![i
i
i
i
i
i
i
i
OUP CORRECTED PROOF – FINAL, //, SPi
introduction
so, he doesn’t give a theory about me or Reginald or anybody else. Saying that “I” just
picks out whoever happens to be speaking isn’t tantamount to saying that anybody
is unimportant. Kaplan’s theory of demonstratives does not imply that Reginald isn’t
special. So much is, I take it, pretty obvious.
Nevertheless, when epistemologists start thinking about contextualism about
“knows”—roughly, the thesis that “knows” is similar to words like “I” and “me” in
that its referent depends on the conversational context—the corresponding point is
not always treated as quite so obvious. For example, echoing Reginald, I have often
encountered a perceived tension between contextualism about “knows” and the idea
that knowledge is important. I encounter this perception more often in conversation
than in print, but Alvin Goldman does give a brief expression to a version of it here:
A popular view in contemporary epistemology (with which I have much sympathy) is that
knowledge has an important context-sensitive dimension. The exact standard for knowledge
varies from context to context. Since it seems unlikely that natural kinds have contextually
variable dimensions, this renders it dubious that any natural kind corresponds to one of our
ordinary concepts of knowledge. (Goldman , p. )1
Assuming that the “popular view” in question is contextualism, Goldman’s fallacy
is the same as Reginald’s: it is a use–mention error. There is no straightforward
connection between the semantic properties of the English word “knows” and the
metaphysical properties of knowledge. (Compare the fact that there is no straight-
forward connection between Kaplan’s observations about indexicals and Reginald’s
beliefs about himself.)2
This book is about the relationship between contextualism about “knows”, on the
one hand, and epistemological theorizing about knowledge, on the other. It is a mistake
to think that there is any very straightforward connection between them, but is there
a subtler one? I shall suggest that there is. In particular, I will argue that there is a
mutually supporting package of views, combining a particular brand of contextualism
about “knows” with a particular interpretation of the “knowledge first” program,
according to which knowledge is a theoretically fundamental and important mental
state. Assuming contextualism, the sentence, “knowledge is a theoretically funda-
mental and important mental state” may be a context-sensitive one—this sentence
could be used to express different propositions in different conversational contexts.
That doesn’t mean it isn’t true and informative and theoretically enlightening, or that
1 In work in progress, Dani Rabinowitz also defends a version of Goldman’s argued incompatibility.
2 Compare also the remarks of Jenkins (, p. ): “Note that none of this context sensitivity [about
‘explanation’ and related terms] gives us any reason to be mind-dependent anti-realists about explanation.
What depends on contexts (and hence on the intentions, interests and so on of the utterer and/or audience)
is what is expressed by terms like ‘good explanation’ and ‘best explanation’, not what counts as a good or
bad explanation once these aspects of meaning are settled”. The thought is also related to Keith DeRose’s
“intellectualist” motivation for contextualism—see DeRose (, Ch. ). However, I will suggest in
Chapter that this motivation is not mandatory for contextualists.](https://image.slidesharecdn.com/3319219-250525220426-f4f3b9fe/85/Contextualising-Knowledge-Epistemology-And-Semantics-1st-Edition-Jonathan-Jenkins-Ichikawa-17-320.jpg)
![i
i
i
i
i
i
i
i
OUP CORRECTED PROOF – FINAL, //, SPi
introduction
roles posited for knowledge—perhaps one need invoke knowledge in order to explain
evidence, or justification, or mental content, etc.
If a picture along these lines is right, then finding such roles for knowledge will
constitute evidence that knowledge is a relatively fundamental state. To be sure,
the Lewisian picture about fundamentality is controversial; adjudicating the serious
metaphysical question about the relationship between theoretical roles and funda-
mentality is well beyond my present scope.7 Even setting such relations aside, the
idea that knowledge connects in deep theoretical ways with action, justification, belief,
evidence, etc. is itself of significant interest.
If contextualism is correct, then the sentence, “knowledge is a relatively funda-
mental state” may itself be a context-sensitive sentence. (I’ll explain contextualism
in much more detail in Chapter .) Contextualists need to have something to say
about how it is intended. As will emerge throughout the book, I think that different
proposed theoretical roles for knowledge ought to be treated differently. There is a kind
of general perception among epistemologists that a knowledge first stance fits poorly
with contextualism. I think this is a mistake.
Contextualism “Evading” Epistemology?
One preliminary worry has to do with the general relevance of contextualism to
epistemology. In effect, it starts with the observations I made above about the
independence of claims about “knows” ascriptions from claims about knowledge,
and concludes from this that insofar as we care about the latter, we ought to
ignore the former. As Ernest Sosa puts it, “[c]ontextualism replaces a given question
[about knowledge] with a related but different question [about ‘knows’ ascriptions]”
(Sosa , p. ). And it is not at all clear that an answer to the latter question will
bear on the former. To adapt one of Sosa’s own pithy examples, Patience might say to
herself:
I am very confident that people often utter truths when they say “Somebody loves
me.” But does anybody at all love me?8
In exactly the same way, one may be convinced by the truth of contextualism, and
rest easy that people often utter truths when making knowledge ascriptions;9 still,
one might wonder whether anybody at all knows anything. (If you’re having a hard
time seeing how this could be an open question, let’s stipulate that I’m writing in
at least a moderately skeptical context now; the thought is that people utter truths
with such ascriptions only in more lax contexts.) It is important to keep clear the
distinction between object language discussion, that uses terms to talk about the world
7 For some discussion, see Jenkins ().
8 Sosa (, p. ).
9 As I use the term, knowledge ascriptions are sentences using the word “knows”.](https://image.slidesharecdn.com/3319219-250525220426-f4f3b9fe/85/Contextualising-Knowledge-Epistemology-And-Semantics-1st-Edition-Jonathan-Jenkins-Ichikawa-20-320.jpg)
![i
i
i
i
i
i
i
i
OUP CORRECTED PROOF – FINAL, //, SPi
“knowledge”
Josephine’s social class is above his own. Suppose also that Ralph loves Josephine, and
knows that he does. Naturally, Ralph believes that he loves someone of a different class.
If all were as it seems, then this belief would amount to knowledge; but in fact, Ralph
has been misinformed about each of their social positions—he himself comes from
a noble family, while Josephine’s roots are working-class. When he believes that he
loves someone of a different class, his belief is true, but it is not knowledge; Ralph is
in a Gettier case. According to invariantist orthodoxy, Ralph’s epistemic position is
sufficient for knowledge in the case where everyone’s social status is as it appears, but
not in the case where it turns out that the evidence about social status was misleading.
So when the environment is less cooperative, it appears that the “epistemic standards”
(in DeRose’s sense) necessary for knowledge are higher. This is a conclusion that
I doubt DeRose would want to accept, as he characterizes contextualism and shifty
invariantism as distinctive in allowing for differing standards governing knowledge
ascriptions.43 But it is not clear how to articulate the notion of an epistemic standard
in a way that makes the mechanisms of IRI views amount to changes in standard,
without categorizing Gettier cases in the same way.
Indeed, one needn’t appeal to Gettier cases to make this point—once the thought
is articulated clearly and laid out for examination, it’s just obvious that different
possibilities can be relevant, depending on the subject’s situation. All theorists will
accept, for instance, that whether one needs to rule out the hypothesis that the animal
before one is a cleverly disguised mule depends at least in part on whether there’s any
reason to think it might be.
Epistemic standards are sometimes glossed as matters of what purely epistemic
factors are required for knowledge or knowledge ascriptions;44 they are also some-
times described as the truth-relevant factors that are required.45 But the notions of
“purely epistemic” factors and “truth-relevant” factors are inevitably left unexplicated;
I myself have never been confident that I have a grip on what they amount to.46 There
is perhaps a response-dependent property in the neighbourhood: one might identify
“epistemic standards” as governing only those features that obviously or unsurprisingly
43 I don’t have the space to make the case here, but my view is that the arguments from e.g. DeRose
(, p. ) turn on this commitment.
44 Fantl and McGrath (, p. ) describe “impurism” as the thesis that “[h]ow strong your epistemic
position must be—which purely epistemic standards you must meet—in order for a knowledge-attributing
sentence, with a fixed content in a fixed context of use, to be true of you varies with your circumstances.”
DeRose (, p. ) likewise characterizes standards as a matter of “how good an epistemic position” one
needs to know.
45 E.g. Stanley (, p. ), Fantl and McGrath (, p. –), Grimm (), DeRose (, pp. &
), Brown (b, p. ).
46 If, as one might’ve found pre-theoretically plausible (obvious?), knowledge itself is both “purely
epistemic” and “truth-relevant”, then it is trivial that a fixed epistemic standard—viz., knowledge—is
necessary and sufficient for knowledge, in all cases. So whether or not one adopts shifty invariantism, one is
committed to the idea that there is a purely epistemic, truth-relevant property that is necessary and sufficient
for knowledge. If knowledge isn’t “purely epistemic”, or if it isn’t “truth-relevant”, then I don’t know how to
determine when those terms apply.](https://image.slidesharecdn.com/3319219-250525220426-f4f3b9fe/85/Contextualising-Knowledge-Epistemology-And-Semantics-1st-Edition-Jonathan-Jenkins-Ichikawa-47-320.jpg)
![i
i
i
i
i
i
i
i
OUP CORRECTED PROOF – FINAL, //, SPi
“knowledge”
context in question, S is the sentence, “Dick knows that Ralph is planning to elope”, w
is the world where things are as they seem, w is the world where Ralph’s conversation
was misleading, and X is the skeptical possibility that Ralph will not elope even though
the evidence is as it is in w:
. X is a possibility in which Ralph is not planning to elope, and it is uneliminated
by Dick’s evidence.
. In w, Buttercup’s utterance of S in C is true.
. So in w, X is an irrelevant possibility.
. In w, X is the case. This explains why in w, Buttercup’s utterance of S in C is
false; so X is relevant.
. So C does not fix whether X is relevant.
If this argument is right, then it would be a mistake to hold a version of Lewisian
contextualism according to which the context fixes the relevant alternatives; if context
is supposed to fix epistemic standards, then epistemic standards cannot be identified
with—or even determine—the sets of relevant alternatives. Insofar as one wishes
to characterize a plausible version of Lewisian contextualism, then, one should not
interpret the view as Jonathan Schaffer does, according to which “[a] sentence of
the form “s knows that p” is true in context c iff s’s evidence eliminates every not-p
possibility relevant in c” (Schaffer , p. ).52
There is, however, a substantive assumption in the argument given above. As
discussion of this point is somewhat technical and might be considered peripheral
to readers not invested in the debate, I’ll discuss it in its own section; some readers
may prefer to skip ahead to §..
. Modality and Knowledge Ascriptions
The argument of the previous section assumed that the very same context, C, might
occur in the relevantly distinct possible worlds w and w. The intuitive thought
underwriting this assumption was that the difference between w and w was in
some sense wholly external to the experience of Buttercup, the speaker; whether
52 Another reason it’s important to be precise about how we think about standards, and just what is
fixed by context, is that failure to do so can make contextualism appear susceptible to objections it needn’t
be vulnerable to. For example, Jason Stanley argues that contextualists can’t comfortably allow that subjects’
practical situations influence how easy it is to satisfy “know”, in cases where the speakers are unaware of
the practical importance to the subject (“Ignorant High Stakes” cases). Stanley writes that for contextualism
to give the right result, “it must turn out that the fact that there is a greater cost to Hannah’s being wrong
affects the semantic content of some of her statements, even when neither she, nor any other conversational
participant, is aware of it. So the contextualist could accommodate Ignorant High Stakes, but only at the
cost of advancing a rather dramatic claim about the potential semantic effects of non-psychological facts
about extralinguistic content.” (Stanley , p. ) Even granting for the purpose of argument that such
an implication is a costly one, the contextualist can avoid it by clarifying that the “standard” provided by
features of the context of which the speaker is cognizant itself encodes a sensitivity to the subject’s practical
situation.](https://image.slidesharecdn.com/3319219-250525220426-f4f3b9fe/85/Contextualising-Knowledge-Epistemology-And-Semantics-1st-Edition-Jonathan-Jenkins-Ichikawa-50-320.jpg)
![i
i
i
i
i
i
i
i
OUP CORRECTED PROOF – FINAL, //, SPi
“knowledge”
Since semantic context-sensitivity is traceable to an individual element, multiple occurrences
of that element in a discourse should be able to take on differing values. In the case of an
utterance such as “This is larger than this”, where two different objects are pointed to by the
person uttering the sentence, this feature is obviously confirmed. But it is present in a broader
range of constructions. (Stanley , p. )
Call the ability of context-sensitive terms to take on different values rather freely
flexibility. Stanley suggests that all context-sensitive discourse is flexible,54 and that
if “knows” is likewise flexible, then contextualism about “knows” does not enjoy the
advantages claimed for it. For example, Stanley writes,
If different occurrences of instances of “knows that p” can be associated with different epistemic
standards within a discourse, some of the paradigm sentences the infelicity of which supposedly
motivates [contextualists’] accounts over rival accounts turn out to be felicitous and potentially
true by contextualist lights. For example, if we have similar behavior to many other context-
sensitive expressions, one would expect the following to be felicitious:
() Bill knows that he has hands, but Bill does not know that he is not a bodiless brain in a
vat. (Stanley , p. , my numbering)
There are two challenges for the contextualist here. First, a major historical motiva-
tion for contextualism has been its supposed ability to resolve skeptical puzzles without
admitting the truth of sentences like Stanley’s (); if Stanley is right that contextualism
predicts () to have easy true readings, a motivation for contextualism is undermined.
Second, more straightforwardly, () is intuitively repugnant; any view that implies it
should be rejected.55
In support of his contention that Lewisian contextualism predicts sentences like
() to be unproblematic, Stanley cites sentences involving explicit quantifiers whose
domains shift mid-sentence. For example, Stanley points out that it’s not difficult to
find a reading of () according to which the two instances of “every” take different
domains:
() Every sailor waved to every sailor.
For example, if two ships have recently crossed paths, it might say that every sailor
on one ship waved to every sailor on the other. More directly to the point, Stanley also
offers this dialogue:
A. Every van Gogh painting is in the Dutch National Museum.
B. That’s a change; when I visited last year, I saw every van Gogh painting, and some
of them were definitely missing. (Stanley , p. )
54 One might perceive an obvious potential objection to this claim in, to use Kaplan’s term, “pure
indexicals”—indexicals like “I” that take their semantic value automatically, without the aid of any kind
of demonstrations. See Stanley’s pp. – for a rejoinder.
55 On both of these points, see especially DeRose ().](https://image.slidesharecdn.com/3319219-250525220426-f4f3b9fe/85/Contextualising-Knowledge-Epistemology-And-Semantics-1st-Edition-Jonathan-Jenkins-Ichikawa-53-320.jpg)
Contextualising Knowledge Epistemology And Semantics 1st Edition Jonathan Jenkins Ichikawa
- 1. Contextualising Knowledge Epistemology And Semantics 1st Edition Jonathan Jenkins Ichikawa download https://ebookbell.com/product/contextualising-knowledge- epistemology-and-semantics-1st-edition-jonathan-jenkins- ichikawa-6638438 Explore and download more ebooks at ebookbell.com
- 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Contextualizing Indigenous Knowledge In Africa And Its Diaspora 1st Edition Ibigbolade Aderibigbe Alloy Ihuah Felisters Kripono https://ebookbell.com/product/contextualizing-indigenous-knowledge-in- africa-and-its-diaspora-1st-edition-ibigbolade-aderibigbe-alloy-ihuah- felisters-kripono-51571386 Contextualising International Law In Northeast Asia Asif H Qureshi https://ebookbell.com/product/contextualising-international-law-in- northeast-asia-asif-h-qureshi-50237176 Contextualising Difficulties In Literacy Development Exploring Politics Culture Ethnicity And Ethics Gavin Reid Janet Soler Janice Wearmouth Gavin Reid Janet Soler https://ebookbell.com/product/contextualising-difficulties-in- literacy-development-exploring-politics-culture-ethnicity-and-ethics- gavin-reid-janet-soler-janice-wearmouth-gavin-reid-janet- soler-51298930 Contextualising Educational Studies In India Research Policy And Practices Pradeep Kumar Choudhury https://ebookbell.com/product/contextualising-educational-studies-in- india-research-policy-and-practices-pradeep-kumar-choudhury-27210276
- 3. Contextualising Backward Classes Discourse G Aloysius https://ebookbell.com/product/contextualising-backward-classes- discourse-g-aloysius-10652788 Contextualising Narrative Inquiry Developing Methodological Approaches For Local Contexts 1st Edition Sheila Trahar https://ebookbell.com/product/contextualising-narrative-inquiry- developing-methodological-approaches-for-local-contexts-1st-edition- sheila-trahar-51260126 New Womens Writing Contextualising Fiction Poetry And Philosophy 1st Edition Subashish Bhattacharjee Girindra Narayan Ray https://ebookbell.com/product/new-womens-writing-contextualising- fiction-poetry-and-philosophy-1st-edition-subashish-bhattacharjee- girindra-narayan-ray-51644438 Reconfigurations Contextualising Transformation Processes And Lasting Crises In The Middle East And North Africa 1st Ed Rachid Ouaissa https://ebookbell.com/product/reconfigurations-contextualising- transformation-processes-and-lasting-crises-in-the-middle-east-and- north-africa-1st-ed-rachid-ouaissa-22489180 Geosocietal Narratives Contextualising Geosciences 1st Edition Martin Bohle https://ebookbell.com/product/geosocietal-narratives-contextualising- geosciences-1st-edition-martin-bohle-36233346
- 6. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi Contextualising Knowledge
- 7. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi
- 8. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi Contextualising Knowledge Epistemology and Semantics Jonathan Jenkins Ichikawa 1
- 9. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi 3 Great Clarendon Street, Oxford, OX DP, 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 © Jonathan Jenkins Ichikawa The moral rights of the author have been asserted First Edition published in Impression: 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 Madison Avenue, New York, NY , United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: ISBN –––– Printed and bound by CPI Group (UK) Ltd, Croydon, CR YY 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.
