price of our own lunch
to some restaurant. Suppose further that I enjoy the thrill of
10f course, in this passage Locke's interest is not in internalistic
judgments, but in moral judgments generally, and his concern with them
is to show that no moral principles are innate. Locke's concern in this
argument is nonetheless similar to ours. His strategy seems to be to
show that even if some moral judgments are universally considered to be
binding, that would not require that those judgments are manifestations
of universally held moral beliefs (nor, therefore, that they are
manifestations of innate moral beliefs). Locke's point, turned around,
is that even if for different agents different ways of deliberating
count as adopting a moral point of view, agents may still consider some
of the same moral judgments to be binding.
Locke, Essay Concerning Human Understanding, vol. 1 (New York: Dover
Publications, Inc., 1959) p. 69 (I.II.5).
112
trying to get away with not paying and do not find the idea of
commercial obligation compelling or cannot understand it, while my
friend regards paying as an obligation. She pays because she
regards any other course as improper. I pay because I do not wish
to upset my friend. The schema 'x ought to pay what she owes'
might yield a true internalistic judgment when instantiated by
each of us and evaluated by each of us, despite our differences in
reasons, ends, and moral outlook (e.g., beliefs about ordering
reasons and ends), and each of us might make this judgment despite
of our having disparate moral beliefs and perhaps differing
capacities to adopt new moral beliefs. The standard she
deliberates in accordance with and considers to accord with her
moral outlook might happen to support behavior in situations of
morally significance (relative to either or both of our moral
outlooks) similar to behavior supported by the standard I
deliberate in accordance with and consider to accord with my moral
beliefs. This is not impossible; the standards do support similar
behavior in this instance (even though we would presumably expect
that they would differ in other cases). We might each approve of
the other’s behavior but not the other's moral beliefs, attitudes,
or character.
Moral relativism might, then, be understood to be a claim
that there are no internalistic schemata of the form 'If x has
certain features F, then x's best reasons support her engaging in
113
behavior falling under the description D' that yield true
internalistic judgments about every agent. Locke's remarks show
us how to distinguish this version of moral relativism from our
relativistic claims about internalistic judgments; even if the
standards by which agents' reasons are to be judged differ, the
behavior supported by agents' best reasons may be the same. The
question of whether an internalistic moral theory sanctions moral
relativism might then be answered by asking whether there are some
situations in which points of view that the theory considers moral
points of view are such that agents who adopt them engage in
similar behavior. More precisely, we should consider whether, for
agents in populations that are of interest to us (e.g.,
populations that exist, have existed, or are likely to exist), the
standards that agents would deliberate in accordance with if they
adopted moral points of view support similar behavior in some
types of cases.
We cannot yet use this characterization of moral relativism
to describe what internalistic moral theories must (at minimum)
deny if they are to be plausible, however. This is because some
universalistic moral theories that would not even be suspected of
having moral relativism as a consequence claim that different
agents are bound by different kinds of moral judgments, and thus
would not meet such a criterion. Our characterization of moral
114
relativism, therefore, requires some adjustment if it is to accord
with common conceptions of implausible moral relativism.
A good example of a universalistic moral theory that claims
that different agents are bound by different kinds of moral
judgments is that offered by G. E. Moore in Principia Ethica.
Moore is not commonly supposed to present a theory that sanctions
moral relativism, yet he explicitly grants that judgments about
conduct and rules governing it are highly dependent upon
circumstances and capacities related to understanding and
reflecting on one’s situation. Moore does not concern himself
with speculation about whether we know of any standards of conduct
that would apply in every situation (i.e., about whether there are
universally applicable internalistic schemata), but he seems to
consider it sufficient to undermine our expectations that there
are any to point out that there are always assumptions about
circumstances, inclinations, and capacities upon which our
judgments about conduct depend.3
Why is Moore not to be considered a moral relativist on these
grounds? A familiar distinction that Moore insists we recognize
makes this clear. Claims that something is good are supposed to
be of two kinds:
They may either assert this unique property
[goodness] does always attach to the thing in
question, or else they may assert that the thing
in question is a cause or necessary condition for
3Principia Ethica (New York: Cambridge University Press, 1960) p. 158.
115
the existence of other things to which this unique
property does attach. . . .Their difference has,
indeed, received expression in ordinary language
by the contrast between the terms 'good as means'
and 'good in itself', 'value as a means' and
'intrinsic value'.4
Moore holds that the truth of claims about what things are good in
themselves does not depend upon the circumstances or other
features of their subjects, even if the truth of judgments about
behavior, which can only be good as a means, is determined
relative to them. We can be relativists about the goodness of
mere means without being relativists about what things are good in
themselves.
Whether or not we wish to call relativism about ascriptions
of goodness to means "moral relativism", moral theories
sanctioning only this kind of relativism render some class of
moral judgments true independently of when they are made, who
their subjects are, and by whom they are made. But this
relativism, relativism restricted to means only, appears to be
fairly trivial. It seems doubtful that there is any reasonable
theory at all according to which both some things are good merely
as means, and the goodness of such things is not determined
relative to the cooperation of the circumstances that make it
reasonable to call them "means". Certainly, the very limited
claim that judgments about merely instrumental goods must be
evaluated relative to agents does not seem to render a moral
4Principia Ethica, p. 21.
116
theory implausible. Since moral relativism is of interest to this
project only to the extent that it has this consequence, I will
adopt the common convention of reserving the term “moral
relativism" for theories according to which judgments about what
is intrinsically good must be evaluated relative to some features
of agents.
The arguments in chapter three suggest that internalistic
moral judgments might sanction moral relativism for judgments
about how agents should behave. The recognition of Moore's
distinction allows us to see that we need not be concerned (on the
grounds of chapter three) that a particular internalistic moral
theory sanctions moral relativism if according to that theory good
conduct is only good as a means. However, for theories according
to which conduct, dispositions, or willings (acts of will) are,
are composed of, or are constituents of something good in itself
(even if they are neither the only things that are good in
themselves, nor the highest good), then we cannot without further
explanation see relativism about what is good conduct as any less
significant a version of moral relativism than relativism about
other intrinsic goods.5 But if we do not rule out conduct as a
source of intrinsic good, then our problem remains: Does the
agent-relativity that arises from defending the Basic
5Moore himself does not seem to rule out that conduct might be good in
itself. In fact he seems to want to allow a great plurality of things
good in themselves.
117
Internalistic Premise imply moral relativism? Moreover, if the
truth of judgments about intrinsic goods that are constituents of
conduct must be evaluated relative to their subjects (as plausible
theories seem to require), and it turns out that no internalistic
schema is universally applicable, then absent an arithmetic of
goods that guarantees the inferiority of goods derived from
conduct, we will not be in a position to maintain that any
judgments about good behavior that actually might occur may be
evaluated except relative to some features of agents.
Certain internalistic theories could avoid the charge of
moral relativism (on the grounds suggested in chapter three) by
denying that conduct can be an intrinsic good. Proponents of an
internalistic version of (traditional) utilitarianism, for
example, need not fear being forced to choose between the Basic
Internalistic Premise and denying moral relativism. They can use
Moore's separation of means from ends to exonerate them from any
charges of moral relativism arising from the agent-relativity of
evaluations of internalistic judgments, or even of internalistic
schemata. What these theories lose in attractiveness even Moore
seems to have noticed: they do not recognize some things as goods
which people tend to agree are intrinsically good.6 Utilitarians
6Perhaps I am overstating a debt to Moore at the expense of recognizing
a debt to Bernard Williams. Moore recognizes at least that classical
utilitarian views in general fail to count enough things as
intrinsically good. Williams points out that these views fail (and must
fail) to recognize as good even as means certain things which are
intrinsically good. (See "A Critique of Utilitarianism". In J. J. C.
118
must, for example, reject absolute claims critics are fond of
presenting as "intuitive": that any act of generosity is good (it
might, after all, inadvertently aid the unnecessary rise of
fascism), that innocent persons may not be treated as criminals
(it might prevent many future wrongs), etc. These limitations
might make a theory implausible as a complete account of what is
good, if the theory cannot provide a satisfying account of
apparently contrary common intuitions. These utilitarians get to
hold on to moral universalism and the Basic Internalistic Premise
if they like, however, and that may count in favor of such
theories (and in favor of any theory that denies aspects of
conduct intrinsic goodness). Granting this, I will restrict our
attention to theories according to which some constituents of
(some) conduct is good not merely as a means.
To defend the internalistic moral theories we are considering
against the charge that they sanction moral relativism, it seems
that we would need to demonstrate that there exist internalistic
schemata which yield true internalistic judgments about every
agent. To do so we would need to look beyond theories of
internalistic judgments to theories of internalistic schemata.
Internalistic moral theories, if they need be concerned with
having moral relativism as a consequence must, therefore, include
not only a theory of internalistic judgments, but also a theory of
Smart and Bernard Williams, Utilitarianism: For and Against (New York:
Cambridge University Press, 1973).)
119
internalistic schemata. Unfortunately, while it is relatively
clear that internalistic judgments are significant in many
received moral theories, it is hardly clear that any more general
notion that would correspond to that of an internalistic schema
has been given so much as explicit mention (even by Locke at his
most pluralistic7). Consequently, it is less clear what
internalistic schemata are like, and hence, somewhat more
difficult to talk about them. To remedy this, I will first try to
determine whether there are conditions under which internalistic
schemata might be universally applicable. Armed with a more clear
formal picture of internalistic schemata, I will then consider the
relationship between internalistic schemata and internalistic
moral theories more directly.
What universally applicable schemata would be like
The schema 'x's best reasons support paying for lunch' is
obviously not a candidate for being a universally applicable
internalistic schema in any reasonable moral theory, since one
does not always owe anything for lunch. But it might be that the
schema 'If x has exhibited behavior such that according to (our)
conventions x owes the cost of her lunch to a restaurant, then x's
best reasons support paying for lunch' is universally applicable.
7Indeed Locke's aforementioned remarks are part of his larger argument
against innate moral principles of all kinds based on the observation
that no such principles are universal. The success of such an argument
would render any search for universal internalistic schemata
uninteresting.
120
To say that it is universally applicable is to say that for any
particular agent in a given population, if that agent has
exhibited behavior such that according to (our) conventions she
owes the cost of her lunch to a restaurant, then she is bound by
an internalistic obligation to pay for lunch (i.e., doing so is
supported by her best reasons). Thus, to determine whether an
internalistic schema is universally applicable, we would have to
look to our moral theory and then to the population of agents to
see whether all of them are capable of adopting a moral point of
view in the relevant situations (i.e., whether each is capable of
deliberating in accordance with some standard which is similar to
one corresponding to the moral theory and such that it maps
reasons for acting to reasons that support acting as described in
the schema). In any case, people who do not eat lunch at
restaurants will not be counterexamples to the universal
applicability of this internalistic schema; the internalistic
schema will merely be irrelevant to the behavior of such agents
since they cannot be the subject of a true internalistic judgment
which follows from an instance of the schema.
One might wonder why it shouldn't turn out that some
internalistic schemata would be universally applicable. It is,
after all, because of the existence of agents who are not capable
of being the subject of any internalistic judgment that most moral
theories include some kind of excusing conditions. One might
121
think that this inclusion rescues such theories from arguments for
moral relativism based on agents' incapacity to adequately assess
morally interesting situations (relative to a particular theory).
Although we should expect that internalistic schemata that do not
include a restriction on the competence of their subjects are
likely to fail to be universally applicable, the addition of "and
the agent is a sane adult" to the description of the subject in
the same way the subject's participation in conventions of owing
money was added in the restaurant case might produce universally
applicable internalistic schemata. We would not in that case
necessarily have to verify whether all agents are capable of
assuming an appropriate moral point of view to determine whether
an internalistic schema of the latter sort is universally
applicable; for such a schema we only need verify it for those who
satisfy the descriptions required in the schema, i.e., for sane
adults.
Straightforward circumstance-relativity of internalistic
judgments appears to be easily accommodated by a theory of
internalistic schemata according to which internalistic schemata
are universally applicable in this way. Competence restrictions,
even simple ones, are more difficult to accommodate, however.
Consider the case of children. We do not expect children to be
completely incompetent in morally interesting situations. We
expect them to be capable of acting properly sometimes and
122
sometimes not. Let us disregard the question of to which
standards we wish to hold them for purposes of punishment or
encouragement, since that depends upon our views about moral
development and education, and confine ourselves to considering
their capacities for having ends, forming reasons for acting,
ordering them, and acting in accordance with some of them.
Certainly infants lack the powers required for moral decision
making. Within a few years they typically become able to order
ends and reasons in some way, and also to do so apparently in
accordance with some moral norms. Whether there are some relevant
capacities some never acquire is at this point an open question.
If all internalistic schemata were universal, and all included "is
a sane adult" as described, then internalistic schemata would
always be merely trivially applicable to children.
To say that some agents are excluded in this way is to say
that one's moral theory is absolutely silent on the behavior of
some portion of the population; that is, that the rules for
ordinary agents simply do not apply to some agents, and not merely
that other rules apply to them. This is the sort of case we
considered at the end of chapter three (the harried mother of
ten). On the current way of thinking, internalistic schemata are
not false when applied to these agents (although internalistic
judgments are). Internalistic schemata are true and universal we
suppose. But just the same they are completely uninteresting to
123
such agents. We may not find this a troubling conclusion for
agents like the harried mother of ten, or, indeed, for Harman's
extraterrestrials. But it is probably an error to regard children
merely as an uncharted and unchartable zone with respect to moral
judgment, despite, for example, a lack of “life projects" or
"dominant ends and reasons" which seem to be necessary for the
exercise of moral decision-making according to some moral
theories. A child's abilities change over time. If we accept
current wisdom in thinking that a child does not suddenly become
capable of normal adult decision-making, but rather progressively
gains relevant abilities, then we should incorporate the various
competences of children into our moral theory.
To exclude children or other agents from being the subject of
any true internalistic judgment is to invite a form of relativism.
Especially in the case of children it would seem that this
relativism would in fact be a form of moral relativism as it is
usually understood. If there are internalistic judgments true of
adults then it would appear that some judgments about children are
internalistic judgments as well.
These observations about children should not be taken merely
at face value. The problem with adding "is a sane adult" to
schemata is its spuriousness. We notice the spuriousness by
realizing that "being a child" does not partition the population
neatly into disjoint groups of children and non-children, and that
124
it does not mark a uniquely significant boundary from the point of
view of moral judgment. Significant boundaries between groups of
agents that depend upon internalistic judgments must be marked by
differences in agents' capacities for being motivated by moral
considerations, and such differences cannot be fixed to features
of agents which are not linked to these capacities.8
We may solve the problem of how to deal with competence
restrictions by identifying particular features which are so
linked. If we do this, then a lack of such features will exclude
agents who are children equally with those of other ages. We
might, for instance, resist the exclusion of children from
internalistic judgments by dividing up the characteristics or
abilities associated with sane adulthood, and depend upon them
variously in internalistic schemata. For example, we might have
schemata such as 'If x is able to visualize herself in the future
p
Hume emphasizes a similar (but more general and more radical) point
about animals.
When any hypothesis, therefore, is advanc'd to explain a
mental operation, which is common to men and beasts, we must
apply the same hypothesis to both . . .The common defect of
those systems, which philosophers have employ'd to account
for the actions of the mind, is, that they suppose such a
subtility and refinement of thought, as not only exceeds the
capacity of mere animals, but even of children and the
common people in our own species; who are notwithstanding
susceptible of the same emotions and affections as persons
of the most accomplish’d genius and understanding. (A
Treatise of Human Nature, L. A. Selby-Bigge, ed., second
edition, with text revised and variant readings by P. H.
Nidditch (New York: Oxford University Press, 1985),
p. 177.)
Hume's point is that if there is not sharp boundary between capacities
within a population, there should be no sharp boundary between the
applicability of judgments which are based on these capacities within
that population.
125
as a consequence of current actions, then x (internalistically)
ought to avoid harm to herself', 'If x is confident in her ability
to make own decisions [relative to some norm], then x
(internalistically) ought to resist demands of those in authority
when they go against her own considered judgment', and 'If x is
able to recognize and control misdirected anger, then x
internalistically ought not to attack infants or animals out of
anger' . What emerges is a group of universal schemata which
divide (but not necessarily partition) the population of agents by
the internalistic judgments of which they may be the subject.9
To assess the extent to which this plan actually would divide
a population, and, hence, lead to moral relativism, will not be
possible to do generally; a moral theory might, after all, be
extremely limited in the judgments it sanctions. Since the
particular relevant capacities are specified by each theory, we
must consider a particular moral theory and a particular
population to determine this. This project will be taken up
beginning in the next chapter.
Earlier I claimed that straightforward circumstance-
relativity seems to be fairly easy to accommodate according to the
9Notice that this indirect moral relativism could arise in the same
manner from theories according to which conduct is never an intrinsic
good or have any intrinsic goods as a part, which were unscathed by
worries that universally applicable internalistic schemata might be
unlikely. Thus they would be subject to any problems that might arise
from this indirect moral relativism. This indirect relativism is not
obviously as likely to emerge for these theories, however, since there
is no obvious need to posit a spectrum of judgments generated by
antecedent filters.
126
plan we have been considering. In a way, this is obviously true,
since circumstances figure much more plainly into the making of
internalistic judgments. The need for the inclusion of promise-
making behavior in the schema 'If x has made a promise, then x's
best reasons support keeping that promise' is obvious; one need
not have compelling reasons to keep promises one has not made, for
example. At the same time, the role of circumstances in rendering
a schema universally applicable is not always obvious, as we saw
earlier in cases in which internalistic judgments appeared to
depend upon one's position in a social network. In such cases, it
is hard to see just how such a position is to be adequately
incorporated into an internalistic schema. To do so would require
a way of capturing complexity sufficient to specify (enough of)
the social structure. Presumably this is not an impossible
enterprise (we might approach it extensionally), but it would
appear to be very difficult.
Circumstance-relativity may also point to problems for this
effort to avoid moral relativism. Our method of dealing with
circumstance-relativity in internalistic judgments on the level of
schemata leaves open the possibility that a theory of
internalistic schemata will be unadaptable to potential radical
changes in the circumstances of agents. For example, if all of
the schemata in the theory concern the behavior of agents who are
in conditions in which they expect their basic needs for survival
127
to be met, then in times of war or famine these schemata may be
completely irrelevant to assessing agents' behavior. If such
radical changes occur, then the content of a theory of
internalistic schemata will change and become less adequate if it
applies nontrivially to agents in a diminishing number of
instances or to a diminishing part of the population. In that
case, it would leave the sort of gap we feared would arise for
moral judgments about children, but for an even larger and more
diverse group of cases.
Do theories of internalistic schemata that follow the model
we have been developing have moral relativism as a consequence
anyway? According to this model, general moral judgments, i.e.,
internalistic schemata, are universally applicable within a
population. The model leaves open the question of whether
schemata actually yield true judgments about every agent, however.
