The Philosophical Quarterly Vol. 58, No. 232 July 2008
ISSN 0031 8094 doi: 10.1111/j.1467-9213.2008.561.x
REPLIES TO HOUGH, BAUMANN AND BLAAUW
By Walter Sinnott-Armstrong
I reply to comments by Gerry Hough, Peter Baumann and Martijn Blaauw on my book Moral
Skepticisms. The main issues concern whether modest justifiedness is epistemic and how it is related
to extreme justifiedness; how contrastivists can handle crazy contrast classes, indeterminacy and
common language; whether Pyrrhonian scepticism leads to paralysis in decision-making or satisfies
our desires to evaluate beliefs as justified or not; and how contextualists can respond to my arguments
against relevance of contrast classes.
I am very grateful to Gerry, Peter and Martijn both for their praise and for their
criticisms. I shall not be able to do justice to their subtle and varied comments in this
short reply, but I shall try to suggest the general direction in which I can offer solu-
tions to the main problems they raise.
I. REPLY TO HOUGH
Hough s central objection is a dilemma. He distinguishes two interpretations of my
claim that modest justifiedness is  independent of whether sceptical hypotheses are
true or whether they can be ruled out epistemically , which he labels  the independ-
ence claim .1 Then he argues that neither interpretation is adequate, so my claim is
defective.
On the stronger interpretation, the independence claim is supposed to mean this:
 S s modest justification in believing p entails that S is extremely justified in believing
the conditional  p is true if any of the modest alternatives are  .2 Hough s example
of Scotland and England (p. 461) is directed against this strong interpretation:  once
I consider the possibility that my senses are systematically deceiving me ... then I am
no longer justified in believing the conditional claim that I am in Scotland, if I am in
Scotland or England . However, I had already rejected this strong interpretation
in my book (p. 110). Modestly justified belief is not the same as extremely justified
belief in a conditional. Thus I avoid this horn of Hough s dilemma.
1
See W. Sinnott-Armstrong, Moral Skepticisms (Oxford UP, 2006, hereafter MS), p. 111.
2
G. Hough,  A Dilemma for Sinnott-Armstrong s Moderate Pyrrhonian Moral Scep-
ticism , The Philosophical Quarterly, 58 (2008) (this issue), pp. 457 62, at p. 460.
© 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
REPLIES TO HOUGH, BAUMANN AND BLAAUW 479
If Hough s alternatives are exhaustive, I am left with his weaker interpretation
(p. 460):  The weaker version of the independence claim is that modest justification
is independent of sceptical hypotheses in so far as modest justification allows
knowers simply to ignore sceptical hypotheses . He calls this claim  more plausible
and even  incontestable . His only objection is that this interpretation  leaves it
unclear why modest justification counts as a kind of epistemic justification (ibid.). I
shall explain why.
My model for modestly justified belief is conditional probability (MS, pp. 110 11).
Imagine a location where on average it snows on 9 days per year, and all of these
days occur during the 3 winter months with a total of 90 days. The probability of
snow on a random day is less than 0.025, but the conditional probability of snow
today, given that it is winter, is 0.1. This conditional probability, in effect, ignores all
of the days when it is not winter. This analogy explains what is meant when the
weak interpretation says that modestly justified belief  allows us simply to ignore
sceptical hypotheses (Hough, p. 460).
In contrast, Hough s strong interpretation makes modestly justified belief ana-
logous to the probability of a conditional. In my example, the probability of the
conditional  If it is winter, it will snow is greater than 0.75, because that conditional
is true by falsity of antecedent during all of the nine months when it is not winter.
That model does not ignore non-winter days in the way in which my model does.
Now Hough asks (ibid.) why the weak interpretation modelled on conditional
probability  counts as a kind of epistemic justification . The answer depends on what
 epistemic means. I contrast instrumentally justified beliefs with epistemically
justified beliefs on the grounds that  epistemic justifiedness is tied to truth and an
epistemic ground for a belief  raises the probability of the truth of the belief or at
least its  conditional probability (MS, p. 64). Since I model modest justifiedness on
conditional probability, the grounds which make a belief modestly justified can also
increase its conditional probability given that sceptical hypotheses are false. That
explains why modest justifiedness is epistemic in my sense.
Hough responds  The worry is that having a high probability of being true given
that sceptical hypotheses are false does not entail having a high probability of being
true . In short, conditional probabilities do not entail unconditional probabilities.