- 10. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi Contents Acknowledgments ix Introduction . “Knowledge” . Contextualism in General . Kaplan on Character and Content . Modals . “Knows” Contextualism and Skepticism . Elusive Knowledge . Quantifiers . Lewis and Lewisian Contextualism . Epistemic Standards . Invariantism with Shifting Standards . Factivity . Modality and Knowledge Ascriptions . Differences between Knowledge and Quantifiers . Knowledge Embedded in Conditionals . Is Contextualism Ad Hoc? . Sensitivity . Two Puzzles . David Lewis on Counterfactuals . Counterfactual Contextualism . Rules for Possibilities . Karen Lewis and Ignorance of Counterfactuals . Knowledge and Sensitivity . Equivocation and Necessary Conditions . Strengthening the Antecedent . Sensitivity, Safety, and Knowledge . Evidence . Motivation for E=K . An Argument Against E=K . Evidence as Non-Inferential? . Alexander Bird and “Holmesian Inference” . Abominable Conjunctions . Non-Contextualist Responses . Contextualist E=K . The Intuitions Again . Evidence as Important . Circularity and Basic Knowledge
- 11. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi vi contents . Lewis and Cartesian Contextualism . Moorean Contextualism . Skeptical Intuitions and Moorean Contextualism . Radical Skepticism . Justification . Initial Clarifications . Desiderata for a Theory of Justification . J=K? . Justification as Potential Knowledge . Is JPK Internalist? . Contextualism . Steven Reynolds . Alexander Bird . Justification as a Normative Status . An Objection . Lotteries . History . Reliability Appendix: Impossible Knowledge, Content Externalism, and JPK . Action . Use of “Knows” . Reasons . Contextualism and Norms . Intuitive Counterexamples to Necessity . Intuitive Counterexamples to Sufficiency . More Specific Theoretical Intuitions . The Thought-Bubble Model of Practical Reasoning . Counter-Closure . Locke on Ethical Theory . Schroeder on Ethical Theory . Reason-To . Internalism and External Redundancy . An Ethical Analogy . A Challenge to Internalist KR . Contextualism and Symmetry . Internalism and Basic Knowledge . Assertion . Stanley and the Certainty Norm . The Factivity Challenge . High-Standards Assertability of Low-Standards Knowledge . DeRose . DeRose . Contextualism and Norms, Again
- 12. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi contents vii . The Method of Cases . KA and Good Enough Positions to Assert . Turri’s “Simple Test” . Incremental Assertion . Contexts and Possibilities . The Incremental Knowledge Norm of Assertion . Schaffer on Contrastivism and Assertion . Explaining Moore-Paradoxicality . Belief . Outright Belief . Shifty Data . Clarke, Sensitivism, and Belief as Credence One . Challenges for Clarke . Contextualism about Belief Ascriptions . “Knows” and “Believes” . Knowledge and Proper Belief . Doxastic States and Epistemology Bibliography Index
- 13. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi
- 14. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi Acknowledgments The kernels of this book were part of the Ph.D. dissertation I wrote at Rutgers in under the supervision of Ernest Sosa. I was thinking at the time about imagination and its many implications for epistemology—among those implications were the mental manipulations of possibilities, and the relationships between far-fetched hypotheses, evidence, and skepticism. One strand of thought from my dissertation developed into a research program on thought experiments, intuitions, and the a priori. My first book, The Rules of Thought (), co-authored with Benjamin Jarvis, was a product of that strand. Another central strand in my dissertation had to do with knowledge, evidence, and relevant alternatives. The present book is a development of that project. As is no doubt obvious, I have been enormously influenced by Timothy William- son’s Knowledge and Its Limits (OUP ) and David Lewis’s “Elusive Knowledge” (Australasian Journal of Philosophy (), –, ). Digesting those works has provided me enormous philosophical nourishment over the past several years; this book is my attempt to share the result. I am grateful to many philosophers and friends who have had useful feedback and conversations with me about the material in this book. In many cases, the book has improved directly as a result. (Many others are prompts for possible future projects.) Some of these people are: Mark Alfano, Maria Alvarez, Nomy Arpaly, Steve Bahnaman, Derek Baker, Derek Ball, Aleksey Balotskiy, Alexander Bird, Liam Kofi Bright, Jessica Brown, Yuri Cath, Nathan Cockram, Stewart Cohen, Roger Clarke, Paul Dimmock, Travis Dumsday, Jeremy Fantl, Claire Field, Daniel Fogal, Rachel Fraser, Mikkel Gerken, Jasper Heaton, Torfinn Huvenes, Alex Jackson, Benjamin Jarvis, Masashi Kasaki, Carly Kocurek, Jennifer Lackey, Maria Lasonen-Aarnio, Karen Lewis, Lauren Leydon-Hardy, Sarah Little, Clayton Littlejohn, Dustin Locke, Errol Lord, Ned Markosian, Aidan McGlynn, Matt McGrath, Robin McKenna, Rachel McKinnon, Lisa Miracchi, Veli Mitova, Chris Mole, Adam Morton, Jennifer Nagel, Shyam Nair, David Plunkett, Kathryn Pogin, Lewis Powell, Baron Reed, Brian Renne, Katernya Sam- oilova, Jonathan Schaffer, Josh Schechter, Susanna Schellenberg, Eric Schwitzgebel, Daniel Star, Chris Stephens, Daniele Sgaravatti, Ernest Sosa, Jason Stanley, Kurt Sylvan, Evan Thompson, Shelley Tremain, Servaas van der Berg, Brian Weatherson, Nathan Weston, John Wigglesworth, Timothy Williamson, Chase Wrenn, Crispin Wright, and Sarah Wright. Thanks to L. Syd Johnson for a drawing of a lion. I am especially grateful to Nate Bemis, Carrie Ichikawa Jenkins, and two anonymous referees, who each gave me thorough feedback on an earlier draft of the whole manuscript. There are many more people who ought to be on the list; I regret that I wasn’t organized and thoughtful enough to remember to include them. (Thank you,
- 15. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi x acknowledgments and I’m sorry.) Thanks also to Peter Momtchiloff for all of his help with the transition from prose to print, and to Bianca Crewe, Kyle da Silva, Graham Moore, and Phyllis Pearson for help with proofreading and indexing. I presented work in progress on various parts of this book at many conferences, workshops, and colloquia. Thanks to helpful audiences at the Arché-Logos conference, the and Joint Sessions of the Aristotelian Society and Mind Association, the meeting of the Canadian Society for Epistemology, the WCPA meeting, the APA Central division meeting, the “Gettier Problem at ” conference in Edinburgh, a epistemology conference in Kyoto, the “Factive Turn” conference in Vienna, and the CPA Annual Congress. Thanks also to workshop participants at UBC, St Andrews, Helsinki, and Northwestern, col- loquium audiences at Indiana–Bloomington, UBC, Victoria, SUNY-Fredonia, CEU, and Lingnan, and seminar participants at Arché and the Edinburgh Epistemology Reading Group. Preliminary work on this book was done at the Arché Philosophical Research Centre at the University of St Andrews, while I was part of an AHRC-funded project on intuitions and philosophical methodology; the bulk of it was written at the University of British Columbia, funded in part by a Hampton Grant. The material in this book is published here for the first time, but some of the philosophical ideas derive from ones given in previously published papers (often with changes in emphasis, and occasionally with changes in view); this will be indicated along the way. Particularly relevant papers are: • “Quantifiers, Knowledge, and Counterfactuals,” Philosophy and Phenomenolo- gical Research, (), March : –. • “Quantifiers and Epistemic Contextualism,” Philosophical Studies, (), June : –. • “Knowledge Norms and Acting Well,” Thought, , March : –. • “Basic Knowledge and Contextualist ‘E=K’,” Thought, (), December : –. • “Justification is Potential Knowledge,” Canadian Journal of Philosophy, (), July : –.
- 16. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi Introduction I begin with an analogy. Diary of a Narcissist Here is a diary entry by Reginald, a confused narcissist. Dear Diary, I am perturbed. As you know, I’ve long found myself, if not perhaps perfection, surely the next best thing to it; I thank Providence every day for raising me up so far above the vulgar. It is no exaggeration to say that hitherto, I have counted myself among the very most beautiful and significant people in all of Creation. But today I happened across a paper by a philosopher called David Kaplan. What I found there shook my deepest convictions to the core. For Kaplan argued that certain words—“demonstratives” or “indexicals”, he calls them—are context sensitive; the referents of these terms can vary according to the conversational context in which they’re used. My first instinct was a sanguine one; his seemed an interesting and plausible semantic thesis. The referent of the word “that”, for example, is simply whatever it is at which my flawless finger happens to be pointing at a given moment when I’m speaking. But it’s not just that! It’s one thing to recognize the general semantic framework—it’s quite another to make particular entries in the list of context-dependent terms. Among Kaplan’s list of context-dependent terms are the very dearest and most important to me! He includes on his list, for example, such touchstones as “I” and “me”! Can you imagine, Diary? I—Reginald the all-right—dependent on such contingencies as conversational contexts? Never in my wildest dreams would I have imagined that anyone would so trivialise me. Needless to say, I am deeply shaken. Must I accept that I am so unimportant? That there is nothing special about me, but rather than I’m merely, whoever happens to be speaking in a given conversation? The thought terrifies me. Tomorrow I shall attempt to rebut Kaplan’s defamatory arguments; tonight I am too shocked. I must rest. Fondly, Reginald Reginald’s error is not difficult to diagnose; he’s very bad at the use–mention distinction. At times, in the passage above, he is using words like “I” and “me”, thus talking about himself; at other times he’s mentioning them, thus talking about those English words. (You are a person, not an English word; “you” is an English word, not a person.) Kaplan () gives a context-sensitive theory about the word “I”; but in doing
- 17. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction so, he doesn’t give a theory about me or Reginald or anybody else. Saying that “I” just picks out whoever happens to be speaking isn’t tantamount to saying that anybody is unimportant. Kaplan’s theory of demonstratives does not imply that Reginald isn’t special. So much is, I take it, pretty obvious. Nevertheless, when epistemologists start thinking about contextualism about “knows”—roughly, the thesis that “knows” is similar to words like “I” and “me” in that its referent depends on the conversational context—the corresponding point is not always treated as quite so obvious. For example, echoing Reginald, I have often encountered a perceived tension between contextualism about “knows” and the idea that knowledge is important. I encounter this perception more often in conversation than in print, but Alvin Goldman does give a brief expression to a version of it here: A popular view in contemporary epistemology (with which I have much sympathy) is that knowledge has an important context-sensitive dimension. The exact standard for knowledge varies from context to context. Since it seems unlikely that natural kinds have contextually variable dimensions, this renders it dubious that any natural kind corresponds to one of our ordinary concepts of knowledge. (Goldman , p. )1 Assuming that the “popular view” in question is contextualism, Goldman’s fallacy is the same as Reginald’s: it is a use–mention error. There is no straightforward connection between the semantic properties of the English word “knows” and the metaphysical properties of knowledge. (Compare the fact that there is no straight- forward connection between Kaplan’s observations about indexicals and Reginald’s beliefs about himself.)2 This book is about the relationship between contextualism about “knows”, on the one hand, and epistemological theorizing about knowledge, on the other. It is a mistake to think that there is any very straightforward connection between them, but is there a subtler one? I shall suggest that there is. In particular, I will argue that there is a mutually supporting package of views, combining a particular brand of contextualism about “knows” with a particular interpretation of the “knowledge first” program, according to which knowledge is a theoretically fundamental and important mental state. Assuming contextualism, the sentence, “knowledge is a theoretically funda- mental and important mental state” may be a context-sensitive one—this sentence could be used to express different propositions in different conversational contexts. That doesn’t mean it isn’t true and informative and theoretically enlightening, or that 1 In work in progress, Dani Rabinowitz also defends a version of Goldman’s argued incompatibility. 2 Compare also the remarks of Jenkins (, p. ): “Note that none of this context sensitivity [about ‘explanation’ and related terms] gives us any reason to be mind-dependent anti-realists about explanation. What depends on contexts (and hence on the intentions, interests and so on of the utterer and/or audience) is what is expressed by terms like ‘good explanation’ and ‘best explanation’, not what counts as a good or bad explanation once these aspects of meaning are settled”. The thought is also related to Keith DeRose’s “intellectualist” motivation for contextualism—see DeRose (, Ch. ). However, I will suggest in Chapter that this motivation is not mandatory for contextualists.
- 18. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction we can only mention, rather than use it. But it does mean that when we use it, we must use it carefully, and attend to potential ambiguities. I have written a book about the relationship between contextualism about “knows” (hereafter “contextualism”) and the knowledge first program because I think both of these views have much to commend them. I have defended versions of both views in print—many of the ideas from these earlier papers are incorporated into this book. Along the way, I will say something about why I find these two views attractive, and I hope that many readers will come to look sympathetically on them, but it is not my primary purpose to argue for either contextualism or the knowledge first stance. Rather, I hope to show that these disparate views, though independently developed, and widely thought to stand in a kind of tension, in fact fit rather well together. Contextualism can help the knowledge first theorist respond to certain important objections to that stance; knowledge first can help provide a theoretical motivation for the contextualist’s claims. Any philosopher who wants to consider one of these views would do well to do so along with the other. Throughout the book I’ll canvass a series of studies of particular issues, exploring ways in which contextualism and the knowledge first stance may synergize. I include a chapter summary at the end of this introduction. First, however, I should set out the important backdrop that is the knowledge first program. Knowledge First It’s not particularly controversial that knowledge is epistemologically interesting; but in what way is it interesting? A prevalent assumption in some of the history of philosophy had it that knowledge is a central explanandum in epistemology: a central task is to explain what knowledge is, how or whether knowledge can be attained, whether and why knowledge is particularly valuable, etc. Paradigmatic of this approach is the “theory of knowledge” literature spawned by Edmund Gettier’s famous paper. The aim of this literature was to provide an “analysis” of knowledge in more fundamental terms—to explain knowledge, for example, in terms of belief, justification, truth, evidence, etc. More recently, some philosophers have attempted to approach epistemological questions concerning knowledge from a different angle. Timothy Williamson is widely credited with this change in perspective; his book, Knowledge and Its Limits, advocated for an approach he calls “knowledge first”.3 The knowledge first stance reverses the traditional order of explanation: knowledge is treated as explanans, rather than as explanandum. The idea here is that knowledge is in some sense fundamental, 3 Contemporary pre-cursors included Zagzebski (), Zagzebski (), and Craig (). There were certainly earlier precedents, e.g. Kneale (, p. ): “According to the view presented here, knowledge is sui generis and the two varieties of belief are to be defined by reference to it.” For an overview and discussion of historical knowledge first ideas, see Marion () and Mulligan ().
- 19. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction and can be used to illuminate other states of epistemological interest. For example, perhaps we should understand evidence in terms of knowledge, instead of vice versa. In our (), Carrie Jenkins and I distinguished between a number of distinct knowledge first theses. In particular, we argued for a clear distinction between the metaphysical and representational claims that travel under that banner. Metaphysical knowledge first claims are claims about knowledge itself; questions about whether knowledge is a mental state, or whether it is a (relatively or absolutely) fundamental feature of reality are foregrounded. Representational claims, by contrast, have to do with how we think about or talk about knowledge; questions about whether the concept knows has believes as a component, or whether knowledge ascriptions typically or invariably proceed by virtue of belief ascriptions, for instance, characterize the issues in discussing representational knowledge first theses. Although Williamson defends views of both families, and although some authors have argued for strong connections between them, Jenkins and I argued that the views are prima facie independent. The version of the “knowledge first” program I am interested in exploring in this book belongs to the metaphysical family. I am more interested, for example, in the idea that knowledge is a mental state that has theoretically significant roles in explaining things like action, belief, and justification, than I am in the idea that knows is a mental state concept, or the question whether knowledge ascriptions proceed via tacit belief ascriptions. Throughout the book, I will be interested in exploring ways in which knowledge connects to other areas. My sympathies lie with the broad methodology of Lewis (), according to which appearance in good theorizing about the world is a mark of fundamentality. Lewis himself focused on the perfectly natural; his view is that the perfectly natural properties are the ones that appear in the basic elements of a minimally adequate theory of the world. In my view, it is reasonably natural to extend that thought to the idea that relatively fundamental properties appear in less fundamental, but genuinely real, theories.4 The idea of the knowledge first project, so interpreted, is that knowledge is more metaphysically fundamental than, for example, was supposed in the project of trying to explain knowledge in terms of justification and belief. This metaphysical claim should be understood in a way similar to the way that many of us think that green is metaphysically prior to grue, or that electrons are metaphysically prior to laptop computers. The approach is plausible to the degree to which metaphysical theorizing in terms of knowledge is fruitful. This is why, for example, Timothy Williamson’s suggestion that citing knowledge typically makes for a better explanation than does citing belief, in the explanation for someone’s action;5 that it appears in the best theory of action explanation is some reason to think knowledge is of some theoretical significance, hence metaphysical fundamentality.6 So likewise for other theoretical 4 The same, I think, goes for Ted Sider’s more recent treatment of similar themes—see Sider (, p. ). 5 Williamson (, pp. –). 6 See Ichikawa and Jenkins (, §§., ) for further discussion.
- 20. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction roles posited for knowledge—perhaps one need invoke knowledge in order to explain evidence, or justification, or mental content, etc. If a picture along these lines is right, then finding such roles for knowledge will constitute evidence that knowledge is a relatively fundamental state. To be sure, the Lewisian picture about fundamentality is controversial; adjudicating the serious metaphysical question about the relationship between theoretical roles and funda- mentality is well beyond my present scope.7 Even setting such relations aside, the idea that knowledge connects in deep theoretical ways with action, justification, belief, evidence, etc. is itself of significant interest. If contextualism is correct, then the sentence, “knowledge is a relatively funda- mental state” may itself be a context-sensitive sentence. (I’ll explain contextualism in much more detail in Chapter .) Contextualists need to have something to say about how it is intended. As will emerge throughout the book, I think that different proposed theoretical roles for knowledge ought to be treated differently. There is a kind of general perception among epistemologists that a knowledge first stance fits poorly with contextualism. I think this is a mistake. Contextualism “Evading” Epistemology? One preliminary worry has to do with the general relevance of contextualism to epistemology. In effect, it starts with the observations I made above about the independence of claims about “knows” ascriptions from claims about knowledge, and concludes from this that insofar as we care about the latter, we ought to ignore the former. As Ernest Sosa puts it, “[c]ontextualism replaces a given question [about knowledge] with a related but different question [about ‘knows’ ascriptions]” (Sosa , p. ). And it is not at all clear that an answer to the latter question will bear on the former. To adapt one of Sosa’s own pithy examples, Patience might say to herself: I am very confident that people often utter truths when they say “Somebody loves me.” But does anybody at all love me?8 In exactly the same way, one may be convinced by the truth of contextualism, and rest easy that people often utter truths when making knowledge ascriptions;9 still, one might wonder whether anybody at all knows anything. (If you’re having a hard time seeing how this could be an open question, let’s stipulate that I’m writing in at least a moderately skeptical context now; the thought is that people utter truths with such ascriptions only in more lax contexts.) It is important to keep clear the distinction between object language discussion, that uses terms to talk about the world 7 For some discussion, see Jenkins (). 8 Sosa (, p. ). 9 As I use the term, knowledge ascriptions are sentences using the word “knows”.
- 21. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction more broadly, and the meta-language, that mentions terms in order to discuss them.10 So it is that Sosa raises the challenge of how it is that contextualism can be relevant for epistemology. Hilary Kornblith has gone so far as to argue that contextualism is “largely irrelevant to epistemological concerns.”11 Contextualism has two kinds of responses available to this worry. One of them is the thesis of this book: that a holistic theory combining contextualism and the uncontroversially epistemologically-relevant knowledge first program is appealing; if contextualism helps make that program more plausible, it is contributing helpfully to epistemology. But there is also a more schematic reply available, which I’d like to articulate now. I agree with Sosa’s claim that contextualism doesn’t bear in any direct way on standard epistemological questions about the nature and extent of human knowledge. But I dispute the inference from this point to its irrelevance for the latter. For even though it doesn’t provide any straightforward evidential support for any particular epistemological view, contextualism is, if true, crucially important for the methodology of epistemology. Anyone interested in understanding knowledge has an interest in thinking clearly about knowledge, and if contextualism is true, then equivocation on “knows” is possible. So if contextualism is true, epistemologists must exhibit sensitivity to this fact.12 Take for example a classical skeptical argument like this one: I have no way to tell whether or not I will unexpectedly drop dead tonight; therefore I don’t know much at all about what I may or may not do tomorrow. Its premise enjoys some intuitive plausibility, but its conclusion is far more skeptical than most epistemologists want to admit. It is tempting to suppose that one has to choose between these attractive ideas; then nonskeptics are burdened with the task of explaining away the attractiveness of the initial claim of ignorance. But if contextualism is right, then one needn’t reject either intuitive starting-point; the argument to the effect that one must is equivocal. Careful attention to the language we use is sometimes the only way to avoid confusion about that which our language is about.13 This is in effect the same observation that Timothy Williamson makes in a different context in The Philosophy of Philosophy: Philosophers who refuse to bother about semantics, on the grounds that they want to study the non-linguistic world, not our talk about that world, resemble scientists who refuse to bother 10 The meta-language uses some terms—consider for example this sentence: “‘Me’ is an example of a context-sensitive term”. This sentence mentions, rather than uses the word “me”, which is why it is about that word instead of me (i.e., Jonathan). But it also uses the word “example” and six other words. 11 Kornblith (, p. ). As an OUP referee points out, in addition to the critique given in the main text, it seems that Kornblith’s argument depends on the dubious (to understate things!) assumption that consideration of skepticism exhausts epistemology’s concerns. 12 See DeRose (, pp. –) for a version of this point. 13 Indeed, Sosa himself recognizes as much in different contexts. Compare his remarks in Sosa (, p. ) about a different kind of semantic issue: “Semantic ascent does have a place in epistemology if only when we attempt to understand persistent disagreement by appeal to ambiguity or context-dependence.”
