https://ebookgate.com/product/reliabilism-andcontemporary-epistemology-essays-1st-edition-
More products digital (pdf, epub, mobi) instant download maybe you interests ...
Epistemology A Contemporary Introduction 1st Edition
Alvin I. Goldman
https://ebookgate.com/product/epistemology-a-contemporaryintroduction-1st-edition-alvin-i-goldman/
Philosophical Applications of Cognitive Science 1st Edition Alvin I. Goldman
https://ebookgate.com/product/philosophical-applications-ofcognitive-science-1st-edition-alvin-i-goldman/
Simulating Minds The Philosophy Psychology and Neuroscience of Mindreading Alvin I. Goldman
https://ebookgate.com/product/simulating-minds-the-philosophypsychology-and-neuroscience-of-mindreading-alvin-i-goldman/
Living Skepticism Essays in Epistemology and Beyond 1st Edition Stephen Hetherington
https://ebookgate.com/product/living-skepticism-essays-inepistemology-and-beyond-1st-edition-stephen-hetherington/
A Beholder s Share Essays on Winnicott and the Psychoanalytic Imagination 1st Edition Dodi Goldman
https://ebookgate.com/product/a-beholder-s-share-essays-onwinnicott-and-the-psychoanalytic-imagination-1st-edition-dodigoldman/
Autobiography and Literary Essays I Autobiography and Literary Essays 1st Edition John Stuart Mill
https://ebookgate.com/product/autobiography-and-literary-essaysi-autobiography-and-literary-essays-1st-edition-john-stuart-mill/
Strategies of Argument Essays in Ancient Ethics
Epistemology and Logic 1st Edition Mi-Kyoung Lee (Ed.)
https://ebookgate.com/product/strategies-of-argument-essays-inancient-ethics-epistemology-and-logic-1st-edition-mi-kyoung-leeed/
Phenomenal reading essays on modern and contemporary poetics 1st Edition Edition Reed
https://ebookgate.com/product/phenomenal-reading-essays-onmodern-and-contemporary-poetics-1st-edition-edition-reed/
Goldman and His Critics 1st Edition Brian P. Mclaughlin
https://ebookgate.com/product/goldman-and-his-critics-1stedition-brian-p-mclaughlin/
Reliabilism and Contemporary Epistemology
This page intentionally left blank
Reliabilism and Contemporary Epistemology
ESSAYS
Alvin I. Goldman
Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education.
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Copyright © 2012 Oxford University Press
Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
www.oup.com
Oxford is a registered trademark of Oxford University Press
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, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data
Goldman, Alvin I., 1938–Reliabilism and contemporary epistemology : essays. p. cm.
ISBN 978–0–19–981287–5 (alk. paper)
1. Knowledge, Theory of. 2. Justification (Theory of knowledge) 3. Social epistemology. I. Title.
BD175.G645 2012 121—dc23 2011022188
ISBN-13: 9780199812875
1 3 5 7 9 8 6 4 2
Printed in the United States of America on acid-free paper
Acknowledgments vii
Introduction 3
1. What Is Justified Belief? 29
2. Immediate Justification and Process Reliabilism 50
3. Reliabilism 68
4. Internalism, Externalism, and the Architecture of Justification 95
5. Toward a Synthesis of Reliabilism and Evidentialism? 123
6. Reliabilism and the Value of Knowledge 151
7. Williamson on Knowledge and Evidence 175
8. Epistemic Relativism and Reasonable Disagreement 197
9. A Guide to Social Epistemology 221
10. Why Social Epistemology is Real Epistemology 248
11. Philosophical Naturalism and Intuitional Methodology 280
Subject Index 317
Author Index 319
This page intentionally left blank
ACKNOWLEDGMENTS
“What Is Justified Belief?” in G. Pappas, ed., Justification and Knowledge (Dordrecht: Reidel, 1979). With kind permission from Springer Science+Business Media B.V.
“Immediate Justification and Process Reliabilism,” in Q. Smith, ed., Epistemology: New Essays (New York: Oxford University Press, 2008).
“Reliabilism,” The Stanford Encyclopedia of Philosophy (Spring 2011 Edition), E. N. Zalta, ed., http://plato.stanford.edu/entries/reliabilism/.
“Internalism, Externalism, and the Architecture of Justification,” The Journal of Philosophy 106: 6 (2009): 309–38.
“Toward a Synthesis of Reliabilism and Evidentialism,” in T. Dougherty, ed., Evidentialism and Its Discontents (New York: Oxford University Press, 2011).
“Reliabilism and the Value of Knowledge,” in A. Haddock, A. Millar, and D. Pritchard, eds., Epistemic Value (New York: Oxford University Press, 2009).
“Williamson on Knowledge and Evidence,” in P. Greenough and D. Pritchard, eds., Williamson on Knowledge (New York: Oxford University Press, 2009).
“Epistemic Relativism and Reasonable Disagreement,” in R. Feldman and T. A. Warfield, eds., Disagreement (New York: Oxford University Press, 2010).
“A Guide to Social Epistemology,” first published under the title “SystemsOriented Social Epistemology,” in T. S. Gendler and J. Hawthorne, eds., Oxford Studies in Epistemology, vol. 3 (New York: Oxford University Press, 2010).
“Why Social Epistemology Is Real Epistemology,” in A. Haddock, A. Millar, and D. Pritchard, eds., Social Epistemology (New York: Oxford University Press, 2010).
“Philosophical Naturalism and Intuitional Methodology,” The Proceedings and Addresses of the American Philosophical Association 84: 2 (2010): 115–50.
The author and publisher would like to thank the publishers for permission to reproduce these papers in this volume.
This page intentionally left blank
Reliabilism and Contemporary Epistemology
This page intentionally left blank
Introduction
Reliabilism is a prominent approach to core topics in epistemology, offering a spectrum of related analyses of knowledge, justification, and evidence.1 These epistemic goods are said to arise from the deployment of mental processes and methods, or the obtaining of suitable conditions, that are conducive to acquiring true belief and avoiding error in actual and/or modally relevant circumstances.
Some historical epistemologies might be viewed as precursors of reliabilism,2 but this introduction concerns the contemporary scene and its twentieth-century background. Ramsey (1931) was the first to offer a (brief) formulation of knowledge reliabilism, but it attracted little notice at the time.3 Forty years later, the times were more propitious, and a flurry of reliabilist approaches to knowledge burst onto the epistemological stage and gathered steam during ensuing decades (Armstrong, 1973; Dretske, 1971, 1981; Goldman, 1975, 1976, 1986; Nozick, 1981; Swain, 1981; Williamson, 2000). Reliabilism about justification—specifically, process reliabilism—was first formulated by Goldman (1979), followed by several permutations and refinements of this initial version (Goldman 1986, 1988, 1992; Alston, 1995; Comesana, 2002; Lyons, 2009). Justification reliabilism is probably more controversial than knowledge reliabilism (Kornblith, 2008) but arguably more challenging and signifi cant. Justification reliabilism is the principal focus of the present volume, though knowledge reliabilism and evidence reliabilism also receive substantial attention (knowledge in essays 6 and 7, evidence in essays 7 and 11).
