Académique Documents
Professionnel Documents
Culture Documents
Epistemology
Dedicated
in
memory of
Horacio
Arl--Costa
Narrowly construed, logical approach to traditional view, the epistemology methods philosophical analytic aim of is to to bring
is
a to of to
analysis.
is
formal epistemology program that by philosophers, statisticians, operations give and with for and
computational
for of example
part addresses the anyone should bother with illustrating, role that
methods by , the
play in specific
traditional epistemologynamely, tist justificationand another addressing ng which has across a making a wide case a fallacy implications range for Finally, for that one of a
programmatic view. logical proposal epistemology, one and traditional approaches into
methodo
incorporates program.
experimental,
Why
be
formal?
you to
of
your
dates back to London lecture first three the tz, speculated different eye, that
perception Hermann
and a nineteenth
perceptions a can of
portion of represented
basic to
Although the
it
took
another color
television,
The y
story for
of how
the to
model story
allegor the to
formal epistemology. impact a formal model on inquiry, and assess the merits of one holds all, by may for reasonable epistemology.
also highlights formal model. But whether is, after that is plenty of to keeping the not
plenty of epistemology formal methods, and committed made their than Why,
that smidgeo
bother being
The given
best years
short ago
answer to by Rich
this
question on
is philosophers
Thomason.1 Thomason, view formal methods a d distraction that the that past to only we are at
advancement,
we have Calculus.
unavailable history,
is
this:
why
limit
your
options?
return to development. people on great a so At sensitive color the had three model the that to primary and went run to nowhere
the
Young--Hel First,
important stages in was Young's idea normal color their eyes. perception This was
idea, but empirical languished wisdom held receptors and blue, wheel and mixed But
designated painters
notwithstanding, Young's tz came along. His ingenious contribution which subjects were match their three that match . the color choice, of which
two of experiments
evidence
replacing it was
represent
2 mixture of motion. mathematical stage of spurred a ced the three basic Helmholtz's model the colors that really set of Maxwell Helmholtz's and things in the second introdu work. color
color photograph within a decade of Halftone printing presses came along soon motion pictures decades later. occurred well when mholz theory physiological trichromatic of color after. followed And yet, before the mechanism vision was
Color television several all last that finally of these stage, underpins
confirmed.
outsized the
Experimental sufficient.
evidence
is
important
but
far
attacked
within appeals
justifying methodology
that Weinberg, Nichols, `Intuition Driven Romanticism (Weinberg the first history will et. part al. of
resonant,
was
stalled
people at
that
time intuitions for color about the that the three set theory
Helmholtz's track by of
colors to
Even ates
this of
history Helmholtz's
complic experim
philosophy. crucial
pointing to the experimental conclusive thesis about in ents from RGB down be for the
basic from
colors, physiological
closely attack.
at
enough, is of experimental
refashioning psychology
(Carlson into an
particu garnere
general conclusions about d from studies undergraduate psychology WEIRD people (Henrich Educated,
nature on who
students,
overwhelmingly Western,
Democratic societies. In the literature, Henrich, Norenzayan found that representative populations in tion, 3 categorization reasoning concepts, remark that studies spatial WEIRD
fairness,
moral IQ.
reasoning, They
no that
obvious a a particular
universal empirical
rather unusual,
analyti are
experimental differences
evidence to in epistemic
in different groups and these claims that epistemic intuitions are universal (Weinberg et. against experimental psychology that used there for are the differences lion's share between of
differences al. is
undermine knock
Stich
would
have
us
believe
intuitions about have observed from one on (NB: `dual ogy,2 East hand, the It Asian and
(Gettier
1963)
they on the
the Indian sub--continent, students epistemic what they intuitions. make of and and
critique of Henrich the dual--process theories not the involve dispute provide judgments 2011).)
psychological
Gigerenzer
problem intuition
with is
this that
line it
philosophy a a would
are universal general facts Rather, what practice claims of of the each of
norms or nature.
non--representative of are is
philosophy, it highly trained, status professors question. college sophomores. a point in critique of
philosophy the
traditional
disappointing
fact
remains
prefer we
2 For overview of dual process theories, see Kruglanski and Orehek (2007), Evans (2008) and Kruglanski and Gigerenzer (2011). 4 The introduction of front, but before considering of the formal models can turn to history. help another on this feature
Counterexamples
are
not
always decisive.
