The Semantics of Knowledge Attributions Michael Blome Tillmann Full Chapter

The Semantics of Knowledge
Attributions Michael Blome-Tillmann

Visit to download the full and correct content document:
https://ebookmass.com/product/the-semantics-of-knowledge-attributions-michael-blo
me-tillmann/
OUP CORRECTED PROOF – FINAL, 13/5/2022, SPi
The Semantics of Knowledge Attributions

The Semantics of
Knowledge Attributions
MICHAEL BLOME-TILLMANN
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Michael Blome-Tillmann 2022
The moral rights of the author have been asserted
First Edition published in 2022
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2022931821
ISBN 978–0–19–871630–3
DOI: 10.1093/oso/9780198716303.001.0001
Printed and bound in the UK by
Clays Ltd, Elcograf S.p.A.
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
For A, A, and A
Contents
Acknowledgements xi
List of Figures xiii
Introduction xv
PART I: EPISTEMIC CONTEXTUALISM
1. Epistemic Contextualism 3
1.1 What is Epistemic Contextualism? 3
1.2 Evidence for Contextualism 7
1.3 Sceptical Puzzles 10
1.4 Error Theory and Contextual Shifts 12
1.5 Closure 14
1.6 The History of Contextualism 15
1.7 Suggestions for Further Reading 20
2. Semantic Implementations 21
2.1 Indexicality: Character and Content 21
2.2 Classical Contextualism 24
2.3 Radical Contextualism 34
2.4 Fancy Contextualism 39
2.5 The Varieties of Epistemic Contextualism 43
3. Versions of Contextualism 46
3.1 The Character of ‘Knowledge’ 46
3.2 David Lewis: Relevant Alternatives 48
3.3 Stewart Cohen: Internalist Evidentialism 55
3.4 Keith DeRose: Safety and Sensitivity 59
3.5 Further Views 64
4. Linguistic Objections 67
4.1 Semantic Blindness 67
4.2 Syntactic Gradability 79
4.3 Overt Restrictors 82
4.4 Free Shifting 85
4.5 A Linguistic Scorecard 86
viii 
5. Philosophical Objections 88
5.1 Over-generation 88
5.2 Too Much of a Concession? 89
5.3 Misconstruing the Sceptic? 95
5.4 Knowledge Norms 96
5.5 Surveys and Experimental Philosophy 100
A Wish List for Contextualists 102
PART II: EPISTEMIC IMPURISM

6. Epistemic Impurism 105
6.1 What is Epistemic Impurism? 105
6.2 Evidence for Impurism 111
6.3 Relation to Contextualism: Impurist Contextualism? 116
6.4 Evaluating the Arguments from Principles 119
7. Problems and Objections 123
7.1 Counterexamples to the Knowledge-Action Principles 123
7.2 High-Attributor/Low-Subject Stakes 127
7.3 Are there Practical Factor Effects? 129
7.4 Unpleasant Consequences: Embeddings 133
7.5 Further Objections 139
7.6 Concluding Remarks 141
PART III: EPISTEMIC RELATIVISM
8. Epistemic Relativism 145

8.1 Truth-Relativism about ‘Knowledge’-Attributions 145
8.2 Motivation 147
8.3 Objections and Replies 149
8.4 Conclusion 154
PART IV: STRICT INVARIANTISM
9. Psychological Invariantism 157

9.1 Lack of Belief 157
9.2 The Availability Heuristic 160
9.3 Epistemic Egocentrism and Epistemic Anxiety 162
9.4 Epistemic Focal Bias 168
9.5 Fragility and Over-generalization 171
 ix
10. Pragmatic Invariantism 173

10.1 Sceptical Pragmatic Invariantism 174
10.2 Moderate Pragmatic Invariantism 182
10.3 A Map of Positions 189
PART V: PRESUPPOSITIONAL EPISTEMIC

CONTEXTUALISM
11. The Presupposition Effect 193

11.1 The Insufficiency of Salience 193
11.2 Concrete and Vivid Salience 197
11.3 The Presupposition Effect 199
11.4 The Indirect Relevance of Stakes and Salience 203
11.5 Charity, Accommodation, and the Deferral Effect 205
11.6 Concessive Knowledge Attributions 210
11.7 Concluding Remarks 211
12. Presuppositional Epistemic Contextualism 213
12.1 From Attention to Presupposition 213
12.2 Presuppositional Epistemic Contextualism 215
12.3 Philosophical Objections: PEC and Non-Elusiveness 216
12.4 Linguistic Objections: Automaticity, Blindness, and
Clarifiability 218
References 225
Index 237
Acknowledgements
This monograph was written over the course of many years in Oxford,
Montreal, Cambridge, and Cologne. I am grateful to everybody at these
wonderful places for their support and input, and especially to Sven
Bernecker, Tim Crane, and Tim Williamson.
For extensive comments on predecessors and earlier versions of one or
another part of this monograph I am indebted to Brian Ball, Ralph
Wedgwood, and, again, Tim Williamson—as well as to an anonymous referee
for OUP. The idea for the book was gifted to me by Daniel Star, at the 2011
Pacific APA Meeting in San Diego. Thanks Daniel! Many other people have
provided input over the years. I am very grateful to all of them, and especially
to Yuval Avnur, Kent Bach, Sven Bernecker, Jessica Brown, Alexander Dinges,
Igor Douven, Julian Dutant, Mikkel Gerken, Thomas Grundmann, John
Hawthorne, Jonathan Jenkins Ichikawa, Christoph Kelp, Dustin Locke,
Ofra Magidor, Matt McGrath, Jennifer Nagel, Duncan Pritchard, Patrick
Rysiew, Bernhard Salow, Jason Stanley, Bruno Whittle, and Crispin Wright.
I apologize to all those who I have unjustifiably failed to mention.
I would also like to thank the organizations and funding agencies that
kindly supported my research: the Social Sciences and Humanities Research
Council of Canada, McGill University, as well as the University of Cambridge,
where I was a Marie Curie Research Fellow from 2014 to 2016, funded by the
European Commission (Marie Curie Actions PIIF-GA-2012-328969
‘Epistemic Vocabulary’).
I have drawn, with kind permission in each case, on previously published
material, as detailed below. I thank the editors and publishers concerned for
permission to use this material; in particular, I thank Blackwell Publishing,
Oxford University Press, Springer Science+Business Media, and Wiley-
Blackwell Publishing. Parts of Chapter 12 are drawn from my paper
‘Knowledge and Presuppositions,’ Mind 118(470): 241‒294 (2009).
Section 2.5 is drawn from ‘Contextualism and the Epistemological
Enterprise,’ Proceedings of the Aristotelian Society, CVII: 387‒394 (2007).
Parts of Section 7.4 are based on ‘Contextualism, Subject-Sensitive
Invariantism, and the Interaction of ‘Knowledge’-Ascriptions with Modal
and Temporal Operators,’ Philosophy and Phenomenological Research
xii 
79(2): 315‒31 (2009), and parts of Chapter 4 are drawn from parts of ‘The
Indexicality of “Knowledge”,’ Philosophical Studies 138(1): 29‒53 (2008),
while Section 5.4 is based on ‘Contextualism and the Knowledge Norms,’
Pacific Philosophical Quarterly 94(1): 89–100 (2013). Parts of Section 10.2.2
are sourced from ‘Moderate Pragmatic Invariantism and Contextual
Implicature Cancellation,’ Analysis (forthcoming), and parts of Sections 6.3‒4
are sourced from ‘Impurism, Pragmatic Encroachment, and the Argument
from Principles,’ Synthese (forthcoming). Chapter 1 is partly based on
‘Knowledge as Contextual’, in: Knowledge in Contemporary Philosophy. The
Philosophy of Knowledge: A History, Vol. IV, Stephen Hetherington and
Markos Valaris (eds.). London: Bloomsbury Academic: 175‒194 (2018).
Parts of Section 12.4 are based on ‘Gradability and Knowledge’, in:
Routledge Handbook on Epistemic Contextualism, Jonathan Jenkins Ichikawa
(ed.). London: Routledge: 348‒57 (2017).
Much of the work in this monograph has been in the making for many
years, and the book itself has been almost finished for the best part of six years.
Sometimes life gets in the way, and in this case, I am mostly lucky that it did.
Lastly, I owe the greatest debt of gratitude to my family: my wife, my
parents, and my two children, for all the fun, love, and support.
M.B.-T.
List of Figures
1. The Kaplan/Lewis picture—character, content, and circumstance

of evaluation 23
2. Classical epistemic contextualism 25
3. Radical contextualism 38
4. Fancy contextualism 42
5. The varieties of epistemic contextualism 44
6. Scorecard of linguistic properties of ‘knows p’ 86
7. Strict invariantism 189
Introduction
Exactly forty-two years ago, in January 1979, David Lewis published his
milestone paper Scorekeeping in a Language Game (Lewis 1979). In it, Lewis,
for the first time, broaches the issue of the semantics of ‘knowledge’-attributions
by discussing how the truth-values of ‘knowledge’-attributions can shift with the
context of utterance:
Take another example. The commonsensical epistemologist says: ‘I know

the cat is in the carton—there he is before my eyes—I just can’t be wrong
about that!’ The sceptic replies: ‘You might be the victim of a deceiving
demon’. Thereby he brings into consideration possibilities hitherto
ignored, else what he says would be false. The boundary [between relevant
and ignored possibilities] shifts outward so that what he says is true.
Once the boundary is shifted, the commonsensical epistemologist must
concede defeat. And yet he was not in any way wrong when he laid claim to
infallible knowledge. What he said was true with respect to the score as it
then was. (Lewis 1979: 355)
Forty-two years later, the semantics of ‘knowledge’-attributions has

become one of the major areas in contemporary epistemology. Seemingly
everybody working in contemporary epistemology has a view on the topic,
or at least aims to make their views compatible with one of the main
contenders in the debate. Surprisingly, however, there isn’t yet a book-
length, systematic, and reasonably concise discussion of the central views
in this field.
This book is meant to fill that gap. It consists of five parts. The first four
parts are devoted to the main contenders in the field—that is, to:
i. Epistemic Contextualism
ii. Epistemic Impurism
iii. Epistemic Relativism
iv. Strict Invariantism
The Semantics of Knowledge Attributions. Michael Blome-Tillmann. Oxford University Press.

