8 BRIDGES BETWEEN
MAINSTREAM AND FORMAL EPISTEMOLOGY
Special issue of Philosophical Studies
An International Journal for Philosophy in the Analytic tradition
Guest-edited by
Vincent F. Hendricks
Abstract
Contributors
Contents
Flyer
Abstract
Contemporary epistemologists are roughly divided into those relying
largely on common-sense considerations and focusing on examples and
counterexamples for advancing or rejecting various epistemological
theses, and those applying a variety of tools and methods from logic,
computability theory or probability theory to the theory of knowledge.
The two sorts, and the traditions to which they hitherto are taken to
belong, have unfortunately proceeded largely in isolation from one
another. But on closer examination the approaches have much in common,
may be bridged for their mutual benefit and the advancement of
epistemology in general. Here are 7 ways of doing it as the invited
papers in this special issue of Philosophical Studies demonstrate the
fruitful interaction between informal considerations and various formal
apparata in order to support, sharpen, undermine, realize, or contribute
in some other pertinent way to fundamental epistemological themes.
CONTRIBUTORS
-
Horacio
Arló-Costa (Carnegie Mellon University)
-
Johan van
Benthem (University of Amsterdam / Stanford University)
-
Luc Bovens
(London School of Economics / University of Colorado at Boulder)
-
Sven Ove
Hansson (Royal Institute of Technology, Stockholm)
-
Vincent F.
Hendricks (Roskilde University)
-
Stephan
Hartmann (London School of Economics)
-
Matthias
Hild (Darden Business School of Administration)
-
Robert
Stalnaker (Massachuetts Institute of Technology)
-
John
Symons (University of Texas, El Paso)
-
Heinrich
Wansing (Technical University of Dresden)
CONTENTS
Rationality and Value: The Epistemological Role of Indeterminate and
Agent-Dependent Values
An
important trend in contemporary epistemology centers on elaborating an
old idea of pragmatist pedigree: theory selection (and in general the
process of changing view and fixing beliefs) presupposes epistemic
values. An elaborated theory along these lines can be found in the
writings of Isaac Levi. More recently both Hilary Putnam and Bas van
Fraassen have also written books defending interesting variants of this
model. And decision-theoretical models of belief change are today more
popular also in other areas, like Artificial Intelligence. This article
focuses on analyzing the case where epistemic values are indeterminate
or vague. According to the theory that thus arises epistemic
alternatives need not be fully ordered by an underlying notion of
information-value.
Independent work by Amartya Sen analyzing game theoretical situations
and strategies leads to an articulation of preferences as chooser- and
menu-dependent. Sen argues quite persuasively that this articulation is
needed in order to have an adequate notion of rationality covering cases
of social interaction where social norms are important. The resulting
notion of value also violates standard consistency conditions (like
a,
t,
WARP, etc) and is therefore compatible with a view of incomplete
preferences where the process of choice is an important concern. Sen
argues that this fine-grained view of preference is more realistic and
that it can be accommodated in a maximizing framework once the
axiomatic structure is
adjusted.
More realistic accounts of the notion of value involved in the process
of fixing beliefs as well as more careful analyses of the notion of
preference of interactive agents suggests a picture where alternatives
need not be fully ordered and where values are both agent- and
menu-dependent. Nevertheless neither the resulting indeterminacy nor the
introduction of more fine-grained contextual parameters conspires
against the fruitful use of values both in epistemology and in
interactive epistemology. On the contrary, it leads to the formulation
of slightly more realistic theories of rationality both for single
agents and for populations of interacting agents.
From Knowing to
Learning
Much of epistemology has
been concerned with what it means to possess knowledge, as a sort
of 24-carat information that is available to us in reliable ways.
Sophisticated new accounts of such intuitions keep appearing until
this day. By contrast, much recent work in epistemic logic has
concentrated on dynamic mechanisms that produce or modify
knowledge and belief - such as speech acts, communication,
observation, learning, or even more radical belief revision. In
that light, the quality of what we *have* resides largely in the
quality of what we *do* - individually, or socially in interaction
with others. In our paper, we discuss the broader philosophical
relevance of current logics of information update and belief
revision, comparing their basic notions to ways of defining
knowledge.
