Philosophical Studies (2006) 130:407–435
Springer 2006

DOI 10.1007/s11098-004-5747-3

J. BROWN

CONTEXTUALISM AND WARRANTED

ASSERTIBILITY MANOEUVRES

ABSTRACT. Contextualists such as Cohen and DeRose claim that the

truth conditions of knowledge attributions vary contextually, in particular
that the strength of epistemic position required for one to be truly ascribed
knowledge depends on features of the attributor’s context. Contextualists
support their view by appeal to our intuitions about when it’s correct (or
incorrect) to ascribe knowledge. Someone might argue that some of these
intuitions merely reflect when it is conversationally appropriate to ascribe
knowledge, not when knowledge is truly ascribed, and so try to accom-
modate these intuitions even on an invariantist view. DeRose (Blackwell
Guide to Epistemology, 1998; Philosophical Review, 2002) argues that any
such ‘warranted assertibility manoeuvre’, or ‘WAM’, against contextualism
is unlikely to succeed. Here, I argue that his objections to a WAM against
contextualism are not persuasive and offer a pragmatic account of the data
about ascriptions of knowledge.

KEY WORDS: Cohen, Contextualism, DeRose, warranted assertibility manoeuvres

Contextualists such as Cohen and DeRose claim that the truth

conditions of knowledge attributions vary contextually, in
particular that the strength of epistemic position required for
one to be truly ascribed knowledge depends on features of the
attributor’s context (‘attributor factors’). Contextualists sup-
port their view by appeal to our intuitions about when it’s
correct (or incorrect) to ascribe knowledge. For instance, in
Cohen’s airport case, Smith truly believes that the New York
flight stops in Chicago on the basis of his flight itinerary. If
Mary and John are in a low standards context in which it’s not
important to them that the flight stops there and the possibility
of error has not been raised in their conversational context, it
seems correct for them to attribute knowledge to Smith.
However, in a different high standards context in which it is
408 J. BROWN

very important to Mary and John that the flight stops there and
the possibility of error has been raised in their conversational
context, it seems correct for them to deny that Smith knows.
The case shows that the intuitive correctness of knowledge
attributions depends on attributor factors. Contextualists
conclude that the truth conditions of knowledge attributions
depend on attributor factors. Contextualists accept that subject
factors such as whether the subject’s belief is true and non-
accidentally so affect the truth of knowledge attributions, but
they argue that these factors play a different role to attributor
factors. On their view, attributor factors, such as the impor-
tance of the issue and whether error possibilities have been
raised, affect the truth conditions of knowledge attributions, in
particular how good an epistemic position the subject must be
in to be truthfully attributed knowledge. Such subject factors as
whether the subject’s belief is true and non-accidentally so af-
fect how good an epistemic position the subject is in and thus
the truth value of knowledge attributions. (DeRose, 1992;
Cohen, 1998.)
Invariantists deny that the truth conditions of knowledge
attributions depend on attributor factors. Nonetheless, they
need to accommodate the intuitions highlighted in contextualist
cases. As Bach (forthcoming (a)) points out, they may do so in
a variety of ways. One option is to argue that the problematic
intuitions arise because we are mistaken about whether the
subject knows. For instance, a sceptical invariantist may argue
that it seems correct to attribute knowledge in the low stan-
dards case because we mistakenly think that the subject knows.
A second option is to argue that the problematic intuitions
reflect when it is conversationally appropriate to ascribe
knowledge, not when knowledge is truly ascribed. On this view,
the contextualist mistakenly takes the fact that the conversa-
tional propriety of knowledge ascriptions depends on context
to show that the truth conditions of knowledge ascriptions
depend on context. Again, there are a number of different ways
to fill out this strategy, say by appeal to the idea that there are
presuppositions or conditions of the felicitous utterance
of knowledge attributions (Bach forthcoming (a)), or by using
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 409
the idea that, via familiar Gricean mechanisms, knowledge
attributions and denials pragmatically convey further claims
and it is the truth value of these further claims which the
problematic intuitions reflect. DeRose (1998, 2002) focuses on
the last of these options and argues that such a ‘warranted
assertibility manoeuvre’, or ‘WAM’, against contextualism is
unlikely to succeed. Here, I argue that his objections to a WAM
against contextualism are not persuasive and offer a pragmatic
account of the data about ascriptions of knowledge.
Note that the characterisation of contextualism and invari-
antism given so far leaves open an issue which has recently
come to the fore: whether the importance of the issue to the
subject of the knowledge attribution, and whether error has
been raised in her context, affects the truth conditions of
knowledge attributions. This issue is not important for the first
part of the paper (sections 1–5) so, for now, I leave it open
where contextualists and invariantists stand on this issue.
However, I discuss this issue in section 6 where it is central to
the argument.

1. DEROSE’S CASE AGAINST A WAM

A WAM can be applied to a wide range of cases including some

not involving the notion of knowledge. For example, consider
the expression ‘it’s possible that p’, understood as expressing
epistemic possibility. Suppose that you ask me whether a col-
league, Carolyn, is in her office. It would seem incorrect for me
to say ‘It’s possible that she is’ if I know that she’s just left to
give a lecture. It also seems incorrect for me to say ‘It’s possible
that she is’ if I know full well that she is in her office. If these
intuitions reflect truth conditions, then ‘It’s possible that p’ is
true iff (1) S doesn’t know that p is false and (2) S doesn’t know
that p is true. However, DeRose argues that at least some of
these intuitions reflect conversational propriety. In particular,
he argues that when a subject knows that p it’s true for her to
say ‘It’s possible that p’ although misleading since it pragmat-
ically conveys the falsehood that she does not know that
410 J. BROWN

p. DeRose explains this pragmatic implication by appeal to the

general rule of conversation, ‘Assert the stronger’. DeRose uses
this WAM to defend the view that ‘It’s possible that p’ is true iff
S doesn’t know that p is false. (DeRose, 1998, p. 197).
DeRose draws from his discussion of the possibility case
three criteria for a successful WAM, namely that it should start
from a conflict of intuitions which requires explanation, that it
should involve explaining away only an intuition of falsehood
but not also an intuition of truth, and that it should appeal to a
general rule of conversation. DeRose argues that the proposed
WAM against contextualism is unlikely to succeed since it fails
to meet these three criteria (1998: §10–11, 2002: §3.2). In sec-
tions 2–3, I argue that the knowledge case does meet the conflict
of intuitions test and that, although it involves explaining away
both an intuition of falsehood and an intuition of truth, this is
not problematic. In section 4, I propose a WAM against con-
textualism which exploits Grice’s cooperative principle.
DeRose uses the notion of warranted assertibility in cha-
racterising the debate between contextualists and invariantists
about whether our intuitions about knowledge ascriptions re-
flect truth conditions. On this characterisation, the invariantist
holds that the contextualist mistakes the fact that the warranted
assertibility conditions of knowledge ascriptions depend on
context for the different claim that the truth conditions of
knowledge ascriptions depend on context. DeRose’s use of the
notion of warrantedly assertibility should not lead us to mis-
takenly think that the debate concerns the notion of epistemic
warrant. The invariantist is not correctly characterised as
holding that whether an ascription of knowledge is epistemi-
cally warranted depends on context; indeed, that’s a contextu-
alist claim about the epistemic property of warrant. Rather, the
invariantist holds that whether an ascription of knowledge is
conversationally appropriate depends on context. That the
relevant notion is that of conversational propriety is also evi-
dent in DeRose’s discussion of ‘It’s possible that p’. On De-
Rose’s view, if a subject knows that p, although it is true for her
to say ‘It’s possible that p’ it is not warrantedly assertible. But,
plausibly, if a subject knows that p, then she has epistemic
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 411
warrant to assert p and so to assert the weaker claim that it is
possible that p. So, when DeRose says that, for such a subject,
‘It’s possible that p’ is not warrantedly assertible, he must mean
that it is not conversationally appropriate. Of course, the
conditions for an utterance to be conversationally appropriate
may include epistemic conditions, for instance DeRose claims
that S is warranted in asserting p only if S knows that p.
Nonetheless, the key issue between contextualists and invari-
antists is whether our intuitions about the correctness of
knowledge attributions reflect conversational propriety or truth
conditions.