Right! This is a feature, not a bug. If being modestly justified did entail a high un-
conditional probability of truth, then it would require the believer to be justified
unconditionally, that is, extremely or without qualification. These are exactly the
implications which a moderate Pyrrhonian seeks to avoid. But why is this supposed
to show that modestly justified beliefs are not justified epistemically?
Hough s answer might be that my model  will not placate sceptics and convince
them that what we have in such cases is an instance of epistemic justification
immune to the sceptical challenge (p. 460; cf. p. 461). Well, my goal is not to
placate radical or dogmatic sceptics. I do want them to agree that some moral beliefs
are modestly justified in my technical sense, but it does not bother me if they deny
that this admission matters to them. However, if dogmatic sceptics go on to
claim that no way of being justified counts as epistemic unless it uses the extreme
contrast class, and thus satisfies them, then I would ask them why they alone are
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
480 WALTER SINNOTT-ARMSTRONG
entitled to define the limits of what is epistemic, as if they set the agenda for all
epistemology.
What makes a reason epistemic is not its ability to satisfy dogmatic sceptics but
rather its connection with truth. The question, then, is whether that connection
must be unconditional. Hough responds (p. 460)  What I do not accept is that a
ground for belief is tied to truth if it increases the conditional probability of the belief
given that sceptical hypotheses are false . I see no reason to restrict what counts as
epistemic to unconditional probabilities. Much scientific evidence works by increas-
ing conditional probabilities (see MS, p. 64). The same goes for evidence in law and
in everyday life.
One can show why conditional probability is enough thus. Suppose we do not
know how likely it is that Henry is coming to our Harry Potter costume party, but
we do know that Henry s favourite character in the novels is Hagrid and that Henry
likes to dress up as Hagrid. Henry s preferences increase the conditional probability
that Henry will come as Hagrid, given that Henry comes at all. Thus Henry s pre-
ference is evidence that Henry will come as Hagrid, given that he comes at all. This
evidence is conditional, but it is still epistemic, because its force as evidence depends
only on its connection with truth and not on any connection with any other value.
Of course, Henry still might not come at all, so we cannot determine the uncon-
ditional probability that Henry will come as Hagrid. We would be in a better
epistemic position if we knew whether Henry is coming. Nevertheless, Henry s
preferences still count as epistemic evidence. If this evidence is not epistemic, what
kind of evidence is it? Since a conditional connection with truth is enough for evid-
ence to be epistemic in this case, an analogous conditional connection with truth is
also enough for modestly justified beliefs to be justified epistemically.
II. REPLY TO BAUMANN
Baumann asks three questions. The first is about crazy contrast classes. I agree that
some contrast classes seem crazy, and {it is vanilla ice cream, it is fish fingers} is one
of them. I also agree with Baumann s account of why this class seems crazy:  there
seem to be no practical interests (or anything else) which would determine this
particular contrast class .3 This might seem to be a problem for my theory, which
implies that I know (or am justified in believing) that I am tasting vanilla ice cream
rather than fish fingers. However, this epistemic judgement can be true, even if it is
odd to utter it because it is practically useless. It is as if someone asked me to
describe my wife, and I said  She s not a fish finger . This strange response is true.
So is  I know that this is vanilla ice cream rather than fish fingers .
We have to be careful, though. Baumann asks  do I know that it is vanilla ice
cream if I can only ... distinguish it from fish fingers, but not, say, from strawberry
ice cream? (p. 464). The problem is that this question asks what I know without
qualification. A Pyrrhonian contrastivist should reject any inference from  I know it
3
P. Baumann,  Problems for Sinnott-Armstrong s Moral Contrastivism , The Philosophical
Quarterly, 58 (2008) (this issue), pp. 463 70, at p. 464.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
REPLIES TO HOUGH, BAUMANN AND BLAAUW 481
is vanilla ice cream rather than fish fingers to  I know it is vanilla ice cream .
Pyrrhonian contrastivists can also explain why we should not interpret any normal
utterance of  I know it is vanilla ice cream as  I know it is vanilla ice cream rather
than fish fingers , since no normal speaker would or should have this contrast in
mind. But then no serious problem for my view arises from this crazy contrast class.
