The Philosophical Quarterly Vol. 58, No. 232 July 2008
ISSN 0031 8094 doi: 10.1111/j.1467-9213.2008.553.x
MODERATE CLASSY PYRRHONIAN
MORAL SCEPTICISM
By Walter Sinnott-Armstrong
This précis summarizes my book  Moral Skepticisms , with emphasis on my contrastivist analysis of
justified moral belief and my Pyrrhonian moral scepticism based on meta-scepticism about relevance.
This complex moral epistemology escapes a common paradox facing moral philosophers.
I. A PARADOX
Many moral philosophers today live a double life. They teach philosophy
classes which study extreme positions, such as moral egoism, nihilism and
relativism. Then they walk across campus to serve on ethics committees in
hospitals or laboratories.
This double life is filled with tensions. Suppose patient Bob tests positive
for HIV, but his doctor Alice thinks that she should lie to her patient about
the test results, because the bad news would hurt him, and doctors should
do whatever is best for their patients. Alice goes to the hospital ethics com-
mittee and asks whether it would be morally wrong for her to lie to her
patient about the test results. It would be very unusual for any member of
the committee to respond that morality is just an illusion, and so nothing is
morally wrong. If such moral nihilism were expressed, it would be dismissed
quickly with laughter or disdain and without argument.
This reaction should be contrasted with that in a philosophy class which
has studied moral nihilism, egoism and scepticism for a whole term. Many
bright students have defended these positions in discussions. In her final
paper, one student, Eve, argues that it is morally wrong for doctors to lie to
their patients, but she does not even mention moral nihilism, egoism or
scepticism. Eve would and should receive a low grade.
A paradox arises when we enter moral epistemology and ask whether the
committee s moral belief and the student s moral belief are epistemically
justified. The hospital ethics committee seems justified in concluding that
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
Published by Blackwell Publishing, 9600 Garsington Road, Oxford ox4 2dq, UK, and 350 Main Street, Malden, ma 02148, USA
MODERATE CLASSY PYRRHONIAN MORAL SCEPTICISM 449
the doctor should not lie, even though the committee never considers moral
nihilism, egoism or scepticism. In contrast, the philosophy student does not
seem justified in reaching the same conclusion, because she has no response
to such extreme positions. This is paradoxical: if neither the philosophy
student nor the committee can rule out moral nihilism, then how can this
inability show that the philosophy student s moral belief is not justified when
the committee s moral belief is justified?
One main purpose of my book Moral Skepticisms (Oxford UP, 2006), was
to resolve this paradox. The solution lies in contrast classes.
II. CONTRAST CLASSES
Contrast classes are apparent in explanations: temperature is the reason why
it is raining as opposed to snowing, but humidity is the reason why it is
raining as opposed to not precipitating at all. Similarly for reasons to act: my
reason to bake my son a birthday cake on May 21 as opposed to June 21 is
that his birthday is May 21, but my reason to bake him a birthday cake on
May 21 as opposed to baking him a pie is his dislike of pie. When surveyed,
all kinds of reasons seem to introduce relativity to contrast classes.
Reasons for belief are no exception. Suppose I see a bird on my lawn and
believe that it is a crow. I can distinguish my actual visual experience from
the experience which I would have if the bird were a cardinal or an ostrich,
but I cannot distinguish my visual experience from the experience that I
would have if the bird were a raven or a rook. Hence my visual experience
gives me evidence and a reason to believe that it is a crow as opposed to a
cardinal or ostrich, but no evidence or reason to believe that it is a crow as
opposed to a raven or rook. I am then justified out of one contrast class but
not out of the other in believing that it is a crow. (Even a reliabilist should
admit this, because a birdwatcher can be reliable at picking out some birds
but not others.) The plausibility and usefulness of these relativized claims
reveals a place for contrast classes in epistemology.
This place can be specified by definitions:
S is justified out of a contrast class c in believing a proposition p when
and only when S is not able to rule out p and is able to rule out the
disjunction of all other members of c.
This account is analogous in some ways to conditional probability. Just as
the conditional probability of p given q is calculated by counting only poss-
ibilities where q is true, so we determine whether S is justified in believing p
out of c by assuming that some member of c is true and then asking whether
S has evidence against the alternatives to p within c.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
450 WALTER SINNOTT-ARMSTRONG
This definition does not depend on any view about scepticism, so it
should be acceptable to sceptics and anti-sceptics alike. Still, it illuminates
scepticism when we distinguish two contrast classes:
The extreme contrast class for a belief that p = all claims contrary to p, including
sceptical scenarios which are systematically uneliminable
The modest contrast class for a belief that p = all claims contrary to p which need
to be eliminated in order to meet usual epistemic standards.