At its inception, process reliabilism was a heretical doctrine, out of sync with the reigning epistemologies of the twentieth century. For starters, it departed from the prevailing view that justification depends entirely on what is “directly accessible” to the mind. Reliability—i.e., conduciveness to truth—is clearly not directly accessible in this sense. Similarly, process reliabilism was controversial because of the role it assigns to psychological processes that
cause or generate beliefs. Previously, justification had been considered exclusively a matter of reasons, not causes.4 Finally, traditional (Cartesian) epistemology held that justification hinges on the state of the epistemic subject at the time of belief, not on its earlier provenance or genesis. For all of these reasons, the philosophical atmosphere was inhospitable, if not downright hostile, to process reliabilism when it first appeared. These theses are still embraced by many if not most internalists. As a result, the debate between internalism and externalism has not disappeared; important issues remain in contention. Even restricting attention to externalism, process reliabilism is not the sole contender. And the precise form it ought to take and its relationship with other currently popular epistemologies remain to be reconsidered, clarified, and refined.
For all of these reasons, I have continued to work on reliabilism—and extensions of reliabilism (e.g., to social epistemology)—until the present day. They are why ten of the eleven essays reprinted here were written and published in just the last few years. For organizational purposes, this introduction groups the essays into five clusters. It highlights their main themes and adds many new reflections not in the essays themselves.
Cluster 1: Essays 1, 2, and 3
“What is Justified Belief?” (essay 1) is the original formulation of process reliabilism (for justification). The core argument in this essay appeals to straightforward, intuitive examples intended to show, first, that the process or method by which a belief is formed—or subsequently retained—is critical to its justificational status (J-status). Proceeding from this point, the question arises: which properties of belief-causing processes are responsible for the J-statuses of the output beliefs? Additional examples are adduced that pinpoint truth-conduciveness as the critical factor. When beliefs are the outputs of highly reliable processes, they are intuitively justified; when they are the outputs of notably unreliable processes, they are intuitively unjustified. These considerations constitute what one philosopher has called the “master argument” for the view: simple considerations that lend powerful prima facie support for process reliabilism. Stated in its simplest form, process reliabilism says that a belief is justified if and only if it is caused by a suffi ciently reliable process, or sequence of reliable processes, where the threshold of sufficiency is left vague (as befits the notion of justification, which itself is vague).5
The remainder of “What Is Justified Belief?” addresses issues of proper formulation, to keep the reliability-style conditions from being either too weak or too strong. At the same time, it is an attractive assumption for epistemological theory in general that justification—a normative status—should supervene on, or be grounded in, non-epistemic states of affairs. So it is natural to place a
constraint on the conditions of justifiedness to reflect this assumption. In a recursive definition of justification, base-clauses cannot advert to conditions that feature epistemic justification. For example, it would be inadmissible for a base-clause to say that a target belief is justified if (A) it is reliably caused and (B) the epistemic subject believes justifiably that it is reliably caused. However, this constraint carries some serious bite. Condition (A) looks too weak to serve as a lone sufficient condition of justifiedness, and condition (B) looks like an attractive supplement. But (B) is excluded by the proposed constraint. If condition (B) were replaced by one that omits the justification qualifier, the prospects would look much less promising. For example, consider condition (B’), that the subject (merely) believes that the first-order belief is reliably caused. Does condition (B’), paired with (A), guarantee justifiedness? Intuitively not. Why should mere belief in reliability—which might itself be a random, unjustified belief—confer justifiedness on a first-order belief? So, these kinds of higher-order requirements are either excluded or unhelpful.
The foregoing comments principally relate to the sufficiency of candidate principles of justification. There is also a problem of necessity. Insisting that subjects believe in the reliability of their processes might require excessive reflection, reflection not customarily realized by many ordinary epistemic subjects, including young children. A different problem, returning to the sufficiency question, is why a single higher level of believed reliability should satisfy philosophers who hanker for meta-level requirements. If plain, vanilla-flavored reliability doesn’t suffice for justification, why should a single meta-level of believed reliability suffice? Why not impose additional meta-level requirements: a second meta-level, a third meta-level, etc.? This road, however, presents a skepticism-inducing regress.
Although positive enhancements of the foregoing type seem too strong, “What Is Justified Belief?” acknowledges a need for an added negative requirement, roughly, a non-defeat requirement. One such requirement was proposed there, and others were proposed elsewhere (see my Epistemology and Cognition, p. 63.) Other refinements include a distinction between processes that belong to our natural endowment and processes, or “methods,” acquired through learning. Epistemology and Cognition proposed that learned methods must be acquired by meta-reliable processes in order to qualify as conferrers of “full” justification.6
Some thorny problems for process reliabilism were acknowledged from the outset. The well-advertised generality problem, for example, was first pinpointed with abundant clarity in “What Is Justified Belief?” Also acknowledged was the fact that process reliabilism is mainly a theory of doxastic (“ex post”) justification rather than propositional (“ex ante”) justification. The question arises, therefore, whether reliabilism can provide a satisfactory account of propositional justification. Such an account was offered, using the previously analyzed notion of doxastic justification.
Successful theories in any domain must show their colors by coping with well-identified problems at least as ably as their competitors. In this vein, process reliabilism, to be successful, must display an ability to solve long-standing problems in epistemology at least as well as its rivals. Essay 2, “Immediate Justification and Process Reliabilism,” tries to demonstrate such an ability. Foundationalism, a vintage approach to the structure of justification, posits a layer of justified beliefs that rests at the bottom of an inferential hierarchy. Beliefs get to be justified either by being basic beliefs, meaning that they are uninferred from other beliefs yet somehow “directly” or “immediately” justified, or by being properly inferred from one or more basic beliefs. Everybody agrees that suitable inference is capable of transmitting justification from antecedently justified beliefs to further beliefs. But how does any non-inferential belief acquire justification? Whichever form of foundationalism one favors—a form that restricts basic beliefs to ones about the subject’s own mental states or a form that allows them to be about the external world—this question is pressing.
Assume that basic beliefs include ones with external-world contents, such as “There is a table before me.” And assume for the sake of argument that the beliefs in question are non-inferential. What property, then, endows them with the status of (even) prima facie justifiedness? Feldman (2003) says that a person is (prima facie) justified in believing “There is a table before me” if the belief is a “proper response” to the person’s visual experience. When is such a response “proper”? According to Feldman’s internalist evidentialism, a doxastic attitude is “proper” just in case it fits the visual experience. But what does fittingness consist in, especially for vision-based attitudes? Consider an unusual proposition one might believe in the same visual circumstances, i.e., “There is a 12-yearold table before me.” Assuming no prior information about the table’s age and no visible date on the table’s surface (etc.), believing that proposition would not qualify, presumably, as a proper doxastic response. Why not? What’s the relevant difference between the two cases? No clear answer emerges from internalist evidentialism.
Process reliabilism offers a more informative diagnosis. The difference it pinpoints between the two cases resides in the fact that no reliable beliefforming process possessed by the subject could be applied to the visual experience to yield a belief in the proposition “There is a 12-year-old table before me.” By contrast, there are features of the imagined visual experience to which a reliable belief-forming process could be applied that would yield belief in the proposition “There is a table before me.” Thus, process reliabilism offers a clear, intuitively compelling explanation of the difference between the two candidates for basic, or immediate, justification. It smoothly handles a tricky puzzle case that outruns the resources of internalist evidentialism, which is forced to offer an unilluminating phrase, “proper response to experience,” containing a mere synonym (“proper”) of the explicandum (“fitting”).7
Let us expand on this theme with material that goes beyond essay 2. Does evidentialism’s tool-kit have the right conceptual tools to mark all the differences between justified and unjustified belief? The key theoretical concept in evidentialism that corresponds to the reliable-process concept in reliabilism is the basing relation concept. Details aside, evidentialists say that a belief B enjoys doxastic justifiedness if and only if (1) B fits the evidence, E, that the subject possesses at the time, and (2) B is based on E. Does this use of the basing relation cover all the necessary ground? I argue in the negative with the help of an example.