- 22. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction about the theory of their instruments, on the grounds that they want to study the world, not our observation of it. Such an attitude ... produces crude errors. (Williamson , pp. –) I do not wish to deny that there is also an important sense in which contextualism is not the answer to deep epistemological questions—I’ll discuss this in much more depth in Chapter . But there is every reason to expect it to be of relevance.14 Context-Sensitive Normative Discourse A second preliminary objection to my project is more specific to the interaction between contextualism and knowledge norms. A significant component of the know- ledge first stance relates knowledge to normative concepts. For example, Williamson () argues that knowledge is the constitutive epistemic norm of assertion, which implies something like this: (N) If and only if S knows p, S is epistemically permitted to assert that p Other knowledge norms have also been proposed, many of which I will consider in detail in this book. But all of them, according to the objection I am now considering, fit badly with contextualism. The objection runs like this: any contextualist who adopts a knowledge norm is committed to problematic contextualism about normative concepts. Suppose, for example, that a contextualist endorses (N). Since the left- hand side is context-sensitive, by the contextualist’s lights, our contextualist must be a contextualist about the right-hand side too. But (the objection continues) it is not plausible to endorse this kind of normative contextualism in the cases at issue.15 The argument is fallacious; it semantically ascends and descends freely, and a contextualist need not accept it. Semantic ascent is the move from an object-language claim to a metalanguage claim, as in the move from the claim that somebody loves me to the claim that an utterance of “somebody loves me” is true; semantic descent is a matter of disquotation, inferring an object-language claim from a metalanguage claim: the utterance of “somebody loves me” is true, therefore somebody loves me. Neither move is generally valid for context-sensitive discourse. (Suppose that nobody loves me, but that some people love Patience. And suppose further that Patience is the person who says “somebody loves me”. Then that sentence is true as uttered, even though nobody loves me.) Returning to the case of (N), observe first that it is stated in the object language; it uses words like “knows” and “permitted”; it does not mention them. It is not about 14 Sosa also gives another argument against the epistemic significance of contextualism in his paper—one that relies on the assumption that contexts in which one engages in epistemology are inevitably skeptical ones relative to which all or nearly all knowledge ascriptions are false. I dispute this assumption in §.. See also Blome-Tillmann (, pp. –). 15 Williamson (b) and Hawthorne (, pp. –) each give arguments in this neighborhood against contextualism. While both these authors have subtle things to say about the relationship between contextualism and knowledge norms, I think both underestimate the extent of the contextualist’s resources.
- 23. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction these words. (There are no quotation marks in (N).) Therefore the move from (N) and contextualism about the left-hand side to contextualism about the right-hand side is invalid. Only a strong metalinguistic generalization of (N) could have these kinds of consequences. In fact, there are at least five options for contextualists who accept object-language claims like (N): . Decline to endorse any metalinguistic principle, instead interpreting (N) as holding in a particular favored context. . Interpret the object-level claim as applying in S’s context, as does DeRose () about assertion. . Endorse the principle in full metalinguistic generality, positing context- sensitivity in the normative language, as per the objection’s suggestion. . Endorse the principle in full metalinguistic generality, positing context- sensitivity in the relevant “assert” language. . Endorse the principle in full metalinguistic generality, holding that the context- invariant right-hand side applies any time the left-hand side is true in any context. These represent a catalog of available tools to the contextualist; there is no reason a contextualist need adopt a uniform treatment for all knowledge norms or other theoretical principles involving knowledge. I will apply diverse strategies through the book, applying a version of () to the knowledge norm of assertion, a version of () to certain proposed connections between knowledge and rational action, and a version of () to the equation of knowledge and evidence, and to the relationships between knowledge, justification, and belief. The point of this section was to give one flavor of the kind of tension that some theorists have perceived between contextualism and knowledge first epistemology. Several similar tensions (and some less similar ones) will be explored throughout the book. Outline of the Book Although contextualism and the knowledge first project have developed independ- ently, and are typically thought to stand in tension with one another, I will argue that this perceived tension is illusory. On the contrary, I mean to make the case in this book for the idea that contextualism and the knowledge first project are complementary: each has something to offer the other. The details will come over the course of the book, but one common theme will be that contextualism helps the knowledge first project to avoid counterintuitive consequences, while the knowledge first project can help certain forms of contextualism answer the challenge that they are ad hoc. I begin in Chapter with a development of the contextualist semantics I prefer. It is inspired by and related to David Lewis’s relevant alternatives approach to knowledge
- 24. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction ascriptions—it descends from an approach I first developed in Ichikawa (a). According to the Lewisian idea, the context-sensitivity of knowledge ascriptions is modeled on the context-sensitivity of modals or sentences involving quantifiers— satisfying “knows p” requires evidence that conclusively rules out “all” not-p cases, but where the domain of the “all” depends in part on the conversational context. I’ll argue in that chapter that a version of this idea captures quite a lot of the intuitive data, and avoids some of the challenges that have been leveled against contextualists. I’ll also argue that there is a significant class of under-explored linguistic data, concerning the interaction of knowledge ascriptions with conditionals, which further motivates this kind of contextualist approach. Chapter picks up on the idea that knowledge requires a certain kind of coun- terfactual connection to the truth. This rather natural idea played significant roles in twentieth-century theorizing about knowledge, but is almost entirely discredited today—not least because it is thought to countenance “abominable conjunctions”. However, the contextualist approach developed in Chapter , when combined with an independently-motivated contextualist approach to counterfactual conditionals, allows the proposed connection to be seen in a new light. I’ll argue that sensitivity, suitably understood, is a genuine necessary condition for knowledge; given con- textualism, counterintuitive consequences—including abominable conjunctions— can be avoided. In Chapter , I turn to two pressing questions about evidence—one concerns Timothy Williamson’s famous suggestion that a subject’s evidence comprises all and only her knowledge (“E=K”). I agree with the letter of this equation (suitably embed- ded within contextualism); but I also suggest that there is good reason to recognize a privileged category of basic knowledge/evidence. This motivates the second central question of the chapter: what is the best way to understand basic evidence? The contextualist approach described in Chapter had it that the truth of a knowledge ascription requires evidence that rules out all relevant counter-possibilities—but this could be understood in terms of various approaches to evidence itself. One of the under-recognized commitments of David Lewis’s contextualism—and of quite a lot of contemporary philosophy—is a kind of “Cartesian” approach to basic evidence, whereby a subject’s subjective experiences are the things that are known first and best. I’ll challenge this assumption in this chapter, arguing that it’s certainly not mandatory, and very plausibly ill-motivated. I’ll consider instead a kind of neo-Moorean disjunct- ivist approach to evidence, according to which those things a subject can see for herself—including propositions about the external world—are basic evidence playing a foundational role. Consequently, the relationship between contextualism and radical skepticism ends up on my view rather different than that which most contextualists so far have posited. Chapter develops a knowledge first theory of epistemic justification within the contextualist framework. Its starting point is the approach to justification defend- ed in Ichikawa (), according to which justification is a matter of “potential
- 25. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi introduction knowledge”—I characterized this notion then as a matter of being intrinsically identical to a possible subject with knowledge. Chapter generalizes from that approach in two ways: First, it motivates and explicates a metasemantic generalization of that view. The result is a contextualist semantics for ascriptions of epistemic justification to match that given for knowledge. Second, it relaxes the internalist assumption I’d previously made, considering various ways to understand the notion of “potential knowledge”, corresponding to various conceptions of basic evidence. One upshot of this generalization is a new theoretical understanding of the kind of approach that could even motivate radically externalist theories of justification, such as the identification of justification with knowledge. The final three chapters turn to proposed knowledge norms. Chapter considers and defends the knowledge norm of practical reasoning, and suggests that, contrary to many authors’ claims, it does not have radical externalist implications about rational action; relatedly, it also doesn’t, when combined with contextualism, require any sort of implausible contextualism about normative discourse about action. I’ll suggest that arguments to the contrary often tacitly assume a highly questionable approach to practical reasoning, which I call the “thought bubble model”. By contrast, in Chapter , I will defend a kind of subject-relative version of the knowledge norm of assertion, justifying and systematizing the connection by reference to Stalnakerian models for conversational contexts. Finally, Chapter defends a contextualist semantics for belief ascriptions, and argues for a systematic normative relation between belief and knowledge, and between “believes” and “knows”.
- 26. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “Knowledge” What is contextualism about knowledge ascriptions, and why might someone want to be a contextualist?1 This chapter lays out contextualism in general (§§.–.), and develops (§§.–.) the particular quantifier-based approach to “knows” contextual- ism I favor. In §§.–. I will go into some details on my preferred understanding of “epistemic standards”. (Readers interested only in the broad shape of the view and the connection to knowledge first epistemology might find it efficient to skip those sections.) I’ll conclude the chapter with a response to an objection to quantifier- based forms of contextualism from Jason Stanley (§.), a semantic motivation for the view (§.), and some remarks on the importance, within this framework, for a consideration of the theoretical roles of knowledge and their relations to contextualism (§.). The rest of the book will be devoted to the exploration of such connections. . Contextualism in General Contextualism about a given bit of natural language is the view that that bit of natural language is context-sensitive. The minimal characterization of contextualism I shall use is this: if a term is context-sensitive, then a sentence containing it will express distinct propositions in different conversational contexts.2 It is easiest to introduce context-sensitivity by way of example. The most uncontroversially context-sensitive terms in English are the indexicals, for instance: • I/you/he/she/him/her • this/that/it • here/there • now/then 1 As I use the term, knowledge ascriptions are sentences using the word “knows”. I’ll discuss my terminological conventions further in §. below. 2 A less minimal characterization of the context-sensitivity of a term X would have it that X itself takes on different semantic values in different conversational contexts. I do not prefer this more committal account, in part because I think views like that of Schaffer () about “knows”, or of Stanley and Szabó () about quantifiers, are best counted as contextualist, even though they hold that “knows” and quantifiers respectively always contribute the same semantic content. (They satisfy the more minimal statement of context-sensitivity because they posit tacit argument places which can be filled by conversational contexts in different ways.)
- 27. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” A given use of the word “she” will refer to some person; but which person is referred to depends on the conversational context. If you ask me how Josephine is doing, and I utter, “she is sad”, I have used those words to say that Josephine is sad. But if I’d uttered the very same words in a different conversational context—say, one in which you’d asked me how Buttercup was doing, those words would not have meant that Josephine is sad; they’d’ve meant that Buttercup is sad. So the sentence “she is sad” might be about Josephine, or it might be about Buttercup, and it’s the conversational context that determines which person it’s about in a given instance. We can think of the person the sentence is about as the referent of a given instance of “she”—since the referent depends on the context in which the word is spoken, contextualism about “she” is true. It is wholly uncontested that indexicals like “she” are context-sensitive. It is only slightly more controversial to extend the treatment to gradable adjectives and quan- tifiers.3 Gradable adjectives are terms that signify properties that come in degrees; the degree necessary to satisfy the semantic value of a given instance of the term will depend on the context in which the term appears. For example, “rich” is a gradable adjective—there are many degrees of wealth, and some of them would suffice for “rich” in the sailors’ conversational contexts, but not in the Lord Commissioner’s. Quantifiers are terms like “all”, “some”, and “everybody”; these are typically understood to range over a domain of quantification that is determined in part by the conversa- tional context. Consider, for instance, a sentence like “everybody is singing”. Its truth requires that every member of a certain domain be singing; but which people are included in the domain—which people “count” as part of “everybody”—depends on the conversational context. In one conversation, it may include the twenty members of the opera company who are currently on stage; there, it is true, as all present are now singing. In another, it may include the baritone who is not on stage during this scene; there, its truth depends on whether he happens now to be singing in his dressing room. In a third context, the domain for “everybody” may include the hundreds of people sitting quietly in the audience; there, it is false. A Sophist might try to make this all sound very mysterious: how can it be that everybody is singing (here indicating the people on stage), even though at the very same time not everybody is singing (here indicating the audience)? If one is unfamiliar with context-sensitivity, and is not careful to distinguish the level of use from the level of mention, one might be bullied into supposing this to show that there are no absolute facts about whether everybody is singing, or that whether everybody is singing depends on which people one is thinking about, or that strictly speaking, one always speaks falsely when saying, “everybody is singing”. But once we understand 3 What dissent there is comes from defenders of “minimal semantics” like Cappelen and Lepore () and Borg (). (Though note that even one of those authors since endorsed orthodoxy in this matter in Cappelen and Hawthorne ().)
- 28. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” these basic tenets of context-sensitivity, the confusions implicit in this sort of reasoning become evident.4 . Kaplan on Character and Content David Kaplan has developed the orthodox framework for context-sensitive language. Kaplan () distinguishes two aspects of meaning, which he calls character and content. The character of a word is its linguistic meaning—it is the sort of thing one might find in a dictionary; it’s that of which one has tacit knowledge by virtue of being a competent speaker of a language. Returning to the example of the previous section, the character of the word “she” is something like, “the female who is salient in the context of the conversation”. The content of a word is its referent—in a sentence about Buttercup, the content of “she” is Buttercup herself.5 In the case of context- sensitive terms like “she”, the content varies according to context. Formally, character is understood as a function from contexts to contents; contents are referents. For example, the character of “I” is the function that returns the speaker in a given context; the content of a particular instance of the word is the person speaking. With Kaplan, I will use “context” to describe the relevant features of the speaker’s conversational situation that play roles in determining contents for context-sensitive language. So the context will include things like the time of the utterance, the speaker, facts about what is being demonstrated, etc. If “knows” contextualism is correct, then the context will also include something like an “epistemic standard”— see §. below. These are technical notions, being used in specific ways. These uses are reasonably well-established, albeit idiosyncratic relative to their ordinary uses. In particular, “context” in colloquial English often just means “situation”. So, for instance, a “contextual theory of personality development” has it that the development of a subject’s personality depends on the subject’s environment.6 This kind of approach is very different from a contextualist theory of knowledge ascriptions—it is not, for instance, any kind of semantic claim about language.7 Throughout this book, I will be reserving “context” for its distinctively linguistic use. The usage I am adopting is, I think it’s fair to say, standard in epistemological discussions of contextualism, but it hasn’t always been so, and it isn’t universally 4 Unger () argues that many gradable adjectives, like “flat”, actually have empty (or nearly-empty) extensions for just this reason. The standard diagnosis, with which I agree, is that Unger is mistaken in declining to invoke the context-sensitivity of these terms to resolve the apparent puzzles. 5 Kaplan does not write univocally about content; he sometimes identifies contents with individuals, as I do in the main text, but sometimes describes them as intensions. See Braun (, §.) for discussion. 6 L’Abate and Bryson (, p. ). 7 Compare also Rachel McKinnon’s invocation of the “context-sensitivity” of assertability—as she makes clear, this is not related to the kind of “context-sensitivity” at issue in contextualism. See McKinnon (, p. ).
- 29. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” adopted even now.8 One sometimes reads epistemologists discussing a view they call “subject contextualism”, distinct from one they call “attributor contextualism”. The latter is their name for what I am calling “contextualism”, but “subject contextualism” uses “context” in the colloquial way gestured at above: it emphasizes features of the subject’s situation. For example, Michael Williams’s “inferential contextualism” is a “subject contextualist” view in that it holds that whether one’s belief amounts to knowledge depends on certain features of one’s inferential and dialectical situation— whether, for instance, one has had one’s belief challenged in certain ways.9 Similarly, David Annis’s () “A Contextual Theory of Epistemic Justification” holds that whether a belief is justified depends on what kinds of doubts have been raised and by whom. I do not prefer to use the name “contextualism” in a broad way that includes such approaches. Indeed, it is not clear that there is anything distinctive about the suggestion that the “subject’s context” is relevant for whether she has knowledge or justification. Rather, once we make explicit that by “context” we mean only “situation”, that suggestion looks to be the mere truism that whether a subject has knowledge depends on the subject’s situation. But of course that is true. For example, whether a subject knows p depends on whether her evidence in favor of p is misleading. No doubt, the so-called “subject contextualists” intended to be expressing the more specific idea that certain interesting features of the subject’s situation are relevant; but this is an extremely different kind of view than that of “attributor contextualism”.10 So I think it’s with good reason that epistemologists tend not to use this language in this way any more. (The current standard term, “subject-sensitive invariantism”, at least does not suggest false similarities with contextualism, but in my view it’s still not particularly clarifying; see §..) I will therefore follow the convention deriving from the s according to which “contextualism” refers to a distinctively linguistic thesis. Also following Kaplan, I will use “index” to describe the situation relative to which a proposition is to be evaluated. For relatively simple engagement with relatively simple sentences like “Ralph is talking to her”, the index, just like the context, will just be something like the actual world and time.11 The context provides that “she” refers to Josephine, so the sentence says that Ralph is talking to Josephine; the index is the actual world and time, and what is said is true just in case the actual world, at that time, is such that Ralph is talking to Josephine. But sometimes we will wish to consider 8 Neta (a, p. ) is one recent invocation of the “subject contextualist” and “attributor context- ualist” language. 9 See Williams () and Williams (a). See Pritchard () for helpful discussion and clarification of Williams’s view (including the label “inferential contextualism”). 10 Pritchard (, p. ) argues that Williams’s view deserves the name “contextualist”, not because it is subject contextualist—he observes as I do that there is nothing distinctive about that commitment— but because of a particular emphasis on particular features of the subject’s situation. But I am not sure why an emphasis on those features represents a point of similarity with “attributor-contextualism”. (In fact, the particular features he cites amount to differences between Williams’s view and that of the “attributor- contextualists” Pritchard discusses.) 11 Whether time should be considered part of the index or not is contested. See e.g. Brogaard () in favor of time-indexed propositions, and Partee () against them. See also Schaffer () for a more comprehensive overview of the debate. This controversy doesn’t matter for my purposes.
- 30. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” whether what is said is true relative to different indexes. We might encode this into the language itself, by using an index-shifting operator—for example, if we say “it is possible that Ralph is talking to her”, the context provides Josephine for “her” just as before, but now we consider whether there is any possible world in which that content is true. Or we may consider the same question given the simpler sentence; when one says, in the actual world (talking about Josephine), “Ralph is talking to her”, one can consider, not only whether that really is true, but whether it could be true (is it true at any index?), or whether it must be true (is it true at every index?), or whether it will be true an hour from now (is it true at the index of the actual world and : p.m.?), or whether it would be true if Josephine had gone ashore this morning (is it true at nearest worlds in which she did?). As highlighted in the Introduction, it is crucial, in thinking clearly about context- ualism, to attend carefully to the distinction between use and mention. Far too often, discussions of contextualism are less than fully explicit on this point—indeed, the time it took to unravel “subject contextualism” from contextualism may naturally be thought of as an instance of this difficulty.12 Throughout this book, I shall endeavor to be as meticulous as possible. Unfortunately and inevitably, with this kind of rigor comes a degree of unloveliness. In my view, in an academic monograph of this sort, the aesthetic price is worth paying; I hope the reader will be convinced to agree. (At any rate, I beg the reader’s indulgence.) I will follow the standard convention of using quotation marks to signify that a word is being mentioned; when I write “knows” without quotation marks, I am using the term: I know that Keith DeRose is a contextualist who thinks “knows” is context-sensitive. For further clarity, I will reserve “say” for a relation to a proposition, and “utter” for a relation to a sentence. I’ll use “knowledge ascriptions” to refer to the assertoric utterance of sentences using “knows”.13 So according to contextualism, what one says when one utters a knowledge ascription will depend on one’s conversational context. . Modals Before turning to “knows”, I’d like to consider one more example of context-sensitivity that does not have anything obviously closely to do with epistemology: modals.14 12 Other instances of this kind of shortcoming are manifest in different ways in Lewis (), discussed in §. (see also Cohen (, fn , p. )), Buford (), discussed in §., and DeRose (), discussed in §.. See also McKinnon (, p. ) (“This is the heart of epistemic contextualism: whether an agent knows depends on the context.”) and Reed (, p. ) (Contextualism implies “that whether someone can correctly be said to have knowledge can change abruptly, even when there is no change in her evidence or reliability.”). Many more examples could easily be added. 13 It’s convenient, if ungraceful, to apply these terms alike to positive sentences like “Hebe knows that Joseph is hungry” and negative ones like “Joseph doesn’t know that Ralph plans to elope”; talk of “knowledge ascriptions” is meant also to cover such cases of knowledge denials. 14 There are relationships between modals and epistemology. One clear example concerns epistemic modals—the epistemic senses of “might” (i.e., Josephine might be outside—I can’t hear her sleeping) and knowledge. As many authors have observed, epistemic modals are naturally understood in terms of knowledge: what “might” be the case is what is consistent with what is known. See e.g. von Fintel and Gillies
- 31. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Modals are terms like “must”, “may”, “necessarily”, “possibly”, “probably”, “ought”, “should”, “often”, “always”, etc. In fact, it is not obvious that these need be construed as a separate category from those discussed above; one possible way—indeed, the way that I like—to understand modals is to think of them as quantifiers ranging over possible worlds. Like quantifiers generally, modals are context-sensitive. Consider a modal sentence like (): () Bill must inspect the winch. We can understand the “must” in this sentence as a universal quantifier over a contextually-influenced set of possible worlds. In some conversational contexts, () will receive a deontic reading, quantifying over those worlds in which Bill performs his duties—it says that in all such worlds, Bill inspects the winch. In other contexts, () might receive a nomic reading; here, the worlds in question might be worlds in which the laws of nature are as they actually are, and Bill’s physical constitution is as it actually is. To say that “Bill must inspect the winch” in such a context would be to say that these features entail that Bill inspects the winch. (Imagine a context in which the relationship between free will and determinism is being considered.) With a bit more imaginative work, we can even contemplate a conversational context in which the relevant set of worlds is the set of metaphysically possible worlds in the sense of Lewis () or Kripke (); then () would say that in all metaphysically possible worlds, Bill inspects the winch. (This is not particularly plausible.) Or () can receive an epistemic reading, if its conversational participants are trying to work out what Bill’s job on the ship must be; as their evidence rules out various hypotheses, they might eventually conclude that “Bill must inspect the winch”—in all worlds compatible with their evidence, Bill inspects the winch. We can think of these categories—deontic, nomic, metaphysical, epistemic—as flavors of modality. But even having settled on a flavor, there is a good case to be made that there is additional flexibility for context-sensitivity at play. Suppose, for instance, that Hebe has been told that Bill performs one of three jobs on the ship, and she’s observed the other two jobs being carried out by other crewmen; it would not be at all surprising under such circumstances to hear her utter (), and, supposing that the relevant appearances were not deceptive, there is a reasonably strong intuition that she could say something true by doing so. Nevertheless, there are possible conversational contexts in which she might proceed more cautiously. She’s been told that Bill performs one of these three jobs, but isn’t it possible that this was a lie? If such possibilities are (). Much of the debate about epistemic modals concerns whose knowledge is relevant here. A rather natural approach is a contextualist one like that of Dowell (); the dialectic between such views and competitors—especially relativist competitors like Egan et al. ()—is very rich, and quite similar to the parallel debate about “knows”; but engaging it in any detail is beyond my present scope. For an overview and defense, see Dowell (). As Anderson () points out, there is also flexibility with regard to which of a given subject’s knowledge is relevant.