© Michael Blome-Tillmann 2022. DOI: 10.1093/oso/9780198716303.001.0001
xvi 
The discussion in these first four parts of the book presents an overview of
the debate, while Part V is devoted to my own view—namely, to what I call:
v. Presuppositional Epistemic Contextualism
As those familiar with the debate will know, this last part of the book develops
a view that is strongly inspired by David Lewis’s 1979 approach. Thus, while
much has been said over the past forty-two years, what has been said at the
very beginning still carries, to my mind, considerable weight in the discussion
of ‘knowledge’-attributions.
To provide a manageable map of the current debate, I have chosen to
proceed in a broadly historical order. I begin with a discussion of epistemic
contextualism (EC) and then approach competing views on the background of
that initial discussion. The individual parts and chapters of the book, there-
fore, build on each other to some degree, but I think that they can nevertheless
be fruitfully read independently of each other.
Finally, let me emphasize that, while this work aims to provide a compre-
hensive overview of the current debate on ‘knowledge’-attributions, it remains
opinionated. While I have tried to be as impartial as possible and to cover all
the major developments in the literature, the result is bound to be
incomplete—especially so given the sheer amount of work published in the
area in recent years.¹ I apologize to all those whose excellent work I have
ignored unjustifiably.
¹ The relevant subcategories in the PhilPapers archive, for instance, list well over 1,500 references at
the time of writing.
PART I
E P I S T E M I C CO N T E X T U A L I S M
1
Epistemic Contextualism
1.1 What is Epistemic Contextualism?
Imagine schoolteacher Jones in the zoo explaining to her class that the animals
in the pen are zebras.¹ Tom is unconvinced and challenges Jones: ‘Are you sure
those aren’t antelopes?’ After Jones has explained the difference between
antelopes and zebras, Tom assures his classmates:
(1) She knows that the animals in the pen are zebras.
Has Tom spoken truly? Surely, Jones’s epistemic position seems good enough
for satisfying the predicate ‘knows that the animals in the pen are zebras’
(henceforth ‘knows Z’): Jones has visual experiences of a black and white
striped horse-like animal, she can discriminate reliably between zebras and
antelopes, she has read the sign on the pen that reads ‘Zebra Pen’, etc. Thus,
Tom’s utterance of (1) seems to be a paradigm case of a true ‘knowledge’-
attribution.
Next, imagine a couple, Bill and Tanya, walking along. Bill, a would-be
postmodernist artist, gives details of his latest ideas: he envisions himself
painting mules with white stripes to look like zebras, putting them in the
zebra pen of a zoo and thereby fooling visitors. Our couple randomly con-
siders Jones, and Tanya claims, at the very same time as Tom is asserting (1):
(2) She doesn’t know that the animals in the pen are zebras.
In Tanya’s mind, for Jones to ‘know Z’, she must have better evidence or
reasons in support of Z than are momentarily available to her. In particular,
Tanya has it that Jones’s evidence must eliminate the possibility that the
animals in the pen are painted mules. As long as her evidence is neutral
with respect to whether or not the animals are cleverly painted mules, Tanya
claims, Jones doesn’t qualify as ‘knowing Z’.
¹ The example is derived from the zebra case in (Dretske 1970).

4  
What is going on in this example? According to our intuitions, an utterance

of (1) is true in the context of the school class, while an utterance of (2) is true
in the context of the artistic couple. Moreover, (2) is the negation of (1): it
doesn’t differ from (1) except for containing the verbal negation ‘doesn’t’. And
since the personal pronoun ‘she’ refers in both contexts to Jones, it seems that
the schoolteacher satisfies the predicate ‘knows Z’ in the context of the school
class but not so in the context of the artists.
How are we to account for these phenomena? First, note again that Tom and
Tanya are talking about one and the same person—Jones—at exactly the same
time.² Thus, we cannot resolve the situation by claiming, for instance, that Jones
‘knows Z’ in one context but not the other because she believes Z in one context
but not in the other.³ Similarly, we cannot plausibly respond that Jones has
certain visual experiences in one context that she lacks in the other, or that she
has the ability to discriminate reliably between certain scenarios in one context
but not the other, or, finally, that she has read the sign on the zebra pen in one
context but not in the other. All factors pertaining to the subject are identical
with respect to both contexts, as the speakers in both contexts—Tom and
Tanya—are talking about one and the same subject at one and the same time.⁴
Thus, what the above example suggests is that the mentioned epistemic
factors—Jones’s visual experiences, her discriminatory abilities, etc.—are suf-
ficient for her to satisfy ‘knows Z’ in one context, but not so in the other. And it
is this view that epistemic contextualism (EC) takes at face value: how strong
one’s epistemic position towards a proposition p must be for one to satisfy
‘know(s) p’ may vary with the context of utterance. In the artists’ conversa-
tional context, Jones needs to be in a stronger epistemic position—she needs
more or stronger evidence in support of Z—than in the school class’s conver-
sational context in order for her to satisfy ‘knows Z’. In fact, some context-
ualists describe the situation by claiming that contexts of utterance are
governed by so-called epistemic standards.⁵ Given this terminology, the epi-
stemic standards in the above example are lower in the context of the school
class than in the context of the artists’ conversation. In the former context, the
² The circumstances of evaluation, as semanticists put it, are identical with respect to both utter-
ances.
³ We shall discuss views aiming to explain the data by postulating differences in the subject’s
doxastic (i.e. belief-related) states in Section 9.1.
⁴ It follows that we cannot plausibly explain the above phenomena by pointing out that Jones is in a
different practical situation with respect to the two contexts, for she simply isn’t. This fact about the
above example creates problems for pragmatic encroachment or impurist accounts of knowledge, which
are the topic of Chapters 6 and 7.
⁵ Cf. (Cohen 1988; DeRose 1995; Lewis 1996).
    5
standards are low enough for Jones to satisfy ‘knows Z’, while in the latter they
are too high: Jones doesn’t, in the artists’ context, satisfy ‘knows Z’ but instead
satisfies ‘doesn’t know Z’.
The notion of an epistemic standard can be explicated in a variety of ways. On
one of the more intuitive ways, which is inspired by relevant alternative
approaches to contextualism, epistemic standards are said to be lower in the
school class’s context because, as David Lewis (1996) puts it, satisfying ‘knows Z’
in that context doesn’t require the elimination of the possibility that the animals
are painted mules, while this is required in the context of the postmodernist
artists. More alternatives must be eliminated or ruled out in the context with the
higher standards than in the context with the lower standards.⁶
Given the hypothesized context-sensitivity of the predicate ‘know(s) p’, it is
in general possible that a subject satisfies the predicate in one conversational
context but doesn’t do so in another, or, in other words, that somebody in a
given context speaks truly when uttering a sentence of the form ‘x knows p’
while somebody in a different context speaks falsely when uttering the very
same sentence—even though both speakers are speaking about the same
subject x at the same time of utterance t. Epistemic contextualism is, as a
consequence, a linguistic or a semantic view—namely, the view that the truth-
values of ‘knowledge’-attributions—sentences of the form ‘x knows p’—may
vary with the context of utterance. According to EC, ‘knowledge’-attributions
are, as Jason Stanley (2005b: 16) puts it, context-sensitive “in a distinctively
epistemological way”: the truth-values of a sentence S containing the predicate
‘know(s) p’ can change with context, independently of whether S is ambiguous
or context-sensitive in any other way.
In a first approximation, we can thus define ‘epistemic contextualism’ as
follows:
Epistemic Contextualism (EC):

The truth-values of ‘knowledge’-attributions may vary with the context of
utterance, where this variance is traceable to the occurrence of ‘know(s) p’
and concerns a distinctively epistemic factor.
Given semantic compositionality—the view that the extensions of complex

expressions are a function of the extensions of their ultimate constituents and
the way in which they are combined—(EC) entails (EC*):
⁶ Strictly speaking, talk of epistemic standards is metaphorical on this view, due to the incommen-
surability of alternatives. Let us ignore this detail until later.
6  
Epistemic Contextualism (EC*):

The contribution that the predicate ‘know(s) p’ makes to the truth-value of a
sentence it is embedded in may vary with the context of utterance in a
distinctively epistemic way.
According to the above definitions, the predicate ‘know(s) p’ adds context-

sensitivity to a sentence it occurs in, and this context-sensitivity is distinctly
epistemic—that is, it goes over and above the context-sensitivity that the verb
contributes to the sentence by virtue of its tense.⁷ (EC) and (EC*) are the final
definitions of epistemic contextualism that I shall use throughout the book.
As we shall see later, in Chapter 2, and in much more detail, there are
several different ways to semantically model the context-sensitivity of ‘know(s)
p’. However, it is worth noting at this point that EC is not a lexical ambiguity
theory—that is, it doesn’t claim that ‘know(s) p’ is assigned multiple conven-
tional meanings in English, as are the lexically ambiguous expressions ‘bank’
or ‘orange’. On the contrary, contextualists have commonly compared
‘know(s) p’ to indexical expressions, such as ‘I’, ‘that’, and ‘today’ or to
gradable adjectives such as ‘flat’ and ‘empty’: these expressions are widely
taken to have only one conventional meaning—what Kaplan (1989) calls their
‘character’—but different contents or semantic values in suitably different
contexts of utterance.
To illustrate this further many contextualists have employed an analogy to
gradable adjectives that will prove fruitful for the moment, even though it will
later turn out to be a merely rough and ready analogy. Just as what counts as
satisfying ‘flat’, ‘empty’, or ‘tall’ may vary with context, contextualists have
argued, what counts as satisfying ‘know(s) p’ may vary with context, too.⁸
Given this analogy, EC can be construed as claiming that the utterances of (1)
and (2) in the above zebra example stand in a relation similar to the relation
between a basketball coach’s utterance of ‘MB-T isn’t tall’ and a jockey coach’s
utterance of ‘MB-T is tall’: while the surface syntax of these sentences suggests
a contradiction, the propositions expressed are compatible, as the semantic
value of ‘tall’ changes with the context of utterance. On the face of it, ordinary
usage of ‘tall’ and ‘knows p’ seem to be similar: both expressions seem to be
context-sensitive.⁹
⁷ Cf. (Schaffer and Szabó 2014) for the above definition of EC.
⁸ For this analogy, see (Unger 1975; DeRose 1995; Lewis 1996; Cohen 1999, 2004a).
⁹ Not all contextualists endorse the analogy to gradable adjectives (see, for instance, (Schaffer and
Szabó 2014)), but the analogy is helpful in illustrating the general concept of semantic context-
sensitivity underlying the view.
   7
One further terminological matter should be mentioned here. The negation

of epistemic contextualism is usually called invariantism, and both of the
terms under discussion here—that is, ‘contextualism’ and ‘invariantism’—
have been coined by Peter Unger (1984). According to Unger’s usage of the
terms, invariantism is simply the view that ‘knows p’ is an ‘absolute’ term—
that is, it applies to a subject only if the subject can satisfy a set of absolute—
and that means in Unger’s case—ascriber-invariant standards. Now, the issue
of how to precisely implement EC and the context-sensitivity of ‘know(s) p’ at
the semantic level shall be discussed in detail in Chapter 2. For now, let us
consider evidence in support of EC that has been put forward by its defenders
in the literature.
1.2 Evidence for Contextualism
The main evidence for EC derives from our intuitions about the truth-values
of ‘knowledge’-attributions in examples such as the above zebra case.¹⁰
However, there are further, more familiar examples that have been presented
in the literature. Consider, for instance, Stewart Cohen’s (1999: 58) Airport
Case:
Airport Case
Mary and John are at the L.A. airport contemplating taking a certain flight to
New York. They want to know whether the flight has a layover in Chicago.
They overhear someone ask a passenger, Smith, if he knows whether the
flight stops in Chicago. Smith looks at the flight itinerary he got from the
travel agent and responds, ‘Yes I know—it does stop in Chicago.’ It turns out
that Mary and John have a very important business contact they have to
make at the Chicago airport. Mary says, ‘How reliable is that itinerary? It
could contain a misprint. They could have changed the schedule at the last
minute.’ Mary and John agree that Smith doesn’t really know that the plane
will stop in Chicago. They decide to check with the airline agent.
(Cohen 1999: 58)
As Cohen’s example suggests, the sentence ‘Smith knows that the flight stops
over in Chicago’ appears true as uttered in Smith’s context but false as uttered
¹⁰ They constitute so-called ‘context-shifting arguments’. See (Hansen and Chemla 2013: 287) for
discussion.
8  
in Mary’s and John’s context. Moreover, it seems as though the practical