Chad justifiably believes two propositions of the form “P or Q” and “not-Q.” So he has first-rate evidence for believing P. Suppose now that he bases a belief in P on his prior belief(s) in these propositions. Is his belief in P doxastically justified? Not necessarily. Suppose Chad has a psychological propensity to reason in the following way: when he believes two premises of the form “P % Q” and “not-Q,” where the symbol “%” can be replaced by any binary truthfunctional connective (disjunction, conjunction, material conditional, etc.), then he infers a conclusion of the form “P.” In other words, he uses a process that “overgeneralizes” disjunctive syllogism. If this is the process whereby he arrives at belief in “P,” then, intuitively, the belief is not doxastically justified. Nonetheless, it stands in the basing relation to perfectly good evidence.8 Evidentialism therefore decrees that the belief is justified; but it isn’t. Substituting a reliable-process test, on the other hand, yields the right result. A disjunctivesyllogism reasoning process will pass the reliability test, and beliefs that result from it will be correctly classified as justified. The overgeneralized reasoning process, by contrast, fails the reliability test, and its output beliefs will be correctly classified as unjustified.
Essay 3 of the volume is an entry on reliabilism from the Stanford Encyclopedia of Philosophy. It pinpoints important objections and counterexamples to reliabilism and surveys existing responses. These include responses to the demon-world problem—a problem for the necessity of reliability—and responses to the reliable clairvoyant problem, a problem for the sufficiency of reliability. These problems continue to stimulate debate over reliabilism.9
Essay 3 also examines the numerous forms of contemporary epistemology that share core features of reliabilism, even ones that fly under different banners and terminologies. The core elements of these theories are saliently reliabilist, although they may use different terminology and propose a variety of “enhancements.” From reliabilism’s perspective, these theories are readily viewed as close cousins.
A clear example is the “tracking” theory of Nozick (1981). In its basic form, the tracking theory is a counterfactual reliability theory of knowledge without any element of causal processes. In its strengthened form, however, a sameness-of-methods provision is added to handle certain counterexamples. Another theory with a strong reliabilist underpinning is Plantinga’s (1993)
“proper functionalism.” Plantinga acknowledges that there is an important truth in reliabilist accounts of warrant, namely, the condition that the objective probability of the belief’s being true must be high. His crucial addition pertains to the existence of a “design plan” for the cognitive faculties by which the belief is produced. Such an enhancement does not appeal to most reliabilists, although an interpretation of teleology in evolutionary terms (not Plantinga’s preferred interpretation) contributes to its appeal for some (e.g., Lyons, 2009).
A third theory with ostensibly close resemblances to reliabilism is the version of virtue epistemology sometimes called “virtue reliabilism” (Greco, 2006). Sosa’s (2007) version of virtue epistemology is a leading example of this type of theory. Although there are clear distinctions in terminology, the theory appears to have substantial commonalities with process reliabilism, depending on one’s emphasis. The theory seems pretty well aimed to the extent that it approximates process reliabilism, but less likely to hit its target to the extent that it diverges from the latter. So, at least, I shall argue.
Theorizing specifically about knowledge, Sosa proposes the conditions of accuracy, adroitness, and aptness. Accuracy consists in truth: a belief must be true to qualify as knowledge. Adroitness is explained as skill or competence aimed at truth. A competence is explained as a disposition that tends to be manifested in successful performance. Since the aim of epistemic adroitness is truth, an associated competence is a tendency to get the truth (or at least avoid error). Obviously, this is reliability. Aptness consists of a belief’s being accurate because of skill or competence, that is, being accurate because of reliability rather than luck. Exactly what work is envisaged for the aptness condition is not wholly clear. If a cognitive competence is something like a reliable beliefforming process, there is nothing mysterious about competences causing beliefs. But most reliabilists think that reliably caused true belief doesn’t suffice for knowledge. A further condition is needed to exclude fake barn cases and other types of Gettier cases. Sosa seems to think that the aptness condition covers not only the process-reliability condition but also the sort of modal condition (e.g., sensitivity, safety, or no relevant alternatives) intended to handle Gettier and fake barn cases as well. It’s hard to see how that is supposed to work, however.
Thus far, the virtue theoretic (VE) approach displays substantial affinities to reliabilism. But other interpretations or points of emphasis might distance it from reliabilism, especially process reliabilism. Sosa (2010) specifies three parameters, or components, of a cognitive competence: (i) its categorical “constitution,” “basis,” or “seat”; (ii) a suitable “condition” of the constitution; and (iii) an appropriate “situation” it is in. A “complete” competence is specified by all three factors. When a competence is triggered by a suitable event, and the condition and situation are suitable, it gives rise to a manifestation. In the cognitive case, this would be a belief.
This account seems to steer the portrait of competences away from anything pertaining to psychological processes or structures (e.g., modules). Insisting on a dispositionalist specification of competence might be designed, in part, to keep epistemic theory and normativity at arm’s length from anything “naturalistic,” like brain regions, mental computations, or cognitive operations. Whether this preference flows from metaphysical theorizing (e.g., the non-reductive character of dispositions) or from methodological scruples of some sort is unclear. But this character of the theory suggests a deliberate “remove” from the spirit of process reliabilism.10
I argue, however, that to the extent that VE distances itself from processes, to that extent it will miss its mark. Without a constraint on processes, it is threatened with excessive laxness. This threat may be illustrated as follows. Arthur has an algebra competence—a disposition, when algebra problems are posed to him, of answering correctly (and believing the answers). In other words, he is disposed to be reliable in answering algebra problems. What are the parameters of the disposition? The basis, seat, or constitution of the competence is presumably his brain.11 The condition associated with the competence is being awake, healthy, and sober (non-inebriated), and the “situation” associated with it is having adequate air and light. Given these parameter settings (all of the parameters VE specifies), Arthur’s algebraic competence is triggered with a new algebra problem. He is shown two equations and asked to solve for X. He proceeds to manifest his competence by forming a belief in a true answer. Does his belief qualify as knowledge? VE, it seems, must answer in the affirmative. Arthur exemplifies the relevant competence conditional on this occasion by exemplifying both the appropriate conjunction of antecedents and the consequent. Moreover, his exemplification of the consequent—his arriving at the true belief—is caused by his exemplification of the conjunction of antecedents (the constitution, condition, and situation provisions). So the resulting belief satisfies the “triple-A” test (accuracy, adroitness, aptness). But does this guarantee “knowledge”?
No. Much depends on the specific process he uses in attacking the algebra problem. Suppose Arthur is lazy today. Instead of applying canonical algebraic procedures to solve the two equations for X, he takes a shortcut. He counts the number of characters in the two equations and determines that there are twelve of them. Then he concludes (bizarrely) that the solution to the problem is “X = 12.” As it happens, “X = 12” is the correct answer. Does Arthur know it? No.