2. THE NEED FOR EXPLANATION

DeRose argues that a good candidate for a WAM should start

from a set of intuitions which, plausibly, cannot all be correct
so that one or more needs explaining away. In the possibility
case we have just such a conflict:
where a speaker knows that p, while it does indeed seem wrong, and may
even seem false to some, for her to assert ‘It’s possible that pind’ it seems just
as bad—in fact worse—and certainly seems false, for her to instead say ‘It’s
impossible that pind’ or ‘It’s not possible that pind’. But, it seems quite un-
likely that both ‘It’s possible that pind’ and ‘It’s not possible that pind’ are
false, so we have good reason to believe that something is not as it seems
here. (1998: p. 198, 2002: §3.2.)

It initially seems that there is also a conflict of intuitions in the

case of knowledge. Contextualists support their position by
arguing that it can explain our conflicting intuitions about
knowledge. For instance, they point out that the sceptic’s denial
that one knows that one is not a BIV seems plausible. Com-
bining this denial with closure, it follows that one lacks
knowledge that one has hands. But, in opposition to this, it
seems highly plausible that one does know such ordinary
propositions as that one has hands. Contextualists argue that it
is an advantage of their response to scepticism that it can ex-
plain this conflict of intuitions (see, e.g., Cohen, 1988; DeRose,
1995, pp. 183–84, 208–214). Similar conflicts of intuition
412 J. BROWN

characterise other key examples used to motivate contextual-

ism. In Cohen’s airport case, Smith’s itinerary for his flight to
New York includes a stop at Chicago. Since this is just the sort
of evidence which would ordinarily be taken as providing
knowledge, we have the intuition that Smith knows that the
flight stops in Chicago. However, when we focus on the context
in which Mary and John have a pressing practical need to get to
Chicago (perhaps they are to perform a life-saving operation
there), it seems wrong to ascribe knowledge to Smith. After all,
the itinerary may contain a misprint, or the route may have
changed since it was printed (Cohen, 1999, 2000).
Although the contextualist cases seem to involve a conflict of
intuitions, DeRose argues that there is no suitable conflict of
intuitions to motivate a WAM against contextualism. DeRose
argues that the contextualist cases involve a shift from a low to
a high standards context such that, in each context, our intu-
itions are not in conflict but in harmony:
In the ‘low standards’ contexts, it seems appropriate and it seems true to say
that certain subjects know and it would seem wrong and false to deny they
know, while in the ‘high standards’ contexts, it seems appropriate and true
to deny that similarly situated subjects know and it seems inappropriate and
false to say they do know. (1998, p. 201, 2002: §3.2.)

For instance, DeRose would hold that the airport case involves
a shift from a low to a high standards context. In the low
standards context, it is intuitive to say Smith knows and
counterintuitive to deny that he knows; in the high standards
context, it is intuitive to deny that he knows and counterintu-
itive to claim that he knows.
However, DeRose’s attempt to avoid a conflict of intuitions
is unconvincing. DeRose is considering an opponent who ac-
cepts contextualist claims about when attributions and denials
of knowledge seem correct, but offers an explanation in terms
of the notion of whether an utterance is conversationally
appropriate. So, this opponent accepts that, say, if one focuses
only on the itinerary information, it seems correct to attribute
knowledge to Smith and incorrect to deny it; and, that if one
focuses only on the practical importance of the stop to Mary
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 413
and John, it seems correct to deny that Smith knows and
incorrect to attribute knowledge. Despite this, both foci are
available to us and, prima facie, the claims made from these
different foci cannot all be true. DeRose could try and remove
the apparent inconsistency by arguing that ‘knows’ expresses
different properties in the low and high standards contexts. On
that view, we would no longer have conflicting intuitions about
whether to ascribe a single property. Rather, we would be
ascribing one property to the subject but denying that she has a
different property and there is no contradiction in that. But, of
course, the claim that context affects which property is ascribed
by attributions of knowledge is a contextualist claim which
would be rejected by invariantists. Thus, to appeal to this idea
would be to beg the question against the invariantist view.

3. EXPLAINING THE CONFLICT

DeRose’s second objection to a successful WAM against con-

textualism concerns the type of explanation such a WAM
would offer of our intuitions. He says that, in the case of a good
candidate for a WAM, the proposed WAM would explain
away an intuition of incorrectness by appeal to a generation of
a false implicature. For instance, the proposed WAM in the
possibility case explains why it seems incorrect for a subject
who knows that p to say, ‘It is possible that p’, by arguing that
although the assertion is literally true, it generates a false im-
plicature to the effect that the subject does not know that p.
However, DeRose argues that the proposed WAM against
contextualism has to explain away not only an intuition of
falsity but also an intuition of truth:
whichever set of appearances the invariantist seeks to discredit—whether
she says we are mistaken about the ‘high’ or ‘low’ contexts—she’ll have to
explain away both an appearance of falsity and (much more problemati-
cally) an appearance of truth (1998: p. 201; see also 2002: §3.2).

For instance, if she seeks to explain away the high standards

intuitions, she needs to explain away both the intuition that it is
correct to deny that the subject has knowledge and that it is
414 J. BROWN

incorrect to ascribe knowledge to the subject. If she seeks to

explain away the low standards intuitions, she needs to explain
away both the intuition that it is correct to ascribe knowledge
to the subject and that it is incorrect to deny that the subject has
knowledge. But, DeRose argues that a fundamental difficulty
faces any attempt to explain away an intuition of truth rather
than falsity. In the case of an apparently false claim, one can
argue that it is literally true although it seems false since it
pragmatically conveys a falsehood. In the case of an apparently
true claim, one could argue that it is literally false but seems
true because it pragmatically conveys a truth. But DeRose ar-
gues that this suggested explanation is unlikely to succeed:

Even if you can come up with a good explanation for why the assertion
would generate some true implicature, this wouldn’t seem to help much. For
don’t we want to avoid falsehood both in what we implicate and (especially!)
in what we actually say? So, it would seem that it would be unwarranted to
assert a falsehood, even if doing so generates a true implicature (1998: p.
200, 2002: §3.2).