interests and goals of the conversational participants or how high the stakes
are with regard to the proposition that the flight will stop over in Chicago
influence the respective contexts’ epistemic standards, and thus whether Smith
satisfies ‘knows’. In Smith’s own context he satisfies ‘knows that the flight will
stop over in Chicago’, but in Mary’s and John’s context, where the stakes are
significantly higher, he doesn’t.
Here is another example reinforcing this point—namely, Keith DeRose’s
(1992) famous Bank Cases, as presented by Stanley (2005b: 3–4):
Low Stakes
Hannah and her wife Sarah are driving home on a Friday afternoon. They
plan to stop at the bank on the way home to deposit their paychecks. It is not
important that they do so, as they have no impending bills. But as they drive
past the bank, they notice that the lines inside are very long, as they often are
on Friday afternoons. Realizing that it isn’t very important that their pay-
checks are deposited right away, Hannah says, ‘I know the bank will be open
tomorrow, since I was there just two weeks ago on Saturday morning. So we
can deposit our paychecks tomorrow morning.’
High Stakes
Hannah and her wife Sarah are driving home on a Friday afternoon. They
plan to stop at the bank on the way home to deposit their paychecks. Since
they have an impending bill coming due, and very little in their account, it is
very important that they deposit their paychecks by Saturday. Hannah notes
that she was at the bank two weeks before on a Saturday morning, and it was
open. But, as Sarah points out, banks do change their hours. Hannah says,
‘I guess you’re right. I don’t know that the bank will be open tomorrow.’
Similar to Cohen’s example, DeRose’s case suggests that it is more difficult to

satisfy ‘knows p’ in a context in which the stakes are higher: in Low Stakes
Hannah satisfies ‘knows that the bank will be open tomorrow’, whereas in
High Stakes she doesn’t, even though she is in exactly the same epistemic
position in both cases. And, again, the defender of EC argues that the differ-
ence in our intuitions in the two bank cases is due to the fact that the relevant
‘knowledge’-attributions are made in different conversational contexts: in the
context of High Stakes, the argument goes, it is considerably more difficult to
satisfy ‘knows that the bank is open on Saturday’ than it is in the context of
Low Stakes.
   9
The above examples and others like them have attracted a large amount of
critical attention in the recent literature. Let us briefly consider four points of
contention. First, the above argument in support of EC takes the form of an
inference to the best explanation. In evaluating the support EC receives from
the examples, we must, therefore, compare EC’s account of the data with
competing explanations offered by rival theories. Such comparisons of
explanatory virtue in philosophy are often difficult and complicated, as it is
not always clear how to compare varying advantages of different theories. We
shall consider the major competitors to contextualism in later chapters to shed
some light on this issue.
A second point of contention concerns the above data’s evidential status.
Some early experimental philosophers have aimed to undermine the above case
for contextualism by arguing that professional philosophers’ intuitions about
the relevant examples do not match the intuitions of the general public. Some
more recent and methodologically superior studies, however, have uncovered
evidence that confirms epistemic contextualism or that is at least compatible
with the view. These issues present interesting and legitimate challenges for the
contextualist and we shall discuss them in more detail in Section 5.5.
A third point that has attracted much discussion in the literature concerns
the exact contextual factors influencing the semantic content of ‘know(s) p’ at
a context. Cohen’s Airport Case and DeRose’s Bank Case suggest that the
practical stakes at the context of ascription—or more broadly, the speakers’
practical interests—influence what we earlier called the context’s epistemic
standards. But it is not obvious that similar mechanisms are in place in the
zebra example from the previous section. Some theorists have, therefore,
proposed that what raises the epistemic standards at a context are exclusively
the alternatives or error-possibilities that are salient or taken seriously at the
context of utterance. The debate about stakes vs. salience is still ongoing, but
we shall see in later chapters (Chapters 7 and 11) that in recent work
significant doubt has been cast on the idea that our practical interests have
an influence on our willingness to attribute ‘knowledge’.
A fourth and final issue to be mentioned here concerns the difference
between first-person and third-person ‘knowledge’-attributions. The
‘knowledge’-attributions in DeRose’s Bank Cases are made from the first-
person perspective—they are so-called self-ascriptions—while the ‘knowledge’-
attributions in the zebra case are third-person ascriptions. Tom ascribes ‘know-
ledge’ to Smith and Tanya ascribes ‘ignorance’—the absence of ‘knowledge’—to
Smith. The difference between first-person ascriptions and third-person ascrip-
tions is important, as some of the rival views to EC—such as the pragmatic
10  
encroachment accounts discussed in Chapters 6 and 7—have the explanatory

resources to explain the former but are struggling with the latter. In determining
which view has the better explanation of the data we thus need to make sure that
we consider both first-person and third-person ‘knowledge’-attributions.
While all of the above issues present interesting and legitimate challenges to
EC, let us leave them to one side for the moment and focus our attention on a
topic that is commonly taken to provide an important philosophical motiv-
ation for EC: sceptical puzzles.
1.3 Sceptical Puzzles
As we have seen in the previous section, EC receives some prima facie support
from examples involving ‘knowledge’-attributions. In addition to this empir-
ical support, contextualists have also frequently argued that there is a second,
independent motivation for EC. EC is an attractive view, the argument goes,
because it provides us with an attractive solution to the sceptical paradox. Even
though the topic of scepticism has somewhat moved into the background in
the recent literature, it is nevertheless important to consider the contextualist
treatment of scepticism in some detail.¹¹ Let us begin the discussion of the
contextualist response to scepticism by formulating the following Sceptical
Argument:
Sceptical Argument:
(i) If I know that I have hands, then I’m in a position to know that I’m
not a handless brain in a vat.
(ii) I’m not in a position to know that I’m not a handless brain in a vat.
(iii) I don’t know that I have hands.
The above argument is valid. If we accept its premises, we must accept its
conclusion, too. Moreover, together with the eminently plausible (iv), the
argument constitutes a paradox:
(iv) I know that I have hands.
¹¹ Peter Ludlow (2005), for instance, defends EC purely on the basis of the linguistic data, leaving
aside the issue of scepticism entirely.
  11
The propositions (i), (ii), and (iv) form an inconsistent set, and so at least one
of them has to be false. But each of the above propositions is, when considered
in isolation, rather plausible. In other words, we tend to intuit that (iv) is
true—call this the anti-sceptical intuition—but we also tend to intuit, when
considering the sceptical argument and in varying degrees, that (iii) is true—
call this the sceptical intuition. How can we resolve this puzzle?
The traditional contextualist response is to claim that the Sceptical
Argument is unsound in conversational contexts that are governed by mod-
erate or everyday epistemic standards but sound in contexts with artificially
high or sceptical epistemic standards.¹² In everyday contexts, the argument
goes, I satisfy ‘knows that he isn’t a handless brain in a vat’ and premise (ii) of
the sceptical argument expresses a falsehood: if I satisfy, in ordinary contexts,
‘knows that he isn’t a handless brain in a vat’, then I also satisfy, in ordinary
contexts, ‘is in a position to know that he’s not a handless brain in a vat’.
Consequently, the Sceptical Argument is unsound in ordinary contexts, and its
conclusion doesn’t follow. Relative to ordinary contexts, I satisfy ‘knows that
he has hands’ and the conclusion of the sceptical argument expresses a
falsehood.
The situation is different in so-called sceptical contexts in which we practice
epistemology and consider and discuss sceptical scenarios such as the brain-
in-a-vat scenario. In such contexts, the argument goes, the epistemic standards
are considerably higher—in fact, outrageously high—to the effect that premise
(ii) expresses a truth in such contexts. For instance, contextualists have argued
that because sceptical possibilities are relevant or attended to in sceptical
contexts, premise (ii) of the Sceptical Argument expresses a truth in such
contexts. Sceptical hypotheses are, after all, uneliminated by our evidence, and
we therefore do not, in sceptical contexts, satisfy the predicate ‘is in a position
to know that s/he is not a handless brain in a vat’. Consequently, when the
sceptic asserts, in her sceptical context, ‘MB-T doesn’t know that he has
hands’, she asserts a truth. However, it is crucial to emphasize that the
truth of the sceptic’s assertion does not affect the truth-values of positive
‘knowledge’-attributions in ordinary contexts.
Another and more elegant way to formulate the contextualist solution to the
sceptical puzzle is by ascending semantically and formulating our sceptical
and anti-sceptical intuitions as follows:
¹² The following is a description of standard contextualist views on sceptical puzzles, as it can be

found—more or less explicitly—in all major writings of contextualists. See, for instance, (DeRose 1995)
and (Cohen 1999).
12  
Anti-Sceptical Intuition (ASI):

People often speak truly when they assert ‘I know p.’
Sceptical Intuition (SI):
People sometimes speak truly when they assert ‘Nobody knows p’ in
contexts in which sceptical arguments are discussed.
The traditional contextualist then claims that both of these intuitions are
correct and only seemingly contradictory. They are correct because the con-
tribution that ‘knows p’ makes to the contents or extensions (i.e. truth-values)
of the sentences it is embedded in varies with the context of utterance;
so when we claim in everyday contexts that we ‘know p’, our utterances
are not in contradiction to our utterance of ‘Nobody knows p’ in sceptical
contexts.
According to the traditional contextualist, it is, therefore, not (iv) which is
shown to be true or (iii) which is shown to be false.¹³ Rather, the traditional
contextualist emphasizes that our sceptical and anti-sceptical intuitions are
exclusively intuitions about the truth-values of utterance tokens, which are by
their very nature situated in particular conversational contexts. Our intuitions
are not about the truth-values of sentences as considered in the abstract in a
philosophical treaties or discussions. Once we appreciate this point and take
into account the context-sensitivity of ‘knowledge’-attributions, the sceptical
puzzle is—the traditional contextualist argues—easily dissolved: the argument
is sound (and its conclusion true) in contexts with exceedingly high or sceptical
epistemic standards, but unsound (and its conclusion false) in contexts with
moderate or everyday epistemic standards.¹⁴
1.4 Error Theory and Contextual Shifts
While initially plausible and elegant, the solution to the sceptical puzzle
just outlined is in fact rather contentious. For if the truth-values of
‘knowledge’-attributions change in a way that allows for both (ASI) and (SI)
to be true, why are we puzzled by sceptical arguments to begin with? Shouldn’t
¹³ As Stewart Cohen (1988: 94) has pointed out, an intellectually satisfying solution to the sceptical
puzzle doesn’t merely block the argument by identifying the false premise. It must also offer an
explanation of why the false premise appeared so plausible at first.
¹⁴ This is a description of standard contextualist views on sceptical puzzles, as it can be found—more
or less explicitly—in all major writings of contextualists. See, for instance, (DeRose 1995) and (Cohen
1999).
     13
we, as competent speakers of English, be somehow sensitive to or aware of the

context-sensitivity of ‘knowledge’-claims and thus also be sensitive or aware of
the fact that the soundness of the Sceptical Argument is a context-dependent
matter? In other words, if EC is true, why are sceptical arguments puzzling?
To account for the puzzling nature of sceptical arguments, the contextualist
is committed to the view that we sometimes lose sight of the context-
sensitivity of epistemic terms. According to the traditional contextualist, we
simply do not realize that the sceptical conclusion is true in the context of
discussions of sceptical arguments only, but false in everyday contexts.¹⁵
Contextualists accordingly pair their semantics of ‘knows p’ with the view
that we are sometimes unaware of or tend to overlook the context-sensitivity
of ‘knows p’. Stephen Schiffer (1996), in an influential and very critical paper,
aptly calls this aspect of standard contextualism its error theory. Ultimately, it
is entirely due to this error theory that EC can claim to be able to account for
both the plausibility of sceptical arguments and our intuition that our every-
day ‘knowledge’-attributions express truths. But how credible is EC’s error
theory, which effectively amounts to the surprising claim that competent
speakers suffer from a form of semantic blindness?
We shall return to the topic of semantic blindness in much greater detail
below (see Section 4.1). In the meantime, note that there is another wrinkle in
the contextualist’s resolution of sceptical puzzles. Remember that the context-
ualist’s account of sceptical puzzles makes crucial use of the intuitive notion of
a shift or a difference of epistemic standards between contexts. But it is
important to note that arguing for the context-sensitivity of ‘know(s) p’ on
the basis of the claim that the Bank Case, the Airport Case, and the Zebra Case
involve contextual variation of epistemic standards is one thing, while claim-
ing that epistemic standards are ‘shifty’ in precisely the way required for a
resolution of sceptical puzzles is quite a different proposition. In fact, for the
contextualist’s treatment of sceptical puzzles to be credible, we need to be told
more about the exact mechanisms underlying contextual shifts and variations
in epistemic standards. In other words, we need to be told more about what
epistemic standards are and how they are determined by context. Why is it
that epistemic standards are ‘high’ in so-called sceptical contexts, and why are
they ‘lower’ in everyday situations?¹⁶
¹⁵ Cf. (Cohen 1988: 106; DeRose 1995: 40).