- 32. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” considered, we might well say that () is false—that Hebe would say something true by uttering: () Bill might not inspect the winch. Even the distinctively epistemic “must”, it seems, is subject to stricter and laxer standards. This brings us rather close to the “knows”-contextualist’s diagnosis of skepticism. Let us consider it now. . “Knows” Contextualism and Skepticism To adopt contextualism about “knows” is to hold that “knows” is context-sensitive; that sentences involving it express different propositions in different conversational contexts. Consider sentence (): () Joseph knows that the Captain has a daughter. According to contextualism, () will express distinct propositions in distinct con- versational contexts. In one “ordinary” context, () requires only that Joseph has some relevant familiarity with the Captain’s familial circumstances, but allows that he is a fallible human with epistemic limitations, while in another “skeptical” context, () might require that Joseph be in a superhuman epistemic position—for example, it might require that Joseph be able to offer non-question-begging evidence against the hypothesis that he is the victim of a large conspiracy concerning the Captain’s family, or even that he is a brain in a vat. If, as it has seemed plausible to many, it is impossible to attain such a standard, () will express an impossible proposition in such a skeptical context. A skeptic argues: • Joseph has no evidence against this possibility: Joseph is a brain in a vat, and there is no Captain or Captain’s daughter. • If there is a possibility w such that proposition p is false, and subject S has no evidence against w, then S does not know that p. Therefore, • Joseph does not know that the Captain has a daughter.15 The skeptic’s argument is deeply controversial; both of its premises are contested.16 I have little at present to add to the vast literature about whether the premises above can plausibly be rejected; I make only two observations. First, the skeptic’s premises 15 This is one of a variety of ways to spell out a skeptical paradox. Another would proceed more directly on the basis of an intuition to the effect that Joseph cannot know the skeptical scenario not to obtain and a kind of closure principle. See e.g. DeRose (). I don’t think these differences in formulation matter for my immediate purposes. 16 Mooreans like Sosa () and Williamson () deny the first premise; sensitivity theorists like Nozick () and Dretske () typically deny the second, thereby denying single-premise closure for knowledge. But I will argue in Chapter that sensitivity theorists need not deny closure or the second premise.
- 33. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” do enjoy a prima facie plausibility; epistemologists argue against them because they appear to lead to unacceptable conclusions, not because they are implausible on their faces. Second, the contextualist can accept both premises of this argument.17 Given these two facts, contextualism does enjoy at least some intuitive advantage over non- contextualist views. The contextualist is able to accept the skeptic’s claims, but to quarantine them—the argument stated above is sound, but its conclusion is given in context-sensitive language; its superficial appearance of overturning tenets of common sense (like the fact that Joseph knows that the Captain has a daughter) is illusory. This has been an important historical motivation for contextualism.18 For my own part, I find its treatment of skepticism compelling, and a good reason to accept contextualism. I shall spell out in more detail how I think skepticism and contextualism play out below. However contextualism per se is not committed to the truth of the skeptic’s claims in her skeptical context. It allows them, but does not require them. If, for instance, we are convinced, on grounds independent of avoiding skepticism, that the first premise in the skeptic’s argument is false—perhaps our best theories of perception and of evidence entail that perceptual content about the external world always counts as evidence—we could still accept contextualism. We would not, presumably, do so in order to avoid external-world skepticism, but this is not the only possible motivation for skepticism. DeRose () argues at length that contextualism is the best way to explain ordinary linguistic use of the term “knows”; the present work argues that contextualism forms an appealing unity with the knowledge first program. More moderate skeptical scenarios that allow direct knowledge of the external world via perception are also sufficient to generate similar puzzles. The conspiracy possibility mentioned above, and the lottery paradox, which I will discuss below, are examples. (I’ll discuss these issues, and whether we should think of moderate and radical skeptical scenarios as importantly different, in much more detail in Chapter .) For now, I will take contextualism’s treatment of skepticism as prima facie motiva- tion for contextualism, and develop my preferred version of the view. My approach is heavily informed by David Lewis’s; as we shall see, it is in many respects a development of his.19 In §., I’ll go on to lay out the Lewisian view to set the stage for my development of it. In §., I’ll contrast the approach with that of the most prolific contemporary defender of contextualism, Keith DeRose. In the final sections of the chapter, I’ll turn to a novel motivation for a contextualism of this kind, and consider some influential objections to the approach. This will set the framework for the 17 Note that so-called “subject-sensitive invariantists” cannot accept the truth of both premises without succumbing to contextualism; neither premise includes any stipulation about whether or not a subject attends to the skeptical possibility, or has any pressing practical interest in whether the Captain has a daughter. (We may suppose that we are running the argument in a way isolated from Joseph’s own attention.) 18 See e.g. DeRose (), Lewis (). 19 The view of the next several sections was first laid out in Ichikawa (a). The core of that paper remains my current view, although the presentation of this book reflects some small changes, some additional commitments, and a difference in emphasis.
- 34. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” synthesis of contextualism and a knowledge first approach that I will pursue over the rest of the book. . Elusive Knowledge According to David Lewis, knowledge ascriptions are best understood on the model of universal generalizations. Lewis gives us this gloss on “S knows that p”: S knows proposition P iff P holds in every possibility left uneliminated by S’s evidence. (Lewis , p. ) This is not a statement of anything in the neighborhood of contextualism; it is a claim about what it takes for a given subject to know a given proposition. The quoted sentence uses the English word “knows”, but it does not treat it as a subject matter. Consequently, it cannot be a statement of, or a commitment to, contextualism—a thesis about a certain English word, not about what it takes for a subject to know a proposition. It can be rather easy to fail to appreciate this point. (Indeed, Lewis himself ignores it throughout most of his paper; see DeRose (, pp. –) for this complaint against Lewis.) But, as emphasized in the Introduction, contextualism implies that general object-level claims about knowledge are susceptible to various interpretations; Lewis intends this one to carry a certain kind of metasemantic generality—but one might sign up to this statement without embracing contextual- ism. Lewis himself wishes to exploit the context-sensitivity of the “every possibility” quantifier in the right-hand side above. Which possibilities are relevant for the purpose of claims about “every” possibility depends on the context of the speaker, and so do which possibilities one must consider to evaluate whether S satisfies “knows p” in a context. Let’s understand Lewis as signing up to some kind of claim like this one: “S knows that p” is true in context C iff “p holds in every possibility left uneliminated by S’s evidence” is true in C.20 As mentioned above, it is largely uncontroversial that quantifiers like “every” are context-sensitive; Lewis attempts to characterize “knowledge” discourse in a paral- lel way. In my first paper on contextualism, I characterized Lewis’s strategy thus: “Lewis’s ‘knowledge’ inherits its context-sensitivity from the context-sensitive ‘every possibility’.”21 I now consider this choice of words regrettable. The word “knowledge” does not contain or otherwise involve the phrase “every possibility”; the context- sensitivity of the latter does not generate or explain context-sensitivity in the former. 20 Lewis (, p. ) remarks that the conscientious reader who wants to understand his suggestion in a more precise way can perform the required “semantic ascent” him or herself; this is the exercise I’ve engaged with in the previous paragraph. In some cases, however, this will be a more difficult task than Lewis seems to have anticipated. It is not at all clear, for instance, that there’s any way, once we are using the language carefully and precisely, to make sense of Lewis’s discussion of dogmatism on his pp. –. 21 Ichikawa (a, p. ).
- 35. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” I do not think—nor, so far as I can tell, would Lewis have thought—that as a psychological description of humans, we tend to evaluate “knowledge” statements by running through context-sensitive claims involving “every”. (Note also that there is no privileged direction in the metasemantic claim above; the “every possibility” claim needn’t be getting at any kind of essence of the “knows” claim.) Instead, I take Lewis’s suggestion to be that we attribute context-sensitivity in “knows” language along the model of context-sensitivity about quantifiers; the mere plausible truth of the meta- semantic claim shows us a way to understand how “knows” might behave in diverse conversational contexts. The familiar context-sensitivity of quantifiers provides us with a model for understanding the context-sensitivity of “knows”. Before pursuing the analogy further, let’s first consider in some small detail context- ualism about quantifiers. . Quantifiers As indicated in §., there are strong reasons to accept contextualism about quantifiers. Even ignoring the context-sensitivity of the gradable adjective “cheap”, what is said by an utterance of () depends on context: () All the plates are cheap. The Captain has been entertaining company in his cabin. In this context, an utterance of () makes a claim about the serving plates at his party. In fact, the insult is unfair; some of the Captain’s plates are cheap, but some are rather dear. As it happens, at the very same time, the boatswain is also hosting a party with some fellow sailors. The boatswain wants his friends to relax; in particular, he doesn’t want them to worry about whether they might break any of the serving dishes. To this aim, the boatswain utters the same sentence: “all the plates are cheap.” His use of () is true, even though the Captain’s rude guest’s was false. Naturally, we’re not too puzzled about () being true in one context and false in another, even though both utterances occur at just the same time, and just the same world; sentences like () are straightforwardly context-sensitive. What is said by an utterance of () depends on the context in which it is uttered. We can think of the character of the expression “all the plates” as a (non-constant) function from contexts of utterance to properties of plates.22 When the Captain’s guest utters (), the context 22 King (, p. ) suggests that such complex expressions do not have Kaplanian characters, but rather that their contents are determined by contents of their simpler constituent parts. (Moreover, according to Stanley and Szabó (), an expression like “all the plates” has among its contsituent parts a silent variable corresponding to a property.) Whether it is strictly correct to think of complex expressions as having substantive characters themselves doesn’t really matter for my purposes; we can still describe a kind of lightweight character in the sense that the stable linguistic meaning of the expression “all the plates” identifies a function from conversational contexts to properties of plates. We may agree with King and Stanley, if we like, that this “character” is explanatorily idle, and that the real semantic work is done by the substantive characters of the constituent parts.
- 36. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” supplies a property that picks out which plates are relevant—the property of being used at the Captain’s party, for instance. When the boatswain utters (), the context supplies a different property—the property of being used at the Boatswain’s party, perhaps. I say that the context interacts with the quantifier to produce a property. Why not say that the context determines a set of plates as the ones relevant for an utterance of “all the plates”? Here are two reasons. First, consider the Captain’s context. Suppose that there are four plates that are relevant, p–p. If the context provides a set of plates, then what is said by () in the Captain’s context is a proposition about p–p. Now consider a counterfactual version of the Captain’s context, in which a different four plates, say, p–p, are served at the party. Supposing that context fixes the set of relevant plates, in this counterfactual case, an utterance of () would have expressed a distinct proposition—one about p–p. So if the admiral utters (), what is said would have been different, had p–p been the plates present. But it is counterintuitive to claim that what is said depends on which plates happen to be around. Now this is admittedly less than a fully probative argument; there is nothing incoherent in the idea that what is said depends on one’s environment, even in ways that the speaker would not recognize. Indeed, it’s clear that some analogous cases hold; what is said by an utterance containing “he”, for example, depends on whether Tom or Dick is the person who happens to be present, whether or not the speaker is sensitive to the difference. Still, I think that there is a sense in which it sounds false to say that what is said (not merely whether it is true) in an utterance of () depends on facts like which plates the Captain has brought out; treating the context as picking out a property of plates, rather than a set of plates, allows us to respect that intuition. But even if you do not share my intuition here, there is also a stronger argument to the same conclusion. Consider the boatswain’s context, where () expresses a truth. What he said by uttering () would have been false if the Captain had come to the party and brought some expensive plates. It’s not only that an utterance of () would have been a saying of something false—it is that what was actually said would have been false. (Remember, as I’m using the terms, sentences are uttered; propositions are said.) The boatswain is saying a proposition that is true, but would have been false if there had been some expensive plates present. But then what is said must not have been a claim only about the plates that were actually used in the Boatswain’s party. Suppose those were p–p, all of which are cheap. If context fixes a set of plates, then an utterance of () in the Boatswain’s context picks out a proposition that is true just in case p–p are all cheap. But this proposition is true, not false, relative to an index in which the Captain crashes the boatswain’s party and brings expensive plate p.23 This question—whether we should think of the contextual domain restriction of quantifiers as providing sets or properties—has a correlate in the case of the Lewisian approach to knowledge ascription: does the context provide a set of possibilities, or something more like a property of possibilities? If what I have said in this section about 23 See Stanley and Szabó (, p. ) for a similar argument.
- 37. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” quantifiers is right, then the analogy suggests that it should be the latter. As I will lay out in §., there are sound epistemological reasons to prefer this version of the view as well—I will suggest that the context provides a parameter that is best modeled as a function from indexes to sets of possibilities. First, however, let’s return to the Lewisian approach to knowledge. . Lewis and Lewisian Contextualism As indicated in §., Lewis treats satisfaction of “knows p” in a context to be equivalent in truth-value to a quantified claim about possible worlds in that context: satisfying “knows p” in context C is a matter of eliminating “all” not-p possibilities, where C also contributes to which possibilities “count” as among “all” the possibilities. Lewis goes on to offer a series of rules that are intended to determine, of a given possibility, whether it is relevant for the purpose of a given knowledge ascription. I shall rehearse Lewis’s rules shortly; first, however, I should like to consider the role of these rules in Lewis’s broader framework. I argued in Ichikawa (a) that the attempt to articulate the rules that govern whether a given possibility is relevant is a supererogatory one, for the purpose of the general Lewisian contextualist approach. As mentioned above, contextualism about quantifiers is uncontroversial.24 Which plates count as among “all the plates” can vary according to the conversational context (as well as the index). We can support and clarify this claim by pointing to possible uses of “all the plates” language that seem to require, for their intuitive truth conditions, that different plates are relevant in different contexts. It is not a commitment of contextualism about quantifiers—a view which enjoys a widespread orthodoxy—to articulate a series of rules that would determine, of a given plate, whether it counts as relevant, given a particular context and index. Natural language is an unwieldy beast; to demand precise rules that explain and predict the domains of its context-sensitive quantifiers is to demand too much. No doubt it would be nice to have them, but the plausibility of contextualism about quantifiers does not depend upon having identified them. It may be that the best that we can reasonably hope to do is to gesture somewhat vaguely at good rules of thumb: plates that are explicitly attended to in a context are relevant; plates that are in relevant respects similar to those explicitly attended to are relevant, etc. In the same way, we should separate Lewis’s ambitious attempt to characterize rules that determine which possibilities are relevant, given a particular context and index, from the more general statement of Lewisian contextualism sketched above, according to which satisfaction of “knows p” requires ruling out those not-p possibilities that are relevant, given a particular context and index. The more general view can be 24 “Uncontroversial” is a context-sensitive gradable adjective; I am writing in a context lax enough for this sentence to be true, despite the existence of some controversy, i.e. the position of Cappelen and Lepore ().
- 38. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” motivated by uses of “knows p” that require, for their intuitive truth values, that different possibilities be relevant in different contexts; it is further clarified by gestures towards what kinds of considerations contribute toward a given possibility’s relevance or irrelevance. But the plausibility of Lewis’s broader framework does not rest upon having articulated a set of rules that make all the right predictions.25 I’ll use “Lewisian contextualism” as a name for this thesis: Lewisian contextualism. The proposition expressed by an utterance of the form “S knows that p” depends on the context in which it is uttered. In particular, the truth conditions require the elimination by S’s evidence of all of some set of alternatives to p; which alternatives are relevant depends in part on the conversational context. As I define it, Lewisian contextualism does not require the completion of the ambitious metasemantic project of articulating necessary and sufficient conditions for relevance. But Lewis himself does attempt this ambitious project, so it is worth saying something about his rules. He offers these four rules:26 • Actuality. The actual world is always relevant. Lewis clarifies that it is the subject’s actuality—i.e., the world the subject is in—rather than the attributor’s, that is at issue here. • Belief. Any possibility the subject believes to obtain, or ought so to believe, is relevant. • Resemblance. Any possibility that saliently resembles another relevant possibility (made relevant in some other way) is relevant.27 • Attention. Any possibility attended to in the attributor’s context is relevant. 25 Blome-Tillmann () rejects some of the details of Lewis’s rules, but retains Lewis’s ambitious project of setting out rules that predict which possibilities are relevant; Ichikawa () argues that Blome- Tillmann’s attempt does not succeed, and that our ambitions should be lowered. One might interpret Blome-Tillmann’s later emphasis on the one-direction “simple view” (e.g. in Blome-Tillmann () and Blome-Tillmann (a)), according to which “S satisfies ‘knows p’ in context C only if S’s evidence eliminates all the ∼p-worlds compatible with what is presupposed in C,” as a move in the direction I suggest. (Still, I don’t think it’s the best one; notice also that, contrary to his remarks in Blome-Tillmann (a, pp. –), the simple view does not itself ensure an “interesting and important philosophical claim about the role of pragmatic presuppositions in the semantics of ‘knowledge’-attributions” or provide an “explanation or resolution of skeptical puzzles” since, as a mere necessary condition for the satisfaction of “knows”, it allows that the role of presuppositions is idle. For example, the simple view is consistent with invariantist radical skepticism.) Even though they’re not counterexamples to the simple view, the counterexamples given to the stronger version of Blome-Tillmann’s view in Ichikawa () would, if successful, demonstrate particular respects in which the simple view fails to explain what needs explaining. 26 Lewis (, pp. –). In fact, Lewis offers seven rules: the four mentioned in the main text, and three permissive rules that offer a defeasible presupposition that certain possibilities are irrelevant (Lewis , pp. –). As far as I can see, the shape of Lewis’s view is the same whether or not the permissive rules are included—we can treat the restrictive rules as providing individually sufficient and jointly necessary conditions for relevance. So I will ignore the permissive rules. (To be frank, it’s never been clear to me why Lewis included them in his list in the first place.) 27 The “other way” proviso is to prevent the full recursive version of the view, which would threaten to extend to all possibilities via small steps of salient relevance.