Traditional ent to
epistemology, epistemic
to solely on good to as be
its
commitm
intuitions, also tends to assess theories, means that But Look RGB a it miss does
the
is by colors which represented never been considered model, nor threat model serve. were to was weighed
limitation to
`counterexample' viewed as a
the physiological theory of originally designed to Rather, against the exceptions the power
model, of
and the
a recognition that theorywhich remained through and the the first last half
the half of
unconfirmed century
neighborhood can
philosophers' down
about
evidence to offer
to a resist contrivedabove
matter how
The problem with philosophy without taking ints into is account that the
aspiring
to
into a ideas. by
formal logic informallybecause to describe define the fundamental Memories are the on right one's
encourages of
philoso their
operations
justified beliefs coherent like one a bump corner this and relianc
all d
means are
that
the
hard, head
unsolve on.
artfully At its
addressed to turn
philosophy
one way to epistemological empirical if you with a empirical theory is cases, that
slide
is
by
constra
considerations and replace those formal model, and constraints, then successful hard won
the theory's aspirations for conceptual judge than the the had
failings. analysis,
In that
other will
words,
success and failure counterexample avoidance. That its vague not is one introduction, idea of recognized
crucial difference which models how to create new model, which is color the the not perception. limitations
exceptions do.
have by
the
merits of
to a illustrative
selected analyti
traditional
epistemic to a beginning as a
setting, introduced
between viewing formal epistemology approach within analytic epistemology and interdisciplinary program, may lead a Pereira simply from type which to viewing research
of methodological naturalism 2005). Here I want the this advantages broader, you would to
programmatic
viewwhatever
prefer to
is there
used is
in often of how
build a problems
rich repertoire together techniques methods (Haenni fixing the may also appear in for are
varying the problem, single problem, which different guises in the methods. An viewing the different begin are of
various advantage
through we can
background
abstraction
is
ignore. Moreover, problem, when reshape your can be philosophical tool for prior to
simply serve of
intuitions. Here the problem with its experimental with WEIRD similar people.
is a partial besets traditional on limits from The not rest claims if model and to epistemic relying entire
reliance
of a grounds a to will
6 While ology, that place this it this for essay presents should be clear is hardly a experimental work is a place formal methods precision, the most a case for formal epistem is too. only be that hampere precisi a
manifesto. There in philosophy, for are intuitions, not the may human well
and it important
friendship,
political
compromiseis
Coherence
and
Dilation
So a view. phy's
far we bird's--eye
have point
discussed of
from philoso
We agreed with critique of philosophy, likely to also found Throughout we how methods used one
the spirit of traditional namely that take us experimental this a might and
In two
shift work
consider a coherence,
theory of
problem for is a
methods being brought to within traditional analytic epistemology. of sound reasoning play. This formal ine, The second example probabilistic and is the an
independence of a stand of
pursued ramifications
Bayesian
Toward
theory of
coherence
In re of tism
1985, to
Laurence BonJour provided the coherence theory his postulates describe a of C. self-- then have this
memories working
(Lewis 1946). Since within the Bayesian epistemology of developing a of coherence along
Lewis--Bon
Much s of of
in a
Bayesian set
epistemology probabilistic
concern measure
whether a be adduced
3 He later despaired of meeting those demands and quit the theory altogether, bu t that is another story. 4 See, for example, Huemer 1997, Cross 1999, Shogenji 1999, Bovens & Hartmann 20 03a, 2003b, 2005, Olsson 2002, 2005, Fitelson 2003, Meijs, 2004, Glass 2006. 7 which of set are is `truth--conducive'that coherence among a of more propositions likely to is, whether those among fully a propositions Bayesian succeeds on results show
higher degr
general is that
in
effect how
any
The question is whether between probabilistically correlated evidence (thought and incremental confirmation (thought ology has to model investigated terms 2002,
there to
assumptions witness
(bw1) A that
is is,
messenger i is positive | each is, Report i(A)) messenger whether whether A and all
that A,
messenger
The ring
idea is whether is or
to to only its
a or by are
conside in not
negation fact A
what might
other be
be
or
what
other
messengers
According pillars
to of
these the if
two most
twin
possibly as well:
truth--conducive,
necessary
conditions
truth and
conducive individual
in
the
compara (2005,
credibility
While these assumptions formal perspective, they te what even shouldin fact circumstances[and
may be
seem seen
restrictive as
from
interesting in
that is (2005, p.
