Epistemology A Contemporary Introduction 1st

Edition Alvin I Goldman download

https://ebookbell.com/product/epistemology-a-contemporary-
introduction-1st-edition-alvin-i-goldman-54220734

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Epistemology A Contemporary Introduction To The Theory Of Knowledge

3rd Robert Audi

https://ebookbell.com/product/epistemology-a-contemporary-
introduction-to-the-theory-of-knowledge-3rd-robert-audi-2534564

Epistemology A Contemporary Introduction To The Theory Of Knowledge

2nd Edition Robert Audi

https://ebookbell.com/product/epistemology-a-contemporary-
introduction-to-the-theory-of-knowledge-2nd-edition-robert-
audi-1494978

What Is Scientific Knowledge An Introduction To Contemporary

Epistemology Of Science Hardcover Kevin Mccain Kostas Kampourakis

https://ebookbell.com/product/what-is-scientific-knowledge-an-
introduction-to-contemporary-epistemology-of-science-hardcover-kevin-
mccain-kostas-kampourakis-10481158

Problems In Epistemology And Metaphysics An Introduction To

Contemporary Debates Steven B Cowan

https://ebookbell.com/product/problems-in-epistemology-and-
metaphysics-an-introduction-to-contemporary-debates-steven-b-
cowan-50218370
Epistemology Contemporary Readings 1st Edition Michael Huemer

https://ebookbell.com/product/epistemology-contemporary-readings-1st-
edition-michael-huemer-1276506

Contemporary Debates In Epistemology Contemporary Debates In

Philosophy 1st Edition Matthias Steup

https://ebookbell.com/product/contemporary-debates-in-epistemology-
contemporary-debates-in-philosophy-1st-edition-matthias-steup-1736886

Reliabilism And Contemporary Epistemology Essays 1st Edition Alvin I

Goldman

https://ebookbell.com/product/reliabilism-and-contemporary-
epistemology-essays-1st-edition-alvin-i-goldman-42854964

What The Ancients Offer To Contemporary Epistemology 1st Edition

Stephen Hetherington

https://ebookbell.com/product/what-the-ancients-offer-to-contemporary-
epistemology-1st-edition-stephen-hetherington-11004692

What To Believe Now Applying Epistemology To Contemporary Issues David

Coady

https://ebookbell.com/product/what-to-believe-now-applying-
epistemology-to-contemporary-issues-david-coady-5478262
 Ep i st e m o l o g y
A Contemporary Introduction

Alvin I. Goldman
and

Matthew McGrath

Ne w York   O x for d

Ox f or d U ni v e r s i t y Pr e s s

00-Goldman-FM.indd 1 16/10/14 12:49 AM

 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 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 © 2015 by Oxford University Press.

For titles covered by Section 112 of the US Higher Education

Opportunity Act, please visit www.oup.com/us/he for the
latest information about pricing and alternate formats.

Published in the United States of America by

Oxford University Press
198 Madison Avenue, New York, NY 10016
http://www.oup.com

Oxford is a registered trade mark 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-
Epistemology : a contemporary introduction / Alvin I. Goldman, Rutgers University
and Matthew McGrath, University of Missouri.
  pages cm
Includes index.
ISBN 978-0-19-998112-0
1. Knowledge, Theory of. I. McGrath, Matthew. II. Title.
BD161.G643 2014
121--dc23
2014015224

Printing number: 9 8 7 6 5 4 3 2 1

Printed in the United States of America

on acid-free paper

00-Goldman-FM.indd 2 16/10/14 12:49 AM

 Table of Contents

Epistemology: A Contemporary Introduction

Alvin I. Goldman and Matthew McGrath iv
Part I: JUSTIFICATION AND KNOWLEDGE:
THE CORE ISSUES  1
1. The Structure of Justification 3
2. Two Debates About Justification: Evidentialism vs.
Reliabilism and Internalism vs. Externalism 25
3. Defining Knowledge 51
4. Skepticism About Knowledge 81

Part II: JUSTIFICATION AND KNOWLEDGE:

SPECIAL TOPICS  105
5. Contextualism, Pragmatic Encroachment,
and the Knowledge Norm of Assertion 107
6. Perceptual Justification 131

Part III: NATURALISTIC EPISTEMOLOGY 159

7. Epistemology, Cognitive Science, and Experimental
Philosophy 161
8. Philosophy’s Intuitional Methodology and the Role
of Science 184

Part IV: SOCIAL EPISTEMOLOGY 203

9. Testimony and Disagreement 205
10. Collective and Institutional Epistemology 224

Part V: PROBABILISTIC EPISTEMOLOGY 249

11. Probabilistic Epistemology 251
Works Cited 283
Index 295

iii

00-Goldman-FM.indd 3 16/10/14 12:49 AM

 Epistemology: A Contemporary
Introduction

Epistemology is a very old field; it is also a young and vibrant field, with
many new directions and fresh ideas. The traditional core themes of episte-
mology, including the challenge of skepticism and the nature of knowledge
and justification, are examined in the first half of the book. Several new
angles and methodologies that arose and matured in recent decades are ex-
plored in the second half. These include naturalized epistemology, experi-
mental philosophy, and social epistemology. Also included is a substantial
treatment of probabilistic epistemology, a favorite subfield of contemporary
researchers that rarely makes an appearance in introductory texts. It is im-
possible, of course, to cover everything. We have gravitated toward the topics
we know best but have tried to preserve a balance between the probable
tastes and interests of potential users and what we ourselves are most ex-
cited about.
Several other balances have also served as desiderata. One is a balance be-
tween accessibility to undergraduates and relevance to (beginning) graduate
students. This has meant keeping the debates challenging and of contempo-
rary relevance while not overwhelming beginning students with excessive
detail. Another balance at which we have aimed is one between a broad and
fair coverage of competing positions while not hiding or refusing to articu-
late our own favored points of view. Philosophy is a more personal field than
most academic fields, and one really gets the feel for it when one sees how a
philosopher sets out and defends a perspective in some depth. The two
authors do not share the same perspective in all matters discussed here. Each
chapter has a primary author responsible for its content, but we have thor-
oughly critiqued one another’s chapters—not to the point of full agreement,
necessarily, but to the point of respectful satisfaction. (Actually, we probably
disagree a lot less than a randomly selected pair of epistemologists.)
Given the amount of material covered in this text, there will inevitably be
different ways to use it, and we encourage instructors to be experimental.
A full-year course could easily cover all eleven chapters. For a one-semester
course, instructors will probably want to select a subset of the chapters.
A course that emphasizes the “core” of epistemology would focus on the first
six chapters. A course that seeks to introduce students to the newest, most

iv

00-Goldman-FM.indd 4 16/10/14 12:49 AM

 Epistemology: A Contemporary Introduction v

cutting-edge topics would invest a lot of time in the last five chapters.
A course that seeks to sample from both of these areas could be constructed
in multiple ways. Instructors should have no trouble designing courses to
their own taste.
There are many people to thank for helpful guidance and/or inputs into
the authorship process. Our editors, Robert Miller, Emily Krupin, Diane
Kohnen, Cindy Sweeney, and Wendy Walker, displayed a nice balance of
urgency and patience. They also obtained for us an excellent cast of reviewers
of the manuscript’s first draft, including James Beebe, David Bennett, Richard
Fumerton, Peter Graham, Kristoffer Ahlstrom, Anna-Sara Malmgren, and
Patrick Rysiew. The book is better because of their comments. We would also
like to thank Marina Folescu, Ted Poston, Paul Weirich, Bob Beddor, and
especially Jack Lyons (who gave us comments on the entire manuscript) and
Patrick Rysiew (who provided extra comments on Chapter 7). Thanks are
also due to the students of Alvin’s epistemology class at Rutgers University,
fall 2013, who road-tested the draft manuscript. Finally, Isaac Choi contributed
editorial assistance and suggestions for end-of-chapter questions for Alvin’s
chapters.

AIG and MM

00-Goldman-FM.indd 5 16/10/14 12:49 AM

00-Goldman-FM.indd 6 16/10/14 12:49 AM
 part I

Justification
and Knowledge:
The Core Issues

01-Goldman-Chap01.indd 1 15/10/14 6:02 PM

01-Goldman-Chap01.indd 2 15/10/14 6:02 PM
 1

The Structure of Justification

Alvin I. Goldman

1.1 T h e Co n ce p t s a n d Q u e s t i o ns
o f E pis t e m o lo g y
Epistemology is the study of knowledge and related phenomena such as
thought, reasoning, and the pursuit of understanding. It is less a study of
customary thinking processes—although they are relevant—than a study of
better versus worse ways to think, reason, and form opinions. Moral theory
reflects on what is right and wrong in the sphere of action, while epistemol-
ogy reflects on what is rational or irrational, justified or unjustified, in the
sphere of the intellect.
Why are matters of the intellect important? This can be approached from
many vantage points of ordinary life. Do I want to make good decisions in
life, decisions that promote my own welfare, my family’s welfare, my com-
munity’s welfare? If so, I had better figure out which of the available choices
would best promote favorable outcomes. In other words, I need to form cor-
rect, or true, beliefs about the consequences that would ensue from the per-
formance of different actions. If I form true beliefs about the consequences of
each choice, I am more likely to make choices that lead me in useful direc-
tions. If I form false beliefs, my choices may be unfortunate however good
my intentions. Accurate beliefs tend to guide us down desired pathways,
inaccurate beliefs down pathways we don’t mean to travel. A student decides
to get trained in a given field because employment opportunities are pre-
dicted to boom in that field within a few years. Training in this field should
bode well for the student if the prediction is true, but a wasted effort if it is
false. One hopes to act on true predictions rather than false ones.
The same point emerges when deciding whom to trust in this or that
domain. In consulting a physician about an ailment, I want her to have exten-
sive medical knowledge plus the skill of applying that knowledge to new
cases. I want her to predict correctly which treatments would cure or allevi-
ate my ailment (and have no serious side effects). The same point extends to

01-Goldman-Chap01.indd 3 15/10/14 6:02 PM

 4 Chapter 1: The Structure of Justification

a choice of a financial advisor or auto mechanic. The claim here is not that
instrumental or practical value is the only value of true belief or the sole
ground of the intellect’s importance. People have intellectual interests not
rooted in practical affairs. We are curious creatures. We want to know, for
example, what caused the dinosaur extinction many millennia ago, even if
knowing this has no immediate action implications. This curiosity does not
rest on the pursuit of practical ends.
All right, you concede, we have reasons to try to get truths, and to enlist the
help of others who are adept at getting truths. But how is truth to be acquired?
Can one ascertain the truth by just reaching out and grasping some facts? It
isn’t clear how one does that. This problem is what epistemology is largely
about. A customary way to approach the issue is to focus on another central
topic in epistemology: justification. If I can get a justified or warranted belief,
rather than a randomly or haphazardly chosen one, such a proposition is
more likely to be true. How, then, do I go about getting justified beliefs? Some
epistemologists link justification to having good evidence. Others link it to
following good methods or procedures. These are among the prospects to be
explored in what follows. In this chapter and Chapters 2 and 6 our discussion
focuses on justification; in Chapters 3, 4, and 5 it focuses on knowledge.
As these introductory remarks make clear, epistemology talks a lot about
belief. What is belief (or opinion)? It is a species of psychological attitude toward
a proposition, where a proposition (roughly) is a content that purports to ex-
press a fact. The statement that the dinosaur extinction was caused by a massive
asteroid impact on the earth purports to state a fact. To believe this proposition
is to mentally “assent” to it, or think that what it states is so. Belief belongs to a
family of psychological attitudes directed at propositions. Other members of
this family include disbelief, which is the mental attitude of rejecting, or denying,
a proposition, and agnosticism, or suspension of judgment, which is the mental
attitude of neutrality, or indecision, with respect to a proposition’s truth. Col-
lectively, these different attitudes (plus more finely graded attitudes such as
being 75 percent convinced of something) are called doxastic attitudes. The term
doxastic comes from the Greek word doxa, meaning opinions.
It is widely thought that forming a belief that is reasonable, warranted, or
justified is the best means available of forming a true belief. Hence, episte-
mology is particularly interested in the question of how one should go about
acquiring a belief that is warranted, justified, or reasonable. Justification
and warrant are examples of terms of epistemic evaluation or appraisal. To
call a person’s belief “justified” is to commend it, or appraise it positively,
along some evaluative or normative dimension, whereas calling a belief
“unjustified” or “unwarranted” is to criticize it, appraise it negatively, along
the same dimension. Suppose I say that John thinks that Gregory is a terri-
ble fellow, but this belief is totally unwarranted because it is wholly based
on Gregory’s appearance. I call it “unwarranted” because judging people by
their looks is like judging a book by its cover. It is a poor basis for judgment
(or belief). Such a normative assessment has parallels with moral discourse,
in which actions are called right or wrong. Epistemologists agree, however,

01-Goldman-Chap01.indd 4 15/10/14 6:02 PM

 1.1 The Concepts and Questions of Epistemology 5

that epistemic terms of appraisal like justified and unjustified, warranted and
unwarranted, are not terms of moral evaluation; they express evaluations
along some intellectual dimension. If you believe in the absence of good
evidence, or by reliance on unsound reasons, this is an intellectual failing
rather than a moral one.
Although justification may be significantly related to truth, truth and jus-
tification are not equivalent concepts. A proposition can be true although
nobody believes it, and it can be true although nobody is justified in believ-
ing it. Consider a precisely delimited expanse of seaside beach. Some propo-
sition of the following form is true: “The number of grains of sand on this
beach is N.” But nobody is justified in believing this truth (with the correct
value of N filled in) because nobody has determined or ascertained the cor-
rect value of N. A major question in epistemology is how justification is re-
lated to truth, but they are not equivalent concepts.
Truth, it may be said, is a purely metaphysical concept rather than an epis-
temological one. Given a proposition, what makes it true or false is simply
the state of the world. Its truth-value is not affected by cognitive relations
people have toward the relevant state of affairs. But cognitive relations to a
proposition are precisely what are crucial in determining justification or
warrant. A person’s justifiedness with respect to a proposition P is never (or
rarely) fixed by P’s actual truth-value. Despite a proposition’s truth, it is pos-
sible for someone to lack any evidence for its truth (as in the grains-of-sand
example). Conversely, it is possible to have highly favorable (though mislead-
ing) evidence that justifies one in believing a proposition despite its falsity.
Thus, truth and justification must be carefully distinguished.
Justified and warranted are not the only terms of epistemic evaluation; an-
other familiar term of epistemic evaluation is rational. Some epistemologists
equate rationality with justification, but we shall keep them distinct. Episte-
mologists who favor rationality-talk as compared with justification-talk
often opt for a more finely delineated range of doxastic attitudes. Instead of
the tripartite classification scheme of belief, disbelief, and withholding (or
suspension), they prefer degrees of belief arrayed along a zero-to-one inter-
val. One represents the highest possible strength of credence in a proposi-
tion, and zero represents the lowest. Our own treatment here will mainly
use the tripartite classification scheme. But in Chapter 11, on probabilistic
epistemology, finer gradations will often be invoked.
In addition to justification, warrant, rationality, and reasonability, a critical
concept in epistemology is knowledge. Indeed, the term epistemology just means
the study of knowledge (episteme, in Greek). According to many theories, there
are intimate relations between knowledge and some of the other epistemic
concepts we have introduced. Here are some points of widespread agreement.
First, knowledge implies truth; you cannot know that P unless P is true. (This
idea is often conveyed by saying that truth is factive.) If you are certain of a
proposition P, in your own mind, you will be inclined to claim to know it. But
if P is not true, as a matter of fact, then you don’t really know it. (One cannot
know what isn’t so.) Second, a person must believe P, or have reasonably high

01-Goldman-Chap01.indd 5 15/10/14 6:02 PM

 6 Chapter 1: The Structure of Justification

credence in P, in order to know it. Knowledge is (partly) a psychological state.

Third, most theories hold that knowledge requires justification. You do not
know that P by simply believing P where P is true. Being justified in believing
it is also necessary. Thus, there appears to be a web of relationships between
knowledge and other important concepts. Closer examination of these mat-
ters occupies Chapters 3, 4, and 5. This book begins with justification, which
some regard as the more fundamental epistemological concept.

1. 2 T h e E pis t e m i c R eg r e ss Pro b l e m
Many justified beliefs attain this status when people infer them from other
things they believe. Inferential relations between beliefs are often expressed
in conversation or debate with others. Suppose Henry says to Tony, “I hear
you believe that New York will win the basketball championship next year.
How can you think that? New York was so miserable last season; how can
you think it will turn things around and win next year?” Tony responds:
“New York landed the top new draft pick in the league this year, and he is
rumored to be the most talented player since Michael Jordan. It has also ac-
quired one of the best big men in the league. With these crucial additions,
I feel confident it will win.”
Tony offers an argument featuring P1 and P2 as premises and C as
conclusion:

P1: New York will have the most talented draft pick in years.
P2: New York also has one of the best big men in the league.
C: Therefore, New York will win the championship next year.

Tony defends C by appeal to P1 and P2, implying that C is a reasonable

conclusion to draw from P1 and P2 (given various unmentioned assump-
tions). C does not deductively follow from P1 and P2, that’s clear. But it may
be a reasonable nondeductive inference from P1 and P2 (given the additional
assumptions).
Our story presents Tony as a speaker who verbally defends his belief to a
challenger, but a similar story might be told where there is no challenger and
Tony makes no verbal defense of C. If he holds his belief in C in utter silence,
but it is (properly) based on (justified) beliefs in P1, P2, and the additional
assumptions, the same justificational upshot will hold with respect to propo-
sition C. Thus, interpersonal speech is not required for there to be justified
inferential belief. Personal justification has no essential tie to a “dialectical”
situation, where verbal reasons-giving takes place.
What is the structure of Tony’s justification for C? As described so far, it
resembles the roots of a tree. In Figure 1.1, the belief in proposition C (labeled
Bel(C)) appears as the top node of a tree, with beliefs in premises P1 and P2
(labeled Bel(P1) and Bel(P2)) shown as downward-branching roots. The
arrows indicate that Bel(P1) and Bel(P2) lend joint support to Bel(C). Inferen-
tial justification is supposed to be transmitted upward from one or more root

01-Goldman-Chap01.indd 6 15/10/14 6:02 PM

 1.2 The Epistemic Regress Problem 7

nodes to at least one higher-level node. However, if the lower-level nodes

have no justification of their own to transmit—in other words, if Tony does
not believe them justifiedly—then the higher-level node(s) cannot inherit jus-
tification from them.