¹⁶ Of course, an account of how epistemic standards shift would also make the contextualist’s
account of examples such as the Bank Case more credible. See Section 3.5.2 for further discussion.
14  
Different contextualists have different stories to tell about what epistemic

standards are and about how they shift and vary. But it is fair to say that the
original approaches proposed in early writings on contextualism (such as
(DeRose 1995) and (Lewis 1996)) must be refined and amended before we
can grant the contextualist that her view offers a solution of sceptical puzzles.
A detailed discussion of these issues and the more recent literature on epi-
stemic standards will be the topic of Chapters 3 and 11.
1.5 Closure
One interesting aspect of the contextualist solution of sceptical puzzles is that

the contextualist resolves sceptical puzzles while fully respecting our intuition
that one can extend one’s knowledge by competent deduction. To see this,
consider the following principle, which is discussed in the literature under the
label Single-Premise Closure:¹⁷
Single-Premise Closure (CL):

If x knows p and x knows that p entails q, then x is in a position to know q.
Here is an instance of (CL) for illustration: if, first, I know that the animal
outside my window is a fox and if, second, I know that its being a fox entails
that it’s not a cat, then I am also in a position to know that the animal outside
my window is not a cat. Of course, I am in a position to know that latter
proposition because I can competently deduce it from (i) my knowledge that
the animal is a fox and (ii) my knowledge that its being a fox entails that it is
not a cat. Single-premise closure captures fairly precisely the intuition that one
can extend one’s knowledge by means of deductive reasoning.¹⁸
While some epistemologists have argued that giving a response to the
sceptic requires us to give up (CL), the contextualist resolution gets by without
any such move. How does the contextualist avoid rejecting closure? Note that,
according to EC, every semantic value that the verb ‘knows p’ can express in a
given context is, loosely speaking, closed under ‘known’ entailment. Here is a
¹⁷ See, for instance, (Lasonen-Aarnio 2008).

¹⁸ Note also that the sceptic can use (CL) when motivating premise (i) of her argument: she can
reason from the assumption that I know that my having a hand entails that I am not a handless brain in
a vat to the conclusion that if I know that I have a hand, then I am in a position to know that I am not a
handless brain in a vat, i.e., to premise (i) of the Sceptical Argument.
    15
more precise and contextualized formulation of the single-premise principle to

illustrate the idea:
Contextualized Single-Premise Closure (CLC):

If x satisfies ‘knows p’ in context C and satisfies ‘knows that p entails q’ in C,
then x is in a position to satisfy ‘knows q’ in C.
(CLC) is a meta-linguistic principle. Loosely speaking, (CLC) says that (CL),

its non-contextualized cousin, expresses a truth as long as the conversational
context is kept fixed.¹⁹ Thus, unlike Fred Dretske (1970) and Robert Nozick
(1981), who reject (CL) in giving a response to the sceptic, the contextualist, by
ascending semantically, merely modifies and clarifies (CL) in a way that
respects our intuitions about the validity of the closure principle.
Contextualists have traditionally taken this to be a great comparative advan-
tage of their theories over epistemological theories that advocate closure
failure.²⁰
1.6 The History of Contextualism
The development of epistemic contextualism took place in several waves and is

a paradigm case of philosophical teamwork and continuous evolution. There
is no singular founder or discoverer of the view. Rather, there are several
theorists who played important roles in developing the view over a period of
by now four decades. The emergence of contextualism during the late 1980s
and early 1990s, was made possible by developments in the philosophy of
language—in particular, by David Kaplan’s and David Lewis’s work on
demonstratives, indexicals, and context-sensitive expressions more generally.
Very early work containing ideas that are vaguely—in fact, very vaguely—
similar to those of contemporary contextualists can be found in the works of
the ordinary language philosophers J. L. Austin (1946) and Norman Malcolm
(1952, 1963). With some creativity, Austin, for instance, can be interpreted as
suggesting, in his own idiosyncratic terminology, that in order to know p one
does not always need to eliminate all alternatives to p,²¹ and Malcolm seems to
¹⁹ See (DeRose 1995; Lewis 1996; Cohen 1999).

²⁰ For discussion of other epistemological puzzles involving closure and contextualism, see
(Lasonen-Aarnio 2017).
²¹ See, for instance, (Austin 1946: 159), where Austin claims in his discussion of the predicate ‘real’:
“The doubt or question ‘But is it a real one?’ has always (must have) a special basis, there must be some
16  
defend an ambiguity theory of ‘knows p’ when he distinguishes between what

he calls a ‘weak’ and a ‘strong’ sense of ‘knowledge’. In yet later work Malcolm
proposes an account of the epistemic modal ‘it is possible that’ that, as Gail Stine
(1976: 260) puts it aptly, “is very much in accord with considerations which go
towards making a proposition a ‘relevant alternative’.” However, neither of these
authors defends a full-blown contextualist account of ‘knows p’.
A milestone in the development of epistemic contextualism is Fred
Dretske’s (1970) paper ‘Epistemic Operators’. In this paper, Dretske develops
what is nowadays known as Relevant Alternatives Theory (henceforth ‘RAT’).
According to RAT, in order to know p one doesn’t need to eliminate or rule
out absolutely every alternative to p. Ruling out the relevant alternatives is
perfectly sufficient. For instance, in order to know that the animal in the pen is
a zebra one doesn’t need to rule out that it is a cleverly painted mule, since that
alternative is irrelevant. Drestke’s relevant alternatives account is thus a close
precursor to epistemic contextualism, but it is important to note that it differs
from EC in one crucial respect. Dretske’s RAT tacitly assumes that whether an
alternative is relevant or not is a contextually invariant matter, so RAT
crucially doesn’t conceive of ‘knows p’ as a context-sensitive expression.
An important consequence of Dretske’s RAT is that it leads to closure
failure. One of Dretske’s main motivations for his approach was to develop a
response to sceptical arguments. The way in which RAT blocked sceptical
reasoning, however, meant that closure would fail. According to RAT, we
know that we have hands, but we are not in a position to know that we aren’t
handless brains in vats. That is so because, according to RAT, we can rule out
the relevant alternatives to our having hands—for instance, the alternative that
we have lost our hands in a terrible car accident—but we cannot rule out the
relevant alternatives to our not being brains in vats—namely, the alternative
that we are brains in vats.²²
Given the intuitive plausibility of the idea that we can extend our knowledge
by means of competent deduction, Dretske’s willingness to reject closure has
widely been taken to be a considerable cost to the view. It is on this back-
ground that the first contextualist versions of relevant alternatives theories
‘reason for suggesting’ that it isn’t real, in the sense of some specific way, or limited number of specific
ways, in which it is suggested that this experience or item may be phoney. Sometimes (usually) the
context makes it clear what the suggestion is: the goldfinch might be stuffed but there’s no suggestion
that it’s a mirage, the oasis might be a mirage but there’s no suggestion it might be stuffed. If the context
doesn’t make it clear, then I am entitled to ask ‘How do you mean? Do you mean it may be stuffed or
what? What are you suggesting?’ ”.
²² It is generally believed that Dretske did not succeed in restricting closure successfully. However,
for an attempt, see (Blome-Tillmann 2006).
    17
were developed. Interpreting Dretske’s paper and also Alvin Goldman’s

(1976), who in passing suggests for the first time that the idea that what is a
relevant alternative might be context-bound, Gail Stine (1976) is the first to
systematically develop this idea. Thus, Stine claims famously that “it is an
essential characteristic of our concept of knowledge that tighter criteria are
appropriate in different contexts. It is one thing in a street encounter, another
in a classroom, another in a law court—and who is to say it cannot be another
in a philosophical discussion?” (Stine 1976: 254). According to this view,
which alternatives are relevant is a function of one’s context: while a given
alternative might be relevant in one context it does not have to be so in
another. Thus, Stine’s paper represents the first formulation and defence of a
contextualist account in the literature.
Moreover, and this is the second important point of divergence of Stine’s
paper from Dretske’s RAT, her account allowed her to retain closure. If we
keep the set of relevant alternatives fixed at a context C, we can ‘know ¬sh’—
the negation of a sceptical hypothesis—at C, simply in virtue of the fact that no
alternative to ¬sh is relevant in C. We can, according to Stine, ‘know p’ despite
the fact that our evidence does not eliminate the possibility that ¬p. For
illustration, consider the case of the sceptical hypothesis: in everyday contexts
sceptical scenarios are irrelevant alternatives and do not have to be eliminated
in order to ‘know’ ordinary propositions. The proposition that we are not
brains in vats can thus be ‘known’ in everyday contexts despite the fact that
our evidence does not eliminate the possibility that we are brains in vats.
Those propositions are simply ‘known’ by default in everyday contexts.²³
Despite the fact that Stine’s account is contextualist in spirit, she does not
ascend semantically and distinguish explicitly between ‘knowledge’ and know-
ledge, and also does not claim explicitly that it is the predicate ‘knows p’ that is
context-sensitive. It is worth noting at this point, however, that the theoretical
framework to make these distinctions was only beginning to emerge when
Stine wrote in the 1970s. One early advocate of contextualism who was aware of
the need for semantic ascent, the need to distinguish between knowledge on the
one hand and the semantics of ‘knowledge’ on the other, was David Lewis.²⁴
²³ Stine (1976: 259) summarizes her view as follows: “(1) With respect to many propositions, to
establish a knowledge claim is to be able to support it as opposed to a limited number of alternatives—
i.e., only those which are relevant in the context. (2) With respect to many propositions—in particular
those which are such that their negations are not relevant alternatives in the context in question—we
simply know them to be true and do not need evidence, in the normal sense, that they, rather than their
negations, are true.”
²⁴ Ironically, Lewis (1996) himself explicitly refuses to systematically ascend semantically in his
seminal paper ‘Elusive Knowledge’.
18  
In his (1979) paper ‘Scorekeeping in a Language Game’ Lewis argues that