conducive
compara
5 See (Pearl 2000) and (Spirtes et. al. 2000) for a thorough treatment. 8 And both yet, while favorable (bw1) and and (bw2) necessary, is the may the least seem case. intuitively The witness
it turns testimony
models for exploring and likelihood (Wheeler 3 and and 4). Scheines
the
relationship 2011
between
forthcoming, the
drop built
the into
conditional
independence (bw2)
conditi
(Myrvold (Wheeler
the
ratio
of over in
evidence association
evidence alone. This relationship the rightmost expansion of measure, For, which is e statements, E 1 single hypothesis, H. defined and E 2,
evidenc
Given one
that to
comparing
another (e.g., tion { E 1, E 3}) by (with respect their to degree of a more (Wheeler focused and focused correla correlation Scheines weaker assumpt respect 1, given that
designated entails
confirmation, ceteris paribus forthcoming, Schlosshauer and Wheeler, ions, when correlation to a then the evidence 2009, Wheeler 2011). focused
What's more,
of an evidence set hypothesis, H) is greater incremental confirmation of is positive (Wheeler and Scheines,
forthcoming).
is fail
not to
that be
witness modelsbecause ceteris is fine in many cases outside of the models. an insight constraints Why? The from imposed reason boils focused
correlation,
is
that
there
is
missing a is
from probabilistic theory not enough to (e.g., the similar the any is reason regard
the association messengers all rather we association. might truth agree of must to the
6 Conditions (A1, A2) in (Wheeler and Scheines forthcoming) which is generalized in (Schlosshauer and Wheeler 2011) 9 the . possibility But, that in (bw2) so was designed to doing, the witness the of possibility prevent of is
Given the
this cause
must
regulating if the
prefer). This requires rethinking the to Lewis--BonJour witness models, and terms moving away from of information seems defining sets, but to have coherence already
BonJour himself
that some
a other
way of beliefs
justification of , rather than appealing contents of other belief was justificationally a general caused ing are in this further generally to with
these perceptual merely the the coherence contents (so would irrelevant), belief that special conditions true that be appeals beliefs way as
beliefs produced
(and well)
perhaps
(2002, p.
206--7).
In what
we think coherence
that is on and
account probabi introducing no to The the indepen help model impossi version think that
crucial to making progress listic theory of coherence, followed a model through on which combines cases to in foul
evidence will affect incremental is hardly comprehensive, and bility there are results intriguing allude to. present impossibility
riddle of
coherence.
Dilating
sets
of
probabilities
on
probability of
definition with
classical event E F of
of the marginal
(IND) 10
Pr( E, F)
Pr( E)
Pr( F).
For example, suppose of a fairly flipped `tails' and F American tosses are of is Those ion, is the quarter
E Euro
is coin
two
.
textbooks too. So often long will as may of to = give an Pr( F) also F the Pr( E), irrelevant between say just alternative that when E F of definit is is E:
epistemically difference E E
probability landing
a ,
Euro same
coin as
has the
is
probability that lands `tails'. knowing how the is irrelevant outcome of Finally, of F a we and
experiment with to estimating fairly tossed quarter. may Eso just long as as
well we avoid
places on
independent
other, that = E)
When Pr,
a are
distribution,
be expressions probabilistic
differences If Pr
between is
of represe events is
belief, then arguably are stochastically independent epistemically (IND) by irrelevant In Pr,
behavioral agent
if and independence
7 Or if there is a distinction draw, it is simply between conditional independen ce (IR) and independence (IND). 11 us to marginal probability since the determined lity of The into and ge factorize that distributions, which some semblance probability of by E taking the and the joint gives of ( E distribution acting like and F) of of by a is the F. the logic probabi
product probability
ability to a product
factorize a of marginal
way and
at that
the it
of through of to
stochastic
the probability allows us to independence for The strategy Spirtes et. to probabilistic
probabilistic reasoning. nets illustrates this perfectly Our own (Pearl 2000, approach
is
another
example
(Haenni
et.
al.
2011).
is this
of on. three
how It
sound
a concept appear
surprisingly, distinct
mathematical to be expressing the lens measure, is that reasoning assumptions glide out a les
notions after three equivalent the of Pr. sound from which same a
of viewed through
means, philosophically, conditional and with So, (EI), those independence will turn
reasoning depend on for assuming a probability nate Assuming degrees distribution. that of
numerically several
determi authors
accommodate approximate for example that pocket: it the probability that coin Instead, is more `tails' Yet even a go is of
idealization single,
landing `tails' the argument for reasonable is opening just a number of to the to plus the crack
probabilistic s is
independence or an example.
several
independence
concept
the to that
interva E, distinc
probability t ways
several but a
8 Pioneers of imprecise probability theory includes B.O. Koopman (1940), Alfred Horn and Alfred Tarski (1948), Paul Halmos (1950), I. J. Good (1952), C.A.B. Smith (1961) , Daniel Ellsberg (1961), clude Isaac Levi ce (2010), Fabio See also Haenni and Henry Kyburg, Jr. (1961). Notable contemporary advocates in (1980), Peter Walley (1991), Teddy Seidenfeld (2010), James Joy Cozman (2000), Gert de Cooman and Enrique Miranda (2007, 2009). et. al. (2011).