The moral of the Arthur story parallels the moral of Chad, the overgeneralizing disjunctive-syllogism reasoner introduced previously. Unless an analysis of knowledge or justification includes a reliable process requirement (or conditions that entail such a requirement), it will be insufficiently demanding. Manifesting a reliable disposition—a condition Arthur meets—is no guarantee of employing a reliable process.
Various responses by proponents of VE are potentially available. They might try to show that, despite appearances, the dispositional analysis already deals with the Arthur case by implicitly excluding unreliable processes of the sort Arthur uses. I don’t see how this would work, but it’s conceivable. It would amount to a tacit admission, however, that using a reliable causal process is critical, precisely what process reliabilism contends. Alternatively, proponents of VE might concede the necessity of adding an explicit provision that requires a reliable process. It might run roughly as follows: “S knows that p only if S forms a belief in p via a reliable process that manifests an epistemic competence (disposition).”12 Of course, this would also concede the central tenet of process reliabilism. If the second tack is taken, moreover, process reliabilists might reasonably plump for a stronger claim. Not only is a reliable process constraint required, but incorporation of process reliabilist conditions would render the original appeal to dispositions unnecessary, or otiose. Of course, this is what process reliabilism has implied right along.
There are different problems, from my perspective, with other kinds of virtue theories. The more they distance themselves from reliability theory by adding virtue-distinctive enhancements, the less persuasive they become (by my lights). Ingredients are often added that comport with Aristotle’s theory of the moral virtues but strike me as excessive or inflated when injected into the epistemic domain. For example, Greco’s (2000) version of virtue theory requires knowledge to be stable rather than fleeting, in conformity with Aristotle’s notion that virtuous character is stable. But does knowledge require stability? No. Pieces of knowledge can rapidly disappear from the mind either because they never get stored in long-term memory, because they are rapidly forgotten, or because of degenerative disease. These sequelae don’t imply that the agent never knew. Another popular enhancement is credit-worthiness: a belief allegedly qualifies as knowledge only if it is “creditable” to the agent. But credit-worthiness is an inappropriate demand for either epistemic justification or knowledge, as argued persuasively by Lackey (2009) with her “Chicago visitor” example.
As we have seen, most avowed reliabilists also propose enhancements to the basic reliabilist schema (viz., “S’s belief that p is caused by a reliable process”). It may therefore be difficult to decide how to label epistemologists who embrace basic elements of reliabilism, add some preferred supplements, but don’t call themselves “reliabilists.” Even when they don’t self-apply the label, they may differ little from self-labeling reliabilists, and it may be reasonable to apply the label to them, especially in a broad survey of the epistemological landscape.
A final observation on reliabilism’s impact on contemporary epistemology is to note how some of the most interesting criticisms of reliabilism lead to reflections that reinforce reliabilism—or reveal that the problems initially pinned on reliabilism afflict other theories equally. A case in point is the problem
of bootstrapping, or easy knowledge. Vogel (2000) initially advertised this problem as peculiar to reliabilism. Cohen (2002) refuted this claim, however, by showing that, far from being unique to reliabilism, the easy knowledge problem arises for any epistemology that rejects a condition like KR.
KR: A potential knowledge source K can yield knowledge for S only if S knows that K is reliable.
Epistemologies that reject KR have what Cohen labels “basic-knowledge structure.” As he indicates, something like KR seems necessary to block the attainment of “easy” knowledge. Avoiding easy knowledge, however, seems to drive one down the road to skepticism. This is persuasively argued by van Cleve (2003). Moreover, even if one adopts KR in order to preclude easy knowledge, a resulting theory will retain a significant strand of reliabilism. Why? Because KR says that first-order knowledge always requires knowledge of source reliability. But if knowledge of source reliability is required, then—by the factivity of knowledge—de facto reliability of one’s source is required. And this is precisely the core (or an essential part of the core) of reliabilism.
Cluster 2: Essays 4 and 5
We turn next to the second cluster of essays in the volume, a pair of articles that extend the debate between externalist reliabilism and internalist evidentialism while also exploring prospects for a limited rapprochement between them. Essay 4, “Internalism, Externalism, and the Architecture of Justification,” explores the debate between internalism and externalism from the ground up. It begins by questioning the standard terms of engagement in the justifiers debate. “Justifiers” is the widely accepted label for properties, conditions, or states of affairs that affect a belief’s J-status positively or negatively. Internalism customarily maintains that all justifiers are internal, and externalism denies this. Under this mode of engagement, the playing field clearly favors externalism (although internalists have not been heard to complain). Essay 4, however, offers to level the playing field. Instead of allowing externalism to win the debate if it finds even a single external type of justifier, the new terms of engagement allow a side to enjoy victory laurels only if it shows that more kinds of justifiers are on its side of the ledger than on the opposite side. This is a generous offer, especially coming from an externalist. Nonetheless, I still contend that externalism wins the battle, even after cheerfully conceding that some (a few) kinds of justifiers are internal. Features of perceptual and memorial experiences occurring at the time of belief do count (subtleties aside) as internal justifiers. Nonetheless, a greater number of justifiers are external rather than internal.
One driving force in generating this upshot is the need for a historical approach to justifiedness. According to a historical (i.e., personal-historical)
approach, past mental states of the subject comprise a crucial class of justifiers, all of which are external in the traditional sense. They are external because, traditionally, states that justify a belief are restricted to those that obtain at the time of belief. This is an example of the sorts of factors that must be included in a comprehensive reckoning of justifier types.
The number of justifier types readily expands, moreover, when viewed through the lens of a right-rule architecture. Within such an architecture, an agent’s doxastic attitudes qualify as justified just in case they conform to what right rules permit, given her evidential situation and the belief-forming processes used. This analytical device provides a natural framework in which to reflect on epistemic normativity. It provides a neutral setting in which the internalism/externalism controversy can be fruitfully examined. It also places at center stage the question of what makes J-rules right or wrong; in other words: what is the criterion, or standard, of J-rule rightness? One of the many considerations supporting externalism is that a plausible rightness criterion must consider true belief acquisition and/or false belief avoidance, both externalist factors. However, most of the core arguments for externalist J-factors depend on factors quite different from the criterion of rightness. Here are two examples. When considering which inferential beliefs are justified, the strength of the (logical, inductive, or probabilistic) support relation between the premises and the conclusion is important. But the amount of such strength is an objective state of affairs, not a mental state of the agent. Internalists sometimes try to twist the consideration in their direction by “internalizing” or mentalizing the support relation. But any such attempt, I argue, will either be insufficiently or excessively demanding, or both. Another consideration arises from the widely recognized fact that inferred beliefs are justified only if the prior beliefs from which they are inferred are also justified. But how did those prior beliefs attain their justifiedness? In general, their justifiedness arises via events that transpired in the agent’s cognitive life prior to the current acquisition of the new belief. This fact leads inexorably to the historicity of justifiedness, a pervasive feature of justification at odds with traditional internalism.
Since externalism doesn’t deny the existence of some internal justifiers, readers should not be astonished to find essay 5 proposing a partial synthesis, or rapprochement, between reliabilism and evidentialism. (Indeed, I would not be surprised if such syntheses become the wave of the future.) At least one other recent article (Comesana, 2010) embraces this approach. Admittedly, acceptance of such a synthesis by evidentialists does not seem to be in the cards. The volume in which essay 5 (first) appears contains a response by Feldman and Conee, who, predictably, disclaim any need for evidentialism to seek help from reliabilism. Of course, one doesn’t expect senior people to change their minds about theses they have long defended. But a younger generation might appreciate the attractions of the synthesis proposed here.