context-sensitivity is a rather widespread phenomenon in natural languages
and briefly discusses epistemic modals. Lewis argues that epistemic modals are
context-sensitive and points out that, in that respect, they resemble the verb
‘knows p’. Only a few years after Lewis’s seminal paper, Peter Unger (1984: 52)
discusses the semantics of ‘knows p’ in great detail and also proposes, similar
to Stine’s view, a contextualized relevant alternatives theory. Here is Unger:
“What competitors are relevant? Like the standards for ruling out taken as
being in force, that, too, depends on the context and, thus, on our interests in
the situation” (Unger 1984: 47). More importantly, Unger (1984: 8–9) also
coined the terms ‘contextualism’—which he conceived of as giving a semantic
explanation of the shiftiness of ‘knowledge’-attributions—and ‘invariantism’—
which, in Unger’s view, must give a pragmatic explanation (in terms of
conversational implicatures) of the mentioned shiftiness.²⁵ Unger (1984) claims
further that there is no principled way to decide between the two types of
view—a position that he dubs ‘philosophical relativity’.
The next important step in the development of contextualism was Stewart
Cohen’s (1988) paper ‘How to be a Fallibilist’, in which Cohen broke new
ground in developing a detailed account of the contextual mechanisms
involved in determining the content of ‘knows p’ at a context, and in providing
more substantial epistemological underpinnings for the view. Cohen, for
instance, was the first to explicitly speak of ‘knows p’ as an indexical expression
(1988: 97): “[T]he theory I wish to defend construes ‘knowledge’ as an
indexical. As such, one speaker may attribute knowledge to a subject, while
another speaker denies knowledge to that same subject, without contradic-
tion.” Cohen is, moreover, the first to apply the contextualist idea to the
resolution of sceptical puzzles as sketched above (Section 1.3). While
Dretske’s RAT was concerned with blocking the sceptical argument or with
refuting the sceptic, Cohen’s ambitions went further. One of the main goals of
his (1988) paper was to resolve the sceptical puzzle, which, according to
Cohen, involves accounting for our intuitions that we ‘know’ propositions in
ordinary contexts, seem to not ‘know’ them in the context of epistemological
discussions, while at the same time respecting our closure intuitions. The
contextualist semantics for ‘knows p’ he proposes in his (1988) paper is
intended to do exactly that.
²⁵ For pragmatic explanations of our truth-value intuitions in examples such as the Bank Case, see
Chapter 10.
    19
After Cohen’s paper and throughout the 1990s and 2000s, contextualism
became in increasingly popular position that attracted lively and controversial
discussion. The view was, for instance, defended in a plethora of papers by
Stewart Cohen (1988, 1998a, 1998b) himself and alternative versions of EC
were developed, most famously, by Keith DeRose (1992, 1995, 1996). In fact,
the next important turn in the debate was DeRose’s (1992) paper, which
introduced the influential Bank Cases and thereby offered an argument for
EC on the basis of linguistic data—independent of the philosophical motiv-
ation uncovered by Cohen (for inscrutable reasons, Cohen’s Airport Case
turned out less influential than DeRose’s Bank Case). A few years later, in
1996, David Lewis then published his enormously influential article ‘Elusive
Knowledge’, which popularized EC even further by systematically developing
a relevant alternatives version of epistemic contextualism.
From 1996 on, in a flurry of philosophical activity, a variety of different
versions of EC have been proposed and different ways to model and imple-
ment the semantics of ‘knows p’ have been discussed. In the late 1990s and
early 2000s, however, a backlash against contextualism emerged. Lead by
influential critics such as Jessica Brown, John Hawthorne, Stephen Schiffer,
Jason Stanley, and Timothy Williamson, the linguistic consequences and
commitments of EC became a topic of intense scrutiny. Contextualists have
responded to these challenges, and contextualism is a view that is, despite an
ongoing controversial discussion, nowadays fairly widely accepted amongst
professional philosophers.²⁶ Amongst the contemporary defenders of EC, some
of whose views we shall look at in subsequent sections, are Elke Brendel, Mark
Heller, Nikola Kompa, Ram Neta, Jonathan Schaffer and Zoltan Szabo,
Jonathan J. Ichikawa, and Michael Blome-Tillmann. The view has, however,
also attracted a large amount of critical attention, mainly by authors such as
Schiffer (1996), Patrick Rysiew (2001), Jeremy Fantl and Matthew McGrath
(2002), John Hawthorne (2004), Jason Stanley (2005b), John MacFarlane
(2005a), Timothy Williamson (2005b), Crispin Wright (2005), and, more
recently, Jennifer Nagel (2010b), to name just a few. The discussion is as lively
as ever at the time of writing with the philpapers.org archive comprising over
900 articles in the subcategory ‘epistemic contextualism’. The discussion of
epistemic contextualism has become part of the philosophical mainstream.
²⁶ Over 40 per cent of respondents to Bourget and Chalmers’ 2009 PhilPapers survey ‘accept or lean
toward contextualism’, while only 31.1 per cent ‘accept or lean toward invariantism’, 25.9 per cent
‘other’, and only 2.9 per cent ‘accept or lean toward relativism’. For further discussion, see (Bourget and
Chalmers 2014).
20  
1.7 Suggestions for Further Reading
• General introductions and overview pieces on epistemic contextualism

include (Pynn 2016) and (Rysiew 2016).
• The most classical readings on contextualism are presumably (Lewis
1979, 1996), (Cohen 1988), and (DeRose 1995).
• For more recent versions of EC, see (Blome-Tillmann 2009c, 2014) and
(Ichikawa 2011a, 2017).
• Discussion of individual cases can be found in (DeRose 1992) and (Cohen
1999). For an excellent discussion of cases from the point of view of
experimental philosophy, and for further references to experimental
work, see (Grindrod et al. 2019).
• For the interaction between contextualism and scepticism, see (Cohen
1998a) and chapter 4 of (DeRose 2018), as well as chapter 2 of (Blome-
Tillmann 2014). For critical discussion, see (Feldman 1999).
• The topic of EC and closure is discussed in detail in (Lawlor 2005), in
chapter 6 of (Blome-Tillmann 2014), in (Lasonen-Aarnio 2017), and in
chapter 5 of (DeRose 2018).
2
Semantic Implementations
The general contextualist idea that the truth-values of ‘knowledge’-

attributions may vary with the ascriber’s context can be semantically mod-
elled in different ways. I shall here consider models that have been proposed
in the literature and briefly evaluate each of them. While epistemic context-
ualists are often somewhat vague about the semantic implementation of their
views, the issue is in fact of central importance. As we shall see in Chapter 3,
whether we can find an empirically plausible and attractive contextualist
way to model the semantics of ‘knows p’ is of crucial importance in
responding to some of the most widely discussed objections to epistemic
contextualism (EC).
2.1 Indexicality: Character and Content
Let us begin with some general background. According to a widely accepted

picture of semantics, the semantic content of a declarative sentence at a
context is a proposition, which in turn uniquely determines the sentence’s
truth-value relative to a circumstance of evaluation. Moreover, on the picture
presupposed here semantic contents are compositional—that is, the content of
a complex expression is a function of the contents of its constituents and of the
way in which they are combined. The content of a subsentential expression is,
as a consequence, its contributions to the propositions expressed by the
sentences they occur in. Given the compositionality of content the content
of a sentences S—that is, the proposition it expresses—can vary with context in
virtue of one constituent of S varying its content with context: if one of the
constituents of S changes its content with context, S will change its content—
and thus its truth-conditions—with context, too.
There are a number of familiar expressions that cause contextual variations
in content and truth-conditions. In contemporary philosophy of language and
linguistics the phenomenon is familiar under the label of indexicality. As
David Kaplan (1989) has pointed out, a number of English expressions

22  
make different contributions to the propositions expressed by the sentences

they are embedded in. In particular, Kaplan focuses his attention on a small
number of expressions that I shall refer to as ‘core indexicals’—namely, the
pronouns ‘I’, ‘he’, and ‘she’, the demonstratives ‘this’ and ‘that’, and a number
of adverbs such as ‘here’, ‘now’, ‘today’, ‘yesterday’, and ‘tomorrow’. The
content (or referent) of each of these expressions is determined by the context
of utterance. If you assert ‘I’m hungry’ the sentence uttered expresses a
different proposition than if I assert it. Similarly, if I utter the sentence ‘It’s
sunny today’ on Friday, the sentence I use expresses a different proposition
than if uttered on Saturday. The contents of sentences containing indexicals
vary with context.
Following David Lewis, Kaplan (1989) offered an intriguing and attractive
account of the semantics of indexicals. According to Kaplan’s analysis, two
different types of meaning are associated with natural language expressions.
First, what Kaplan calls an expression’s character; and, second, its content.
While the content of a declarative sentence is the proposition it expresses at a
context and the content of a subsentential lexical item its contribution to the
proposition expressed by the sentence at the context, character is more closely
associated with an expression’s conventional meaning or its lexicon entry.
According to Kaplan, character is precisely that aspect of meaning of an
expression that is known by competent language users. Content, on the
other hand, is that aspect of meaning that a particular expression has as
interpreted on the backdrop of a particular context. Thus, the character of ‘I’
in English can be paraphrased roughly as ‘the speaker at the context of
utterance’, whereas its content in a particular context will usually be a
person—namely, the context’s speaker.
What is crucial about Kaplan’s view, however, is that an expression’s
character determines, together with an utterance context, the expression’s
content. Characters are functions from contexts to contents. For instance, the
character of ‘I’ is a function that takes a context C and returns the speaker at C,
while the character of ‘now’ is a function that takes a context C and returns the
time of utterance of C. The character of a subsentential expression together
with the context C determines the expression’s content in C. The content of a
subsentential expression is the expression’s contribution to the content of the
sentence in which it is embedded. If an expression is itself a declarative
sentence S, then its character determines, in conjunction with a context C, a
proposition p as the content of S in C, which in turn has a truth-value relative
to a circumstance of evaluation—which we can think of as an ordered pair of a
possible world w and a time t. A proposition is, on this view, true at a world w
:    23
Context
Character
of Utterance
jointly determine
(indexical resolution)
Circumstance of
Content
Evaluation
jointly
determine
Truth-Value
Figure 1. The Kaplan/Lewis picture—character, content, and circumstance of

evaluation
at a time t. The general ideas underlying the Kaplan/Lewis picture can thus be
illustrated as in Figure 1.
Once the distinction between character and content is in place, we can
define indexical terms as expressions that have an unstable character—that is,
a character that does not map all contexts on the same content. Non-indexical
expressions, on the other hand, are expressions whose character is stable in the
sense that their characters are constant functions—functions that map all
contexts on the same content. In what follows it will be useful to have an
explicit definition of indexicality along these lines:
Indexicality:
An expression e is an indexical iff it has an unstable Kaplan character.
The above definition allows for non-referential terms to be indexicals. While

the paradigm examples of indexicals discussed in the literature are usually
referential terms such as ‘I’ and ‘here’ or the demonstratives ‘this’ and ‘that’,
the above definition is perfectly compatible with the idea that non-referential
terms such as adjectives (‘tall’, ‘flat’, ‘local’, ‘nearby’) or quantified noun
phrases (‘everybody’, ‘nobody’) are indexicals. In what follows I shall, there-
fore, use the term ‘indexical’ in this wider, non-referential sense—that is, as
describing the phenomenon of having an unstable Kaplan character.
24  
2.2 Classical Contextualism
As we have just seen, some sentences have contextually variable truth-values

because they have contextually variable contents, where the variability in
content is to be traced back to the unstable Kaplan character of one of their
constituents. One important and influential way to model EC rests on this
Kaplanian analysis of indexicality. The idea is that the truth-values of
‘knowledge’-attributions vary with context because ‘knows p’ is an indexical
in Kaplan’s sense: it has an unstable character and thus changes its content
with context. I shall call this view indexicalism or classical contextualism:
Indexicalism (Classical Contextualism):

‘Knows p’ has an unstable Kaplan character.
According to indexicalism, there is no such thing as the semantic content of