The answer to that question will depend upon the particular moral
theory and the particular population. Similarly, the model leaves
open other important questions. It might turn out that according
to some theories that follow the model, different agents will be
the subjects of very different moral judgments. It may be that
some theories partition the population into disjoint groups
according to which schemata apply nontrivially. Finally, it might
turn out that there are some agents of whom no internalistic
schema yields true judgments; this too will depend upon the theory
128
of internalistic schemata and the population. The last two
possibilities most strongly suggest that we might have to embrace
a form of moral relativism as it is usually understood. But even
if the moral theory and the population are such as to make these
conditions obtain, why should we think that a significant form of
moral relativism results? Why should it, if it does not when
excusing conditions are permitted? In building this model of
theories of internalistic schemata, all we have done is exploited
the idea of excusing conditions and clarified its use by
incorporating excusing conditions into internalistic schemata.
Whether we should call it moral relativism thus seems to have more
to do with which members of the population get excluded then
whether any are excluded. And that can only be settled by
examining a particular moral theory and a particular population.
The possibility of universality in theories about
internalistic judgments
I have argued that the question of whether an internalistic
moral theory sanctions moral relativism is a question not of
whether the population is divided by what internalistic judgments
might apply to them (because we should expect that they will be in
any case), but of whether internalistic schemata partition the
population into disjoint groups,10 and in that case, of how
10I.e., so that agents fall into exactly one such group according to
which schemata apply to them. This would have the effect of allowing us
to think of agents as differing in kind according to the sorts of moral
obligations by which they may be bound.
129
exactly they do. Before we can ask whether a particular moral
theory about internalistic judgments sanctions moral relativism,
we need to know who according to the theory the possible subjects
of internalistic judgments are. To see whether moral relativism
emerges, we then want to ask whether, according to the moral
theory, internalistic schemata partition a population of interest
to us in unacceptable ways. If, for example, agents' variations
in motivation-affecting capacities entail that they are
partitioned into disjoint groups according to disjoint classes of
obligations by which they might be bound, so that there aren't any
obligations that are even potentially binding for every
(commonsense) moral agent, then that would certainly render true a
form of moral relativism that runs strongly against commonsense
moral intuitions.
Let us, for convenience, call a schema that applies
nontrivially to all agents who are the subject of some true
internalistic judgment nontrivially universal. The most
straightforward way to defend a moral theory against the charge of
sanctioning moral relativism is to show that if the theory does
partition the population into disjoint groups according to
disjoint classes of obligations by which they might be bound
(because of variations in agents' motivation-affecting
capacities), then all members of the population about whom some
internalistic judgment is true are in the same partition; that is,
130
to show that there are internalistic schemata which are
nontrivially universal. This we probably cannot hope to
accomplish, thanks to the equal reliance of internalistic schemata
on circumstances and other features of agents. Even in a
uniformly competent population, if one member never in fact
promises (according to our convention of promising), then even the
schemata 'If x has promised to p, then x internalistically ought
to p' could not be nontrivially universal. A nontrivially
universal internalistic schema would have to have only extremely
weak restrictions on circumstances and competences (relative to
the circumstances in which members of the population find
themselves and the competences they have).
Even if nontrivially universal schemata are unlikely to be
found for a given population, we still may be able to argue
against a similar form of moral relativism in a similar manner.
There might be groups of internalistic schemata such that every
subject of some true internalistic judgment in a particular
population is the subject of some internalistic judgment following
from an instance of a schema in the group. Let us call such
groups nontrivially universal because of their similarity to
nontrivially universal schemata. The idea behind shifting our
attention to nontrivially universal groups of schemata is to try
to capture the idea that everyone with the same global capacities
for being motivated to fulfill their obligations is a potential
131
subject for internalistic judgments following from the same
schemata. Unfortunately, there is no obvious way to distinguish
divisions among domains of standards that arise from differences
in circumstances from those that arise from differences in
capacities for being motivated to fulfill their obligations. So
the thought is that when two agents have the same capacities for
being motivated to fulfill their obligations, there will be
nontrivially universal groups for the population including just
them. There is no reason to believe that nontrivially universal
groups would emerge for such a population otherwise. We might
look at nontrivially universal groups as a filtering tool; their
emergence does not indicate any real universality holds of
theories and populations which lack nontrivially universal
schemata, but their emergence does indicate that, for a given
theory and population, agents variations in motivation-affecting
capacities will not force the theory to partition the population
into disjoint groups according to disjoint classes of obligations
by which they might be bound.
Just as was the case for nontrivially universal schemata, we
cannot in general expect to find nontrivially universal groups.
There do seem to be attainable conditions under which nontrivially
universal groups of schemata would emerge for a given population,
however. Nontrivially universal groups of schemata would be
possible for exactly those pairs of populations and internalistic
132
theories such that every member of the population who is the
subject of some true internalistic judgment has the capacity to
deliberate in accordance with a standard that accords with some
internalistic judgment following from a schema in the group and
does not conflict with any internalistic judgment following from a
schema in the group, and such that deliberating in accordance with
the standard constitutes adopting a moral point of view with
respect to the theory.
In what cases would standards have such similarities?
Even though sketchy, our characterization of plausible
internalistic moral theories so far suffices to ensure that they
will tell us something about internalistic schemata. At the very
least, an internalistic moral theory will include restrictions on
internalistic schemata that arise from the fact that, for a given
internalistic moral theory, being similar to a standard
corresponding to the theory is a well-defined property. How
exactly this fact implies restrictions on internalistic schemata
will, of course, depend upon the particular theory. For example,
consider a certain nihilistic moral theory characterized by
“anything goes". Since any standard corresponds to this
nihilistic theory, the theory will obviously not include any
internalistic schemata.
Under what conditions may an internalistic moral theory
include internalistic schemata? For some theories, claiming that
133
all moral agents can assume a certain minimal perspective from
which to deliberate provides support for internalistic schemata.
For example, given a population that is radically egoistic, an
egoistic theory of moral judgments (e.g., one that claims that
agents morally ought always to try to advance their own interests)
may yield a substantial theory of internalistic schemata.11 If
every agent is always capable of choosing to act in a way that she
believes will advance her own interests, and some activities are
so much to an agent's advantage that everyone will perceive such
activities as in their advantage, then engaging in such activities
could bo seen as choiceworthy by all agents. In that case,
engaging in such activities could be morally obligatory for all
agents.
Claiming that all moral agents can assume a certain minimal
perspective from which to deliberate will not in general guarantee
that an internalistic moral theory may include internalistic
schemata, however, because a moral theory may demand more from
some than from others. For example, one’s realization that there
is more of moral interest than meets the eye may make inaction no
longer permissible.12 Consider, again, an egoistic internalistic
moral theory. Suppose that there is a perspective that everyone
can adopt that would lead everyone to act similarly in similar
11Against such defenses of egoistic views of course are denials of the
requisitely strong egoistic assumptions. But these defenses may have
any advantages that accrue from being an internalistic theory.
1 ?
xxEven if the realization itself is not required.
134
circumstances. For instance, suppose that any person is capable
of dwelling on the payoffs of a given choice of behavior, and,
where the costs are not great, of thereby finding aiming at the
highest payoff most advantageous, and, therefore, most
choiceworthy. Presumably, according to our egoistic moral theory,
an altruistic person would benefit from adopting such a
perspective. Though ordinarily she would be tempted to act
contrary to her own interests, by adopting the universally
accessible perspective of dwelling on the payoffs of her choices,
the altruist will (by hypothesis) be moved to act morally, that
is, to try to advance her own interests. Such a perspective
might, morally speaking, be the best one for the altruist to
adopt.
It would not follow from the universal accessibility of this
perspective, the uniformity of the behavior of agents who adopt
it, or from the claim that for some agents the universally
accessible perspective amounts to a moral point of view, that
everyone ought either to adopt it or to behave as if they did.
This is because adopting the perspective may well not be in every
agent's interest or perceived interest. Consider, for example,
that one consequence of human agents' dwelling on the payoffs of
their choices seems to be that they come to consider taking risks
with low expected values but large payoffs as in their advantage.
Adopting the universally accessible perspective of dwelling on the
135
possible payoffs might, therefore, lead agents to be moved to take
such risks. Adopting the universally accessible perspective might
lead agents to play multimillion-dollar state lotto games, for
example. People who are aware of the low expected value of a
lotto ticket would be acting contrary to their perceived self-
interest if they buy a lotto ticket (rather than something with a
higher expected value), however. Thus, according to our egoistic
moral theory, it would be wrong for such agents to dwell on the
large payoff. The universally accessible perspective would not be
appropriate for such agents according to our egoistic moral
theory. It would not necessarily be wrong, according to the
theory, for someone who genuinely did not believe the expected
value of a lotto ticket to be very low, to buy a lotto ticket or
to dwell on the large payoff. Adopting the universally accessible
perspective may even be morally best for some agents--the altruist
considered above, for instance. But the existence of a
universally accessible moral perspective does not guarantee that
everyone should either adopt it or act as if they did, even if it
is morally appropriate for some agents.
As I remarked above, the egoistic moral theory may
nonetheless guarantee the existence of nontrivially universal
schemata by claiming not only that agents are similar in their
capacity to adopt some moral perspective, but that agents perceive
some things in relevantly similar ways. Perhaps no agent would
136
perceive refusing her money to a known murderer pointing a gun to
her head as being in her interest. In that case, our egoistic
moral theory might reasonably support the internalistic schema
that no one ought to refuse her money in such a situation. This
kind of defense of the existence of nontrivially universal
schemata depends upon more than merely assuming that agents are
capable of adopting a minimal perspective from which to
deliberate, however.
Moreover, it is more difficult to see how such additional
constraints might be defended for moral theories advanced more
commonly than egoism. For example, even if it were reasonable to
claim that every human agent can adopt a perspective from which
she would be moved by generalized benevolence or pity, it seems
doubtful that every human agent would agree about what would best
advance the interests of humanity generally.13 Similarly, even if
it were reasonable to claim that every human agent is capable of
refusing to act irrationally, it seems doubtful that every human
agent is capable of judging the same actions irrational.14
Thus, since correspondence of a standard to a moral theory
is, at least formally, determined relative to agents, claiming
that all moral agents can assume a particular minimal perspective
13This is suggested, for example, by the common criticisms of
utilitarianism that it requires that agents have the capacity to view
what they value as commensurable, and that it requires that agents have
computational abilities beyond the reach of many human agents.
14See, for example, Daniel Kahneman, Paul Slovic, and Amos Tversky,
Judgment under Uncertainty: Heuristics and Biases (New York: Cambridge
University Press, 1982).
137
from which to deliberate does not guarantee that all agents are
capable of deliberating in accordance with a standard similar to
one corresponding to a particular moral theory. Neither,
consequently, does making such a claim guarantee that a particular
moral theory may include internalistic schemata.
For an internalistic moral theory to include some
internalistic schemata, the theory must have a more complex kind
of regularity (rather than a minimal universally accessible
perspective). An internalistic moral theory will include some
internalistic schemata for a population P if it satisfies two
conditions:
(i) For any agent in the population P, there is a capacity for
being motivated (actually, a set of capacities, where that
is understood very broadly) such that she has it, and such
that her exercising that capacity would yield that she
deliberates in accordance with a standard similar to one
that corresponds to the theory.
(ii) There exists some agent A, set of reasons for acting g, and
course of action D such that in a situation in which A has
exactly the reasons in g, A's best reasons (relative to the
theory) support course of action D, and such that for no
agent B is it the case that in a situation in which B has
exactly the reasons in g, B's best reasons (relative to the
theory) do not support course of action D.
The existence of an element that satisfies condition (ii)
does not, of course, guarantee that an internalistic moral theory
will include a nontrivially universal schema. Such a schema will
be nontrivially universal if, in addition, for any agent C in P,
in a situation in which C has exactly the reasons in g, C's best
reasons (relative to the theory) support course of action D. As
138
we recognized earlier, nontrivially universal schemata are only
possible if we assume an implausible degree of homogeneity in a
population. Nontrivially universal groups of internalistic
schemata emerge when there are sets of reasons for acting g1( g2,
gm and action-descriptions Dj, Dm such that:
(a) Each schema in the nontrivially universal group of schemata
applies to some agent in some case; i.e., for each i such
that 1< i < m, there is an agent A_* in P such that in a
situation in which Ai has exactly the reasons for acting in
g^, Ai's best reasons (relative to the theory) support a
course of action falling under description .
(b) Every agent in the population is the subject of some true
moral judgment that follows from a schema in the
nontrivially universal group of schemata; i.e., for any
agent A in P, there is some i, 1< i^m, such that there is
a situation in which A has exactly the reasons for acting
in gj, and A's best reasons (relative to the theory)
support a course of action falling under description Dj.
(c) No moral point of view will lead an agent adopting it to
choose to act contrary to any judgment following from a
schema in the nontrivially universal group of schemata;
i.e., there is no agent A in P such that for some i,
1 < i < m, in a situation in which A has exactly the reasons
for acting in g_j, A's best reasons (relative to the theory)
do not support a course of action falling under description
D i-
To justify the assumption that there is some nontrivially
universal group of internalistic schemata, a moral theory must
address two problems. The first is to show that condition (c) is
met. The most straightforward way this might be guaranteed is if
all acceptable standards (those such that some agent's
deliberating in accordance with it constitutes her adopting a
moral point of view) are subsets of a single one-to-one
139
mapping.1516 A theory need not meet condition (c) so
straightforwardly, of course.17
Once we guarantee that (c) is met, we can address questions
about partitioning. We can ask whether there are nontrivially
universal groups of internalistic schemata, and whether they are
significant or merely spurious (i.e., if they do not describe
significant partitions of the population.) When we ask about how
significant a partition is available, we are asking about whether
a partition arises merely from differences among agents in minor
particulars of their capacities or circumstances, or in capacities
important for making moral evaluations, or in the framework of
circumstances comprising all of an agent's experience. We are now
in a position to consider how particular internalistic moral
15Does it in that case follow that either there is a nontrivially
universal schema or that the population is partitioned? Technically,
yes. But should partitioning of this kind emerge, it need not trouble
us because we have nontrivially universal groups of schemata. For
instance, if you get into your particular situations and me mine, but
because of the particulars of the definition of standards the domains of
our standards don't actually overlap. But in this case you might get
into promising situations just as I do, and in those cases, the view
might direct both of us to do as we each promised. A partition that
separates us solely based on the minute differences in the particulars
of our circumstances would not appear to be significant.
16For theories according to which acceptable standards are one-to-one
mappings, g6g] , g 2 ........g, only if for any i, j such that Sj and Sj
each correspond to the theory for some agent in the population P, either
Si(g)=Sj(g), or g e dom(Sj) , or g 6 dom{Sj) .
More generally, for theories according to which standards need not be
one-to-one, g e g 1 ,g2. • • • . 9 m only if for any i, j such that Sj and
Sj each correspond to the theory for some agent in the population P, and
appropriate k, either there exists an 1 such that either Sjfg)^ and
S■ , • (g) 2 each support the same courses of action or g e dom{Si) , or
ge dom(Sj), where we assume standards are indexed and that S^fg)* is the
k1 1 1 image of g in Sp
140
theories might avoid sanctioning moral relativism. I take this up
beginning in the next chapter.
141
Chaptar Five:
Strategies Internalistic Moral Theories Might Use to Avoid Moral
Relativism
In this chapter, I will explore the possibility that an
internalistic moral theory might avoid moral relativism. The
question about moral relativism that is most important for
assessing the plausibility of a moral theory is whether the theory
requires that we evaluate moral judgments differently for
different agents. The arguments of chapters three and four show
how this question can be answered by answering other questions.
First, in chapter three, I argued that we have reason to believe
it will matter for the evaluation of a moral judgment who its
subject is, if capacities involved in making decisions matter for
the evaluation. In the first section of chapter four, I suggested
that even if relevant capacities vary among agents, these
differences may not matter for the evaluation of a moral judgment
if there is a coincidence of choices selected in situations of
moral interest through the exercise of (globally) different
capacities (and, therefore, who the subject of a moral judgment is
will not matter for its evaluation on the grounds of chapter
three). I then suggested that we need not fully investigate this
possibility (which would require us to investigate particular
descriptions of particular capacities associated with various
theories), because it is doubtful that for the full range of moral
judgments there will be such overlap. Fortunately for the
142
prospects of internalistic moral theories, it is also doubtful
that we want or expect it; the inclusion of some capacities in the
preconditions for applying a schema to get a judgment seems to be
reasonable and in accord with common sense. Since it is
reasonable that the applicability of moral judgments to different
agents will vary according to certain of the subjects' capacities,
the important question about moral relativism is really how the
identity of the subject of a moral judgment affects the way in
which we evaluate the judgment. More specifically, an
internalistic moral theory will only be plausible if agents are
divided by the judgments that apply to them in ways that fit with
our moral intuitions.
At the end of chapter four, I proposed a set of minimum
criteria that internalistic moral theories must meet if they are
to be plausible. Plausible internalistic moral theories should
not partition a population of human agents into disjoint groups
according to disjoint classes of obligations by which they might
be bound. In other words, at least some very general sorts of
moral judgments must be universally applicable, or, more
precisely, at least some moral considerations must be universally
appreciable as such. Thus, plausible internalistic moral theories
should (at minimum) include what I have called nontrivially
universal groups of internalistic schemata, where nontrivially
universal groups of internalistic schemata are supposed to
143
characterize classes of moral judgments that derive from certain
uniformities in human capacities or features that affect
motivation. According to the model developed so far, a plausible
internalistic moral theory will provide an account of what agents'
internalistic obligations are, and a defense of this account,
which will show that the theory satisfies both the completeness
and congruence restrictions, i.e., that it renders the right moral
obligations internalistic (relative to our pretheoretical moral
intuitions).1 I have argued that the latter will require (at
minimum) (i) an account of what a moral point of view involves for
a given agent in a given situation; (ii) a defense of the claim
that an appropriately wide class of human agents can assume a
moral point of view; and (Hi) a defense of the claim that the
theory includes nontrivially universal groups of internalistic
schemata.
Since no received moral theory has been offered explicitly as
a theory of internalistic schemata, it will take some elaboration
to see if any received internalistic moral theories are plausible.
In this chapter, I will make a start at elaborating some received
internalistic moral theories and assessing their plausibility. I
will approach the question of whether an internalistic moral
theory can avoid sanctioning moral relativism by considering some
examples of simplified versions of received moral theories. These
■*•366 chapter one.
144
simplified theories will involve moves that one might use to
defend against moral relativism that are available for some of the
most common received internalistic moral theories. It will,
therefore, be helpful to consider these simplified theories in
some detail.
A good place to begin is with a theory that provides a
relatively clear account of what agents' internalistic obligations
are. Let us consider a utilitarian theory, and consequently begin
with the claim:
(U) An agent ought to engage in some behavior if and only if
doing so is in the public interest (as defined by a
particular utility calculus).
For simplicity, let us restrict our attention to populations that
include only agents who are subjects of some internalistic
judgment, and consider a theory about internalistic judgments
exclusively. Since we want to exclude from consideration non-
internalistic moral judgments, we must begin with a more
restrictive claim than (U):
(Cl) An agent has an internalistic obligation to engage in some
behavior if and only if doing so is in the public interest
(as defined by a particular utility calculus) .