So, let me restate a few of the challenges to the viability of (pure) internalist evidentialism posed in essay 5. First, evidentialists have thus far declined to provide any definition of “evidence,” a task they can hardly shirk. This task is pressing, given the centrality of the evidence concept (some evidence concept) in their account of justification. But they have not clarified what they mean by “evidence.”13 Essay 5 offers two types of definition that evidence theorists might find congenial but that (mentalist) evidentialists cannot adopt. One definition is in terms of reliable indicatorship; the other is in terms strengthening justification. The first option is unavailable to internalists because they want no truck with any truth-linked concept like reliable indicatorship. Nor can they accept a definition in terms of justification-enhancement or warrant-enhancement, which would create an unsavory circularity in their analysis of justification.
The reply to these contentions by Feldman and Conee is perplexing. They quote several dictionary definitions of “evidence” and find none that corresponds to the options I offer. In particular, they find none that corresponds to the “reliable mark of the truth” definition I consider. They oddly neglect the fact that I also consider a “justification-strengthening” definition and argue that this would pose a serious circularity problem for their evidential analysis of justification. Their published response ignores this argument. As it turns out, most of the dictionary definitions they adduce are readily interpretable as versions of this second definitional approach. Here are a few of them: “an appearance from which inferences may be drawn”; “. . . a fact that gives proof or a reason for believing” (emphasis added); “a thing or things helpful in forming a conclusion or judgment”; and “ground for belief or disbelief” (emphasis added). If these definitions are the kinds Feldman and Conee favor, they will pose exactly the circularity problem I introduced.
They seem to give favorable attention to one truth-linked approach. Considering the definition quoted from Kelly (2008), construing evidence as “a guide to truth,” they take no exception to this definition; they only take exception to my paraphrase that adds the term “reliable,” i.e., “a reliable sign, symptom, or mark of the truth.” Their complaint is that “reliability” is absent from the original and (they imply) is an unwarranted elaboration. This is extremely odd. What does it mean to say of a fact that it’s a reliable indicator of the truth? Presumably, it means that the fact in question correctly points to the target truth in question, perhaps because it instantiates a type that is reliably correlated with another type that the target truth instantiates. For example, the fact that the cross-section of a tree-trunk has 250 rings is well-correlated with the fact (the truth) that the tree is 250 years old. This seems hardly different from Kelly’s “a guide to truth” (a definition that elicits no complaint from Feldman and Conee). But truth, as much as reliability, is one of the external factors that Feldman and Conee need to exclude from their definition of evidence. How can they comfortably accede to Kelly’s definition but complain
about my elaboration? Either definition is externalist, which means that any account of justification in terms of this notion of evidence would become externalist.
Perhaps Feldman and Conee think that “guide to truth” means something like “warrant, or justification, for believing something to be true.” However, then we are back to the family of definitions considered in the preceding paragraph, which simply reintroduces the other horn of the definitional dilemma I presented.
Ultimately, Feldman and Conee don’t endorse any unambiguous definition of “evidence.” The meaning they attach to this key term remains a mystery. I reiterate the point that until a suitable definition is provided, we cannot assess their evidentialist theory. Its meaning remains obscure, and its claim to offer a non-externalist conceptual basis is precarious.14
An equally serious lacuna in the conceptual foundations of their approach is the notion of “fittingness.” However, since my “synthesis” floated in essay 5 provisionally buys in on experiential and inferential (doxastic) fittingness, I will not press this point. It would have been interesting, however, to learn more of what Feldman and Conee think about the proposed synthesis. Do they deny the intuitive plausibility of distinguishing the two dimensions of justifiedness highlighted by the Shirley-Madeleine example? Even if they have no opinion about its intuitive plausibility, or how I elucidate it, other epistemologists can be expected to weigh in on the matter.
Cluster 3: Essays 6 and 7
The next two essays (6 and 7) focus on issues related to reliabilism as a theory of knowledge or evidence rather than justification. Some critics of knowledge reliabilism claim that it fails a crucial test that confronts all putative accounts of knowledge, viz., to explain why knowledge has more value (indeed, more epistemic value) than mere true belief. Reliabilism’s alleged inability to explain this excess value is advanced by several epistemologists, especially proponents of virtue epistemology such as Swinburne, Zagzebski, and Kvanvig. As observed near the outset of essay 6, evidence in support of the surplus-value-ofknowledge principle is not really thick on the ground. Nonetheless, it is conceded for the sake of argument.
Reliabilism’s alleged deficiency vis-à-vis the value challenge is often presented in terms of the “swamping problem” (Swinburne, 1999; Kvanvig, 2003). The critics argue that reliabilism can only account for the extra value of knowing if the extra value resides in the reliable process that produces a knowledgequalifying belief. But how does the generating process acquire the extra value? Swinburne puts the point this way:
[Clearly] it is a good thing that our beliefs satisfy the reliabilist requirement, for the fact that they do means that . . . they will probably be true. But, if a given belief of mine is true, I cannot see that it is any more worth having for satisfying the reliabilist requirement. So long as the belief is true, the fact that the process which produced it usually produces true beliefs does not seem to make that belief any more worth having. (1999: 58)
The idea, it would seem, is that the process’s value is wholly acquired from the true belief that it causes, which confers on it instrumental epistemic value. This instrumental value, however, would be “swamped” by the value of the true belief from which it is derived. Zagzebski (2003) makes the same point with the help of an espresso analogy. A good cup of espresso has value, but it acquires no additional value when produced by a reliable espresso machine. A good cup of espresso is just as good when produced by an unreliable machine as by a reliable one.
Essay 6, co-authored with Erik Olsson, offers two solutions to the valueof-knowledge problem. The first of these, the conditional probability solution, was devised and defended by Olsson. The second solution, labeled the “value autonomization” solution, is one I devised and defended. I now regret the assignment of central importance to the “autonomization” idea. However, another element, also featured in this solution, continues to strike me as important, and I’ll argue here that it goes a long way toward solving the challenge to reliabilism.
Before revamping this second solution, let me emphasize the point that the so-called swamping problem, if it’s a genuine problem at all, is equally problematic for many theories of knowledge, not just reliabilism. It is even a problem for many internalist theories, as Pritchard (2011: 10) points out. For example, BonJour’s (1985) defense of his coherentist theory of justification indicates clearly that the value of coherence is purely instrumental to the goal of true belief. Being a justificationist about knowledge, BonJour would presumably account for the extra value of knowledge in terms of the belief’s coherence with the rest of his belief system. If this is all correct, then BonJour’s coherentist theory equally faces the problem of explaining how extra value can arise from justification in virtue of this instrumental character vis-à-vis true belief. In short, if the swamping problem has any genuine bite, it must have equal bite for many different kinds of theories, not just causalist/externalist theories like process reliabilism.