‘knows p’. Rather, the content of ‘knows p’—and thus the content of sentences
containing it—depends on the utterance context. Indexicalism accordingly
explains the contextualist’s claim that the truth-values of ‘knowledge’-attributions
vary with the utterance context by postulating that their contents (the proposi-
tions they express) vary with the context of utterance—independently of whether
they contain further indexicals, are ambiguous, or context-sensitive in any other
way. Consider the chart in Figure 2 in which the distinctively epistemic context-
sensitivity of ‘knows p’ affects first the content (in virtue of indexical resolution)
and then the truth-value of ‘knowledge’-attributions. The distinctively epistemic
effects of the context of utterance flow, as it were, through the chart along the
dashed lines.
As the term ‘classical contextualism’ suggests, indexicalism is the standard
semantic model of epistemic contextualism. In fact, EC has traditionally
simply been equated with indexicalism and even today almost all participants
in the debate have in mind indexicalism when discussing the view. EC is
usually simply understood to be the view that the truth-values of ‘knowledge’-
attributions vary with context because their contents vary with context.
However, as we shall see below, there is an alternative way to model the
context-sensitivity of ‘knows p’, without assuming that the predicate ‘knows
p’ is an indexical. The uncritical identification of EC with indexicalism is
therefore somewhat misleading. Before discussing non-indexical versions of
EC below in Sections 2.3 and 2.4, let us distinguish between three different
versions of indexicalism and consider their respective strengths and
weaknesses.
  25
Context
Character
of Utterance
jointly determine
(indexical resolution)
Circumstance of
Content
Evaluation
jointly
determine
Truth-Value
Figure 2. Classical epistemic contextualism
Classical contextualism comes in three major different implementations,

what I shall call strict indexicalism, scalar indexicalism, and quantificational
indexicalism. The different ways to model classical contextualism can be
distinguished according to the analogy to more recognized indexicals that
they pursue. The difference between the first of these views and the second two
is that the indexicality is traced back to a hidden variable in the logical from of
‘knowledge’-attributions on the latter two views but not on the former. Let us
consider strict indexicalism, the least plausible of the views, first.
2.2.1 Strict Indexicalism
According to strict indexicalism, ‘knows p’ functions very similarly to the

core indexicals ‘I’, ‘here’, ‘she’, and ‘tomorrow’. On the view at issue, ‘knows
p’ simply denotes different contents in different contexts, where these con-
tents are construed as (more or less complex) relations between subjects and
propositions. Thus, much like ‘I’ denotes different people in different con-
texts and ‘here’ different places, ‘knows’ denotes different knowledge-
relations. I shall call this view strict indexicalism, because there is, according
to the view, a rather close or strict analogy between core indexicals and
‘knows p’:
26  
Strict Indexicalism:
‘Knows p’ denotes different relations in different contexts.¹
While, on the face of it, strict indexicalism offers an attractive model for the
contextualist, the view has fallen into considerable disrepute recently. The
reasons why shall be discussed in more detail in Chapter 4. The main line of
argument against strict indexicalism, however, rests on the observation that
‘knows p’ behaves semantically quite differently from core indexicals. Without
going into too much detail at this point, remember the above discussion of the
contextualist’s error-theory and the phenomenon of semantic blindness.
According to EC, competent speakers are sometimes unaware of the fact
that ‘knows p’ has changed its content with context. Competent speakers of
English, however, are usually fully aware of the fact that core indexicals, such
as ‘I’ and ‘here’, change their content with context. The analogy between
‘knows p’ and core indexicals isn’t a very good one.²
2.2.2 Scalar Indexicalism: Gradable Adjectives
Some expressions of natural language English are more subtly indexical than
core indexicals. Epistemic contextualists have, therefore, thought of them as
better candidates for modelling the context-sensitivity of ‘knows p’. One
analogy that has historically been particularly popular amongst contextualists
is the analogy between ‘knows p’ on the one hand and gradable adjectives such
as ‘flat’, ‘empty’, or ‘tall’ on the other. How can we model the context-
sensitivity of such adjectives?
It is often held among semanticists that the contents of gradable adjectives
have semantic links to scales measuring the gradable property associated with
the adjective at issue. For instance, the content of ‘tall’ is taken to have a link to
a scale of height, the content of ‘flat’ a link to a scale of flatness, and the content
of ‘empty’ a link to a scale of emptiness.³ According to such scalar analyses of
gradable adjectives, (1) is to be analysed as in (2), where ‘δF’ denotes a function
mapping objects onto values of a scale of flatness and the variable ‘vminFC’
¹ See (Schaffer and Szabó 2014) for this definition.

² Keith DeRose (2009: 25, 188, 259) sometimes writes as if he favours strict indexicalism, but in fact
intends, in his (2009), to stay neutral with respect to the semantic implementation of EC.
³ See (Kennedy 1999) for such an account of gradable adjectives.
  27
denotes a value on that scale separating the domain of ‘flat’ into its positive
and negative extension in context C:⁴
(1) A is flat.
(2) ≥ (δF (A); vminFC).
More intuitively, (2) is to be read as follows:
(3) The value A takes on a scale of flatness is at least as great as the minimal
value required for satisfying ‘flat’ in context C.
According to the scalar analysis, positive ‘flat’-ascriptions have a logical form

similar to the logical form of comparative ‘flat’-ascriptions. To see this, note
that the scalar analysis assigns the logical form as depicted in (5) to the
comparative statement (4):
(4) A is flatter than B.

(5) > (δF (A); δF (B)).
Again, more intuitively, (5) is to be read as in (6):
(6) The value A takes on a scale of flatness is greater than the value B takes
on a scale of flatness.
According to the scalar analysis, at the level of logical form, positive

‘flat’-ascriptions comprise a contextually provided comparison value that is
unarticulated at the level of surface structure.
Similarly to the semantics of gradable adjectives, some epistemic context-
ualists have argued that ‘know(s) p’ is also associated with a hidden argument
place for standards of epistemic strength. According to this scalar account, a
‘knowledge’-ascription like ‘x knows p’ has an implicit argument place for an
epistemic standard the value of which is determined contextually. Thus, even
though no part of the English sentence explicitly denotes the argument place
for an epistemic standard, it does have a representation at the level of logical
⁴ There will probably be no definite cut-off point for any gradable adjective here, but rather an area
where it is unclear whether the adjective applies or not, i.e. a penumbra. Gradable adjectives are vague.
However, I take it that vagueness and context-sensitivity are two distinct semantic phenomena.
28  
form.⁵ To illustrate this view further, consider the following account of

‘knowledge’-attributions, according to which (7), which is to be read as in
(8), gives the logical form of ‘knowledge’-ascriptions:
(7) ≥ (δES (bx); vminKC).

(8) The value x’s true belief b takes on a scale of epistemic strength is at least
as great as the minimal value required for satisfying ‘know’ in context C.
On this view, just as the content of ‘flat’ is semantically linked to a scale of

flatness, the content of ‘knows p’ is semantically linked to a scale of epistemic
strength, the degree of epistemic strength required for a belief to satisfy
‘know(s) p’ varying with context.⁶
We are now in a position to define scalar indexicalism as follows:
Scalar Indexicalism:
‘Knows p’ is a scalar predicate that has a hidden argument place for a
contextually determined standard of epistemic strength.
We shall see below, in Chapter 4, that there are convincing objections to scalar
indexicalism. Let me mention just a few problems here. First, it is not entirely
clear how to model epistemic standards in a scalar way, since, intuitively,
standards do not always allow for a linear ordering and may thus appear
incommensurable (especially when modelled in terms of relevant alternatives).
Moreover, the analogy between ‘knows p’ and gradable adjectives, while more
attractive than the one between ‘knows p’ and core indexicals advocated by the
strict indexicalist, doesn’t carry as far as the contextualist might wish.⁷
⁵ Cp. (MacFarlane 2011b: 537): “Although no part of the English sentence explicitly denotes the
third relatum, it might be denoted by an ‘aphonic’ element in the deep syntax [logical form], or the
speaker might simply expect hearers to be able to fill in the blank using contextual cues.”
⁶ The versions of EC to be found in (Cohen 1988), (DeRose 1995), and (Cohen 1999) can be read
along these lines. So it is not quite clear how DeRose intends to square his account with the idea that it
is a version of strict indexicalism, but see also fn. 7 below.
⁷ Note that, despite appearances, the strict indexicalist can in fact uphold the analogy between
‘knows’ and gradable adjectives, but that commits her to a non-standard semantics of gradable
adjectives. On the view I have in mind the predicates ‘flat’ and ‘tall’ have different properties as
semantic contents in different contexts: while the property of being taller than 6 foot is denoted by
‘tall’ in a context in which students are discussed, the property of being taller than 6 foot 9 inches is
expressed by ‘tall’ in contexts in which basketball players are the topic of conversation. Such a view,
while a theoretical possibility, has gone out of fashion in semantics and linguistics (see (Kennedy
1999)). Note also that the view cannot explain why ‘know(s) p’ and gradable adjectives behave
differently from core indexicals.
Another random document with
no related content on Scribd:
Abschiednehmen von den Angehörigen, als ginge es direkt in den
Schützengraben und ins Granatfeuer.
Unter Palmen am Euphratufer.
Achtes Kapitel.
Im Reich der Palmen.
N och zwei Tage hielt ich mit dem deutschen Geschwader

gleichen Schritt, dann aber gab ich das Rennen auf und ließ die
so angenehme Reisegesellschaft im Stich. Ihr lag daran, so schnell
wie möglich flußabwärts zu kommen, und sie konnte die geselligen
Abendstunden durch Schlafen am Tage einbringen. Meine Arbeit
aber erforderte angestrengte Aufmerksamkeit über Tag und Ruhe bei
Nacht. Bei dem Dorf Do-er verabschiedeten wir uns, und am Morgen
des 25. Aprils lag meine Fähre wieder einsam am Ufer. Als die
Sonne über dem Horizont heraufstieg, stand ich auf dem Gipfel des
nächsten Kalksteinfelsens. Vor mir breitete sich die Wüste aus, flach,
öde und unfruchtbar. Kleine Stücke hübsch geschliffenen
Feuersteins lagen überall umher. Kein lebendes Wesen war zu
sehen. Bald trieb mich die zunehmende Hitze wieder hinab zu dem
kühlen Strom unter den Schatten meines Zeltdaches. Der Frühling
hatte endlich gesiegt; gestern Mittag zeigte das Thermometer über
25 Grad im Schatten, und bei der kleinen Stadt Mejadin hatten uns
die ersten Palmen daran erinnert, daß wir uns wärmeren Gegenden
näherten. Noch standen sie vereinzelt, aber am folgenden Tage
zeigten sich bei dem Dorfe Rawa die ersten Palmenhaine, und
abwärts von Ana, das ich am 27. April erreichte, bilden diese
köstlichen Oasen den herrlichen, das Auge immer wieder
entzückenden Schmuck des Euphratufers. Mit ihrem Auftreten wird
zugleich die Schiffahrt gefährlicher, denn die Bewässerung der
Palmengärten erfordert umfangreiche, oft weit in den Strom
vorstoßende, die Fahrstraße einengende Wasserschöpfwerke.
Das erste Wasserschöpfwerk.