Bel(C)

Bel(P1) Bel(P2)
FIGURE 1.1 Belief in C is Justified by Inference from Justified Premises P1 and P2.

Might the root structure of Tony’s justification run deeper? If his beliefs in
premises P1 and P2 are justified, how was their justification acquired?
Perhaps Tony had other justified beliefs from which P1 and P2 were properly
inferred. His belief in P1, for example, may have been based on his justified
beliefs about the new draft pick’s previous scoring records and the tall re-
cruit’s rebounding prowess. To depict this situation in the diagram, we would
expand the root structure by adding another level of branching roots, roots
that extend downward from Bel(P1) and from Bel(P2). Of course, each addi-
tional root node must also be justified if Tony’s beliefs in P1 and P2 are to
acquire justification from them.
What emerges here might be called a regress of reasons, grounds, or justi-
fication. How does this regress of justification continue? This question poses
what is called the regress problem in epistemology. It was posed in ancient
times by Sextus Empiricus, who asked about the chain of “proofs” for a per-
son’s assertion. Here we focus on beliefs rather than assertions and reasons
rather than proofs, but the core idea—and worry—is much the same. Can the
regress of reasons continue indefinitely, with additional beliefs being in-
voked at each step, with no repetition and no end points? Or must the regress
eventually terminate along each root, where a terminus is a justified belief
that acquires its justificational status from an epistemic source distinct from
inference? What are the possible structures here, and which possibility cap-
tures the real structure of inferential justification?
Here are three views about the correct structure of a regress—that is, a struc-
ture that enables justified beliefs to be derived from other (justified) beliefs:

1. Infinitism. The correct structure is an unending continuation of rea-

sons, without repetition or end. A justification tree for each inferen-
tially justified belief contains roots that never end and never contain
repetitive nodes.
2. Foundationalism. The correct structure is a tree with roots of finite length.
Each root has a terminal, or final, node that represents a justified but

01-Goldman-Chap01.indd 7 15/10/14 6:02 PM

 8 Chapter 1: The Structure of Justification

noninferred belief. Such beliefs are foundational, or basic, beliefs. All jus-
tified beliefs that are not themselves basic, however, ultimately derive
their justification from one or more other beliefs that are basic. So, all
justification ends at—or begins with—basically justified beliefs.
3. Coherentism. The correct structure is one in which some of the roots
circle, or loop, back on themselves. In other words, some nodes reap-
pear earlier in the chain. Thus, the regress has no end points. At the
same time, it is not infinitely long. In rejecting end points, coherentism
joins infinitism in rejecting the notion that justification depends on
there being foundational, or basic, beliefs.

A fourth response to the regress problem belongs in a separate category

because it does not purport to “solve” the problem of justification—that is, to
show how inferential justification is possible (and feasible):

4. Skepticism About the Regress. None of the first three solutions to the re-
gress problem is satisfactory. One cannot get justified belief via any of
the structures they describe. Because those other three solutions
jointly exhaust the possibilities for a positive solution, there is no way
by which inferential beliefs can be justified.

In claiming that none of the three solutions works (for reasons to be exam-
ined below), the skeptical position implies that justification can never be de-
rived by inference. This response is skeptical because it denies justification for
a huge swath of beliefs—all beliefs based on reasoning—that we normally
consider justified.
Before examining these rival responses to the regress problem, let us note a
shared feature of the positive solutions. All of these theories restrict beliefs
that do the justifying to ones held at the same time as the target belief. In the
case of Tony’s basketball championship belief, our question is whether his
belief at the time of the conversation is justified. Obviously, Tony’s justification
for C depends on whether he is justified in believing that the new draft pick
is as talented as he claims. This is something he might be justified in believing
in, say, December, although he wasn’t justified in believing it in October.
The three positive theories under discussion are all “classical” theories of
justification in the sense that they center on synchronic reasons. A synchronic
reason for a belief is a factor occurring at the same time as the target belief.
Earlier or later beliefs make no justificational difference in the traditional
view. Presumably everybody would agree that Tony’s belief that New York
will win the championship cannot be justified in the way indicated if Tony’s
beliefs about the new players either weren’t really in place at the time he be-
lieved C or weren’t yet justified at that time, say, because the draft of the
super-­talented player had not yet occurred or the player had not yet signed a
contract. Surely the supporting premise beliefs must be held and justified at
the time of the conclusion belief; so the synchronic view would have it. We
can agree that justified premise beliefs that only occur after a target belief

01-Goldman-Chap01.indd 8 15/10/14 6:02 PM

 1.3 Infinitism 9

cannot help the latter belief (at the earlier time). However, this leaves open
the possibility that earlier beliefs and their justificational statuses can influ-
ence the justificational status of a later belief (albeit indirectly). That is a pros-
pect we shall encounter in connection with a diachronic theory of justification
to be considered in Chapter 2. Diachronic theories as such do not solve the
regress problem, certainly not simply by virtue of being diachronic. The
point is only to indicate that theories of justification need not be purely syn-
chronic. All classical theories of justification, however, have been synchronic,
including infinitism, foundationalism, and coherentism.

1.3 I n fi n i t ism
Infinitism is rarely given serious consideration. Two problems jump out on
first encounter. First, if a justification tree goes back infinitely far (on one or
more roots), there seems to be no point at which justification originates. But
if there are no originating places for justification—places where justification
gets “kick-started”—how can there be any justification to transmit to higher
nodes? Second, if a tree contains an infinite chain of reasons, doesn’t this
imply infinitely many beliefs? But, surely, no human person has infinitely
many beliefs.
Does infinitism have any good response to these problems? In response to
the first problem, it may be argued that justification need not originate any-
where; starting points aren’t needed. Just as time and the universe may each
be eternal (in both directions), with no real starting points, so justification
can lack starting points. In the case of synchronic justification, of course, the
issue of temporal origination doesn’t arise. What the critic of infinitism
denies is that all beliefs on inferential trees are epistemically dependent on
others. The critic insists that some beliefs must be independently justified if
they are to generate any justificational “juice” to transmit to the others. This
is not a proof, but it seems intuitively compelling.
What of the criticism that nobody possesses infinitely many beliefs?
Here the infinitist has a bit of maneuvering room (Klein 2005). The princi-
pal maneuver appeals to a standard epistemological distinction between
two kinds of justification. Doxastic justification is a property of existing
beliefs; propositional justification arises from the state or condition of an
epistemic agent that entitles them to believe a proposition (even if they
don’t opt to believe it). Proposition P may be justified “for” such an agent in
the sense that they would be justified in believing it in virtue of their condi-
tion or situation. If you are shivering with cold on a wintry night, you are
justified in believing that you feel cold, even if you fail to notice this feeling
and don’t assent to the proposition. Once it is conceded that propositions
can be justified even without belief in them, infinitism can be interpreted
as the view that there are infinitely many propositions (rather than beliefs)
that form chains of inferential justification. Even if no finite being like us
has infinitely many beliefs, this leaves open the possibility that there are
infinitely many propositions we are justified in believing at a given time.

01-Goldman-Chap01.indd 9 15/10/14 6:02 PM

 10 Chapter 1: The Structure of Justification

The propositions might be arrayed in a chain of (potential) inferential de-

pendence. Not that we actually execute all of the corresponding inferences,
but we would be justified in executing any finite segment of them.
Does a switch from an infinite chain of beliefs to an infinite chain of prop-
ositions really work? What the infinitist presumably holds is not only that a
proposition can be justified without itself being believed but also that it can
receive (inferential) justification from other propositions without the latter
being believed. So infinitism supports the viability of a chain of proposi-
tions, each of which bears (or possesses) justification but none of which is
believed. This is extremely dubious: Unless a proposition is believed, it is
doubtful that it can convey inferential justifiedness to other propositions.
Don’t some propositions in the chain have to be believed? Perhaps so, the in-
finitist might reply, but this only requires one proposition to be believed, not
infinitely many of them. Is this correct? If a single believed proposition is
inferentially justified, doesn’t it have to be inferred from other beliefs earlier
in the chain? But then we are off and running on the same sort of infinite
regress of beliefs the infinitist sought to avoid.
Of course, we have already encountered an example in which a proposi-
tion is deemed (propositionally) justified without receiving that justifica-
tional status from a belief. That’s the feeling-cold example of two paragraphs
back. The problem is that no infinitist can embrace this sort of case as proto-
typical, because it features a proposition that acquires justification from a
nondoxastic source, and hence is an example of basic, or foundational, justi-
fication. This supports not infinitism but one of the rival solutions instead
(foundationalism).

1.4 Co h e r e n t ism
We proceed to the coherentist solution to the regress problem. Coherentism
is a minority view today but historically was an influential theory and doubt-
less retains some of that luster. In terms of the regress problem, it is com-
monly conceptualized as tolerating circular inference. But this may not be
the best way to introduce it; let us consider another perspective.
Coherentism depicts a body of justified beliefs as a holistic system whose
parts mutually support one another. There are no “privileged” beliefs that
play the role accorded to foundational beliefs under foundationalism
(i.e., beliefs that obtain justification from outside the doxastic system and
transmit this justificational juice to the remaining elements of the system).
Under coherentism, all justification arises from interrelations among beliefs
(or other doxastic states). Moreover, the justificational interrelations are re-
ciprocal or bidirectional rather than unidirectional.
A good metaphor for coherentism is a house of mutually supporting cards
leaning against one another. Such a structure stands upright because its com-
ponents are each supported by all of the others, or at least many of the others.
Support runs in many directions. In a four-card house consisting of A, B, C,
and D, support might run from A, B, and C to D, from B, C, and D to A,

01-Goldman-Chap01.indd 10 15/10/14 6:02 PM

 1.4 Coherentism 11

and so forth. There are no “foundational” cards that support others but are
themselves wholly unsupported. The very formulation of the regress prob-
lem assumes that justificational support is linear—that is, runs in a single
direction. Coherentism challenges this assumption.
Now consider how justificational coherence might be spelled out, meta-
phors aside. Unfortunately, there is little consensus here. One ingredient that
presumably contributes to system coherence is logical consistency. Unless
the beliefs in a total doxastic system are mutually consistent (i.e., their con-
junction does not imply a contradiction), the system isn’t coherent. But this is
an extremely weak condition. Thus, consider the following set of proposi-
tions as someone’s total system:

1. There are cats.

2. There are bananas.
3. Sometimes it rains.
4. There is a sun.
5. There is a moon.

These five propositions constitute a consistent system, because their con-

junction entails no contradiction. But is this consistent system a highly coher-
ent one? No; at least not in the rich sense of coherence that coherentists
typically advance. The system fails to be coherent because (i) no proposition is
deducible from any of the remainder (either singly or in conjunction with one
another) and (ii) none of the propositions would have its probability raised by
taking the truth of the others as given. In short, the five propositions are
mutually independent, both logically and probabilistically. Coherence, how-
ever, is supposed to be the opposite of this; a coherent system should display
a high degree of mutual interdependence.
Here is an illustration of a system with greater interdependence:

6. Edna loves cappuccinos.

7. On Tuesday afternoon Edna had a yen for a cappuccino.
8. On Tuesday afternoon Edna thought that the closest place to get a
good cappuccino was Café Nero.
9. On Tuesday afternoon Edna went to Café Nero for a cappuccino.

Like system (1)–(5), this system of propositions is logically consistent. In ad-

dition, however, it also features several probabilistic support relations among
various members of the system. The truth of (6) raises the probability of (7); the
truth of the conjunction of (7) and (8) considerably raises the probability of (9);
and arguably (9) raises the probability of (7). So there is a fair amount of mutual
coherence among these propositions, and hence the system as a whole has a
fair degree of systemic coherence (especially for a small system).
Now comes the jackpot question: Does a system’s level of coherence guar-
antee or contribute materially toward a positive justificational status for the
propositions that constitute the system? Does the fact that a belief system is

01-Goldman-Chap01.indd 11 15/10/14 6:02 PM

 12 Chapter 1: The Structure of Justification

highly coherent guarantee, or even make it likely, that each of its members
(or most of them) is justified? No.
Notice first that the propositions constituting a consistent system need not be
true, and similarly for the members of a coherent system. System (6)–(9) is more
coherent than system (1)–(5), but this is no clue as to which system has more
true members. Now turn to justification. Would the fact that somebody believes
all and only the members of (1)–(5) and another person believes all and only the
members of (6)–(9) provide any clue about which believer has a larger propor-
tion of justified beliefs? No. It all depends on how the beliefs were formed,
or arrived at, by the two believers, and nothing said thus far speaks to their re-
spective processes or methods of belief formation. The believer in propositions
(1)–(5) may have formed each of her beliefs by careful observation, whereas the
believer in propositions (6)–(9) may have formed his beliefs by mere wishful
thinking. Thus, the justificational quality of the member propositions does not
correlate with system coherence. Contrary to coherentism, nothing follows
about a belief’s justifiedness—certainly not its doxastic justifiedness—from the
fact that it belongs to a system that ranks high on a systemic-coherence scale. It
may also fail to be propositionally justified if the relation of coherence among
the propositions is too complex or obscure for the subject to appreciate it.
This point may be reinforced by considering the following possibility. Al-
though we may believe each member of a coherent system, we might not
recognize or (intellectually) appreciate the fact that they cohere. We might
form each belief separately from the others, without recognizing their logical
and/or probabilistic relations. In this scenario, the fact that our beliefs are
mutually coherent contributes nothing to their level of justifiedness. Their
justifiedness depends on whatever leads (or causes) us to believe them. Since,
by hypothesis, we are not influenced by their coherence relationships, those
relationships have no impact on their justifiedness.
We can highlight these points by tweaking an example often used against
coherentism. Fiona loves to fantasize. She sits in her armchair weaving com-
plex stories that rival the best detective mysteries for intricacy of plot and
depth of detail. Fiona’s fantasy is, in effect, itself a well-composed novel with
a high degree of coherence (the plot “hangs together”). Moreover, the fantasy
is so vivid and realistic that Fiona believes each of the elements in the story.
Is she justified in believing them (to be true)? Surely not. Yet each such ele-
ment is a member of a highly coherent system. The system would rate very
high on the coherence dimension, judged, at least, by their internal relations
with one another.
Here are two additional problems for coherentism, the first commonly
called the isolation objection. According to coherentism, a given belief’s jus-
tificational status is exclusively a function of the believer’s other belief
states. Psychological states of other kinds, such as perceptions and feelings,
are irrelevant. According to coherentism, your feeling cold right now has no
bearing on any of your beliefs’ justificational statuses, even the status of a
belief to the effect that you feel cold or feel hot. How you actually feel makes
no difference, because justification is exclusively a function of inferential

01-Goldman-Chap01.indd 12 15/10/14 6:02 PM

 1.5 Foundationalism and Basic Beliefs 13

relations between beliefs, and neither feeling cold nor feeling hot is a belief.
Similarly, if we hear what sounds like a motorcycle careening down the
street, hearing this sound is justificationally irrelevant to any motorcycle
beliefs we may have. This approach seems wrong-headed. Beliefs that are
misaligned with concurrent experience (especially vivid experience) are
unjustified. After all, perceptual experience is input from the world, and
ignoring such experience does not conduce to sound belief. Some episte-
mologists (e.g., BonJour 1985) try to tinker with coherentism to avoid this
counterintuitive feature, but none of them has had much success.
Finally, a seemingly small defect in coherentism shows how dramatically
off-track it is. Coherentism says that a belief is justified if and only if it be-
longs to (is a member of) a highly coherent total system of beliefs. An obvious
consequence of this is that all members of a highly coherent system are justi-
fied and all members of a weakly coherent system are unjustified. In short, all
members of any total system have the same justificational status. This makes
little sense: Normal people have a mix of justified and unjustified beliefs, a
mix that coherentism has no obvious resources to accommodate.

1. 5 F o u n dat i o n a l ism a n d B a si c B e l i e fs
Foundationalism’s response to the regress problem says that every root in a
tree of (successful) inferential justification terminates after finitely many
steps. Each stopping point (or starting point, one might say) is a belief that
possesses justification it does not get from further (premise) beliefs. Such
starting points, or “basic” beliefs, have two crucial properties: (A) they are
uninferred and (B) they are justified. In addition, foundationalist theories
say that there are inferential relations between basic and nonbasic beliefs
such that enough of the latter are justified to avert the threat of global skepti-
cism. Roughly speaking, a wide swath of our commonsense beliefs qualify
as justified under foundationalist principles.
To succeed, a developed form of foundationalism must address four ques-
tions (we shall frequently use the terms immediate and mediate justification for
basic and nonbasic justifiedness, respectively):

1. What does it take for a belief to attain the status of “justified” in an

unmediated or noninferential fashion?
2. Which types of belief qualify as immediately justified? (For example,
what types of propositional content lend themselves to immediate
justification?)
3. What strength of justification must immediately justified beliefs pos-
sess according to foundationalism? Must they exhibit the highest
grade of justification (i.e., certainty or infallibility)?
4. Assuming that many immediately justified beliefs are available, does
this enable epistemic agents to draw enough reasonable inferences to
further beliefs to dispel the specter of skepticism often laid at founda-
tionalism’s doorstep?

01-Goldman-Chap01.indd 13 15/10/14 6:02 PM

 14 Chapter 1: The Structure of Justification

The foregoing questions pose the constructive tasks confronting the foun-
dationalist program. Can it provide satisfactory answers? Foundationalism’s
critics have attacked it with energy and zeal, challenging the very possibility
of its delivering the goods. We shall give examples of some roadblocks
erected by the critics. Do these roadblocks eviscerate all prospects for a suc-
cessful foundationalism, or do the opponents’ criticisms fall apart when ex-
amined closely?