To call certain standards  usual is not to endorse them but only to say that
most people use them. In this non-normative sense, the usual epistemic
standards do not require believers to eliminate sceptical scenarios, so the
extreme and modest contrast classes differ.
These two contrast classes can be used to distinguish two epistemic
judgements:
S is justified out of the extreme contrast class in believing p
S is justified out of the modest contrast class in believing p.
Judgements of this form are denied by two kinds of relativized moral
scepticism:
Scepticism about extremely justified belief = no belief is justified out of the
extreme contrast class
Scepticism about modestly justified belief = no belief is justified out of
the modest contrast class.
The former does not entail the latter, so a compromise is possible:
Moderate scepticism about justified belief = scepticism about extremely
justified belief but not about modestly justified belief.
This compromise should be attractive to those who admit that we cannot
rule out extreme sceptical scenarios, such as deceiving demons and brains in
vats, but who also think that we can sometimes rule out all but one member
of the class which needs to be considered in order to meet usual epistemic
standards.
III. MORAL CONTRAST CLASSES
This structure applies neatly to moral beliefs. Just add  moral before  belief
in the definitions.
The only complication is that moral beliefs are relative to three kinds of
contrast classes. A moral believer might be justified in believing that one act
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
MODERATE CLASSY PYRRHONIAN MORAL SCEPTICISM 451
as opposed to a second act is morally wrong, without being justified in
believing that the first act as opposed to some third act is morally wrong.
Hence we need a class of contrasting acts. Similarly, a moral believer might
be justified in believing that an act has one moral status (such as being
morally right) as opposed to a second moral status (such as being morally
wrong), without being justified in believing that the act has the first status as
opposed to some third status (such as being supererogatory). Thus we need a
class of contrasting moral statuses. Again, a moral believer might be justified
in believing in one moral theory as opposed to a second moral theory,
without being justified in believing in the first theory as opposed to some
third theory (if, for example, he has a counter-example which refutes the
second but not the third). So we need a class of contrasting moral theories.
For any of these contrasts, moderate moral scepticism follows from these
definitions if some moral believers can sometimes rule out all but one of the
alternatives in the modest contrast class, but no moral believer can ever rule
out all of the extreme alternatives lying outside the modest contrast class but
inside the extreme contrast class. My book argues for this compromise.
Although other extreme moral views could work, I focus on moral nihil-
ism, which claims that nothing is morally wrong. It is important that real
people really believe in moral nihilism on the basis of real arguments. Hence
moral nihilism cannot be dismissed on grounds which some (such as Peirce)
cite for dismissing general sceptical scenarios, such as deceiving demons and
brains in vats, which nobody really believes in or argues for.
But can moral believers rule out moral nihilism? No. To rule out a mem-
ber of a contrast class, believers need some conflicting ground which does
not beg the question. Some moral believers try to rule out moral nihilism by
appealing to moral intuitions, but those moral intuitions are predicted and
explained by moral nihilists with the help of biology and psychology, so it
begs the question to cite moral intuitions against moral nihilism. Moral
nihilism also cannot be ruled out by logic or semantics, because it is logically
and semantically consistent when developed properly. Furthermore, no non-
normative premises, inference to the best explanation, or contractualist
framework can justify a denial of moral nihilism. To show all of this, I run
through naturalism, normativism, intuitionism and coherentism in part II of
my book. I cannot repeat those arguments here, so I shall simply take it for
granted that we cannot rule out moral nihilism. It follows that we are not
extremely justified as moral believers.
None the less, in my closing chapter, I show how some of us can some-
times use coherentist methods to rule out all but one of the alternatives
in the modest moral contrast class. These methods are exemplified in
properly structured ethics committees. The main philosophical trick involves
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
452 WALTER SINNOTT-ARMSTRONG
second-order beliefs about reliability. Again I cannot go into detail here, so I
shall simply take it for granted that some moral believers are modestly
justified. The result is moderate scepticism about justified moral belief.
IV. RELEVANCE AND CONTEXTUALISM
Everything I have said so far could be accepted by a wide variety of
epistemologists, including invariantists and contextualists, because I have
discussed only relativized epistemic judgements with explicit contrast classes.
The classic disputes in epistemology arise over unrelativized epistemic
judgements without explicit contrast classes.
Some epistemologists might hold that unqualified epistemic judgements,
such as  S is justified in believing p , are primitive and should not be ana-
lysed in terms of contrast classes at all. This dodge is hardly illuminating.
Moreover, because so many epistemic judgements (and judgements about
reasons of other kinds) are relative to contrast classes, the burden of proof
seems to lie on anyone who claims that any epistemic judgements are excep-
tions to that general pattern. Besides, almost any account of  S is justified in
believing p will be equivalent to some account in terms of contrast classes.