To be plausible, it seems that DeRose’s argument should be re-

stricted to non-figurative uses of language. In a variety of figu-
rative uses, e.g. irony, metaphor and hyperbole, a literally false
utterance may seem apt because it pragmatically conveys a truth.
Indeed, in introducing the notion of a conversational implicature,
Grice applied it to precisely such examples. Still, DeRose may say
that the knowledge attributions which are his focus plausibly in-
volve non-figurative uses of language. Is DeRose’s argument
plausible when restricted to non-figurative uses?
It may seem so if we interpret the notion of warranted as-
sertibility as the notion of epistemic warrant. For, a subject is
not epistemically warranted in asserting p if she takes p to be
false or, on factive accounts of epistemic warrant, if p is false.
However, as we saw earlier, the relevant notion is not that of
epistemic warrant but rather conversational propriety. So our
question is whether, for non-figurative uses of language, one
can explain away an intuition of correctness by saying
that although the relevant utterance is false it seems
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 415
conversationally correct since it pragmatically conveys a truth.
Suppose that a false utterance pragmatically conveys a truth. If
speakers concentrate on what the utterance pragmatically
conveys rather than on what’s literally said, or if speakers
mistake what the statement pragmatically conveys for what it
literally says, then the utterance will seem correct even though it
is literally false. Indeed, we will see that there is a wide range of
non-figurative uses of language where one standard approach
involves supposing that a literally false statement seems correct
because it pragmatically conveys a truth. The existence of such
cases is a corollary of the existence of cases in which a true
statement seems incorrect since it pragmatically conveys a
falsehood. It is no part of the brief of this paper to determine
whether this approach is the right one to take about the rele-
vant cases. Rather, the point is to show that DeRose’s dismissal
of the possibility of a successful WAM against contextualism is
too quick. For what he sees as a serious problem for such a
WAM – that it would involve arguing that a false utterance can
seem correct because it pragmatically conveys a truth – is part
of a standard and well-entrenched approach in the philosophy
of language.

3.1. Definite Descriptions

On the Russellian approach, the semantic value of a definite

description of the form ‘the F’ is equivalent to the semantic value
of the expression ‘the unique F’. Thus, for instance, the semantic
content of (1) is equivalent to the semantic content of (2):
(1) The man in the corner drinking a martini is a lawyer.
(2) There is a unique man in the corner drinking a martini and
he is a lawyer.
Donnellan (1966) challenged this treatment by arguing that
definite descriptions may be used ‘referentially’ to refer to a
salient object regardless of whether that object uniquely fits the
description. For instance, (1) could be used to refer to a salient
man regardless of whether he is the unique man in the corner
416 J. BROWN

drinking a martini. Further, the truth value of the utterance

intuitively turns on the properties of this salient individual, even
if there is some other man who uniquely fits the description.
A Russellian may accommodate the referential use of defi-
nite descriptions by arguing that even though the expression
‘the man in the corner drinking a martini’ literally denotes the
individual who uniquely fits this description, the expression
may be used to pragmatically convey information about a
salient individual regardless of whether he fits the description
(e.g., Kripke, 1977; Bach, 1984; and forthcoming, (b)). Suppose
that the salient man is not a lawyer, but the individual who
uniquely fits the description is a lawyer. In this case, an utter-
ance of (1) is literally true but pragmatically conveys the false
claim that the salient individual is a lawyer. As a result, al-
though the utterance is literally true it seems incorrect. Further,
and more importantly for our purposes, this pragmatic defence
of the Russellian view seems committed to the existence of cases
in which an utterance is literally false but seems correct since it
pragmatically conveys a truth. Consider a different party at
which the salient person is not drinking a martini and is a
lawyer, but the man who uniquely fits the relevant description is
not a lawyer. On the proposed pragmatic account, in this sec-
ond scenario, an utterance of (1) is literally false but conveys
the truth that the salient man is a lawyer and, as a result, seems
intuitively correct.

3.2. Impliciture

There seem to be numerous cases in which what a speaker

means goes beyond what she literally says, where what’s said is
determined compositionally by the semantic contents of the
constituents of the sentence uttered as a function of their syn-
tactic relationship (this compositional determination is relative
to disambiguation and an assignment of referents to indexicals,
etc) (Sperber and Wilson, 1986; Bach, 1994). For instance, in
normal circumstances, a speaker uttering (3) would mean, and
be taken to mean, that she has eaten recently:
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 417
(3) I have eaten.
Thus, her utterance would seem correct if and only if she has
eaten recently. However, there is no obvious component of the
utterance which corresponds to the qualification ‘recently’
(though see Stanley and Szabo, 2000). Such examples are
ubiquitous. In a suitable context, the following utterances could
be used to mean the expanded claim indicated by the bracketed
material: ‘It’s raining’ [here], ‘Everyone got an A’ [everyone on
the course], ‘Howard is too old’ [to be leader of the Tory party], etc.
One way to treat such examples is to take our intuitions
about the correctness and incorrectness of the relevant utter-
ances as indicating the truth conditions, and thus the content,
of what’s said. On this view, what’s literally said by an utter-
ance of (3) in normal circumstances is that the speaker has
eaten recently. As a result, what’s said by an utterance is not
wholly determined by the constituents of the sentence uttered
together with its syntax, relative to a given disambiguation and
assignment of referents to indexicals, etc. Rather, what is said is
partly determined by the extralinguistic context (e.g., Sperber
and Wilson, 1986; Carston, 1988).
However, Bach defends the alternative view that what’s said
is a function of the semantic content of the constituents of the
sentence uttered as a function of their syntactic relationship,
given disambiguation and assignment of referents to indexicals.
On this view, what is said by an utterance of (3) is that the
speaker has eaten at some time or other. This view may seem to
conflict with our intuition that, in ordinary circumstances, an
utterance of (3) is correct if and only if the speaker has eaten
recently. However, the proposed view explains these intuitions
by claiming that, in ordinary circumstances, an utterance of (3)
pragmatically conveys that the speaker has eaten recently.
(Bach, 1994, 2001; Borg, forthcoming.)
Bach coins the term ‘impliciture’ to cover such examples. In
both impliciture and Gricean implicature, the speaker exploits
Grice’s cooperative principle to pragmatically convey some-
thing other than what she literally says. In Gricean implicature,
the implicatum is conceptually independent of what’s said, and
418 J. BROWN