9 Recent textbook treatments include Paris 1994, Halpern 2003, Haenni et. al. 20 11. 12 common lity a one is assignment of the to interpret [ l, u] to an an interval event E = E {Pr1, the of probabi by Pr2,, upper Pr( E). to
infimum of E and of
Pr( E) = l = infPrPr Pr( E) (lower probability) Pr( E) = u = supPrPr Pr( E) (upper probability)
set
of
classical
Bayes
agents who
express
judgments a model
about for
E; Pr could studying in Pr
also
be Bayesian
interpr
credal probabi
underlying consider
mechanics is not
interpretation.
Although described
imprecise probability as `exotic', in Bayesianism the set drops Pr From a different one end
we can condition
(Cx) two
a closed convex set measures Pr1, Pr2 all = (Cx) the 0 r Pr1+
when, in
for 1, is
Condition in Pr,
convex mixtures, is
set of imprecise we in
probabilities credences.
is
interpreted holds.
We E
say just
that in
an case
event
dilates
the
event
In case
just
in
proper subset of to E
B is outcomes. B strictly
13 The remarkable specter of estimate of e, no matter the thing about turning E into outcome. strict dilation is a more precise a less precise the estimat
recount
Peter
that of
a the
fair coin
coin is
is
tossed
a in
fair such
toss, a
the that
performed first Let We toss. H 1, T know (PrA) but second. second toss estimat
outcome about the possible possible pair and toss, toss the the of that so is
or degree of the 1, H 2, T 2 denote the outcomes the coin toss e is for for is a the between extreme the puts the fair fair first
outcome of a precise
constraint prior to
1 ( a) Pr A ( H 1) = Pr A ( H 1) = Pr A ( H 1) = = Pr 2 A ( H 2 ) = Pr A ( H 2 ) = Pr A ( H 2 ). or between the ce by pair of tosses. Model A's ignoran However, degree of little is dependence known about the direction
1 ( b) Pr
A ( H 1, H 2 ) = 0, and Pr A ( H 1, H 2 ) = Pr A ( H 1) = 2
that toss
A is
the
outcome
of
and
can
yielding
( c) ( i) Pr A ( H 2 | H 1) = Pr A ( H 2, H 1) / Pr A ( H 1) = 0 and ( ii) Pr A ( H 2 | H 1) = Pr A ( H 2, H 1) / Pr A ( H 1) = 1 So, learning dilates A's estimate of value within the interval An analogous learns that the toss the are A's second unit the argument outcome the second toss [0,1]. holds of two there first about if the outcomes toss the to way can the 1991, instead first to any A partition turn vacuous pp. 298--9). out, although that the initially first toss PrA( H 2) lands heads = ,
is tails. Since these outcome space, i.e., no other precise ways the probability
One is
way that
to they
the two
hypotheses toss. Each specifies that the be ( ii) be on tails says heads this
mechanism
mechanism: certain
of the possibly
definitive information, is insufficient to determine in t. Arguably, the first signals toss a what then,
that is
gap in
in her belief.
the more
reproduce is no the
connection toss.
a is
What we example
1 (
a ) Pr ( H A 2 ) = Pr A ( H 2 ) = Pr A ( H 2 ) = 2 Pr A ( H 1) = 0, Pr ( H A 1) = 1, well. which But appears to the explanation dilate the second toss we provided for as
the this
is since
not we
to Yet,
the tosses are events are appears dilate the of be?! that sharp another,
imprecision independent
What this
is interesting setting is of
`independent
event' in
dilation is for stochastically independent 2.1--2.3), paradoxical drawn. What is that there the is flips after
conclusions critics al.'s theorems tell where So, second could there it coin this is would toss be?
dependent
How
The ion
us there
to are
our
earlier a single,
discuss
several independence concepts rather than unified independence concept (Kyburg and Within the probability independence Pittarelli, imprecise setting, entails Cozman 2012, Wheeler,
epistemic but it
epistemic nor,
irrelevance
shockingly, independence!