To meet the requirements of our model of plausible internalistic
moral theories, the theory must guarantee the existence of
nontrivially universal groups of internalistic schemata. Suppose
that the theory does so by providing a defense of the following
two claims:
145
(C2) All agents (in fact) have the public interest (as defined
by our utility calculus) as an end (at any time), i.e.,
they have motivating attitudes reflecting this.
(C3) At any given time, there is a way of thinking about,
considering, or imagining one's situation and the ends that
one has such that one will act on reasons for acting
aligned with the public interest (as defined by our utility
calculus) if one is engaging in this (i.e., one will have
reasons for acting such that the set of those reasons R
will be in the domain of a standard S that corresponds to
our theory, and one's behavior is guided by a reason for
acting in the set to which R is mapped by standard S),2
There is, in fact, a fairly well-disseminated moral theory
which appears to make these three claims; it is unfortunately a
certain reading of Hume (based mostly on the Enquiry Concerning
the Principles of Morals) which affords only a very poor fit with
much of what Hume wrote. I will call this view NQH (for Not Quite
Hume) .3 In NQH, the process whereby an agent comes to feel the
compellingness of the public interest is called sympathy with the
public interest.4
2It might be helpful to notice that claim (C3) involves going beyond the
model's merely descriptive use of standards. As I remarked in chapter
three, there is no difficulty in claiming that agents act in accordance
with a standard because their actions are guided by the reasons which
characterize the standard. The model does not rely upon this assumption
because it requires unnecessarily strengthening the model. But the
model is certainly compatible with it.
3The view expressed in A Treatise of Human Nature, for example, is not
utilitarian, and, hence, would be incompatible with claim (Cl).
According to Hume's Treatise view, being in the public interest does not
necessarily make an action right (let alone obligatory), because an
agent's character, rather than the consequences of her behavior, is the
primary object of interest for moral judgment. The primacy of character
also implies that, according to Hume's Treatise view, even an action
that is contrary to the public interest may still be morally obligatory
if refraining from it could only flow from a bad character (one the
survey of which would arouse disapprobation). For a specific case in
which NQH and Hume's Treatise view conflict, see below, pp. 161-163.
4To emphasize that "sympathy1 ' and "sympathy with the public interest"
are used in this discussion in an idiosyncratic manner (following Hume),
I will italicize them. In addition, since, in this idiosyncratic,
Humean usage, sympathy is a process, I will use the expression "to
146
According to this view, a guarantee that there is some
(nonempty) nontrivially universal group of internalistic schemata
is supposed to arise from both the existence of an end (or a
motivating attitude) that all subjects of internalistic judgments
are assumed to share and the existence of a uniformly available
way of appreciating whatever potentially unique ends each has.
The relevant assumptions are claims (C2) and (C3). Claim (C3) can
be seen as asserting that our utility calculus describes a certain
type of human behavior, namely, the behavior of human agents when
they are engaging in sympathy with the public interest. Claim
(C2) can be seen as ensuring that claim (C3) might be true, that
is, as ensuring that some action in accordance with our utility
calculus is always an option. Assuming that our utility calculus
involves only agent-independent calculations, claims (Cl), (C2),
and (C3), taken as described, jointly ensure that nontrivially
universal groups emerge from NQH; because any acceptable standard
(for any agent, relative to NQH) is a subset of a single class
defined by our utility calculus, the existence of a group of
schemata that meets condition (c) is ensured (i.e., one such that
no moral point of view will lead an agent adopting it to choose to
act contrary to any judgment following from a schema in the
nontrivially universal group of schemata5).
engage in sympathy" rather than "to sympathize", to avoid confusion with
the ordinary, non-Humean idea.
5See p. 139 (chapter four).
147
NQH is obviously a very strong theory that assumes the
abilities of members of a population meet strict requirements. It
also ultimately depends for the hard work on the utility calculus,
a dubious sort of construction. The purpose of considering NQH is
not to examine it as a viable moral theory, however, but merely to
provide a start at understanding how more sophisticated
internalistic moral theories might provide a defense of the
existence of nontrivially universal groups of internalistic
schemata. At this point we need only ascertain what the
requirements are for nontrivially universal groups, not to assess
theories' overall plausibility, usefulness, or adequacy. NQH's
problems might even be an asset, since they are easy to isolate.
To ask whether there is a significant nontrivially universal
group of schemata for NQH, then, we need not concern ourselves
with theoretical issues; the definition of NQH itself guarantees
that there will be nontrivially universal groups of internalistic
schemata. The problem for NQH is that it may be too strong a
theory to hold of human populations of interest to us. To address
this problem I will try to characterize the kinds of populations
for which claims (C2) and (C3) might hold.
Of what kinds of populations might NQH hold?
To render claims (C2) and (C3) plausible will require a
positive argument, including some kind of argument in favor of its
particular choice of utility calculus. Our utility calculus would
148
need to be defended as an account of internalistic moral
obligation. The formal structure of NQH is dependent on the
existence of some agent-independent method for judging actions
which NQH, being a utilitarian theory, specifies by a particular
utility calculus. I will assume that we need not get more
specific about our utility calculus here. Although, Hume
certainly offers us no such calculus, there are possibilities for
such a calculus, even if their adequacy is uncertain.6 Lacking a
sufficiently specific account of our utility calculus, we will be
concerned with objections that arise granting that we have a
successful argument for our utility calculus, even though this may
well be thought to be the most serious problem for defending NQH.
One observation that appears to undermine the possibility of
claim (C2) applying to ordinary (known) human populations has been
used to criticize utilitarianism on various grounds throughout its
history. This is the plain observation that some ordinary persons
appear not to have the public interest as an end. A basic
assumption of NQH is that this is no more than appearance;
everyone cares about the public interest, and, therefore,
sometimes has some kind of reason to act so as to enlarge the
public interest (though not necessarily reasons for acting in
particular instances). To make plausible a version of claim (02)
6Jeremy Bentham offers one, for example. See An Introduction to the
Principles of Morals and Legislation, in The Utilitarians (Garden City,
New York: Doubleday & Company, Inc., 1961), pp. 37-39.
149
strong enough for a theoretical guarantee of nontrivially
universal groups of internalistic schemata, we must first account
for such appearances.
NQH is ready with an explanation. By engaging in sympathy
with the public interest, an agent comes to want to do things that
will enlarge it. An agent's failure to engage in sympathy with
the public interest with any frequency will result in her
preoccupation with other ends, and consequently in a lack of
reasons for acting arising from her attachment to the public
interest, or possibly in the swamping of any such reasons for
acting that do arise. Since claim (C2) is an assumption about
motivating attitudes and not reasons for acting, the failure to
arise of reasons for acting reflecting the postulated motivating
attitude does not contradict it.7
An alternative explanation of (utilitarianly) bad behavior is
the failure of an agent to recognize situations as being morally
interesting, i.e., the failure of an agent to see her situation as
one in which any course of action promotes the public interest.
This, of course, is not an appearance which casts doubt upon the
70f course, if there are agents who frequently fail to get reasons for
acting from a motivating attitude it is claimed that they have, we
should ask for the grounds upon which we might claim they have it. To
defend NQH, we would presumably need to claim with Hume that human
agents do, at least sometimes, demonstrate that they have the public
interest as a motivating attitude, if only by taking an interest in
whether others advance or detract from it. See, for example. An Enquiry
Concerning the Principles of Morals, in Enquiries Concerning Human
Understanding and Concerning the Principles of Morals, 1777. Reprint,
L. A. Selby-Bigge, ed., with text revised and notes by P. H. Nidditch
(New York: Oxford University Press, 1983), p. 219, p. 226.
150
universality of having the public interest as an end. Thus, it
does not suggest that there is any absurdity in assuming that all
agents have the public interest as an end.
It is the combination of claims (C2) and (C3) that warrants
greater concern. The type of motivating attitude that "having the
public interest as an end” entails is such that for anyone
possessing it, engaging in sympathy with the public interest will
result in behavior that is in the public interest. As was pointed
out in chapter two, this is not ordinarily the case with
motivating attitudes. Either the end (the public interest) or the
process (sympathy with the public interest) must be special in
that even agents' strongly or compulsively desired ends do not
interfere. First, reasons for acting so as to favor the public
interest will arise under certain conditions (even in spite of the
presence of strongly or compulsively desired ends). Second, those
reasons for acting cannot fail to be efficacious when sympathy
with the public interest occurs (i.e., the action that results
will be guided by those reasons). Is there anything that could
account for the motivational effectiveness of the combination of
having the public interest as an end and engaging in sympathy with
the public interest?
If this special effect is supposed to derive from the end
alone, then we would need to have an account of how the public
interest is different from other ends, and to explain why it is
151
motivationally more effective than other ends, especially those
that are compulsively desired. Arguments against the possibility
of a successful account of this sort are fairly common. Apparent
counterexamples include cases that are said to demonstrate
“weakness of will". Suppose, for example, I know I ought to pay
you the money I owe you while I have it. Instead I frivolously
spend it on myself. A reason for acting that might be described
as "an interest in not spending the money on myself" or "an
interest in setting aside this money to give to you" might, and
let us suppose in this case does, arise from the public interest--
in particular, from the increase in societal stability to which
conscientious debt payment leads. Nonetheless this reason for
acting fails to result in action because of some other reason for
acting to which I am more favorably disposed at the time I am
deliberating. Of course, it might be that the public interest
could produce a reason for acting that I would not resist in a
choice between it and the frivolous reason for acting that in fact
moved me. But this would not alter the fact that in this case I
had a less motivationally effective reason for acting which arose
from the public interest.
We might deny that my actual reasons for acting to reserve
the money to pay my debt arose from my having the public interest
as an end. To make this claim would require a test for
determining which reasons for acting that favor the public
152
interest arise from having the public interest as an end, and
which merely coincidentally support actions which would favor the
public interest. We cannot in general assume that reasons for
acting which the agent sees as arising from that end only would be
included among the former. The case of my shirking my obligation
to repay you, on this account, is one in which I fail to acquire
reasons for acting that I would if I were considering my situation
differently. It begins to sound as though this view will reduce
members of a large part of the population almost to the status of
Sisyphus--to being agents whose deliberations typically occur
outside of the reach of any moral considerations. Though it is
possible to hold this view, it hardly seems worth taking
seriously; what, if not my obligation to pay my debts
conscientiously, is a moral consideration relevant to my
deliberation in the case described above? It does not seem
reasonably deniable that an agent can have reasons for acting
arising from having the public interest as an end, and choose to
act either according to that reason for acting or not. Nor,
therefore, does it seem reasonably deniable that however
motivationally effective an end the public interest is, it cannot
guarantee the motivational efficacy of reasons for acting that
arise from it.
Nor does it seem that we need to rely on the assumptions of
NQH to see that the motivational efficacy of an end cannot ensure
153
the motivational efficacy of reasons for acting that arise from
it. Certainly we need not in the case of any end which in general
generates reasons for acting for morally good behavior; in any
such case, we can offer as a counterexample an instance of
"weakness of will" as we did above. It is not enough to say that
the end imparts generic motivational strength to reasons for
acting that derive from it. We must at least give conditions
under which it does so. Otherwise, we are left to (implausibly)
deny that unpersuasive reasons for acting, like my weak
recognition of my obligation to pay my debts conscientiously,
cannot emerge from the same end that successful and morally worthy
reasons for acting might. I assume that we cannot genuinely claim
that the kind of moral failure illustrated by this case of
"weakness of will" is impossible.
In NQH, the existence of conditions under which reasons for
acting arising from the public interest are or become
motivationally efficacious are guaranteed by claim (C3). These
conditions are supposed to obtain when an agent engages in
sympathy with the public interest. Notice that the conditions
required to guarantee the motivational efficacy of some reason for
acting arising from the motivating attitude postulated by claim
(C2) (the public interest) may either be conditions on the
formation of a reason for acting, so that it is especially
forceful, or conditions on the consideration of the reasons for
154
acting, so that a reason for acting that might under other
conditions not lead to action would lead to action in the given
conditions. An example of the former sort of condition would be a
version of Kant's action from the motive of duty alone. When an
agent gets a reason for acting to follow some course of action
that is her duty, solely because she recognizes that it is the
only course of action available to her which can be the act of a
free agent, (in the absence of external interference) it is
supposed to result in the agent's trying to pursue that course of
action. If a similar reason for acting arises through another
process, it is not supposed to be strong enough to reliably trump
other reasons for acting that might arise; for example, if my
interest in conscientiously paying back what I owe you arises
through my wanting to please you, or even through my thinking that
conscientious payment of debts conduces to a stable society
(assuming this is the full extent of my consideration of the
behavior), then other of my reasons for acting may be more
efficacious. An example of the second type of condition,
conditions on deliberation under which certain reasons for acting
become efficacious regardless of their origin, can be found in
Hume, and, of course, in NQH. The relevant condition is the
occurrence of the process called sympathy with the public
interest.
155
In NQH, the public interest is defined by our utility
calculus. Thus, claim (03), like the other assumptions, depends
upon the still unspecified utility calculus. But claim (C3) also
depends upon a related process, sympathy with the public interest.
At the level of generality of claim (C3), sympathy with the public
interest is exactly the sort of process which led us to wonder
about the possibility of counterintuitive partitioning. To
determine whether sympathy with the public interest could
guarantee in all agents the motivational efficacy of a particular
kind of reason for acting, we will therefore need to get more
specific about what it is.
To avoid a technical discussion of Hume's account as much as
possible, I will construct an account and attribute it to NQH,
rather than to Hume. Sympathy is a process whereby one comes to
experience a passion (emotion, sentiment) because one attributes
that passion to another agent (using imagination). (I will ignore
the technical question of how this is supposed to occur.) One
engages in sympathy with the public interest by engaging in
sympathy in accordance with conventions that aim at correcting for
one's biases for those with whom one easily identifies (e.g.,
those close in time and space, friends, etc.). The aim is to
experience passions that would be aroused by a general survey of
the situation about which one makes a judgment. The passions in
question must, therefore, be ones that any agent [in the
156
population] is (at least in principle) in a position to
experience.8 We may, for example, observe of our behavior with
respect to a man we judge virtuous that we can imagine the
passions of those with whom he interacts to be pleasurable ones:
We may observe that, in displaying the
praises of any humane, beneficent man, there is
one circumstance which never fails to be insisted
on, namely, the happiness and satisfaction,
derived to society from his intercourse and good
offices. To his parents, we are apt to say, he
endears himself by his pious attachment and
duteous care still more than by the connexions of
nature. His children never feel his authority,
but when employed for their advantage. With him,
the ties of love are consolidated by beneficence
and friendship. The ties of friendship approach,
in a fond observance of each obliging office, to
those of love and inclination. His domestics and
dependants have in him a sure resource; and no
longer dread the power of fortune, but so far as
she exercises it over him. From him the hungry
receive food, the naked clothing, the ignorant
and slothful skill and industry. Like the sun,
an inferior minister of providence he cheers,
invigorates, and sustains the surrounding world.9
We do not generally recognize the emergence of painful passions in
those with whom the virtuous man interacts, although they may
emerge (if, for example, these agents compare themselves with
him) .
In NQH, things that promote the public interest, and thus are
supposed able to produce pleasant passions in us by sympathy with
the public interest, are determinable by our utility calculus.
But engaging in sympathy with the public interest need not itself
8I.e., it must be possible to attribute these passions to any such
agent.
9An Enquiry Concerning the Principles of Morals, p. 178.
157
involve the utility calculation. We observe and/or imagine that
the behavior of a person produces pleasant passions in someone.10
Sympathy with the public interest makes us then experience the
pleasant passion as we consider her behavior.
The most obvious sort of constraint imposed on the population
by the assumption of the emergence of nontrivially universal
groups in NQH is one concerning the universality of certain
passions. For each moral judgment that occurs, anyone in the
population must be able to experience some of the pleasant and
unpleasant passions that make the moral judgment true.
We need not assume that every pleasant or unpleasant passion
that may be experienced by anyone warrants some moral judgment,
however. To see this, consider for any such passion some behavior
for which the difference in utility according to our utility
calculus between its performance and nonperformance is the utility
(or disutility) of the production of that passion in one person.
An example of such a case would be the following. Suppose that I
have stolen a pen from the university bookstore, and that I am now
10If our utility calculus is an aggregate act utilitarian calculus, it
is unlikely that any such observation will be possible; for such a
calculus, no one person need experience a (discriminable) pleasant
passion on account of an action for it to be a good action, e.g., social
goods that do not consist in pleasant passions (though they may
contribute to the production of such passions). In that case, one would
most likely use imagination according to conventions only, and not
observation, to experience sympathy with the public interest. This is
not awkward when seen within Hume's view, since sympathy in Hume's
account is a type of activity of imagination. It is arguable that no
"observations" beyond those of objects of the imagination are needed
even with a different understanding of "the public interest".
158
considering whether or not to tell you about it. I know that if
you knew, you would neither turn me in nor tell anyone of my
shameful behavior, but that it would upset you to know of my
misdeed. Let us suppose that the (unpleasant) feeling of aversion
my act of telling you would cause in you is the only passion,
pleasant or unpleasant, that would be produced by either my
telling you or not. If our utility calculus is an aggregate act
utilitarian calculus, then it would seem likely that every passion
(experienced by anyone) might warrant some moral judgment. In the
case at hand, such a calculus would yield that my best reasons
support keeping my shoplifting a secret. According to NQH, since
there is no other passion that would distinguish between my
telling you and not, this must be because through sympathy one can
experience the passion that you would feel if I told you.
A more Humean utility calculus need not posit marginal
differences in utilities based on a single person's experiences,
however. Thinking about my behavior may always lead one1* - to
think of the disadvantage to the public interest of my being a
shoplifter. In that case, whether I tell you or not, the only
thing I have best reasons for doing (until I stop being a
shoplifter) is to try to make myself stop being a shoplifter.12
1 Whether one is an observer or me.
12Notice that, on such a view, my best reasons would not necessarily
support trying to prevent people from thinking about shoplifters. This
is because on this view the moral judgment that I ought not to shoplift
would be true in virtue of what people (including me) feel when they
assess my behavior and character while engaging in sympathy with the
public interest. The truth of the judgment that I ought to stop being a
159
If our utility calculus is not aggregate, then we need not expect
marginal differences in the production of pleasant or unpleasant
passions between courses of action to correspond to differences in
utility as measured by our utility calculus. Whether or not every
passion plays a role in some moral judgment will depend upon our
utility calculus; there is no obvious reason why we need assume
such a strong condition unless we are aggregate act utilitarians.
Nonetheless, according to NQH, internalistic schemata have
some passions as their source. Consequently, to be capable of
appreciating relevant moral considerations in situations of moral
interest, an agent must be capable of experiencing such passions.
To illustrate, let us suppose that some agents were not capable of
this. For example, suppose that we thought both that envy is an
unpleasant passion from which moral judgments follow, and that
there are saints who are incapable of experiencing envy. The
first assumption seems quite sound; envy is a common and
unpleasant emotion which arises easily in people, often in whole
segments of a population (as of famous people, wealthy neighbors).
I would imagine that we should expect even a non-aggregate utility
calculus to allow the emergence of moral prohibitions against
excessive ostentation when it would induce a lot of envy (and
little else). The second assumption is more suspect, but it is
certainly an empirical question whether some people are incapable
shoplifter would thus be unaffected by whether I ought also to prevent
other people from thinking about shoplifters.