may have no constituents in common with what’s said. For

instance, by saying ‘It’s raining’, one might implicate that one
won’t be playing tennis that day. However, in impliciture
what’s meant is ‘built up from the explicit content of the
utterance by conceptual strengthening… which yields what
would have been made explicit if the appropriate lexical
material had been included in the utterance’ (Bach, 2001). For
instance, a likely impliciture of an utterance of ‘It’s raining’ is
that it’s raining here.
What’s important for our purposes is that Bach’s view not only
creates cases in which a literally true utterance seems incorrect
since it pragmatically conveys a falsehood but also cases where a
literally false utterance seems correct since it pragmatically con-
veys a truth. Suppose with Bach that an utterance of ‘I’ve eaten’
literally says that the speaker has eaten at some time or other in
the past but, in normal circumstances, pragmatically conveys
that the speaker has eaten recently. On this view, if I have not
eaten recently, my utterance of ‘I’ve eaten’ is literally true since I
have eaten at many points in the past, but seems incorrect since it
pragmatically conveys the falsehood that I have eaten recently. In
the same circumstances, my utterance of ‘I’ve not eaten’ is liter-
ally false since I have eaten on many occasions in the past, but
seems correct since it pragmatically conveys the truth that I have
not eaten recently. Bach explicitly considers the objection that,
on his view, the literal truth conditions of an utterance may di-
verge from the truth conditions intuitively ascribed; an utterance
treated by his account as literally true may seem false and an
utterance treated by his account as literally false may seem true.
He dismisses this objection, treating both types of case alike. He
argues that, in cases of impliciture, we focus on what’s prag-
matically conveyed, not what’s said, and it is the truth value of
what is pragmatically conveyed by an utterance which our intu-
itions of truth value reflect. (For similar comments, see Borg,
forthcoming.)
We have seen that it is part of a common strategy in the
philosophy of language to argue that the intuitive truth value of
an utterance may reflect not its literal truth value but rather the
truth value of what it pragmatically conveys and, in particular,
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 419
that an utterance may seem true although it is literally false
since it pragmatically conveys a truth. Given this, DeRose fails
to undermine the proposed WAM against contextualism by
arguing that it involves explaining away intuitions of truth as
well as intuitions of falsity.

4. A WAM AGAINST CONTEXTUALISM

According to DeRose’s third objection to a WAM against

contextualism, no one has yet offered such a WAM which ap-
peals to a general rule of conversation. However, Rysiew (2001)
offers a WAM against contextualism which appeals to the
Gricean rule that participants in a conversation should make
their contribution relevant. Rysiew’s WAM makes use of a
notion employed in contextualist accounts, the notion of
strength of epistemic position. According to contextualism, the
strength of epistemic position required for a subject to count as
knowing depends on context. On Rysiew’s proposed invarian-
tist account, there is a context-invariant level of strength of
epistemic position required for a knowledge attribution to be
literally true. However, Rysiew argues that an attribution of
knowledge may sometimes pragmatically convey that the sub-
ject has a stronger epistemic position. Rysiew uses this frame-
work to explain away contextualist cases. On his moderate
invariantist view, in both the easy and tough contexts of the
contextualist’s cases, the attribution of knowledge is literally
true. However, in tough, the knowledge ascription pragmati-
cally conveys that the subject is in a stronger epistemic position
than she actually is and thus seems incorrect.
Contextualists offer various different accounts of the notion
of strength of epistemic position. Some contextualists explain
the notion in terms of the range of alternatives which the
subject can rule out (e.g. Cohen, 1988; Lewis, 1996). By con-
trast, DeRose (1995) explains the notion in terms of the range
of possible worlds across which the subject’s beliefs track the
truth. Rysiew explains the notion of strength of epistemic
position in terms of the notion of ruling out alternatives,
420 J. BROWN

however he points out that this is inessential to his account and

invites others to substitute their own preferred explanation (pp.
487–488).
On his proposal, a subject knows that p iff she can rule out
the relevant alternatives to p, i.e. those alternatives which ‘we
(normal) humans take to be the likely counter-possibilities to
what the subject is said to know’ (p. 488). However, even an
irrelevant alternative may be salient in a conversation, say be-
cause it has been mentioned (p. 488). In such a context, Rysiew
claims that an attribution of knowledge pragmatically conveys
that the subject can rule out the salient alternative(s). Consider
DeRose’s bank case in which DeRose and his wife are dis-
cussing whether to queue to deposit their pay cheques in the
bank today or wait till the next day, a Saturday. DeRose ini-
tially claims to know that the bank will be open on Saturday on
the basis that the bank was open when he was there just two
weeks ago on Saturday. However, after his wife reminds him of
the importance of paying in the cheques before Monday and
points out that banks do sometimes change their hours, De-
Rose says ‘Ok, I don’t know the bank will be open. I better
check’. Rysiew argues that, throughout the conversation, De-
Rose’s claim that he knows that the bank will be open on
Saturday is true for DeRose can rule out the relevant alterna-
tives. However, once DeRose’s wife makes salient the irrelevant
error possibility that the bank has changed its hours, an
assertion of knowledge by DeRose would pragmatically imply
that DeRose can rule out this possibility. Since DeRose cannot
do so he no longer claims to know (though the claim would be
literally correct). Instead, he makes the false assertion that he
does not know, pragmatically conveying the true claim that he
cannot rule out the possibility that the bank has changed its
hours (p. 490). (Contextualist cases provide better support for
contextualism when formulated in the third, rather than the
first, person. However, here I follow Rysiew’s discussion of
DeRose’s original first person formulation of the case.)
As DeRose (2002) points out, a good WAM against con-
textualism should show how the supposed pragmatic implica-
tions of knowledge ascriptions follow from the semantics of
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 421
‘knows’ plus general conversational principles, rather than
introducing a second pragmatic meaning for ‘knows’ which is
unrelated to its semantic meaning. Rysiew’s attempt to fulfil
this condition of adequacy occurs in the following discussion of
the bank case:
Witness, then, how natural it would be for the speaker’s audience to reason
as follows: He has just said ‘I [guess I] don’t know that the bank’ll be open
tomorrow.’ And he has said this in response to my raising a doubt as to
whether he can really be so sure—after all banks, as I’ve just reminded him,
do change their hours. Presumably (on the assumption that he’s conforming
to the CP [Grice’s cooperative principle]), he wouldn’t have said what he has
unless he thought that there were possibilities incompatible with the bank’s
being open tomorrow—specifically, that it has recently changed its
hours—that he could not rule out. (p. 491).