1.6 B a si c B e l i e fs A s Se l f - J us t i f y i n g B e l i e fs ?
Focusing on the first (and central) question, one answer is that immediately
justified beliefs are self-justifying beliefs. The term self-justifying often crops
up in attempts to explain what is distinctive to immediate justifiedness. Self-
justifying sounds like the right contrast with inference-based justifiedness,
since inference is always a relation with other beliefs. So perhaps immediate
justifiedness is what accrues when a belief is self-justifying. But what does it
mean for a belief to justify itself? Is this even possible? Perhaps a belief justi-
fies itself in case its occurrence logically entails the truth of its content.
A belief to the effect that one has some beliefs seems to satisfy this condition.
To be sure, few beliefs exemplify this property: Believing it is going to rain
tomorrow does not entail it will rain tomorrow. Still, in those few cases in
which a belief has this property, maybe it is immediately justified.
Would this answer to question (1) satisfy the goals of foundationalism? It
is widely assumed that paradigm cases of immediately justified beliefs con-
cern our own current (nonfactive) mental states—for example, “I believe I am
in pain,” or “I believe I want an espresso.” But the proposed definition will
not guarantee this result, first because the desired logical entailment will
rarely hold. Does the fact that I believe I am sad logically entail that I am sad?
That is highly dubious. Perhaps we are always correct when we classify our
own current mental states—but is this logically, or even metaphysically,
guaranteed? These are problematic theses.
However, even if it is true in such cases that believing we are in mental
state M guarantees that we are in M, why should this imply justifiedness of
the belief? For any truth L of logic, my believing L (trivially) entails its truth
(because it is necessarily true). But it hardly follows that I am justified in be-
lieving it. L might be such a complex proposition that although I manage to
grasp its content, I fail to understand how or why it is true. Thus, even when
the content of a belief guarantees its truth, a bit of reflection shows that this
does not guarantee its justifiedness.
Here is a slightly different way to understand self-justification. If person S
believes that she is currently in mental state M and S is in M, then S is justified
in believing this proposition. Under this second conception, self-­justification
works a little differently: It isn’t the believing that confers justifiedness, but
the belief’s being true that confers justifiedness. This idea was defended by
Roderick Chisholm (1977). Chisholm offered the following formula: “What
justifies me in thinking I know that a is F is simply the fact that a is F.”

01-Goldman-Chap01.indd 14 15/10/14 6:02 PM

 1.7 Debunking Some Debunkers of Immediate Justification 15

Chisholm conceded that the formula does not apply to every proposition
that instantiates it:

Thus, in answer to “What justification do you have for counting it as evident

that there can be no life on the moon?” it would be inappropriate—and
presumptuous—­simply to reiterate “There can be no life on the moon.” But we
can state our justification for certain propositions about our beliefs, and certain
propositions about our thoughts, merely by reiterating those propositions.
(21, italics added)

Unfortunately, Chisholm offers no explanation of why it is appropriate

to defend the justification for certain beliefs by reiteration but not others.
Why does it work for first-person mental-state propositions but not for
third-person mental-state propositions? What is the crucial difference be-
tween them?
Chisholm introduced a phrase to distinguish the one class of proposi-
tions or states of affairs from the other class. States of affairs are called self-­
presenting when it is appropriate to reiterate the fact in defense of a
justification claim. This provides a label, but not the slightest explanation of
the alleged epistemic difference. Perhaps such an explanation is intended
when Chisholm says that a self-presenting state is one that is “apprehended
through itself” (22). But this obscure notion is never explained. If the phrase
of immediate justification is not explained more clearly, a defense of im-
mediate justification remains elusive.

1.7 D e b u n k i n g So m e D e b u n k e r s
o f I m m e d i at e J us t i fi c at i o n
Perhaps the case for foundationalism takes on an obscure character from
these unhelpful attempts to clarify the nature of immediate justification. In
principle, however, this might not be a hopeless task. To get some apprecia-
tion for what foundationalism is after, consider first the following analogy.
In European history becoming a monarch was usually a matter of lineage:
One became a monarch by standing in the right lineal relationship to some-
one else who had been a monarch. Not every monarch, however, became
one by inheritance; some became monarchs by leading a conquest of new
territory. So, in addition to monarchy being passed on by inheritance, it was
also sometimes acquired de novo. This parallels the idea of justificational
status (J-status) being generated without any justificational “juice” being
transmitted from one possessor to another. This makes good sense in the
abstract, and there seems no reason to preclude it, at least in principle, in the
realm of epistemic justification.
Nonetheless, many epistemologists are pessimistic about the prospects for
foundationalism. Several have tried to debunk immediate justifiedness by
demonstrating its inherent unrealizability. William Alston (1983) has sur-
veyed some of these attempted debunkings and undertaken to debunk those

01-Goldman-Chap01.indd 15 15/10/14 6:02 PM

 16 Chapter 1: The Structure of Justification

debunkings. Let us consider a few of these critical assaults and Alston’s re-
joinders to them.1 We begin with the ground rules Alston lays down, more
specifically the core notions of mediate justification (MJ) and immediate jus-
tification (IJ):

(MJ) S is mediately justified in believing that p if and only if S is justified in be-

lieving that p by virtue of some relation this belief has to some other justified
belief(s) of S.
(IJ) S is immediately justified in believing that p if and only if S is justified in
believing that p by virtue of something other than some relation this belief has
to some other justified belief(s) of S. (Alston 1989, 58)

Immediate justification is defined here purely negatively: It is justification

that does not involve any inferential relation or other relation to other justi-
fied beliefs. This skeletal definition does not supply an answer to question (3)
posed above; it does not provide positive conditions for immediate justified-
ness. Alston goes on to argue that the relevant debunkers he addresses make
sneaky attempts to substitute a different conception of immediate justifica-
tion than the simple conception, (IJ). In effect he accuses debunkers of qui-
etly seeking to impose more stringent conditions on immediate justification
than foundationalism needs.
The basic maneuver may be illustrated in terms of an analogy. You are
going to a concert, ticket in hand. You approach the entrance to the concert
venue and hand your ticket to the ticket-taker standing at the door. The ticket-
taker says, “Well, OK. You have a ticket; but that’s not enough for entry. You
must also produce proof that you personally purchased this ticket from an
authorized ticket agency plus the date of your purchase. No such proof is on
the ticket itself.” You would be outraged, of course. Permission to enter public
events of this kind is normally guaranteed by presentation of an applicable
ticket. Further proof of how and when the ticket was acquired is not needed.
Such a surprise “upping of the ante” would be frowned upon in everyday
life. Analogously, Alston’s plan is to rebut certain debunkers of immediate
justification who (in his view) unfairly and unreasonably “up the ante” for
immediate justifiedness.
Alston picks out several debunkers who offer “level ascent” arguments
against the possibility of immediate justification. Level ascent is exemplified
in our concert ticket example. The first-level requirement for admission—
normally the only requirement—is presentation of a legitimate ticket. The
ticket-taker goes up a level by requiring not only a legitimate ticket but proof
that it is a legitimate ticket. A similar inflation of reasonable requirements
for immediate justification is what Alston claims to detect in the work of
certain critics.
The first such critic is Wilfrid Sellars (1956). Alston describes what Sellars
says about a certain approach to observational knowledge, which might ac-
count for how such knowledge can be immediate. This approach would say,
roughly, that an observer has perceptual knowledge of an object’s being

01-Goldman-Chap01.indd 16 15/10/14 6:02 PM

 1.7 Debunking Some Debunkers of Immediate Justification 17

green if, in the presence of green objects, he reliably forms beliefs that they
are green. Sellars, arguing to the contrary, says that Jones does not know that
the object is green unless he is able to reflect on his own performance and
take the formation of his statement or belief as a reason for supposing that a
green object is present. Alston diagnoses the maneuver as follows:

Sellars is clearly denying that observational knowledge is or can be immediate

knowledge, as that term was explained above [in (IJ)]. His reason for denying it
clearly falls under our Level Ascent rubric. One’s belief counts as knowledge
only if one knows something about the epistemic status of that belief, viz., that
it counts as a reliable sign of the fact believed. And, equally clearly, this move
could be used against any claim to immediate knowledge. (1989, 66)

As further support for his interpretation and rejection of Sellars’s level

ascent argument, Alston accuses Sellars (and others) of confusing two
subtly different concepts that need to be kept separate. The two concepts
are, first, the concept of a belief’s being justified and, second, engaging in
some kind of activity vis-à-vis the belief that reflects on it and establishes
its legitimacy. In the concert ticket case, there is a difference between the
ticket’s being legitimate and the person’s having an additional document
to prove or establish the ticket’s legitimacy. Similarly, Alston argues, it is
one thing for a belief to have a certain epistemic property (e.g., being justi-
fied); it is quite another for the believer to have a higher-level justification
for believing or asserting that it is legitimate. Immediate justification is a
first-level property like having legitimacy. When one insists, as Sellars ap-
pears to do, that having legitimacy also involves having a higher-level
justification about the belief’s legitimacy, this inevitably brings in addi-
tional beliefs and keeps the original belief from being immediately justi-
fied. For the foundationalist, the proper solution is to reject the insistence
on any extra, higher-level requirement.
Another epistemologist who tries to debunk basic justification by deploy-
ing a level ascent maneuver is Laurence BonJour (1985), who at the time was
a coherentist. BonJour begins with the usual assumption that foundational-
ism can succeed only if there is a certain class of beliefs that are immediately
justified. Moreover, given his discussion of justifiedness, a belief’s being jus-
tified implies that it is highly likely to be true. Thus, whatever feature you
might specify that promises to legitimize a belief as immediately justified,
possession of that feature must imply that the belief is highly likely to be
true. Call your favorite candidate for the correct feature F. Then for a belief
to qualify as basic, says BonJour, the premises of the following argument
must themselves be at least justified:

i. Belief B has feature F.

ii. Beliefs having feature F are highly likely to be true.

Therefore, B is highly likely to be true.

01-Goldman-Chap01.indd 17 15/10/14 6:02 PM

 18 Chapter 1: The Structure of Justification

BonJour then continues as follows:

And if we now assume, reasonably enough, that for B to be justified for a par-
ticular person (at a particular time) it is necessary, not merely that a justification
for B exist in the abstract, but that the person in question be in cognitive posses-
sion of the justification, we get the result that B is not basic after all since its
justification depends on that of at least one other empirical belief. (1985, 5–6)

In other words, BonJour is saying that it’s not enough for the premises of
his argument to be true—for example, that B has feature F (being likely to be
true). It is also necessary that the believer be justified in believing that B is
likely to be true. But this is an extra, higher-level belief that is said to be
needed for first-order justification. If this were indeed necessary, what is ini-
tially claimed to be a basic belief depends for its justification on the justified-
ness of a higher-level belief as well. If this were so, it would obviously
undermine foundationalism. The argument is again a level ascent style of
argument. It says that the justifiedness of any first-level belief B requires the
epistemic agent also to have a justified higher-level belief B* to the effect that
B is likely to be true. It is not enough for the first-level belief, B, genuinely to
be likely to be true. In addition, the agent must have a different belief, B*, that
justifies him or her in believing that B is likely to be true. This automatically
implies that B is not immediately justified, at most only mediately justified.
Alston’s response to BonJour is now predictable: He challenges BonJour’s
crucial step of requiring that for S’s first-level belief to be justified there must
be a justified higher-order belief to the effect that the first-level belief is
highly likely to be true. “In other words, in order that I be justified in accept-
ing B, I must know, or be justified in believing, the premises of the above
argument. And why should we suppose that?” (1989, 73). By Alston’s lights,
BonJour just begs a crucial question, and doesn’t really argue for it. Alston
concludes, “Thus in BonJour, as in Sellars, the contention that putatively im-
mediate knowledge really rests on higher-level reasons itself rests on a foun-
dation of sand” (77). Alston offers no positive defense of foundationalism by
specifying a particular criterion for immediate justifiedness, but he does
poke a large hole in the indicated attempts to refute the very possibility of
foundationalism.2

1. 8 Str e n g t h , Co n t e n t s, D e fe at e r s,
a n d B o os t e r s
Our discussion has focused almost entirely on our earlier question (1): What
are the necessary and sufficient conditions for obtaining immediate justifica-
tion? Let us now say a few words about questions (2) and (3) in our original
list of questions (Section 1.5).
Descartes was responsible for jump-starting the foundationalist picture
of knowledge and justification. He also led the way with the idea that the
starting point of the epistemic justification project resides in our own states

01-Goldman-Chap01.indd 18 15/10/14 6:02 PM

 1.8 Strength, Contents, Defeaters, and Boosters 19

of mind. His famous maxim “I think, therefore I am” and the struggles he
inaugurated about how to proceed from mental states to states of the exter-
nal world pervaded several centuries of epistemology. In the twentieth
century, however, foundationalists started to rethink matters and decided
that even physical object propositions (“There’s a pear on the table”) might
be an immediately justified belief. A resolution of this debate is still up in
the air.3
Roughly the same story can be told for debates over question (2). Descartes
again set things in motion by identifying the epistemological project with a
quest for certainty. He argued that certainty could be attained in the case of
first-person current mental propositions, and certainty was just the strength
of justification to demand for basic beliefs. In the last century, however, epis-
temologists began to have second thoughts: Perhaps immediate justifiedness
need not require the highest level of justifiedness. There is no obvious con-
nection between the type of belief content and this required justificational
strength, although some such connections have long been defended.
A less explored topic is the impact of so-called defeaters on the problem of
justificational types. In foundationalist lore, there are two kinds of (positive)
justificational statuses, either of which can characterize a justified belief.
There is immediately justified belief and mediately justified belief. This is an
exhaustive and exclusive typology. All justified beliefs are justified in one
manner or the other; no justified belief has its justification in both ways. But
this cannot be an adequate portrait of the epistemic landscape. In fact, the
justificational source for many beliefs can be both immediate and mediate,
both direct and inferential.
How can this transpire? First, let’s introduce some terminology. When a
proposition gets an “injection” of justification from some source, let us say it
is prima facie justified for the epistemic agent. Prima facie justification is pro-
visional, or tentative, justification, as opposed to justification on balance, or
all things considered. Whether the J-status of the proposition (for the given
agent) changes to on-balance justified may depend on other sources. Some of
those other sources might cut against the proposition, thereby reducing its
initial positive justifiedness. If that happens (to a sufficient extent), we say
that the undercutting or undermining sources defeat the prima facie justifica-
tion arising from the original source. The beliefs (or whatnot) that perform
the undermining are called defeaters. We might also introduce a term for fac-
tors that add justificational support to a proposition: let us call them boosters.
Both pieces of terminology can also be used in association with the language
of “reasons.” Doxastic or nondoxastic reasons may be boosters for believing
p. If there are both positive and negative sorts of reasons and the negative
ones are strong enough, they defeat the prima facie reasons in favor of p. In
that case the ultima facie (all things considered) J-status of the proposition is
unjustified. Alternatively, though an initial injection of justifiedness may be
insufficient to bring a proposition across the threshold of justifiedness, get-
ting a booster from a new source may add enough to bring the proposition
across the justificational threshold.

01-Goldman-Chap01.indd 19 15/10/14 6:02 PM

 20 Chapter 1: The Structure of Justification

Suppose you are looking at the sky and spot a plane with an unfamiliar
contour. Is it one of those new-fangled 829s, you wonder? That’s what they
look like, but the plane is partly in the clouds, it’s getting dark, and your
distance vision was never so good anyway. But you now recall reading that
this new model was going to be flight-tested in your area today. So you con-
clude, firmly, that what you are seeing is an 829. Your belief is justified but its
source of justification is twofold: both noninferential and inferential. The
first source is visual, presumably falling in the category of immediate justi-
fiedness. The second source is background information, falling in the cate-
gory of mediate justifiedness. The two sources jointly supply enough
justificational “juice” to make the belief ultima facie justified. But neither
source alone would reach across that threshold. A satisfactory theory of jus-
tification, obviously, must be capable of handling cases of this kind. If foun-
dationalism is to be such a theory, it had better be able to handle them.
Fortunately for foundationalism, this seems feasible with only a slight ad-
justment. Foundationalism can simply take the properties of immediate and
mediate justifiedness to pertain not only (or even primarily) to beliefs, but to
sources, factors, or components of justifiedness. Thus, the same belief can
have both a mediate source of justifiedness and also an immediate source.
Coherentists will here jump in eagerly. They will point out that for any
belief one is tempted to form by an immediate source, its justificational
status can always in principle be influenced by other beliefs one already
holds. “This is just what we have been saying all along,” coherentists will
exclaim. A careful epistemological should resist this claim. This is not
exactly what coherentists have said right along; at least it’s not the entirety
of what they say. Another, stronger thesis they endorse is that inference,
or inferential relations to other beliefs, is the only source of justification.
Coherentism denies the existence of any source of justifiedness other than
inference. So what was conceded in the previous two paragraphs is not
tantamount to coherentism. But it does give ground to coherentism at
least to some degree. It also seems to require foundationalism to counte-
nance the fact that many basic beliefs owe their justification, in part, to
inferential sources, albeit these basic beliefs would also have some justi-
fiedness arise from noninferential sources.

1.9 M ov i n g fro m t h e F o u n dat i o ns

to H i g h e r Flo o r s (Q u e s t i o n 4)
Cartesian foundationalism is the variant of foundationalism that confines
basic beliefs to beliefs about first-person current mental states. Setting aside
the question of how beliefs about mental states gain immediate justifiedness,
we now ask how a subject can proceed from such beliefs to propositions
about material things in the external world that also qualify as justified.
Clearly, this requires there to be legitimate methods for inferring such
external-­world propositions from propositions about current experience. Are
there such methods? Obviously, we cannot logically deduce external-world

01-Goldman-Chap01.indd 20 15/10/14 6:02 PM

 1.9 Moving from the Foundations to Higher Floors (Question 4) 21

propositions from ones about our own inner experiences. According to many
epistemologists and philosophers of science, however, our most important
and interesting inferences are not deductive anyway. A standard kind of
inference—­used both in everyday life and in science—is inference to the best
explanation (also called abductive inference).
After this morning’s snowfall, I see some tracks in the snow. I infer that the
tracks were made by a deer because they look like deer tracks and deer fre-
quent my neighborhood. This looks like a good inference and it’s clearly an
explanatory inference. I regard the deer hypothesis as a good explanation of
the observed tracks in the snow. But my neighbor tells me: “A company
called ‘Toys for Tricksters’ recently marketed a machine that simulates deer
tracks. Its tracks in the snow are indistinguishable from genuine deer tracks
(and leave no telltale wheel tracks). These machines have become very popu-
lar among jokesters.” My neighbor claims that my tracks were produced by
such a machine; or, when pushed a bit, he says that it is perfectly possible
that they were so caused. In either case he concludes that I am not justified in
believing it was a deer.
How should we adjudicate this issue? Some inference principle is needed
here, one for explanatory inference in particular. Here is a candidate
principle:

(JEI) Justification by Explanatory Inference. If hypothesis H purports to ex-

plain S’s evidence E, and there is no incompatible hypothesis H′ that
provides a better or equally good explanation of E, then S is justified in
believing H on the basis of E. [OR: . . . then S is justified in believing
hypothesis H on the basis of E if and only if there is no incompatible
hypothesis H′ that provides a better or equally good explanation of E.]