The same goes for Baumann s crazy moral contrast classes. On my view, I am
justified in believing that it is morally acceptable to offer my thirsty friend some
petrol to drink rather than the lethal insecticide. This epistemic judgement is true,
even though it is very odd, misleading, and possibly even immoral to utter it or to
think in terms of this contrast class. It is like saying that it would be better to blind all
of the witnesses to my crime than to kill all of these witnesses. The former is better, I
suppose, but we should not be choosing among these options in the first place.
So how do I answer Baumann s question? Are there constraints on admissible
contrast classes? That depends on what  admissible means. Crazy contrast classes
are admissible in the sense that some judgements with them are true. None the less,
they are inadmissible in the sense that it would be odd, misleading, and sometimes
immoral to utter those judgements or to think in terms of those contrast classes.
Does this answer admit that there are constraints on relevant contrast classes, as
Baumann suggests? No, because my theory was only about relevance to the truth of
epistemic judgements. We can understand crazy contrast classes without calling
them irrelevant in that epistemic sense.
Other classy Pyrrhonian sceptics might disagree with me here. They might deny
the truth of  I know it is vanilla ice cream rather than fish fingers at least when I
cannot distinguish vanilla from strawberry. They can still reject the notion of
relevance among admissible contrast classes. However, this concession that crazy
contrast classes are irrelevant seems to me to conflate epistemic relevance with
practical and moral admissibility. I find it more illuminating to keep them separate,
even if it does lead to me to defend the truth of some odd judgements.
Baumann s second question is about indeterminacy. In his example (p. 466), an
agent is driving on a street which is deserted in the sense that there is almost never
anyone on or next to this street, and nobody is on or next to this street when the
driver enters it (and the driver knows this). However, a child suddenly and unex-
pectedly runs into the street, and the driver cannot stop before he hits and kills the
child. The theoretical problem arises because this deserted street lies within a larger
densely populated neighbourhood (and the driver knows this). The driver seems
reckless, and hence responsible, if we think at a higher level of generality (the
neighbourhood) but not if we think at a lower level of generality (the street).4
It is tempting to respond that an agent is responsible if and only if his act is reck-
less at either level, but that seems wrong when the driver looked carefully and saw
nobody before accelerating to 20 mph. Another possible response is that an agent is
blameworthy if and only if his act is reckless at the smallest level of generality. If the
smallest level were the street, this second response would remove the driver s
4
This puzzle about responsibility has illuminating parallels to the problem of generality for
reliabilist theories of knowledge. See E. Conee and R. Feldman,  The Generality Problem for
Reliabilism , in Conee and Feldman (eds), Evidentialism (Oxford UP, 2004), pp. 135 58.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
482 WALTER SINNOTT-ARMSTRONG
responsibility. Unfortunately, there is always an even smaller level of generality, such
as the metre before and after the spot where the child runs out. That area is not
deserted, so the driver is responsible again according to this second response; and
yet that conclusion can still seem wrong, depending on details. A third response
introduces degrees of responsibility. A driver is most responsible if both the street and
the neighbourhood are crowded, slightly less responsible if the street is crowded
but the neighbourhood relatively deserted, moderately responsible if the neighbour-
hood is crowded but the street is deserted, and not at all responsible if both the street
and the neighbourhood are deserted. However, this response just assumes a solution
to Baumann s case where the neighbourhood is crowded but the street is deserted.
The question is why the driver should be held responsible at all in that case.
Baumann s solution is to relativize judgements of moral responsibility to under-
lying reference classes. He realizes that this additional relativization is compatible
with my suggestion that we relativize epistemic judgements to contrast classes. Fine;
but then (p. 467) he concludes  I think all this also makes plausible a certain version
of contextualism . Why contextualism rather than contrastivism? A contextualist
about these matters would hold, presumably, that which reference class (or level of
generality) is relevant varies with context, so it is true in some contexts and false in
other contexts to say that Baumann s driver is responsible. Baumann does not
describe contexts which affect our judgements of responsibility in these ways. Maybe
that is not what he means. In any case, it seems to me more plausible to go
contrastivist: the driver is reckless and responsible relative to one reference class but
not relative to the other.5 Then we might also go Pyrrhonian: suspend belief about
which reference class is the relevant one and about whether the driver is responsible
without qualification. If this is plausible, then Baumann s addition of another kind of
relativization fits neatly with the structure of my moral epistemology.
Baumann s third question is how my Pyrrhonian contrastivism avoids paralysis.