- 39. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” The rules of actuality and belief explicitly depend on the subject, rather than the attributor. In the Kaplanian terms, they are a matter of index, rather than context. The rule of attention, by contrast, has everything to do with the attributor, rather than the subject, so it is a matter of context instead of index. The rule of resemblance has a more complex, hybrid structure; the language of “salience” suggests that which resemblance relation is at issue is a matter of context; but once that relation is fixed, which possibilities do resemble which others will be a matter of index.28 I will take care in the following section to separate the respects in which the context is relevant from those in which the index is relevant; there is a sense in which it is misleading to list all four of these rules together. Or at least, it obscures a crucial feature of Lewisian contextualism. Before moving on, I pause to note four more respects in which David Lewis’s commitments in “Elusive Knowledge” outstrip those of Lewisian contextualism, as I’ve characterized it. (Each is discussed in more depth in Ichikawa (a).) First, Lewis signs up to a particular conception of possibilities: a possibility, for Lewis, is a centered (metaphysically) possible world. Consequently on Lewis’s view there are no possibilities, and so no “relevant” possibilities, in which, e.g., Hesperus is not Phosphorus. So it is impossible, by Lewis’s own lights, to rule out the possibility that Hesperus is a star without ruling out the possibility that Phosphorus is a star. So anyone who knows the former must also know the latter; this is a counter- intuitive result. But it is open to the proponent of Lewisian contextualism to construe possibilities more liberally. If, for example, there is a coherent notion of “rational possibility” or “conceptual possibility” that outstrips metaphysical possibility, there is nothing preventing the Lewisian contextualist from using this more liberal notion in his account. I think these are important notions—this is one of the central ideas in Ichikawa and Jarvis ()—and I prefer to invoke them in my own implementation. So I say that one’s evidence might eliminate the possibility that Hesperus is a star without eliminating the possibility that Phosphorus is a star; in such cases, one might well know that Hesperus is not a star, but remain ignorant that Phosphorus is not a star. I do not plan essentially to be relying on such cases throughout the book; I merely record my own opinion on this controversial matter. If one prefers, with Lewis, to bite the bullet on cases like this, one need not quarrel with the project of this chapter. Second, Lewis commits to a particular understanding of what it is for evidence to rule out a possibility. Evidence, Lewis says, comprises sensory and introspective experience; for some evidence to rule out a possibility is for the obtaining of the evidence—i.e., the fact that the subject has such-and-such experience—to entail that the possibility in question is non-actual. As Lewis puts it: When perceptual experience E (or memory) eliminates a possibility W, that is not because the propositional content of the experience conflicts with W. (Not even if it is the narrow content.) 28 For more on the complexity of this feature of Lewis’s view, see Cohen () and Ichikawa (a, §).
- 40. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” The propositional content of our experience could, after all, be false. Rather, it is the existence of the experience that conflicts with W: W is a possibility in which the subject is not having experience E. Else we would need to tell some fishy story of how the experience has some sort of infallible, ineffable, purely phenomenal propositional content . . . Who needs that? Let E have propositional content P. Suppose even—something I take to be an open question—that E is, in some sense, fully characterised by P. Then I say that E eliminates W iff W is a possibility in which the subject’s experience or memory has content different from P. I do not say that E eliminates W iff W is a possibility in which P is false. (Lewis , p. ) It is open to a defender of Lewisian contextualism to follow Lewis here, or to invoke a different understanding of what it would take for the evidence to rule out a hypothesis. For example, it is open to a Lewisian contextualist to invoke a richer understanding of “evidence” than Lewis’s own internalist conception; I will have much more to say about this topic (including how the approach fits with a knowledge first theory of evidence) in Chapter ; for now, I’ll leave open questions about the nature and extent of evidence. Third, as several authors have emphasized, Lewis’s Rule of Attention is too weak— it’s not enough, for a possibility to be relevant, that it receive any attention whatsoever. One can hear a possibility mentioned without its becoming relevant. A more moderate version of Lewis’s view would have it that if attributors take a possibility seriously, then it is relevant.29 Fourth, the Lewisian contextualist might embrace Lewis’s general framework, according to which knowledge ascriptions require that alternatives among some contextually-influenced class of possibilities be eliminated by evidence, without fol- lowing Lewis in supposing that this evidential state is sufficient for the truth of a knowledge ascription. Lewis embraced the surprising implication of his view that satisfaction of “knows p” in a context does not require that a subject believe that p— “I even allow knowledge without belief, as in the case of the timid student who knows the answer but has no confidence that he has it right, and so does not believe what he knows” (Lewis , p. )—but this possibility is no commitment of the more general statement of Lewisian contextualism. In Ichikawa (a) I suggested adding a belief requirement and a basing requirement to the Lewisian framework. Here is my current proposal: S knows that p just in case, for some evidence E, (i) S believes that p, where that belief is properly based on evidence E, and (ii) all the E cases are p cases. 29 See Oakley () and Blome-Tillmann (, p. ). Note that taking possibility p seriously doesn’t require taking every possibility in which p seriously. Neta (, pp. –) seems to assume the contrary, arguing that if one is considering the possibility that one doesn’t have hands, one can’t “know” it not to obtain, since one can’t rule out the possibility that one is a brain in a vat. But in the Lewisian framework, one may take the possibility that one doesn’t have hands seriously by taking seriously the possibility that one was born without hands, or that one lost one’s hands earlier in life. These possibilities are plausibly ruled out by one’s perceptual experience; things would appear differently if they obtained.
- 41. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Strengthening the Lewisian condition for the truth of a knowledge ascription also avoids another problem for Lewis’s own view: the implausible implication that the possession of evidence E is luminous, in Timothy Williamson’s () sense.30 For suppose that some subject has evidence E. Then there are no possibilities (in the relevant set or otherwise) consistent with the subject’s having that evidence, in which the subject does not have E. So regardless of what possibilities are included as relevant, “knows that she has E” is satisfied. This is an implausible result; adding a belief and proper basing requirement avoids it.31 So this is the approach I will continue to defend. . Epistemic Standards It is common to characterize contextualism in terms of “standards” for knowledge or knowledge ascriptions. The notion is intuitive enough, but the extant literature about epistemic standards is unclear in crucial ways. Much of the confusion, I think, comes from the fact that the term is asked to cover two subtly, but importantly, different phenomena. As getting the correct notion of “standards” right will have significant implications throughout the book, I will spend §§.–. developing it. Readers interested in a quicker and broader treatment could skip ahead to §. now. To see why one might be tempted to explain contextualism about “knows” in terms of standards, consider the oft-cited analogy of contextualism about “knows” with contextualism about gradable adjectives like “flat” or “tall”.32 The conversational context, it is natural to think, determines just how tall one has to be to satisfy “tall” in that context. This threshold is very naturally thought of as a “standard” for tallness. Just as a university might set standards for applicants, admitting all and only students who score at or above the standard, so might a conversational context set standards for “tall”.33 And maybe the same goes for “knows”. 30 In fact, it has an even stronger implication than this already implausible one. If having evidence E is luminous, then any subject who has E is in a position to know that she has E. As I go on to explain in the main text, on Lewis’s view, any subject who has E isn’t merely in a position to know that she has E—she must actually know that she has E. 31 Ichikawa (a) gave the characterization above, minus the word “properly”—it required only of the relevant E that the subject base a belief that p on E. I am now convinced that this was a mistake. It’s not merely possible for one to fail to know something about one’s evidential state; the “Speckled Hen Problem” shows that there are cases in which one’s evidence doesn’t even put one in a position to know certain features about one’s evidence. (For the canonical presentation of the case and the related worry, see Chisholm () and Sosa ().) Assume for the purpose of argument Lewis’s phenomenological conception of evidence, and consider the experience one has upon looking at a hen with thirty-eight visible speckles; one’s evidence entails that one has the visual experience as of thirty-eight speckles, but given ordinary discriminatory abilities, one won’t know that one has the visual experience as of thirty-eight speckles (for all one knows, it’s the experience as of thirty-seven speckles). One won’t even know this if one somewhat bizarrely forms the rash belief that one has the experience as of thirty-eight speckles on the basis of one’s experience. This isn’t proper basing. Thanks to Baron Reed and Karen Lewis for bringing this issue into clearer focus for me. See Ichikawa and Jarvis (, pp. –) for my own approach to the problem of the Speckled Hen, which precisely emphasizes the distinction between sufficient evidence and (what I am here calling) proper basing. 32 See e.g. Cohen () and Halliday (). 33 However, as Kompa (, p. ) emphasizes, even in cases of gradable adjectives, the relationship between contexts and standards can be exceedingly complex.
- 42. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Keith DeRose often invokes the language of “standards”. Here, for example, is a passage in which he characterizes contextualism about “knows” as a theory according to which the truth-conditions of knowledge-ascribing and knowledge- denying sentences (sentences of the form “S knows that p” and “S does not know that p” and related variants of such sentences) vary in certain ways according to the context in which they are uttered. What so varies is the epistemic standard that S must meet (or, in the case of a denial of knowledge, fail to meet) in order for such a statement to be true. In some contexts, “S knows that p” requires for its truth that S have a true belief that p and also be in a very strong epistemic position with respect to p, while in other contexts, an assertion of the very same sentence may require for its truth, in addition to S’s having a true belief that p, only that S meet some lower epistemic standard. (DeRose , pp. –) (my emphasis) I do not deny that talk like this can be useful. In fact, given my own preferred use of the relevant terms, which I’ll outline below, a statement like this one from DeRose will come out as a true description of my preferred form of contextualism. But we should treat “standards” talk cautiously, for at least two kinds of reasons. First, as Stanley (, Chapter ) has emphasized, knowledge ascriptions seem to behave in ways very different from gradable adjectives. For example, we don’t seem to have any very natural way of talking about the relevant quantity that is suggested to come in degrees. “Strength of epistemic position”, unlike “height”, is a philosopher’s term of art. We also don’t have natural ways of expressing that one has more or less of the quantity in question, the way we so easily do with “taller” and “shorter”. Relatedly, it is not a commitment of contextualism that the so-called “epistemic standards” parameter be susceptible to a total ordering. The Lewisian view, for example, is most naturally developed in a way in which it can’t. Suppose that Buttercup sees the lights on in the Captain’s cabin; consider the sentence: () Buttercup knows that the Captain is awake. There are four contexts, c–c, according to Lewisian contextualism, for which each of these is true: c. () requires for its truth that Buttercup’s evidence eliminate the possibility that Buttercup herself is asleep, merely dreaming that the Captain is awake with the lights on. (It also requires that more moderate skeptical scenarios, like those mentioned below, be ruled out.) c. () permits the proper ignoring of the dreaming hypothesis, but it does require eliminating certain specific skeptical possibilities, such as the possibility that the Captain might be sleeping somewhere other than his cabin. c. () permits the proper ignoring of the possibility that the Captain is not in his cabin, but requires eliminating different specific skeptical possibilities, such as the possibility that the Captain has fallen asleep with the lights on. (It allows that the possibility that the Captain is sleeping elsewhere can be ignored.) c. () permits the proper ignoring of all possibilities in which the cabin light is on but the Captain is asleep. So “knowledge” requires only eliminating the possibility that the lights are off.
- 43. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Buttercup is dreaming C2 C4 C3 C1 Captain is sleeping elsewhere Captain is sleeping with the lights on Figure .. Four possible sets of relevant possibilities If it isn’t obvious that there are four contexts like this, stipulate that the light is a very good indicator of the Captain’s wakefulness—good enough to make the existence of c plausible. (This will be a rather nonskeptical context.) Then we can generate contexts with each of the other three features by adding salience of various kinds of error possibilities to the nonskeptical context. Fig. . illustrates the point at issue. The entire diagram represents the space of all possibilities; the circles labeled with contexts indicate which worlds are relevant for the purpose of knowledge ascription (). The dots correspond to particular skeptical scenarios—so the possibility that the Captain is sleeping elsewhere is among the relevant possibilities in c and c; its being outside the c and c represents that it is properly ignored in those contexts. There is a fine intuitive sense in which c involves a higher epistemic standard than the others—it is a skeptical context, where a subject must rule out the possibility that she is dreaming, in order for a knowledge ascription to be true of her. And c is very naturally understood as one that has a lower standard than the others. But these are all the comparisons that can straightforwardly be made. Notice that although c and c each require the consideration of certain slightly extraordinary possibilities, neither requires strictly more than the other. It does not seem right to say that either invokes a higher epistemic standard than the other: c demands more in some respects; c demands more in others. So if we want to talk about contexts supplying epistemic standards, we must be careful not to presuppose that these “standards” can always be lined up in order.34 We can describe a partial order well enough: standard s is stronger 34 DeRose (, p. ) countenances cases where “it’s unclear how the two epistemic positions we’re evaluating compare with one another,” but seems to think that the unclarity at issue is merely epistemic.
- 44. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” than standard s if and only if satisfying s entails satisfying s, but not vice versa. On the Lewisian framework, this will come approximately to the idea that the possibilities countenanced as relevant by s are a superset of those that are relevant according to s. But there will be some incomparability. This observation fits somewhat uncomfortably with Keith DeRose’s treatment of epistemic standards, which characterizes them in terms of reliability throughout modal space. In explaining his notion of a strength of epistemic position (to which epistemic standards are meant to correspond), DeRose writes: An important component of being in a strong epistemic position with respect to P is to have one’s belief as to whether P is true match the fact of the matter as to whether P is true, not only in the actual world, but also at the worlds sufficiently close to the actual world. That is, one’s belief should not only be true, but should be non-accidentally true, where this requires one’s belief as to whether P is true to match the fact of the matter at nearby worlds. The further away one can get from the actual world, while still having it be the case that one’s belief matches the fact at worlds that far away and closer, the stronger a position one is in with respect to P. (DeRose , p. ) According to DeRose, what it is to have a stronger epistemic position is to rule out the alternatives in broader sets of possible worlds. So what it is for knowledge ascriptions to require stronger epistemic positions is for them to require the ruling out of alternatives in broader sets of possible worlds.35 An epistemic standard, then, could be thought of as a kind of measurement of modal distance—how far into modal space must one avoid error? If one draws a diagram of modal space, the radius of a circle centered on actuality could correspond to an epistemic standard. See Fig. .. But if (as seems to follow from Lewisian contextualism) there can be contexts like c and c above, one cannot always represent the space of relevant possibilities as spheres; the possibilities corresponding to c and c have the same area and are both centered on actuality, but they are not congruent.36 On DeRose’s view, situations like the one described in Fig. ., which involve overlapping sets of relevant alternatives, can never arise. One must ask which skeptical possibility is nearer in modal space: a case where On the view I’m articulating, there are genuinely incommensurate strengths of epistemic standards; it is not merely a matter of our failure to recognize which is stronger. 35 See Blome-Tillmann () for an argument against this way of explaining standards. I find his argument somewhat compelling, although it probably turns on an open question about just how the modal similarity metric in question is supposed to be established. 36 Schaffer () also criticizes DeRose’s approach along these lines. Note however that he is using “standards” in a stronger sense than I am—the idea that epistemic standards are what shifts is by definition, in Schaffer’s framework, a matter of modal distance. Compare also the discussion of “epistemic standards” in Neta (), who argues that it is that which counts as evidence, rather than the epistemic standards, that shifts according to context—Neta’s assumption seems to be that the “standard” is or supervenes on the strength of evidence that is required for knowledge. On the more general notion of “standards” I am using, an approach like Neta’s according to which contextual variation makes for a difference in what is “known” because it makes for a difference in what is “evidence” will count as one in which the standards change.
- 45. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” radical skeptical scenario moderate skeptical scenario moderate standard actuality nonskeptical standard skeptical standard Figure .. DeRose’s model of epistemic standards as modal distances the Captain is sleeping somewhere else, or one where he is sleeping in his cabin with the lights on? If one is relevant, then anything else closer is relevant too.37 I’ll remain officially neutral in this book as to whether there can be a strict ordering on contexts of this kind, or whether the conversational pragmatics governing the set of relevant alternatives can take these kinds of less symmetrical shapes. Although I do think it would be a mistake to assume a strict ordering, I’ll return to this question in Chapter , where we will see that there may be certain advantages to enforcing a more symmetrical approach. The second kind of reason we must be careful about “standards” language is that conversational context is not the only thing that can influence which alternatives are relevant. Invariantists sometimes talk about varying “standards” too. Getting clear on what this suggestion amounts to is crucial for distinguishing contextualism with rival views like “interest-relative invariantism”. I turn to these issues now. 37 Perhaps one might hold on this view that we don’t typically know which is a nearer skeptical possibility. Then if one is relevant, we may not know whether the other is relevant. In such cases, if Buttercup’s evidence eliminates one skeptical possibility but not the other, we wouldn’t be able to know whether, in certain contexts, the knowledge ascription could be true.
- 46. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” . Invariantism with Shifting Standards Consider, for example, “interest-relative invariantism” (IRI), the family of views that has recently been defended by Stanley (), Fantl and McGrath (), and Weatherson ().38 According to these views, some practical considerations, such as how important a question is to a subject, play roles in determining what a subject must be able to rule out in order to have knowledge.39 IRI is often described as similar to contextualism in that differing “standards” govern the truth of a knowledge ascription, with the difference being that contextualists think thespeaker’s interests are relevant for determining the standards, while IRI has it that the subject’s interests are what matter. Features other than practical interests are also sometimes thought, by other kinds of “shifty invariantists”, to play such roles. For example Hawthorne () suggests that which possibilities are salient to a subject make a difference for what one has to do in order to know; Wright () says that features of the subject’s social roles make a difference.40 So if the epistemic standard is characterized in terms of which possibilities must be ruled out, then shifty invariantism, like contextualism, will posit shifts in epistemic standards. DeRose identifies and accepts this consequence of his view, writing that shifty invariantism “agrees with contextualism that varying epistemic standards govern whether a speaker can truthfully claim ‘I know that p’. ”41 It is worth appreciating, however, that if epistemic standards are characterized in this way, then it is not only contextualists and shifty invariantists who are committed to varying epistemic standards for knowledge or knowledge ascriptions. Probably every nonskeptic is committed to this kind of “shifting standards”. Certainly, anyone who embraces orthodoxy about Gettier cases must also think that which possibilities a subject must rule out in order to satisfy “knows” can vary according to the subject’s situation.42 For example, suppose that Ralph believes on excellent grounds that 38 This family of views is sometimes labeled “subject-sensitive invariantism” or “SSI”. This is not a particularly apt name, as it is trivial that whether a subject has knowledge depends in part on features of the subject. Cf. Stanley (, p. ). My preferred “interest-relative invariantism” makes explicit that knowledge depends on facts about the subject’s interests, not just about the subject generally. It should be noted, however, that this name has its shortcomings too; the view of Hawthorne (), for instance, is typically categorized alongside the others mentioned in the main text, even though Hawthorne’s suggestion is that knowledge depends upon what is salient to the subject; it is something of a stretch to characterize this as a fact about the subject’s practical interests. I use “shifty invariantism” as a more general term. But in light of what I will go on to say about epistemic standards, interest-relative invariantism, and Gettier cases, articulating what it is that characterizes shifty invariantism in general is a surprisingly difficult question. 39 Because these views are invariantist, it is more natural to speak at the object level about what is required for knowledge, rather than for the truth of a knowledge ascription. Strictly speaking, however, one could put things into the metalinguistic terms. According to IRI, whether a knowledge ascription is true of a subject depends in part on the subject’s practical interests. The IRI theorist simply disquotes: in general, if invariantism of any kind (including IRI) is correct, a knowledge ascription is true of a subject iff the subject has knowledge. 40 The “methodological contextualism” of Wright (), which she derives from Williams () and Williams (), is also a shifty invariantist view of this kind. 41 DeRose (, p. ). 42 See Brueckner () on this point.