According to Grice’s cooperative principle, one should

‘make [one’s] conversational contribution such as is required,
at the stage at which it occurs, by the accepted purpose or
direction of the talk exchange in which [one is] engaged’
(Grice, 1975). Rysiew’s general idea seems to be that it follows
from this injunction that DeRose should say something that
immediately addresses the point his wife has just made,
namely that the bank may change its hours, and that his
utterance will be interpreted in that light. While this idea
seems plausible, it would be nice to have it fleshed out in some
more detail. In particular, how does DeRose’s wife’s men-
tioning an error change the direction or purpose of the con-
versation in such a way that an assertion of knowledge
pragmatically conveys that the speaker can rule out that error,
and a denial of knowledge pragmatically conveys that the
speaker cannot rule out that error? I will attempt to fill in the
details of how the pragmatic implications of knowledge
attributions are generated in the course of developing Ry-
siew’s account to deal with a rather different objection,
namely that it cannot, as he hopes, accommodate the way in
which practical importance affects knowledge attributions.
Rysiew suggests that the practical importance of an issue
affects knowledge attributions by making the possibility of
422 J. BROWN

error salient.1 For instance, he says that, in the bank case, ‘as
soon as [DeRose’s] wife has brought to salience the importance
of their depositing their paychecks before Monday morning – if
they don’t, some ‘large and important’ checks they’ve written
will bounce, etc. – it’s going to be natural for them to consider
some hitherto unconsidered possibilities (p. 490)’.2 However,
one cannot explain the full impact of practical importance on
knowledge attributions in this way. The classic cases which
Cohen and DeRose use to motivate contextualism, the airport
and bank cases, involve both error being raised and practical
significance (Cohen, 1999, pp.58–59; DeRose, 1992, p. 913). If
the practical importance of the relevant issue were to affect
knowledge attributions only by making error salient, why do
these two key cases not only involve error being raised but also
the practical importance of the issue?
Further, in many cases, the impact of error possibilities on
knowledge attributions seems to be affected by whether the
issue in question is also practically important. As we have seen,
mentioning an error possibility tends to undermine the relevant
knowledge attributions, where for now it’s left open whether
that’s because mentioning an error tends to make the relevant
knowledge claims false or merely unwarranted. Contextualists
accept that after an error possibility has been mentioned, its
knowledge-undermining tendency may be resisted, say by
questioning the likelihood of that error.3 I will argue that the
likelihood and success of this kind of attempt to resist the
knowledge-undermining tendency of a mentioned error is af-
fected by whether the issue in question is practically important.
Consider a variant of Cohen’s airport case in which Mary and
John don’t figure. Instead, Smith is talking to his travelling
companion Jones. As they both know, it’s unimportant to them
whether the flight stops in Chicago. In response to Jones’ idle
inquiry, (‘Out of curiosity, does the flight stop in Chicago?’)
Smith says ‘Yes, I know it stops there – look my itinerary says
so’. Suppose that Jones now mentions the possibility that the
itinerary contains a misprint. Mentioning this error possibility
tends to undermine the relevant knowledge attribution. How-
ever, when it’s unimportant that the flight stops in Chicago, the
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 423
knowledge-undermining tendency of mentioning this error
possibility may well be resisted. For instance, Smith may reply,
‘Well how likely is that?’ Now consider a different situation in
which it is very important to Smith and Jones that the flight
stops in Chicago, say because they are heart surgeons due to
perform an emergency operation there. Suppose as before that
after Smith points out that his itinerary contains a Chicago stop,
Jones mentions the possibility of itinerary error. In principle,
Smith could try to resist the knowledge-undermining tendency
of the mentioned error possibility by saying, ‘How likely is
that?’. However, in a situation in which it is very important that
Smith and Jones get to Chicago, this attempt to resist Jones’s
knowledge-undermining manoeuvre seems much less successful.
Given the stakes, it is important that Smith and Jones consider
even unlikely error possibilities. As a result, it is harder to resist
Jones’ knowledge-undermining manoeuvre merely by pointing
out that it’s unlikely that the itinerary is wrong.
It seems, then, that practical importance affects the likely
success of an attempt to resist the knowledge-undermining
tendency of mentioned error and, thus, how likely it is that such
an attempt will be made. To say this is not to say that if the
issue is not important, raising error never undermines knowl-
edge. When the relevant issue is not practically important, the
knowledge-undermining tendency of a mentioned error need
not be resisted and, even if it is resisted, further knowledge-
undermining manoeuvres may be made. (E.g. after Smith has
pointed out that itinerary error is pretty unlikely, Jones may
make a further knowledge-undermining manoeuvre, say by
asking ‘Do you know that the itinerary doesn’t contain a
mistake?’). Further, when the issue is practically important,
although it is important to consider some unlikely error pos-
sibilities, it doesn’t follow that it’s important to consider every
possibility of error no matter how unlikely. Despite these
caveats, a distinction remains. Considered in isolation from any
further conversational manoeuvres, pointing out that the rele-
vant error possibility is unlikely is in general more powerful in
resisting the knowledge-undermining tendency of a mentioned
error when the issue is not practically important as compared to
424 J. BROWN

when it is practically important. The reason is simple: when the

issue is practically important we should consider some unlikely
error possibilities (though not all).
Given that the dialectical impact of a mentioned error pos-
sibility is affected by whether the issue is practically important,
I suggest that one cannot, as Rysiew hopes, capture the impact
of practical importance on knowledge attributions by suppos-
ing it functions merely to make error salient. I will develop
Rysiew’s account both to better capture the impact of practical
importance on knowledge attributions, and to further clarify
how the pragmatic implications of knowledge attributions are
generated by Grice’s rule of relevance. In doing so, I avoid
Rysiew’s explication of the notion of strength of epistemic
position in terms of the notion of ruling out alternatives and,
instead, use the explication central to one of the most detailed
current contextualist views, that offered by Keith DeRose.
DeRose’s explicates the notion of strength of epistemic po-
sition in terms of the range of possible worlds across which
one’s belief matches the facts. He claims that ‘[A]n important
component of being in a strong epistemic position with respect
to p is to have one’s belief as to whether p match the fact of the
matter as to whether p is true, not only in the actual world, but
also at the worlds sufficiently close to the actual world’,
(DeRose, 1995, p. 204). For DeRose, context determines how
strong an epistemic position one must be in to count as
knowing, that is it determines the size of the ‘sphere of possible
worlds, centered on the actual world, within which a subject’s
belief as to whether p is true must match the fact of the matter
in order for the subject to count as knowing’ (p. 206). As the
standards rise the size of this sphere of ‘epistemically relevant
worlds’ increases. Although DeRose uses his account of the
notion of strength of epistemic position to develop his
contextualist position, it could instead be used as part of a non-
sceptical invariantist account. According to the suggested in-
variantist view, there is a context-invariant range of possible
worlds across which the subject’s belief must match the facts in
order to constitute knowledge, although in some contexts, via
Grice’s rule of relevance, an attribution of knowledge may
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 425
pragmatically convey that the subject’s belief matches the facts
across a wider range of possible worlds.
As background to the account, consider one of Grice’s
examples of how the rule of relevance works. Suppose that a
man is standing beside a car in obvious need of repair and asks
as passer-by, ‘Is there a garage nearby?’ In this situation what’s
conversationally relevant is not merely whether there is a gar-
age nearby, but whether it’s open. As a result Grice argues that
although the response ‘Yes, there’s one round the corner’ lit-
erally states only that there is a garage there, it would prag-
matically convey that the garage is open (1975, p. 66). For, if
the passer-by did not think that it is possible that the garage is
open then her utterance would infringe the maxim, ‘Be Rele-
vant’. Thus, the passer-by’s utterance would be intuitively
incorrect if she knows that the garage is shut. Rather, in such a
scenario, and assuming that she believes that there is no other
nearby open garage, it would be correct for her to reply ‘No,
there’s no garage nearby’. While this utterance is literally false,
it pragmatically conveys the true claim that there is no open
garage nearby.
Using Grice’s rule of relevance we can explain the contex-
tualist intuitions about the bank case. In low, it seems appro-
priate for DeRose to self-attribute knowledge that the bank is
open on Saturday, although in high, it seems inappropriate for
DeRose to self-attribute knowledge. DeRose explains the case
by claiming that, between low and high, there is an increase in
the strength of epistemic position required for an attribution of
knowledge to be true, i.e. an increase in the range of worlds
over which the subject’s belief must match the facts in order to
count as knowledge. As a result, DeRose’s self-attribution is
literally true in low, but would be false in high. The proposed
non-sceptical invariantist account holds that, in both low and
high, there is a context-invariant level of strength of epistemic
position required for DeRose to know, i.e. there is a context-
invariant range of worlds across which DeRose’s belief must
match the facts in order to constitute knowledge. On this view,
DeRose’s self-attribution is true in both low and high. How-
ever, the invariantist argues that in high, an attribution of
426 J. BROWN