He argues (p. 468) that whenever we face any choice (or at least any choice between
what is morally required and what is in the agent s self-interest), we can always
formulate two arguments which lead to contrary conclusions that we should and
should not do the moral act. These conflicting arguments do not actually lead to
paralysis or an inability to act, since we can choose and act even if we cannot give
any reason for choosing one act over another. Unlike Buridan s ass, humans can
choose arbitrarily, and this ability is useful. None the less, Baumann s point is that
contrastivists cannot have any adequate reason to choose either action or to prefer
either of the conflicting arguments to the other.
The crucial premise in Baumann s second argument is (B2)  Even if someone
badly needs my help and I could help with little effort, it is still unclear whether I
5
If an agent is responsible for a harm only if that agent s act causes that harm, and if
causation is best analysed along contrastivist lines, as argued by J. Schaffer,  Contrastivist
Causation , Philosophical Review, 114 (2005), pp. 327 58, then responsibility has another con-
trastivist dimension as well. Moreover, if responsibility requires freedom, and if freedom is
always freedom from a certain range of factors within a contrast class, as I argue in  A
Contrastivist Manifesto , Social Epistemology (forthcoming), then responsibility has yet another
contrastivist dimension. It should not be surprising that complex notions like responsibility
are contrastivist in multiple ways.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
REPLIES TO HOUGH, BAUMANN AND BLAAUW 483
have any obligation to help; if this is unclear and if there is a (small) cost to helping,
then I should, all things considered, not help (p. 468). This conjunctive premise is
problematic in several ways.
First, to say that it is unclear whether I have an obligation to help does not imply
that I definitely do not have an obligation to help. Despite its unclarity, there still
might really be an obligation to help. If there is an obligation to help, then this
obligation outweighs the admittedly small costs. That makes it hard to see how any-
one could conclude that I definitely should not help that person. All that follows, if
anything, is that it is unclear whether I should help. This conclusion would still be a
problem for my view, however, so I cannot stop here.
More importantly, it is unclear what the term  unclear means or why moderate
Pyrrhonian contrastivists have to admit that it is  unclear whether I have any ob-
ligation to help . Baumann s idea seems to be that moderate Pyrrhonian contrast-
ivists see every obligation as unclear because they suspend judgement about (i) and
deny (ii):
(i) I am justified without qualification in believing that I have an obligation to help
(ii) I am extremely justified in believing that I have an obligation to help.
Nevertheless, moderate Pyrrhonian contrastivists can still accept both of the
judgements (iii) amd (iv):
(iii) I am modestly justified in believing that I have an obligation to help
(iv) I have an obligation to help.
I think (iii) and (iv) are enough for moderate Pyrrhonian contrastivists to hold that it
is clear that I have an obligation to help in Baumann s case. If so, they can reject
the part of Baumann s premise (B2) before the semi-colon. Judgements (iii) (iv) cure
paralysis.
Baumann might respond that even if I endorse (iii) (iv), as long as I disavow
(i) (ii), it is still unclear to me whether I have an obligation to help. To avoid verbal
quibbles, the term  unclear can be used in this more demanding way. The fact that
my obligation is unclear in this new sense still does not show that I should not help.
It is not even odd to hold (iv) while disavowing (i) (ii). Although Pyrrhonian con-
trastivists suspend belief about unqualified epistemic claims, nothing in their view
makes them suspend belief about substantive moral claims. They can believe in
moral obligations just as strongly as anyone else. Thus they can deny the part of
premise (B2) after the semi-colon on this new use of  unclear .
Either way, the solution for moderate Pyrrhonian contrastivists is to reject the
second argument in Baumann s dilemma. The same solution applies to my doctor
example. Baumann asks  Which  knowledge should she take as the basis of
action? Contrastivism cannot recommend one rather than the other (p. 469).
Granted. Contrastivism is a theory of knowledge and justified belief. As such, it
implies nothing about the basis of action and cannot recommend either belief as the
basis for action. That is a question for theories of action or of practical reasoning.
None the less contrastivism is compatible with plausible answers to Baumann s
question. Since we are never justified extremely or without qualification in believing
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
484 WALTER SINNOTT-ARMSTRONG
anything, it would be fruitless to require beliefs to be justified in those ways before
they could be used as a basis for choice and action. The most that a plausible theory
of action and practical reasoning can require is modestly justified beliefs. This
requirement can be met, and I see no reason why moderate Pyrrhonian contrastiv-
ists cannot embrace this requirement without suggesting that the modest contrast
class is the relevant one for epistemology.