Thus there is no good reason to avoid contrast classes in the analysis of
unqualified epistemic judgements.
With contrast classes, several options arise, but I propose this analysis:
S is justified (without qualification) in believing p = S is justified, out of
the contrast class which is relevant, in believing p.
This proposal should be obvious, since to call a contrast class  relevant here
is just to say that a believer needs to be able to rule out all other alternatives
in that contrast class in order to be justified without qualification in believing
a member of that contrast class. What is at issue is relevance to the status of
being epistemically justified, not relevance to practical purposes, to morality,
to religion, or to anything other than epistemic justifiedness.
This analysis is compatible with a variety of views about which contrast
class is relevant. Invariantists hold that a certain contrast class is always
relevant. Extreme invariantists see the extreme contrast class as always relevant.
Modest invariantists see the modest contrast class as always relevant. Context-
ualists then deny invariantism and hold instead that which contrast class is
relevant varies from context to context. The standard version of contextual-
ism claims that the modest contrast class is relevant in everyday contexts,
and the extreme contrast class is not relevant in everyday contexts but is
relevant in some philosophical contexts.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
MODERATE CLASSY PYRRHONIAN MORAL SCEPTICISM 453
In my book, I argue against all of these views. Since extreme invariantism
implies that all contrary propositions are always relevant, it conflicts with the
common practice of dismissing some outlandish sceptical alternatives as
irrelevant. On the other hand, modest invariantism conflicts with the
common practice of recognizing that some alternatives which we had not
previously recognized as relevant are relevant after all. Thus neither varia-
tion on invariantism captures our common epistemic practices. I do not see
how either kind of invariantism improves upon our common epistemic prac-
tices. For these reasons, I reject invariantism.
I also reject contextualism. Since contextualists claim that which contrast
class is relevant varies with context, they need to say which contrast class,
whose context, and why. For example, if Sue believes that her pet is a dog,
does the relevant contrast class include dingoes and hyenas? What about
similar species unknown to Sue or to anyone? Contextualists can stipulate
answers, but they cannot justify any particular answer, so they cannot
adequately specify which contrast class is relevant. Moreover, if Todd judges
whether Sue is justified in believing that her pet is a dog, is it Todd s context
or Sue s context that determines whether Sue is justified without quali-
fication? The answer matters if Todd worries about dingoes and hyenas, but
Sue does not. Yet different philosophers give different answers, and I see no
firm basis for either answer, so I see no way for contextualists to specify
whose context is crucial. Finally, even if we settle on a particular person s
context, we still need to know which features of that context matter. Is
the context determined by what people say publicly or what they believe
privately? Is the context identified by the person s interests or purposes
(which can work against real interests)? Does one person s epistemic context
depend on beliefs and purposes of other people or groups? Do physical or
institutional surroundings affect a person s epistemic context?
Answers to these questions sometimes affect whether a given contrast
class is relevant, and hence whether a given believer is justified without
qualification, according to contextualists. Thus contextualists cannot avoid
answering such questions (along with other questions like whether a believer
can be in more than one context at once). As always, contextualists can pick
an answer and stick with it, but I see no way for them to show why any
feature of the context determines which contrast class is relevant or who is
justified without qualification. Of course, contextualists might surprise me
with an argument I never imagined. Still, in the absence of any such argu-
ment, contextualists should not get away with vague hand-waving declara-
tions about context and contrast without specifying which contrasts, whose
context, and why. Until contextualists answer these questions and justify
their answers, contextualism remains at best underdeveloped and dubious.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
454 WALTER SINNOTT-ARMSTRONG
V. PYRRHONISM
After rejecting invariantism and contextualism, what option is left? My
answer is Pyrrhonian. Since I see no firm enough basis for claiming that any
contrast class is relevant or for claiming that any contrast class is not
relevant, I suspend belief about which contrast class is relevant and about
whether any contrast class is relevant, even in a given context. Sometimes
a certain contrast class will seem relevant or irrelevant, but that is only
appearance. I neither assert nor deny that any contrast class is ever really
relevant.
On my analysis (above), the epistemic judgement that S is justified with-
out qualification in believing p presupposes that some contrast class is
relevant. Hence I also suspend belief about whether S is justified without
qualification in believing p, for any S and for any p. Such unqualified epi-
stemic judgements are not needed in order to locate a believer s epistemic
position, which is the job of epistemology. Relativized epistemic judgements
do that job perfectly well, so nothing essential is lost by giving up unqualified
epistemic judgements.
Is this scepticism? Not as scepticism is usually understood by its oppon-
ents. Anti-sceptics normally present scepticism as the dogmatic claim that
nobody is justified in believing anything (or that nobody knows anything).
This is a straw man. No real sceptic has held such dogmatic scepticism for
more than a short while. All real sceptics are Pyrrhonian sceptics like me. I
do not deny justified belief. Instead, I suspend belief about justified belief.