160
of experiencing envy. But given this assumption, saints are not
capable of being motivated by sympathy with anyone's envy to
prevent acts such as those of (mere) excessive ostentation,
because saints do not experience envy.
For illustration, consider the case of Mother Theresa's visit
to Bangla-less, a small country bordering Bangladesh distinguished
by even more severe poverty (because its insignificance tends to
preclude its receiving aid) and its being a monarchy under the
rule of a king whose already enormous personal wealth is
incremented by a centuries old tax on the kingdom's subjects of
95% of their production, the historical purpose of which is to
support the king's excesses. After a devastating typhoon, M.T.,
who is one of the saintly few who do not experience envy, visits
to coordinate a meager food and medicine relief effort. She must
meet with the king to discuss planning issues. To this end the
king invites her to a private feast in her honor in a public
square in the desperately poor capital city. We suppose that
M.T.'s presence at the feast would not affect the success of the
relief efforts or the Bangla-lessi's opinions of M.T.; the king,
after all, only discusses issues of public relevance at feasts.
Their feasting in the proposed manner, however, would produce envy
in the hearts of the many hungry witnesses in the city. According
to our utility calculus, let us suppose, M.T. should rather than
agree to meet at a feast in public, try to convince the king to
161
meet her privately. When M.T. engages in sympathy with the public
interest, however, she does not experience the unpleasant passion
of envy, even if she imagines that there is such a passion and
that feasting in public view would arouse it in people. M.T. in
this case would be a counterexample to either claim (C2) or claim
(C3). Thus, NQH does not apply to a population including both the
subjects of Bangla-less and M.T.
Notice that there is a disparity between NQH and Hume's own
view. According to Hume,
no action can be virtuous, or morally good,
unless there be in human nature some motive to
produce it, distinct from the sense of its
morality.13
Nonetheless,
a person may perform an action merely out of
regard for its moral obligation, yet still this
supposes in human nature some distinct
principles, which are capable of producing the
action, and whose moral beauty renders the action
meritorious.14
Even lacking the original passions which are the source of the
moral worth of a course of action, an agent may find the action
worthy because of its moral obligation. This is possible for all
morally worthy courses of actions on Hume's view because of the
conventions that emerge from the recognition that there is a
nearly universal regularity in human nature. Hume's paradigm
example of this is the natural love of a parents for their
13Treatise, p. 479.
14Treatise, p. 479.
162
children.15 When the love is absent, even if the love is
impossible, a person can through sympathy with the public interest
do what one ought for the child (feed, clothe, teach, etc.)
because of the existing conventions. The parent deficient in love
can imagine the pain that her neglecting to conscientiously feed
and clothe her child would cause the child, realizing that because
of the convention of assuming a natural inclination of parents
toward doing so for their children assumed to follow from a
natural love, that no one else will; or she can realize that her
neglecting her child's needs would be a sign to others that she
lacks a natural love for her child, and imagine that the survey of
this would cause uneasiness in them. Through conventions, then,
sympathy will still provide the deficient parent with a way of
looking at her situation such that acting morally will be
choiceworthy.
This is not the case for NQH, however. According to NQH,
some courses of action may be morally obligatory but such that
some agents are not capable of finding them so. This obtains in
the Bangla-less case, for instance. Unlike the deficient parent,
M.T. lacks nothing expected of her by anyone, and thus cannot feel
secondary, conventional pressures to act according to our utility
calculus.
15See, for example, Treatise, p. 478.
163
The Bangla-less case presents no counterexample for Hume's
Treatise view because, unlike NQH, according to it there is
nothing wrong with M.T.'s meeting the king at the public feast.
That is because Hume's view (in the Treatise at least) is not a
simple utilitarian one. Since the envy of the subjects of Bangla-
less would not amount to a mark on M.T.'s character or motives
(even the subjects would not blame her for the accepted
ostentatious conventions followed by the king), there is nothing
morally wrong with her attendance, even though that feast would
not occur if she decided not to attend. There is something wrong
with scheduling the feast perhaps, or continuing the convention of
meeting at ostentatious feasts, etc., but these would amount to
marks against the character of the king, or against those of his
entourage, and are not connected to assessments of M.T.'s
behavior.16
Even in NQH, some morally worthy courses of action will
(presumably) be supported by conventions, so that a person need
not be capable of experiencing the passion initially responsible
for their moral worth. Just as for Hume’s view, we might imagine
that the moral obligations of parents to care for their children
arise fairly straightforwardly. But we cannot assume that this is
the case in general because there is not a convention available to
16Actually, the case might be more complicated. If M.T. was subject to
other scrutiny in other countries, then the case would, of course, be
further complicated.
164
back up every state of affairs to be preferred. Since our utility
calculus ranks states of affairs for individual courses of action,
we cannot expect all moral obligations to be derivable from a
conventional derivative passion in addition to the original
passion from which they spring. Consequently, we have to be more
concerned about the prospect of some agents in the population
lacking the ability to experience certain passions; if, for
instance, this obtains in a population for a passion which is the
unique passion from which a moral judgment follows, then NQH will
not hold for the population.
Another way of putting the point is as follows. Any action
that maximizes utility (according to our utility calculus) is
morally obligatory according to NQH. In order for it to be
possible for an agent to be motivated to pursue a morally
obligatory course of action by engaging in sympathy with the
public interest, the agent must be capable of experiencing at
least some of the passions the production of which makes it the
case that the course of action maximizes utility. Consequently,
any agent in a population that satisfies NQH must be capable of
experiencing some of these passions. For courses of action which
would maximize utility without the existence of conventional
social penalties to back them up, but which also would maximize
utility because of the conventionally derived social consequences
alone, agents who are capable of experiencing only the
165
conventionally derived passions can be motivated to act morally by
engaging in sympathy with the public interest. In Hume's Treatise
account, this is the case for parents' duties to care for their
children. On a utilitarian theory, it is likely that some moral
obligations will not be backed by conventional social penalties,
however. Thus, if some agents are not capable of experiencing
some morally significant passions that are not conventionally
derived, it is likely that there will be some courses of action
that are obligatory according to NQH, but such that some agents
will be incapable of being motivated to pursue them by engaging in
sympathy with the public interest.
What does this mean for the possibility of nontrivially
universal groups in NQH? How likely is it that there are saints
(by our definition)? The real Mother Theresa herself seems to
deny being unlike other people in any such way; for instance, she
has said that she experiences resistance to acting with her
conscience. But even if there are not saints, the problems
associated with developing capacities for various passions will be
relevant here. We may think that children gradually gain the
capacity to experience the various passions, or that some people
are physiologically incapable of various passions. For the
reasons discussed in the Bangla-less case, the consequentialism of
NQH would appear to make it likely that if there are such agents
166
in the population, then there will be counterexamples either to
claim (C2) or to claim (C3).
We may, however, with Hume, think that the passions are
universally available, for human agents at least. Hume himself is
unwilling even to grant that humans and animals differ in the
passions they experience, and makes an analogy from the
differences between humans and animals to the differences among
humans.17 Animals, he thinks, experience at least some of the
same passions as humans, such as the pride nightingales take in
their singing, hounds in their sense of smell, or male peacocks in
their beauty.18 Even if we accept Hume's attribution of humanly
experienced passions to animals, it does not appear that animals
are capable of sympathetically experiencing passions in the manner
that humans can (although they may, and, according to Hume, do
engage in sympathy). It is unlikely that an unspectacular male
peacock will exhibit any signs that it experiences love for a more
beautiful rival, though perhaps it will appear to experience
humility or envy. Indeed, such instances did not escape Hume's
watchful eye, for he notes that
Envy and malice are passions very remarkable in
animals. They are perhaps more common than pity;
17See, for example, Treatise, p. 610.
18Hume does not say that all of the passions of humans are experienced
by all creatures, but neither does he deny this. But what kind of
evidence might be given for the experiences of humility in ants? The
answer would seem to be that we cannot reasonably take any position on
the capacities of some animals for some passions.
167
as requiring less effort of thought and
imagination.19
Similarly, there are variations in the abilities of human agents
to effectively experience passions, including passions that would
affect an agent's capacity for making moral judgments in NQH.
That there are people on earth who experience very little envy, or
who are so well-fed as to find it difficult to imagine feeling
envy at someone's eating abundantly would seem to indicate that
such differences would jeopardize NQH's applicability to ordinary
human populations. We are familiar with altruists, if not with
saints, and with egocentric agents, if not with isolated atomic
agents. Limitations on agents' effective abilities to experience
passions in various circumstances, if not their susceptibility to
experiencing the passion in general, give us reason to think that
NQH will partition agents into disjoint groups according to the
types of obligations by which they may be bound, that is, so that
agents will be divided by the kinds of moral obligations by which
they may be bound into groups with no common members. (We do not
yet have reason to think that NQH will partition agents into
disjoint groups according to disjoint classes of obligations by
which they might be bound, however; that is, we do not yet have
reason to think that there are no kinds of obligations that bind
agents in different groups.) This point becomes even more certain
19Treatise, p. 398.
168
when we look at the role played by the understanding in Hume's
account of moral experience.
According to Hume, the differences in the way animals
experience the passions are explained by the differences and
deficiencies of animals in relation to humans in their abilities
to imagine and think. Animals are not, for example, able to
comprehend or attend to complex relations and conventions, and
hence cannot make moral judgments which depend upon them. But
furthermore:
Men are superior to beasts principally by the
superiority of their reason; and they are the
degrees of the same faculty, which set such an
infinite difference betwixt one man and another.
All the advantages of art are owing to human
reason; and where fortune is not very capricious,
the most considerable part of these advantages
must fall to the share of the prudent and
sagacious.20
Animals do have reason; they learn from experience relations of
cause and effect. The reason of an animal is limited in that its
inferences never rest on
any process of argument or reasoning, by which he
[it] concludes that like events must follow like
objects, and that the course of nature will
always be regular in its
operations. . . .Animals, therefore, are not
guided in these inferences by reasoning
But:
Neither are children: Neither are the generality
of mankind, in their ordinary actions and
conclusions: Neither are philosophers
themselves, who, in the active parts of life,
20Treatise, p. 610.
169
are, in the main, the same with the vulgar, and
are governed by the same maxims.21
The difference between the understandings of various humans and
between humans and animals Hume discusses briefly, in a footnote.
Agents differ in understanding for reasons including: (i)
differences in attention, memory, and observation, which affect
the number and legitimacy of inductively formed hypotheses22; (ii)
in capacities for manipulating (mentally) chains of consequences
of various lengths and complexities; and (iii) in the degree of
bias from (Humeanly defined) illegitimately-formed beliefs.23 All
of these differences affect the influence of an agent's
understanding on inferences that are not directly influenced by
reason, which are the most frequently occurring type of
inferences, and comprehends those common to both humans and
animals. We suppose any human agent to be superior to any animal
in all capacities relevant to this part of the understanding.
Human agents also are superior or inferior to one another with
respect to the degree to which they possess each of these
capacities.
The limitations enumerated render animals incapable of
experiencing passions with causes "plac'd either in the mind or
external objects"; the causes of theirs "lie solely in the
21An Enquiry Concerning Human Understanding, p. 106.
22Humeanly defined by custom.
23An Enquiry Concerning Human Understanding, p. 107.
170
body."24 Virtue and vice, for example, because they are qualities
of character, can only be ascribed to an agent through the use of
understanding and imagination in addition to a sympathetic
experience of passions. Thus, these limitations prevent animals
from making inferences of the complexity required for making moral
judgments. Since animals have a more limited set of causes of
their passions, the range of circumstances in which they are
capable of experiencing particular passions is more limited than
that of humans. A population including animals will therefore not
satisfy NQH.
That a population including animals will not satisfy NQH is
obviously not a very interesting conclusion by itself. The point
has wider application, however. The list of relevant limitations
makes it reasonable to join Hume in the conclusion that human
beings also differ among themselves in their abilities to make
moral judgments. Infants would appear to have the same status as
many animals; they experience passions and learn by experience,
but appear to lack the attention span needed for observing complex
relations (whether or not they possess the mental capacity for
processing them inferentially) • The capacity for moral judgment
would seem to be out of the question at such an early stage since
even the simplest perception of self-consciousness is a very
complex observation. In addition, since engaging in sympathy with
24Treatise, p. 326.
171
the public interest is, in part, a convention-driven activity, it
involves an awareness of social relations that very young children
lack, if for no reason other than inexperience. From infancy to
adulthood the understanding develops. Since the type of reason
involved in ordinary actions is not supposed to be syllogistic,
but is rather supposed to be of the very same sort through that
development, the part of adult understanding most relevant to
experiencing the passions is on a continuum with that part of the
understanding of animals. From infancy, human agents acquire more
and more skill at forming causal chains (even if perhaps only to
some limit).
For any utility calculus likely to be proposed, variations in
the understanding of agents seem to virtually guarantee that human
populations will include a wide range of abilities to experience
morally relevant passions in situations of moral interest with
respect to NQH. If an agent cannot, for example, see that the
result of taking out a large loan now will almost certainly be
that it will be in default two years hence, then she is not
capable of appreciating relevant moral considerations with respect
to NQH, even though she might be quite susceptible to experiencing
the pain and disutility of the action's likely consequences were
she to imagine them as such. Even the best judgment of such an
agent is not good enough to reflect an appreciation of the
relevant moral considerations according to NQH, because of the
172
limitations of her understanding.25 (I am assuming that aversion
to risk is not an issue in this case. The agent in question
neither expects to win the lottery nor receive an inheritance in
that time. She merely does not realize that all similar loans
have ended in default. Perhaps she takes as evidence that she can
pay off the loan that she is able to get it.)
There is a second side to the influence of variations in
human agents' understanding on the partitioning question. In
addition to its causing variations in agents' effective abilities
to experience passions, the existence of these variations seems to
diminish the possibility that agents in human populations are
uniformly capable of experiencing passions through the complex
activity of the understanding (imagination) called sympathy with
the public interest. The complexity of the process will certainly
exclude very young children from being capable of appreciating
(any) moral considerations. The complexity of our utility
calculus will determine how large a portion of the population will
be excluded completely from being capable of appreciating any
moral considerations, and whether the rest of the population is
partitioned by how well they are able to do so. This last point
leads us back where we started: with one of the classic arguments
against utilitarianism as a theory of moral experience, that it is
too complicated to have anything to do with the behavior of the
250r of her experience. The part of the understanding at work here
includes this.
173
average person. We cannot comment further on this without a more
specific characterization of our utility calculus.26
Our straw man NQH has not surprisingly been knocked down.
But some of what we discovered for NQH carries over to other, more
solid theories. Most straightforwardly, it carries over to some
more reasonable Humean theories. Even when we jettison our
utility calculus, variations in agents' effective abilities to
experience passions in diverse circumstances seems to be a source
of partitioning for any Humean account that depends upon the
sympathetic experience of passions (using the imagination and
understanding) for the appreciation of moral considerations.
Obviously, to make this point more generally than for theories
that accept Hume's theory of human nature, we would have to use an
appropriate theory of human nature. Nonetheless we can make the
argument for the more Humean of Humean theories, just as we have
for NQH. Moreover, for Hume's own view (the one given in the
Treatise at least) problems for assuming nontrivially universal
groups of internalistic schemata emerge for populations of
interest arise not only from the dependence of the capacity to be
26A related potential source of partitioning arises if Hume is taken
straightforwardly when he remarks that animals "are but little
susceptible either of the pleasures or pains of the imagination"
(Treatise, p. 397). If the experiences of such pleasures and pains are
less vivacious for infants, and progressively more so with improvements
in an agent's understanding, then an agent's ability to experience
passions through sympathy with the public interest will certainly vary
through development; for only a lively imagined passion (which is itself
pleasant or painful) will become an actual passion through sympathy. In
that case, the capacity to appreciate (experience) moral considerations
will certainly vary through an agent's development. So much the more
so, given the complexity of the notion of the public interest.
174
motivated by sympathy with the public interest on the mechanism of
sympathy, but from that capacity's dependence on the mechanism of
reason's corrective influence in moral reflection.
Although the Humean strategy seems to commit a theory to
partitioning agents in ordinary human populations into disjoint
groups according to the kinds of moral judgments that apply to
them, it does not necessarily partition them into disjoint groups
according to disjoint classes of obligations by which they might
be bound. Internalistic moral theories that follow this strategy
thus do not necessarily sanction implausible moral relativism.
Such theories nonetheless seem unable to adequately answer the
charge that they sanction moral relativism, however; given that
agents are partitioned according to the kinds of moral judgments
that may apply to them because of differences in various
motivation-affecting capacities, the burden of proof seems to rest
on the side of the internalistic moral theory. To defend such a
theory, we should at least be able to provide some reason to think
that there will be some kinds of moral judgments that bind all
agents despite their varying capacities. The Humean strategy
seems unable to offer any such reasons because the array of
motivation-affecting capacities that internalistic judgments
depend upon is so large.
We should also notice that arguments for the existence of
nontrivially universal groups of internalistic schemata are
175
unavoidably dependent upon some part of a theory of human nature
supplied by a particular moral theory. Our results are to this
extent limited--they cannot be made more generally applicable to
the theories considered as independent of a theory of human
nature. Internalistic moral theories are to the same extent
necessarily limited, however; the part of a theory of human nature
needed for us to determine whether agents are partitioned into
disjoint groups according to the kinds of moral judgments that
apply to them is the part that these theories depend upon to make
intelligible their defenses of the Basic Internalistic Premise
(i.e., that there are some moral obligations such that the fact
that an agent ought to D entails she has sufficient motive for
doing D). We should not accept a defense of the Basic
Internalistic Premise that uses the claim that recognizing that
something is good can be motivationally efficacious, unless we can
determine (even if only by empirical methods) whether the
capacities involved in this mechanism (of recognizing things as
good and being motivated by such recognition) are different for
different members of a population of interest, and if so whether
it follows from this that populations of interest might be
partitioned into disjoint groups according to the kinds of moral
judgments that apply to them.
Without such an account, it must also remain an open question
whether, for a given theory and population, variations in agents'
176
motivation-affecting capacities cause agents to be partitioned
into disjoint groups according to disjoint classes of obligations
by which they might be bound. The charge of moral relativism,
thus, cannot be answered without such an account.
The Kantian strategy
An internalistic moral theory can plausibly be maintained
only in conjunction with some part of a theory of human nature
that can justify assuming that populations of interest are not
partitioned into disjoint groups according to disjoint groups of
internalistic schemata that apply to them nontrivially. I have
argued that although the Humean strategy can, in principle, be
made into a plausible internalistic moral theory, we lack a theory
of human nature that is sufficiently detailed for the Humean's
needs. Moreover, as was remarked above, it is difficult to see
how the account can be filled in.
The reason for the Humean strategy's dim prospects for
avoiding partitioning is ultimately that sympathy with the public
interest is a very complicated process, depending on a wide array
of human capacities. We have no reason to think that humans
satisfy the required regularities in advance of comprehensive
empirical study of human psychology and motivation. This does not
mean these defenses must fail, but it does mean that the Humean
should regard the truth of the Basic Internalistic Premise (i.e.,
the claim that there are some moral obligations such that the fact
177
that an agent ought to D entails she has sufficient motive for
doing D) as an open question.