In the end, I agree with Baumann when he says (ibid.)  The root of both versions
of the problem lies in the  absoluteness which moral action requires . I escape the
problem by denying that moral action requires absoluteness, if that means extremely
or unconditionally justified belief. We do need to use some modest contrast class
when making decisions, but that does not make that contrast class absolutely
relevant for epistemology. Some physicists used to claim that their science requires
absolute time, space and motion. Now we recognize that absoluteness is not really
required for physics. Still, we pick a frame when we have to decide how fast to drive
on a highway. Analogously, moral epistemologists should give up the assumption
that absoluteness in justified belief is required for rational choice or for ethics, but
they can still use modest contrast classes for making decisions in everyday life.
III. REPLY TO BLAAUW
Blaauw begins by characterizing my brand of contrastivism as revisionist. He defines
revisionist contrastivism as a conjunction:  one could defend the position that [i]
contrastivism about knowledge is not embedded in the way in which people actually
use  knows , but that [ii] there are compelling reasons why it would be better to
adopt the contrastive  knows as opposed to the binary  knows  .6 I accept the
positive part, [ii]. Still, I do not want to commit myself to the negative part, [i],
partly because I am not sure exactly what  embedded means.
In one sense, an account of knowledge or justified belief is embedded in actual
use only if the whole account never goes beyond or behind patterns of use shared by
all competent speakers of English (or other languages?), and only if the account
never suggests that any aspect of common language ever leads us astray. If that is
what  embedded means, I happily admit that my account is not embedded in
common usage and that my view is revisionist to that extent. But then it is not clear
why Blaauw thinks  this seems ... to be a problem for Sinnott-Armstrong s position
(p. 473). He never says why, and I see no reason to treat common language as in-
fallible or as the sole source of epistemological insights.
Nevertheless, my contrastivist account is supposed to be connected in a more
limited way with common language. My account analyses what we are talking about
when we talk about justified belief, as  Water is H2O analyses what we are talking
about when we talk about  water . To that extent, I do think that contrast classes are
 embedded in common language. That does not imply, however, that speakers
consciously think in terms of contrast classes, any more than ancient speakers or
6
M. Blaauw,  Contesting Pyrrhonian Contrastivism , The Philosophical Quarterly, 58 (2008)
(this issue), pp. 471 7, at p. 472.
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
REPLIES TO HOUGH, BAUMANN AND BLAAUW 485
young children who talk about water without knowing that water is H2O think in
terms of H2O.
As a result, I reject Blaauw s tests for ellipses. Blaauw admits that  John is tall is
elliptical. Then he claims (p. 473) that when people say  John is tall , (a)  On reflec-
tion, most speakers will recognize that tallness attributions are usually made against
the background of a set of alternatives ; (b) the speaker  must have some idea of
what the relevant contrast class ... is ... at the very least ... upon reflection ; (c)  when
we do make the contrasts explicit, this will not elicit surprised reactions ; and (d)
when we ask  Tall for what? , most speakers will understand this question (unlike the
question  Justified with respect to what? ).
These claims are predictions, so I tested them. I asked one friend whether
another friend is tall. The other friend is six feet three inches in height. The first
friend answered,  Yes, he is tall . Then I asked  Tall for a what? . Her face expressed
surprise. Then she asked  What do you mean? . Eventually she said  Tall compared
to other people I know . But when I asked  Are you sure that you did not mean
 Tall for a man ? , she admitted uncertainty. Upon more reflection, she became
confused and reported that she did not understand the question. Additional
exchanges produced similar reactions, but I did not do a controlled study. Still, it is
at least not clear that, even after reflection, common speakers will understand or will
not be surprised when we ask them to spell out the ellipses that Blaauw himself
admits.
Besides, many other ellipses are less obvious. If someone asks how fast my car is
moving, and I reply  Relative to which frame? , many speakers will look surprised
and will not understand what I am asking. Still, this does not rule out the hypothesis
that  Your car was going at 100 kilometres per hour is elliptical for  Your car was
going at 100 kilometres per hour relative to the road (or  to the earth ?). Closer to
epistemology, if someone asks  Why is it snowing? , and I answer  Are you asking
why it is snowing instead of raining or why it is snowing instead of not precipitating
at all? , I suspect that many speakers will be surprised and will not understand my
question. Still, this hardly rules out the philosophical claim that explanations are
relative to contrast classes.