Schon am fünften Tag meiner Euphratfahrt, dann am 18. April
kurz vor dem Dorfe Säbcha, war ich an den ersten, noch primitiven
Schöpfwerken, Denkmälern einer tausendjährigen Tradition,
vorübergekommen. Damit sich das Berieselungswasser über das
Ufer verbreiten kann, muß es ziemlich hoch hinaufbefördert werden;
dazu eignet sich nur ein Ufer, das sich etwa drei Meter über den
Wasserspiegel erhebt und senkrecht ausgewaschen ist. Wo es fehlt,
muß es künstlich geschaffen werden. Die Schöpfeinrichtung besteht
aus einem kunstlosen Holzgerüst, in dessen Oberteil ein Seil über
eine Rolle läuft. Nach dem Wasser zu hängt daran ein großer
Ledersack; das innere Seilende wird von einem Zugtier — Ochse
oder Pferd — oder durch Menschenkraft landeinwärts gezogen und
so der Wassersack emporgewunden. Seine Öffnung ist durch ein
Holzkreuz aufgespannt, damit er sich ordentlich füllen kann, und
wenn er oben ankommt, ergießt er durch einen rüsselartigen, durch
eine zweite Rolle und ein dünneres Seil regulierten Fortsatz seinen
Inhalt in den Anfang des Bewässerungskanals, ein kleines, aus
Weidenruten geflochtenes, mit Lehm gedichtetes Becken. Nach
fünfzehn oder zwanzig Schritt kehrt das Zugtier um, der Sack sinkt
durch seine Schwere wieder hinab und füllt sich aufs neue. So geht
das den ganzen Tag. Häufig zwingt die Laune des Flusses dazu, die
Schöpfmaschine zu verlegen. Steigt der Strom und überschwemmt
er die Ufer, so können Tiere und Menschen ruhen; die befruchtende
Flut findet von selbst ihren Weg zu den Äckern, und die eigentlichen
Bewässerungskanäle stehen dann wie schmale Hafendämme über
der weiten Wasserfläche.
Das Wasser wird mit Hilfe von Ochsen oder Pferden gehoben.
Die knarrende Musik dieser Wasserhebewerke hatte mich von
Tag zu Tag begleitet, und ihre derben Holzgerüste glichen besonders
bei Nacht den Skeletten vorweltlicher Tiere. Bald sah ich auch zwei,
drei Säcke an jedem Holzgestell auf- und abwärts steigen; jeder
Sack hatte sein Zugtier und seine menschliche Bedienung, und jede
Rolle knirschte ihre eigene Melodie.
Im Bezirk Chreta sah ich das erste Paternosterwerk. Pferde mit
verbundenen Augen drehen ein wagerechtes Rad; dessen Zähne
greifen in ein senkrechtes Rad, über das die Kette mit den
Wassereimern läuft. Solch eine Einrichtung heißt Näura oder Dolab,
während die einfachere mit den Ledersäcken Dschird genannt wird.
Nicht weit hinter Do-er stand mitten in der Strömung ein einsamer
Pfeiler aus gebrannten Ziegeln. Sale behauptete, das sei der Rest
eines alten Turmes; wahrscheinlicher aber war es das Überbleibsel
eines der großen Wasserschöpfwerke, wie ich sie bald in vollem
Betrieb zu sehen bekam.
Wasserschöpfwerk.
Am Morgen des 26. Aprils war der Fluß wieder um 20 Zentimeter

gestiegen, und von den überschwemmten Weizenfeldern waren am
Abend feuchte Nebel und Myriaden von Mücken aufgestiegen, so
daß ich zum erstenmal mein Moskitonetz hatte gebrauchen müssen.
Wir glitten in ruhiger Fahrt an der schroffen Bergwand Baghus
vorüber und kamen dann an eine große, mit Gras und Gebüsch
bewachsene Insel, die den Euphrat in zwei Arme teilte. Wir bogen
links ein und sahen hier zum erstenmal die Wasserkraft selbst zum
Heben des Wassers ausgenutzt. Eine Steinmauer war in den Fluß
hineingebaut; ihre Spitze bildete ein großes Rad von etwa acht
Meter Durchmesser. In der Peripherie des Rades hingen längliche
Tontöpfe, die bei der Umdrehung das Wasser schöpften und in eine
Rinne entleerten, die rechtwinklig von dem über den Kamm der
Mauer laufenden Kanal ausging. Jetzt stand das Rad still, weil das
Wasser über seine Achse gestiegen war; fließt es unter der Achse,
dann dreht der Strom das Rad Tag und Nacht, und unermüdlich
ergießen die Schöpfeimer ihren kostbaren Inhalt in die Kanalrinne,
wenn auch immer etwas daneben fließt, sobald sich der
Schöpfeimer bei der Umdrehung zu neigen beginnt. Das Rad ist aus
rohem Treibholz zusammengezimmert, die Speichen sind krumm
und schief, aber wenn es nur halbwegs rund ist, erfüllt es seine
Aufgabe.
Ein Paternosterwerk unter einem Maulbeerbaum am Euphratufer.

Solcher Wasserschöpfwerke oder Dolabs sah ich auf meiner
Weiterfahrt im Reich der Palmen eine Unzahl. Je nach ihrer Länge
sind die Mauern durch Bogenwölbungen unterbrochen, damit sie
dem ungeheuren Druck besonders bei Hochwasser standhalten und
der Strom ungehindert hindurch kann. Ich zählte bis zu siebzehn
solcher Mauerbogen und bis zu vier an kleineren Mauerarmen
untergebrachten Rädern an einem Dolab. Mauern und Räder sind im
Laufe der Zeit grünschwarz geworden von Flechten und Moosen. Oft
sind sie auch nur noch Ruinen; der Strom hat die Räder, manchmal
auch nur die Töpfe zerbrochen oder mit sich geführt, die Mauern
zerstört, und nur einzelne Pfeiler halten sich noch inmitten der
brausenden Strudel. Auf einem solchen einsamen, vor Raubtieren
sichern Pfeiler nistete ein Storchenpaar. Am Anfang der Mauer am
Ufer steht meist eine Palme oder mehrere wie eine Schildwache.
Wasserrad eines Dolabs im Gang.

Ein Dolab mit vier Rädern.
Sind mehrere Räder am selben Mauerrahmen im Gang, so
entsteht ein fürchterliches Konzert wie von Ferkeln, Katzen und
Hyänen. Aber den Fellachen ist dieser ohrenbetäubende Lärm die
liebste Musik; da strömt Leben auf die Äcker, da gibt es üppige
Weizenernten und Brot für die Familie, und sie legen ihr Worte unter
und singen mit. Die Eingeborenen könnten mit verbundenen Augen
eine Strecke den Fluß hinabfahren, die ihnen vertraute Musik der
einzelnen Dolabs verriete ihnen, wo sie sich befänden. Wo die
Schöpfwerke zerfallen sind, kann der Ackerbau am Ufer nicht
gedeihen.
Die Dolabs werden möglichst bis in die stärkste Strömung
hineingeführt und treten gern paarweise auf, gleichzeitig auf beiden
Ufern. Das verschmälert die Fahrstraße oft bedeutend, steigert die
Stromschnelle, die Geschwindigkeit der Räder und damit die Masse
des hinaufbeförderten Wassers; man merkt deutlich, wie zwischen
zwei solchen Mauerspitzen der Strudel zunimmt und der Euphrat in
stärkeren Schlagwellen geht, und ich konnte noch von Glück sagen,
daß wir bei der häufigen Unaufmerksamkeit Sales ohne ernstlichen
Schiffbruch das Dorf Ana erreichten, das sich am Fuß der rechten
Uferfelsen, von Palmen beschattet, unendlich lang ausdehnte. Es
waren spannende Augenblicke, bis es Sale und den Ruderern
endlich gelang, zwischen zwei Dolabs, die hier dicht hintereinander
die Fahrstraße einengten, aus der reißenden Strömung
herauszukommen und, ohne gegen die nächste Mauer getrieben zu
werden, am Ufer zu landen.
Sale mit einem Ruder auf der „Kommandobrücke“.

Links die Hütte mit dem Arbeitstisch.
Straße in Ana.
Solange wir mit den Deutschen zusammengefahren waren, hatte
Sale seinen Stolz darein gesetzt, durch Geschicklichkeit und
Dienstfertigkeit seine Kameraden aus Der-es-Sor zu überstrahlen.
Jetzt, als niemand mehr da war, vor dem er paradieren konnte, hatte
sein Eifer merklich nachgelassen. Nur seine allzeit frohe Laune und
seine Sangeslust waren auf gleicher Höhe geblieben. Den ganzen
Tag schallten die gutturalen Laute und klingenden Vokale seiner
Muttersprache nur so über den Euphrat. Wenn er aber gerade nicht
sang, sein helles Lachen nicht von den Felsen widerhallte oder er
seine Gefährten nicht durch Märchen und Geschichten erheiterte,
mußte ich Obacht geben, denn dann schlief mein Kapitän gewiß in
aller Gemütsruhe am Steuerruder und ließ die Fähre treiben, wohin
der Strom sie führte. Saßen wir dann auf einer Schlammbank oder in
dem dichten Tamariskengebüsch eines schmalen Seitenarmes fest,
oder drehte sich die Fähre über einem saugenden Strudel, so tobte
und fluchte er im schönsten Arabisch auf die armen Kameraden, die
seinen Schlaf mit doppelter Arbeit bezahlen mußten. Kamen wir aber
glücklich los, sofort war er wieder eitel Sonnenschein und sang
unverdrossen seine Lieder. Machte ich ihm dann Vorwürfe, er sei ein
schlechter Kapitän und so etwas wäre seinem Vorgänger, dem
Türken Mohammed, nie widerfahren, so lachte er voll glücklichen
Übermuts und bewies mir mit sprudelnder Beredsamkeit, seine
Kurven seien Muster von Geschicklichkeit und Eleganz.
Eines Tages begann er die Leute am Ufer anzurufen: „Habt Ihr
etwas von Ben Murat gehört?“ oder „Habt Ihr Ben Murat gesehen?
Sagt ihm, daß ich in acht Tagen zurückkomme!“ Niemand kannte
Ben Murat, und Sales Kameraden wollten sich jedesmal totlachen.
Schließlich fragte ich, wer denn der vielberufene Ben Murat sei. Da
mußte er selbst so lachen, daß er das Steuer fahren ließ und in einer
Ecke zusammensank. Als er endlich wieder zu Atem kam, vertraute
er mir an: seines Wissens gebe es in der ganzen Gegend keinen
Ben Murat; er treibe nur mit den Uferleuten Spaß. Mehrere Tage
spukte Ben Murat an Bord und beruhigte sich erst, als wir Ana
erreichten. Denn hier wechselte ich zum letztenmal meine
Besatzung für die Strecke bis Feludscha.
Ahmed Apti.
Der neue Kapitän war ein Greis von siebzig Jahren, sehr
wortkarg und ernst, aber er hatte sein Leben lang den Euphrat
befahren, kannte, wie die neuen Ruderer Ismail Ben Halil und
Dschemi Ben Omar versicherten, jede Biegung, jede Insel, jede
Sandbank im Strom, ja jede Palme am Ufer, so daß es in ganz
Mesopotamien keinen Schiffer gebe, der mit ihm zu vergleichen sei.
Sale hatte mich gebeten, ihm den Traum seines Lebens zu erfüllen
und ihn nach der berühmten Stadt Bagdad mitzunehmen; auch
könne ich, meinte er, einen Koch gut gebrauchen, hütete sich aber
einzugestehen, daß er von Kochkunst keine Ahnung hatte. Ich ließ
ihn gewähren, und da meine neue Besatzung kein Wort Türkisch
verstand, leistete er mir als Dolmetscher wertvolle Dienste. Außer
dem neuen Gendarmen namens Saalman, einem Araber aus Mosul,
der seine Mutter in Bagdad besuchen wollte, erhielten wir aber noch
einen Passagier, und zwar diesmal einen Soldaten von Major von
Schrenks Batterie, die, wie ich jetzt erfuhr, nur zwölf Stunden voraus
war. Ahmed Apti, so hieß er, war auf unerklärliche Weise
zurückgeblieben und bat mich himmelhoch, ihn doch ja
mitzunehmen, damit er noch rechtzeitig seine Kameraden einholen
könne.
Kapitän Ali am Steuerruder.

Kapitän Alis Ruhm bewährte sich denn auch glänzend schon in
dem schwierigen Moment, als wir Ana verließen und um die
nächsten Dolabmauern herum die Fähre mit einer kurzen, aber
gewaltigen Anstrengung wieder in den Strom hineinbugsieren
mußten. Ruhig und seiner Sache sicher stand er am Steuerruder,
und seine braunen Falkenaugen bemerkten jede Tücke des Stromes
und jede Lässigkeit oder Dummheit der Mannschaft. Der Euphrat
stand jetzt 3,23 Meter über dem normalen Niederwasserstand, und
unser leichtes Fahrzeug wurde von der brausenden Flut so
pfeilschnell entführt, daß die wunderbare Schönheit der Ufer bei Ana
mit ihrer Palmenpracht wie ein Traum an mir vorüberflog. Nicht
minder malerisch waren kleine Inseln, die zum Schutz gegen das
Hochwasser rundherum mit Steinmauern umzäunt waren; bei tiefem
Wasserstand mußten sie wie kleine Festungen aussehen. Direkt an
Ana schloß sich das Dorf Dschemile, und die ununterbrochenen
Palmengärten schienen kein Ende zu nehmen. Erst bei Wadi Gaser
hörten sie wieder auf. Der Reichtum an Zelten, der oberhalb Der-es-
Sor die Ufer belebte, war jetzt völlig verschwunden, hier gab es nur
feste Dörfer mit Palmenhainen und Feldern und Wasserwerken oder
wüstenstille Ufer. Nomaden, die zeitweise am Flusse hausten, hatten
sich, wie ich in Ana hörte, in die Steppe zurückbegeben. Auf einer
jähen Felsenspitze des linken Ufers lag das Heiligengrab Habibi
Nedschar. Dort liege der Baumeister der Arche Noah begraben,
erklärte mir Ali, der in der Tat jede Einzelheit an den Ufern kannte,
und Sale fügte hinzu, die Arche sei ein Schahtur gewesen ganz wie
der unsrige, aber wohl ein paar Kilometer lang. „Wann war das?“
fragte ich. „Das ist mindestens schon zweihundert Jahre her“,
antwortete Sale in tiefstem Ernst.
Mehrfach bemerkte ich auf der Wasserfläche lange, dunkle
Streifen, die sich teilten und wieder vereinten, über Wasserstrudel
hinzogen und Inseln bildeten. Erst glaubte ich, es seien verweste
Pflanzen. Dann aber zeigte sich, daß es lauter Heuschrecken waren,
tote und lebende, die in verschiedenen Stadien der Ermattung hilflos
im Wasser zappelten und schwammen. Auf ihren Raub- und
Freßfahrten hatten sie sich verflogen und waren ein Opfer des
Stromes geworden. Auch meiner Fähre hatte sich das Gesindel bald
bemächtigt; in Bataillonen saß es auf den Relingen und dem
Zeltdach; es war nicht durch die Luft gekommen, sondern ein Teil
der unglücklichen Schwimmer hatte sich an die Fähre angeklammert
und war an der Reling emporgeklettert. Hier trockneten sie nun an
der Sonne und schöpften neue Kraft nach diesem unbehaglichen
Abenteuer.
Die Hitze hatte in den letzten Tagen mächtig zugenommen; am
28. April zeigte das Thermometer fast 53 Grad, und was auf meinem
Schreibtisch der Sonnenglut ausgesetzt war, begann zu brennen. An
diesem Tage erfuhr ich durch einen türkischen Offizier, der vor fünf
Tagen Bagdad verlassen und bei Ismanije am Ufer sein Zelt
aufgeschlagen hatte, daß die Engländer bei Kut-el-Amara
rettungslos eingeschlossen seien und das Begräbnis des
Feldmarschalls unter großem militärischen Pomp stattgefunden
habe. So drang das Echo der großen Weltereignisse auch in diese
meine Stromeinsamkeit.
Am 29. April landeten wir bei dem herrlichen Garten Misban,
dessen Besitzer, ein weitberühmter Mann, seinen Sohn sandte, um
mich zum Gastmahl einzuladen. Da es aber schon zu spät am
Abend war, machte ich diesem kleinen Märchenschloß erst am
anderen Morgen einen Besuch.
Durch einen langgestreckten Vorhof kamen wir zunächst in einen
Stall, wo einige braune Vollblutstuten an den Krippen standen, und
dann in den eigentlichen Hof, wo Misbans Sohn und Diener mich
empfingen und ins Haus geleiteten. In einem großen prächtigen
Zimmer saß ein würdiger Alter, der Bruder des Hausherrn, mit
gekreuzten Beinen auf einem Teppich und las laut, mit knarrender,
eintöniger Stimme, in einem Buche, wobei er den Körper auf und ab
wiegte. Jetzt stand er auf, hieß mich willkommen und führte mich an
den erhöhten Ehrenplatz, von wo man durch vergitterte Fenster eine
herrliche weite Aussicht auf den majestätischen Strom hatte.
Schaker, 14jähriger Araber aus Ana.
Das ganze Zimmer war mit Teppichen belegt. In seiner Mitte

erhob sich ein viereckiger Herd, auf dem Palmenholz glühte und die
Kaffeekanne brodelte. Nach einer Minute trat auch Misban selber
herein, vornehm und würdig wie ein Herrscher, in weißem,
goldgesäumtem Mantel, ein weißseidenes Tuch mit Silberringen um
den Kopf. Mit der verbindlichen Höflichkeit eines reichen Arabers
erkundigte er sich nach meiner Fahrt und erzählte dann bei Kaffee
und Zigaretten von sich und seiner Besitzung.
Sein Vater hat die Oase vor siebzig Jahren angelegt; seit
zwanzig Jahren bewirtschaftet er sie selbst. Sein vollständiger Name
ist Misban Ibn Schoka. Er hat vier Söhne und sechs Diener, die mit
ihren Familien, insgesamt dreißig Personen, hier wohnen. Er besitzt
tausend Palmen und ist dabei, seine Plantagen noch zu erweitern.
Ich sah gerade eine Fähre mit Palmenschößlingen landen, die
angepflanzt werden sollten.
Misban und sein Bruder.

Der Spaziergang durch den Garten war köstlich. Treibhauswarm,
schwer und still hing die Luft zwischen Mandel-, Orange- und
Maulbeerbäumen, und die schönen zackigen Blätter der Palmen und
ihre zarten, noch weißen Datteltrauben hoben sich scharf von dem
blauen Himmel ab. Das Korn stand hoch und sollte in einigen
Wochen geschnitten werden, einen Monat später der Weizen. Erst
im September glänzen die Datteln braungelb wie Bernstein und sind
dann reif zur Ernte. In Misban gediehen auch Zucker- und
Wassermelonen, mehrere Arten Trauben, Zwiebeln und Bohnen und
viele andere Küchengartengewächse.
Die Einkünfte der Oase sind schwankend, aber im allgemeinen
gut. Die Dattelernte allein beträgt im Durchschnitt vierzig
Kamellasten, jede zwei türkische Pfund in Gold oder etwa vierzig
Mark wert. Die gesamte Ernte berechnete der Besitzer, alles
eingerechnet, auf vierhundert Pfund. Aber nur die Hälfte davon
wurde alljährlich verkauft, die andere an Ort und Stelle verbraucht.
Misban Ibn Schoka.

Misbans Garten war kein Kavekhane, kein Wirtshaus, wo man
einkehrte, um sich ein Abendbrot zu bestellen. Aber wer vorüberfuhr,
war Misbans willkommener Gast, und es verging kaum ein Tag, an
dem nicht jemand eine Weile auf den bequemen Sofas vorne am Kai
rastete, bis die Sonne unterging und Abendkühle eintrat. Zuweilen
kamen sogar die Karawanenleute von der einige Stunden entfernten
großen Landstraße zwischen Bagdad und Aleppo herüber, um hier
auszuruhen und ihre Tiere zu tränken. An glühenden Sommertagen,
wenn die Hitze über der trockenen Wüste zittert und der feine Staub
von den Tritten der Kamele aufwirbelt, muß es allerdings ein
Hochgenuß sein, von der Uferhöhe aus Misbans Palmen ihre
Kronen über reifenden Äckern wiegen zu sehen. Da winkt Ruhe, da
kann man sich satt trinken und im kühlen Schatten die Mühsal der
Wüste vergessen! —
Ein Schahtur landet am Ufer der Oase Misban.

Eins der schönsten Landschaftsbilder, das der Euphrat zu bieten
hat, ist die kleine Stadt Hit, ein uralter Ort, dessen Asphaltquellen vor
Jahrtausenden das Erdpech lieferten, das die Baumeister der
babylonischen Königspaläste brauchten. Über einem wogenden
Palmenmeer thront sie auf einem Hügel, an dessen Fuß ein Minarett
trotzig seine weiße Spitze erhebt. Sie ist die erste arabische Stadt,
die Seehandel treibt und an deren Kai sich ein regelrechtes
Schiffsleben entwickelt. Am Kopf der Schiffbrücke, die beide Ufer
verbindet, liegen zahlreiche arabische „Meheile“, große Kähne mit
spitzen Vorder- und Hintersteven, schräg stehenden Masten und
langen zusammengerollten Rahesegeln vor Anker, die in voller Fahrt
mit vom Wind geblähten weißen Segeln, Schaum am Vordersteven,
von überaus malerischer Wirkung sind. Matrosen waten im
Uferwasser geschäftig hin und her. Mächtige Schollen zähflüssiges

The Semantics of Knowledge Attributions Michael Blome Tillmann Full Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Semantics of Knowledge Attributions Michael Blome Tillmann Full Chapter

Uploaded by

Copyright:

Available Formats

The Semantics of Knowledge

Attributions Michael Blome-Tillmann

The Semantics of Knowledge Attributions

PART I: EPISTEMIC CONTEXTUALISM

PART II: EPISTEMIC IMPURISM

PART III: EPISTEMIC RELATIVISM

8. Epistemic Relativism 145

PART IV: STRICT INVARIANTISM

9. Psychological Invariantism 157

10. Pragmatic Invariantism 173

PART V: PRESUPPOSITIONAL EPISTEMIC

11. The Presupposition Effect 193

1. The Kaplan/Lewis picture—character, content, and circumstance

Take another example. The commonsensical epistemologist says: ‘I know

Forty-two years later, the semantics of ‘knowledge’-attributions has

The Semantics of Knowledge Attributions. Michael Blome-Tillmann. Oxford University Press.

v. Presuppositional Epistemic Contextualism

1.1 What is Epistemic Contextualism?

¹ The example is derived from the zebra case in (Dretske 1970).

The Semantics of Knowledge Attributions. Michael Blome-Tillmann. Oxford University Press.

What is going on in this example? According to our intuitions, an utterance

    5

Epistemic Contextualism (EC):

Given semantic compositionality—the view that the extensions of complex

Epistemic Contextualism (EC*):

According to the above deﬁnitions, the predicate ‘know(s) p’ adds context-

   7

One further terminological matter should be mentioned here. The negation

1.2 Evidence for Contextualism

in Mary’s and John’s context. Moreover, it seems as though the practical

Similar to Cohen’s example, DeRose’s case suggests that it is more difﬁcult to

   9

encroachment accounts discussed in Chapters 6 and 7—have the explanatory

1.3 Sceptical Puzzles

(iv) I know that I have hands.

¹² The following is a description of standard contextualist views on sceptical puzzles, as it can be

Anti-Sceptical Intuition (ASI):

1.4 Error Theory and Contextual Shifts

     13

we, as competent speakers of English, be somehow sensitive to or aware of the

¹⁵ Cf. (Cohen 1988: 106; DeRose 1995: 40).

Different contextualists have different stories to tell about what epistemic

One interesting aspect of the contextualist solution of sceptical puzzles is that

Single-Premise Closure (CL):

¹⁷ See, for instance, (Lasonen-Aarnio 2008).

    15

more precise and contextualized formulation of the single-premise principle to

Contextualized Single-Premise Closure (CLC):

(CLC) is a meta-linguistic principle. Loosely speaking, (CLC) says that (CL),

1.6 The History of Contextualism

The development of epistemic contextualism took place in several waves and is

¹⁹ See (DeRose 1995; Lewis 1996; Cohen 1999).

defend an ambiguity theory of ‘knows p’ when he distinguishes between what

    17

were developed. Interpreting Dretske’s paper and also Alvin Goldman’s

In his (1979) paper ‘Scorekeeping in a Language Game’ Lewis argues that

    19

1.7 Suggestions for Further Reading

• General introductions and overview pieces on epistemic contextualism

The general contextualist idea that the truth-values of ‘knowledge’-

2.1 Indexicality: Character and Content

Let us begin with some general background. According to a widely accepted

The Semantics of Knowledge Attributions. Michael Blome-Tillmann. Oxford University Press.