There is an alternative strategy that relies upon a less
complicated array of capacities to define moral points of view-- \
those that characterize practical reasoning. This strategy is not
necessarily at odds with Hume's. Hume himself defines moral
points of view partly by the exercise of reason as a corrective
influence. For Hume, reason is yet another faculty by which the
possibility of moral obligation must be conditioned, however. The
situation is more straightforward for a Kantian view. For such
views, reason is the sole faculty that defines moral points of
view. Although reason itself involves a complicated array of
capacities, I will argue that this strategy provides some basis
for arguing against the charge that human agents will be
partitioned into disjoint groups according to the kinds of moral
judgments that apply to them.
The most straightforward version of the Kantian strategy
fares no better than Hume's at providing a defense of the
existence of nontrivially universal groups of internalistic
schemata.27 This is the view that identifies a single perspective
as most rational, and thus, as legitimately moral. It claims (i)
that all normal human agents are capable of being rational to the
27As I remarked above, this is not to say that it must ultimately fail,
but only that it seems on the surface likely to fail, and that the
resources required to defend the existence of nontrivially universal
groups of schemata are not currently available.
178
degree required by morality--so that every normal human agent can
see, for example, which types of behavior are objectively
choiceworthy and which are objectively unchoiceworthy; and (ii)
that the recognition that some behavior is objectively
choiceworthy or objectively unchoiceworthy will move one to engage
in it or to avoid engaging in it, respectively. The most serious
problem with this view is that normal human agents differ in their
abilities to reason and to apply reason practically, and these
differences clearly affect what they judge rational and
irrational. For example, as clear as it was to Kant that suicide
for the purpose of ending suffering is irrational, many people
thoughtfully and reflectively believe that just the opposite may
be true (and that Kant was wrong).28
A more sophisticated Kantian strategy is more promising.29
For illustration, consider a view that defines moral points of
view as those that involve the use of reason to criticize
prospective choices of behavior.30 More specifically, define
Internalistic Kantianism (IK, for short) as a theory that claims
28The point is not merely that many people don't conform to Kant's
conception of rationality, but that many genuinely find his position
unconvincing.
29I am especially indebted to Barbara Herman for helping me to see how
the more plausible Kantian strategy might work.
30This is not the only view available to the internalistic Kantian. I
take as an example a Kantian view that makes very limited claims about
the capacities of human agents, because it serves to illustrate the
benefits of the Kantian strategy fairly straightforwardly, and requires
less controversial claims about human capacities than would a stronger
position. However, from a Kantian perspective, this view may seem to be
one of the less attractive internalistic options, because it is
incompatible with some of the most important Kantian claims (for
instance, that moral obligations are categorically binding).
179
that (i) an agent has an internalistic obligation not to engage in
some behavior if and only if she can recognize that the behavior
is irrational; and (ii) recognition that some behavior is
irrational will move any person to avoid engaging in it. IK will
thus define a moral point of view as one that enables an agent to
recognize irrational behavior. The hope of Internalistic Kantians
is that despite agents' differences in their abilities to reason
and to apply reasoning practically, they will recognize some of
the same sorts of behavior as irrational if they consider it. For
example, it might be thought that everyone can see that killing an
innocent person is irrational (and that everyone would see this if
they engaged in rational criticism of their behavior). Given that
we accept IK's assumption that a person would not choose to engage
in behavior she recognizes is irrational, we will, consequently,
have a defense of the existence of nontrivially universal groups
of schemata.
As was the case for NQH, observations about how people behave
present a straightforward but inadequate reason for thinking that
IK's defense of the existence of nontrivially universal groups of
schemata is mistaken. Since thinking some behavior irrational
does not always motivate people to refrain from engaging in it,
one might think that recognition of the irrationality of some
behavior is by itself motivationally inert--that it can only
180
motivate people for whom caring about being rational is a
motivating attitude.
Internalistic Kantianism denies that recognition of the
irrationality of some behavior is motivationally inert.
Internalistic Kantians will claim that even if some people deny
that they care about being rational, that does not mean that they
are (or would be) motivationally unaffected by the recognition
that some behavior is irrational. The best evidence for this is
that it seems that any agent who engages in some behavior because
she believes it furthers some end will (other things being equal)
cease to pursue it if she recognizes that it does not in fact
further that end. Internalistic Kantians will presumably say that
the recognition that their behavior is irrational is precisely
what affects agents' behavior in such cases. Even if we do not go
so far as to deny the existence of agents who act irrationally
while fully recognizing their behavior as such, it would not be
particularly radical to claim that agents who fail to satisfy this
minimal condition of rationality might not be possible subjects of
ordinary (internalistic) moral judgments.31
The question of whether IK plausibly defends the existence of
nontrivially universal groups of internalistic schemata thus
31Such observations may suggest that a less plausible (single moral
point of view) version of an internalistic Kantian theory assumes too
much because it assumes that everyone can recognize the same things as
irrational, even though they don't seem to be uniformly capable of being
motivated to refrain from the same types of behavior that Kant
considered irrational. The prospects of IK do not seem to be affected
by such observations, however.
181
hinges on whether there are any kinds of behavior that are
(nearly) universally recognizable as irrational.
To answer this last question, one might suggest an appeal to
common sense. People generally seem to behave as though they
uniformly think some types of immoral behavior are irrational.
Consider, for example, the kinds of judgments typically passed on
criminals. For the most serious crimes--serial murder, for
instance--the most common response seems to be to consider the
criminal insane in the sense of being incomprehensible or
impossible to understand, but sane in the sense of knowing what
she is doing. In more Kantian terms, we might say that the serial
killer is considered to be acting irrationally, but also to be
rational in the sense that she chooses from her available options,
and in conformity with a plan that she also chooses. We assume
that she could have done otherwise, and that she would have had
she reflected differently on what she was doing.
People do, of course, have strong and mixed feelings about
those who commit crimes as serious as serial killing. It might be
thought that their feelings about such criminals' capacities for
rational action are so mixed with reactions of anger and horror
that people do not judge with any accuracy that these criminals
are capable of acting rationally. The suggestion is presumably
that people only insist on thinking such criminals are capable of
acting rationally so that their crimes will not go unpunished.
182
This thought only seems to challenge the point being made here if
we suppose that people are always mistaken when they think that
serial killers are capable of acting rationally. But making this
claim amounts to assuming what this argument is trying to show; it
supports the Internalistic Kantian's argument for the existence of
nontrivially universal groups of internalistic schemata directly,
because it assumes an even stronger claim: that engaging in serial
killing is irrational, and that if someone does not recognize
serial killing as irrational, then she is not capable of rational
agency (rather than that the latter might, at most, follow from
her being incapable of recognizing serial killing as irrational).
Commonsense judgments about other seriously criminal behavior
also seem to assume that we think such behavior irrational, and
generally observably so. Consider, for example, that, although
they may often engage in minor legal and moral transgressions,
most people are generally law-abiding and do not engage in
egregiously immoral behavior. Why? Is it just because most of us
think ourselves incapable of avoiding punishment? It seems not--
otherwise our reactions toward criminals would resemble jealousy
far more than outrage or frustration. Moreover, most people don't
normally consider engaging in seriously immoral or criminal
behavior. This does not seem only to be because people are
conditioned not to notice profitable opportunities or not to
consider them if they are noticed (though people certainly do tend
183
to be so conditioned), but to be because such behavior is not
generally seen by people as a genuine option. This absence may be
thought to suggest an incompatibility--or, in more Kantian terms,
that seriously immoral or criminal behavior is in some way
inconsistent with the way they conceive of themselves when acting.
If human agents generally must in acting conceive of themselves in
similar ways, then we may all rely upon similar assumptions. If
those assumptions turn out to conflict with assumptions involved
in choosing to engage in certain kinds of behavior, then, (nearly)
all human agents might, in principle, be capable of coming to see
behavior of those kinds as irrational.32
Against such arguments, Hume's famous remark that, "'Tis not
contrary to reason to prefer the destruction of the whole world to
the scratching of my finger,"33 might be offered. Hume's point
does not seem to apply here, however. The Internalistic Kantian
strategy outlined here does not depend upon the possibility of
deriving a contradiction from reason alone, independently of the
conditions of human agency. The Internalistic Kantian suggestion
is rather that those conditions themselves create some kind of
contradiction with some types of (immoral) behavior.34 (The most
obvious candidates seem to be suicide and murder.)
32This is just the sort of position Thomas Nagel takes in The
Possibility of Altruism (Princeton, New Jersey: Princeton University
Press, 1978) (though Nagel's view is not identical with IK).
33Treatise, p.416.
34The idea is similar to that of Kant's contradiction in will test.
184
At the same time, there is a danger to this strategy if we
attempt to derive the whole of morality from this form of
contradiction, as, for example, Thomas Nagel does in The
Possibility of Altruism (in which he presents an internalistic
Kantian moral theory that follows a strategy similar to this one) .
Attempting to derive the whole of morality from this form of
contradiction presents a problem for Internalistic Kantianism, at
least on the surface, because human beings differ in their
abilities to reason and to apply reason practically. If the
conditions of human agency upon which all morally relevant
contradictions depend are too complicated, then we cannot expect
them to be recognizable as such by most human agents. If not,
then most people would be unable to recognize the contradictions
involved in immoral behavior, and would therefore be unable to see
the relevant behavior as irrational. This is incompatible with
IK, however, because IK is internalistic.
If the Kantian strategy follows Nagel's model, then the
problems involved in defending the existence of nontrivially
universal groups of internalistic schemata thus become similar to
those that arose for the Humean strategy. The hope held out by
the Kantian strategy came from its seeming to offer an account of
how different agents with different capacities might come to
similar conclusions in practical reasoning. If the account of the
nature of rational criticism of behavior is unresponsive to such
185
differences, then it seems unlikely to offer any great advantage
over the Humean strategy.
In addition, it is hard to see how a Kantian theory can
really be internalistic. Internalistic Kantian theories seem to
rely upon the possibility of an ideal theory. But if there is an
ideal theory, why isn't it true? Why should we think that agents
are all right in their assessments when they apply reason to the
best of their ability given that there is a more perfect reason to
which their reasoning merely (at best) approximates? If we answer
this last question negatively, we will not be espousing an
internalistic theory.35 The internalistic Kantian's commitment to
assuming the existence of internalistic obligations seems to force
her to a characterization of moral thinking that, like Nagel's,
assumes universal access to judgment according to the ideal
theory, and thus lacks the advantages of the Kantian strategy over
the Humean one.
Even if Internalistic Kantians can defend the existence of
nontrivially universal groups of internalistic schemata, a
successful defense of their existence is not by itself enough to
guarantee that an internalistic moral theory is plausible. A
guarantee of their existence is merely a minimal condition for the
plausibility of an internalistic theory. Internalistic Kantianism
35See chapter one.
186
may meet this minimal condition, but still fail to be plausible
because it fails to satisfy either the congruence or completeness
restriction, i.e., it may still fail to render the right moral
obligations internalistic (relative to our pretheoretical moral
intuitions).36 There is, in fact, reason to worry that such a
theory will satisfy these restrictions. An internalistic Kantian
theory that fares better than the theories that follow the Humean
strategy by avoiding the problem of Nagel's version of the
strategy will necessarily ascribe differences in moral thinking to
different agents. Consequently, even if such a theory guarantees
that there are nontrivially universal groups of schemata (i.e.,
some types of behavior everyone can choose or avoid, and will
choose or avoid if they adopt a moral point of view), and even if
we grant the more contentious claim that agents are always capable
of being moved in accordance with reason to the extent that they
can apply it in deliberating, we should still expect that there
will be other types of behavior that agents will differ in their
moral appraisals of. The possibility thus remains that different
people will be bound by different moral obligations in a way that
does not match our moral intuitions. I will return to this issue
in chapter six.
To defend the Basic Internalistic Premise (i.e., the claim
that there are some moral obligations such that the fact that an
36See chapter one.
187
agent ought to D entails she has sufficient motive for doing D)
using the claim that recognizing that something is good can be
motivationally efficacious, an internalistic moral theory must
include a part of a theory of human nature. For internalistic
theories that follow both the Kantian and the Humean strategy,
however, there is a tension between being able to characterize
human nature sufficiently thoroughly to defend the existence of
nontrivially universal groups of internalistic schemata, and
respecting the diversity in motivacion-affecting capacities of
ordinary human agents. A natural response to this kind of
observation is a third strategy for defending the existence of
nontrivially universal groups of internalistic schemata.
The social programming strategy
The third traditional approach to guaranteeing the existence
of nontrivially universal internalistic schemata or groups of
internalistic schemata for a population is to tailor the
population to fit the moral theory, and/or to tailor the moral
theory to fit the population. The foremost problem for using the
latter method by itself is that to restrict a theory adequately
for this purpose would apparently ruin the prospects of the
theory's applicability across populations normally considered to
be of interest. Thus, it would not help an internalistic moral
theory avoid the charge of implausible moral relativism. The
former method, to tailor populations of interest to fit a moral
188
theory--by itself or in combination with tailoring a moral theory
to fit the population--is a more promising strategy, and has found
expression in a number of philosophers’ works.37 (Hobbes is
usually taken to be the paradigm.) The fundamental idea is that
given a minimal amount of fairly uncontroversial theory about
agents in populations taken to be of interest, we can construct a
system of conventions backed by coercive measures that are
sufficient to guarantee the existence of nontrivially universal
groups either because they create reasons for acting in every
agent in the population that will be persuasive when considered
from any moral perspective, or because they make some reasons for
acting that agents in the population already have more persuasive
(when considered from any moral perspective) than they otherwise
would be.
A good example of the type of effort that would be required
for a theory to follow this strategy is described fairly
straightforwardly by Hobbes. Although Hobbes is concerned with
describing a program for how certain motivation-affecting features
(such as, ends, reasons for acting, and circumstances) across a
potentially diverse population can be aligned, it is not at all
clear that he can consistently be read as offering an
37Not necessarily as part of an internalistic program, however. For
example, Aristotle, Hobbes, and Bentham all offer proposals of this sort
(and it would probably be an error to regard any of them as offering an
internalistic moral theory).
189
internalistic moral theory.38 Let us consider, then, a version of
Hobbes' program that might provide support for an internalistic
moral theory. The internalistic Hobbesian would begin with the
assumption that human beings in populations of interest
(inhabitants of predominantly Christian European countries in the
seventeenth century) all share certain motivating attitudes that
they regard as morally significant: an interest in the lives (and
among Christians, in the salvation) of themselves (and, perhaps,
of their families). Of course, uniformity of motivating attitudes
could not be enough to guarantee the existence of nontrivially
universal groups of schemata because whether an agent has
sufficient motive for behaving in some manner depends on more than
just her motivating attitudes.39 The shared motivating attitudes
are supposed to guarantee some kind of motivational uniformity
among nearly all agents (in populations of interest) through
38Some passages in Leviathan (C. B. Macpherson, ed. (New York: Penguin
Books, 1986)) certainly suggest that Hobbes' account is internalistic--
particularly those in which he seems to claim that one can only be
obligated to do what it is reasonable for one to do. See, for example,
Leviathan, 1.13, pp. 186-187; 1.14, p.196; 11.26, p. 314.
I do not, however, wish to defend the internalistic view as
interpretation of Hobbes' own position. There are various reasons
against reading Hobbes as offering an internalistic theory; for
instance, his account of what it is for something to be reasonable for
one to do relies upon claims about what human agents can reasonably
will, rather than what human agents actually will. See, for example,
Leviathan, 1.14, p.192. Moreover, if Hobbes is read as offering an
internalistic moral theory, then he perhaps should not be read as
offering one that guarantees the existence of nontrivially universal
groups of schemata using the social programming strategy alone, because
all human agents are morally bound by the Fundamental Law Of Nature
independently of any external influences (though, of course, its
manifestations in action may vary according to the particulars of civil
law). (See Leviathan, 1.14, pp. 189-190.)
39I discuss this point in some detail in chapter three.
190
uniform coercion arising from credible threats of an absolute
monarch who demonstrates and reinforces his power over the lives
and salvation of citizens. An internalistic moral theory might
use such an argument to show that, despite their differences in
motivation-affecting capacities and interests, nearly all agents
(in populations of interest) will be capable of recognizing as
morally obligatory any morally obligatory behavior that is backed
up by such threats, because the threats themselves will amount to
moral considerations that make engaging in such behavior morally
obligatory.40 Therefore, nearly all agents will have sufficient
motive for engaging in such behavior when they adopt a moral
perspective.41
The primary problem with an internalistic moral theory that
tries to use this strategy to defend against the charge of
sanctioning counterintuitive moral relativism is that if we are
not in the conditions that make internalistic morality possible
(as apparently we are not, for the internalistic Hobbesian), then,
internalistic moral judgments simply do not apply to us. In that
case, we do not seem to be any better off in defending our
40Though they are not the only moral considerations that make engaging
in such behavior morally obligatory.
41The sovereign of course is not a part of the population, so this plan
cannot actually generate any kind of total (nontrivial) universality if
the sovereign is a monarch. Some groups of internalistic schemata might
be nontrivially universal in the unlikely case that particular monarchs
find themselves similar to their subjects in situation and motivational
attitudes in some cases.
191
internalistic moral theory against the charge that it sanctions
counterintuitive moral relativism.
It may be helpful to briefly consider Aristotle's view, since
it uses social programming in a way that makes internalistic moral
judgments appropriate. According to Aristotle, everyone ought to
act like the excellent person. Internalistic judgments could
reasonably be expected to hold only of the excellent, however.
Social programming thus may make an internalistic moral theory
hold of more people; by making more people excellent, more people
are capable of being motivated by the (true) recognition that some
behavior is good. Nonetheless, most people (those who are not
excellent) cannot be the subjects of true internalistic moral
judgments. This is not a problem for Aristotle's account, of
course, because it does not assume that moral judgments are
generally internalistic--there are supposed to be other sorts of
reasons for why one should be excellent. For an internalistic
moral theory, however, being forced to excuse from internalistic
moral judgment those who have not been made into proper subjects
for internalistic judgments invites a charge of moral relativism
that seems to run at least as much against commonsense moral
intuitions as the one to which the social programming strategy was
supposed to respond.
192
Chapter Sixt
Can an Internalistic Moral Theory Be Plausible?
I have argued that neither the Basic Internalistic Premise,
(BIP) There are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for
doing D,
nor the traditional method of defending it entails moral
relativism.1 Thus, plausible internalistic moral theories are not
impossible for the reasons usually given. However, I have also
argued that given the most common strategy for defending the Basic
Internalistic Premise, a reasonable characterization of moral
thinking and deliberation will characterize different behavior as
moral thinking (or deliberation) depending upon the agent. Though
these processes may yield the same results in similar
circumstances, it seems unlikely that members of ordinary human
populations will be sufficiently uniform in capacities and
circumstances that they will always be subject to the same kinds
of internalistic moral requirements. Moreover, as I argued in
chapter five, on some of the most commonly advanced types of
theories, it is difficult to argue for even weaker uniformities.
Thus, it seems that internalistic moral theories, even those that
are structured so as to avoid sanctioning moral relativism, are
likely to imply that human populations are partitioned according
to the kinds of internalistic schemata that will yield judgments
1Not, at any rate, on any of the usual interpretations of moral
relativism that are obviously incompatible with commonsense moral
intuitions.
193
about them, because of the diversity in capacities and
circumstances of ordinary human agents.
Sanctioning such partitioning might be thought to undermine
the plausibility of a moral theory. One might, for example, begin
with the assumption, common among moral philosophers, that it
cannot be the case that two conflicting principles can both be
legitimately moral. This assumption seems to be incompatible with
the claim that two perspectives that lead agents who adopt them to
make different kinds of choices can both be legitimately moral
points of view. W. D. Ross's view is a fairly representative of
this traditional kind of position. Rules involved in
interpartitional conflict would clearly fall into the last of the
his categories of moral rules:
There are . . .some whose correctness is self
evident . . .; some whose rightness can be
deduced from a self-evident rule by applying the
rule to the universal conditions of human nature;
some which can be derived from a self-evident
rule by applying the rule to the actual
conditions of the particular society; and some
which cannot be justified even on that basis and
must be discarded as based on incorrect views
about human nature or physical nature, or on
views which were true in past conditions of
society but have ceased to be true today.2
Ross himself even claims that the reconciliation of conflicts not
unlike those suggested by the claim that agents are partitioned
according to the kinds of moral obligations by which they may be
2W. D. Ross, Foundations of Ethics: the Gifford Lectures (New York:
Oxford University Press, 1960), p. 313.
194
bound3 yields to moral philosophy its "practical value”, "by
removing a great discouragement from the moral life.”4 If Ross is
right, then sanctioning partitioning in populations of interest
will certainly count against a moral theory.
Unlike extreme moral relativism, however, partitioning might
also be thought to enhance a theory's plausibility without
challenging commonsense intuitions. The idea that human
populations are partitioned by the kinds of judgments that apply
to them is, after all, not a new one; its roots extend at least to
Aristotle.5 W. D. Falk articulates the intuition that human
populations are partitioned by the kinds of judgments that apply
to them quite forcefully:
We cannot divorce the moral eligibility of
actions from their effects, from the means to be
employed and the alternatives to be sacrificed:
any of these factors may vary with the
circumstances of the case. Nor can we divorce it
from the way people are made, since their
dispositions may vary, both by nature and by
education, and what is good for one need not be
good for another. Nor, finally, can we divorce
3Conflicts "between the rules current in a single society . . .[and]
between the rules accepted in different societies". W. D. Ross,
Foundations of Ethics, p. 313.
4W. D. Ross, Foundations of Ethics, p. 314.
5For Aristotle, what kind of behavior is appropriate to a person depends
upon her vulnerabilities, her experience, and her personality. What the
excellent person should do is what she sees to be fine. The brave, for
instance, should "stand firm against what is and appears frightening to
a human being, because it is fine to stand firm and shameful to fail."
(Aristotle, Nicomachean Ethics, trans. Terence Irwin (Indianapolis:
Hackett Publishing Company, 1985), p. 78.) Those who are not
excellent--including, for example, those who are rash, or ignorant
either of their vulnerabilities or of what is fine--cannot do what the
excellent person should do. What is appropriate to the rash may be to
defer to the judgment of the brave when deciding whether to stand firm
in the face of danger, because she is a poor judge of whether or not
such behavior is fine or shameful.
195
it from people's insight, since we can have no
duty of which we are ignorant, and hence what we
ought to promote is what as far as we can see we
must do, and not what we would have to do if we
had perfect insight. Of some people we can
expect no more than that they should be true to
the values in which they have been brought up or
which prevail in their social environment, and we
may even think that, on reflection, they ought
often to accept the limits of their own
understanding and not try to evolve new
evaluations of their own. Of others we will
think that they have not only a right but a duty
to question and to reverse preconceived notions
of the eligibility of actions in the light of
some new insight into their nature and effects.
On the basis of natural morality no accepted
standard is sacrosanct, and the search for the
ideal morality, based on perfect insight, is a
never-ending adventure charged with
responsibility and as full of potential perils as
rewards.6
If we agree with Falk, then we are likely to think that a moral
theory's partitioning populations of interest according to certain
of their capacities is an asset rather than a flaw.7 Moreover, we
will see it as an asset to internalistic theories that they allow
us to arrive at Falk's intuition in a principled manner;
constructing an internalistic moral theory leads us naturally to
support Falk's intuition. In addition, our analysis of
internalistic moral theories provides a substantive interpretation
of what the truth of such an intuition might consist in; different
agents are partitioned according to the kinds of judgments that
6W. D. Falk, "Morals without Faith," in Ought, Reasons, and Morality,
pp. 178-179.
7In considering whether such partitioning is required by internalistic
moral theories, on some level we have been elaborating a way of finding
room for Falk's observations in received moral theories--how his
spectrum from "primary morality" to "mature morality" might be
represented from within them.
196
will apply to them because of their possessing differing
capacities affecting motivation.
Falk’s version of the intuition might also be thought to be
too radical, however, because he claims that populations will be
partitioned according to what kind of moral perspective is
appropriate for their deliberations, even though the different
perspectives may be irreconcilable, even in particular instances
of judgment.8 If so, then we have to accept that there are
irreconcilable conflicts among moral obligations or values. Such
conflicts are sometimes thought to imply the unpopular
existentialist conclusion that in cases in which one must act in
the face of such conflicts, morality can provide no guidance.
The classic example is Jean-Paul Sartre's case of the student
trying to decide whether to stay to support his mother or leave
the country to join the Free French forces. Sartre argues that in
this case, because of the conflicting values, moral thinking
cannot help to decide.9 The student, it is supposed, might
conscientiously adopt a perspective that will lead him to stay to
support his mother, or equally conscientiously adopt a perspective
that will lead him to leave the country to join the Free French
Nonetheless Falk seems to think that the levels are ordered, with each
perspective better or worse than another, each a better or worse
approximation to the ideal moral perspective. Since the appropriate
■sub-ideals" for moral deliberation for agents in different partitions
may be irreconcilable, however, the claim of linearity does not weaken
the partitioning.
9Jean-Paul Sartre, "The Humanism of Existentialism," in Essays in
Existentialism (Secaucus, New Jersey: Citadel Press, 1965), pp. 42-45.
197
forces. Sartre concludes from this that morality is in the hands
of the agent, to be invented through action rather than merely
manifested by it.10 Although Sartre's example involves a case in
which two different perspectives are supposed to be equally
appropriate for a single agent, the case reveals an analogous
problem for interpartitional conflicts of moral perspectives. For
our purposes, the case suggests that, since two agents may
(equally conscientiously) choose differently, when deciding to act
we cannot look to an independent fact about whether one or the
other choice is the morally correct one, because there is no such
fact. More precisely, once we assume that there are two truly
moral perspectives such that an agent who adopts the one that is
appropriate for her will choose to act contrary to how an agent in
similar circumstances for whom the other perspective is
appropriate would choose if she adopted the other, we deny that
either of the agents' choices is the morally correct one. It
would seem, then, that morality is not what helps one decide in
such cases; if anything decides how one will act in cases of
interpartitional conflict, it would seem to be one's capacities
and other features that affect how one may be moved to act--that
is, whatever determines one's appropriate moral perspective.
10Since "The content [of a judgment about an action] is always concrete
and thereby unforseeable; there is always the element of invention. The
one thing that counts is knowing whether the inventing that has been
done, has been done in the name of freedom." ("The Humanism of
Existentialism," p. 59.)
198
There are many perspectives from which conclusions like
Sartre's will seem unpalatable. Regardless of whether we join
Sartre in drawing the stronger conclusion that morality is in the
hands of the agent, however, we may think that entailing the
weaker claim, that moral thinking cannot guide us in difficult
decisions, is by itself enough to render a moral theory extremely
unpalatable. In particular, we may wonder what purpose moral
thinking can have if it is not to help us make difficult
decisions.
Falk responds to this challenge by denying that the fact that
in some cases there may be irreconcilable conflicts of values
implies that in such cases moral thinking does not help to
decide.11 He points out that even when faced with such decisions
we are not completely indifferent to the alternatives when they
conflict. Falk claims that in such cases moral thinking involves
choosing conscientiously, even if in the face of uncertainty.
Morality is thus not silent on our decision, even though no single
choice is the morally right one (and the choices are
incommensurable). The point becomes particularly potent given
that the conflicts with which we are concerned are between and not
within partitions. (Although there is no reason to suppose in
iiiphe reply is one commonly advanced by those who posit conflicts among
values. See also, for example, Thomas Nagel's "The Fragmentation of
Value" (in Mortal Questions (New York: Cambridge University Press,
1991)), Michael Stocker's Plural and Conflicting Values (New York:
Oxford University Press, 1990), pp. 179, 194-195.
199
advance that irreconcilable conflicts of values will not occur
within a partition, the problems that these conflicts create will
be no different than those that arise for moral theories that do
not sanction partitioning.) In a situation in which there are
interpartitional conflicts of moral values, a member of a given
partition will have an opinion about which choices are best (or
can come to have one). For such cases, Falk's argument seems to
be strongest.
The case for Falk's reply is not as strong as it might seem,
however. Suppose that we accept partitions like those that Falk
describes. In particular, suppose that some people will be bound
to follow conventional norms in cases in which others will be
bound to question them. In such cases the values (or orderings of
values) of members of one partition will conflict with those of
the other, even though both values are properly considered moral
values. Consider a case of interpartitional conflict of values
analogous to Sartre's case. A conflict of this sort might arise
over whether to submit to conscription in an unjust war (or, more
generally, whether to obey laws that are prima facie unjust) . A
moral theory that sanctions partitions might be forced to grant
that for some it will be right only to submit and for others only
to refuse, i.e., that obedience to one's (not unjust) government
and conscientious action are incommensurable in a significant way
200
(even when they conflict).12 If so, then deciding whether or not
to submit to conscription in an unjust war may well be a decision
which remains difficult regardless of whether one has decided
which is the better course of action, because knowing that others
will disagree after equally conscientious deliberation may
undermine one's confidence in one's own judgment. Such difficulty
seems to arise through various mechanisms. One may wonder how
others can conscientiously disagree with one when one has so much
in common with them. When morality seems to ask more of oneself
than of others, one may even resent that one is not a person who
could find the other course preferable or wish one were relevantly
different. One may respond to such thoughts by wondering who one
is, and, as a consequence, one may cease to be certain that one
has been adequately reflective in judging the case. When a
judgment that an agent ought to follow a particular course of
action follows from a nontrivially universal schema or a schema in
a nontrivially universal group (that is, when the judgment
involves moral considerations that are universally recognizable as
such), then this problem will not arise; one's judgment in such
cases is likely to seem justified, or, in any case, it is unlikely
12Contrast this point with Bernard Williams' point about remainders (see
"Ethical Consistency," in Problems of the Self (New York: Cambridge
University Press, 1993), pp. 172-173). Williams' point is that in
cases in which an agent has an opinion about what it would be best to
do, the value of a rejected alternative does not vanish. The point I am
making here is that in some cases an agent may be unable to come to a
decision about what it would be best to do because of her awareness of a
conflict of values.
201
to seem immediately suspect. For judgments that are only possible
among agents in an one's partition, however, (i.e., those
following from an internalistic schema in a nontrivially universal
group with respect to members of the partition including oneself,
but not in a nontrivially universal group for other populations of
interest) we cannot expect that one will be equally confident in
one's judgment. What one will think will depend upon the
situation--in particular, whether one thinks the conflict offers
reason to doubt one's judgment. Moral thinking, even if,
following Falk, it includes conscientious deliberating, will not
necessarily help to decide such conflicts.
Falk's reply to Sartre thus seems to reject Sartre's
conclusions more straightforwardly than ordinary experience
suggests is possible. Real and difficult conflicts sometimes show
us that the way we view ourselves, or at least how we effectively
view ourselves when we deliberate, is less defensible than we seem
to assume. If, for example, one regards oneself as a free agent
who can choose and be fully responsible for her choices, then the
occurrence of conflicts of obligations can show one the inadequacy
of the view: one sometimes finds that, through no fault of one's
own, one has no acceptable options. These conflicts provide good
reasons to wonder if the way one regards one's reasons while
deliberating is appropriate; in our example, one's commitment to
acting the way a "free" agent would might be undermined by a
202
realization that one does not know what a "free" agent in this
sense would be, i.e., how any agent confronted with such a
conflict might act freely and be fully responsible for her
choices. Falk is right to point out that, contrary to Sartre, it
does not follow from such conflicts that we cannot reasonably
guide our deliberations; we do have ways of deciding. However,
neither does it follow, as Falk would have it, that we can
reasonably regard our best judgment always and with equal
confidence as an adequate guide, regardless of whether it is an
easy activity or a struggle to exercise it, or whether it directs
us contrary to our culture, religious beliefs, family bonds, or
majority opinion. To a certain extent, conflict is able to
undermine our entire way of looking at ourselves and our behavior.
Conflict certainly can illuminate for us our uncertainty about
whether the use of reason, imagination, faith, emotion, or
something else constitutes our best judgment. And where we
knowingly allow some (personally) customary way of regarding our
reasons and ends to determine our behavior in situations in which
the decision to do so matters, we have already made a choice to
which, contrary to Falk, we may not feel committed. Falk's reply
to Sartre is thus unsatisfying.
We are now in a position to better respond to the challenge
inspired by Sartre, however. The worry was that moral theories
that sanction partitioning are unable to yield to moral thinking
203
the role of helping us to make difficult decisions, at least not
for those involving interpartitional conflicts of obligations. In
considering conflicts of this type, we noticed that ordinary
experience does not support the intuition that moral thinking is
supposed to fill this role; it is contrary to experience to assume
that there is in every instance of conflicting values a best
choice. For example, conscientious refusal and submission to
conscription to fight a prima facie unjust war are both difficult
behaviors to choose. One is likely to feel far less justified in
acting in either of these ways than, say, in choosing not to
murder for expedience or in saving a drowning child. We seem to
expect that such decisions are necessarily difficult because they
involve irreconcilable conflict.13
If we take such a view of difficult conflicts of values, then
it would seem that, far from making a moral theory contrary to
commonsense intuitions, entailing such irreconcilable
interpartitional conflicts of values may allow moral theories that
sanction partitioning to provide good accounts of commonsense
intuitions in cases in which many theories are notoriously
130nce again, it may be helpful to compare the point with Williams'
point about remainders ("Ethical Consistency", p. 173). The point here
is not that in cases of interpartitional conflict agents may or will
have regrets no matter how they choose. The point is rather that agents
deliberating in the face of such conflicts may be unable to experience
any choice as best because of their recognition of the conflict.
204
inadequate (i.e., moral theories that posit a single or small set
of things of intrinsic value).14
Moreover, moral theories that sanction partitioning give us
an account of how some very difficult cases fit with more
straightforward ones, even though difficult and straightforward
cases often seem to involve obligations with the same sources.
Cases in which agreement beyond a narrow partition of the
population is impossible, are often in the periphery of what we
regard as moral thinking. As individuals we may find that we have
exactly the strength of commitment to certain "peripheral" norms
as for globally shared moral norms, even where we know others will
not be committed at all. One may feel equally committed to a norm
expressing a duty not to steal as to one expressing a duty to be
agreeable at dinner (though one may of course think one more
important than the other). We may have these same strong
commitments because for us (individually) the commitments may have
similar sources for all norms. This can be explained on the
account developed here; agents are in the same partition because
14In addition, it is often argued that not only is there nothing
problematic about positing conflicts among moral values generally, but
that commonsense demands that we do. See, for example, Michael
Stocker's Plural and Conflicting Values.
Although Stocker's treatment of the issues involved in positing
conflicts among values is very thorough, I have focused on Falk's more
vague treatment for two reasons. My first reason is that the case for
positing interpartitional conflicts of values seems to be an especially
strong one. The more important reason is that Stocker does not directly
address interpartitional conflicts of values, and the view of conflicts
of values that he advances may be incompatible with the claim that there
are interpartitional conflicts of values (as given). (Support for the
latter,, for instance, is suggested by his claim that we must deny that
incommensurable values are incomparable (p. 176).)
205
they have similar motivation-affecting capacities of moral
significance. Other agents in other partitions cannot share their
experiences in every case. It might be that the most significant
point of contrast someone is able to distinguish between being
agreeable at dinner and refraining from stealing is that one
pleases mother and the other pleases God. For such a person, the
strength and content of her commitments to each norm are bound to
differ.
A second possible advantage of internalistic moral theories
Another potential advantage for moral theories that support
partitioning seems to be that they might provide principled
support for agent-centered features. The advantage of having
agent-centered features is supposed to be that such features
provide a way of guaranteeing that agents may have consuming
personal commitments. By incorporating agent-centered features, a
moral theory is able to claim that moral principles cannot
generally require agents to give up what they care deeply about.
This in turn provides a basis for claiming that people need not
consider their deepest and most important commitments to be
subject to revision.15
15This, in turn, might be thought attractive because such commitments
are what give us reason to live and to act in the first place. (See,
for example, Bernard Williams' "Persons, Character, and Morality" (in
Moral Luck (New York: Cambridge University Press, 1986).)
206
Samuel Scheffler isolates two routes by which moral theories
might describe some agents as acting morally when they fail to
produce the state of affairs which is best overall (for our
purposes, this need not be taken in a strictly consequentialist
sense) by preferring to pursue their own interests: moral theories
might incorporate either “agent-centered-restrictions" or "agent-
centered-prerogatives". By "agent-centered restrictions" what are
meant are
restrictions on action which have the effect of
denying that there is any non-agent-relative
principle for ranking overall states of affairs
from best to worst such that it is always
permissible to produce the best available state
of affairs so characterized.16
Agent-centered prerogatives merely "have the function of denying
that one ought always to do what would have the best outcome
overall".17 They do not deny that one may always do what will
produce the state of affairs which is best overall.
Internalistic judgments can be true only if their subjects
are capable of preferring to act in accordance with them (though
not necessarily through the recognition of the truth of the
judgment) . Some moral judgments will not be true if their
subjects have commitments that are too entrenched and too firm to
be subject to being reevaluated in the face of other
considerations; to this limited degree we can say that the Basic
16Samuel Scheffler, The Rejection of Consequential ism (New York: Oxford
University Press, 1987), pp. 2-3.
^ The Rejection of Consequential ism, p. 17.
207
Internalistic Premise sanctions agent-centered restrictions. Some
moral judgments will not be true if their subjects have
commitments that are too entrenched to be seen by them as
recalcitrant in the face of other considerations; assuming the
Basic Internalistic Premise, commitments may in this way be seen
as carving out agent-centered prerogatives for some agents. To
assume the Basic Internalistic Premise, then, is to assume that at
least formally there are agent-centered prerogatives and
restrictions.
These observations do not appear to support the assumption
that either agent-centered restrictions or agent-centered
prerogatives will arise substantively (rather than merely
formally) given the Basic Internalistic Premise, only that they
may. Nor do they appear to support the assumption that, if agent-
centered restrictions or prerogatives do arise, they will differ
substantively across persons, i.e., that they will be truly agent-
centered rather than merely agency- or human-agency centered.18
18What I have in mind can be illustrated by the following example.
Suppose it is not possible for a human being to both act morally and not
love some other human being. Suppose also that human agent A loves B
and is not capable of seriously reconsidering her commitments regarding
B when considering general ways of behaving or during particular
instances of deliberation; her love for B is not subject to being given
up in the face of considerations that merely support courses of action
which are mutually exclusive of some arising from the love for B. For
example, A will not hesitate to choose to act to save B's life over
saving the lives of perfect strangers. (A may of course choose to save
the lives of perfect strangers over merely doing something nice for B,
but this choice hardly seems to need to turn on B's giving up or
modifying her commitments to B.) No true internalistic judgment will
assert that A ought to prefer the lives of perfect strangers to B's.
The love for B turns out then to be one of those commitments which seems
to carve out at least an agent-centered prerogative for A. But suppose
208
Whether either assumption follows from the Basic Internalistic
Premise will be determined by the answer to the question of
whether partitioning of agents according to which kinds of
judgments apply to them arises according to internalistic moral
theories. If any substantive kind of agent-centered restriction
or prerogative does arise given the Basic Internalistic Premise,
then we will have found a principled route to assumptions like
Scheffler's.19 Similarly, the more easily accessible question of
whether agent-centered restrictions or prerogatives are entailed
by a particular internalistic moral theory will be settled
affirmatively if the theory sanctions partitioning;20 in that case
Scheffler's agent-centered features will, consequently, arise for
that theory in a principled manner.
What motivates Scheffler's interest in the two agent-centered
features is the thought that there is room within any reasonable
morality for consuming personal commitments. This type of
intuition springs up most often in the form of a criticism of
utilitarianism: that it asks too much of people; that most normal
human agents, at any rate, ones that we are inclined to think of
that, along the lines on which we began, every human being capable of
good actions loves another with exactly the firmness of A's love for B.
Then every person will have an "agent-centered" prerogative analogous to
that carved out by A's love for B. The description of the prerogative
therefore need not in such a case be agent-relative at all, but only
relative to the type of relationship from which the prerogative springs.
19In any case, the search for evidence of partitioning is analogous to
the endeavor of evaluating whether Scheffler's agent-centered features
are properly called agent-centered.
20In that case, it will also answer the question of whether either
feature is truly agent-centered affirmatively.
209
as good or healthy, have commitments which they weigh more heavily
in relation to other commitments than would be judged reasonable
or fair by other agents. Bernard Williams and Peter Railton are
recent proponents of arguments along these lines. (Even a century
before, arguments of this kind were apparently rehearsed with
sufficient frequency that Sidgwick uses the plural to describe
proponents (though he cites as an instance just J. Grote).21)
Even if the intuition underlying these arguments is
granted, one need not think it requires that a moral theory must
incorporate agent-centered features to be plausible, however.
Sidgwick, for example, grants the intuition, but argues that his
utilitarianism leaves sufficient room for agents to prefer their
interests despite its lack of truly agent-centered features.22 If
Sidgwick is right, then the emergence of agent-centered features
will therefore not necessarily be a prerequisite for the
possibility of a moral theory's permitting consuming personal
commitments. To notice that Sidgwick is right about the latter,
however, we need not consider the merits of Sidgwick's larger
21Henry Sidgwick, The Methods of Ethics, p. 432. Actually, Sidgwick
himself uses the same kind of argument against James Martineau's
system--see proper spheres argument in ”'Idiopsychological Ethics'" in
Mind 12 (1887) .
22Sidgwick's arguments would, if successful, appear to seriously
undermine the intuitive support Scheffler claims for the need for agent-
centered features. Scheffler's treatment of Sidgwick's response is
remarkably unresponsive to this point. See The Rejection of
Consequential ism, p. 16.
Despite this, (as I will argue) intuitive support for Scheffler’s two
features, precisely because they are only formally agent-centered is not
undermined by Sidgwick's arguments.
210
argument. In fact, we need look no farther than Scheffler's two
features. As we have already noticed, his "agent-centered"
restrictions and prerogatives are really only formally agent-
centered. Whether or not they are truly agent-centered (rather
than merely agency- or human-agency-centered) will not depend upon
whether a given moral theory permits consuming personal
commitments, but on how relevantly homogeneous populations of
moral agents are. To permit consuming personal commitments, only
formal agent-centricity is required. Consequently, Scheffler's
two features do capture the intuition underlying them, regardless
of whether on a given moral theory which manifests such features
they turn out to be truly agent-centered.
Thus, the Basic Internalistic Premise, following the strategy
developed, provides a basis for assuming that agents may have
consuming personal commitments, regardless of whether partitioning
arises or how much partitioning arises. Of course, whether it
does so in a way that is attractive cannot be determined in
general, but only for particular theories.
211
Chapter Savant
A Related Challenge to Traditional Internalistic Moral Theories
I have argued that internalistic moral theories can be
defended against the most common charge against them--that they
are implausible because they sanction moral relativism. Moreover,
as I argued in chapter six, they can be considered especially
attractive for reasons that connect with the motivation for that
line of attack. An internalistic moral theory can only be
defended against the charge of sanctioning an implausible version
of moral relativism if it is much more detailed and complicated
than moral theories usually are, however. At the very least,
internalistic moral theories should suggest that the Basic
Internalistic Premise,
(BIP) There are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for
doing D,
is true in light of the possible relativistic consequences. I
have suggested that at least one type of internalistic moral
theory does this--a particular sort of Kantian theory.
Although no received internalistic moral theory seems to be
sufficiently thorough to be truly defensible, I have argued that
the problem is primarily a matter of missing accounts--
particularly in the Kantian case--rather than of some sort of
principled incompatibility with commonsense moral intuitions, as
the original charge suggests. Although it may be hard for
particular theories to provide a positive defense of the claim
212
that they do not sanction moral relativism, internalistic moral
theories do not necessarily sanction moral relativism.
There is, however, another charge of principled
incompatibility with commonsense moral intuitions that can be
leveled against the internalistic moral theories with which we
have been concerned. Like the most common one, this charge
involves the question of whether they sanction moral relativism.
Unlike the common one, however, this charge is not lodged against
the Basic Internalistic Premise, but against theories that defend
it using the claims that (a) recognizing that something is good
can be motivationally efficacious; and (b) there are some good
things such that we are able to recognize them as good.
Although the argument seems to be firmly rooted in
commonsense thinking about morality, the only place I have seen it
offered explicitly is in the early editions of Sidgwick's Methods
of Ethics A The substance of the charge is that internalistic
moral theories that defend the Basic Internalistic Premise using
claims (a) and (b) cannot credibly defend the claim that some or
all moral judgments are objectively binding. Since Sidgwick's
version of the argument is not explicitly about what I have
labeled internalistic moral theories, I will present the argument
in two parts. First, I will reconstruct Sidgwick's argument,
which concerns a particular case of internalistic obligation. I
■^See, for example, the introduction to the first edition (London:
Macmillan and Co., 1874), especially pp. 6-13.
will then present a related and more general argument, more or
less independently of Sidgwick's treatment.
Since my interest here is in developing a certain line of
argument, and not in defending an interpretation of Sidgwick, I
will ignore exegetical issues, and merely develop one
interpretation of the argument. The case Sidgwick's argument
concerns involves a normal, mature moral agent, A, who wants to do
what is right. Sidgwick begins by making three assumptions about
morality and motivation:
(1) There is a single class of moral obligations in every
situation (for every agent). It cannot be the case that
(at one time, in one situation) an agent ought to D and
ought not to D; D is either in the class of moral
obligations or not. (Of course, it might be that an agent
ought to do either D or not D, but not both.)
(2) Some moral obligations entail that acting so as to fulfill
them is subjectively most reasonable. In particular, some
moral obligations by which A is bound are such that A is
capable of seeing acting so as to fulfill them as most
reasonable.
(3) A has sufficient motive for engaging in some behavior if
and only if doing so is subjectively most reasonable.
From assumptions (1), (2), and (3), we can conclude:
(4) In cases of obligation of the sort described in (2), A can
only have sufficient motive for engaging in behavior that
is a member of a single class of subjectively most
reasonable behavior (in the sense described in (1)). In
such cases, it cannot be that (at one time, in one
situation) it is subjectively most reasonable for an agent
to D and not to D; D is either in the class of subjectively
most reasonable behavior or not. (Of course, it might be
that what is subjectively most reasonable for the agent is
to do either D or not D, but not both.)
Conclusion (4) is problematic because of Sidgwick's
assumptions about what he calls "the dualism of practical
214
reason1,2--the claim that both duty and self-interest are equally
subjectively reasonable:
(5) There is a dualism of practical reason. In particular,
behavior that promotes the happiness of humans generally,
and which is therefore morally obligatory, and behavior
that promotes one's own happiness, are both equally and
maximally subjectively reasonable.3
In addition, Sidgwick takes for granted what seems natural enough:
(6) Promoting one's own happiness may conflict with promoting
the happiness of humans generally, and it may do so in
cases of obligation of the sort described in (2).
It follows from (4), (5), and (6) that:
(7) Neither that it promotes one's own happiness nor that it
promotes the happiness of humans generally by itself
determines whether some behavior is subjectively most
reasonable.
What then does make behavior subjectively most reasonable for A
and, hence, morally obligatory for A when it is required by a
moral obligation of the sort described in (2)? (4), (5), and (6)
immediately preclude our claiming that what is subjectively most
reasonable for A is what promotes both her own and the general
happiness. They also preclude our simply claiming that what is
subjectively most reasonable is anything that promotes either A's
or the general happiness. If we were to claim that what is
subjectively most reasonable is anything that promotes either A's
or the general happiness, then we would be committed to claiming
^The idea of which is, as Sidgwick points out, to be found in Butler.
See, for example, The Methods of Ethics, seventh edition, pp. 508-509;
Joseph Butler, Five Sermons Preached at the Rolls Chapel (Indianapolis:
Hackett Publishing Company, 1985), especially pp. 17-18.
3Sidgwick’s "dualism of practical reason’ is actually further
complicated by the subjective reasonableness of pursuing both one's own
perfection and the perfection of human beings generally.
215
that, in cases in which promoting A's happiness requires that A
not promote the general happiness, refraining from promoting her
own happiness and refraining from promoting the general happiness
each might be subjectively most reasonable for A. We would thus
be committed to claiming that promoting neither A's nor the
general happiness is subjectively most reasonable, in
contradiction to (5). It looks, then, as if we must deny that A
might be bound by a moral obligation of the sort described in (2).
Given (3), i.e., that A's having sufficient motive for engaging in
some behavior depends upon it being subjectively most reasonable,
this would mean that we would have to deny that A might be bound
by an internalistic moral obligation.
Rather than deny that agents may be bound by internalistic
moral obligations, Sidgwick concludes that we should instead
reject (2) and (3). By rejecting both (2) and (3), we at least
avoid directly denying that A might be bound by an internalistic
moral obligation. Having rejected (2) and (3), we might, for
example, claim that agents have sufficient motive for acting in
any way that is subjectively reasonable (regardless of whether it
is subjectively most reasonable); in that case, Sidgwick's
argument would present no difficulty for claiming that some moral
obligations are such that agents bound by them have sufficient
motive for acting so as to fulfill them. The price of rescuing
216
internalistic moral obligations would seem to be the loss of our
model of moral motivation, however.
There are, of course, other ways to avoid being forced to
deny that A might be bound by an internalistic moral obligation--
we might deny (1), for example. Rather than consider whether
Sidgwick's argument can be accommodated by an internalistic
theory, however, it will be useful to consider a related argument
that applies to a wider range of internalistic moral theories.
A related and more general argument
An argument that is related to Sidgwick's and more general
seems to show that traditional internalistic moral theories rely
upon models of moral motivation that should be rejected. More
precisely, the argument that I will develop in this section seems
to show that internalistic moral theories that defend the Basic
Internalistic Premise using the claims that (a) recognizing that
something is good can be motivationally efficacious, and (b) there
are some good things such that we are able to recognize them as
good, cannot credibly defend the claim that some or all moral
judgments are objectively binding--at least given the usual
interpretation of this claim, so that it implies that there are
some moral judgments that are necessarily binding on all human
agents, irrespective of their differences.
I have argued that, to be plausible, internalistic moral
theories that defend the Basic Internalistic Premise using claims
217
(a) and (b) should consider different and possibly incompatible
ways of looking at one's situation to amount to legitimately moral
points of view.4 Plausible internalistic moral theories that
follow this model, therefore, seem to be committed to holding that
moral judgments can only be assessed relative to agents or to
groups of agents who share certain capacities (that determine
which moral perspective is appropriate for them). Thus, if a
moral judgment is binding according to any single moral
perspective, that would not by itself suggest that the judgment is
universally binding, but, at most, that the moral judgment is
binding for agents for whom that perspective is appropriate. To
credibly claim that some moral judgments are objectively binding,
proponents of internalistic theories will, consequently, need to
have something further to say about such judgments.
It is possible for plausible internalistic theories to avoid
the unpalatable consequence of extreme moral relativism (on these
grounds) if there turn out to be moral judgments that apply to all
real human moral agents. An internalistic moral theory can
support the claim that there are such judgments, if according to
the theory there are some moral judgments that are binding
according to every moral perspective that is appropriate for some
living mature moral agent. An internalistic moral theory might
provide an empirical defense of the latter claim, along the lines
4See chapter three.
218
of Sidgwick's argument that all commonsense moral principles
effectively support utilitarianism, or an a priori defense, along
the lines of Butler's argument that conscience, self-interest, and
utilitarianism are reconciled by the moral government of God.5
Even if such strategies can be used to defend an
internalistic moral theory against the charge of sanctioning moral
relativism, however, merely showing that there are some moral
judgments that are binding according to all moral perspectives
that are appropriate for some living mature moral agent does not
provide a principled reason for why some moral judgments apply
universally. Merely showing that there are judgments that are
binding on all living mature moral agents therefore does not show
that some moral judgments are objectively binding, that is,
necessarily binding on all agents irrespective their differences.
For example, suppose that we observe that a prohibition on
torturing the innocent is sustained by every moral perspective
appropriate for some living human being. Suppose that according
to the moral perspective appropriate to agent J, the prohibition
is derived from a right of all human beings to noninterference,
and according to the moral perspective appropriate to agent C, the
prohibition is derived from the harmfulness of torture. Without
any argument beyond this observation, we have no explanation of
why the prohibition on torturing the innocent is universally
51 consider some other strategies in chapter five.
219
binding. We would not, for instance, have an account of how a
prohibition on torturing the innocent differs from imperatives
which issue from some but not all legitimate moral perspectives;
for example, from an absolute prohibition on deception. Merely
demonstrating that no living mature moral agent is morally
permitted to torture the innocent is not enough to provide a
principled reason for why this is the case. Thus, it provides no
demonstration that some mature moral agent might be morally
permitted to torture the innocent; for example, it might be
permitted by the moral perspective appropriate for some agent who
will come to moral maturity in the future (or whose life has
already passed and whose appropriate moral perspective is long
forgotten). Without a principled reason for why judgments that
are binding on all living mature moral agents are binding on all
living mature moral agents, we can have no reason to think that
judgments that are effectively universally binding are therefore
necessarily universally binding on all agents, irrespective of
their differences.
Might a plausible internalistic moral theory have a
principled reason to offer? The most straightforward reason one
might offer is that only one moral perspective generates the true
moral obligations. In that case, conformity with that moral
perspective is what determines which judgments necessarily apply
universally. Assuming that the universal bindingness of these
220
judgments is not defended on the basis of agent-dependent
considerations, this defense would allow us to defend such
judgments as objectively binding.
The most straightforward reason is apparently not available
to internalistic moral theories, however. Given that to be
plausible internalistic moral theories must claim that different
moral perspectives are appropriate for different agents, it will
not in general be possible for them to claim in addition that one
of the moral perspectives that is appropriate for some agent
generates the moral judgments that necessarily apply universally.
To do so would be to deny that moral perspectives that are
appropriate for different agents may be incompatible. A
particular internalistic theory may, of course, provide grounds
for denying this. Although we should not assume that there are no
such theories that are plausible, the burden of proof seems quite
clearly to rest on internalistic theories. As I have already
argued,6 even on some of the most promising internalistic moral
theories, the variation among mature human agents in factors that
affect their capacities for recognizing the moral obligations by
which they are bound and in factors that affect their being
motivated to fulfill them seems sufficiently wide to offer many
opportunities for such conflict between legitimately moral
perspectives. Moreover, to reap the pluralistic benefits that
6See chapter five.
221
were claimed in chapter six to accrue from sanctioning different
moral perspectives requires that the moral perspectives are
incompatible.
The challenge to principled objectivity posed by positing
different moral perspectives is further complicated by Sidgwick's
assumption of "the dualism of practical reason.” Sidgwick's
argument from "the dualism of practical reason" suggests that
plausible internalistic moral theories are committed to an even
stronger claim about the differences among legitimately moral
perspectives than my arguments suggest. As I argued in chapter
three, plausible traditional internalistic theories, i.e., those
that defend the Basic Internalistic Premise (i.e., the claim that
there are some moral obligations such that the fact that an agent
A ought to D entails A has sufficient motive for doing D) using
the claims that (a) recognizing that something is good can be
motivationally efficacious, and (b) there are some good things
such that we are able to recognize them as good, will hold that
what one ought to do is what one sees as most choiceworthy when
one adopts a moral point of view. A moral point of view, in turn,
is supposed to have some special status among possible ways of
looking at one's situation--it might be defended as the most
reflective, rational, or distinctly human perspective, depending
on the theory. If we agree with Sidgwick that both duty and self-
interest are equally subjectively reasonable, and therefore that
222
they are equally subjectively choiceworthy, then plausible
internalistic theories must therefore posit at least two distinct
moral perspectives, corresponding to viewing one's situation with
a regard to duty and viewing one's situation with a regard to
self-interest. Moreover, since, as Sidgwick points out, duty and
self-interest can and sometimes do conflict (on the surface at
least), even for a single agent there will be no single
legitimately moral point of view that is appropriate to her.
If both duty and self-interest are in fact equally
subjectively choiceworthy, then the fact that a moral judgment is
binding from the point of view either of duty or of self-interest
would, therefore, not by itself suggest that the judgment is
objectively binding, because the other perspective might sanction
a competing judgment. On the other hand, it does seem to be
possible to claim that moral judgments in cases in which it can be
demonstrated that both points of view support the same kinds of
behavior are objectively binding.
"The dualism of practical reason" presents a more serious
problem for internalistic moral theories than merely that they
might fail to suggest that moral judgments are objectively
binding, however. Since both competing points of view are
legitimately moral, and plausible traditional internalistic moral
theories7 define what agents (internalistically) ought to do by
7By traditional internalistic moral theories, I mean those that defend
the Basic Internalistic Premise (i.e., the claim that there are some
223
how they behave when they assume a moral point of view,
internalistic moral theories seem to be committed to claiming that
in some cases judgments of the form "A ought to do D“ and "A ought
not to do D" (and hence, "It is not the case that A ought to do
D") will both be true.8 Internalistic moral theories that try to
defend the objectivity of moral judgments thus seem to be
committed to absurdity.