The relativity of epistemic judgements to contrast classes fits this pattern of
hidden ellipses. As Blaauw says (p. 473), it  would elicit surprised reactions to point
out this relativity, and  very few people would even understand  the question when
asked which contrast class they are saying that a believer is justified out of. However,
this shows only that this relativity is embedded deeply, not that it is not embedded at
all. What elicits surprise depends on what people are used to. If contrastivism
catches on, then the surprise will fade. Even if the surprise never fades, that will not
refute contrastivism about knowledge and justified belief, any more than relativistic
physics is refuted by the surprise that some people still feel when asked  Relative to
which frame? . All the surprise shows is that they are in the grip of a false pre-
supposition and need more time to get comfortable with the complexities of the
world.
Blaauw s objection in his żII can be handled in a similar way. He complains
(p. 475) that Pyrrhonians  cannot slake our desire to evaluate whether our beliefs are
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
486 WALTER SINNOTT-ARMSTRONG
justified or not without qualification. Of course not! Pyrrhonians do not try to
satisfy that desire, because they see that desire as illegitimate. When physicists and
others first began to recognize that speed is relative to a frame, some people kept
asking  But how fast is it going absolutely, not relative to any frame? . Relativistic
physicists could not satisfy that desire for an answer about absolute speed. But that
inability was no mark against relativistic physics. Similarly, Pyrrhonians should not
be criticized for their inability to satisfy a desire for answers about who is justified
without qualification, when the whole point of their theory is to reject such
questions. If contrastivism catches on, such desires will fade.
Does this inability show that  Pyrrhonism cannot do justice to a crucial aspect of
our epistemic lives (ibid.)? No. To do justice, Pyrrhonians need to explain why
people desire unqualified epistemic assessments, but they can explain that desire by
a common tendency to presuppose that some contrast class is the relevant one.
Pyrrhonians can also explain why people believe unqualified epistemic judgements,
because those judgements are so close to qualified epistemic judgements which are
true (see MS, p. 106). If that did not count as doing justice to this aspect of our
epistemic lives, it would be hard to see how relativistic physicists could do justice to
many people s desires for assessments of absolute speed. That analogous case shows
how a theory can do justice to an aspect of our lives without endorsing the truth of
every presupposition behind that part of our lives.
This point comes out in Blaauw s own example (p. 475). Jane is justified in be-
lieving that the train station is beneath the airport as opposed to above or next to the
airport. Being modestly justified in this way is enough for Jane s practical purposes,
even if she is not extremely or unqualifiedly justified in believing that the train
station is beneath the airport. Thus, even if Jane wants to say that she is justified
without qualification, she does not need that judgement. She can get along in
the world well enough without anyone s slaking her desire for being absolutely
justified.
Blaauw s final objections are aimed at my arguments against real relevance. First
(p. 476), Blaauw worries that my arguments and my approach  can be generalized to
all sorts of philosophical arguments and positions . So? I think it is a point in favour
of Pyrrhonian contrastivism that it also solves other philosophical puzzles. Philo-
sophers often deal with reasons of various kinds: reasons to act, reasons to believe,
reasons why (explanations), and so on. If reasons are relative to contrast classes, as I
claim (MS, p. 84; see also  A Contrastivist Manifesto ), then it should be no surprise,
and certainly no objection, that my approach generalizes to other philosophical
puzzles. After all, ancient Pyrrhonism was also applied to all philosophy.
In response to my second argument, Blaauw claims  Pyrrhonists seem to be in the
same predicament as contextualists (p. 476). However, contextualists claim that
believers are justified unqualifiedly in certain contexts only if they can rule out all
other members of a certain contrast class. Thus they cannot determine whether a
given believer is justified in the way that matters to them unless they can specify
what is and is not in the relevant contrast class. Pyrrhonian contrastivists need not
carry any such burden, because they never claim that a believer is justified
unqualifiedly, so they never have to say which of various contrast classes is relevant
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
REPLIES TO HOUGH, BAUMANN AND BLAAUW 487
in order to say whether the believer is justified in the qualified ways that are all that
matter to Pyrrhonians (see MS, p. 102).