There are two keys to my solution as I would now develop it. The first key—already contained in the original paper—is type instrumentalism. In launching the value-based critique of reliabilism, critics probably assume that, under reliabilism, the “extra” value that accrues to a true belief produced by a reliable process derives from that token reliable process. Moreover, they
probably assume that this token process acquires its value by standing in a singular causal relation to the true belief in question. This is what I would call token instrumentalism. The puzzle about token instrumentalism is how the process token can “return” any of the value it inherits from the true belief it causes so that the latter’s total value exceeds the value it would have were it instead caused by an unreliable process.
However, a token reliable process might acquire “derivative” epistemic value in an entirely different way without encountering the same problem. It is widely assumed in moral theory that something’s being a token of a valuable type automatically confers value (at least prima facie value) on that token. For example, an action’s being a case of promise keeping automatically confers (prima facie) value on that action token. Similarly, if we assume that being reliable makes some type of belief-forming process epistemically valuable (for example, taking visual appearances at face value), then any token of such a belief-forming type would also have epistemic value. Thus, to qualify as epistemically valuable, a token belief-forming process would not have to succeed in generating any true belief. It would not have to stand in a singular causal relation to a true belief. Value might accrue to it in a different, two-step fashion. In step 1, the type acquires value because many instances of it produce token true beliefs. In step 2, a token of the process type acquires epistemic value because it instantiates that valuable type, whether or not it produces a true belief of its own. If all this is correct, then it’s false that the only way for a belief-forming process-token to acquire (instrumental) epistemic value is for it to cause an epistemically valuable upshot, for example, a true belief.
This point helps position reliabilism for an escape from the swamping problem. If the source of a process token’s value needn’t be a true belief that it produces, then there’s no guarantee that the value of the resulting true belief must swamp, or trump, that of the token. There remains, however, a nagging problem. How can the value of the token increase the value of the true belief?
Value theory routinely assumes that value transmission runs “from” achieved goals, outcomes, or ends “to” the means that produce them. How could it countenance the idea that value transmission runs in the opposite direction, viz., from means to ends? That is what would be required if a (valuable) reliable process is to increase the value of a true belief by virtue of causing it. Is this scenario the slightest bit plausible?
I think it is. Here is an example to illustrate the indicated pattern of transmission. Any work by Rembrandt, whatever its intrinsic aesthetic quality, will be assigned a high degree of value simply because it was produced by Rembrandt. The fact that Rembrandt painted it is enough to confer significant value on it, as its price in the art market would reflect. The extra value assigned to a genuine Rembrandt need not be merely a recognition of market value. Instead, it may be a recognition of what explains that market value, i.e., the fact that people admire or esteem such a work because it’s the product of a masterful hand (and eye).
Here, then, is an instance of the pattern in question. A highly prized process of generation—being painted by Rembrandt himself—elicits enhanced admiration or esteem for the product. If this transpires in the case of Rembrandt’s paintings (or Robert Frost’s poems), why can’t it transpire in the case of reliably produced true beliefs? In the latter case, a true belief’s “total” (epistemic) value could come from two independent sources: first, the “final” value of the true belief itself, and second, the instrumental value of a token process that produces it. Since these could well be independent sources, the reliably produced true belief could easily contain more value than a similar true belief produced by an unreliable process. Thus, reliabilism easily meets the challenge of the swamping problem.
I now digress a bit by addressing another challenge to reliabilism based on a theory of value, though this problem makes no appearance in any of the papers of the present volume. It is fashionable nowadays to seek ways to bridge or unify the normative fields of ethics and epistemology. One proposed unification compares epistemic teleology (or consequentialism) to moral teleology (or consequentialism). Reliabilism is often said to be a consequentialist approach. It posits, as core epistemic values (final values), states of true belief and error avoidance. Justifiedness of belief is some sort of deontic status, such as being permitted, or entitled, to believe. This kind of deontic status is ultimately rationalized, or derived, from the conduciveness of a belief—or, rather, conduciveness of the belief-forming process—to the production of good epistemic consequences, in the form of true beliefs or the absence of false beliefs. Viewing reliabilism in this light is encouraged by formulations in terms of a right-rule architecture, as is offered in essay 4 (and in Epistemology and Cognition). As we shall now see, however, this teleological or consequentialist rendering of reliabilism is used by some critics as a basis for putative counterexamples to it.
Here is an example found, in various permutations, in Firth (1981), Fumerton (2001), Berker (unpublished), and other critics of reliabilism. A scientist seeks a grant from a religious organization. Although she is an atheist, she realizes that her only chance of getting funding from the organization is to form a belief in God’s existence. She knows that if she received the grant, she would use it to further her research, and this would allow her to form many new true beliefs and revise many previously held false beliefs on matters of intellectual significance. Would she be justified, in these circumstances, in forming a belief that God exists? Reliabilism, according to these interpreters, implies that she would be so justified, because so many true beliefs would result, along with the abandonment of many false beliefs. All of these consequences would yield great epistemic value, so epistemic teleology would certainly permit it to be done and perhaps even require it. Reliabilism must deem it to be justified. Intuitively, however, it isn’t at all justified.
This, however, is an inaccurate portrait of reliabilism and its implications, at least on my rendering of reliabilism. As I have presented it, process reliabilism
(PR) does not warrant the specified conclusion in the scientist case. There are three ways the alleged interpretation fails to comport with PR. First, the scientist example proceeds on the assumption that reliabilism is a value-maximizing kind of theory. But PR does not equate justified belief with belief that maximizes (final) epistemic value. This was clear in Epistemology and Cognition, where right J-rules—conformity with which yields justifiedness—are ones that authorize processes with a high truth ratio, not necessarily a high quantity of true belief (Goldman, 1986: 106). The high truth-ratios standard gives more weight to error avoidance than to truth acquisition. Belief-forming processes can earn high truth ratios by forming beliefs very cautiously and carefully— generating very few beliefs, hence rarely falling into error, but also rarely generating truths. Thus, unlike the most famous forms of moral consequentialism, e.g., utilitarianism, it is not a value-maximizing theory.
Second, one could only impute to reliabilism the implication that the scientist’s belief is justified in the specified example by ignoring the process component of reliabilism. PR does not determine a belief’s justificational status simply by its direct or indirect causal consequences (actual or hypothetical). In fact, it doesn’t look at the consequences of the target belief at all. Instead it looks—most immediately—at the process (or processes) of which the target belief is an effect. If the process that causes the belief is a token of a type such that its (the type’s) belief-upshots have a high truth ratio, then the belief is justified. If not, not. What is the belief-forming process in the scientist case? Since the presented case is one of propositional justification, the case is one where the agent has not (yet) formed a belief. Therefore, some assumption must be made about how—i.e., by what psychological route—she would form the belief if she adopted it. What might such a psychological route be? Presumably, something like “being driven to believe a proposition by irrelevant goals or desires”—irrelevant, that is, to the truth of the believed proposition. No such candidate process would have a very high truth ratio. So PR implies that the scientist would be unjustified in forming a belief in God’s existence. That such a belief would nonetheless have favorable downstream epistemic consequences has no bearing, under PR, on its justificational status.15 Only upstream cognitive activity—i.e., psychological activity actually or potentially involved in the target belief’s production—is relevant to that belief’s justificational status. Notice that the anticipated downstream consequences (true beliefs in scientific propositions) would presumably be formed by very different belief-forming processes than the one that would generate a belief in God. So, even if those subsequent, science-generated beliefs would themselves be justified, this in no way suggests that the scientist’s belief in God (were she to form this belief in the manner specified) would be justified.