The difference between denial and suspension is crucial, because dogmatic
denials take on a burden of proof which is avoided by suspension of belief.
My scepticism is also moderate. I suspend belief only about whether
anyone is justified without qualification. This suspension is compatible with
relativized epistemic judgements, such as that Sue is justified in believing
that her pet is a dog as opposed to a cat, and that she is not justified in be-
lieving that her pet is a dog as opposed to a dingo or hyena. More generally,
Pyrrhonian suspension of all unrelativized epistemic judgements is compat-
ible with a moderate scepticism which claims that no believer is ever
justified out of the extreme contrast class but some believers are some-
times justified out of the modest contrast class. These relativized epistemic
judgements do not presuppose that any contrast class is relevant, so they are
defensible even if relevance is indeterminable. Moderate scepticism can thus
be combined with Pyrrhonian scepticism.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
MODERATE CLASSY PYRRHONIAN MORAL SCEPTICISM 455
VI. MORAL PYRRHONISM
This general epistemology applies easily to moral beliefs. The crucial
question then is whether moral nihilism is a relevant alternative to common
moral beliefs. Extreme invariantists say  Yes . Modest invariantists say  No .
Contextualists say  Sometimes . Moral contextualism is developed most fully
by Mark Timmons, who writes
Regardless of whether members of the [ethics] committee could defend their basic
moral beliefs against sceptical challenges, or even whether they had justifying reasons
that they might be able to rehearse, I submit that the moral conclusions reached by
the committee were justified ones. In more of a detached context, in which sceptical
challenges get their hold, those same beliefs would be in need of justification.1
In my terminology, what Timmons claims is that moral nihilism is relevant
in detached contexts but irrelevant in engaged contexts, such as ethics
committees.
But why is moral nihilism not relevant in engaged contexts? Timmons
says it is not, but why must anyone agree? After all, moral nihilism is logic-
ally inconsistent with common moral beliefs in engaged contexts. Moreover,
some real people really believe in moral nihilism for real reasons, even if
they do not assert it openly in engaged contexts. Why must they be present
or outspoken for their views to be relevant? Admittedly, the purposes of a
hospital ethics committee would be undermined if its members had to wait
until they could rule out moral nihilism. Such practical considerations do
affect which contrast classes seem relevant and whether it is useful to take
them to be relevant, but maybe all this shows is that ethics committees need
to act without justified beliefs. None of it shows that moral nihilism really is
irrelevant to whether engaged moral believers are justified epistemically.
I also know of no adequate way to establish the opposite claim, that
moral nihilism really is relevant epistemically. In engaged contexts, most
people feel justified in forming moral beliefs while recognizing that others
believe in moral nihilism and admitting a lack of good argument against
them or their view. Even in contexts where philosophers discuss moral
nihilism, the fact that someone mentions moral nihilism suggests that the
speaker thinks or hopes that it is relevant, but that hardly shows that it really
is relevant. If the mere mention of a sceptical alternative made the alter-
native relevant, then a defence attorney could simply mention the hypo-
thesis of a deceiving demon, and that mention would be enough to ensure
1
M. Timmons, Morality without Foundations (Oxford UP, 1998), p. 221.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
456 WALTER SINNOTT-ARMSTRONG
that the jury is not justified in believing that the defendant is guilty. Ana-
logously, mentioning moral nihilism does not make it relevant. Moreover, it
is hard to imagine any other way to settle the controversial issue of whether
moral nihilism is relevant.
Consequently I suspend belief about whether moral nihilism is or is not
relevant epistemically. Hence I suspend belief about whether any moral be-
lief is epistemically justified without qualification. That makes me a moral
Pyrrhonist. Nevertheless, I can still consistently deny that any moral belief is
justified out of the extreme contrast class, which includes moral nihilism. I
also claim that some moral beliefs are justified out of the modest contrast
class, which does not include moral nihilism. That makes me a moderate
moral sceptic. My use of contrast classes makes my view classy, of course.
All together, this explains my title, for my view ends up as moderate classy
Pyrrhonian moral scepticism.
This conclusion solves the opening paradox. In that case, the ethics com-
mittee seems justified if we think of the modest contrast class. The philosophy
student seems unjustified if we think of the extreme contrast class. This
combination is not contextualist, because none of these claims changes its
truth-value as context varies. The committee and the student are both justi-
fied out of the modest contrast class but not out of the extreme contrast
class, regardless of context. That is how I avoid the troubles for contextual-
ism. Confusion arises only if we ask whether the doctor and the student are
really justified without qualification. When Pyrrhonists reject that crude
language game, they explain away the appearance of paradox.
Dartmouth College, New Hampshire
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly