A Coherentist Theory of Normative Authority

Author(s): Linda Radzik

Source: The Journal of Ethics, Vol. 6, No. 1 (2002), pp. 21-42
Published by: Springer
Stable URL: http://www.jstor.org/stable/25115713
Accessed: 11/12/2009 17:34

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=springer.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Ethics.

http://www.jstor.org
 LINDA RADZIK

A COHERENTIST THEORY OF NORMATIVE AUTHORITY*

(Received 6 October 2000; accepted in revised form 12October 2001)

ABSTRACT. What an "ought"

makes claim authoritative? What makes a particular norm

genuinely for an agent? This paper argues that normative authority can best
reason-giving
be accounted for in terms of the justification of norms. The main obstacle to such a theory,

however, is a regress problem.The worry is that every attempt to offer a justification for an

"ought" claim must appeal to another "ought" claim, ad infinitum. The paper argues that
vicious regress can be avoided in practical reasoning in the same way coherentists avoid
the problem in epistemology. Norms are justified by their coherence with other norms.

KEY WORDS: authority, coherentism, justification, normativity, norms

What is it that makes an "ought" claim authoritative? In the course of

daily decision-making, I encounter a multitude of norms (imperatives or
standards) that command me to act in one way or another. How do I tell
which of these are legitimately reason-giving, genuinely binding on my
behavior? What features does a norm have to have such that I really ought
to follow it?
The idea that we can be bound by "ought" claims is familiar. But, of
course, we do not believe every norm legitimately restricts us. Consider
the following:

A: One ought not be cruel.

B: Wives should be submissive to their husbands. (Call this the
Submissive Wife Norm.)

We tend to think the first norm makes a genuine claim on us. I think that I
really ought not be cruel, that I ought to accept A as a guide to my action. B
*
For their helpful comments in conversation and on earlier drafts of this material,
I thank David Copp, Christopher Deabler, Stephen Finlay, Keith Lehrer, Matthew
McGrath, Michael O'Rourke, Geoffrey Sayre-McCord, Schmidtz, David
David Silver,
Allen Thompson, an anonymous reviewer for this journal and the late Jean Hampton. I
am also grateful for support received from the Research Enhancement Program, Office of
the Vice President for Research and Associate Provost for Graduate Studies, at Texas A&M

University.

The Joumal ?fEthics 6: 21-42, 2002.

WCl
wV ? 2002 Kluwer Academic Publishers. Printed in theNetherlands.
 22 LINDA RADZIK

does not make a legitimate claim on me, however. I do not believe that there
is anything wrong, stupid or inappropriate about my being an equal partner
in a marriage (quite the contrary). My question, then, is what accounts for
this difference between A and Bl Why is it that I really ought to accept A
and not Bl What makes some norms authoritative?
One might think the difference is that A is a requirement of morality
while B is not. But this does not seem right, for two reasons. First of all,
the following norms seem to be legitimate and forceful, but they are not
moral:

C: One ought to eat green, leafy vegetables.

D: One ought to conduct drug efficacy studies in a double-blind environ
ment.

Secondly, we might well wonder whether this norm is one we really ought
to accept:

E: One ought to be moral.

So if the authoritative/nonauthoritative distinction is not the same as

the moral/nonmoral distinction, what is it?What
separates the norms that
we really ought to live by from the ones that we can discount as guides to
our own actions? My view, and the view shared by Christine Korsgaard,
David Copp and Jean Hampton in their recent books,1 is that normative
authority is a kind of justification. Authoritative "ought" claims have a
kind of justification that non-authoritative ones do not have. The difference
between A and B is that there is a good reason
to accept A as a guide to
my actions but not B. The challenge, then, is to say exactly what sort of
justification this is and what it takes for a norm to be justified in this sense.
The most tempting way to develop such a theory of the justification
of "ought" claims is to identify some particular norm (e.g., the Hypothet
ical Imperative, the Categorical Imperative, the realist norm "One ought
to preserve and promote what has objective value," the virtue theorist's
norm "One ought to become eudaimon") as an unequivocally authoritative
guide to action. Then one argues that this norm is the source of all genuine
justification. All other norms to which an agent might be subject are justi
fied only insofar as they follow from, or are at least permitted by, this
authoritative norm.

1 Christine
Korsgaard, The Sources of Normativity (New York: Cambridge University
Press, 1996); David Copp, Morality, Normativity and Society (New York: Oxford Univer

sity Press, 1995); and Jean Hampton, The Authority of Reason (New York: Cambridge
University Press, 1998). Hampton reserves the label "justification" for another concept,
but the idea is the same.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 23

This approach authority can be understood,

to normative I believe, as a
close analogue to foundationalist theories in epistemology. These theories
of epistemic justification argue that there are some beliefs (called found
ational or basic beliefs) that are justified in themselves. All other beliefs
gain their justification only insofar as they follow from the foundations. In
the case of theories of normative authority, some norm (or set of norms)
is presented as an incorrigible guide to action. It provides the foundation
upon which all other norms are based. Defenders of different theories of
normative authority then proceed to attack each other's alleged founda
tions as not really incorrigible after all. Kantians argue that the Humeans
are wrong to think that desires are basic sources of reasons for action.
Humeans charge that Kantians cannot
show that the Categorical Imperative
is really rationally required. Both of these camps attack the realists and the
virtue theorists as relying on a metaphysics that is too queer to be believed.
One alternative that is generally either overlooked altogether or
dismissed in a few lines is a coherentist theory of normative authority.2
Such a theory would be modeled on another part of the epistemology liter
ature. Coherentist theories of epistemic justification give up the search for
beliefs that are inherently justified and instead claim that a particular belief
is justified for an agent if and only if it coheres well with the other things
the agent believes. An analogous theory of normative authority would
claim that whether a particular norm is an authoritative guide for action
for an agent will depend on how well it fits with the other norms to which
the agent is committed.
My aim in this paper is to show that coherentist theories of the authority
of practical reason deserve to be treated as serious contenders in the current
debate. In order to do this, I will present what seem to be five criteria of
adequacy for any theory of normative authority. I will develop a version
of a coherentist theory, called Reflective Endorsement Coherentism, and
argue that it is particularly well-suited to satisfy these criteria. I will then
examine some of the consequences of the theory and argue that, contrary
to what some critics seem to think, coherentism does not commit us to
anything unacceptable. In order to effectively argue that a coherentist
theory is actually the best theory of normative authority on offer, I would
have to show why the foundationalist theories are flawed. Although I
have attempted such a critique elsewhere,3 in this essay I will aim at the
more modest goal of explaining what motivates a coherentist to
approach
normative authority.

2
Cf. Copp, Morality, Normativity and Society, pp. AX-A2.
3 Linda
Radzik, "Incorrigible Norms: Foundationalist Theories of Normative
Authority," The Southern Journal of Philosophy 38 (2000), pp. 633-649.
 24 LINDA RADZIK

Sincebe relying on analogies with epistemic

I will justification in order
to discussthe sort of justification that is related to normative authority, let
me here emphasize the disanalogies. Epistemic justification has to do with
the justification of beliefs, where the goal is to believe propositions only if
they are true. The discussion of normative authority has to do with practical
reason. The topic is the justification of norms, or "ought" claims, as guides
to action. Unlike beliefs and propositions, norms are not the sort of thing
that have truth values. They are standards for behavior, imperatives restated
in the indicative mood.4 They prescribe action rather than describe the
world, so the sort of justification involved must be distinctively practical.
Perhaps the difference between
epistemic and authoritative practical justi
fication can most easily be shown with an example where they come into
conflict. For instance, a cancer patient might be epistemically justified in
believing that she will die within six months but unjustified from a practical
point of view in doing so, since she is more likely to survive if she guides
her belief-forming behavior by the norm "One ought to be optimistic." So,
although thinking about epistemology will prove helpful in developing a

theory of the authority of practical reason, they are significantly different

subjects.

II

The general hypothesis that we are working with is that a norm is author
itative if and only if it is justified in some important sense. The task is to
figure out what sort of justification this could be. A first crucial step is to
determine whether the sort of justification that accounts for the authority of
"ought" claims is a "first person" or a "third person" kind of justification.
The distinction that I have in mind is most familiar from (but certainly
not limited to) debates about epistemic justification. There, the main
point of disagreement between the "internalists" and the "externalists" is
whether an agent's reasons for belief must be things to which the agent has
epistemic access.5 Internalists believe that agents must have first person
access to a reason for belief. For example, my belief that there is a red ball
in front of me can only be justified by something else that is actually in
my epistemic grasp, e.g., the belief "I see a red ball" or a percept with a
4
Copp, Morality, Normativity and Society, p. 19; David Copp, "Explanation and Justi
fication in Ethics," Ethics 100 (1990), pp. 237-258; Allan Gibbard, Wise Choices, Apt
Feelings (Cambridge: Harvard University Press, 1990), pp. 46, 70.
5
See, e.g., William P. Alston, "Concepts of Epistemic Justification," Epistemic Justi

fication: Essays in the Theory of Knowledge (Ithaca: Cornell University Press, 1989),
pp. 81-114.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 25

distinctive red quale. I call such a view a first person conception of justi
fying reasons. For an externalist about epistemic justification, however, the
agent need not have epistemic access to the reasons for her belief. What
justifies my belief that there is a red ball may be something I have no
access to at all. For example, according to the externalist, the fact that
my eyes are reliable detectors of nearby, well-lit, mid-sized objects may
justify my belief. I need not know or believe that my eyesight is reliable; it
is enough that it is reliable. I call this sort of account, where epistemic
access to justifying factors is not required, a third person account of
reasons.6

The shape our analysis of normative authority takes will be greatly

determined by the conception of justification on which we settle. Are
authoritative reasons for action first person or third person reasons? Is an

"ought" claim binding on us in virtue of something that is necessarily

within our epistemic grasp, or not? I will defend the position that the
authority of practical reason is a first person kind of justification. The agent
for whom the norm in question is binding must have epistemic access to
the justifying factors, at least in the sense that she would be able to see
why the norm is justified were she to make a sincere effort at doing so.
In The Sources of Normativity, Korsgaard argues that a theory of the
authority of practical reason must address the agent in her first person
point of view because the problem of authority comes up in its most
in the The - "Is
pressing form first person context.7 question morality
-
really authoritative?" is most naturally interpreted as a first person sort
of worry. It is the question an agent asks when morality is commanding
her to do something difficult, when she is being ordered to make a great
sacrifice. She wants to know "Do I really have to be moral?" If the question
about authority is primarily a first person question, a question that the agent
asks about herself, then the answer had better be accessible to that agent.
Whatever it is that makes morality binding on this agent had better be

6
Though thefirst person/third person distinction ismost explicitly drawn in the debates
about epistemic justification, the distinction appears in the case of other types of justific
ation as well. For instance, there are also debates about whether the moral of
justification
actions is first or third person. For an action to be morally justified, must the actor be able
to detect what it is that makes the action morally justified? Or need the action simply have
certain properties, regardless of whether the agent can detect or appreciate the import of
those properties? These questions underlie the more familiar debate in ethics: Is an action

morally justified because of the content of the agent's intentions or because the action has

objectively good consequences? Though the first person/third person distinction deals with

epistemic access to reasons, the reasons in question may themselves be epistemic reasons,
moral reasons, prudential reasons, or what have you.
7
Korsgaard, The Sources of Normativity, pp. 16-17.
 26 LINDARADZIK

something that she can grasp and appreciate the import of, otherwise her
question has not really been answered.
Another reason for thinking that the relevant notion of justification must
be first person accessible is if one wants the notion of normative authority
to be a regulative notion. By "regulative" I mean that the authority of
practical reason would be a concept that people could usefully think about
in order to decide what to do.8 If thinking about normative authority is
supposed to be of any use to me when I am trying to make choices, I had
better be able to tell what is authoritative and what is not.
Not everyone who writes about of practical reason agrees
the authority
that it should be a first person, regulative conception of justification.9
However, I believe that the idea that normative authority must be first
person accessible is widely, if tacitly, held. In fact, it seems to be what leads
the majority of theorists to believe that the source of normative authority
must be internal to the agent. If the justifying
somehow conditions for
norms to the agent, if they are based on facts completely
are external inde
pendent from him and his mental life, then there is a significant epistemic
gap between the agent and the conditions that oblige him to act in certain
ways. That leads to the objectionable result that he could have obligations
to act that he has no way of knowing about.10 So theorists limit them
selves to working with what is internal to the agent. They try to find the
source of normative authority within the agent's own deepest motivations
or, on Kantian accounts, the structure of his autonomous will. The rest of
this essay will proceed from the assumption that the authority of practical
reason is best analyzed by a first person sort of justification.

Ill

Another thing we can confidently say about the kind of justification that
accounts for normative authority is that it must be comprehensive.11 To
8
The regulative/nonregulative distinction comes from Alvin Goldman's discussion
of different conceptions of epistemic justification in Epistemology and Cognition (Cam
bridge: Harvard University Press, 1986), p. 25.
9
See Copp,
Morality, Normativity and Society, Chapter 3.
10 " "
I take to be a corollary
this of the 'ought' implies 'can' principle. But this too is
controversial. See Roy Sorensen, "Unknowable Obligations," Utilitas 1 (1995), pp. 247
271. As I am only claiming that a first person conception of justification is compelUng,
I will not try to resolve this debate here. Even Sorensen acknowledges that most people
beUeve unknowable obligations are objectionable (p. 248).
11 David
Copp, "Gyges's Choice: Overridingness and the Unity of Reason," Social
Philosophy & Policy 14 (1997), pp. 86-101. See also Linda Radzik, "Justification and
theAuthority of Norms," Journal of Value Inquiry 34 (2000), pp. 451^461.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 27

explain what I mean by this, let me first point out that there are many
kinds of justification. There is epistemic justification, prudential justifica
tion, moral justification. We distinguish among these different notions by
reference to characteristic interests that define a particular evaluative point
of view. Epistemic justification lays out what one would be warranted in

believing, assuming one wants to believe what is true. Prudential justifica

tion says what to do if one wants to optimize one's own self-interest. Moral
justification tells one what to do in order to promote either human welfare
or happiness or respect for agents as ends in themselves (this is a matter of
great debate in ethical theory). I will call these "interest-driven" notions of
justification. There is a long list of these types of justification. There is also
logical justification, political justification, and justification from the point
of view of etiquette. In fact, it seems one may take any possible interest
and find an evaluative point of view connected with it. Take the goal of
being a mafioso. We can talk about mafia justification and sort out different
actions in terms of whether they are mafia justified or not - whether they
fit the standards for a successful mafioso or not.
If there really are amultitude of conceptions of justification, then saying
that normative authority is a matter of justification does not get us very far.
Even the Submissive Wife Norm, mentioned in the first section of this
essay, is justified in some sense. It is justified from the point of view
of certain fundamentalist religious ideologies. It is offensive to people
occupying these evaluative points of view when a wife's career determines
where the family moves, or when she chooses to keep her surname as a
symbol of her individuality. But that does not mean that the Submissive
Wife Norm is authoritative. Why should an agent care if she is unjustified

according to that standard?

There are many different interest-defined notions of justification, but,
rather than any of them capturing the idea of normative authority, it seems
that we need the concept of normative authority in order to decide among
them. to depend on my judgments
I need about what is authoritative in
order that I really have reason to do what is morally
to decide justified
but not what is justified by a male supremacist religious ideology. I need
a notion of authority in order to decide which of these interests ground
justification worth having and which do not.
If normative authority really is a matter
of justification, it must be a
very different kind of justification than the interest-driven ones. It must
be justification from some more comprehensive - a
point of view point
of view from which we can look over all the interest-defined evaluative
schemes and judge which ones should be allowed influence over our
 28 LINDA RADZIK

choices. authority needs

Normative to be characterized in terms of an

all-things-considered sort of justification. And so that we avoid begging

the question about which sorts of interests really ought to influence us,
this thorough-going justification must also be interest-neutral. It cannot be
defined in terms of some particular concern or goal. Rather, as David Copp
writes, a comprehensive conception of justification

would take the verdicts by all the special

given [evaluative] standpoints [the moral, the

epistemic, etc.] regarding any situation where an agent needs to choose; it would evaluate
these verdicts without any question-begging; and it would produce an overall verdict as to
what the agent is to do.12

Following Copp, I call this comprehensive evaluative point of view "justi

fication simpliciter."13 This is the sort of justification that seems to account
for normative authority. The only "ought" claims that an agent really ought
to accept are ones that are justified simpliciter.

IV

A third requirement that we can put on a theory of the authoritative justi

fication of norms is that it be transparent.14 The transparency requirement
says that the criteria for justification must not be such that the justified
norms lose all their force for an agent once the agent comes to understand
why the norms are justified. An example of a non-transparent theory of
authority would be one that said that all the norms that we usually take to
be authoritative are simply the ones that maintain traditional power struc
tures. If we really believed this to be the source of authority then we would
probably no longer find these norms to be authoritative. Transparency is
important because we want a theory of justification simpliciter in order to
account for and vindicate our intuition that some norms are authoritative
while others are not. Thistask cannot be completed if, in coming to know
this theory, we lose that very intuition. A non-transparent theory of justi
fication simpliciter is practically indistinguishable from skepticism about
the authority of "ought" claims.
A natural way of restating the transparency requirement seems to be in
terms of reflective endorsement.15 A norm cannot really be justified for an
12
Copp, "Gyges's Choice: Overridingness and the Unity of Reason," p. 94.
13 and of Reason,"
Copp, "Gyges's Choice: Overridingness the Unity p. 94.
14 the Limits
Korsgaard, The Sources of Normativity, p. 17; Bernard Williams, Ethics and

of Philosophy (Cambridge: Harvard University Press, 1985), pp. 101-102.

15 Lectures 3 and 4.
Cf. Korsgaard, The Sources of Normativity,
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 29

agent if, when she reflects on what it is that makes the norm justified, she
can no longer endorse it. The requirement of reflective endorsement plays
a key role in several theories of normative authority, including those of
both David Hume and Immanuel Kant, and more recently in the works of
Bernard Williams and Korsgaard.16 It is not much of an oversimplification
to say that these theorists think that a norm is binding on an agent only
if it she can endorse it upon reflection. They simply differ over what the
relevant kind of reflection entails.

The fourth requirement for a justification-based theory of authority is that

itmust avoid infinite regress.17 The regress problem in providing a justi
fication simpliciter for norms is parallel to the familiar regress problem
that plagues the epistemic justification of belief. There, the worry is that
an agent can only justify one belief with something else he believes, and
so is in danger of winding up in an infinite regress of reasons for belief.
As in the epistemic case, the threat of regress for a theory of normative
authority follows from the commitment to a first person conception of
justification. If justification simpliciter must be first person accessible, this
means that the agent must be in a position where he could, if he sincerely
tried, offer justifying in support of the norm that binds him. But
reasons
it seemsthat the only thing that he can offer as justifying reasons for one
norm are other norms that he accepts. We ask, "Why accept norm NT and
he responds, "Because N has property P and one should accept norms that
are P." But then he must say what justifies that norm - "Why should you
accept norms that are P?" The agent is in danger of falling into an endless
chain of norms and "why" questions. But an infinite justification seems to
be no justification at all.
If the problems in epistemology and practical reason are similar,
perhaps the solutions are similar as well. Historically, the most popular
strategy for responding to the regress problem has been foundationalism.
As Imentioned in Section 1, the main idea behind foundationalist models
of epistemic justification is that the chain of justifying beliefs will come to
16
Williams, Ethics and the Limits of Philosophy. Korsgaard argues that all of these
theories are versions of reflective endorsement in Korsgaard, The Sources of Normativity,
Lectures 3 and 4.
17
Copp, Morality, Normativity and Society, pp. 31-49; Korsgaard, The Sources of
Normativity, p. 30; Linda Radzik, "A Normative Regress Problem," American Philosoph
ical Quarterly 36 (1999), pp. 35^7.
 30 LINDA RADZIK

an end if there are beliefs

that are justified in themselves. Theorists who
want to provide a justification of practical reasons that can avoid infinite
regress try something similar. They look for norms that are, somehow,
insusceptible to the question "Why should I accept this as a guide to my
action?" I think there is good reason to doubt that there are any such incor
rigible norms, but I will not go into that issue here.18 Instead, I would
simply like to present another possible solution to the regress problem.
In epistemology, foundationalism lost its position of dominance years
ago. One of the main competitors these days is coherentism.19 The idea
here is that what justifies a particular belief for an agent is the way in
which the belief fits with the rest of the things that the agent believes. A
particular belief is justified if it coheres with rest of the agent's belief-set.
But the belief-set is not counted as foundational, as beyond doubt itself.
The agent can question each of these supporting beliefs as well, but she
can only do so from the point of view of other beliefs that she holds. In this
way, the believer is (to use Otto Neurath's a sailor repairing
image20) like
his ship while at sea. Any plank in his ship is subject to evaluation and
replacement, but in order to examine one of the planks, the sailor must
be standing on some other plank. Similarly, the only way an agent can
question and evaluate the soundness of her beliefs is by depending on other
beliefs of hers. What else could she use?
This coherentist model can be adapted to the project of the justifica
tion simpliciter of norms as well. The basic idea is the same. The sort of

justification we are looking for is a first person kind of justification. This

means that the relevant kind of evaluation must be something that the agent
is capable of performing himself. In determining whether he should accept
some particular norm N as a guide to his behavior, he has to rely on the
other norms that he accepts. How could he refrain from using them and
yet still come to a decision? N will be justified if and only if it coheres
well with the norms
he accepts. But the norms that do the justifying in one
instance can themselves be subjected to evaluation in the next. The regress
is avoided in that justifying reasons are provided for each norm, though the
set of norms is finite and no norm is taken as justified in itself. This model
of justification will be more fully developed in a moment.

18
But cf. Radzik, "Incorrigible Norms: Foundationalist Theories of Normative

Authority."
19
Laurence BonJour, The Structure of Empirical Knowledge (Cambridge: Harvard

University Press, 1985); Keith Lehrer, Theory of Knowledge (Boulder: Westview Press,
1990).
20
Otto Neurath, "Protokollsaetze," Erkenntnis 3 (1932), pp. 204-214.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 31

VI

A fifth criterion of adequacy on a theory of justification simpliciter is

reflexivity. The reflexivity requirement is a consequence of the compre
hensiveness and transparency conditions. The standard of justification
simpliciter can be viewed that tells us which norms we ought
as a norm
to accept. Since are subject to a demand
all norms for justification, we
can ask whether we have reason simpliciter to accept this norm - whether
we have reason to be justified simpliciter in our normative acceptances.
Comprehensiveness says that an adequate standard must have the resources
to answer this question. The transparency condition adds the requirement
that this question be answered in the affirmative. After all, if our theory
implied that we do not have reason simpliciter to accept the norms that we
are justified simpliciter in accepting, then we would no longer be able to
feel the force of norms that are justified simpliciter. We would say, "Norm
N is justified simpliciter, but so what?" The normative force of N would
"evaporate in our hands," as Korsgaard would say.21 In order to prevent
this loss of transparency, it must be the case that the alleged standard
of justification passes its own test. Any adequate account of normative
authority must be capable of accounting for its own justification. It must
be reflexive.
Many people tend to reject the significance of this sort of self-support.
In his book Morality, Normativity and Society, Copp dismisses reflexive
standards of justification with the following counterexample:
F: A standard is to be subscribed to if it can be expressed in English in
fewer than twenty words.22

This nineteen-word-long norm is self-supporting but no one could seri

ously think it to be
justification-conferring. But reflexivity is only a
necessary condition on a theory of normativity, not a sufficient one. A
theory also has to meet the other four criteria (at least) in order to be
acceptable, and this includes the transparency requirement. Copp's norm
F is certainly not transparent.

VII

A coherentist theory is particularly well-suited to meet the requirement of

reflexivity. To illustrate how, I will first need to fill out a particular version
21
Korsgaard, The Sources of Normativity, p. 30.
22
Copp, Morality, Normativity and Society, p. 4.
 32 LINDA RADZIK

of normative coherentism. The variation that I advocate blends a Reflective

Endorsement approach with a Lehrerian model of coherentism.23 Let's
start with the basic idea of Reflective Endorsement:

Reflective Endorsement: A norm is justified for an agent if the agent would endorse the
norm upon reflection.