If we substitute for “H” the hypothesis that the deer tracks were caused by
a deer, and substitute for “E” the observed deer tracks, does JEI sanction my
believing the deer hypothesis? Or does the deer-track simulator hypothesis
constitute a better or equally good alternative explanation of the tracks? If
my believing the deer hypothesis does not satisfy JEI, but JEI is retained as
the relevant inference principle, then I should withhold judgment rather
than flat-out believe it.
What is involved in one hypothesis being explanatorily superior to an-
other? We shall briefly address this question here and return to it in greater
depth in Chapter 4 (Section 4.5). Here are a few principles that have won
some degree of acceptance among philosophers (Beebe 2009):

Explanatory Simplicity: Other things being equal, a theory that posits

fewer primitive explanatory notions should be preferred to one that
posits more.
Coherence with Background Knowledge: Other things being equal, a theory
that fits better with other widely accepted theories and background
knowledge should be preferred to a theory that fits less well.

01-Goldman-Chap01.indd 21 15/10/14 6:02 PM

 22 Chapter 1: The Structure of Justification

Explanatory Depth: Other things being equal, a theory that provides a more
illuminating explanation of the relevant data should be preferred to a
theory that provides a less illuminating explanation.
Avoidance of Ad Hoc Elements: Other things being equal, a theory that has
fewer ad hoc elements should be preferred to a theory that has more ad
hoc elements.

How might these ideas be applied to the problem of inferring proposi-

tions about the external world from propositions about our own inner
experiences? Most philosophers would say that the existence of genuinely
material things (chairs, rocks, buildings) is the best explanation, for each
of us, of our collection of mental experiences. But George Berkeley dis-
agreed. As a better explanation he offered the hypothesis of a nonmaterial
thing—­namely God (an infinite mind)—that causes our sequences of ex-
perience. He defended this partly by appeal to the greater simplicity of
the God hypothesis and partly by appeal to the principle of greater coher-
ence with background knowledge. We already know that minds can cause
experiences, but we don’t already know that material substances can
cause experiences. So it is unclear whether foundationalism can use an
inference-to-the-best-­explanation principle (whether JEI or another such
principle) to show how beliefs about the external world can justifiedly be
based on beliefs about our own mental experiences (the foundations).
We shall return to this question in Chapter 4.

QUESTIONS
1. The first section of the chapter draws a parallel between moral rightness and
wrongness, on the one hand, and epistemic rightness and wrongness on the
other. The latter is usually described in terms of being justified or unjustified,
rational or irrational in holding certain beliefs. At the same time it was said that
moral and epistemic normativity are different species of normativity. Can we
really hold both of these things? How can there be a strong commonality be-
tween moral and epistemic normativity if they are two different species of right-
ness and wrongness? How would you explain what is common to the two
realms and what is different?
2. Descartes seemed to regard the foundations of his belief system as beliefs he
could trust as the basis for other beliefs, on the analogy of what a house-
owner does in relying on the foundations of the house to support its higher
stories. Does this mean that the foundations must always be stronger than
the higher floors? How would this analogy play out in the realm of belief and
justification? Must a foundational, or “basic,” belief be better justified than
any nonbasic justified belief? Descartes said that his foundational belief was
“I think.” Is that better justified than any other (nonbasic) belief one might
have? What property does this belief have that sets it apart from nonbasic
justified beliefs and make its justification so strong? Does this property
(or properties) hold for all basically justified beliefs? What should a wise
foundationalist hold?

01-Goldman-Chap01.indd 22 15/10/14 6:02 PM

 Further Reading 23

3. Suppose you are a member of a jury that is trying a murder charge against a
defendant. As often happens in such cases (at least on television), much seems to
depend on whether the prosecution’s case “hangs together” better than the rival
story of what transpired presented by the defense. The first story, of course, in-
cludes a proposition about how the defendant did the deed. The second story
includes propositions about how the defendant was nowhere near the scene of
the crime. Aren’t you, as a juror, better justified in believing whichever story is
the more coherent one, the one that “hangs together” better? If this is right,
doesn’t it prove the truth of coherentism? How might a foundationalist reply to
this argument?
4. Sticking with the murder case example, suppose the prosecution contends that
its own story of what transpired is simpler than the story told by the defense and
it has fewer ad hoc elements. Suppose you agree with these claims. Is it clear that
this should convince you of the prosecution’s story? Why is a simpler story and
one with fewer ad hoc elements more worthy of belief? Either explain why it is
more belief-worthy, or indicate why it is a mistake to assign simplicity and mini-
mization of ad hoc elements such heavy weight in determining justification.
5. At the end of Section 1.4 it was objected that according to coherentism all mem-
bers of a highly coherent system will be justified and all members of a weakly
coherent system will be unjustified. In other words, for any system, all of its
member beliefs have the same justificational status. This is an objection because
it’s clear that this is not how justification usually works: Normally people have
some justified and some unjustified beliefs, despite having just one current
system. This appears to be a serious problem given the formulations of coherent-
ism presented in the chapter. But maybe a new formulation of coherentism could
avoid this problem. Can you suggest one?

FURTHER READING
Alston, William (1980). “Level Confusions in Epistemology.” Midwest Studies in
Philosophy 5, 135–150. Reprinted in Alston (1989). Epistemic Justification: Essays
in the Theory of Knowledge. Ithaca, NY: Cornell University Press.
BonJour, Laurence (1985). “The Elements of Coherentism.” Chapter 5 in The Structure
of Empirical Knowledge. Cambridge, MA: Harvard University Press.
Haack, Susan (1999). “The Foundherentist Theory of Epistemic Justification.” In
Louis Pojman (ed.), The Theory of Knowledge: Classical and Contemporary Readings
(2nd ed.). Belmont, CA: Wadsworth.
Huemer, Michael (2001). Skepticism and the Veil of Perception (pp. 98–118). Lanham,
MD: Rowman and Littlefield.
Klein, Peter (2005). “Infinitism Is the Solution to the Regress Problem.” In M. Steup
and E. Sosa (eds.), Contemporary Debates in Epistemology (pp. 131–140). Oxford:
Blackwell.
Lyons, Jack C. (2009). Perception and Basic Beliefs (Chapters 1, 2, and 4). New York:
Oxford University Press.
Pollock, John, and Joseph Cruz (1999). Contemporary Theories of Knowledge (2nd ed.,
Chapter 2, “Foundations Theories”). Lanham, MD: Rowman and Littlefield.
Williams, Michael (2001). Problems of Knowledge. New York: Oxford University Press.

01-Goldman-Chap01.indd 23 15/10/14 6:02 PM

 24 Chapter 1: The Structure of Justification

NOTES
1. Note that Alston’s official target in his paper is knowledge rather than justifica-
tion. But since the present chapter is devoted to justification rather than knowl-
edge, our exposition will be an adaptation of Alston’s discussion.
2. By contrast, Jack Lyons does specify a clearly stated criterion for immediate
justifiedness. See Lyons (2009).
3. We return to this topic in Chapter 6.

01-Goldman-Chap01.indd 24 15/10/14 6:02 PM

 2

Two Debates About Justification:

Evidentialism vs. Reliabilism
and Internalism vs. Externalism
Alvin I. Goldman

2.1 J us t i fi c at i o n a n d E v i d e n ce
As we have seen, foundationalism and coherentism are the leading theories
of the structure of justification. They also constitute the dominant players in
what might be called traditional epistemology. Although they disagree
sharply on the question of structure, they share a number of assumptions.
Much of this chapter focuses on contrasts between traditional justification
theories, mainly represented by mentalist evidentialism, and untraditional (or
less traditional) theories, mainly represented by process reliabilism. A lively
opposition between these two theories is revealed both by a head-to-head
debate between them and by an associated debate between the broader
­approaches they exemplify, internalism and externalism.
The term justifier refers to anything that helps make a belief state justified
or unjustified. In other words, it’s anything that contributes, positively or neg-
atively, to the justificational status (J-status) of a target belief. ­Foundationalism
and coherentism, even between them, offer only two types of justifiers. The
first type is (other) belief states of the subject, beliefs from which the target
belief can be inferred. These are the things to which we usually appeal when
asked or challenged to defend a specified belief. When I reply, “I believe P
because of X, Y, and Z,” the reasons cited are normally propositions I believe.
(Sometimes I also know them, but that implies belief in them.) A second type
of justifier is experiential states, such as perceptual and memory experiences.
(What is meant by a memory experience is a conscious seeming-to-­remember
episode.) As we saw in Chapter 1, coherentism restricts justifiers to beliefs
(or other doxastic states) exclusively. It holds, in effect, that all justifica-
tion takes place by inference, or inferential relations. By contrast, foundation-
alism allows justification to be conferred by experiences as well as by
(other) beliefs.
One thesis shared by traditional foundationalism and coherentism is that
all justifiers are mental states of one sort or another—more specifically,

25

02-Goldman-Chap02.indd 25 15/10/14 6:02 PM

 26 Chapter 2: Two Debates About Justification

mental states of the epistemic subject. Only a subject’s own mental states can
make a difference to whether the subject is justified or unjustified in holding
a given belief. Another point on which foundationalism and coherentism
concur is the assumption that only states subject S is in at t are ones that
affect the justificational status (J-status) of S’s belief in P at t. Consider Jane, a
25-year-old who believes today that when she was seven years old her
­feelings were badly hurt by her big sister making a snide and derogatory
comment. Jane has no memory experience today of those long-ago feelings,
or of the sound of her sister’s words being uttered. She just believes that this
incident occurred. Is Jane warranted in believing that the incident occurred?
In addressing this question traditional epistemology instructs us to ignore
any mental state or event Jane underwent when she was seven. Such past
states certainly matter to the truth of her memory belief (which isn’t here in
question), but they make no difference to the J-status of her current belief
(precisely the matter that is in question). According to traditional theories (in
particular, foundationalism and coherentism), only an agent’s own current
mental situation affects the justifiedness of one of her current beliefs. This
may be called the current-time-slice assumption of traditional epistemology.
(“Solipsism of the moment” is another label for this general type of view,
which has dominated modern philosophy.)
The idea harks back to Descartes and other philosophers of his period,
who conceptualized epistemology as the task of asking whether—and if so,
how—we can start from what we think and experience at any moment and
rationally reconstruct our entire corpus of belief from that position. The
“problematic” of epistemology, in other words, is to confront this question:
How can one legitimately infer outward (to the external world), forward
(to the future), and backward (to the past) from the indicated, very limited,
dataset? These are all the data we have to go on, according to the tradition, so
we’d better be able to show that these inferences are legitimate. Skepticism
must be answered by showing how all (or most) of one’s commonsense
­beliefs can be justified on the basis of one’s own current mental states.
Before proceeding further let us remind ourselves of an important distinc-
tion drawn in Chapter 1 between two types of justification: doxastic versus
propositional justification. Doxastic justification applies to beliefs actually
held by a subject, not merely a belief he or she could hold. In speaking of
doxastic justifiedness, what is evaluated is a (token) belief, not the subject or
the epistemic situation of the subject. By contrast, propositional justification
applies to a proposition, a subject, and his or her epistemic situation; it is
­applicable even if the subject has no belief in the specified proposition. To say
that a subject is propositionally justified with respect to P is to say (roughly)
that it would be appropriate for him or her to believe P (given his or her epis-
temic situation). Both concepts of justifiedness are widely used, but some
theorists find the doxastic sense of justification more important or congenial
and others find the propositional sense more congenial. The best theories try
to accommodate both senses of justifiedness but may give priority to one
or the other.

02-Goldman-Chap02.indd 26 15/10/14 6:02 PM

 2.1 Justification and Evidence 27

Let us now return to the kind of tradition being exposited earlier and
r­ eformulate it in terms of evidence, a concept widely used in epistemology.
There are different conceptions of evidence, but for present purposes we use
the term in a sense in which it refers exclusively to mental states. Earl Conee
and Richard Feldman (2004) defend a theory called evidentialism (or mentalist
evidentialism), which has a lot in common with traditional theories, especially
foundationalism. According to evidentialism, all (positive or negative) justi-
fiers of a belief held by epistemic agent S at time t are evidential states S is in
at t. Such states, as we have seen, are exhausted by experiences and beliefs (or
doxastic states more generally).
What must the relation be between S’s belief in P at t and S’s evidential
states at t in order for the belief in P to be justified? (Propositional justification
is the species of justification of central focus here.) The answer offered by evi-
dentialism is that belief is (would be) the appropriate or fitting attitude to
adopt given the evidence. In other words, there must be a relation of fitting-
ness that holds between the attitude of belief (directed at P) and S’s total evi-
dential states at t. Here we have an exceedingly simple (or simple-looking)
theory of justification, which uses only two concepts: evidence and fitting-
ness with the evidence.
Of course, the notion of fittingness is only apparently simple. On reflec-
tion, it proves to be fairly elusive, especially when pressed theoretically.
­Fundamentally, there are two kinds of fittingness: inferential and noninfer-
ential. Inferential fittingness holds between propositional states (like beliefs)
when there is a strong enough “support” relationship between the contents
of the evidential beliefs and the target hypothesis. If the evidential beliefs
logically entail the hypothesis, then the support relationship is presumably
strong enough to warrant belief in the hypothesis. But other kinds of support
also sometimes warrant belief—for example, support of an inductive or
­explanatory variety. An example of inductive support is this:

  E1: In the past whenever smoke emanated from a certain location, fire was
also present at that location.
E2: Smoke is now observed at location L.
 H: Fire is also present at L.

If E1 and E2 are the believed propositions and H is the hypothesis, H is

i­nductively supported by the conjunction of E1 and E2. Inference to the best
explanation, discussed in Section 1.9, provides another pattern of the sup-
port relation, at least in favorable cases. So we appear to have a reasonably
clear conception of what inferential fittingness consists in (ignoring prob-
lems and complications arising in these areas).
Noninferential fittingness is a more obscure and debatable relation. This is
the category to which fittingness with experience would belong. One question
here is whether experiences (e.g., visual experiences, auditory experiences,
being in pain, etc.) have propositional contents, or contents of any sort. If
they lack propositional contents (i.e., contents expressible by a declarative

02-Goldman-Chap02.indd 27 15/10/14 6:02 PM

 28 Chapter 2: Two Debates About Justification

sentence), how can they be “commensurable” with doxastic states (with prop-
ositional contents) so that the latter can fit or fail to fit the experience? And
what shall we say about the epistemic impact of different perceptual modali-
ties (e.g., vision and touch)? If vision and touch provide different “deliver-
ances” about the shape, texture, or identity of an object of interest, which
modality’s deliverance should be assigned greater evidential weight? What
conclusion “fits” the two perceptual deliverances taken together? Also, do
perceptual experiences have such qualities as clarity and vividness, and if so
how do these qualities affect the noninferential appropriateness of having a
belief in their contents as opposed to suspending judgment? Finally, exactly
which kinds of experience influence justifiedness? Is having a premonition
that something bad will happen a type of experience that makes it fitting
and proper (intellectually speaking) for one to believe that something bad
will happen?
All such problems must be addressed by an evidentialist theory that
makes fittingness a key concept. There is another important complication,
however. Although evidentialists often say that beliefs are evidence, this
cannot be strictly correct. Return to the smoke/fire example. Is it correct that
believing E1 and E2 provides evidence for H? Not necessarily. Belief per se
should not be considered an evidential state; only justified belief deserves to
count as evidence. If my beliefs in E1 and E2 are unjustified, they don’t give
me any justification for believing H. This complicates the evidentialist story
in important ways, as we shall see below.

2. 2 Pro b l e ms f o r E v i d e n t i a l ism
2.2.1 Is Evidential Co-presence Necessary and/or Sufficient?
Let us examine mentalist evidentialism by asking whether the conditions
it specifies are necessary and/or sufficient for justifiedness. Evidentialism
­implies that if a person S believes proposition P and this belief obtains in S’s
mind at the same time as he or she undergoes or instantiates a set of eviden-
tial mental states for which belief is the fitting response, then this is sufficient
for S to be justified in believing P. Is this correct?
Jack has worked hard for Congressional candidate Cindy, and it is now
election night and he is watching the returns. They are coming in fast from
many precincts. Some are rather encouraging if examined carefully. A cor-
rect tabulation of those numbers would support substantial confidence.
However, Jack is very excited, and although he sees these numbers, he makes
a sloppy calculation and comes up with a discouraging but faulty total. Jack’s
attention, meanwhile, is focused more on other precincts. Although their
numbers are scanty, Jack is sure they will come in massively for Cindy. It is
mere wishful thinking on his part, however, that makes him confident of
this. As a result of this wishful thinking, however, Jack thinks Cindy will
win the election. Is his belief justified?