An
Impossibility Result for Coherence Rankings
We receive
items of information from independent and partially reliable sources
(I&PR). What determines our degree of confidence (DoC) that these items
are true? E.g. we are trying to determine the locus of a disease
on the human genome. Before running tests there are certain loci we
consider more and less likely. We run n tests to identify areas where
the disease is located with I&PR instruments. The instruments identify
possible loci. Let there be some overlap between the reported areas.
Prima facie there are three factors that determine our DoC that the
tests are providing correct data, i.e. that the locus of the disease is
in the overlap:
(i) How expected are the results? Compare two cases. The only
difference is that in case (1) the overlapping area is likely whereas in
case (2) it is unlikely. Our DoC will be lower that the locus of the
disease is in the overlap in (2) than in (1).
(ii) How reliable are the instruments? Compare two cases. The
only difference is that in (1) the tests are quite reliable, while in
(2) they are quite unreliable. Our DoC will be lower that the locus of
the disease is in the overlap in (2) than in (1).
(iii) How coherent is the information? Compare two cases. The
only difference is that in (1) the tests point to precisely the same
area A, whereas in (2) they make broad sweeps and the overlap coincides
with A. Our DoC will be lower that the locus of the disease is in the
overlap in (2) than in (1).
We construct a Bayesian model to determine our DoC that information from
I&PR sources is true and show that the following set of prima facie
reasonable conditions, constitutive of the Bayesian Coherence Theory
of Justification (BCTJ), are jointly inconsistent:
(i) Separability. The more coherent the information is, the more
justified we are in believing it, ceteris paribus.
(ii) Holism. The relation ‘...is more coherent than...’ is
defined over information sets of any size n.
(iii) Ordering. The relation ‘...is more coherent than...’ is an
ordering.
(iv) Probabilism. Coherence rankings are determined by a
probabilistic measure defined over information sets.
The reason one is tempted to endorse BCTJ is that it does indeed hold
true for the case of n = 1, 2. However, BCTJ is an ill-fated
conjecture, since it breaks down for all cases n > 2. This is
also why attempts to construct a viable probabilistic measure of
coherence have been fatal. We discuss some admittedly costly escape
routes to salvage BCTJ.
Epistemic Coherence and its Formalization
The degree
to which our beliefs are mutually compatible and/or supportive is an
essential component of a comprehensive account of the mechanisms of
belief. It is also an aspect that is unusually well suited for formal
treatment. However, in most formal accounts, rather simplistic
approaches to coherence have been chosen. In probabilistic
representations, coherence is typically an all-or-nothing issue. In
non-probabilistic belief revision, it has mostly been taken for granted
that logically closed belief sets represent a coherentist and belief
bases a foundationalist model of belief. In the present contribution it
is argued that a more elaborate representation of coherence is needed,
and an outline of such a representation is presented.
TBA
Knowledge and Belief, Rational Action and Interaction
Interactive epistemology - the application of epistemic logic and
semantics to game theory - throws light both on some patterns of
strategic reasoning (such as backward induction reasoning) and also on
some general questions about the nature of knowledge and the relation
between knowledge and belief. Specifically, the effect of the assumption
that the players in a game have common knowledge that they all are
rational depends on exactly what is assumed about knowledge. In this
paper, I will explore some different assumptions about knowledge, and
trace some of their consequences in a game theoretic setting. I will
suggest that the application of epistemic logic in a theory of practical
strategic reasoning helps to connect formal questions about the logic of
knowledge with some classical problems of substantive epistemology.
Doxastic decisions, epistemic justification, and the logic of agency
A prominent issue in mainstream epistemology is the controversy about
doxastic obligations and doxastic voluntarism. In the present paper it
is argued that this discussion can greatly benefit from forging links
with formal epistemology, namely the modal logic of agency, belief, and
obligation. We shall suggest a stit-theory-based semantics for deontic
doxastic logic and claim that this is helpful and illuminating in
dealing with the mentioned intricate and important problems from
mainstream epistemology. Moreover, it is argued that this linking is of
mutual benefit. Simple observations concerning other-agent nestings of
stit operators may help clarifying issues about making belief and
responsibility for beliefs of others.
Keywords: doxastic obligations, doxastic voluntarism, doxastic logic,
deontic logic, agency, stit-theory
|