If a norm N would be endorsed

upon reflection, that means that there is
something norms
in the set of beliefs
and that the agent accepts that lends
some sort of support to N. There are two important concepts here that we
must define: "accepting/endorsing a norm" and "supporting a norm."
To accept, or endorse, a norm is to hold it as a guide to action. Endorse
ment is a kind of psychological state, which is tied tomotivations to act and
dispositions to avow support for the norm. If a person endorses the norm
"One ought to eat green, leafy vegetables," then he will think of himself
and others as having reason to eat such vegetables. He will advise others
to eat them and do so himself, unless he suffers from weakness of will, or
he finds that another norm overrides the green-vegetable-eating norm in a
particular situation, or he is setting out to do something that he ought not
do. As Allan Gibbard says, "Accepting a norm is whatever psychic state
... rise to this of avowal of the norm and governance
gives syndrome by
it."24

The second concept that is important to the idea of Reflective Endorse

ment is that different accepted norms can support, or fail to support,
one another. In general, one norm N is supported by another norm M
if following N satisfies the standard M sets forth. So the norm "I ought
to eat green, leafy vegetables" is supported by the norm "I ought to do
what serves my self-interest" if eating green, leafy vegetables serves my
self-interest. If N gets such support within an particular agent's accept
ance set, let us say that N is defeasibly justified simpliciter for that agent.
The positive support could come from a sophisticated chain of practical
reasoning or from an acceptance as simple as "If it feels right, do it" or
"If everyone to accept N, do so as well." So, a norm that is
else seems
"reflectively endorsed" is one that is both accepted itself and supported
by something else that the agent accepts. A reflectively endorsed norm is
defeasibly justified.
23
Lehrer,Theory of Knowledge. The notions of "acceptance," "competition," "defeat,"
etc., that I will define below in order to flesh out my theory are all practical analogues of

epistemic notions that Lehrer works out in this text. Keith Lehrer, Self-Trust (New York:
Oxford University Press, 1997), Chapters 1-4 apply his model of coherence to practical
reasoning, but in a significantly different way than I do here. See footnote 28 below.
24
Gibbard, Wise Choices, Apt Feelings, p. 75.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 33

The reason
that reflective endorsement seems to bring only defeasible
justification is that it seems that an agent could reflectively endorse two
norms that compete with one another. Norms N\ and N2 are in competition
if one prohibits the other (e.g., if N\ says "Don't accept norm N2") or if
they direct the agent to perform incompatible actions (e.g., ifN\ says "Do
y" and N2 says "Don't do y"). Some subset of the agent's accepted norms
might lead her to endorse N\ while another subset tells her to endorse N2. If
one of these norms is to have, not just defeasible, but decisive justification,
itmust also be ultimately undefeated by any other defeasibly justified norm
in the set of acceptances. N2 will defeat N\ when something else in the
set of defeasibly justified norms, call it N$9 directs the agent to weigh N2
more heavily than N\. Of course, defeaters may themselves be defeated
by something else in the acceptance set. This is why I say that in order
for a norm to be fully justified it must be both positively supported and
ultimately undefeated.25 Given that we are interested in which norms an
agent is justified in accepting in a first person sense, the only norms that
are potential defeaters are ones to which the agent has epistemic access.
Potential defeaters must also have at least defeasible justification them
selves if they are to be able to outweigh the justification of any other norm.
So the pool
of potential defeaters must come from the set of norms the
agent accepts, plus the ones that she has reason to accept on the basis
of norms she actually accepts. A norm is decisively justified for an agent
if she would it upon reflection
endorse and that endorsement would be
ultimately undefeated
by the rest of her acceptance set. Call this theory of
justification "Reflective Endorsement Coherentism" (REC).
So, according to REC, we are bound by justified norms by virtue of our
own acceptances. If an agent acts in violation of a norm that would cohere
with his own acceptances upon reflection, then he is doing something of
which he himself disapproves (or of which he would disapprove were he
to reflect on it). Breaking one's obligations amounts to betraying what one
cares about. The price is self-alienation and self-hatred. Korsgaard argues
that the norms to which an agent is committed form his sense of identity. If
this is right, and it seems to be, then violating authoritative norms amounts
to ceasing to be oneself anymore. It is a kind of self-destruction.26

25
For the moment, I would like to leave open the possibility of genuine normative
dilemmas, cases in which an agent holds two conflicting norms with equal of
degrees
defeasible justification and in which neither norm can be ultimately defeated by some
thing else in the acceptance set. For an introduction to the debate about whether there are
genuine dilemmas, see Christopher W. Gowans (ed.), Moral Dilemmas (New York: Oxford
University Press, 1987).
26
This is the way Korsgaard spells out what itmeans to be bound by authoritative norms
in her own, Kantian theory of normative authority (Korsgaard, The Sources of Normativity,
 34 LINDA RADZIK

this point, one might

At object that REC cannot stop the vicious regress
because it is itself a norm that stands in need of justification. Specifically,
it is this norm:

RE\: One ought to accept norms that one would endorse upon reflection.

But why should the agent trust her own practices of reflective endorse
ments? Perhaps they lead her astray. This objection is the same as the
one that Korsgaard levels against Reflective Endorsement theories in The
Sources of Normativity.21
coherentist move
The is to point out that the Reflective Endorsement
theorist has an answer to this call for justification. An agent is justified in
accepting RE\ if he has a good reason to reflectively endorse norms in just
-
the way that he does i.e., if his reflection is conducted in accordance with
norms that are themselves reflectively endorsed and ultimately undefeated.
Reflection is a norm-guided activity. We each accept that some methods of
reflection are better than others. For example, we usually value the results
of reflection performed when calm and sober than when agitated or drunk.
In other words, we adopt norms of reflection for ourselves. The agent can
justify RE\ if he can claim,

RE2. I ought to accept the norms of reflection that I do.

RE 1 is a norm of reflection that the agent accepts. It is supported by RE2.

Of course, appeal to the norm RE2 raises another call for justification.
Why should I accept /??2? It seems the agent needs to be able to justify
RE2 by appeal to some other norm that he accepts. He can do that. The
reason for accepting RE2 is RE\. He ought to accept the norms of reflection
that he does because when he reflects on those norms, he endorses them,
and he ought to accept norms that he endorses upon reflection. So RE\
supports RE2 while RE2 supports RE\. Following Keith Lehrer's lead, I
will refer to this as the "loop of reflective endorsement."28 In this picture,

Lecture 3). It should be emphasized that there is a difference between violating one's
commitments (self-alienation) and simply changing one's commitments (becoming a new

self). For a discussion of how REC distinguishes between justified and unjustified changes
in one's commitments, see footnote 35 below.
27
Korsgaard, The Sources of Normativity, pp. 88-89.
28 uses a loop similar to this one to tie up his
In Self-Trust, Lehrer theory of practical
wisdom (Chapter 2). Lehrer's loop involves the notion of being trustworthy in judging
oneself trustworthy. My theory has an advantage over Lehrer's in that reflective endorse
ment is a more determinate concept than trustworthiness. Lehrer resists simple reductions
of trustworthiness to non-normative, naturalistic notions such as reliability. For him,
trustworthiness is a primitive concept (cf. Chapter 3).
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 35

reflective endorsement is like a tribunal that passes judgment on its own

procedures.29
Here is another way of highlighting the same idea. When Korsgaard
critiques the Humean view that basic desires are foundational sources of
normative authority, she points out that "When desire bids, we can indeed
take it or leave it. And that is the source of the problem."30 Desires are not

incorrigible, that is why we have to struggle with the question of normative

authority. We must ask ourselves which desires we really have good reason
to act on. A parallel complaint against grounding the justification of norms
in the agent's own acceptances would be, "When the norms we accept
command us to perform some action, we can take it or leave it." This seems
right. But, notice, to say that none of the norms we reflectively endorse
is unquestionable does not mean that we can step completely outside of
our set of accepted norms to decide what to do and what to accept. Any
normative acceptance can be questioned and potentially rejected, but the
coherentist will insist that any particular normative acceptance can only be

questioned and rejected on the grounds of something else that the agent
accepts. What else could the agent use?
According to REC, a norm is authoritative for an agent if and only if
it, and everything that is counted as a reason for it, is supported by norms
that one would endorse upon reflection, where the process of reflection
would itself be reflectively endorsed and ultimately undefeated. Notice,
this entails that any agent who would not accept the loop of reflective
endorsement (RE\ and RE2) upon reflection is not an agent for whom
norms are justified or unjustified. Does the loop requirement amount to a
loophole? Can an agent be released from all obligations by simply rejecting
RE\ and RE2 upon reflection? In theory, this is possible. But it seems to
me that no agent could really do so. To reject the loop of endorsement is
to disavow one's own powers of judgment. It is to deny the legitimacy of
trusting oneself. Perhaps it is possible not to trust one's own reflection,
but it is certainly not an attractive option. In fact, it seems it would be
tantamount to ceasing to be an agent anymore.
line of thought brings me into a measure
This of agreement with both
Lehrer and Korsgaard. Lehrer argues that agents must judge themselves
to be trustworthy. That is why he makes trustworthiness the key concept
in his theory of practical wisdom.31 In Reflective Endorsement Coher
entism, taking oneself to be trustworthy amounts to accepting that one's

29
The image is due to Onora O'Neill, Constructions of Reason (New York: Cambridge
University Press, 1989), Chapter 1.
30
Korsgaard, The Sources of Normativity, p. 97.
31
Lehrer, Self Trust, Chapter 2.
 36 LINDA RADZIK

own can justify the acceptance

reflection of norms. In her Kantian theory,
Korsgaard also draws a tight connection between normative authority and
agency. It is because we are agents, creatures who can reflect on and decide
what to do, that we must concern ourselves with whether our reasons for
action are justified or not.32

VIII

How well does

the coherentist version of Reflective Endorsement meet
the criteria of adequacy that we have placed on a theory of normative
authority? These were first person accessibility, comprehensiveness, the
ability to avoid infinite regress, transparency and reflexivity. I will examine
each of these in turn, starting with reflexivity.
The reflexivity requirement demands that a standard of justification
simpliciter be able to account for its own justification. This is precisely
what the loop of reflective endorsement allows. To put this in an even
simpler form, notice that what I have presented as a mutually supporting
pair of norms, RE\ and RE2, can also be stated as a single self-supporting
norm. If we ask why one should accept RE\ the answer is, "When I reflect
on RE\,1 endorse it and (RE\) one ought to accept norms that one endorses
upon reflection."
As I mentioned
in the last section, reflexivity is only a necessary
condition on a theory of normativity, not a sufficient one. Copp's alleged
counterexample, norm F, is reflexive but clearly an unacceptable candidate
for the standard of justification simpliciter. There is an important differ
ence between the reflexive Reflective Endorsement standard and Copp's
counterexample, a contrast
that highlights the difference between coher
entist and foundationalist theories. Copp's standard is automatically or

inherently self-supporting, Reflective Endorsement is not. To say that RE\,

or any other norm, is justified because it is endorsed upon reflection is to
acknowledge that other norms play a role in the justification of that norm.
Reflection is a norm-governed process. Norms from the set of acceptances
are what reflection appeals to in determining whether to endorse RE\. In
reflecting, an agent may come to believe that certain norms she uses for
32
Korsgaard, The Sources of Normativity, pp. 92-94. The difference between REC
and Korsgaard's Kantian theory, however, is that she moves from the conclusion that I
must take my own reflection to be justification-conferring to the claim that I must take

anyone's reflection to be justification-conferring (cf. Lecture 4). I argue that Korsgaard

cannot defend this claim without falling into the regress problem in "Incorrigible Norms,"
Section 3.iv. The argument there is similar to the argument I will give in section VIE of
this paper, when I consider attempts to "objectivize" REC.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 37

reflection are prejudiced, insufficiently thorough, childish, or inconsistent;

and, given that she accepts norms that say that one ought not be prejudiced,
careless, childish, or inconsistent, this will prevent her from endorsing
those particular norms of reflection. So, REi's self-support is not at all
automatic or necessary. However, as I said in the last section, an agent for
whom RE\ is not self-supporting would hardly count as an agent. An agent,
if she is to be an agent, must be able to trust her own practices of reflective
endorsement in general, even if she recognizes that she is untrustworthy
in certain contexts.
Still, the fact that REi's self-support is not automatic
makes it a much
more substantive standard of justification than Copp's
standard F. It keeps RE\ from being trivial or viciously circular.
REC clearly satisfies the requirement that the reasons for accepting
norms be first person accessible. Its starting point is the agent's own set of
accepted norms. Justification is constructed out of this set by a process of
reflective endorsement. In defining the sort of epistemic access to reasons
that is required, we said that the reasons must be accessible to someone
who makes a sincere effort at detecting them. This is the sort of epistemic
state picked out by the term "reflection." We believe propositions form
desires and internalize normsautomatically
quite at times. In reflection,
we think about these different
acceptances and desires with a degree of
detachment and ask, "What reason do I have for this?" According to REC,
the answer to that question is found in the agent's own set of acceptances.
The comprehensiveness of the theory is also fairly straightforward. A
standard of justification simpliciter is comprehensive if and only if it can,
without begging the question, account for the justification of all types
of norms - moral norms, norms of prudence, norms that enjoin us to
adopt epistemically justified belief-forming habits, etc. Because reflective
endorsement is a very general kind of attitude, the REC standard can be
- be itmoral or
taken to a wide range of subjects. It seems any kind of norm
-
epistemic, prudential or logical can be endorsed (or potentially rejected)
upon reflection. Question-begging is avoided because the REC standard
can also be applied to itself and it is not guaranteed that itwill pass its own
test.

Earlier out how a coherentist

I pointed theory like REC avoids regress.
The theory provides a way to answer every request for justification in terms
of the other norms that agent reflectively endorses, though the set of norms
is finite and no norm, not even RE\, is treated as inherently justified.
The final criterion of adequacy is transparency. A theory of normative
authority is not transparent if, were the agent to believe that this is really
what gives her reason to accept norms, she could no longer feel the force
of reasons or see any norms as justified. In a way, Reflective Endorsement
 38 LINDA RADZIK

Coherentism has a built-in safeguard against failures of transparency. To

say that a theory of normative authority is not transparent is to say that its
standard of justification would not survive one's reflection. But Reflective
Endorsement Coherentism's standard
of justification just is, in part, the
requirement that the standard survive
reflection. By requiring the loop of
reflective endorsement, the theory requires that the agent find the results of
her own reflection compelling.
Still, are ways
there in which one can charge REC with a failure
of transparency. In fact, transparency differs from the other criteria of
adequacy discussed in this essay in that its satisfaction is not demonstrable.
I cannot prove that any agent who really understands the REC theory and
believes it to be true would still be able to feel the binding force of justified
norms. I can only respond to variouscharges that agents would find REC
non-transparent for some particular reason by clarifying the theory in a
way that undermines the worry. In what follows, I will look at some likely
objections.
An objector might start with the observation that we have general but
firm intuitions that certain sorts of people are reasonable. If REC concludes
that these people fail to have reason simpliciter to follow norms that we

confidently believe to be authoritative, then REC is not a good theory

of normative authority. The most tempting of such objections focuses on
moral norms. Many theorists will want to say that any theory of normative
authority that is not compatible with a plausible theory of morality will not
be satisfactory. They will charge that any theory of normative authority
that cannot preserve the intuition that we have reason to be moral fails the
transparency condition.
It seems REC cannot avoid the Knave Problem, that is, it cannot deny
that there might be some knavish person for whom there is no authorit
ative reason to be moral. It also cannot
the possibility
deny that there are
some people for whom immoral norms (e.g., the Submissive Wife Norm
mentioned at the beginning of the paper) are authoritative. REC, as it
stands, is a subjectivist theory. Which norms are justified for a particular

agent will depend on what is in his set of normative acceptances. There

might be some knave for whom the norms of morality do not survive
reflection, or for whom immoral norms do survive reflection.
It is worth
highlighting what motivates the subjective character of this
theory of normative authority. One might try to develop an alternative,

objectivist version of Reflective Endorsement Coherentism that places

more limits on the form that reflection may take or on the content of
the acceptance system. For instance, one might turn REC into a kind
of ideal spectator theory, requiring that only reflection conducted when
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 39

calm and fully informed is capable of generating justification. This sort of

move would likely help with the case of the Submissive Wife Norm, for
instance, since it seems that such a norm would only come to be reflect
ively endorsed and ultimately undefeated in the acceptance set of a person
with false beliefs about the nature of women, the will of God, or other
such issues. Another objectivist theory might stipulate that only acceptance
systems that include the Categorical Imperative can ground justification. A
third might say that acceptance systems must include a norm that tells the
agent to be sensitive to what is objectively valuable.
Theproblem with placing such restrictions on the shape reflective
endorsement or an acceptance system must take is that it is not clear how
these requirements are to be justified. Why ought I accept the results of
calm and informed reflection only? Why must I accept the Categorical
Imperative? Why should my acceptances track facts of value? Of course,
these norms might be accepted and justified through reflective endorse
ment. But to propose them as separate requirements is to suggest that
they are relevant to normative authority whether are accepted under
they
reflection or not. If these norms cohere with the agent's acceptance set,
then stating them as separate requirements is unnecessary. But if they do
not cohere, then they are arbitrary from the point of view of the
agent.33
We fall back into the regress.
The regress problem is what pushes us toward a subjectivist theory of
justification simpliciter. It is what prevents us from getting the (admit
tedly desirable) result that all agents have an authoritative reason to be
moral. However, I do not think that the Knave Problem should be taken
as sufficient reason to reject Reflective Endorsement Coherentism as non
transparent. As I will argue, REC actually does a good job, on the whole,
of accommodating our intuitions about morality.
For instance, it seems REC can conclude that people generally have
reason simpliciter to be moral. One way it will do this is by taking
advantage of the fact that being moral is generally in one's self-interest.
Norms that tell one to serve one's
self-interestown (at least where
"interest" is given a subjectivist reading) are ones that we would expect to
be reflectively endorsed by most people. The things that people most care
about form their interests. The things that they most care about just are,
presumably, the things that they will endorse upon reflection. If morality
is a system of norms that usually is in line with what most agents count
as their interests (and the social needs of individuals combined with the

33
This response to attempts to objectivize REC follows the pattern of Lehrer's response
to attempts to objectivize coherentist theories of first person (Lehrer,
epistemic justification
Theory of Knowledge, p. 101).
 40 LINDA RADZIK

social enforcement of moral norms makes this likely), then people usually
have reason to be moral.34
Another of the stable
intuitions that we have about morality is that
sometimes people do not think that they have a reason to be moral when
they really do. REC can account for the possibility of such errors. In
general, the reflective endorsement theory says that x (an act, belief,
acceptance of a norm) is wrong for a person if and only if x violates some
norm which is justified for her. For example, Mary might accept the norm
"I may take office supplies home from work with me" even though she
also accepts, with decisive justification, the moral norm "I ought not steal."
The pilfering norm is accepted in violation of the moral norm. Mary ought
not take office supplies, even though she does not believe that this is the
case. Mary probably fell into this error because
she failed to recognize
that taking office supplies is stealing, perhaps because it is easy to think
of businesses as impersonal. Mary does not see that taking office supplies
amounts to taking the possessions of her employer or taking money out of
the pockets of the stockholders of the corporation, even though everything
she needs to make this connection is within her epistemic grasp. REC can
recognize these errors as errors.
One might that errors are too easy to fix under REC. If the source
charge
of my obligation by moral norms ismy own reflection, then what is to stop
me, when being moral becomes hard, from simply releasing myself from
these obligations? If obligations can be dissolved so easily, then are they
really binding or authoritative after all? Doesn't it seem that any obligation
that I can simply release myself from at will is really no obligation at all?
It is, of course, possible under REC that an agent who comes to see that
being moral is costly in a particular circumstance will enter this inform
ation into his reflection and stop endorsing the moral norm in question
(or at least count the reason for following the norm as overridden in this
circumstance). But REC can account for the intuition that a person usually
cannot simply release himself from an obligation. There are times in which
morality is really binding.

34
Perhaps REC can also
take advantage of the fact that most people most of the time
will accept that they should not
allow their normative acceptances to be influenced by false
beliefs about natural facts or false beliefs about objective values (if there are any). This
might suggest that most people will not be decisively justified in accepting an immoral
norm like the Submissive Wife Norm, which will only seem plausible to people with false
beliefs about women. The idea is that the Submissive Wife Norm will be defeated by the
agent's commitment to true belief. This seems to assume, however, that the falsity of the
relevant beliefs is within the epistemic grasp of the agent in question. Whether that is true
is too big a question to pursue here.
 A COHERENTISTTHEORYOF NORMATIVEAUTHORITY 41

REC can accommodate this judgment because it can assign error to

the ways in which
agents change their acceptance systems. Whether I can
justifiably alter my reflective judgment of a norm from endorsement to
rejection will depend on which norms of reflection I endorse. Most agents
will endorse norms that will restrict their ability to justifiably shrug off
obligations. For example, they will accept that general principles of action
are only worth having when one abides by them over the long term, that

changing one's principles in circumstances that are stressful and unnerving

will lead to regret later (and that regret should be avoided), etc. Of course,
it is possible that there will be agents who have no such acceptances and
no reason to refrain from capricious changes in their acceptance sets. But
most people will not be like this. For most people, obligations, moral and
otherwise, cannot be easily shed.35
Another thing that I would like to point out is that Reflective Endorse
ment Coherentism is consistent with a realist theory of moral truth and
with practices of punishment. Though the REC theorist cannot say that the
knave has reason simpliciter to avoid unkindness and injustice, there is no
obstacle to her truthfully labeling the knave immoral when he is unkind
and unjust. Furthermore, all of us who have reason to be committed to
morality have reason to interfere with and even to punish the knave in
hopes of influencing his future behavior.
Really, the only thing that REC will not allow us to do is to say truth
fully, "The knave has reason simpliciter to be moral." Given everything
that we can say and do with regard to knaves, this does not strike me as
too great a loss. REC is compatible with a morality worth having. The
Knave Problem is not a sufficient reason to reject REC, especially if, as I
believe but am not able to argue for here, the only alternative seems to be
the normative skepticism that the regress problem threatens to bring.

IX

In this essay I have argued that Reflective Endorsement Coherentism is

a good candidate for a theory of normative authority by showing that it
35 are norms
The idea that there of reflection will allow us to distinguish between
various kinds of changes that agents make to their acceptance sets. Some alterations in
our normative commitments should be considered violations of what we care about - the
sorts of errors that are a kind of self-alienation.
These are changes that are unjustified by
the lights of one's own acceptance set. Other to our acceptance sets can be seen
changes
to be improvements. For example, imagine someone who is committed to moral equality
eliminating homophobic norms from his acceptance set. In this way, his acceptance set
becomes more consistent and, assuming that he accepts that he ought to be consistent, we
can say that he has his acceptance set.
improved
 42 LINDA RADZIK

meets five
important criteria of adequacy on such a theory - that it is
first person accessible, comprehensive, transparent, reflexive and that it
avoids the regress problem. The tie REC provides between reflection and
normative authority seems appropriate since, as Korsgaard argues, it is our
capacity to be reflective that sets the problem of answering the question of
authority in the first place.36

[0]ur capacity to turn our attention onto our own mental activities is also a capacity to
distance ourselves from them and call them into question. I perceive, and I find myself
with a powerful impulse to believe. But I back up and bring that impulse into view and
then I have a certain distance. Now the impulse doesn't dominate me and now I have a

problem. Shall I believe? Is this perception really a reason to believe?37

The reflective mind, because it is reflective,Skepticism needs a reason.

about the authority
of practical reason, Korsgaard continues, is not a meta
physical skepticism. It is not the doubt that there are values or reasons "out
there in the world." Rather, normative skepticism is the worry that when I
reflect on my impulses to believe, act, or accept norms, I will not be able
to come to any resolution.38 It is the fear that once I really look at my
principles, I will not be able to care about them anymore.39
But, as Korsgaard writes, "[i]f the problem springs from reflection, then
the solution must do so as well. If the problem is that our perceptions and
desires might not withstand scrutiny, is that they might."40
then the solution
Reflective Endorsement Coherentism provides a description of the way in
which a norm can withstand scrutiny. The norm can be such that it, and
everything that is counted as a reason for it, is supported by a norm that
one would endorse upon reflection, where the process of reflection would
itself be reflectively endorsed.

Department of Philosophy
Texas A&M University
College Station, TX 77843-4237
USA
E-mail: Iradzik?philosophy.tamu. edu

36
Korsgaard, The Sources of Normativity, pp. 92-94.
37
Korsgaard, The Sources of Normativity, p. 93.
38
Korsgaard, The Sources of Normativity, p. 94.
39
Korsgaard, The Sources of Normativity, pp. 13-14.
40
Korsgaard, The Sources of Normativity, p. 93.