knowledge would pragmatically convey the falsehood that

DeRose is in a stronger epistemic position, that his belief
matches the facts across a wider range of worlds.
The invariantist can explain how this pragmatic implication
is generated by appeal to Grice’s rule of relevance. Recall that,
in the bank case, DeRose’s wife points out that it’s practically
very important that they deposit the cheques before Monday
(or there’ll be big trouble with the bank). By doing so, she
makes it clear that what’s relevant to the conversation is a very
strong epistemic position; in other words what’s relevant to the
conversation is whether DeRose’s belief matches the facts not
only at the actual and nearby worlds, but also at some further
away worlds which are ordinarily too far away to undermine
knowledge. Specifically, given that DeRose’s wife mentions the
possibility that the bank has changed its hours since DeRose’s
last visit, what’s relevant is whether DeRose’s belief matches
the facts out to the possible world in which the bank has
changed its hours since DeRose’s last visit. As a result, De-
Rose’s assertion ‘I know that the bank is open on Saturday’
pragmatically conveys that he is in a very strong epistemic
position with respect to that proposition, that his belief matches
the facts out to the world in which the bank has changed its
hours since the last visit. For, if DeRose did not think that he is
in such a position, then his utterance would break Grice’s
maxim of relevance. Further, given the context, DeRose’s
assertion ‘Ok, I don’t know that the bank is open on Saturday’
pragmatically conveys, though it does not literally state, that
DeRose is not in a very strong epistemic position with respect
to the relevant proposition, that his belief does not match the
facts out to the world in which the bank has changed its hours.
The account can be used to fill out Rysiew’s suggestion that
raising the possibility of error may change the pragmatic
implications of knowledge attributions. Consider first a case in
which error possibilities are mentioned in a context in which the
issue is practically important. As we have seen, in such a con-
text, a knowledge attribution pragmatically conveys that the
subject has a strong epistemic position, that her belief matches
the facts across a range of worlds including worlds ordinarily
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 427
regarded as too far away to undermine knowledge. If, within
this context, some specific error scenario is raised, then a
knowledge attribution also pragmatically conveys that the
subject’s belief would match the facts even in this error sce-
nario. This is what is happening in the bank and airport cases.
Next consider a case in which an error scenario is raised
although the issue is not practically important. We have seen
that raising an error possibility in such a context has a different
dialectical impact than when the issue is practically important;
in particular, it seems that the knowledge-undermining force of
a mentioned error may be more easily resisted by pointing out
that the error is unlikely. Nevertheless, raising an error possi-
bility even when the issue is not practically important does tend
to undermine knowledge attributions. We can explain this by a
mechanism similar to that used in explaining why practical
importance affects knowledge attributions.
When a subject mentions an error possibility in a situation in
which the issue is not practically important, she suggests indi-
rectly that what is relevant to her interests is a very strong
epistemic position with respect to the relevant proposition.
Consider a version of the bank case in which it’s not practically
important that the bank is open on Saturday. After DeRose
claims to know the bank is open, his wife mentions that banks
do sometimes change their hours. On the proposed invariantist
account, the literal truth of DeRose’s claim to know that the
bank is open does not require that DeRose’s belief match the
facts in the possible scenario in which the bank has changed its
hours since the last visit. Why, then, does DeRose’s wife raise
that possibility? Plausibly, she is not merely interested in whe-
ther DeRose knows, but whether he has a stronger epistemic
position. Thus, after she has mentioned the change in hours
scenario, a knowledge attribution would pragmatically convey
that the subject’s beliefs match the facts in a larger range of
worlds including the possible world in which the bank’s hours
have changed since the last visit.4 However, the pragmatic
implication is rather weaker than in a case in which a subject
has made it clear that she regards the issue as practically
important. For, raising error without making it clear that the
428 J. BROWN

issue is practically important is a more indirect and weaker way

of making a strong epistemic position relevant. As a result,
when a subject only mentions the possibility of unlikely error,
other participants may reply by making the knowledge claim
while simultaneously suggesting that the mentioned error sce-
nario is too far away to undermine knowledge.