Berker stresses that justificational assessments of belief must observe the principle of “the epistemic separateness of propositions.” This principle states that when determining the epistemic status of a belief in a given
proposition, “it is epistemically irrelevant whether or not that belief conduces (either directly or indirectly) toward the promotion of true belief and the avoidance of false belief in other propositions beyond the one in question” (unpublished, p. 29). Moreover, he contends that veritistic epistemic teleology flouts this principle and that reliabilism is a species of veritistic epistemic teleology. These contentions are mistaken. As explained previously, PR does not flout the principle, because it is not really a species of veritistic epistemic teleology.
I turn next to essay 7, a critical examination of Williamson’s (2000) theories of knowledge and evidence. Because Williamson’s global view is reliabilist through and through, large parts of it are very congenial, at least in general contour. Nonetheless, I argue that certain details of the theory are less successful than those of broadly similar theories I have defended, for example, in the specific modal requirements appropriate to knowledge attainment. I also suggest that Williamson misses the boat by downplaying the process element in PR. In response, Williamson (2009: 307) has replied that he did not intend to distance himself from the process element. If so, all well and good, but this fact was not readily detectable.
The final two sections of essay 7 address the topic of evidence. Williamson holds that a person’s body of evidence is all and only what she knows (E = K). A battery of intriguing and powerful arguments were offered in Knowledge and Its Limits to show that no rival view could account for the intuitive data better than (the quite surprising and initially counterintuitive) E = K thesis. I respond by arguing that at least one rival view performs at least as well as E = K in accounting for Williamson’s “data,” namely, the view that evidence is non-inferential propositional justification (E = NPJ). This dispute is relevant to Williamson’s knowledge-first treatment of epistemology, of which I am not yet persuaded. In his response, Williamson (2009) repeats his claim that evidence is factive, and that this feature favors a knowledge account as opposed to the E = NPJ view. He also recognizes the possibility of strengthening E = NPJ by adding the factivity component. He doesn’t say much to favor his preferred view over this alternative. He only remarks that E = TNPJ is a “rather unnatural hybrid” (2009: 311). He doesn’t spell out, however, what makes this particular hybrid “unnatural” or what makes E = K natural.16
Cluster 4: Essays 8, 9, and 10
The three essays in this cluster belong in the category of social epistemology, a subfield of epistemology to which I have recently devoted at lot of attention (more than the present volume indicates). Essay 8 addresses two rather distinct topics in social epistemology, peer disagreement and epistemological relativism. Here I confine my summary remarks to relativism.
One way to characterize epistemological relativism is to treat it as the conjunction of three theses:17 (1) Different communities often have differences in the epistemic rule systems they endorse. (2) There are no objective facts that make one such system right and incompatible systems wrong. (3) There is no such thing as universal epistemic rightness; epistemic rightness is properly understood as relativized either to communities or to epistemic agents. For example, an epistemic system Σ is right relative to community C if and only if C endorses Σ. The first of these theses would be widely accepted, by almost everybody. But a majority of mainstream epistemologists, I conjecture, would reject the second thesis, and the third thesis would be dramatically unpopular among them. Nonetheless, all three theses find considerable support outside of philosophy, and at least the first two are warmly received by an important subset of philosophers of science. Historically oriented philosophers of science, in particular, are uncomfortable with the notion that a unique set of norms is properly associated with “science.” Scientific norms and methods change over time, so that what is accepted in one historical period (or by one scientific specialty) is superseded, amended, or never adopted in others. Many philosophers of science would find it naïve to claim that there is a uniquely correct system of scientific rules (in Plato’s heaven, for example). Bayesian philosophers of science, of course, won’t share this stance.
Confronting a possibly irresoluble divide between absolutism and relativism, as formulated here, essay 8 proposes a novel version of relativism that might be palatable even to staunch objectivists (advocates of uniqueness) about epistemic normativity. The proposed conception of relativism might accommodate both perspectives simultaneously. It does not propose to relativize all epistemic normativity, but it leaves space (of sorts) for such relativization.
The relativization proposal floated in essay 8 is spurred by the notion of iterative justification. In the case of first-order justification, a belief or other doxastic attitude qualifies as justified because right epistemic rules imply permissibility of the attitude, given the agent’s epistemic situation or circumstances. Things are more complicated, however, when we turn to higher-order justification. Even if there is a uniquely correct system, Σ1, what obtains if and when a subject justifiedly believes (the falsehood) that Σ2 is the right system instead (where Σ2 clashes with Σ1 on important points)? A subject might easily be justified in believing a wrong system to be right if, for example, members of her epistemic community go around asserting that Σ2 is the right system (especially in pedagogical contexts). Assume that the objectively correct system, Σ1, contains a rule of testimonial acceptance that isn’t unduly demanding. Then multiple testimonies by a subject’s parents, teachers, and/or neighbors could easily justify her in believing that Σ2 is the right system.18 By contrast, people living in other epistemic communities, where nobody goes around promoting Σ 2 as right, won’t have the same justification—or any justification—vis-à-vis
the rightness of Σ2. So the community in which an agent is embedded can make a big difference with respect to which epistemic system she is justified in believing to be right.
Can this make a difference to the justificational status of ordinary, nonepistemic propositions (as opposed to propositions about system-rightness)? Yes; at any rate it can make a difference to the second-order justificational status of such propositions. Let Fiona be in epistemic circumstances R with respect to proposition P. And suppose that the right system, Σ1, implies that Fiona, being in R, is unjustified in believing P. However, system Σ2 implies that Fiona in R is justified in believing P. Moreover, Fiona is justified in believing that Σ2 is the right system. Then Fiona has second-order justification for believing P: she is justified in believing that she is justified in believing P (JJp). Now consider epistemic agent Fabrizio, who is also in R with respect to P but is not justified in believing that system Σ2 is right (nor justified in believing any other system to be right that licenses belief in P when in R). Then Fabrizio does not have second-order justification with respect to believing proposition P, although he is in the same (first-order) epistemic circumstances with respect to P as is Fiona.19 This is the sense in which justification—at least second-order justification—may be “relative” to one’s community. This kind of relativity holds, however, only at the meta-level (as far as our present assumptions are concerned). We are assuming that there is such a thing as objective first-order justification, which is fixed by a uniquely correct epistemic system. Thus, there are two species of justifiedness (first-order and higher-order), one that fixes justifiedness by reference to objective standards and others that fix justifiedness by reference to factors that can be heavily influenced by one’s embedding community. These two species of justifiedness can co-exist, a possibility not previously noticed in the literature (as far as I know).
The remainder of essay 8 makes provisional suggestions for applying these ideas to the problem of peer disagreement. Here, however, my focus has been on its novel way of interpreting epistemological relativism. My proffered interpretation, of course, is far weaker than traditional epistemological relativisms. But it is interesting to explore diverse approaches to the subject.
Essays 9 and 10 provide overviews and taxonomies of social epistemology.20 Each is perhaps best viewed from the perspective of my book Knowledge in a Social World (Goldman, 1999). That book offered a systematic treatment of social epistemology in terms of a veritistic framework, a framework that generalizes reliabilism by describing valued epistemic outcomes in terms of attitude/truth-value pairs. It presented social epistemology as a field devoted to the veritistic assessment of a wide range of social practices and institutions.21 But veritism is just one of many possible frameworks, criteria, or measures of epistemic success. In the new framework for social epistemology of essays 9 and 10, a wider range of measures of epistemic appraisal is offered, not just externalist measures like truth-relatedness but any measures gener-
Another random document with no related content on Scribd:
Fig. 1.