The self-effacing alternative
Although Sidgwick's rendering of this problem is not well-
known, his solution is very much so: he proposes that objective
moral truth may be inaccessible to ordinary agents, and,
therefore, that attempting to apply the true moral theory (which
describes what one objectively ought to do) does not generally
constitute adopting a point of view that is morally best to adopt.
That Sidgwick's proposal is commonly regarded as notorious
and unappealing is testimony to the pervasiveness of internalistic
intuitions--for if being morally obligated never implies that
simply in virtue of being obligated one has sufficient motive for
acting to fulfill it, it should not even be surprising if
moral obligations such that the fact that an agent A ought to D entails
A has sufficient motive for doing D) using the claims that (a)
recognizing that something is good can be motivationally efficacious;
and (b) there are some good things such that we are able to recognize
them as good.
8And, a fortiori, there will be some cases in which the judgments "A
ought to do D" and, for similarly situated B, "B ought not to do D" will
both be true. Sidgwick seems to think that this is the more obviously
unacceptable result. I have already argued that this kind of conflict
need not in general be considered problematic. See chapter six.
224
attempting to apply the true moral theory does not generally lead
one to act morally. Nonetheless the aim of Sidgwick's proposal is
to accommodate such intuitions without giving up moral
objectivity, in light of the challenges presented by “the dualism
of practical reason" and the existence of competing commonsense
approaches to moral decision-making. To analyze the prospects of
internalistic moral theories, it will, therefore, be useful to
consider Sidgwick's solution. Once again, since my interest here
is in developing a certain line of argument, and not in defending
an interpretation of Sidgwick, I will ignore exegetical issues,
and merely develop an interpretation of the argument.
The overall structure of Sidgwick's argument is fairly
straightforward. Even in cases in which equally appropriate
legitimately moral points of view conflict, the possibility of
objectively binding moral judgments requires that it cannot be the
case that an agent morally ought both to engage in some behavior
and to refrain from it. As a consequence, Sidgwick begins with
the assumption that a moral theory must decide the question of
what agents morally ought to do in such cases. The problem for
moral theories that take seriously competing moral points of view
is to explain how to decide this question. Such theories
obviously cannot simply claim that the question is decided by
which perspective is most appropriate because, by hypothesis, the
competing perspectives are equally appropriate. Sidgwick's
225
solution is to offer an account of agents' moral obligations that
is independent of what makes points of view moral or choiceworthy.
The independent explanation that Sidgwick offers is that a version
of utilitarianism, Universalistic Hedonism, is the true moral
theory--that is, that this version of utilitarianism describes the
moral judgments that are objectively binding. This independent
account of agents' moral obligations is then supposed to justify
the claim that acting morally is most choiceworthy even in cases
in which contradictory courses of action are subjectively
choiceworthy. If his arguments are successful, Sidgwick thus
explains how his account provides a defense of the Basic
Internalistic Premise (i.e., the claim that there are some moral
obligations such that the fact that an agent A ought to D entails
A has sufficient motive for doing D) that can be reconciled with
the dualism of practical reason (i.e., the claim that duty and
self-interest are equally subjectively reasonable) and the
existence of competing commonsense moral points of view. Since it
is primarily the reconciliation of the latter claim with the Basic
Internalistic Premise that is of interest for defending an
internalistic moral theory, I will focus on that aspect of his
account.
To understand Sidgwick's solution, it is helpful to consider
how Sidgwick sees the tension between competing moral points of
view and objectivity. Sidgwick bases the problem of competing
226
moral points of view on the assumption that if some course of
action is right it must be possible to see that it is most
choiceworthy.9 He assumes that the perspectives that present
themselves to agents as ultimately reasonable (reasonable on
reflection) are prima facie moral points of view. Since when
applied in practice, these points of view sometimes support
incompatible courses of action, Sidgwick assumes that the
different prima facie moral points of view cannot all give correct
accounts of what the right courses of action are. When taken by
themselves, these points of view are equally choiceworthy, and
hence prima facie all equally legitimate moral points of view.
Because of "the dualism of practical reason" and the
arguments I advanced in chapter three, points of view that have
equal prima facie claims to being considered moral points of view
seem to support conflicting behavior (in some cases). As a
consequence, if one accepts the Basic Internalistic Premise, then
one cannot plausibly maintain that one of these points of view
defines the true moral theory in virtue of its having a superior
prima facie claim to being considered a moral point of view than
competing points of view. Sidgwick thus argues instead that if
there is a theory of which all of these prima facie moral points
of view can be viewed as approximations or projections, then that
theory will be the most choiceworthy, and, hence, most justified
9See, for example. The Methods of Ethics, seventh edition, pp. xvii-xx,
pp. 5-14.
227
theory because it provides the best systematization of what we
find most choiceworthy. Sidgwick then argues that utilitarianism
is such a theory. We need not consider the question of whether
this is a good argument in support of the truth of utilitarianism,
nor whether utilitarianism really does best systematize
commonsense morality here. The important point in understanding
Sidgwick's strategy and assessing its efficacy is that the
argument in support of it is independent of whether action
according to utilitarianism is subjectively most choiceworthy.
Let us grant, then, for the sake of understanding Sidgwick's
strategy for reconciling internalistic intuitions with
objectivity, that utilitarianism is the best systematization of
commonsense morality.
By defending a moral theory as most justified on the basis of
its ability to account for the general accuracy of points of view
likely to be considered legitimately moral points of view, prima
facie moral points of view are yielded a kind of primacy over the
best moral theory; whenever the best moral theory supports
behavior that departs radically from what one would choose if one
adopted them, the guidance of prima facie moral points of view is
reasonably to be preferred.10 Sidgwick's strategy thus ensures
10 The Methods of Ethics, seventh edition, pp. 460-475. This part of
the strategy is the least explicitly treated by Sidgwick. Sidgwick does
argue that the guidance of prima facie moral perspectives is reasonably
to be preferred, but he seems to argue on grounds other than that his
methodology requires it.
228
that, unless the case is one in which prima facie moral points of
view conflict, when one adopts a prima facie moral perspective,
the behavior that one finds most choiceworthy is morally
obligatory according to the most justified moral theory, and thus
is objectively morally obligatory. Nonetheless no single prima
facie moral point of view will yield true moral judgments in all
cases.
In cases in which adopting different prima facie moral points
of view would lead one to choose to behave differently,
utilitarianism, the true moral theory, describes which behavior is
most choiceworthy, and hence, morally, to be preferred. Since,
presumably, it would be behavior one would choose if one adopted
at least one of the competing prima facie moral points of view,
one has sufficient motive for engaging in it. Sidgwick's strategy
thus supports the Basic Internalistic Premise (i.e., the claim
that there are some moral obligations such that the fact that an
agent A ought to D entails A has sufficient motive for doing D) .
Sidgwick's defense of the Basic Internalistic Premise does not
follow the traditional strategy of using the claims that (a)
recognizing that something is good can be motivationally
efficacious, and (b) there are some good things such that we are
able to recognize them as good, however.
Because it defends the Basic Internalistic Premise,
Sidgwick's theory has the advantages claimed for internalistic
229
theories in chapter one. Sidgwick's utilitarianism is thus
deserving of serious consideration for precisely the sort of
reasons that such consideration is generally denied it: the
intuitions that motivate internalism. In addition, it defends
true moral judgments as objectively binding, and thus, may be the
best hope for a theory that allows one plausibly to maintain
internalistic intuitions.
There remains an importantly unsatisfying feature of
Sidgwick's theory, of course. Although it renders true moral
judgments objectively binding, it does not render to any competing
prima facie moral point of view the status of best. Neither does
it provide a principled reason to consider any prima facie moral
point of view privileged. From a commonsense perspective, these
are not small defects. For example, on Sidgwick's account, the
only thing wrong with being a thorough-going egoist (assuming one
is sufficiently enlightened so as to be a reflective and
successful one) is that doing so does not necessarily enable one
to conform in all cases to the best systematization of what we
find most choiceworthy . However, it is not commonly supposed
that this is the only thing wrong with being a thorough-going
egoist. Rather, it is usually thought that, however enlightened
and reflective one is, thinking only in terms of oneself is small-
minded and morally deficient in some way. A moral theory is
ordinarily expected to provide a criticism of this, and even an
230
antidote to it--neither of which Sidgwick's can provide. Part of
the appeal of traditional internalistic moral theories is that
they can characterize what it means to be a good person in a way
that meets these expectations. On such accounts, a good person
may be characterized as the most rational, reflective, respectful,
conscientious, or sympathetic. Similarly, such accounts can
characterize the egoist as deficient in these respects, and
criticize or exhort her accordingly.
As significant a defect as this may be, self-effacing
internalistic moral theories can be used to argue that some moral
judgments are objectively binding. Sidgwick's argument seems to
show that traditional internalistic moral theories are not
compatible with a principled defense of the claim that any moral
judgment is objectively binding. If traditional internalistic
moral theories are incompatible with such arguments, then they
cannot be used to demonstrate the superiority of moral points of
view over competing perspectives. In that case, the advantage
traditional internalistic moral theories have over self-effacing
ones will evaporate as well--leaving self-effacing internalistic
moral theories as the only hope for defending the Basic
Internalistic Premise and the veracity of the intuitions that
motivate it. To be plausible, then, a traditional internalistic
moral theory must have a response to the line of argument sketched
here.
231
Are traditional internalistic moral theories compatibla with
principlad objectivity?
The most obvious strategy for challenging the line of
argument sketched in the last section is to simply deny "the
dualism of practical reason,* and, more generally, to deny that
genuinely conflicting points of view can be equally reasonable or
moral. Even if we grant that agents (dispositionally) have
sufficient motive for acting in different ways, we could claim
that some dispositions are better than others. We could, then,
claim on these grounds that the choices the preferred points of
view would lead to are most choiceworthy, and, hence, that such
behavior is morally preferable. Arguing in this way would allow
us to claim without difficulty that some moral judgments are
objectively binding.
This move forces us back to the claim that traditional
internalistic moral theories are not flawed, but incomplete. As I
have already argued, internalistic theories that may reasonably
ask us to accept such a promissory note must provide an account of
what it means to adopt a moral point of view that is sufficiently
detailed to justify our assuming that legitimately moral points of
view do not conflict. I have also argued that, moreover, the
burden of proof rests on such theories to the extent that to be at
all plausible in advance of filling in the missing accounts, these
theories must at least suggest that the Basic Internalistic
232
Premise (i.e., the claim that there are some moral obligations
such that the fact that an agent A ought to D entails A has
sufficient motive for doing D) is true in light of the possible
relativistic consequences.
To adequately defuse the argument from Sidgwick's dualism of
practical reason will be difficult, however. Presumably, it will
involve a demonstration that self-interest and duty simply are not
equally subjectively reasonable or choiceworthy--a claim it is
hard to imagine opponents could conceivably become convinced of.
In addition, such theories would presumably lack any agent-
centered features that might be thought to enhance the
plausibility of a theory on the grounds discussed in chapter six.
There is another strategy that traditional internalistic
moral theories can use to defend some moral judgments as
objectively binding in a principled manner. A traditional
internalistic moral theory can undercut the Sidgwickian argument
by claiming that in cases in which legitimately moral points of
view support incompatible courses of action, no single course of
action is morally required--it could be claimed that in such
cases, no single point of view is morally best.
To illustrate this strategy, consider again an example from
chapter six: the case of deciding whether to obey one's
government's order to enlist in the military to fight in an unjust
war. Suppose that a particular theory considered to be
233
legitimately moral both a point of view that would lead one to
consider more important adherence to the generally just system of
laws governing one's country, and a point of view that would lead
one to consider more important one's duty to act to mitigate the
effects of an immoral war effort or to work to end such a war.
Such a theory could deny that either point of view is morally to
be preferred in this case. This would not simply amount to saying
that in cases of conflict, anything goes. If neither submitting
to conscription nor refusal is morally best, neither is
necessarily on a par with choosing an alternative, for instance,
suicide or terrorism. Actions one would not choose if one adopted
any moral point of view would still be inferior to actions
supported by some moral point of view (even though all competing
alternatives may seem equally bad when one adopts a particular
moral point of view). In addition, by claiming no single moral
point of view is morally best to adopt in such cases, we need not
be indifferent to the point of view an agent adopts in such cases.
There is a moral difference between conscientiously submitting to
conscription or conscientiously refusing, and doing either for
less noble reasons. Moral points of view would, on such accounts,
be privileged as such, even if they are not privileged among each
other.
Similarly, although traditional internalistic moral theories
have not tended to accept Sidgwick's claim that duty and interest
234
are equally subjectively reasonable or choiceworthy,11 this
strategy allows us to accept "the dualism of practical reason,"
and even to think that fulfilling one's duty and promoting one's
own interest have equal moral standing. In cases in which the two
perspectives conflict, neither behavior supported by a self-
interested point of view nor that supported by that of duty is
morally preferable. The dualism of practical reason thus does not
present an argument against the possibility of defending some
moral judgments as objectively binding in a principled way.
Traditional internalistic moral theories that follow this
strategy will, presumably, expect agents to be capable of
recognizing competing legitimately moral points of view. Since
agents must be able to recognize relevant good things as good in
cases in which they are bound by internalistic moral obligations
(claim (b)), they must be capable of distinguishing cases in which
they genuinely recognize some behavior as choiceworthy from those
in which they do not genuinely recognize behavior as morally
choiceworthy, in particular, from cases in which they consider
some behavior choiceworthy but only because they adopt a
particular but unprivileged moral point of view. According to
accounts that follow this strategy, to realize that there are
11At least in the way that Sidgwick means it--that they are equally
reasonable guides to action. For instance, Joseph Butler (whom Sidgwick
credits with articulating what Sidgwick later called "the dualism of
practical reason") claims that although duty and self-interest are
equally subjectively reasonable ends, we can see that it is not
reasonable for human beings to try directly to pursue self-interest in
their actions, and that we can see that acting conscientiously is.
235
legitimately moral points of view that compete with those
adopts oneself may be a necessary part of what it means to
adopting a truly moral point of view in the first place.
Works Cited
Aristotle. Nicomachean Ethics. Translated by Terence Irwin.
Indianapolis: Hackett Publishing Company, 1985.
Bentham, Jeremy. An Introduction to the Principles of Morals and
Legislation. In The Utilitarians, 5-398. Garden City, New
York: Doubleday & Company, Inc., 1961.
Butler, Joseph. Five Sermons Preached at the Rolls Chapel.
Indianapolis: Hackett Publishing Company, 1985.
Davis, Nancy. “Acting Utilitarians." Pacific Philosophical
Quarterly 66 (1985): 125-140.
Falk, W. D. “Fact, Value, and Nonnatural Predication." In Ought,
Reasons, and Morality: the Collected Papers of W. D. Falk,
99-122. Ithaca, New York: Cornell University Press, 1986.
"Hume on Is and Ought." In Ought, Reasons, and Morality:
the Collected Papers of W. D. Falk, 123-142. Ithaca, New York:
Cornell University Press, 1986.
"Morals without Faith." In Ought, Reasons, and Morality:
the Collected Papers of W. D. Falk, 163-179. Ithaca, New York:
Cornell University Press, 1986.
"On Learning about Reasons." In Ought, Reasons, and
Morality: the Collected Papers of W. D. Falk, 67-81. Ithaca,
New York: Cornell University Press, 1986.
"'Ought' and Motivation." In Ought, Reasons, and
Morality: the Collected Papers of W. D. Falk, 21-41. Ithaca,
New York: Cornell University Press, 1986.
Frankena, William K. "Obligation and Motivation in Recent Moral
Philosophy." In Essays in Moral Philosophy, edited by A. I.
Melden, 40-81. Seattle: University of Washington Press, 1958.
Hare, R. M. The Language of Morals. London: Oxford University
Press, 1952.
Harman, Gilbert. "Moral Relativism Defended." Philosophical
Review 8A (1975): 3-22.
Hobbes, Thomas. Leviathan. Edited by C. B. Macpherson. New
York: Penguin Books, 1986.
Hume, David. An Enquiry Concerning the Principles of Morals. In
Enquiries Concerning Human Understanding and Concerning the
Principles of Morals, 169-323. 1777. Reprint. Edited by L.
237
A. Selby-Bigge. Third edition, with text revised and notes by
P. H. Nidditch. New York: Oxford University Press, 1983.
A Treatise of Human Nature. Edited by L. A. Selby-Bigge.
1888. Second edition, with text revised and variant readings
by P. H. Nidditch. New York: Oxford University Press, 1985.
Kahneman, Daniel, Paul Slovic, and Amos Tversky. Judgment under
Uncertainty: Heuristics and Biases. New York: Cambridge
University Press, 1982.
Kripke, Saul A. Wittgenstein on Rules and Private Language.
Cambridge: Harvard University Press, 1982.
Lewis, David. "Dispositional Theories of Value." Proceedings of
the Aristotelian Society Supplement 63 (1989): 113-137.
Locke, John. Essay Concerning Human Understanding. 2 vols.
Annotated by Alexander Campbell Fraser. Oxford University
Press. Reprint. New York: Dover Publications, Inc., 1959.
Martineau, James. Types of Ethical Theories. 2 vols. Oxford:
Oxford University Press, 1885.
Moore, G. E. Principia Ethica. New York: Cambridge University
Press, 1960.
Nagel, Thomas. "The Fragmentation of Value." In Mortal
Questions, 128-146. New York: Cambridge University Press,
1991.
The Possibility of Altruism. Princeton, New Jersey:
Princeton University Press, 1978.
Nowell-Smith, P. H. Ethics. Baltimore: Penguin Books, Inc.,
1954.
Parfit, Derek. Reasons and Persons. New York: Oxford University
Press, 1984.
Pritchard, H. A. "Duty and Interest." In Readings in
Contemporary Ethical Theory, edited by Wilfrid Sellars and John
Hospers, 690-703. New York: Appleton-Century-Crofts, 1970.
Ross, W. D. Foundations of Ethics: the Gifford Lectures. New
York: Oxford University Press, 1960.
The Right and the Good. Indianapolis: Hackett Publishing
Company, 1988.
238
Sartre, Jean-Paul. “The Humanism of Existentialism." In Essays
in Existentialism, 31-62. Secaucus, New Jersey: Citadel Press,
1965.
Scheffler, Samuel. The Rejection of Consequentialism. New York:
Oxford University Press, 1987.
Sidgwick, Henry. "'Idiopsychological Ethics.'" Mind 12 (1887).
The Methods of Ethics. First edition. London: Macmillan
and Company, Ltd., 1874.
The Methods of Ethics. Seventh edition. London:
Macmillan and Company, Ltd., 1907. Reprint. Indianapolis:
Hackett Publishing Company, 1981.
"Some Fundamental Ethical Controversies." Mind 14 (1889):
473-487.
Smart, J. J. C. and Bernard Williams. Utilitarianism: For and
Against. New York: Cambridge University Press, 1973.
Stocker, Michael. Plural and Conflicting Values. New York:
Oxford University Press, 1990.
Williams, Bernard. "Ethical Consistency." In Problems of the
Self, 166-186. New York: Cambridge University Press, 1993.
"Persons, Character, and Morality." In Moral Luck, 1-19.
New York: Cambridge University Press, 1986.
239

Linked assets