5. TWO OBJECTIONS CONSIDERED

Before leaving the details of this pragmatic account, I will

consider two objections. According to the first, a standard test
of whether an utterance pragmatically implies some claim is
whether one can ‘cancel’ the implication (Grice, 1989, p. 44).
However, it is said, the relevant pragmatic implication cannot
be cancelled. On the proposed account ‘S knows that p’ may
pragmatically convey that S is in a very strong epistemic posi-
tion, that her belief matches the facts across a wide range of
worlds, including some so far away they are not normally taken
to undermine knowledge. Contrary to the objection, it may be
possible to cancel this implication. To my ear, the claim that ‘S
knows that p, but S is not in a really strong epistemic position
with respect to p’ or ‘S knows that p, but her belief wouldn’t
match the facts in a really distant possible world’ do not seem
obviously inconsistent or uncomfortable.5 Although intuitions
about this proposed cancellation may well differ, the defence of
the pragmatic account does not turn on whether there are cir-
cumstances in which the relevant cancellation can be comfort-
ably made. Grice himself accepted that some implicatures
cannot be ‘comfortably’ cancelled (1989, p. 46). Indeed, we
should expect an uncomfortable cancellation in a case where
speakers tend to confuse what’s literally said by an utterance
with what it pragmatically conveys.
A second, distinct objection to the pragmatic account ab-
stracts from the details of the account, and argues that any
pragmatic account cannot hope to explain the power of the
sceptical argument when merely considered in thought, rather
than uttered in conversation. Even a successful pragmatic ac-
count offers an explanation only of a subject’s making, or
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 429
refraining from making, a verbal attribution of knowledge. In
response, note first that the same objection may be levelled at
some contextualist accounts. For instance, it’s unclear how the
power of the sceptical argument in thought is explained by
DeRose’s rule of sensitivity: ‘When it is asserted that some
subject S knows (or does not know) some proposition p, the
standards for knowledge (the standards for how good an epi-
stemic position one must be in to count as knowing) tend to be
raised, if need be, to such a level as to require S’s belief in that
particular p to be sensitive for it to count as knowledge’ (1995,
p. 206). DeRose himself accepts that his account needs to be
extended to cover scepticism in thought. DeRose provides
several suggestions, of which the following could also be ap-
plied to extend the pragmatic account: ‘our judgement
regarding whether something can or cannot be truly asserted
(under appropriate conditions) might be held to affect our
judgement regarding whether it’s true or false, even when we
make this judgement in solitude, with nothing being said at all’
(1995, p. 187). Second, even if the pragmatic account were to
fail to explain the power of the sceptical argument in thought it
would still undermine the contextualist account. As DeRose
and other contextualists admit, contextualism would seem ad
hoc if it were required only to solve the sceptical problem. If
contextualism were to invoke just two standards for the truth of
knowledge attributions, the sceptical, introduced in philo-
sophical discussions of scepticism, and the ordinary, used in all
other contexts, then contextualism fails to make it clear why we
are taken in by the sceptic’s argument, rather than recognising
that she is merely raising the standards for the truth of
knowledge attributions (1998, p. 195). But if we anyway need to
posit a variety of standards and a mechanism for raising
standards even in ordinary, nonphilosophical conversations,
then the contextualist diagnosis of scepticism is much more
plausible. So, it’s an important part of the defence of contex-
tualism that it is required to explain not only the power of the
sceptical argument, but also our attributions of knowledge in
everyday nonphilosophical contexts. If a pragmatic account
430 J. BROWN

could explain these everyday attributions, the force of the

contextualist’s case would be much undermined.

6. WEATHERSON’S CASES

We have seen how an invariantist can use a WAM to explain

away third person contextualist cases. In this last section, I
show how the suggested WAM can also deal with another
important type of case discussed in the literature, due to Brian
Weatherson.6 Weatherson’s cases are distinct from standard
contextualist cases in that they suggest that the intuitive cor-
rectness of an attribution of knowledge to a subject may depend
on whether, in the subject’s context, the issue is important and
error has been raised. Consider a variant of DeRose’s bank case
which is the same as the original except that, after his wife has
pointed out how important it is that the cheque is paid in before
Monday and has raised the possibility of error, DeRose
nonetheless decides to leave paying in the cheque until Saturday
because of the long queues. Further, the bank is open on Sat-
urday morning and DeRose deposits the cheque then. On
Saturday evening, the following conversation takes place:
Friend: Why didn’t you deposit the pay cheque Friday afternoon if it was so
important?

DeRose: Because I knew the bank would be open on Saturday.7

As Weatherson points out, DeRose’s claim to know seems just

as unacceptable on Saturday evening after the cheque has been
paid in as it would have been on Friday afternoon. In other
words, DeRose’s self-attribution seems incorrect although, in
the context of attribution, the stakes are low and error is not
salient.
Weatherson used the case to argue against contextualism
and in favour of one kind of invariantism which he calls ‘sub-
ject sensitive invariantism’, or SSI for short. This position de-
nies that the truth value of an attribution of knowledge depends
on attributor factors but allows that it is affected by the
importance of the issue to the subject, and whether error has
been mentioned in the subject’s conversational context [for
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 431
examples of this kind of view see Hawthorne (2004) and Stanley
(forthcoming)]. Against Weatherson, DeRose argues that the
contextualist can plausibly explain Weatherson’s case. The core
contextualist idea is that the epistemic standards for an attri-
bution of knowledge depend on the attributor’s context which
includes the direction or purpose of the attributor’s conversa-
tion. Although the purpose of the attributor’s conversational
context often determines that the truth conditions of knowledge
attributions are fixed by the attributor’s context, it is compat-
ible with contextualism that they should sometimes by fixed by
whether, in the subject’s context, the issue is important and the
possibility of error has been raised (DeRose, 2004b).8
It is important to see that the proposed pragmatic account
can dealt with Weatherson’s case, even though it is conceived
as a ‘classic invariantist view’, which holds that the truth value
of an attribution of knowledge depends on neither the
attributor’s context, nor on whether, in the subject’s context,
the issue is important and error has been raised. On the
proposed account, in both contexts of the original bank case,
DeRose knows that the bank will be open but, in the high
context, the attribution seems incorrect since it conveys the
falsehood that DeRose is in a very strong epistemic position.
This account can easily be extended to deal with Weatherson’s
variant. The Saturday evening conversation starts with De-
Rose’s friend’s query ‘Why didn’t you deposit the pay cheque
Friday afternoon if it was so important?’ This query makes it
clear that what’s relevant to the Saturday evening conversa-
tion is whether DeRose’s Friday action (not paying in the
cheque) was appropriate in the light of how high the stakes
were to him then. As a result, what’s relevant to the conver-
sation is whether, on Friday, DeRose had a very strong
epistemic position with respect to the claim that the bank
would be open on Saturday. Although the attribution ‘I knew
the bank would be open’ is literally true, in the context it
pragmatically conveys the falsehood that DeRose was in a
very strong epistemic position. As a result, the attribution
seems incorrect even though it is literally true.
432 J. BROWN

7. CONCLUSION

DeRose thinks that the prospects of a pragmatic explanation of

contextualist data are very poor. However, his arguments are
not convincing. The standard cases used to support contextu-
alism are characterised by a conflict of intuitions ripe for a
pragmatic explanation. Such an explanation would involve
explaining away both intuitions of incorrectness and intuitions
of correctness. However, there is no reason to suppose that the
latter task is more problematic than the former. Rather, prag-
matic treatments standardly suppose that a literally false
statement may seem correct because it pragmatically conveys a
truth. I have developed Rysiew’s pragmatic account of the
contextualist data which appeals to the Gricean notion of rel-
evance, stressing practical importance as well as the salience of
error. We have seen that this pragmatic account can naturally
deal with both third-person contextualist cases in which our
intuitions about the correctness of knowledge attributions are
affected by attributor factors, and Weatherson’s cases in which
our intuitions about the correctness of knowledge attributions
are affected by subject factors.9