FIG. II.
Fig. II illustrates the side view of the angle of 135 deg. as applied on the draft, and shows the reduction of the square in front and below the waist, as well as the addition behind and below the waist. The diagram on this figure is a simple vest, on the square of 17½, and hence the under arm cut is 3¼ at the circle.
DIA. I.
Dia. I is intended for illustration only. It is on a square of 20½, and the ½ is intended for the extra seam which the frock coat requires. The center of the back and center of the front run in the same direction as they must run when on the body. It also illustrates the spring over the hips as the body actually requires it.
The gore between the side-piece and back illustrates what may be buckled up on a vest. The front represents a straight single-breasted coat, as worn by the military or clergy, and may be depended upon if placed as in Dia. II. The front of such a coat must have a large gore in the center, which gore must be made at least 1¼ to 1½ inches, seams included, and must start well up above line 15, and the edges cut oval, not hollow; and the canvas and all padding and lining must be cut and worked in the same way, in which case the front requires no drawing in. The stay is put on close merely to keep the edge from stretching.
Above line fifteen the button-holes are cut, and the buttons are set plumb on the front line, but below line 15 the button line turns backward as shown. To meet the collar in front, nick one seam back of the front line, for military, but for a clergyman’s coat place the nick ¾ back of said line. The top of the skirt laps 1 inch in front of the forepart, and drops ½ inch on the bottom of the front. The standing collar must be a straight piece, and its width depends upon regulation. For a clergyman’s coat the standing collar should not be less than ¾ inches nor more than 1 inch, made up. The front of such a collar may be run straight upward with a line parallel to the front line of 135 deg., but may be shaved backward on top according to regulation or taste.
The place for the shoulder straps on military or society coats is the spot between the lines represented by 60 deg. from the front and 60 deg. from the back, which space of 7½ deg. represents the top of
the shoulder. For practical purposes, use the back part of Dia. II for such a coat, and shape the front like Dia. I.
This Dia. I is made to show all the angles which start from the point of 135 deg., and which are required for a coat or vest. That angle which forms the shoulder seam, and which is also marked square of 20½, is 45 deg. from the center of back, but for some reason it was forgotten to be so marked. The lines of this diagram may be used to draw the angles for any coat or vest, by simply making them long enough for the size required, the same as in Dia. XII-A. Drawn long enough to cover both shoulders, as shown in Fig. I, the outside lines of this angle of 135 deg. is to be used to measure the shoulders as well as to draw all the other lines, or angles for drafting.
A circle drawn from this point, with a half diameter of 8 inches, will be large enough, and it may be cut out of solid paste-board, but better, of tin or zinc on which the lines are correctly drawn. The location of the forepart and shoulders is the same as in Dia. II, or any other diagram. The back and the sidepiece on the square of 20½ are in a different position only. The front is just as good as any diagram can be made for a military or for a clergyman’s coat, when placed as in Dia. II. The back and the sidepiece is for illustration only.
That back which rests on the angle of 135 deg. is in all respects correct, but the back which rests on the sidepiece has one incorrect line, and this is the line forming the shoulder seam. It is correct at the armhole, but should run up to the point of the square of 20½, so that all three lines meet, and it will be found that from that point, down to line 9 over the front, will be 14⅜, but line 9 itself gives the correct distance from the top and back corner of the square.
By close observation it is also seen that the height of back above line 9 is only ⅜ more in Dia. I than it is in Dia. IV, or on the vest, and also that the side of the back of Dia. I is reduced again in length ½ inch by a gore from the armhole to the shoulder blade, all of which proves that both coat and vest are the same thing on the
same square, and that the heights of back and front are the same, and in order that the vest is covered by the coat, the neckhole for the vest is cut ½ lower.
If everything is considered at the waist, the coat is only a trifle larger than the vest, and that, what the front of a coat has more than a vest, is used for the lapel. Though Dia. I has a square of 20½, and Dia. IV has only a square of 20, both have the same width when all the seams are sewed up. All of which shows that, as the coat must pass over the vest, the vest must be cut at least one size smaller than the coat, and I have found it correct. In other words: The difference between the breast measure over the vest, and that under the vest, is about 1 inch in the whole breast measure. Dia. I gives a good deal of information, though it may be useless for practical cutting.
DIA. II.
This is a plain double-breasted frock; and the diagram is plain and requires little extra description, except the front of the waist and lapel. The fore-part has no gore at the waist, and the reduction is made in front. It will be seen that the double-breasted front of Dia. II is only about 1 inch wider than the single-breasted front of Dia. I, and that the gore in Dia. I is balanced in Dia. II by the large gore between the lapel and fore-part upward, and if it is desired to cut a small gore at the waist, then all what that gore takes up must be allowed in front.
The lapel in Dia. II will roll anywhere down to the second or third button, but if a gore is cut at the waist the roll will go down to the waist seam; and if the collar is pretty well rounded in front the lapel may be run down to the bottom. All frock coats have the center of the back pretty well thrown out at the bottom of the skirt, and this extra width is again reduced between the back and front skirt, which operation allows that seam to be oval-shaped in order to fit over the seat. The frock coat, being open behind, must receive its oval form over the seat in the seam between the back and front skirt, because that seam is sewed together and will hold its shape, whereas if the roundness were placed in the center behind it would simply produce a curved edge, but not an oval shape for the seat. The lap between the forepart and the skirt represents the extra length over the oval chest, and for a full chest, and if the chest is not very full, the lap of 1 number may be reduced to ½ at the front edge, but must remain 1 number at the plumb line base.
Dia. II is intended for the normal form, and for the following measure; Breast, 35; waist, 32; hip, 34; seat, 36; length of legs, 32 to 33; form straight without being over-erect. The back and the sidepiece is intended for a close fit, or as fine work ought to fit. If a looser fit is required, the gores between the back and the sidepiece
and between the fore-part and the sidepiece should be made a trifle smaller from the hollow of the armhole downward, which will give greater ease to the armhole and is better than to cut the armhole more forward. If a waist is prominent, say nearly as large as the breast, the underarm gore may be made as small as ⅛ to ¼ at line 17½ and run out to nothing at line 20; while a very large waist may require a lap of 1 inch, at the waist seam, starting said extra width at the bottom of the armhole.
This diagram shows the sleeve and the armhole different than Dia. VII. The armhole is cut out to the front sleeve base line and the sleeve and the armhole laps 2½ at 60 deg., all of which gives both armhole and sleeve a trifle larger and that armhole requires not so much stretching as that of Dia. VII. The top sleeve and armhole nicks are connected by a right angle from the center of back through the angle of 60 deg., and they will fit pretty close together, and in fact may be taken together on all sleeves and armholes, basting forward and backward to meet the other nicks.
NOTE: Elsewhere it is stated, that Dia. II must be considered the parent pattern of all others, which is quite true, though the armhole of Dia. II is larger than the others.
FURTHER: In cutting the gore between the back and sidepiece, point 5 must be quite slightly touched by both, and it may be better to have ⅛ gore there on all coats except for the stooping form. See Dia. II B.