does
epistemic The
independence
entail stochastic
10 A version of this is discussed by Walley (1991), White (2010), Sturgeon (2010 ), and Joyce (2010). 15 reasons for be dealt the ions this with are in technical, another point (if which and will essay.11 is that have to But
reasoning sets of
extend to measures.
observation is a
suggests the side issue, and is that concepts. for horn, other:
over a
For orthodox reveals a (IND), (EI), properties collapsed distribution, Bayesianism if you and which when Pr.
theory
are for
you
that degrees
your
elicitation of belief
numerically collapsing
Bayesians: behavior of
In so far interpretation of do
as you
you provide
events as (EI)?
completely
follow from
In dence
that a
there
many
indepen
reaching consequences. Although this discovery through diagnosing ure on point listic
concerns reasoning.
FIE--model Snow up
of
inquiry long broad and of ago types hard each all, 1959 that of universities yet
is the
some way
basis we
for do.
grouping Physics,
academi
11 However, if you drop (Cx), then Stochastic independence does not entail Epist emic independence. See Wheeler forthcoming for an example. 16 chemistry, experimental and le part e. and biology are science. Although the the pillars skills aim to the by experimental of an evidence The quarrel of reconci evidenc the data; literat indicat
humanities there are branches ure. It importance ion of is. front The and
placed on experimental how scientific a hard sciences put center, whereas either do experimental some often not or philosophy,
extent, viewed as
whether or part
philosophy of the
is part sciences.
is a Mathematics
familiar has no
picture, more to
it do
with experimental cooking is experimentation as cooking is academic we as subject, In closing, should instead
data as
and
professional But
to suggest that academic disciplines rather than two. There This Reason there thoselik this
dividing into three are formal disciplines, experimental three--way proposed (1990, 16) activities is much disciplines, distinction by to that to
university, way of
disciplines, nature.
particular Call
Mathematics empirical
a formal discipline, largely the traditional leading disciplines. literature sometimes at the an a various observed fine But are
empirical disciplines, and letters are exemplars of all fields draw each category. concerned with the upon
mathematics and psychology with interpretation, and psychology interested society or r, or series of and in literature the facts
It help
is in
unclear organizing
FIE--model as
would
Kyburg idea
when an
The Or,
It is results
epistemology and
sciences,
statistics, Rather,
is that inquiry,
harnessing together etive skills do not match our That that 17 the way departments Snow he was must we and right go
organize believe
References BonJour, Laurence 1985: al Knowledge. Cambridge, University BonJour, ionalism Sosa 1999, Press. Laurence 1999: and Coherentism'. pp. 117--42. 2002: Epistemology. Oxford: Rowman `The In Dialectics Greco and of Foundat The MA, Structure Harvard of Empiric
BonJour, Laurence and Littlefield. Bovens, Luc Epistemology. University Bovens, Luc Riddle of 601--33.
and Stephan Hartmann Oxford: Oxford Press and Stephan Coherence'. Mind, Hartmann 112, pp.
2003a: Bayesian
2003b: `Solving
the
2006:
`An
Impossi
Studies,
and D. intelligence:
2001: of
Uncertainty the
in
conference
(UAI--2001).
Francisco:
person in 75,
Foundations Chicago
of
Probabi
the
Provabl 120(2),
Cozman, Fabio 2000: pp. 199--233. Cozman, Favio dence, and Synthese, Cross, ve
Intelligence,
Distributions, Indepen
and
Truth
Conduci
186--93. Crupi, 'On V., K. Bayesian Tentori, Measures empirical 229--52. Clark Nets and of M. Gonzalez 2007: Evidential Support: of Science Propert
issues'. Philosophy
`Linearity pp.
Variables'. de
Cooman, `Symmetry In
and Enrique Miranda models versus models of Wheeler Enrique 2007, Miranda
symmetry'. de Cooman,
`Forward Planning,
of
Statistical
Douven, Igor and ese, 156(3), Earman, John tion of Cambridge, 18 Eells, and Ellery and Asymmetries
`Measuring
A Critical Theory.
Branden Fitelson in Evidential 107(2), `Risk, Ambiguity, Journal of 643--69. 2008: and
Support'. Philosophical Studies, Ellsberg, Axioms'. Economics, Evans, of Daniel 1961: Quarterly 75: pp.