My use of the phrase  modest contrast class might create the opposite im-
pression. However, I use that label only for convenience. Nothing essential to my
view hinges on whether any alternative is or is not in the modest contrast class. For
example, is  Geoff in the modest contrast class when I wonder whether I know the
name of the waiter who speaks his name  Jeff without spelling it? It does not matter
at all to me as a Pyrrhonian. I am justified in believing that the waiter s name is  Jeff
as opposed to  Tom ,  Dick or  Harry . If you ask about  Fred and  Jules , then yes,
I also am justified in believing that the waiter s name is  Jeff as opposed to  Fred
and  Jules as well as  Tom ,  Dick or  Harry . As you add names, I add relativized
judgements, until you come to  Geoff . I am not justified in believing that the
waiter s name is  Jeff out of any set that includes  Geoff in addition to  Jeff . Once
we get all of these contrast classes distinguished and know which ones I am justified
out of and which ones I am not justified out of, then it does not matter at all which
of these contrast classes, if any, is the modest contrast class. Thus, Pyrrhonians are
not  in the same predicament as the contextualist .
In response to my third objection to contextualism, Blaauw replies (p. 476)  it will
be quite clear, in any given the context in which one asserts that someone is justified,
which alternatives should go into the modest contrast class . I grant that speakers
who call believers justified without qualification often treat a certain contrast class as
relevant. That is the point of my analysis of such unqualified judgements. Still, that
point is only about what appears relevant. Pyrrhonians can grant that contrast classes
often appear relevant to some people. That concession does not imply that any
contrast class is really relevant. The point of my arguments against real relevance is
that different contrast classes will appear relevant to different people depending on
their positions and purposes, and there is no way to get behind these perspectives to
determine which of these different contrast classes really is relevant. What might
seem clear to some is not really clear at all.
Blaauw applies his claim to my example of the waiter who speaks his name  Jeff
without spelling it. Blaauw (p. 476) asks  Is it a problem that the customer is likely to
be unable to list all non-Jeff male names? . No. The problem I raise is not about
anyone s inability to take the time to list all names. My point is that, even when
presented with certain alternatives, we often have no good reason either to exclude
them or to include them. Blaauw seems to think he has a reason, for he says  What is
the content of the modest contrast class out of which the customer is justified in
believing that the waiter s name is Jeff ? In this case, in which nothing out of the
ordinary is happening, the intuitive answer will be  all male names  (ibid.). I see no
basis for confidence here. It is not even clear to me whether Blaauw means all
possible male names (including names used only on the other side of the world) or
only local male names. How local do they have to be? Why only male names rather
than all names? Why not female names? So I find this intuition questionable.
Finally (p. 477), Blaauw argues that  Geoff is a relevant alternative in my
example. As Blaauw reports, my reply is that, if  Geoff is a relevant alternative, then
I do not know the waiter s name right after the waiter has told me his name in a
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly
488 WALTER SINNOTT-ARMSTRONG
context where I do not care at all how the waiter spells his name and where the
spelling  Jeff is much more common locally than the spelling  Geoff . In such
circumstances, it strikes me and many others as odd to deny that I know the waiter s
name. Blaauw, however, says  this seems to me to be the right result if we are talking
about how the waiter s name is spelt (p. 477). But we are not talking about how his
name is spelt. We are talking about what his name is. In my example, I do not care
how his name is spelt. Then Blaauw adds (ibid.)  I do think that we can be justified in
believing that the waiter s name sounds like  Jeff  . Again we are not talking about
how his name sounds. We are talking about what his name is. Blaauw tries to avoid
the problem by changing the question. The question is whether I know the waiter s
name. Contextualists can, of course, declare that I know or that I do not know the
waiter s name, but they cannot give an adequate reason for either answer as
opposed to the other. This is why I prefer Pyrrhonism, which rejects the question
and thereby avoids the need to give either answer.
IV. CONCLUSION
I shall close by again thanking my commentators. Moderate classy Pyrrhonian
moral scepticism is a difficult view to understand, because it requires significant con-
ceptual revision. Their comments have, I hope, enabled me to clarify some aspects
of my account and thereby to make it more comprehensible and palatable, although
many problems undoubtedly remain.
Dartmouth College, New Hampshire
© 2008 The Author Journal compilation © 2008 The Editors of The Philosophical Quarterly