02-Goldman-Chap02.indd 28 15/10/14 6:02 PM

 2.2 Problems for Evidentialism 29

According to evidentialism, it appears that Jack’s belief is justified, because

the numbers he saw from the first group of precincts did support belief in
her ultimate victory. On the other hand, he didn’t pay much attention to
them. And the numbers from the second batch of precincts are scanty, and
prove nothing. But that should not matter under evidentialism, because the
first batch of numbers constitutes adequate evidence to support confidence
in Cindy’s victory. Thus, Jack’s belief in victory meets evidentialism’s suffi-
cient conditions for justifiedness. Intuitively, though, this verdict is wrong.
Jack’s belief is not justified. (At least in the doxastic sense of justifiedness, his
belief is not justified.) It is based on sloppy calculation and wishful thinking.
Surely something has gone awry.
What has gone wrong, it appears, is that the mere co-presence of a belief
with evidential states that it fits is not enough for its justifiedness. It is crucial
that appropriate evidential states actually cause the belief via suitable think-
ing. This does not happen in the election returns example. Jack occupies an
evidential position that might potentially lead to a justified belief, but he has
misused this evidential position, so his confidence in Cindy’s victory is the
product of highly flawed thought processes. Evidentialism falters here by
failing to incorporate a suitable causation condition. It tries to make do with
co-presence, but this does not work.
Evidentialism has tried to accommodate this problem by introducing
­another concept in the vicinity of justification called well-foundedness and
by including something like a causation condition as necessary for well-­
foundedness. However, the specified condition is not exactly causation. The
proposed condition is that the belief be “based on” appropriate evidence,
where the basing relation is left unanalyzed. Unless a causal condition is in-
cluded, however (whether or not it is embedded in the basing relation), it
looks like the election case cannot be handled appropriately. This is already
an important shift from the tradition in which it used to be claimed that
mental causation and “discovery” are purely psychological matters that have
no bearing whatever on justification and no place in a theory of justifiedness
(Reichenbach 1938). Causal theories of justification, of the sort we shall exam-
ine shortly, run contrary to this tradition (Goldman 1967).
Turn next to necessity: Is it a necessary condition of a belief’s being justified
that it fit some set of evidential states co-present with it? Here is a counterex-
ample. Nora’s friend Amy asks her: “What is the circumference of the Earth?”
Nora replies: “It is approximately 25,000 miles.” Amy continues, “Really?
What makes you think that? What is your evidence for it?” Nora replies:
“I don’t know what my evidence is. I can’t think of any evidence right now.
It’s just one of those facts that I know, or at least believe.” Suppose, indeed,
that Nora has no recollection of reading it anywhere, or hearing somebody
say it. In short, Nora no longer retains any specific evidence about the Earth’s
circumference, although she did once learn it from an authoritative source.
As far as her current mental states are concerned, the only relevant state is
the belief itself. So she has neither inferential (belief-based) evidence for

02-Goldman-Chap02.indd 29 15/10/14 6:02 PM

 30 Chapter 2: Two Debates About Justification

the proposition nor noninferential (experiential) evidence for it. Nonetheless,

it is perfectly plausible that her belief in this proposition is justified. If she
learned it originally from an authoritative source and retained it ever since,
she is still justified in holding it. If this is right, the current possession of evi-
dence that a belief fits is not a necessary condition for the belief’s being justi-
fied. (Do you share this intuition? If not, more argument would be needed.)
A defender of evidentialism might reply as follows. Even if Nora has no
current justified belief about the source of her belief in the circumference
proposition, she must surely have a justified current belief that her memory
beliefs are usually reliable. Isn’t this a current evidential belief that helps sup-
port her circumference belief, since it is a belief of hers? Perhaps the italicized
belief represents the needed evidence. However, it does not seem necessary
for Nora’s belief in the circumference proposition to be justified that she
should have such a belief about her memory’s reliability. Maybe she just isn’t
reflective in this sort of way, or isn’t being reflective about it at the moment.
Even if she possesses this general belief about her memory, which could be
elicited if she were prompted, it plays no causal role in her current belief.
Thus, it is arbitrary to insist that this is what confers justifiedness on the
belief. If the assumed story is that Nora did read some authoritative source to
the effect that the Earth’s circumference is 25,000 miles but later forgot the
identity of that source, it might be reasonable to conclude that this past piece
of evidence acquisition played a crucial missing role in Nora’s being justified
later on. But this does nothing to show that she needs co-present evidential
states to have a justified belief now.
Next we consider an argument for saying that some justified beliefs are not
accompanied by any (appropriate) evidential states whatever. Almost all epis-
temologists agree that people have justified beliefs about their current mental
states; for example, “I now have a headache” or “I think that my social secu-
rity number is such-and-such.” It is also generally agreed that such ­beliefs
enjoy noninferential rather than inferential justification. So, if all noninferen-
tial evidence (about contingent propositions, at any rate) consists in experien-
tial states, the present cases would also have to involve experiential evidence.
Finally, a fairly standard view is that beliefs of the foregoing kind are justified
because one introspects the mental states that the beliefs ­describe. Does intro-
spection involve experiential evidence? This is dubious. Although some phi-
losophers view introspection as a quasi-perceptual process (an “inner sense”),
it surely differs from standard perceptual processes in lacking any associated
sensory “feel.” Introspection involves no distinctive qualitative character
analogous to the qualitative character associated with vision or hearing.
This suggests that no experiential evidence is present in introspection.
What, then, is the experiential state that renders an introspection-based
belief justified?
Evidentialists might retort: “It is the target mental state itself. In the fore-
going examples this would be the headache itself, or the judgment that my
social security number is such-and-such.” One objection to this is that it
seems improper for a target state to be evidence for itself. The very idea of a

02-Goldman-Chap02.indd 30 15/10/14 6:02 PM

 2.2 Problems for Evidentialism 31

state’s conferring evidence on itself is highly suspect. As we saw in ­Chapter 1,

Chisholm proposed to deal with the problem of ending evidential ­regresses
by quipping “what justifies me in thinking I feel sad is that I do feel sad.” This
is clever, but Chisholm offers no explanation of when or why it should work.
Can we equally say “what justifies me in thinking that the fusiform gyrus of
my right occipital cortex [a certain brain region] is being activated is that the
fusiform gyrus of my right occipital cortex is being activated”? Surely not.
This certainly does not work for random brain states, if any. An evidentialist
needs to explain why not. How do brain states differ from feelings of sadness
so that the formula “works” for the latter but not for the former? Moreover,
in the social-security-number belief case the target state is not an experien-
tial one, so how can it provide experiential evidence in the way that percep-
tual or memory states do?
Finally, this subsection turns to a problem of explanation rather than a
counterexample. How can evidentialism explain why certain experiences
confer evidence while others do not? Philosophical theories aim to explain
things, especially to illuminate underlying reasons that lie behind intuitively
familiar data. Here the thing to be explained is an intuitive datum men-
tioned earlier: Although premonitions are experiential states, they do not
confer justification. If one has a premonition that P, this lends no legitimate
evidential support to P. Why is this so? It is hard to find an answer in eviden-
tialism’s toolkit. By contrast, a competing theory we shall soon examine has
a simple and straightforward answer.

2.2.2 What Is Meant by “Evidence”?

When we engage in philosophical theorizing, we want to be careful about
our theoretical terms. A theoretical term is often open to different definitions
or interpretations, and the choice of definition can be quite important. One
definition might suit a given theoretical position well and another might suit
it badly. So it is always appropriate to ask proponents of a philosophical
theory to specify the meanings they attach to their terms.
These points apply to evidentialism. It is obviously important, in trying to
assess the merits of this theory, to get straight about the intended meaning of
“evidence.” Although evidentialists offer clear examples of what they con-
sider evidence, they do not provide any definition. Definitions of “evidence”
have been offered by other philosophers, however, and it is instructive to see
if any of these could work for evidentialism. Here is one proffered definition:

X is evidence for Y just in case possessing X enhances a person’s justification for

believing Y.

Notice that this definition uses “justification” in the analysans (the clause
that does the analyzing). There is nothing wrong with this in the abstract, but
here we are discussing a theoretical approach that seeks to analyze justifica-
tion in terms of evidence. So it would be circular to turn around and define

02-Goldman-Chap02.indd 31 15/10/14 6:02 PM

 32 Chapter 2: Two Debates About Justification

evidence in terms of justification. It is OK to define either term by means of

the other but not to do both at once. That engulfs one in circularity.
What is so bad about a circular pair of definitions? Suppose I define wizard
as someone who produces magical events, and then define magical as events
produced by a wizard. If I tried to figure out whether there are any wizards
or magical events, this pair of definitions would be singularly unhelpful.
Spotting events I suspect to be magical, the second definition tells me that
I need to determine whether any of them is caused by a wizard. When I try
to determine of any individual whether he or she is a wizard, the first defini-
tion instructs me to determine whether the individual produces any magic.
Progress seems impossible.
Let us try a different definition of evidence. It is a matter of common
­observation that lightning is evidence of impending thunder. To call light-
ning “evidence” here seems to mean that lightning is a reliable indicator of
thunder, where a reliable indicator is a sign that invariably, or regularly, gets
things right. What is meant by “gets things right” is, roughly, “tells the truth.”
Thus, this sense of “evidence” can also be captured by the phrase “reliable
mark of the truth.”
Can evidentialists adopt this as their preferred sense of “evidence”?
­Accepting this definition would imply that a mental state (e.g., a perceptual
experience) qualifies as an evidential state only if it reliably indicates the
truth—presumably, the truth about the external world. But given the posited
relationship between justification and truth, beliefs will get to be justified at
least partly in virtue of states having truth-indicating properties, so truth-
indicatorship would become relevant to the J-status of beliefs. It would follow
from this that truth-indicatorship is a “justifier,” something that helps deter-
mine the J-status of beliefs. But truth-indicatorship is not a mental state, so in
the end, the evidentialist’s core claim that only mental states are justifi-
ers would be undercut. Thus, evidentialists cannot use this definition of
“­evidence.” This is a fundamental theoretical problem for evidentialism:
How can they define their central theoretical term so as to avoid circularity
but not undercut core features of their theory?

2.3 J us t i fi c at i o n a n d Pro ce ss Re l i a b i l i t y
Let us make a new start toward a theory of justification. The example of Jack
the election worker illustrates that the J-status of a belief depends on how it
is formed, or caused. The fact that Jack’s confidence in his candidate’s victory
is caused by wishful thinking is clearly what leads us to assess his belief as
unjustified. Evidentialism tries to accommodate this point by saying that a
belief’s being justified (“well-founded,” in their terminology) requires not
only that the belief fit the subject’s evidence but also that it be based on this
evidence. Does this cover the point adequately? It handles the case of Jack
because although he possesses some evidence (i.e., the favorable numbers
he substantially ignores) that his victory belief fits, it isn’t based on that
­evidence. So far, so good.

02-Goldman-Chap02.indd 32 15/10/14 6:02 PM

 2.3 Justification and Process Reliability 33

Now consider another case. A simple rule of deductive logic called “dis-
junctive syllogism” licenses the following pattern of inference as valid:

Either P or Q.
Not-Q.
Therefore, P.

Suppose that Chad justifiably believes a proposition of the form “P or Q” and

justifiably believes “not-Q.” Then he has first-rate evidence for believing P;
indeed, his evidence logically entails P. Hence, if he forms a belief in P based
on this evidence, mustn’t this belief of Chad be justified? Is that necessarily
the case? No.
Suppose Chad is psychologically “wired” to deploy the following reason-
ing process under many circumstances. When he believes two premises of
the form “P % Q” and “not-Q,” where the symbol “%” is replaceable by any
binary truth-functional connective (disjunction, conjunction, material condi-
tional, etc.), then Chad infers a conclusion of the form “P.” In other words,
Chad employs a reasoning principle that resembles disjunctive syllogism but
greatly (and ludicrously) overextends it. If deployment of this overextended
principle is the process by which Chad arrives at his belief in P in the forego-
ing example, then his belief is definitely not justified. It follows from this that
it is not a sufficient condition for a belief to be justified that it both fits one’s
evidence and is based on that evidence (Goldman 2012, 7).
As this and previous cases indicate, it is quite crucial to the J-status of a
belief how it is causally produced. It is not just a matter of the inputs to the
causal process (e.g., their being suitable evidential states). It is crucial that the
belief-forming process used be of an appropriate, or suitable, kind. Which
belief-forming processes are suitable and which are unsuitable?
In trying to answer this question, let’s review some of the examples al-
ready covered. Forming beliefs by wishful thinking is clearly unsuitable.
Forming beliefs by using invalid reasoning processes like Chad’s overgener-
alized inference schema is also unsuitable. Beliefs formed by these kinds of
processes are always unjustified, no matter what prior mental states (or evi-
dence) they take as inputs. What are examples of suitable belief-forming pro-
cesses? Presumably, ordinary perceptual processes are suitable. We normally
judge the outputs of such processes (perceptual beliefs about the external
world) to be justified. Similarly, we tacitly assume that introspection is a suit-
able belief-forming process. Finally, using valid inference schemas to guide
our reasoning is surely justification-preserving. If one starts with justified
beliefs in the premises and applies a valid reasoning process to it, the output
of the reasoning process will also be justified.
What do suitable (“good”) belief-forming processes have in common that
makes them all good (i.e., justification-conferring)? What do unsuitable
(“bad”) belief-forming processes have in common that makes them all bad
(i.e., unjustification-conferring)? The answers seem to leap out. The good
processes are highly reliable processes: A high proportion of the beliefs they

02-Goldman-Chap02.indd 33 15/10/14 6:02 PM

 34 Chapter 2: Two Debates About Justification

generate are true. The bad processes are unreliable processes, or at least not
very reliable ones: The proportion of true beliefs they generate is not very
high, some of them less than 50 percent. Thus, wishful thinking is bad
­because the beliefs it outputs are only occasionally true. The overextended
version of disjunctive syllogism is bad because it would frequently lead to
erroneous conclusions even when its inputs are true. By contrast, forming
beliefs by standard perceptual processes (in favorable viewing conditions) is
generally reliable. And using disjunctive syllogism (in the strict sense) will
always yield true conclusion-beliefs if the premise-beliefs are true. It’s a
good-making feature of a belief-forming process that it tends to produce a
high ratio of true beliefs and a bad-making feature of a belief-forming pro-
cess that it tends to produce a low ratio of true beliefs. Very plausibly, this is
why belief-outputs of the former type of processes are justified and belief-
outputs of the latter type of processes are unjustified. A theory of justifica-
tion incorporating these core ideas is naturally dubbed process reliabilism
(Goldman 1979).

2.4 F o r m u l at i n g Pro ce ss Re l i a b i l ism

How should such a theory be formulated? For starters, consider the ­following
principle:

Belief B is justified if and only B is produced by a reliable belief-forming

­process—that is, a process that has a tendency to generate (belief) outputs with
a high percentage of truths.

What is a high percentage? Eighty percent? Ninety percent? The theory can
afford to be vague in this matter. After all, the concept (or property) being
analyzed, justification, is itself a vague concept, so it is appropriate for its
analysis to be correspondingly vague.
However, there is another problem. Some belief-forming processes are rea-
soning processes. These processes operate on prior beliefs of the subject
(premise beliefs), which themselves can be either justified or unjustified and
either true or false. Focus on their truth-values. Where the input beliefs are
true, a “perfect” reasoning process (i.e., a deductively valid process) would
have all of its output beliefs true as well. And we might expect inductively
strong processes to have most of their output beliefs be true. But we cannot
expect this for the class of cases in which the input beliefs include falsehoods.
Even a first-rate inference process can’t be expected to produce true outputs
when applied to false inputs. As the saying goes, “garbage in, ­garbage out.”
So let’s distinguish between processes that take prior beliefs as inputs and
processes that don’t. Processes of the latter type should be required to be
unconditionally reliable. Processes of the former type, on the other hand,
should only be required to be conditionally reliable—that is, to have a high
proportion of true outputs across the range of cases in which their belief

02-Goldman-Chap02.indd 34 15/10/14 6:02 PM

 2.4 Formulating Process Reliabilism 35

inputs are all true. Once this distinction is drawn, we can advance two prin-
ciples of justification (doxastic justification is the main species of justification
at issue here):

(UR) A belief B is justified if it is produced by a belief-forming process that is

unconditionally reliable.
(CR) A belief B is justified if (i) B is produced by a belief-forming (or belief-­
retaining) process that is conditionally reliable, and (ii) all of the belief inputs to
the belief-forming (or belief-retaining) process that causes B are justified.

Notice that (CR) does not specify that all the input beliefs must be true; it
only requires that they all be justified. This guarantees, however, that the
input beliefs must have been produced by prior uses of “good” processes too,
even if, as it happened, those good processes did not (all) produce truths on
those occasions. The rationale here is that one cannot expect epistemic agents
to do more than use good epistemic procedures. The idea behind reliabilism
is that good epistemic procedures are either unconditionally reliable or con-
ditionally reliable.
It follows from (UR) and (CR) that a person can arrive at a justified belief
by a series of belief-forming or belief-retaining processes that take place over
a considerable period of time. People routinely make inferences scattered
over a period of time, each new inference drawing on old beliefs that may
have been formed quite a while earlier. Memory plays a fundamental role in
such a series—not memory in the sense of conscious episodes but memory
in the sense of a preservative or retentive process that takes belief states at
one point in time and maintains them, or “carries them over,” to a later time.
(Alternatively, an imperfect memory process may result in belief modifica-
tion over time.) With this in mind, (UC) and (CR) can be fused into a single
reliabilist principle of justification:

(R) A belief B (at time t) is justified if and only if B (at t) is the output of a series
of belief-forming or belief-retaining processes each of which is either uncondi-
tionally or conditionally reliable, and where the conditionally reliable processes
in the series are applied to outputs of previous members of the series.

A simple illustration of this principle is as follows. Suppose Mary believes

that her favorite Hollywood couple is getting divorced. She acquired this
belief from reading a Hollywood magazine, Gossip, which reported that
­Kenneth and Gwyneth had filed for divorce. She also believes that Gossip is
very trustworthy in such matters, which she had established previously.
Mary first used perceptual processes to form the belief that Gossip reports an
impending divorce (=D) and from this she inferred the truth of D. Each of
these cognitive steps involved a reliable process (unconditional in the per-
ceptual step and conditional in the inferential step). Her belief in D was then
stored in memory and retained for at least a week. Memory storage, let us
assume, is a conditionally reliable belief-retaining process; that is, its later

02-Goldman-Chap02.indd 35 15/10/14 6:02 PM

 36 Chapter 2: Two Debates About Justification

outputs are usually true if the earlier inputs to it were true. Then, by princi-
ple (R), Mary’s preserved belief in D is justified. Suppose that Mary no longer
remembers a week later that her belief in D was acquired by reading Gossip.
At the later time of belief, she no longer possesses any such evidential source
that inferentially supports D. The reliabilist principle (R), however, does not
require co-present evidence. According to (R) the history, or etiology, of a
person’s belief in a proposition is also relevant to its J-status. Given that Mary
originally formed her belief in D based on good evidence, and nothing has
transpired to undercut or defeat that evidence, (R) is perfectly consistent
with Mary’s being justified at the later time in believing D. This seems like
the intuitively right result. Unlike evidentialism, then, process reliabilism
delivers a correct verdict about Mary. (At least so it strikes the present author.)
Notice that process reliabilism marks a substantial departure from earlier
theories we considered (foundationalism, coherentism, and evidentialism) in
introducing a novel category of justifiers, viz. belief-forming processes.
Under process reliabilism, psychological processes used in acquiring and/or
retaining beliefs are critically relevant to their J-status. Thus, belief-forming
and belief-retaining processes are justifiers, or J-factors. They are not eviden-
tial justifiers; they are not pieces of evidence, or reasons, to which a subject
would appeal to defend his or her beliefs. (They just “operate” on such evi-
dence or reasons to produce beliefs or other doxastic attitudes.) Nonetheless,
they are relevant to the J-statuses of the beliefs they produce, which is what
it takes to qualify as justifiers.
Process reliabilism is mainly a theory of what makes beliefs justified or
unjustified—beliefs that are actually held. Thus, it is primarily a theory of
doxastic justifiedness. However, we can also ask a question about proposi-
tional justifiedness. We can ask whether a person is justified to believe P
even when he or she does not in fact believe it. A visual experience someone
undergoes while viewing a mountain landscape may entitle him or her to
believe that there are three peaks in the distance, although he or she hasn’t
paid attention to the number of peaks or formed any belief on that question.
We might nevertheless say that he or she is justified (or warranted) to believe
that there are three. Since a proposition that is not believed has no causal his-
tory, however, principle (R) cannot be used directly to assess its J-status for S.
What needs to be asked, therefore, is whether the subject in question is in
mental states that could be used as inputs to a reliable process and, if so used,
would output a belief in the designated proposition. That’s a rough sketch of
how process reliabilism can be adapted to handle propositional justification
(Goldman 1979).