J. St. Reasoning,
B. T. Judgement, Review of
Cognition'. Annual
Psychology, A
Ewing, Alfred C. 1934: London: Methuen. Fitelson, Branden Coherence'. Analysis, 63, Gendler, Studies Tamar in
Oxford: Oxford University Gettier, Edmund 1963: ge?' Analysis, 23, Good, the I. Royal J. 1952: Statistics
Justified 121--123.
True
Belief Knowled
`Rational Society,
107--14. Greco, Guide John to and Ernest Sosa (eds) Epistemology. Malden, 1999: MA: The Blackwell
Blackwell. Glass, D. Relations H. to 2006: Fuzzy `Coherence Similarity Measures and and their
in 26, Clark by
Knowledge Bases'. Artificial pp. 227--49. 1998: `What Observation of Went and
Intelligence on 1--32.
Science,
Glymour, K. Kelly Academic Haenni, R. Williamson Probabilistic . Halmos, Paul Nostrand Halpern, ge, MA:
Romeyn, Probabilistic
G. Logic
and
J. Library
Networks,
Dordrecht:
1950: Measure Theory. Reinhold Company. Joseph 2003: Reasoning MIT Press.
York:
Hartmann, Stephan, Marcel Weber, Wenceslao z, Dennis Dieks, Thomas Uebe (eds) 2011: and Explanation, Old Ones Prediction, Reconsidered. and Confirmation:
Gonzale Trends
Dordrecht:
Hendricks, Vincent and phy. Rolskilde: Automatic Henrich, 2010: The World? Behavioral Joseph, Weirdest and
33(2--3), in
Horn, Alfred and Alfred Tarsk Algebras'. Transactions of AMS. 64(1): 467--97.
`Measures
`Probability Journal
and of
Coherence
Justifi
35,
pp.
Imprecise Making'.
and J. Monotone
Charact pp.
Models'. The
Annals of
1994:
`What
Price algebra
Coheren
axioms and of
41(2): 269--92. and Gerd Judgments Gigerenzer Are Based Review, 2011: `Intuitive 118(1): `Partitioning the
on Common Principles'. Psychological 97--109. Kruglanski, Domain of Dual mode Review of 291--316. Kyburg, Jr. H. of Rational University Kyburg, Jr. Oxford: Press. H. E. 1990: Oxford University E. and Transactions and Science Press. M. on Arie Human
alternatives'. Annual
and CT:
and Pittarelli A.
Kyburg, Jr., H. Bayesianism'. IEEE Systems, Levi, MA: Lewis, and Isaac MIT Man,
Cybernetics of
Knowledge.
C. I. Valuation.
Meijs, n's
`A of
to pp.
Hartman
Evidenc
is the Journal
Problem of of Philosophy.
246--72. J. 2005: Against Justification. Oxford: Press. The Uncertain Companion. University Reasoner's Press. in the laborat of Companion: Coherence: Oxford Truth, Probabi
Cambridge 1986: of
D. O. influences
on social psychology's view Personality and Social Psychology, Schlosshauer, `Focused and of 51, pp.
nature'. Journal
515--30. and Gregory Confirmation, Variable 276--92. `Krister 159--167. Wheeler 2011:
Maximillian Correlation,
the Jigsaw Puzzle of Science, 78(3), pp. Krister Symons 2005, Teddy, Mark `Coherent under Teddy of 2005: pp.
Segerberg, ks and Seidenfeld, Kadane 2010: Functions 157--76. Seidenfeld, for sets Annals of
Uncertainty'.
Truth
Structure
7,
pp.
C. P. Printing.
and R. Search.
S. 2010: Gendler
126--49.
Walley, Peter 1991: Statistical Reasoning Probabilities. London: Chapman and Hall. Weinberg, `Normativity Philosophical Wheeler, ation' . Philosophy Wheeler, and the Evidence'. Philosophy J., and Topics, S. Nichols and Epistemic Intutions'. 29, pp. `Focused Journal 60(1), pp.
with
Imprecise
S.
Stich
2001:
429--60. Correlation for the 79--100. Calibration for the 2008: and Confirm
`Objective Bayesian of Non--convex The British Science, to Lus and Journal appear. Moniz Pereira Epistemic 315--28. Richard Causation'. Dieks, and Scheines
Wheeler, Gregory and `Methodological Naturalism Internalism'. Wheeler, `Coherence, Hartmann, Wheeler, e and under White, e'. review. Roger In 2010: `Evidential Gendler and Synthese, Gregory Association, 163(3): and and
Uebe,
symmetry Hawthorne
and
mushy
credenc
2010,
pp.
Williamson, Bayesianism.
21