- 47. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Josephine’s social class is above his own. Suppose also that Ralph loves Josephine, and knows that he does. Naturally, Ralph believes that he loves someone of a different class. If all were as it seems, then this belief would amount to knowledge; but in fact, Ralph has been misinformed about each of their social positions—he himself comes from a noble family, while Josephine’s roots are working-class. When he believes that he loves someone of a different class, his belief is true, but it is not knowledge; Ralph is in a Gettier case. According to invariantist orthodoxy, Ralph’s epistemic position is sufficient for knowledge in the case where everyone’s social status is as it appears, but not in the case where it turns out that the evidence about social status was misleading. So when the environment is less cooperative, it appears that the “epistemic standards” (in DeRose’s sense) necessary for knowledge are higher. This is a conclusion that I doubt DeRose would want to accept, as he characterizes contextualism and shifty invariantism as distinctive in allowing for differing standards governing knowledge ascriptions.43 But it is not clear how to articulate the notion of an epistemic standard in a way that makes the mechanisms of IRI views amount to changes in standard, without categorizing Gettier cases in the same way. Indeed, one needn’t appeal to Gettier cases to make this point—once the thought is articulated clearly and laid out for examination, it’s just obvious that different possibilities can be relevant, depending on the subject’s situation. All theorists will accept, for instance, that whether one needs to rule out the hypothesis that the animal before one is a cleverly disguised mule depends at least in part on whether there’s any reason to think it might be. Epistemic standards are sometimes glossed as matters of what purely epistemic factors are required for knowledge or knowledge ascriptions;44 they are also some- times described as the truth-relevant factors that are required.45 But the notions of “purely epistemic” factors and “truth-relevant” factors are inevitably left unexplicated; I myself have never been confident that I have a grip on what they amount to.46 There is perhaps a response-dependent property in the neighbourhood: one might identify “epistemic standards” as governing only those features that obviously or unsurprisingly 43 I don’t have the space to make the case here, but my view is that the arguments from e.g. DeRose (, p. ) turn on this commitment. 44 Fantl and McGrath (, p. ) describe “impurism” as the thesis that “[h]ow strong your epistemic position must be—which purely epistemic standards you must meet—in order for a knowledge-attributing sentence, with a fixed content in a fixed context of use, to be true of you varies with your circumstances.” DeRose (, p. ) likewise characterizes standards as a matter of “how good an epistemic position” one needs to know. 45 E.g. Stanley (, p. ), Fantl and McGrath (, p. –), Grimm (), DeRose (, pp. & ), Brown (b, p. ). 46 If, as one might’ve found pre-theoretically plausible (obvious?), knowledge itself is both “purely epistemic” and “truth-relevant”, then it is trivial that a fixed epistemic standard—viz., knowledge—is necessary and sufficient for knowledge, in all cases. So whether or not one adopts shifty invariantism, one is committed to the idea that there is a purely epistemic, truth-relevant property that is necessary and sufficient for knowledge. If knowledge isn’t “purely epistemic”, or if it isn’t “truth-relevant”, then I don’t know how to determine when those terms apply.
- 48. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” influence knowledge or knowledge ascriptions. But there is no clear role for such response-dependent properties in serious theorizing about knowledge. So I don’t find it useful to categorize contextualism and shifty invariantism alike as involving the use of different “epistemic standards”. Notice also that it should be possible to endorse both contextualism and e.g. the kind of interest-sensitivity characteristic of IRI. The “invariantism” in “interest- relative invariantism” is the negation of contextualism, so one can’t coherently be a contextualist who endorses IRI, but the semantic thesis of invariantism about “knows” is perfectly consistent with the metaphysical thesis of “pragmatic encroachment” posited by IRI47—namely, that knowledge depends on the practical situation of the subject.48 But if such pragmatic encroachment is characterized as the view that the epistemic standards are fixed by the practical situation of the subject, then one cannot consistently hold contextualism, if it is characterized as the view that the speaker’s context fixes the standards. The same goes, mutatis mutandis, for other kinds of shifty invariantism, such as the social-sensitivity view of Wright (). One could describe the “interest-relative contextualist” view I am mentioning as a view according to which the speaker’s context and the subject’s practical situation both play roles in fixing the epistemic standards necessary for the truth of a knowledge ascription, but this obscures the fundamental difference between contextualism— a semantic thesis about the verb “knows”—and pragmatic encroachment—a meta- physical thesis about knowledge. It better gets at what is distinctive of these views, I suggest, to describe contextualism as the view that different standards govern knowledge ascriptions in different contexts, while pragmatic encroachment holds that practical considerations about a subject are relevant to determining whether that subject meets a given standard. Consequently, I do not draw the straightforward connection between epistemic standards and contrast classes. According to orthodoxy, whether there are fake barns around influences which worlds are relevant; according to pragmatic encroachment theorists, the practical importance of a question to the subject also influences which worlds are relevant. I do not describe either kind of subject-sensitivity as a shift in the standards that must be met for knowledge. I recognize this as a terminological departure from DeRose, whose seminal role in setting out this terrain admittedly entitles him to a degree of stipulative authority over the vocabulary in the area. Nevertheless, I consider this departure sufficiently theoretically motivated to justify my slightly heterodox use of terminology here. My view is that epistemic standards are simply functions from indexes to contrast classes.49 One can recover the idea of a stronger or weaker standard thus: standard A is 47 The term “pragmatic encroachment” is due to Jonathan Kvanvig in a “Certain Doubts” blog post in June . 48 Cf. McGrath () and Fantl and Mcgrath () who also point out this consistency. 49 DeRose also thinks of standards as such functions—see DeRose (, pp. –). But as explained in the main text, his view—at least the version in DeRose ()—has it that the functions have rather different characters to the ones I have in mind.
- 49. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” stronger than standard B (and B is weaker than A) iff, for every index i, A(i) is a proper superset of B(i). This will have the implication that the satisfaction of standard A will entail the satisfaction of standard B. Since I agree with DeRose that contextualism is well-characterized as the thesis that knowledge ascriptions require the satisfaction of different standards in different contexts, I also agree with DeRose that context- ualism allows for different alternatives to be relevant in different contexts. In fact, my approach is easily read as a generalization on DeRose’s remarks here. He thinks that standards interact with indexes to generate sets of relevant worlds by drawing a sphere with a particular radius around them in modal space; I think standards interact with indexes to generate sets of relevant worlds in a more complicated way. But unlike DeRose, I do not consider shifty invariantism to be like contextualism in allowing different standards to govern different knowledge ascriptions about different situations. If invariantism is true, there is only one epistemic standard; so for any given index, there is only one set of relevant alternatives. But the index encodes information about the practical situation of the subject, which may be relevant for the question whether a given standard is met. . Factivity Something DeRose and I do agree about, however, is that it won’t do to identify standards with sets of relevant possibilities.50 One problem with this idea is that, if standards are supposed to be set by contexts, it asks contexts to do work that indexes should be doing. The easiest way to see this, though not the only way, is by considering factivity. Whether a knowledge ascription is true depends in part on whether the content of that ascription is true at the index at which the knowledge ascription is evaluated. That is to say, in the Lewisian framework, the world of the index is always relevant, for any knowledge ascription.51 But it is not very plausible that in general, the conversational context will supply the needed facts about the subject’s situation. To see this, suppose that Dick has heard Ralph discussing a plan to elope, and that Buttercup knows that Dick has done so. On this basis, it is plausible to suppose that there is a context in which Buttercup could truly say, “Dick knows that Ralph is planning to elope.” But consider the same sentence, in the same context, in a world where things seem the same to both Dick and Buttercup, but where Ralph is not planning to elope. (Perhaps his discussion of such putative plans was intended to deceive would-be meddlers like Dick.) In this world Ralph is not planning to elope; so factivity demands the relevance, for Buttercup’s knowledge ascription, of the possibility that Ralph will not elope. Since this possibility is a case where Ralph has no plans to elope, and is uneliminated by Dick’s evidence, Buttercup’s knowledge ascription in this context must be false. Schematically, where C is the nonskeptical 50 See DeRose (, pp. –). 51 Cf. Lewis (, pp. –); see also my §..
- 50. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” context in question, S is the sentence, “Dick knows that Ralph is planning to elope”, w is the world where things are as they seem, w is the world where Ralph’s conversation was misleading, and X is the skeptical possibility that Ralph will not elope even though the evidence is as it is in w: . X is a possibility in which Ralph is not planning to elope, and it is uneliminated by Dick’s evidence. . In w, Buttercup’s utterance of S in C is true. . So in w, X is an irrelevant possibility. . In w, X is the case. This explains why in w, Buttercup’s utterance of S in C is false; so X is relevant. . So C does not fix whether X is relevant. If this argument is right, then it would be a mistake to hold a version of Lewisian contextualism according to which the context fixes the relevant alternatives; if context is supposed to fix epistemic standards, then epistemic standards cannot be identified with—or even determine—the sets of relevant alternatives. Insofar as one wishes to characterize a plausible version of Lewisian contextualism, then, one should not interpret the view as Jonathan Schaffer does, according to which “[a] sentence of the form “s knows that p” is true in context c iff s’s evidence eliminates every not-p possibility relevant in c” (Schaffer , p. ).52 There is, however, a substantive assumption in the argument given above. As discussion of this point is somewhat technical and might be considered peripheral to readers not invested in the debate, I’ll discuss it in its own section; some readers may prefer to skip ahead to §.. . Modality and Knowledge Ascriptions The argument of the previous section assumed that the very same context, C, might occur in the relevantly distinct possible worlds w and w. The intuitive thought underwriting this assumption was that the difference between w and w was in some sense wholly external to the experience of Buttercup, the speaker; whether 52 Another reason it’s important to be precise about how we think about standards, and just what is fixed by context, is that failure to do so can make contextualism appear susceptible to objections it needn’t be vulnerable to. For example, Jason Stanley argues that contextualists can’t comfortably allow that subjects’ practical situations influence how easy it is to satisfy “know”, in cases where the speakers are unaware of the practical importance to the subject (“Ignorant High Stakes” cases). Stanley writes that for contextualism to give the right result, “it must turn out that the fact that there is a greater cost to Hannah’s being wrong affects the semantic content of some of her statements, even when neither she, nor any other conversational participant, is aware of it. So the contextualist could accommodate Ignorant High Stakes, but only at the cost of advancing a rather dramatic claim about the potential semantic effects of non-psychological facts about extralinguistic content.” (Stanley , p. ) Even granting for the purpose of argument that such an implication is a costly one, the contextualist can avoid it by clarifying that the “standard” provided by features of the context of which the speaker is cognizant itself encodes a sensitivity to the subject’s practical situation.
- 51. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Ralph has been secretly deceptive or not isn’t relevant to what kind of context she is speaking in. (Ralph is not a participant in the conversation.) But in fact, this was a stronger assumption than I needed. The key assumption is that Buttercup expresses the same proposition in w as she does in w. (This is weaker than the assumption that the contexts are identical, because some differences in conversational context can be idle for the purpose of a given knowledge ascription. So what is really at issue here is whether the two contexts are the same in all relevant respects—i.e., whether the knowledge ascription expresses the same proposition in each of them.) The argument assumes that what Buttercup says does not depend on whether Ralph’s conversation was sincere. Whether what she says is true depends on Ralph, but not what she says itself. It should be admitted, however, that these kinds of higher-order metasemantic intuitions are rather theoretical; someone might deny the intuitions cited above about whether Buttercup expresses the same proposition in distinct possible worlds which she cannot distinguish. I think this would be a counterintuitive move, but not a disastrous one. Jonathan Schaffer has told me in conversation that this is his preferred way to think of things. So, adapting his version of the view to my example, Schaffer would insist that the contexts mentioned in () and () above are in important ways distinct: the context mentioned in () is one in which Buttercup is discussing Dick’s position vis-à-vis Ralph in w; the context mentioned in () is one in which Buttercup is discussing Dick’s position in w. This difference in context does make for a change in the relevant alternatives. So what is said is different between w and w. But there is another reason not to go this way, in addition to its counterintuitive consequences about what is said. Taking the Schaffer route commits one to choosing between implausible verdicts about modal profiles of propositions expressed by knowledge ascriptions. Consider again Buttercup’s knowledge ascription in distinct worlds w and w. In w it is true, and in w it is false; Schaffer thinks that this is because she expresses different propositions—the sentence expresses p in w, and p in w, where p is stronger than p in that it entails elimination of the possibility that Ralph was lying. Consider p, the true proposition Buttercup expressed with “Dick knows that Ralph is planning to elope” in w. Its truth requires Dick to rule out cases where Ralph is (a) not planning to elope, and (b) straightforward and honest in his reports about his intentions. Dick can do this in w, since he heard Ralph say that he was planning to elope. That’s why p is true in w. But notice that p also looks to be true in w, where Ralph was lying about his plans to elope. For p does not require Dick to rule out the possibility that Ralph was lying. On the Schaffer model, where epistemic standards are fixed by contexts, and identify sets of relevant alternatives, we must say that the proposition expressed in w is one that doesn’t imply elimination of that skeptical possibility. But this means that, relative to a world where that possibility obtains—in this case, w—that proposition may still be true, even though that said to be “known” is false. That is to say, what Buttercup says with her utterance of “Dick knows that Ralph is planning to elope”
- 52. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” in w—p—is true in some worlds where Ralph isn’t planning to elope, viz., w. This is nothing less than a violation of factivity in the straightforward sense: it is false, on the view I am here sketching, that the proposition expressed by “S knows that q” entails that q. There are possible worlds—w in my example—where the proposition is true, but q is false. This by itself is a serious theoretical cost; it also has particular counterintuitive implications. For example, when Buttercup expresses p in w, it is extremely natural to suppose that if Ralph had been lying, what Buttercup said would have been false. But what Buttercup said, on the view we’re considering, was p; and p is true at worlds like w where Ralph was lying. The considerations just raised should put pressure on most contextualists away from thinking of contexts as establishing sets of relevant alternatives, the way that Schaffer does. However, I do not expect these considerations to have much sway against Schaffer himself. For Schaffer is independently committed to necessitarianism about propositions—the view that all propositions are ultimately propositions about particular possible worlds, and so have their truth values necessarily (Schaffer ). So Schaffer is already committed to the view that Buttercup’s sentence in w, “Dick knows that Ralph is planning to elope”, expresses a proposition that is true in w where Ralph is not planning to elope. On Schaffer’s view, the proposition in question is something like the proposition that in w, Dick knows that Ralph is planning to elope, rather than sitting around and not doing or saying anything about eloping; and this proposition is necessarily true. Even relative to worlds in which Dick has never heard of Ralph, it’s still true that in w, Dick knows that Ralph is planning to elope. Whether this amounts to a violation of factivity turns on sensitive questions about what exactly factivity requires—it is a view according to which the proposition expressed by “S knows that p” does not entail that p, in the usual sense that there are possible worlds where the former proposition in question is true but the latter is false. But if it is a failure of factivity, it is one that the necessitarianist is already committed to. Insofar as one wishes not to commit the Lewisian contextualist to necessitarianism about propositions, one has good reason not to think that context fixes the relevant alternatives. . Differences between Knowledge and Quantifiers Jason Stanley has argued that knowledge ascriptions interact with contexts in a way quite different from the way standard quantifiers like “every” and “some” do. Accord- ing to Stanley, this undermines the Lewisian strategy of understanding knowledge as context-sensitive along the model of quantifiers.53 Stanley defends the following claim about context-sensitive discourse in general: 53 In fact, Stanley’s argument is intended to generalize to all forms of contextualism, but it is the Lewisian strand that interests me here.
- 53. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Since semantic context-sensitivity is traceable to an individual element, multiple occurrences of that element in a discourse should be able to take on differing values. In the case of an utterance such as “This is larger than this”, where two different objects are pointed to by the person uttering the sentence, this feature is obviously confirmed. But it is present in a broader range of constructions. (Stanley , p. ) Call the ability of context-sensitive terms to take on different values rather freely flexibility. Stanley suggests that all context-sensitive discourse is flexible,54 and that if “knows” is likewise flexible, then contextualism about “knows” does not enjoy the advantages claimed for it. For example, Stanley writes, If different occurrences of instances of “knows that p” can be associated with different epistemic standards within a discourse, some of the paradigm sentences the infelicity of which supposedly motivates [contextualists’] accounts over rival accounts turn out to be felicitous and potentially true by contextualist lights. For example, if we have similar behavior to many other context- sensitive expressions, one would expect the following to be felicitious: () Bill knows that he has hands, but Bill does not know that he is not a bodiless brain in a vat. (Stanley , p. , my numbering) There are two challenges for the contextualist here. First, a major historical motiva- tion for contextualism has been its supposed ability to resolve skeptical puzzles without admitting the truth of sentences like Stanley’s (); if Stanley is right that contextualism predicts () to have easy true readings, a motivation for contextualism is undermined. Second, more straightforwardly, () is intuitively repugnant; any view that implies it should be rejected.55 In support of his contention that Lewisian contextualism predicts sentences like () to be unproblematic, Stanley cites sentences involving explicit quantifiers whose domains shift mid-sentence. For example, Stanley points out that it’s not difficult to find a reading of () according to which the two instances of “every” take different domains: () Every sailor waved to every sailor. For example, if two ships have recently crossed paths, it might say that every sailor on one ship waved to every sailor on the other. More directly to the point, Stanley also offers this dialogue: A. Every van Gogh painting is in the Dutch National Museum. B. That’s a change; when I visited last year, I saw every van Gogh painting, and some of them were definitely missing. (Stanley , p. ) 54 One might perceive an obvious potential objection to this claim in, to use Kaplan’s term, “pure indexicals”—indexicals like “I” that take their semantic value automatically, without the aid of any kind of demonstrations. See Stanley’s pp. – for a rejoinder. 55 On both of these points, see especially DeRose ().
- 54. i i i i i i i i OUP CORRECTED PROOF – FINAL, //, SPi “knowledge” Stanley suggests that in the natural reading of this dialogue, the “every” in (A) and the “some” in (B) each range over all van Gogh paintings in existence, but that the “every” in (B) is restricted to those paintings that were in the museum at the time of the visit. I agree. So Stanley is right that the domain of “every” can become larger or smaller over a relatively short period within a conversation. But this does not imply, as Stanley suggests it does, that Lewisian contextualism predicts abominable conjunctions to be acceptable. It does not even imply that there is any significant disanalogy between knowledge ascriptions and quantifiers. Stanley has shown that “every” is at least somewhat flexible, but he has not shown that it is as flexible as the disanalogy would demand.56 Consider this sentence: () Bill’s evidence eliminates all possibilities in which he lacks hands, but Bill’s evidence does not eliminate all possibilities in which he is a handless brain in a vat. This sentence is closely connected to the abominable conjunction () mentioned above, given Lewisian contextualism. But it is not at all easy to find a true reading of (). In fact, () feels like a contradiction, much as the simpler () does: () Every sailor is on deck, but some sailors are below deck. But () and () are not obviously coherent; one cannot easily find true readings of them. This is a point of analogy, not of disanalogy, between quantifiers and “knows”; for () works just the same way.57 So it does not seem to me that contextualists should accept that there is enough flexibility in “knows” to predict the felicity of abominable conjunctions. This is not to insist, however, that “knows” is totally inflexible, either—and the same goes for the case of quantifiers. For example, if one adds supporting context to help make clearer the apparent shift in epistemic standards, it might be possible to get something like the mid-sentence shifts Stanley describes. I think () is much preferable to (), for instance: () Since he certainly wouldn’t fail to notice an amputation, Bill knows he has hands, but since he can’t prove it from first principles, Bill doesn’t know he’s not a disembodied brain in a vat. Or, sticking to more mundane skeptical scenarios: 56 I first gave this argument in defense of Lewis against Stanley’s attack in Ichikawa (a, pp. –). For a similar response, see Blome-Tillmann (, pp. –). 57 Gauker () argues that Stanley’s own semantics for quantifiers (Stanley and Szabó ) implies that there are true readings of () and (), and uses this fact to motivate his own alternative. He also points out (p. ) that “knows” patterns with quantifiers in this way.
- 55. Exploring the Variety of Random Documents with Different Content
- 56. a cheery laugh, taps Tottie's cheek with his forefinger, and cries, in a tone of satisfaction, 'Now, I've got it!' (Tottie wishes she had.) 'Now, I've got it,' cries the old man again; 'all complete.' Tottie shifts restlessly in her high chair. 'And Tottie shall see me make it,' says Ben, with beaming face, rubbing his hands, and shifting the fruit and the spice about much the same as if they form pieces of a puzzle, and he has found the key to it. 'Especially,' adds Ben, 'as Tottie will sit still, and won't touch.' 'No, I never!' exclaims Tottie. This is Tottie's oath, which she is much given to swearing when her honour is called into question. Tottie's 'No, I never!' is in her estimation worth a volume of affidavits, but it is much to be feared that her sense of moral obligation is not of a high order. 'And as Tottie's a good little girl----' 'Tottie's a dood little girl!' There is no expression of doubt in the nods of the head with which Tottie strengthens this declaration. 'And'll sit still, she shall see me make it.' The good old fellow laughs. He does not seem to realise how difficult is the task he has set Tottie. To sit still, with these treasures in view! Here an agonising incident occurs. A small piece of candied sugar has become detached from one of the halves of lemon-peel, and Ben Sparrow, with an air of abstraction, picks it up, and puts it--
- 57. in his own mouth! Tottie watches him as he moves it about with his tongue, and her own waters as the sweet dissolves in her imagination. She knows the process as well as Ben, and appreciates it more, and she sighs when the candy is finally disposed of. 'You see, Tottie,' says Ben, taking her into his confidence, 'business is very slack, and Christmas is coming, Tottie.' Tottie gives a nod of acquiescence.
- 58. 'So I think to myself--another nod from Tottie; she also is thinking to herself--'if I can put some thing in the window that'll make the people look at the figs----' Here Tottie introduces an artful piece of diplomacy. 'Tottie can spell fig,' she says, and proceeds to do it smilingly--'F-I-G, fig.' But Ben, intent upon his scheme, does not see the point of Tottie's interruption, and proceeds: '--Something that'll make 'em look at the figs, and the currants, and the raisins--something new and spicy'--(Ben laughs at this joke, and repeats it)--'something new and spicy, perhaps it'll wake 'em up, and bring 'em in here instead of going to another shop. For they want waking up, Tottie, they want waking up badly.' Solemn nods from Tottie proclaim the serious consideration she has given to the general sleepiness and indifference of Ben Sparrow's customers. Ben Sparrow picks up a fat currant and contemplates it with as much interest as a geologist would contemplate a new fossil. Tottie's eyes follow his movements; she sits like Patience on a monument, and another sigh escapes her as Ben Sparrow (again abstractedly) puts the currant in his mouth, and swallows it. Draw a veil mercifully over Tottie's feelings. 'It was in the middle of the night,' says Ben Sparrow with all the impressiveness demanded by the historical fact, 'that I first thought of making ME, and putting ME in the window to attract custom. I was a good deal puzzled about my legs, and my stomach got into my head, and I couldn't get it out; but little by little all my limbs and every other part of me came to me until the idea was complete. And now we'll try it--now we'll set to work and make a MAN! And if you're a good girl, and'll sit still, you shall see ME made.'
- 59. Tottie's experience in literature is very limited--extending no farther, indeed, than b-a-t bat, c-a-t cat, r-a-t rat, d-i-g dig, f-i-g fig, p-i-g pig--and she knows nothing of the terrible story of Frankenstein; therefore, she is not at all frightened at the idea of seeing a man made, nor has she any fear that it will turn out to be a monster. On the contrary, if Ben Sparrow's thoughts would only take a benevolent turn in the shape of a fig for Tottie, or a few plums for Tottie, or some candied sugar for Tottie, she would be prepared to enjoy the feat which Ben is about to perform as much as if it were the best bit of fun in the world. 'Now, then,' commences Ben, with a whimsical glance at Tottie, who smiles back at him like a true diplomatist, 'I don't know what part is generally made first, but perhaps it'll be as well to commence with the stomach. Here it is--here's my stomach.' He takes one of the halves of the candied lemon-peel, and places it before him, round side up. 'There's a little too much sugar in me,' he says, with a more whimsical glance than the first; 'it'll make me rather too heavy, I'm afraid. And besides, Tottie, it ain't true to nature. My inside ain't got such a coating as this.' He breaks a piece of candied sugar from the inside of his stomach, looks at Tottie, notices her wistful eyes, and gives it to her. She eats it eagerly, and so quickly as to cause amazement to Ben Sparrow, who says, 'You shouldn't eat so fast, Tottie. Good little girls don't eat so fast as that.' Tottie, with feminine duplicity, accepts this warning in an inverted sense, and cries, with her mouth full of sugar, 'Tottie's a dood little girl!' as if indorsing a statement made by her grandfather. But Tottie's thoughts are not upon the good little girl; at
- 60. the present moment she resembles a savage. She has tasted blood, and thirsts for more. 'It's a fatter stomach than mine,' proceeds Ben, laying his hand upon his stomach of flesh, the stomach he came into the world with; 'it's rounder and plumper, and would fit the Lord Mayor or an alderman, but it'll do, I daresay. Now for my neck.' He picks up the thickest piece of cinnamon, and measures it with his eye, breaking the stick in two. 'I mustn't make my neck too long- -nor too short--and I take the thickest piece, Tottie, because it's got to support my head. Like this.' He makes a hole in the end of the lemon-peel, and sticks the cinnamon in firmly. 'Now to stick my head on, Tottie.' He selects the largest of the fat figs, and attaches it to his neck. 'What's the next thing? My eyes, to be sure. Currants.' Remarkably like eyes do they look when they are inserted in the face of the fat fig. Then he takes a clove for his nose, and, making a thin slit in the fig for his mouth, inserts an appropriate morsel of mace. All this being successfully accomplished, he holds himself up (as far as he goes) for his own and Tottie's inspection and approval. Tottie claps her hands, and laughs, but subsides into a quieter humour at a guilty thought that steals into her mind. She thinks what a delightful thing it would be to take her grandfather (as far as he goes) and eat him bit by bit. 'I begin to look ship-shape,' observes Ben Sparrow, gazing admiringly at the unfinished effigy of himself. 'You see, Tottie, what the people want nowadays is novelty--something new, something they haven't seen before. Give them that, and you're all right' (Which vague generality appears to satisfy him.) 'Now, here it is-- here's novelty--here's something they've never seen before; and if this don't bring custom, I don't know what will.'
- 61. Tottie gives a grave and silent assent; she cannot speak, for her mind is bent upon cannibalism. She is ready to tear the old man limb from limb. 'But,' continues Ben Sparrow, unconscious of the horrible thought at work in the mind of the apparently innocent child before him, 'I must get along with myself, or I shall never be finished. I haven't been in any battle that I know of, and I wasn't born a cripple, so my limbs must be all right when I appear in public. Now for my arms. More cinnamon! I think I may call cinnamon my bones.' When two pieces of cinnamon are stuck into the sides of the candied lemon-peel, they look so naked that he says, 'I must put sleeves on my arms.' And impales raisins upon them, and sticks five small slips of mace in each of the last raisins, which serve for fingers. 'Now for my legs, and there I am. More cinnamon!' Two sticks of cinnamon stuck in the bottom of his candied stomach, and then clothed with raisins, form his legs, and there he is, complete. 'I think I'll do,' he says complacently. At this moment a voice calls 'Shop!' and a fairy, in the shape of a shoeless ragged girl, taps upon the counter. Ben Sparrow goes into the shop to serve, and Tottie is left alone with his effigy. Now it has been mentioned above that Tottie has a vice, and this is it: she is afflicted, not with a raging tooth, but with a tooth so sweet as to weaken her moral sense, so to speak: she is unable to resist temptation when it presents itself to her in the shape of sweetmeats or fruit, and her notions as to the sacredness of such-like property are so loose that (no one being by to see her do it) she helps herself. And yet it is a proof that she possesses a wakeful
- 62. conscience, that she turns her back upon herself when she pilfers, as if she would wish to make herself believe that she is unconscious of what she is doing. Thus, seeing, say, a bowl of currants near, and no person within sight, she will approach the bowl stealthily, and, turning her back to it, will put her hand behind her, and take a fistful, with an air of thinking of something else all the while. And it is a proof that the moral obligation of her conscience is not entirely dormant, that, after the act is committed and enjoyed, she will, under the influence of a human eye, instantly defend herself without being accused, by 'No, I never! no, I never!' This express admission of guilt she can no more resist than she can resist the temptation itself. At the present time the sweet effigy of Ben Sparrow is lying within reach upon the table. Shutting her eyes. Tottie stretches out her hand, and plucking her grandfather's left leg bodily from his candied stomach, instantly devours it, cinnamon, raisins, and all-- and has just made the last gulp when Ben Sparrow, having served his customer, reënters the parlour. He casts a puzzled look at his dismembered effigy, and mutters, 'Well! if I didn't think I had made my two legs, may I be sugared!' Which sweet oath is exactly appropriate to the occasion. Then he turns to Tottie, who is gazing unconsciously at vacancy, with a wonderfully intense expression in her eyes, and she immediately shakes her head piteously, and cries, 'No, I never! no, I never!' Ben Sparrow, having his doubts aroused by this vehement asseveration of innocence, says mournfully, 'O, Tottie! Tottie! I didn't think you'd do it! To begin to eat me up like that!' But Tottie shakes her head still more vehemently, and desperately reiterates, 'No, I never! no, I never!' With the frightful consciousness
- 63. that the proofs of her guilt are in her inside, and that she has only to be cut open for them to be produced. Ben Sparrow, with a grave face, makes himself another leg, moving himself, however, out of Tottie's reach with reproachful significance. An unexpected difficulty occurs at this point. Being top- heavy he cannot balance himself upon his legs; but Ben is of an ingenious turn of mind, and he hits upon the expedient of shoring himself up from behind with stout sticks of cinnamon. Then, setting himself up, he gazes at himself in admiration. Tottie's eyes are also fixed upon the effigy; it possesses a horrible fascination for her. HERE AND THERE ARE FORGET-ME-NOTS. All night long Saul Fielding kneels by the side of his bed, absorbed in the memory of the woman whom he loves, and who, out of her great love for him, has deserted him. At first his grief is so great that he cannot think coherently; his mind is storm-tossed. But after a time the violence of his grief abates, and things begin to shape themselves in his mind. The night is cold, but he does not feel the winter's chill. The wind sighs and moans at his window, but he does not hear it. As it leaves his lattice, and travels through the courts and streets, it bears upon its wings the influence of the grief it has witnessed, and it sobs to the stone walls, 'There kneels a man in woe!' It gathers strength when it leaves the packed thoroughfares, which, huddled together like a crowd of beggars, seem to seek warmth in close contact, and becomes angry when it reaches the wide streets, angrier still when it reaches the woods, where the trees tremble as it rushes past them. Say that it rushes onward and still onward, and that we have the power to follow it--that we see it
- 64. merge into other winds, and become furious--that we see its fury die away--that we leave the winter and the night behind us--that we travel ahead of it over lands and seas until we come to where spring and daylight are--that we travel onward and still onward, until noon and spring are passed, and we come to where bright sun and summer are. Where are we? Thousands, upon thousands of miles away; but the time is the same, for as the warm wind kisses us we look back and see the man kneeling by the side of his bed. It is winter and night, and there kneels the man. It is summer and day, and here is another man among the mountains lying on the earth, looking at the clouds. And the time is the same. The thoughts of both these men are in the past. What connection can there be between these two in such adverse places, seasons, and circumstances? They have never touched hands. What link can bind them? Heart-links? Perhaps. It would not be so strange. It may be that at this present moment, in some distant part of the world of which we have only read or dreamt, links in your life's chain and mine are being forged by persons whose faces we have never seen. He is desolate. Jane has gone from him. She has left words of comfort behind her, but he may never look upon her face again. She has given him a task to fulfill. 'If I have done my duty by you,' she said, 'and I have tried to do it, it remains for you to do your duty by me.' He will be true to his dear woman, as she has been to him. He will strive to perform the task she has set before him--he will strive to find a way. Ay, if he dies in the attempt. He will consider presently how he shall commence. In the mean time, he must think of Jane. He falls into a doze, thinking of her, and with her in his mind the past comes to him. The aspirations which filled his boyish mind--his love for books--his desire to rise above his surroundings--his reasonings upon the relation of this and that, and his theoretical conclusions, which were to suddenly divert the common custom of things, as if a creation could in a moment crumble into dust the growth of centuries--his delight when he found that he was an orator, and could move an assembly of men to various passions--his
- 65. meeting with Jane---- He went no farther. The memory of her as she was when he first saw her, a bright flower--ah, how bright, how trustful and womanly!--stopped farther thought, and for a time no vision appears of his downfall, his weakness, his disgrace, his sinking lower, lower, until he is almost a lost man. It comes to him presently with all its shame; but when he wakes, the chaos of images in his mind resolves itself into this: his life is before him, full of weeds, like an untended garden, but here and there are Forget-me-nots, and each one bears the name of Jane. The morning light steals in upon his vigil, and still he has not decided how or in what way he shall commence his new life. In truth he is powerless. He has no weapons to fight with. His old confidence in himself, his pride, his strength of will, are covered with the rust of long weakness. Rising from his knees, he breaks the crust of ice upon the water in his pitcher, and bathes his face. The cold water seems to bring strength to him. He looks about the room, and everything within the poor walls speaks of Jane's love and care for him. The fire is laid with the last few sticks of wood and the last few lumps of coal. The old kettle, filled, is on the hob. The last pinch of tea is in the cup; the remains of the loaf are on the table. Not a thing is forgotten. 'Dear woman!' he murmurs. 'It is like you!' He paces the room slowly, striving to think of some path by which he can obtain a home for Jane, and thereby win her and reward her. It is useless, he knows, to seek for work here in the neighbourhood where he is known. He is known too well, and has sunk too low. Who would believe in his profession of amendment? Besides, what is the use of trying? He is of the same trade as George Naldret, and even George, a better workman than he, has resolved to leave, and try his fortune elsewhere, because of the difficulty he finds in saving sufficient money to buy a home for the girl he desires to marry. Even George is compelled to emigrate---- He stops suddenly in the middle of the room, and draws himself up with a spasmodic motion. Jane's words come to him: 'It is a blessing for many that these new lands have been discovered. A man can commence a new life there, without being crushed by the misfortunes or faults of the past, if he
- 66. be earnest enough to acquire strength. It might be a blessing to you.' 'A new life in a new land!' he says aloud. 'All the weakness and shame of the past wiped away because they will not be known to those around me. I should feel myself a new man--a better man; my strength, my courage would come back to me!' So strong an impression does the inspiration of the thought make upon him, that he trembles with excitement. But can he leave Jane--leave the country which holds her dear form? Yes, he can, he will; the memory of her will sustain him; and she will approve, as indeed she has done already by her words. 'It is the only way!' he cries; 'the only way!' Thus far he thinks, and then sinks into a chair, despairing. The means! How can he obtain the means? He has not a shilling in the world, nor any friends powerful enough to help him. Heaven's gate seems to be more easily accessible to him than this new land across the seas. But he does not allow himself to sink into the lowest depth of despondency. Jane stands before him; her words are with him; like wine they revive his fainting soul. 'Come, Saul,' he cries aloud to himself, resolutely. 'Come--think! Cast aside your weakness. Be your old self once more!' These words, spoken to himself as though they came from the lips of a strong man, sound like a trumpet in his ears, and really strengthen him. Again he thinks of George Naldret. 'Mr. Million gave him his passage ticket,' he says; 'would Mr. Million give me one?' No sooner has he uttered the words than the current of his thoughts is diverted, and he finds himself speculating upon the cause of Mr. Million's generosity to George. Friendship? No, it can scarcely be that. There can be no friendship between George and Mr. Million. Kindness? Perhaps; and yet he has never heard that Mr. Million was noted for the performance of kindly actions. These considerations trouble him somewhat on George's account, although he cannot explain to himself why the fact of Mr. Million giving George a free passage ticket to the other end of the world should cause him uneasiness. 'I wonder how it came about,' he thinks. 'I never heard George speak of emigrating until the ticket was promised to him. At all events, if George has any claim upon him, I have none. But Mr. Million is a public man, and may be in favour of emigration. It will cost him but little to assist me. There are Government emigration
- 67. ships which take a man over for almost nothing, I have heard. A line of recommendation from Mr. Million in my favour would be sufficient, perhaps. I will try; I will try. If I knew a prayer that would make my appeal successful, I would say it.' BATTLEDOOR AND SHUTTLECOCK. As a public man, James Million, Esquire, M.P. for Brewingham, felt it necessary to his position to spend two or three hours in his study every morning, and to 'make-believe' to be busy. Had you asked James Million what he was, he would not have told you that he was a brewer or a capitalist, but would have replied briefly and emphatically, 'A public man, sir.' Now, to be a public man, you must have a shuttlecock; and whether it was that Mr. Million had a real sympathy for the institution known as the working man, or because the working man drank large quantities of Million's Entire and Million's Treble X, it is certain that he set up the working man as his shuttlecock; and it is quite as certain that he set it up without in the least understanding it, being, indeed, a most unskilful player at any game in which his own interests were not directly involved. The game of battledoor and shuttlecock is a popular one with us from childhood upwards, but I am not aware that any close observer and noter of curious things has ever calculated how many shuttlecocks an ordinary battledoor will outlast. Popular as the game is with children, it is more popular with public men, who, battledoor in hand, are apt (in their enthusiasm and love for the game) to run into exceedingly wild, extremes when a new shuttlecock, with spick and span new feathers, is cast among them. Such a superabundance of energy do they in their zeal impart into the game that they often
- 68. sorely bruise the poor shuttlecock, and so knock it out of all shape and proportion that the members of its family find it impossible to recognise it. How many a poor shuttlecock have you and I seen on its last legs, as one might say, in a desperate condition from being much hit and much missed and much trodden into the mud, and with feathers that would rival those of a roupy old hen in the last stage of dissolution! and looking upon it in melancholy mood, may we not be excused for dwelling sadly upon the time (but yesterday!) when its feathers were new and crimson-tipped, and when it proudly took its first flight in the air? In appearance, James Million, the eminent brewer, was a small, flabby man, with a white face on which the flesh hung loosely. It had been said of him that his morals were as flabby as his flesh--but this was invented by a detractor, and if it conveyed any reproach, it was at best a hazy one. He had a curious trick with his eyes. They were sound and of the first water--not a flaw in them, as diamond merchants say; but whenever there was presented for his contemplation or consideration a question of a perplexing or disagreeable nature, he would close one of his eyes, and look at it with the other. It was a favourite habit with him to walk along the streets so, with one eye closed; and a man who set himself up for a satirist, or a wag, or both, once said: 'Jimmy Million is so moral that he doesn't like to look on the wickedness of the world; so he shuts one eye, and can only see half of it, and thereby saves himself half the pain.' To James Million, as he sits in his study, comes a servant, who, after due tapping at the door, so as not to disturb the ruminations of the legislator, announces a man in the passage who desires to see Mr. Million. 'Name? asks Mr. Million.
- 69. 'Saul Fielding,' answers the servant, and adds, 'but he says he does not think you know him.' 'What does he look like?' The servant hesitates; he has not made up his mind. Although Saul Fielding is shabbily dressed, he is clean, and Jane's watchful care has made his wardrobe (the whole of which he wears on his back) seem better than it is. Besides, there is 'an air' about Saul Fielding which prevents him being placed, in the servant's mind, on the lowest rung of vagabondism. 'Is he a poor man? Is he a working man? demands Mr. Million impatiently.
- 70. 'He looks like it, sir,' replies the servant, not committing himself distinctly to either statement. Mr. Million has an idle hour before him, which he is not disinclined to devote to the workingman question, so he bids the servant admit the visitor. 'Wait a minute,' says Mr. Million to Saul Fielding as he enters the room. Mr. Million evidently has found some very knotty problem in the papers before him, for he bends over them, with knitted brows and studious face, and shifts them about, and makes notes on other pieces of paper, and mutters 'Pish!' and 'Psha!' and 'Very true!' and 'This must be seen to!' with many remarks indicative of the engrossing nature of the subject which engages his attention. After a sufficient exhibition of this by-play, which doubtless impresses his visitor with a proper idea of his importance and of the immense interest he takes in public matters, he pushes the papers aside with a weary air, and looks up, with one eye closed and one eye open. What he sees before him does not seem to afford him any comfort: for it is a strange thing with public players of battledoor and shuttlecock, that although they have in theory a high respect for their shuttlecocks, they have in absolute fact a very strong distaste for them. Seeing that he is expected to speak, Saul Fielding commences; he is at no loss for words, but he speaks more slowly than usual, in consequence of the heavy stake he has in the interview. 'I have ventured to call upon you, sir,' he says, 'in the hope that you will take some interest in my story, and that you will extend a helping hand to a poor man.' Somewhat fretfully--for careful as he strives to be, Saul Fielding has been unwise in his introduction, which might be construed into an appeal for alms--somewhat fretfully, then, Mr. Million interposes with
- 71. 'A working man?' 'I hope I may call myself so--although, strictly speaking, I have done but little work for a long time.' Mr. Million gazes with curiosity at his visitor, and asks, in a self- complacent, insolent tone, as if he knows all about it, 'Not able to get work, eh? 'I have not been able to get it, sir.' 'But quite willing to do it if you could get it?' 'Quite willing, sir more than willing--thankful.' Saul Fielding knows that already he is beginning to lose ground,' but his voice is even more respectful and humble than at first-- although the very nature of the man causes him to speak with a certain confidence and independence which is eminently offensive to the delicate ears of the friend of the working man. 'Of course!' exclaims Mr. Million triumphantly and disdainfully. 'The old cry! I knew it. The old cry! I suppose you will say presently that there is not room for all, and that there are numbers of men who are in the same position as yourself--willing to work, unable to obtain it.' Saul Fielding makes no reply; words are rushing to his tongue, but he does not utter them. But Mr. Million insists upon being answered, and repeats what he has said in such a manner and tone that Saul cannot escape. 'I think, sir, that there are many men who are forced to be idle against their will; that seems to be a necessity in all countries where population increases so fast as ours does. But I don't complain of that.'