NOTES
1
See also Cohen: ‘In the case of Mary and John it was the importance of
arriving on time that made the chance of error salient’ (2000, p. 98).
2
Rysiew says that the knowledge claim conveys the further implicature that
DeRose’s epistemic position is strong enough to warrant putting off
depositing the cheques til the following day (p. 490). However, this doesn’t
answer the objection below that Rysiew’s account makes it puzzling why
raising irrelevant error alone has a different impact on knowledge than
mentioning irrelevant error in a context in which the issue is practically very
important. For, on Rysiew’s view, in both cases the attribution of knowl-
edge is literally true but carries a false implicature. The only difference is
that, in the first case, the attribution implicates the false claim that the
subject can rule out the salient error possibilities whereas, in the second, the
attribution also implicates the false claim that the subject’s epistemic posi-
tion is strong enough to warrant some specific action.
3
DeRose (2004a) discusses what happens in cases where ‘the sceptic has
executed a manoeuvre that has a tendency to raise the epistemic standards,
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 433
and that her opponent has responded by executing a manoeuvre which has
at least some tendency to keep lower, ordinary standards in place’.
4
One can have a purely theoretical interest in whether one’s epistemic
position with respect to a proposition is very strong even when the issue is
not practically important.
5
Compare Rysiew’s claim that, in some circumstances, the assertion ‘S
knows that p, but S cannot rule out all the bizarre ways in which p may be
false’ does not seem inconsistent or uncomfortable (2001, p. 495).
6
See his comments on the web and DeRose’s response at http://
www.brown.edu/Departments/Philosophy/tar/Archives/002074.html
7
This is part of Weatherson’s dialogue, which also includes a third char-
acter Suzanna.
8
For some problems with SSI, see DeRose (2004b) and Cohen (forth-
coming).
9
Many thanks to the Leverhulme Trust for supporting this research, and to
Kent Bach, Alex Barber, Emma Borg, Stewart Cohen, Duncan Pritchard
and Jenny Saul for useful comments which helped improve earlier versions
of the paper.

REFERENCES

Bach, K. (1994): ‘Conversational Impliciture’, Mind and Language 9, 124–

162.
Bach, K. (2001): ‘Speaking Loosely: Sentence Nonliterality’, MidWest
Studies in Philosophy 25, 249–263.
Bach, K. (Forthcoming (a)): ‘The Emperor’s New ‘Knows’’, in G. Preyer
and G. Peter (eds.), Contextualism in Philosophy: On Epistemology, Lan-
guage and Truth, Oxford University Press.
Bach, K. (Forthcoming (b)): ‘Descriptions: Points of Reference’, in A.
Bezuidenhout and M. Reimer (eds.), On Descriptions: Semantic and
Pragmatic Perspectives, Oxford: OUP.
Borg, E. (2004): Minimal Semantics. Oxford: OUP.
Carston, R. (1988): ‘Implicature, Explicature and Truth-theoretic Seman-
tics’, in Kempson (ed.), Mental Representations: The Interface between
Language and Reality (pp. 155–181), Cambridge: CUP.
Cohen, S. (1988): ‘How to be a Fallibilist’, Philosophical Perspectives 2, 91–
123.
Cohen, S. (1998): ‘Contextualist Solutions to Epistemological Problems:
Scepticism, Gettier and the Lottery’, Australasian Journal of Philosophy
76, 289–306.
Cohen, S. (1999): ‘Contextualism, Scepticism, and the Structure of Rea-
sons’, Philosophical Perspectives 13, 57–89.
434 J. BROWN

Cohen, S. (2000): ‘Contextualism and Scepticism’, Philosophical Issues 10,

94–107.
Cohen, S. (Forthcoming): ‘Knowledge, Speaker and Subject’, Philosophical
Quarterly.
DeRose, K. (1991): ‘Epistemic Possibilities’, Philosophical Review 100, 581–
605.
DeRose, K. (1992): ‘Contextualism and Knowledge Attributions’, Philoso-
phy and Phenomenological Research 52, 913–929.
DeRose, K. (1995): ‘Solving the Sceptical Problem’, Philosophical Review
104, 1–52. Page references from K. DeRose and T. Warfield (eds.), (1999):
Scepticism, (pp. 183–219), Oxford: OUP.
DeRose, K. (1998): ‘Contextualism: An Explanation and Defence’, In Greco
and Sosa (eds.), Blackwell Guide to Epistemology, Oxford: Blackwell.
DeRose, K. (2000): ‘How Do We Know We’re Not Brains In Vats’, Spindel
Supplement to the Southern Journal of Philosophy.
DeRose, K. (2002): ‘Assertion, Knowledge and Context’, Philosophical
Review.
DeRose, K. (2004a): ‘Single Scoreboard Semantics’, Philosophical Studies
119, 1–21.
DeRose, K. (2004b): ‘The Problem with Subject-Sensitive Invariantism’,
Philosophy and Phenomenological Research LXVIII, 346–350.
Donnellan, K. (1966): ‘Reference and Definite Descriptions’, Philosophical
Review 75.
Dretske, F. (1970): ‘Epistemic Operators’, Journal of Philosophy 67, 1007–
1023.
Grice, H.P. (1975): ‘Logic and Conversation’, In P. Cole and J. Morgan
(eds.), Syntax and Semantics, vol. 3, London: Academic Press. Page ref-
erences from R. Harnish (ed.), (1994) Basic Topics in the Philosophy of
Language (pp. 57–73), Hertfordshire: Harvester Wheatsheaf.
Grice, H.P. (1989): Studies in the Way of Words. Cambridge: HUP.
Kripke, S. (1977): ‘Speaker’s Reference and Semantic Reference’, MidWest
Studies in Philosophy 2, 255–276.
Hawthorne, J. (2004): Knowledge and Lotteries. Oxford: OUP.
Lewis, D. (1996): ‘Elusive Knowledge’, Australasian Journal of Philosophy
74, 549–567.
Nozick, R. (1981). Philosophical Explanations. Oxford: OUP.
Rysiew, P. (2001): ‘The Context-Sensitivity of Knowledge Attributions’,
Nous 35, 477–514.
Saul, J. (2002): ‘What is said and Psychological Reality; Grice’s Project and
Relevance Theorists’ Criticisms’, Linguistics and Philosophy 25, 347–372.
Schaffer, J. (Forthcoming): ‘Scepticism, Contextualism and Discrimination’,
Philosophy and Phenomenological Research.
Sosa, E. (1999): ‘How to Defeat Opposition to Moore’, Philosophical Per-
spectives 13, 141–154.
CONTEXTUALISM AND WARRANTED ASSERTIBILITY MANOEUVRES 435
Sperber, D. and Wilson, D. (1986): Relevance. Cambridge: HUP.
Stanley, J. (2004): ‘On the Linguistic Basis for Contextualism’, Philosophical
Studies 119, 119–146.
Stanley, J. (Forthcoming): ‘Context, Interest-Relativity and Knowledge’,
Philosophical Studies.
Stanley, J. and Szabo, Z. (2000): ‘On Quantifier Domain Restriction’, Mind
and Language 15, 219–261.
Stine, G. (1976): ‘Scepticism, Relevant Alternatives and Deductive Closure’,
Philosophical Studies 29, 249–261.
Weatherson, B. (2003): ‘Thoughts, Arguments and Rants: the Bank Cases’,
http://www.brown.edu/Philosophy/tar/Archives/002074.html.

Department of Philosophy
Bristol University
9 Woodland Road
Bristol BS81TB
UK
E-mail: j.a.brown@bristol.ac.uk