2. 5 Re l i a b i l ism ’s At t r ac t i o ns a n d Pro b l e ms
Process reliabilism has many attractions as a theory of justification, some of
which have been noted. It also faces a number of significant problems. Let us
survey some of its chief advantages and most prominent problems, and ask
how reliabilism might address the latter.

02-Goldman-Chap02.indd 36 15/10/14 6:02 PM

 2.5 Reliabilism’s Attractions and Problems 37

If the preceding discussion was right, reliabilism delivers accurate ­verdicts

about J-statuses in a large number of cases. In addition, it has two attractive
features that merit emphasis. First, it goes a long way toward reducing threats of
skepticism. Epistemologists generally welcome effective rejoinders to skeptics.
(At issue here is justification skepticism. Reliabilism can also be formulated,
with important additions, as a theory of knowledge, as we’ll see in Chapter 3.
Reliabilism is congenial to several leading responses to ­knowledge-skepticism
as we’ll see in Chapter 4. Section 1.6 posed a traditional skeptical challenge. It
argued that external-world beliefs can be justified only if they offer better expla-
nations of everyday experience than competing ­hypotheses such as Cartesian
demons or envatted brains. Indeed, many skeptics argue that a subject must be
able to exclude such competitors on explanatory grounds. These stiff require-
ments demanded by skeptics are hard to meet. But is it appropriate to impose
them? Reliabilism proposes a milder requirement for justification, namely the
reliability or truth-­conduciveness of the belief processes used. This condition is
much more easily satisfied than the hyperdemanding conditions often pro-
posed by skeptics.
A second widely acknowledged desideratum for a theory of justification is
that it underwrite a close link between justification and truth. Contemporary
epistemologists admit that a belief’s being justified rarely guarantees the
truth of its content. Still, a belief’s being justified should make its truth highly
probable. For many theories, however, it is difficult to see how this result can
be secured. BonJour, an erstwhile coherentist, explicitly accepts the truth-
link desideratum, and sets out to show why his own theory has this conse-
quence (1985). Unfortunately, it is very unclear how coherentism guarantees
this. By contrast, reliabilism seems to secure it easily. If a belief is caused by
reliable processes, doesn’t this essentially guarantee a substantial probabil-
ity of the belief’s being true?
As a first line of criticism, some critics turn the tables on reliabilism by
disputing the adequacy of its response to skepticism. They argue that it sets
too weak conditions for justification, allowing warrant to be attained on the
cheap. Serious skeptical worries are just dodged, not really met. One such
line of criticism starts by saying that being caused by a reliable process is not
enough for justification. The subject must also meet a “meta-justification”
requirement: he or she must be justified in believing that his or her belief is
reliably caused. Furthermore, such critics continue, this meta-justification
condition cannot be met. In particular, it cannot be met without appealing to
the self-same element of reliability that is under challenge—and this would
be an unsatisfactory solution.
Reliabilists are not lacking in responses here. Reliabilism about justifica-
tion is partly motivated by the idea that we need an analysis of justification
in nonjustificational and nonepistemic terms. Causal reliability can meet this
desideratum because it makes no appeal to justificational notions. So if we
have to honor a meta-justification requirement, it might take the form of a
requirement that the subject must believe reliably that his or her first-order
belief is reliably caused. But why, exactly, would the satisfaction of this

02-Goldman-Chap02.indd 37 15/10/14 6:02 PM

 38 Chapter 2: Two Debates About Justification

higher-level requirement guarantee justifiedness if satisfaction of a first-­

order reliable belief formation fails to do so? If “bare” reliability of first-order
belief cannot do the job, why would a single higher-order element of meta-
reliability succeed? Why not require an infinite series of higher-order reli-
ably formed beliefs? Yet this would clearly be excessive. So we should reject
the demand for a meta-level requirement at the very start. A related condi-
tion, however, may well be appropriate, namely the much weaker require-
ment that the subject not believe (or not reliably believe) that his or her
first-order belief is unreliably caused. This “negative” condition, simply an
anti-defeat condition, can certainly be incorporated into reliabilism without
adverse consequence.
A second line of skepticism-related criticism also concerns the adequacy of
the reliable-causation condition. According to one critic, this feature of relia-
bilism guarantees only the possibility that humans meet the justification
­requirement for their ordinary beliefs. Guaranteeing this mere possibility,
however, should not satisfy a skeptic (Wright 2007, 31). A skeptic wants to be
given something stronger than this. He or she wants to see that it is not
merely possible for people to be justified in their everyday beliefs; he or she
wants to be shown they are so justified. At any rate, the skeptic wants to be
shown that they are justified in believing (if they are reflective enough) that
they are so justified. And he or she wants to challenge the very possibility
of this. But even Wright concedes that the skeptic is in no position to assert
the impossibility of our having such second-order justifiedness when it is
­characterized in reliabilist terms. The skeptic cannot preclude the possibility
that we should have reliably caused beliefs to the effect that our ordinary
beliefs are reliably caused. What an internalism-driven skeptic wants to
claim is that mere higher-order reliability does not adequately express and
resolve his or her dissatisfaction. However, Wright seems to concede that it
isn’t ­obvious how the skeptic’s dissatisfaction should be expressed. Wright
offers no formulation of the dissatisfaction (in internalist terms) that he him-
self finds satisfactory.
A third line of criticism has a different focus. It casts doubt on the idea that
token belief-forming processes can be assigned unambiguous degrees of re-
liability, an assumption that seems fundamental to process reliabilism. With-
out such assignments, the question of whether a given belief is caused by a
sufficiently reliable process (wherever the threshold for reliability is set)
cannot be resolved. The ambiguity arises from the fact that any token belief-
forming process can be “typed” in many different ways. For example, the
token process in which George arrives at the belief that he has won a lottery
by reading this in The New York Times might be an instance both of the type
forming a belief by reading The New York Times and of the type forming a belief by
reading. The first type might have a very high reliability, enough to meet the
requisite standard of justifiedness. The second type, by contrast, may not be
sufficiently reliable to qualify for justifiedness. But which type should be
used in fixing a token belief’s justifiedness? Reliabilism hasn’t supplied
any (agreed-upon) formula for choosing a process type at the “right” level of

02-Goldman-Chap02.indd 38 15/10/14 6:02 PM

 2.5 Reliabilism’s Attractions and Problems 39

generality. This is called the generality problem. One response by reliabilists is

to point out that the generality problem, though genuine, is not peculiar to
reliabilism. According to this view, in some form it is every epistemology’s
problem, not one peculiar to reliabilism.1 This response can be supported by
pointing out that every epistemology must acknowledge that a belief is (dox-
astically) justified only if it stands in the basing relation to states of evidence-
possession or reasons-possession that the epistemic subject is in. But the
basing relation involves some sort of causal process, and any instantiation of
a causal process must be suitably typed. This introduces the generality prob-
lem in (roughly) the same way it arises for process reliabilism.
Many attempts to solve the generality problem have been tried. A novel
approach emerges if one considers a general approach to justification (and
other philosophical topics) that focuses heavily on the attributor, a maneuver
we will consider below. For now let us postpone discussion of the general-
ity problem until we have placed some methodological considerations on
the table.
Our fourth problem for reliabilism was, historically, one of the first that was
raised. It is called the new evil-demon problem. Descartes entertained the pos-
sibility of an evil demon that systematically deceived him, causing even his
perceptual experiences to mislead him about his surroundings. Contemporary
critics of reliabilism use the evil-demon hypothesis to pose a counterexample
to reliabilism. Since the character in the evil-demon world is systematically
deceived by his perceptual experiences, his perceptual belief-forming pro-
cesses must be unreliable. Hence, according to reliabilism (it seems), he isn’t
justified in believing their outputs. But, these critics protest, such beliefs should
intuitively be classified as justified, presumably because agents in an evil-­
demon world have the same experiences as people in the actual, normal world.
Ostensibly, this is a major flaw in reliabilism.
At this point it may be wise to take a step back and assess the methodology
we are using (not just in this immediate context, but more broadly). In assess-
ing the merits or demerits of each theory, the evidence to which we appeal is
to ask what judgments, verdicts, evaluations, or attributions people make, or
are inclined to make, about the target belief’s epistemic status. In so doing,
however, a theorist doesn’t directly detect the belief’s epistemic status.
It cannot be “read off” from the case description whether the belief held by a
victim of the demon is justified or unjustified. The theorist’s “immediate”
evidence is what he or she herself intuitively judges and what other people
who consider the same question judge. Do they attribute justifiedness or un-
justifiedness to the victim? Or what are they inclined to attribute? In effect,
the data to which we appeal are data about justification attributions. It is
natural to suggest, then, that what we should be seeking is a theory that
­explains, or accounts for, these attributions made by people who are given
the cases in question. Theories of this sort need not focus exclusively on
the epistemic subject whose belief justifiedness is in question; they might
equally focus on the attributor.2 Let us see how this strategy plays out in the
evil-demon case.

02-Goldman-Chap02.indd 39 15/10/14 6:02 PM

 40 Chapter 2: Two Debates About Justification

Here is a plausible-looking story (though admittedly a speculative one).

The epistemic properties people associate with various belief-forming pro-
cesses are likely to be properties that those processes display in ordinary,
actual-world settings. As encountered in the actual world, perceptual pro-
cesses are by and large impressively reliable over a fairly wide range of cir-
cumstances. Thus, people might naturally think of common perceptual
process types (e.g., seeing, hearing, feeling) as good ways of forming ­beliefs—
good in the sense that they usually deliver the truth (which is what people
seek for the most part). Other belief-forming process types, such as forming
a belief by hunch or guesswork, are presumably thought to be bad ways of
forming beliefs, bad because of their error-proneness. Now if these people
are presented, in conversation, with descriptions of imagined scenarios in
which the protagonist’s situation is highly unusual (e.g., an evil-demon type
of situation), they might not evaluate the protagonist’s performance in terms
of what regularities prevail in his or her unusual world. Rather than aban-
don the classifications they have stored up based on actual-world experience,
they might continue to apply them even to the bizarre hypothetical world.
They might classify perceptually formed beliefs as justified and beliefs
formed by guesswork as unjustified, even if these processes are described
as having reversed reliabilities in the imagined scenarios. This pattern of
justification attribution might be captured by a “two-stage” theory of justifi-
cation attribution. In the first stage the potential attributor forms appraisals
of ­belief-forming process types in terms of their observed (or inferred) reli-
ability in the actual world. In the second stage the potential attributor is in-
vited to make attributions of justifiedness or unjustifiedness about real or
imagined epistemic subjects. The theory we are developing conjectures that
people would make this choice by relying heavily on their mentally stored
lists of approved and disapproved process types (Goldman 1992). This
­account would readily explain the pattern of people’s attributions in the new
evil-demon case. In particular, it predicts that attributors would give a “justi-
fied” verdict for the victim’s belief because perceptual belief-forming types
will be on attributors’ approved lists.3
Approved-list reliabilism might also help us understand what transpires
in another style of (putative) counterexample to reliabilism. Laurence ­BonJour
advanced the example of a reliable “clairvoyant” subject, Norman, who
forms beliefs—he knows not how—about the current whereabouts of the
President, despite having no perceptual experiences of seeing the President
at the time (or any other standard bit of evidence) (1985, 38–45). As it hap-
pens, Norman in fact possesses an extremely reliable clairvoyant power, and
this power is what generates his belief about the President’s whereabouts.
Norman just suddenly finds himself believing that the President is in
New York City. He has no evidence or reasons for thinking that he has any
special cognitive power for remote detection—but he does have such a
power. ­According to the original formulation of reliabilism, the theory seems
committed to saying that Norman’s clairvoyant belief is justified, but most
epistemologists find this counterintuitive.

02-Goldman-Chap02.indd 40 15/10/14 6:02 PM

 2.5 Reliabilism’s Attractions and Problems 41

The question now arises whether the new form of reliabilism under con-
sideration here (approved-list reliabilism) also predicts the “wrong” verdict
(i.e., “justified”). The answer seems to be no: Clairvoyance would presum-
ably not be on the approved list of normal attributors. Indeed, both clairvoy-
ance and other mysterious and suspect powers would probably be on a
normal attributor’s disapproved list (along with extrasensory perception,
telekinesis, etc.). This would explain the tendency of attributors to deny that
Norman’s belief is justified.4
Finally, let us return to the generality problem to see what light might be
shed on it by the attributional focus. Rephrased for this approach, the princi-
pal question for approved-list reliabilism is whether attributors judge a belief
to be justified when and only when they also judge it has been produced by
a reliable process. As Erik Olsson (forthcoming) suggests, this is a question
one might test for “experimentally,” using the following method. One might
first select two groups of subjects and confront them with, say, twenty epi-
sodes of ordinary life involving a person coming to believe something. The
episodes could be presented as film sequences. Each subject in the first group
is asked to state independently for each episode whether the character in that
episode is justified in his or her belief. Similarly, each subject in the second
group is asked to state independently for each episode whether the character
in that episode acquired his or her belief in a reliable way. Suppose the ex-
periment yielded the result that there was a high correlation between the
reports in the first group and the reports in the second group. In other words,
more or less the same episodes that were classified as justified by all or
almost all members of the first group were also classified as having been
acquired in a reliable way by all or almost all members of the second group.
This would be clear support, at least prima facie support, for the reliabilist
theory (at least in its original, one-stage version). This begins to give us an
idea of how reliabilism might be tested, and it isn’t obvious, argues Olsson,
that it would be refuted. We shall return to this topic with further develop-
ments in Chapter 7.
A fifth problem for process reliabilism focuses on a theory of epistemic
value to which it is allegedly tied. According to these critics, however, alle-
giance to this theory of epistemic value creates a serious problem for reliabi-
lism’s verdicts about justification.5 The theory of value in question is epistemic
teleology, or epistemic consequentialism, which says that the possession
of true beliefs and the avoidance of false beliefs are valuable ends in them-
selves. According to the critics, reliabilism about justification is a theory of
when a belief is “right” or “appropriate” very similar to the ethical theory of
utilitarianism, which views an action’s moral status as a function of the
causal effects of that action. Applied to the epistemic case, this is said to yield
intuitively incorrect verdicts of justifiedness. Consider a hypothetical case in
which you are a longstanding atheist, persuaded of this view by careful
intellectual reflection. Now you have an opportunity to be awarded a
­
­research grant offered by a religious organization, but winning this grant is
contingent on your adopting a belief in God. By a kind of surrender of the

02-Goldman-Chap02.indd 41 15/10/14 6:02 PM

 42 Chapter 2: Two Debates About Justification

intellect to desire, your keen desire to win the grant changes your belief
around so that you now believe in God. Finally, suppose that you win the
grant, and the research it supports generates an abundance of widely ­believed
scientific truths. Does this show that your belief in God is justified? Certainly
not, say the critics; but it looks like reliabilism is committed to this palpably
incorrect verdict.
Clearly the critics say that the belief is not justified. But does process
­reliabilism imply otherwise, as the critics imply? No. As clearly indicated by
the principles stated in Section 2.4, process reliabilism is a historical theory of
belief justifiedness, which implies that only processes upstream of a target
belief, not downstream from it, determine its justificational status. In the crit-
ics’ example, however, the crucial process that leads ­directly to the belief in
question (belief in God’s existence) is quite clearly a defective process. Surren-
dering one’s intellect to desire, whatever that amounts to, is clearly defective.
The process reliabilist will explain such defectiveness in terms of unreliability.
This feature is salient in the case, and process reliabilism would clearly invoke
this feature to issue a verdict of “unjustified” about this belief. It does not
matter at all (for reliabilism’s verdict) that many true-belief states are the con-
sequences of your forming a belief in God (at the time in question). According
to process reliabilism, it is the causes of the target belief (within the agent) that
are critical to its J-status, not its consequences.

2.6 I n t e r n a l ism vs . E x t e r n a l ism: Fr a m i n g t h e De b at e

Evidentialism belongs to a family of justification theories called internalism
and process reliabilism belongs to a family of justification theories called
externalism. Internalism and externalism are rival approaches to the nature of
justifiers, or J-factors. In the original terms of debate, internalism has claimed
that all justifiers are “internal” and externalism has denied this. Externalists
assert that at least some justifiers are “external.” Since mental states are para-
digm cases of internal factors, (mentalist) evidentialism is a prime example
of internalism. Since truth (or reliability) is a paradigm case of an external
factor, reliabilism is a prime example of externalism. Examining the (some-
times fierce) debate between adherents of these two approaches can deepen
our understanding of issues in the theory of justification.
It might help to conceptualize the topic of epistemic justification in explic-
itly normative terms. Ethical theory regularly talks about morality in terms
of rules. For example, an action may be considered morally right just in case
it conforms to right moral rules and wrong in case it violates, or fails to con-
form to, such rules. Analogously, in epistemology we can say that a doxastic
attitude toward a proposition is justified just in case it conforms to what is
permitted by correct epistemic rules and unjustified in case it doesn’t con-
form to such rules. Presumably, correct rules permit or decline to permit
doxastic choices based on an agent’s epistemic conditions or circumstances.
If S’s circumstances include the possession of a certain body of evidence

02-Goldman-Chap02.indd 42 15/10/14 6:02 PM

 2.6 Internalism vs. Externalism: Framing the Debate 43

relevant to P, then the correct rules may permit S to ­believe P. If a different

body of evidence is possessed, the correct rules may not permit him or her
to ­believe P. What kinds of conditions or circumstances are featured in cor-
rect rules and therefore play a role in permitting or prohibiting doxastic
decisions? Those types of conditions or circumstances are the justification-­
makers or -breakers; that is, they are the “J-factors” (or some of them,
anyway). We want to determine the kinds of factors these J-factors are—in
particular, whether they are internal or external states of affairs.
What is meant by calling a J-factor an “internal” factor? There are two
main approaches here. As we have seen, one approach holds that internal
factors are mental states. An older approach is more flexible; it holds that
­internal factors are anything directly accessible to the agent at the time of
doxastic decision making. What does it mean to be “directly accessible”? It
includes being available to introspection but is usually meant to be wider
than that. Sometimes it is said to be anything available “to reflection,” where
reflection may include a priori cognition. We shall return to this in the course
of our discussion.
Why do (some) epistemologists wish to restrict J-factors to internal states,
whether mental states or other directly accessible matters? Sometimes the
argument proceeds by straightforward argument “from cases.” In defending
internalism, for example, Earl Conee and Richard Feldman (2001, 236–238)
present six pairs of contrasting cases. In each pair one person is justified in
believing a certain proposition and the other person is unjustified in believ-
ing it. Conee and Feldman contend that these contrasts in J-status are best
explained by supposing that internal differences account for the epistemic
difference. Here is one pair of such cases.
A novice and an expert bird-watcher both get a good look at a bird in a
nearby tree. The expert immediately knows that it is a woodpecker, because
he has fully reasonable beliefs about what woodpeckers look like. The novice
has no good reason like that, so he is not justified in believing it is a wood-
pecker. Conee and Feldman comment: “The epistemic difference between
novice and expert arises from something that differentiates the two inter-
nally. . . . The novice would gain the same justification as the expert if the
novice came to share the expert’s internal condition concerning the look of
woodpeckers” (2001, 237).
One can see the merit in this line of argument. However, externalists
­(especially reliabilist externalists) have a response. They can say that what
differentiates the expert and the novice is that the expert’s beliefs about what
woodpeckers look like give him a process that is reliable at distinguishing
between woodpeckers and non-woodpeckers. If his visual representation of
the observed bird is compared to his stored representation of a prototypical
woodpecker, there will be a reasonably good “match.” Arriving at a belief
that the observed bird is a woodpecker in this fashion would be a reliable
process. The novice would lack any such process. Thus, the externalist would
explain the difference in terms of the ability to make accurate—that is,

02-Goldman-Chap02.indd 43 15/10/14 6:02 PM

Random documents with unrelated
content Scribd suggests to you:
Each promised to restore to the other whatever castles or towns
they had won; and it was also agreed, that the eldest son of the
count should marry the duke's eldest daughter, who was to give her
annually six thousand francs, and a certain sum in ready money on
the day of her marriage.
This treaty having been drawn up by their most able counsellors,
was signed by them, and then they mutually pardoned each other
for whatever they might have done amiss. The young lady was
delivered into the hands of the count, and all the articles of the
treaty were duly observed, to the great joy of their subjects, who
now found themselves free from all the vexations they had suffered
in consequence of the late warfare between their lords.

CHAP. XXXIII.
THE DUCHESS OF BURGUNDY IS BROUGHT TO BED OF A SON IN
THE TOWN OF GHENT.
On the 14th of April in this year, the duchess of Burgundy was
brought to bed of a son in the town of Ghent. His godfathers were,
the cardinal of Winchester, and the counts de St Pol and de Ligny,
brothers,—and the countess de Meaux was the godmother. He was
christened Josse, although neither of the godfathers bore that name,
but it had been so ordered by the duke and duchess. They all
presented very rich gifts to the child.
This year, the duke, with the consent of the estates, renewed the
coin; and golden money was struck, called Riddes[8], of the value of
twenty-four sols in silver coin called Virelans[9]. All the old money
was called in at a fourth or fifth part of its value, and recoined. At
this time, there were great quarrels between the towns of Brussels
and Mechlin, insomuch that a severe war took place between them.
In like manner, there was much dissention among the Ghent-men,
so that several officers were banished from the town.

CHAP. XXXIV.
A PEACE CONCLUDED BETWEEN THE DUKE OF BAR AND THE
COUNTS DE ST POL AND DE LIGNY.
A treaty of peace now took place between the duke of Bar and the
two brothers, the counts de St Pol and de Ligny, who had for some
time been at war,—by which the whole country of Guise, parts of
which had been conquered by sir John de Luxembourg, count de
Ligny, and which was the hereditary inheritance of the duke of Bar,
was given up to the said sir John de Luxembourg, in perpetuity to
him and his heirs.
For the greater security of the above, the duke freely gave up the
castle of Bohain, in the presence of many of his nobles and officers
of the county of Guise, whom he had ordered thither for the purpose
of witnessing it, as well as several imperial and apostolical notaries.
There were likewise some discussions relative to Joan de Bar,
daughter of sir Robert de Bar, count of Marle, and the portion of
property she was to have in the duchy of Bar, in right of her said
father. There were also some proposals for a marriage between the
second son of the count de Saint Pol and one of the youngest
daughters of the duke of Bar: but these two articles were deferred
to the next time of meeting. When this negotiation had lasted some
days, and the duke had been most honourably and grandly feasted
by the two brothers in the castle of Bohain, he departed thence,
according to appearances, highly pleased with them, and returned to
his duchy.

FOOTNOTES:
 [8] Riddes,—of the value of five shillings.—Cotgrave.
[9] Virelans. Q.

CHAP. XXXV.
A WAR TAKES PLACE BETWEEN SIR JOHN AND SIR ANTHONY DU
VERGY AND THE LORD DE CHASTEAU-VILAIN.
In this same year, a great discord arose between sir John and sir
Anthony du Vergy, burgundian knights, and the lord de Château-
Vilain, which ended in an open war. The lord de Château-Vilain, the
more to annoy his enemies, turned to the party of the king of
France, together with sir Legier d'Estouteville, Jean de Verpelleurs,
and some other gentlemen, who had long been his allies and
wellwishers. By this conduct they broke their oaths to the duke of
Burgundy, their natural lord, with whom the lord de Château-Vilain
had been on the most intimate terms.
This lord also returned the badge of the duke of Bedford which he
had long worn, which made the duke very indignant; and he blamed
him greatly in the presence of the person who had brought the
badge, saying that he had thus falsified the oath he had made him.
The duke of Burgundy was likewise very much displeased when it
came to his knowledge, and he sent pressing orders to all his
captains in Burgundy to exert themselves to the utmost in harrassing
the lord de Château-Vilain. In obeying these orders, the country of
Burgundy suffered much,—for the lord de Château-Vilain had many
castles in different parts of it, which he garrisoned with his friends.
By the forces of the duke, assisted by the lords du Vergy and others
of the nobles of Burgundy, he was so hardly pushed that the greater
part of his castles were conquered and demolished, namely, Graussy,
Flongy, Challancy, Villiers le Magnet, Nully, the castle of St Urban,
Blaise, Saint Vorge, Esclaron, Varville, Cussay, Romay, Vaudemont,
and Lasoncourt.
The siege of Graussy lasted more than three months under the
command of Jean du Vergy, the principal in this quarrel, having with
him sir William de Baufremont, William de Vienne, sir Charles du
Vergy, and twelve hundred combatants. The lord de Château-Vilain,
with the heir of Commercy and Robert de Vaudricourt, and sixteen
hundred fighting men, marched to raise the siege, when a grand
skirmish took place, but only one man was killed.
The lord de Château-Vilain, however, finding that he could not
attempt to raise the siege without very great danger from the
strength of his enemies, retreated to the place whence he had
come; and shortly after, sir Denis de Sainct-Flour, who commanded
within the castle, capitulated to surrender the place, on the garrison
being allowed to march away in safety with their lives and baggage.
Having concluded this treaty, sir Denis went to the king of France,
who had him beheaded for several charges that had been made
against him, and also for having put his wife to death.
At this time, some captains of the duke of Burgundy took by storm
and by scalado the town of Epernai, belonging to Charles duke of
Orleans, a prisoner in England, in which every disorder was
committed as in a conquered town.

CHAP. XXXVI.
A TREATY OF PEACE IS CONCLUDED BETWEEN THE DUKE OF
BURGUNDY AND THE LIEGEOIS.
At the end of this year, a peace was concluded between the duke of
Burgundy and the Liegeois. Many meetings had been held before the
two parties could agree on terms: at last, it was settled that the
Liegeois should pay the duke one hundred and fifty thousand nobles
by way of compensation for the damages they had done to his
country of Namur by demolishing his castles, and other mischiefs.
They also consented to raze to the ground the tower of Mont-
Orgueil, near to Bovines, which they held, and which indeed had
been the chief cause of the war.
They completely fulfilled all the articles of the treaty; and the
pledges for their future good conduct were John de Hingsbergh their
bishop, Jacques de Fosseux, and other nobles of the country of
Liege. For the more effectual security of this treaty, reciprocal
engagements were interchanged between the parties; and thus the
Liegeois who had been in very great alarms and fear, were much
rejoiced to have peace firmly established throughout their territories.

[A.D. 1433.]

CHAP. XXXVII.
THE DUKE OF BEDFORD, WHO STYLED HIMSELF REGENT OF
FRANCE, MARRIES THE DAUGHTER OF THE COUNT DE SAINT POL.
At the commencement of this year, John duke of Bedford espoused,
in the town of Therouenne, Jacquilina, eldest daughter to Pierre de
Luxembourg count de St Pol, and niece to Louis de Luxembourg
bishop of Therouenne, chancellor of France for king Henry, and also
to sir John de Luxembourg.
This marriage had been long negociated by the bishop, who was
very eager to bring it about, and he was at that time the principal
minister and adviser of the said duke. The duke of Burgundy was not
in that country when it was solemnized,—but hearing of it on his
return, he was displeased with the count de St Pol for having thus,
without his knowledge or advice, disposed of his daughter.
The wedding-feasts were celebrated in the episcopal palace of
Therouenne; and for the joy and happiness the duke felt in this
match (for the damsel was handsome, well made and lively), and
that it might be long had in remembrance, he presented to the
church of Therouenne two magnificent bells of great value, which he
had sent thither from England at his own cost.
Some days after the feasts were over, he departed from
Therouenne.

CHAP. XXXVIII.
THE TOWN OF ST VALERY, IN PONTHIEU, IS WON BY THE FRENCH.
At this time, sir Louis de Vaucourt and sir Regnault de Versailles,
attached to king Charles, accompanied by about three hundred
combatants, surprised about day-break, and took by scalado the
town of St Valery in Ponthieu. The town was governed for the duke
of Burgundy by Jean de Brimeu, and great mischiefs were done
there by the French according to their custom of dealing with
conquered towns.
The capture of this place alarmed the whole country round, and not
without cause; for within a few days they greatly reinforced
themselves with men at arms, and commenced a severe war on all
attached to the English or Burgundians. The most part of those in
the neighbourhood entered into an agreement for security with
them, for which they paid heavy sums of money.
At this time also, by means of Perrinet Crasset, governor of la
Charité on the Loire for king Henry, was that town and castle given
up. It was strongly situated, and had not been conquered during the
whole of the war.
 CHAP. XXXIX.
THE DUKES OF BEDFORD AND OF BURGUNDY GO TO SAINT OMER.
Toward the end of May in this year, the dukes of Bedford and of
Burgundy went to St Omer to confer together on several public
matters, and to consider on certain angry expressions that had been
used and reported on both sides. The cardinal of England was with
the duke of Bedford, and very desirous to bring these two dukes to a
right understanding with each other. However, though these two
noble princes were come to Saint Omer for this purpose, and though
it had been settled that they were to meet at an appointed time
without either being found to wait on the other; nevertheless, the
duke of Bedford expected that the duke of Burgundy should come to
him at his lodgings, which he would not do. Many of their lords went
from the one to the other to endeavour to settle this matter of
ceremony, but in vain.
At length, the cardinal waited on the duke of Burgundy, and,
drawing him aside, said in an amicable manner, 'How is this, fair
nephew, that you refuse to compliment a prince who is son and
brother to a king, by calling on him, when he has taken so much
trouble to meet you in one of your own towns, and that you will
neither visit nor speak to him?' The duke replied, that he was ready
to meet him at the place appointed. After a few more words, the
cardinal returned to the duke of Bedford; and within a short time,
the two dukes departed from St Omer without any thing further
being done, but more discontented with each other than before.

CHAP. XL.
THE DEATH OF JOHN DE TOISY BISHOP OF TOURNAY.—GREAT
DISSENTIONS RESPECTING THE PROMOTION TO THE VACANT
BISHOPRICK.
In this year, died in the town of Lille, at a very advanced age, master
John de Toisy bishop of Tournay, and president of the duke of
Burgundy's council. John de Harcourt, bishop of Amiens, was
nominated by the holy father the pope to succeed him, which much
displeased the duke of Burgundy, for he was desirous to have
promoted to it one of his counsellors, called master John Chevrot,
archdeacon of the Vexin under the church of Rouen. The duke had
spoken on this subject to the bishop of Amiens, that when it should
become vacant he might not apply for it; and it was reported, that
de Harcourt had promised not to accept thereof. However, when he
had been translated to Tournay, the duke ordered all his subjects, in
Flanders and elsewhere, not to pay him any obedience; and in
addition, the whole, or greater part of the revenues of the bishoprick
were transferred to the duke, to the great sorrow of the bishop.
Hoping, nevertheless, to devise some means for a reconcilement, he
resided a long time in Tournay as a private person, where he was
obeyed, and much beloved by the burghers and inhabitants.
During this interval, the archbishoprick of Narbonne became vacant,
and, through the solicitations of the duke of Burgundy, it was given
to John de Harcourt by the pope, and the bishoprick of Tournay to
the before-mentioned Jean de Chevrot. This translation was made by
the holy father to please all parties, more especially the duke of
Burgundy; but it was very unsatisfactory to Jean de Harcourt, who
refused to be translated, saying, that the pope had only done it to
deprive him of his bishoprick of Tournay.
The duke, seeing that he would not comply, was more angered
against him and the townsmen of Tournay than before, and in
consequence, forbade his subjects to carry any provisions to
Tournay, under pain of confiscation and corporal punishment. He had
it also proclaimed, that all persons should give to his officers
information where any property lay belonging to the burghers of that
town, that it might be confiscated.
Very many mischiefs were done for the space of four or five years,
on account of this discord. During which time, the count d'Estampes
was sent into Tournay with a large company of knights and esquires,
to take possession of the bishoprick for Jean de Chevrot, although
John de Harcourt was in the town. It happened therefore, that when
the count d'Estampes had ordered master Stephen Vivien to take
possession of the cathedral, the greater part of the townsmen, to
shew their discontent at the proceeding, rose in rebellion, and
advanced to the cathedral, where Vivien, seated on the episcopal
throne, was going through all the ceremonies and acts that he had
been ordered to do in the name of Jean Chevrot, in taking
possession of the bishoprick.
The populace no sooner witnessed what he was about than they
rudely pushed him from the throne, and tore his surplice and other
parts of his dress. Many, in their rage, would have put him to death
if the officers of justice had not laid hands on him and carried him
off as their prisoner, giving the crowd to understand that he should
be judicially punished to their satisfaction.
John de Harcourt, on whose account this riot had been raised,
restrained them as much as he could by gentle remonstrances, and
begging of them to return to their houses, for that all would end
well, and he would legally keep possession of his bishoprick. After
some little time, the commonalty retired, and the magistrates and
principal inhabitants made the best excuses they could to the count
d'Estampes for this riot,—for they were afraid they should fare the
worse for it in times to come. The count d'Estampes, finding nothing
effectual could be done, departed, and returned to the duke of
Burgundy at Arras, and told him all that had passed in Tournay. He
was much vexed thereat, and issued stricter orders than before to
distress the town, so that from this quarrel respecting the two
bishops very many persons suffered great tribulations.
Even after the peace was concluded between king Charles and the
duke of Burgundy, the king was much displeased at the conduct of
the duke respecting Tournay, and was desirous of supporting the
claim of John de Harcourt.
John de Harcourt perceiving that the duke was obstinately bent on
having Jean de Chevrot bishop of Tournay, and that he should not be
allowed to enjoy peaceably the revenues of the bishoprick, and that
withal his lands in Hainault had been seized on and confiscated by
the duke, departed from Tournay, and went with a few attendants to
the king, who gave him a most gracious reception, and he then
continued his journey to his archbishoprick of Narbonne. Thus did
Jean de Chevrot gain the bishoprick of Tournay, who sent thither, to
take possession, a canon of Cambray, named master Robert
d'Auclair. He was at this time very courteously received there, and
obeyed as his procurator.

CHAP. XLI.
THE FRENCH MAKE MANY CONQUESTS ON THE CONFINES OF
BURGUNDY.
About this time, ambassadors were sent from the three estates of
the duchy and county of Burgundy to the duke, to remonstrate with
him on the great damages the partisans of king Charles were doing
to his country by fire and sword, more especially his brother-in-law
the duke of Bourbon. They told him, that they had already taken by
force many towns and castles, and were daily making further inroads
into the country, which must be totally destroyed unless a speedy
remedy was applied. They concluded by requesting most humbly,
that he would, out of his grace, raise a sufficient body of men, and
that he would personally march to their assistance.
The duke, having heard their harangue, assembled his council, and
then determined to collect men at arms from all his dependancies in
Brabant, Flanders, Artois, Hainault and other parts. Clerks were
instantly employed to write letters to the different lords, knights and
esquires, who had usually served him in his wars, to assemble as
many men at arms and archers as they could raise, and be ready to
march with him at the beginning of the month of May, whither he
might be pleased to lead them. The captains, on receiving these
orders from their prince, made every diligence to obey them; and
several soon brought their men into the field, which harrassed much
the countries of Picardy, Ponthieu, Artois, Tournesis, Ostrevant,
Cambresis, Vermandois and the adjoining parts, for the duke had not
been equally diligent in completing his preparations, so that these
men remained wasting the countries aforesaid for upwards of a
month.
At the end of May, the duke having assembled, from divers parts, a
great quantity of carriages, stores and artillery, set out from the
town of Arras on the 20th day of June, attended by many of his
captains. He was also accompanied by his duchess, who had a
numerous attendance of ladies and damsels, to the amount of more
than forty; and they were lodged in Cambray, where sir John de
Luxembourg met him, and requested that he would come to his
castle of Bohain, to which the duke assented.
On the morrow, when the duke and duchess had heard mass in the
church of our Lady at Cambray, and afterward taken some
refreshment, they set out for the castle of Bohain, where they were
joyfully and honourably received by sir John de Luxembourg, count
de Ligny, and the countess his lady. They and their attendants were
plentifully and nobly served with all sorts of provisions that were in
season; and they remained there for two days, taking their pleasures
in the chace and other amusements.
In the mean time, the captains and men at arms advanced into the
Rhetelois. The duke and duchess, on leaving Bohain, went to
Peronne, and thence through Champagne, passing near to Rheims.
There were with him full six thousand combatants, as well men at
arms as archers, the principal leaders of whom were the lord de
Croy, sir John de Croy his brother, sir John de Hornes seneschal of
Brabant, the lord de Crequi and his brother, sir John bastard de St
Pol, his brother Louis, the lord de Humieres, sir Baudo de Noyelle,
the lord de Crevecoeur, Robert de Neufville, Lancelot de Dours,
Harpin de Richammes, and many other nobles, as well knights as
esquires. When the duke marched through Champagne, he formed
his troops into a van guard, a main body, and a rear guard.
Sir John de Croy commanded the first under his brother,—and he
had with him Harpin de Richammes. During the march, all the
baggage was placed between the van and main body; and the
duchess, then far gone with child, was there also, with her women,
and near to the duke.
The army marched in this array before the town of Troyes, that was
held by the French, and advanced to Cappes on the line to
Burgundy. Many of the burgundian lords now joined him, to whom
he gave a gracious reception,—and having called a council of war,
resolved on their future proceedings.
It was settled that the duchess should fix her residence with her
attendants at Châtillon-sur-Seine, while the duke marched to lay
siege to Mussi-l'Evêque, in the possession of the French. Great
preparations were made, and many pieces of artillery were pointed
against the gates and walls. The garrison once intended making an
obstinate defence; but when they saw how numerous and well
appointed were the duke's forces, and found they had no hope of
succour, after eight days siege, they capitulated to surrender the
place on having their lives and fortunes spared. On the conclusion of
this treaty, they marched away under the duke's passports for St
Florentin.
When the duke had appointed a new garrison, he went to the
duchess at Châtillon, and his men at arms advanced toward the
county of Tonnerre.

CHAP. XLII.
 THE DUKE OF BURGUNDY RECONQUERS SEVERAL PLACES WHICH
THE FRENCH HAD WON IN BURGUNDY.
When the duke of Burgundy had sojourned some days at Châtillon,
he ordered the duchess to go to Dijon, where she was most
honourably received, and he himself went after his army. He had
Lussigines and Passy besieged; and the first was so hard pressed
that the garrison surrendered on having their lives spared, but giving
up their effects. Those of Passy also gave hostages to surrender on
the first day of September following, unless the duke and his army
should be fought "withal and beaten by his adversaries before that
time.
Many other castles and forts held by the French, who were much
alarmed at the great power of the duke of Burgundy, were yielded
up to him, namely, Danlermoine, Herny, Coursaint, Scealefloug,
Maligny, Saint Phalle, Sicry, Sabelly and others, to the amount of
twenty-four. After these surrenders, the duke went to Dijon, and his
captains and men at arms were quartered over the country. Sir John
de Croy was the commander in chief at all these sieges of places
that submitted to the obedience of the duke of Burgundy.

CHAP. XLIII.
GILLES DE POSTELLES IS ACCUSED OF TREASON TO THE DUKE OF
BURGUNDY, AND BEHEADED.
In this year, a gentleman of Hainault was accused of treason against
the duke of Burgundy. His name was Gilles de Postelles, who had
been brought up as a dependant on the dowager-countess of
Hainault, aunt to the said duke. He was charged with having
practised with divers of the nobles of that country to put the duke to
death by shooting him with an arrow, or by some other means,
while hunting in the forest, whither he would accompany him.
For this cause, he was arrested in the mansion of the countess, at
Quesnoy, by sir Willian de Lalain bailiff of Hainault. When he had
been strictly examined and tortured, he was beheaded and
quartered in the market-place of Mons, and his quarters were sent
to be placed in the four principal towns of that country. One of his
servants was beheaded with him; but John de Vendeges, to whom
he had discovered his plot, fled the country, and afterward, by
means of different excuses, and through the interest of his friends,
was pardoned by the duke. The countess of Hainault was strongly
suspected of being implicated in this affair, but nothing was clearly
proved against her.

CHAP. XLIV.
THE FRENCH WIN BY SCALADO THE TOWN OF CRESPY IN THE
VALOIS.—OTHER MATTERS.
While these things were passing, a party of king Charles's adherents
won by scalado at day-break, the town of Crespy in the Valois from
the English. The bastard de Thian was governor; and he, with part
of the garrison, and the inhabitants, were made prisoners:
innumerable mischiefs were done to the town, for the French treated
it in their usual manner to a conquered place.
On the eve of the feast of the Ascension, in this year, the
commonalty of Ghent rebelled against the duke's officers and the
magistrates. But the principal sheriff posted himself with the banner
of the counts of Flanders in the market-place well accompanied,
before the rebels had time to collect together, who, perceiving that
they could not now carry their intentions into effect, fled from the
town: some of them, however, were taken, and punished by the
magistrates of Ghent.
In these days, the town of Bruyeres, in the Laonnois, was won from
king Charles by sir John de Luxembourg's men, commanded by
Villemet de Hainau, governor of Montagu. This capture caused great
alarm in the adjoining places, for they expected a strong garrison
would be posted therein to attack them; and they, consequently,
reinforced themselves as much as they could, to be enabled to resist
them.

CHAP. XLV.
THE DUKE OF BURGUNDY KEEPS HIS APPOINTMENT BEFORE
PASSY.—HE BESIEGES THE TOWN AND CASTLE OF AVALON.
When the first day of September was come, the duke of Burgundy
(having previously sent his orders to all those who had been
accustomed to serve under him) made his appearance before Passy,
according to the terms of the capitulation.
He was there joined, by orders of king Henry, by the lord de l'Isle-
Adam, marshal of France, and sir John Talbot, with sixteen hundred
combatants. The duke received them joyfully, and made very
handsome presents to these lords and to their men. The French,
however, did not appear; and the garrison, in consequence,
surrendered the place to the duke of Burgundy, and marched away
under his passports.
The duke then sent a detachment to surround Avalon, of which was
captain one called Fort Espice, having under him two hundred men
at arms, the flower of the army and renowned in war. They made an
obstinate defence.
The principal burgundian lords among the besiegers were the lord
de Charny, Philibert de Vaudray, and others,—from Picardy were sir
John bastard de St Pol, the lord de Humieres, and many noblemen,
who advanced with great courage and encamped near to the
ditches. Several engines were pointed against the gates and walls,
and damaged them greatly, breaches being made in divers parts.
The besiegers now thought to take the place by storm, and made a
vigorous attack, but were gallantly repulsed. However, the garrison,
foreseeing that they could not hold out longer, and having no hopes
of succour, they fled by night in much disorder, through a postern
that had been neglected by the enemy. Their flight was soon known,
and the Burgundians lost no time in arming and pursuing them, so
that falling courageously upon them, they took and slew many. Fort
Espice and some others saved themselves by flight. The town was
now suddenly attacked, and won without resistance. The wife of Fort
Espice was made prisoner, with many of his men and some
peasants,—and every thing that was found in the place was
plundered and carried away.

CHAP. XLVI.
PIERRE DE LUXEMBOURG, COUNT DE ST POL, BESIEGES THE
TOWN OF ST VALERY.—THE DEATH OF THE COUNT DE ST POL.
In the month of July of this year, Pierre de Luxembourg, count de
Saint Pol, accompanied by lord Willoughby, an Englishman, and
twelve hundred combatants of the two nations, laid siege to the
town of Saint Valery; in which were, on the part of king Charles, sir
Louis de Vaucourt, Philip de la Tour and sir Regnault de Versailles,
with a garrison of three hundred men.
They pointed artillery against the walls and gates; and after the
siege had lasted for three weeks, the before-named knights entered
into treaty with Robert de Saveuses, who had been commissioned by
the count de St Pol for the purpose, and agreed that they would
surrender the place at a fixed day, should they not be relieved
before then, on receiving a sum of money, and liberty to depart in
safety with their prisoners and baggage. As no one appeared to their
succour, they marched away, under passports, to Beauvais.
Shortly after, sir Louis de Vaucourt and sir Regnault de Versailles
were met by one called Le Petit Roland, on the road to Senlis, who,
though of the same party, from a private quarrel, attacked them with
the men he was leading to Chantilly; and in the end he defeated and
robbed them, making sir Regnault his prisoner.
The count de St Pol, having re-garrisoned St Valery, gave the
command of it to sir Robert de Saveuses. On marching thence, he
fixed his quarters at a large village called Blangy, in the county of
Eu, with the intent to besiege the castle of Monchas, held by sir
Regnault de Fontaines for king Charles. Sir Regnault, not wishing to
wait the event of a siege, capitulated with the commissioners of the
count to surrender the place on the 15th day of next October,
provided that neither king Charles nor any of his partisans should be
in sufficient force to offer him combat on that day before the castle
of Monchas, or on the plains of Santhois near to Villiers-le-Carbonel,
one league distant from Haplaincourt. This treaty was confirmed, the
26th day of August, by the count, and hostages given on each side
for its due performance.
On the last day of this month of August, while the count was
encamped near to Blangy, and giving his orders for besieging the
castle of Rambures, he was taken suddenly ill, and died almost
instantly.
His men and all the English captains were grieved at heart for his
loss, and retired to the garrisons whence they had come. His
household had the body transported to St Pol, where it was interred
in front of the great altar of the abbey-church of Cercamps, of which
his ancestors had been the founders. His eldest son, Louis de
Luxembourg, then about fifteen years of age, took possession of all
his estates and lordships, and thenceforth was styled the Count de
St Pol.
 CHAP. XLVII.
THE LORD DE LA TRIMOUILLE IS ARRESTED IN THE KING'S
PALACE, AND MADE TO SURRENDER HIS PRISONER THE VISCOUNT
DE THOUARS.
While these things were passing, king Charles resided chiefly at the
castle of Chinon, and with him was the lord de la Trimouille, his
principal adviser, but who conducted public affairs much to the
dissatisfaction of Charles d'Anjou, and many other great lords.
They also hated him from their friendship to the lord d'Amboise
viscount de Thouars, whom he had detained in prison from the time
the lord de Lessay and Anthony de Vivonne had been beheaded
through his means at Poitiers, and also because the constable, by
reason of his interference, could not regain the good graces of the
king.
Having therefore formed their plan, the lord de Bueil, sir Peter de
Verseil, Pregent de Coetivy and other barons, to the number of
sixteen, entered the castle of Chinon, and went to the chamber of
the lord de la Trimouille, whom they found in bed. They made him
prisoner, and carried him away, taking from him the government of
the king. He afterward, by treaty, surrendered to them the lord
d'Amboise, and promised never to return to the king, yielding up
many forts that he held as security for keeping the said treaty.
Shortly after, the constable was restored to the good graces of his
monarch, who was well satisfied to receive him, although he was
much vexed at the conduct that had been held to the lord de la
Trimouille: nevertheless, new ministers were appointed for the
management of his affairs.
At this time, Philip lord de Saveuses resided in Mondidier with a
sufficient garrison to oppose the French in Compiègne, Ressons,
Mortemer, Bretueil, and other places. These had made an excursion
to the amount of about one hundred and fifty combatants into the
country of Santhois, where they were met by the lord de Saveuses,
who slew or made prisoners the greater part: the rest saved
themselves by flight.
In this year, died in his town of Avesnes, in Hainault, the count de
Penthievre, who had been deprived of the duchy of Brittany, as has
been elsewhere fully related. A great mortality took place throughout
almost all France, as well in large towns as in the country; and there
prevailed also great divisions between the nobles and gentlemen
against each other, so that neither God, his church, nor justice, were
obeyed or feared, and the poor people were grievously oppressed in
various ways.

CHAP. XLVIII.
WILLIAM DE COROAM PUTS TO FLIGHT JOHN BEAURAIN.—SIR
JOHN DE LUXEMBOURG RECONQUERS THE CASTLE OF
HAPHINCOURT.
About this period, William de Coroam, an Englishman, in company
with Villemer de Hainault, and some others of sir John de
Luxembourg's captains, with three or four hundred combatants,
overthrew and plundered near to Ivoy, between the Ardennes and
Champagne, from five to six hundred men, whom John de Beaurain,
and divers captains, had assembled in hopes of conquering them.
John de Beaurain, however, and others, saved themselves by the
fleetness of their horses.
In the month of September, the castle of Haphincourt, seated on the
river Somme, two leagues distant from Peronne, was taken by a
partisan of king Charles, called Martin le Lombard, and his
accomplices. Within the castle was sir Pierre de Beausault, a noble
and ancient knight, with his lady, the mother to sir Karados de
Quesnes.
The whole of the country of Vermandois was much alarmed at this
conquest, for the inhabitants feared it would open an easy entrance
for the enemy into those parts. They, however, lost no time in
sending notice of it to sir John de Luxembourg, who, in a few days,
assembled eight hundred Picards, and marched them, in company
with his nephew the young count de St Pol, sir Simon de Lalain, the
lord de Saveuses, and other noble captains, to the castle of
Haphincourt, and had his artillery instantly pointed against the walls.
His attacks were so severe on the garrison that they were forced to
surrender at discretion, when some were hanged and others
strangled. As for Martin, Jacotin and Clamas, they obtained their
liberty on paying a heavy ransom. The castle was delivered into the
hands of Jean de Haphincourt, and the knight and lady sent away.
After this exploit, sir John de Luxembourg returned with his nephew,
and the other captains, to the places whence they had come.

CHAP. XLIX.
THE COUNTS DE LIGNY AND DE ST POL KEEP THE APPOINTED DAY
AT VILLIERS LE CARBONEL, AND AFTERWARD DEFEAT THE FRENCH
FROM THE GARRISON OF LAON.
On the 15th day of October, the young count de St Pol, sir John de
Luxembourg, count de Ligny, with from four to five thousand
combatants, whom they had summoned from Picardy and Hainault,
under the command of sir William de Lalain, sir Simon his brother,
the lord de Mailly, sir Colart de Mailly his brother, the lord de
Saveuses, Valleran de Moruel, Guy de Roye, and others expert in
arms, marched to keep the appointment at Villiers le Carbonel,
according to the capitulation signed at the castle of Monchas in
Normandy. They were also joined by twelve hundred English, under
the orders of the lord Willoughby and sir Thomas Kiriel.
Neither sir Regnault de Fontaines, governor of Monchas, nor any
others on the part of king Charles made their appearance at Villiers
le Carbonel; and thus their hostages were left in very great danger.
The two counts, however, remained all that day in battle-array on
the plain, and toward evening quartered themselves and their men
in the adjoining villages, seeing there was not a probability of an
enemy shewing himself. On the morrow, they returned, by a short
march, to the place whence they had come.
Within a few days after this, when the two counts were at Guise,
news was brought them, that the lord de Penesach, governor of
Laon, had made an excursion, with four or five hundred combatants
from different garrisons into the country of Marle, and had nearly
taken Vervins, the hereditary inheritance of Joan de Bar, sir John's
daughter-in-law, and had set fire to the suburbs of Marle.
Sir John was much troubled on receiving this intelligence, and
instantly mounted his horse, together with the count de St Pol, sir
Simon de Lalain, and those of his household. He sent in haste for
reinforcements from all his garrisons that were near, and sir Simon
ordered his men, who were quartered in a village hard by, to follow
without delay; so that he had very soon upwards of three hundred
fighting men, whom he boldly marched to meet the enemy.
He overtook them on their retreat at Disy, not far from Laon; and
although they were very superior in numbers, he no sooner saw
them than without waiting for the whole of his men to come up, he
most gallantly charged them, and did wonders by his personal
courage. The French took to flight, even under the eyes of their
commander, excepting a few, who were defeated, and the most part
put to death, to the number of eight score. The principals were,
Gaillart de Lille, Anthony de Bellegarde, de Mony, le borgne de Vy,
Henry Quenof from Brabant, and others, to the number aforestated.
From sixty to eighty were made prisoners, the greater part of whom
were on the morrow hanged; among them was one named
Rousselet, provost of Laon. A gentleman of arms, called L'Archenciel
was taken in the engagement, but given up to sir Simon de Lalain,
whose life he had formerly saved at St Vincent, as has been related.
In return, sir Simon was desirous of saving his; but he could not
succeed, for sir John de Luxembourg caused him to be put to death,
which angered greatly sir Simon, but he could not remedy himself.
The French were pursued as far as Laon, and many killed and taken.
On this day, the young count de St Pol was entered a warrior,—for
his uncle made him slay several, in which he took much delight.
After the defeat, they all returned to Guise in high spirits on account
of their happy success.

CHAP. L.
LA HIRE AND OTHER FRENCH CAPTAINS OVERRUN ARTOIS AND
CAMBRESIS.
In the month of September, of this year, La Hire, with others of king
Charles's captains, such as Anthony de Chabannes, Blanchefort,
Charles de Flavy, Regnault de Longueval, and full fifteen hundred
combatants, whom they had assembled in Beauvais, crossed the
Somme at Cappy into Artois, and made a number of peasants
prisoners, who were unsuspicious of such an inroad, and returned
with them and their plunder to Beauvais, where they were all
quartered. They also made great seizures of men and cattle in the
Cambresis, by whose ransoms they acquired large sums of money.
They again took the field, but after some little time they divided; and
Anthony de Chabannes with Blanchefort and their men went toward
Cambray, and, passing by it, they took the straight road to Haspres,
as a free fair had been held the preceding day at the town of Ivoy;
and because the townsmen would not compound according to their
pleasure, they burnt most part of the town and the church.
They then advanced to Haspres, which was full of people and
merchandize, and entered it by surprise. They made many prisoners,
but several retired with some monks into a strong tower, which was
long attacked in vain by the French. In revenge for not being able to
gain it, they plundered all they could lay hands on in the town, and
then set it on fire, by which several houses were destroyed, with the
church and abbey of St Akaire. They also committed other enormous
mischiefs.
Having packed up their plunder, they departed, and, traversing the
Cambresis, took many prisoners, and burnt numbers of houses, and
went to lodge at Mont St Martin[10], where La Hire was waiting for
them. On this same day, La Hire had set fire to the town of
Beaurevoir, the mill, and a very handsome country-seat called La
Mothe, situated near to the town, and belonging to the countess de
Ligny. Many detachments scoured the country, committing
numberless mischiefs without opposition; for sir John de
Luxembourg was absent with his nephew the young count de St Pol
on business relative to matters that had happened in consequence
of the death of sir Peter de Luxembourg his father.
This was the cause why the French met with no resistance on this
expedition wherever they went. From Mont St Martin they took the
road toward Laon, carrying with them multitudes of prisoners and
great herds of cattle. They halted at Cressy-sur-Serre, and thence,
without any loss, returned to Laon, where they divided their spoils,
and went to the different garrisons whence they had come.
About this period, the lords de Croy and de Humieres returned, with
about two thousand horse, from Burgundy, where they had been for
a considerable time under duke Philip, assisting him in his various
conquests from the French.
The duchess of Burgundy was delivered of a son at Dijon, who was
knighted at the font: his godfathers were Charles count de Nevers,
who gave him his own name, and the lord de Croy. He was also
made a knight of the order of the Golden Fleece, and in addition the
duke his father gave him the county of Charolois.
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