- 72. 'O!' cries Mr. Million, opening both his eyes very wide indeed. 'You don't complain of that! You are one of those glib speakers, I have no doubt, who foment dissatisfaction among the working classes, who tell them that they are down-trodden and oppressed, and that masters are fattening upon them! I should not be surprised to hear that you are a freethinker.' 'No, sir, I am not that,' urges Saul Fielding, exquisitely distressed at the unpromising turn the interview has taken; 'nor indeed have I anything to complain of myself. I am too crushed and broken-down, as you may see.' 'But if you were not so,' persists Mr. Million, growing harder as Saul grows humbler, 'if you were in regular work, and in receipt of regular wages, it would be different with you--eh? You would have something to complain of then doubtless. You would say pretty loudly that the working man is underpaid, and you would do your best to fan the flame of discontent kept up by a few grumblers and idlers. You would do this--eh? Come, come,' he adds haughtily, seeing that Saul Fielding does not wish to answer; 'you are here upon a begging petition, you know. Don't you think it will be best to answer my questions?' 'What is it you wish me to answer, sir?' asks Saul Fielding sorrowfully. 'The question of wages. I want to ascertain whether you are one of those who think the working classes are underpaid.' Saul Fielding pauses for a moment; and in that brief time determines to be true to himself. 'Jane would not have me do otherwise,' he thinks. 'I think, sir,' he says, firmly and respectfully, 'that the working classes--by which I mean all in the land who have to work with their hands for daily bread--do not receive, as things go, a fair equivalent for their work. Their wages are not sufficient. They seem to me to
- 73. be framed upon a basis which makes the work of ekeing them out so as to make both ends meet a harder task than the toil by which they are earned. The working man's discontent does not spring from his work; he does that cheerfully almost always. It springs out of the fact, that the results of his work are not sufficient for comfort, and certainly not sufficient to dispel the terrible anxiety which hangs over the future, when he is ill and unable to work, perhaps, or when he and his wife are too old for work.' 'O, indeed!' exclaims Mr. Million. 'You give him a wife!' 'Yes, sir; his life would be a burden indeed without a woman's love.' Mr. Million stares loftily at Saul Fielding. 'And children, doubtless!' 'Happy he who has them! It is Nature's law; and no man can gainsay it.' The theme possesses a fascination for Saul Fielding, and he continues warmly, 'I put aside as distinctly outrageous all that is said of the folly and wickedness of poor people marrying and having large families. This very fact, which theorists wax indignant over-- theorists, mind you, who have wives and families themselves, and who, by their arguments, lay down the monstrous proposition that nature works in the blood according to the length of a man's purse-- this very fact has made England strong; had it been otherwise, the nation would have been emasculated. Besides, you can't set natural feeling to the tune of theory; nor, when a man's individual happiness is concerned, can you induce him to believe in the truth of general propositions which, being carried out in his own person as one of the units, would make his very existence hateful to him.' Mr. Million opens his eyes even wider than before; such language from the lips of the ragged man before him is indeed astonishing.
- 74. 'What more have you to say? he gasps. 'You will want property equally divided----' 'No, sir, indeed,' interrupts Saul Fielding, daring to feel indignant, even in the presence of so rich a man, at the suggestion. 'The man who makes honestly for himself is entitled to possess and enjoy. I am no socialist.' 'You would, at all events,' pursues Mr. Million, 'feed the working man with a silver spoon?--You would open the places of amusement for him on the Sabbath?' 'I would open some places and shut others.' 'What places, now?' 'The museums, the public galleries. I would give him every chance--he has a right to it--to elevate himself during the only leisure he has.' 'And in this way,' demands Mr. Million severely, 'you would desecrate the Sabbath!' For the life of him Saul Fielding cannot help saying, 'A greater desecration than even that can be in your eyes takes place on the Sabbath in places that are open in the name of the law.' 'You refer to----' 'Public-houses. If they are allowed to be open, what reasonable argument can be brought against the opening of places the good influence of which is universally acknowledged? It is the withholding of these just privileges that causes much discontent and ill-feeling.'
- 75. This is quite enough for Mr. Million. This man, ragged, penniless, has the effrontery to tell the rich brewer to his face that he would have the public picture-galleries and museums of art opened on the Sabbath-day, and that he would shut the public-houses. Mr. Million can find no words to express his indignation. He can only say, stiffly and coldly, 'I have heard quite enough of your opinions, sir. Come to the point of your visit. You see'--pointing to the papers scattered about the table--'that I am very busy.' 'I came, sir,' he says sadly, 'in the hope that, seeing my distress, you would not have been disinclined to assist me--not with money, sir,' he adds swiftly, in answer to an impatient look of dissent from Mr. Million, 'but with your good word. But I am afraid that I have injured my cause by the expression of my opinions.' 'In what way did you expect that I could aid you?' asks Mr. Million carelessly, as he settles himself to his papers. 'I have been especially unfortunate in my career, sir. As I told you, I am willing to work, but am unable to obtain it. If I could emigrate; if I could get into a new country, where labour is scarce, things might be better for me.' The poor man is helpless at the rich man's foot; and the rich man plays with him, as a cat with a mouse. 'Well,' he says, 'emigrate. The country would be well rid of such as you.' Saul Fielding takes no notice of the insult. He is not to be turned aside from his purpose, although he knows full well that he has missed his mark. 'I have no means, sir; I am poor and helpless.'
- 76. 'How do you propose to effect your object, then?' 'There are Government emigrant ships which take men out, I have heard, for very little--for nothing almost. A line of recommendation from you would be sufficiently powerful, I thought, to obtain me a passage.' 'Doubtless, doubtless,' this with a smile; 'but you are a man of some perception, and having observed how utterly I disagree with your opinions--which I consider abominable and mischievous to the last degree--you can hardly expect me to give you the recommendation you ask for. May I ask, as you are a perfect stranger to me, for I have no recollection of you in any way, to what I am indebted for the honour you have done me by choosing me to give you a good character?' 'You are a public man, sir, and I have heard a friend to the working man. And as you had helped a friend of mine to emigrate by giving him a free passage in a ship that sails this week----' 'Stop, stop, if you please. I help a friend of yours to emigrate by giving him a free passage! I think you are mistaken.' 'If you say so, sir, I must be. But this is what George Naldret gave me to understand.' 'And pray who is George Naldret?' demands Mr. Million haughtily; 'and what are his reasons for emigrating?' 'George Naldret,' returns Saul Fielding, in perplexity, 'is almost the only friend I have in the world, and he is emigrating for the purpose of putting himself into a position to marry more quickly than his prospects here will allow him.' 'As you are introducing me,' says Mr. Million, with an air of supreme indifference, 'to your friends, perhaps you would like also to
- 77. introduce me to the young lady--for of course' (with a sneer) 'she is a young lady--he desires to marry.' 'Her name is Sparrow--Bessie Sparrow, granddaughter to an old grocer.' Mr. Million becomes suddenly interested, and pushes his papers aside with an exclamation of anger. 'What name did you say?' 'Miss Bessie Sparrow.' The rich brewer ponders for a moment, evidently in no pleasant mood. Then suddenly rings a bell. A servant appears. 'Is my son in the house?' 'Yes, sir.' 'Tell him to come to me instantly.' Saul Fielding waits gravely. Seemingly, he also has found new food for contemplation. Presently young Mr. Million appears. 'You sent for me, sir.' 'Yes, James. Do you know this person?' with a slight wave of the hand in the direction of Saul Fielding, as towards a thing of no consequence. Saul Fielding knows that his mission has failed, but does not resent this contemptuous reference to him. He stands, humble and watchful, before father and son. 'I have seen him,' says young Mr. Million, 'and I should say he is not a desirable person in this house.' 'My opinion exactly. Yet, influenced by some cock-and-a-bull story, he comes here soliciting my assistance to enable him to emigrate.
- 78. The country would be well rid of him, I am sure; but of course it is out of my power to give such a person a good character to the emigration commissioners.' 'Out of anybody's power, I should say,' assents young Mr. Million gaily. 'To what cock-and-a-bull story do you refer?' 'He tells me--which is news to me--that I have given a free passage ticket to a friend of his, George--George--what did you say?' 'George Naldret, sir.' Saul Fielding supplies the name in a manner perfectly respectful. 'Ay--George Naldret. Such a statement is in itself, of course, a falsehood. Even if I knew George Naldret, which I do not, and desired to assist him, which I do not, the fact of his being engaged to be married to any one of the name of Sparrow--a name which means disgrace in our firm, as you are aware-would be sufficient for me not to do so.' Young Mr. Million steals a look at Saul Fielding, whose face, however, is a mask; and in a hesitating voice says: 'I think I can explain the matter; but it is not necessary for this person to remain. You do not know, perhaps, that he was the chief mover in a strike a few years ago, which threatened to do much mischief.' 'I am not surprised to hear it,' says the rich brewer; 'the opinions he has expressed have prepared me for some such statement concerning him. He would desecrate the Sabbath-day by opening museums and picture-galleries, and he would curtail the liberty of the subject by closing public-houses, and depriving the working man of his beer! Monstrous! monstrous! He has nothing to say for himself, I suppose.' 'No, sir,' answers Saul Fielding, raising his head, and looking steadily at young Mr. Million, 'except that I believed in the truth of
- 79. what I told you, and that I don't know whether I am sorry or glad that I made the application to you.' The rich brewer has already touched the bell, and the servant comes into the room. 'Show this person to the door,' Mr. Million says haughtily; 'and if he comes again, send for a policeman. He is a dangerous character.' Saul Fielding's lips wreathe disdainfully, but he walks out of the room, and out of the house, without a word of remonstrance. This chance has slipped from him. Where next shall he turn? He walks slowly onwards until he is clear of the rich brewer's house, and then stops, casting uncertain looks about him. As a sense of his utter helplessness comes upon him, a young woman brushes past him without seeing him. He looks up. Bessie Sparrow! She is walking quickly, and seems to see nothing, seems to wish to see nothing. Without any distinct purpose in his mind, but impelled by an uncontrollable undefinable impulse, Saul Fielding turns and follows her. A gasp of pain escapes him as he sees her pause before Mr. Million's house. She rings the bell, and the door is opened. She hands the servant a letter, and the next moment she is in the house, shut from Saul Fielding's view. The terror that comes upon him is so great that the street and the sky swim before his eyes, and he clings to a lamp-post for support. 'O, George!' he groans. 'O, my friend! How will you bear this? Good God! what bitterness there is in life even for those who have not fallen as I have done!' TOTTIE'S DREAM.
- 80. When Tottie was put to bed, it was no wonder that she was haunted by the sweet effigy of old Ben Sparrow, and that his stomach of candied lemon-peel and his head of rich figs and currants presented themselves to her in the most tempting shapes and forms her warm imagination could devise. As she lay in bed, looking at the rushlight in the washhand basin, the effigy appeared bit by bit in front of the basin until it was complete, and when it winked one of its currant eyes at her--as it actually did--the light of the candle threw a halo of glory over the form. Her eyes wandering to the mantelshelf, she saw the effigy come out of the wall and stand in the middle of the shelf; and turning to the table, it rose from beneath it, and sat comfortably down; with its legs of cinnamon and raisins tucked under it like a tailor. When she closed her eyes she saw it loom in the centre of dilating rainbow circles, and in the centre of dark-coloured discs, which as they swelled to larger proportions assumed bright borderings of colour, for the express purpose of setting off more vividly the attraction of the figure. Opening her eyes drowsily, she saw the old man come down the chimney and vanish in the grate, and as he disappeared, down the chimney he came again, and continued thus to repeat himself as it were, as if he were a regiment under full marching orders. Whichever way, indeed, Tottie's eyes turned, she saw him, until the room was full of him and his sweetness, and with his multiplied image in her mind she fell asleep. No wonder that she dreamt of him. Tottie and Bessie slept in the same room, and Tottie dreamt that long after she fell asleep--it must have been long after, for Bessie was in bed--she woke up suddenly. There she was, lying in bed, wide awake, in the middle of the night. The room was dark, and she could not see anything, but she could hear Bessie's soft breathing. She was not frightened, as she usually was in the dark, for her attention was completely engrossed by one feeling. A frightful craving was upon her, which every moment grew stronger and stronger. This craving had something horrible in it,
- 81. which, however, she did not quite realise. In the next room slept old Ben Sparrow, who, according to the fancy of her dream, was not made of blood, and flesh, and bone, but of lemon-peel, fig, and currants and raisins. All the sweet things in the shop had been employed in the manufacture, and there they lay embodied in him. Tottie knew nothing of theology; knew nothing of the value of her soul, which, without a moment's hesitation, she would have bartered for figs and candied lemon-peel. And there the delicious things lay, in the very next room. If she could only get there!--perhaps he would not miss an arm or a leg. But to eat the old man who was so kind to her! She had a dim consciousness of the wickedness of the wish, but she could not rid herself of it. Thought Tottie, 'He won't know if he's asleep, and perhaps it won't hurt him. I know it would do me good.' Her mouth watered, her eyes glistened, her fingers twitched to be at him, her stomach cried out to her. She could not withstand the temptation. Slowly and tremblingly she crept out of bed, and groped along the ground towards the door. Bessie was asleep. Everybody was asleep. The house was very quiet. Everything favoured the accomplishment of the horrible deed. 'Nobody will know,' thought Tottie. Thoroughly engrossed in her desperate cannibalistic purpose, and with her teeth grating against each other, Tottie turned the handle of the door and opened it; but as she looked into the dark passage Ben Sparrow's door opened, and a sudden flood of light poured upon her. It so dazzled her, and terrified her, that she fled back to her bed on all-fours, and scrambled upon it, with a beating heart and a face as white as a ghost's. Sitting there glaring at the door, which she had left partly open in her flight, she saw the light steal into the room, and flying in the midst of it, old Ben Sparrow. He was not quite as large as life, but he was ever so many times more sweet and delicious-looking. As old Ben Sparrow appeared, the room became as light as day, and Tottie noticed how rich and luscious were the gigantic fig which formed his head, the candied lemon-peel which formed his stomach, the raisins which clothed his legs and arms; and as for the ripeness of his dark, beady, fruity eyes, there was no form of thought that could truly express
- 82. the temptation that lay in them. Ben Sparrow hovered in the air for a few moments, and then steadied himself, as it were: he stood bolt upright, and, treading upon nothing; advanced slowly and solemnly, putting out one leg carefully, and setting it down firmly upon nothing before he could make up his mind to move the other. In this manner he approached Tottie, and sat down on her bed. For a little while Tottie was too frightened to speak. She held her breath, and waited with closed lips for him to say something. But as grandfather did not move or speak, her courage gradually returned, and with it, her craving for some of him. She became hungrier than the most unfortunate church-mouse that ever breathed; her rapacious longing could only be satisfied in one way. Timorously she reached out her hand towards his face; he did not stir. Towards his eyes; he did not wink. Her finger touched his eye; it did not quiver--and out it came, and was in her hand! Her heart throbbed with fearful ecstasy, as with averted head she put the terrible morsel in her mouth. It was delicious. She chewed it and swallowed it with infinite relish, and, when it was gone, thirsted for its fellow. She looked timidly at the old man. There was a queer expression in his fig face, which the loss of one of his eyes had doubtless imparted to it. 'It doesn't seem to hurt him,' thought Tottie. Her eager fingers were soon close to the remaining eye, and out that came, and was disposed of in like manner. Tottie certainly never knew how good Ben Sparrow was until the present time. She had always loved him, but never so much as now. The eyeless face had a mournful expression upon it, and seemed to say sadly, 'Hadn't you better take me next?' Tottie clutched it desperately. It wagged at her, and from its mace lips a murmur seemed to issue, 'O, Tottie! Tottie! To serve me like so this!' But Tottie was ravenous. No fear of consequences could stop her now that she had tasted him, and
- 83. found how sweet he was. She shut her eyes nevertheless, as, in the execution of her murderous purpose, she tugged at his head, which, when she had torn from his body, she ate bit by bit with a rare and fearful enjoyment. When she looked again at the headless figure of the old man, one of the legs moved briskly and held itself out to her with an air of 'Me next!' in the action. But Tottie, hungering for the lemon-peel stomach, disregarded the invitation. It was difficult to get the stomach off, it was so tightly fixed to its legs. When she succeeded, the arms came with it, and she broke them off short at the shoulder blade, and thought she heard a groan as she performed the cruel operation. But her heart was hardened, and she continued her feast without remorse. How delicious it was! She was a long time disposing of it, for it was very large, but at length it was all eaten, and not a piece of candied sugar was left. As she sucked her fingers with the delight of a savage, a sense of the wickedness of what she had done came upon her. Her grandfather, who had always been so kind to her! She began to tremble and to cry. But the arms and legs remained. They must be eaten. Something dreadful would be done to her if they were discovered in her bed; so with feverish haste she devoured the limbs. And now, not a trace of the old man remained. She had devoured him from head to foot She would never see him again--never, never! How dreadful the table looked, with him not on it! How Tottie wished she hadn't done it! She was appalled at the contemplation of her guilt, and by the thought of how she would be punished if she were found out. In the midst of these fears the light in the room vanished, and oblivion fell upon Tottie in the darkness that followed. I CAN SEE YOU NOW, KISSING HER LITTLE TOES.
- 84. The next day, being George's last day at home, was a day of sorrow to all the humble persons interested in his career. He, was to start for Liverpool by an early train on the following morning, and was to pass his last evening at Ben Sparrow's, with the old man and Bessie and Tottie and his mother and father. He had decided to bid Bessie good-bye in her grandfather's house. Bessie was for sitting up all night, but he said gently, 'I think, Bessie, that mother would like to have me all to herself the last hour or two. You know what mothers are! By and by, heart's treasure! you will have the first claim on me; but now mother looks upon me as all her own, and it will comfort her heart, dear soul! to let it be as I say.' There were tears in George's eyes as he looked down upon the face of his darling, and his heart almost fainted within him at the thought of parting from her. And, 'Do you love me, Bess?' he asked, for the thousandth time. 'With all my heart and soul,' replied Bessie, pressing him in her arms. And so, with his head bowed down to hers, they remained in silent communion for many minutes. They were sitting in Ben Sparrow's parlour, and the old man had left the young people by themselves, finding occupation in his shop, in the contemplation of his effigy, and in weighing up quarters of a pound of sugar. There was a woful look in Ben Sparrow's face as he stood behind his counter; times were hard with him, and his till was empty. 'Bess, darling,' said George, waking up from his dream. She raised her tearful eyes to his. He kissed them. 'As I kiss away your tears now, my dear, so I will try to take sorrow and trouble from you when we commence our new life.' 'I know it, George; I know it,' she said, and cried the more.
- 85. 'But that is not what I was going to say. I was going to say this. Listen to me, dearest: If it were not for you, I shouldn't go; if it were not for you, I should stay at home, and be content. For I love home, I love the dear old land, I love mother and father, and the old black cat, and the little house I was born in. And it's because of you that I am tearing myself from these dear things. I am going to earn money enough to make a home for you and me; to make you more quickly all my own, all my own! How my heart will yearn for you, dear, when I am over the seas! But it will not be for long; I will work and save, and come back soon, and then, my darling, then!----' The tenderness of his tone, and the tenderness there was in the silence that followed, were a fitter and more expressive conclusion to the sentence than words could have made. 'I shall say when I am in the ship, I am here for Bessie's sake. When I am among strangers, I shall think of you, and think, if I endure any hardship, that I endure it for my darling--and that will soften it, and make it sweet; it will, my dear! I shall not be able to sleep very much, Bess, and that will give me all the more hours to work--for you, my darling, for you! See here, heart's treasure; here is the purse you worked for me, round my neck. It shall never leave me--it rests upon my heart. The pretty little beads! How I love them! I shall kiss every piece of gold I put in it, and shall think I am kissing you, as I do now, dear, dearest, best! I shall live in the future. The time will soon pass, and as the ship comes back, with me in it, and with my Bessie's purse filled with chairs and tables and pots and pans, I shall see my little girl waiting for me, thinking of me, longing to have me in her arms as I long to have her in mine. And then, when I do come, and you start up from your chair as I open the door!----Think of that moment, Bess--think of it!' 'O, George, George, you make me happy!' And in such tender words they passed the next hour together, until George tore himself away to look after some tools, which he was to take with him to coin chairs and tables and pots and pans with. But if he did not wish his tools to rust, it behoved him not to
- 86. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com