Vous êtes sur la page 1sur 27


First published Wed Mar 9, 2011

Abduction or, as it is also often called, Inference to the Best Explanation is a type
of inference that assigns special status to explanatory considerations. Most
philosophers agree that this type of inference is frequently employed, in some form
or other, both in everyday and in scientific reasoning. Hoever, the exact form as
ell as the normative status of abduction are still matters of controversy. !his entry
contrasts abduction ith other types of inference" points at prominent uses of it,
both in and outside philosophy" considers various more or less precise statements of
it" discusses its normative status" and highlights possible connections beteen
abduction and Bayesian confirmation theory.
#. Abduction$ !he %eneral Idea
o #.# &eduction, induction, abduction
o #.' !he ubiquity of abduction
'. Explicating Abduction
(. !he )tatus of Abduction
o (.# *riticisms
o (.' &efenses
+. Abduction versus Bayesian *onfirmation !heory
Academic !ools
,ther Internet -esources
-elated Entries
1. Abduction: The General Idea
.ou happen to /no that !im and Harry have recently had a terrible ro that
ended their friendship. 0o someone tells you that she 1ust sa !im and Harry
1ogging together. !he best explanation for this that you can thin/ of is that they
made up. .ou conclude that they are friends again.
,ne morning you enter the /itchen to find a plate and cup on the table, ith
breadcrumbs and a pat of butter on it, and surrounded by a 1ar of 1am, a pac/ of
sugar, and an empty carton of mil/. .ou conclude that one of your house2mates got
up at night to ma/e him2 or herself a midnight snac/ and as too tired to clear the
table. !his, you thin/, best explains the scene you are facing. !o be sure, it might
be that someone burgled the house and too/ the time to have a bite hile on the
1ob, or a house2mate might have arranged the things on the table ithout having a
midnight snac/ but 1ust to ma/e you believe that someone had a midnight snac/.
But these hypotheses stri/e you as providing much more contrived explanations of
the data than the one you infer to.
3al/ing along the beach, you see hat loo/s li/e a picture of 3inston *hurchill in
the sand. It could be that, as in the opening pages of Hilary 4utnam5s 6#78#9, hat
you see is actually the trace of an ant craling on the beach. !he much simpler, and
therefore 6you thin/9 much better, explanation is that someone intentionally dre a
picture of *hurchill in the sand. !hat, in any case, is hat you come aay
In these examples, the conclusions do not follo logically from the premises. :or
instance, it does not follo logically that !im and Harry are friends again from the
premises that they had a terrible ro hich ended their friendship and that they
have 1ust been seen 1ogging together" it does not even follo, e may suppose,
from all the information you have about !im and Harry. 0or do you have any
useful statistical data about friendships, terrible ros, and 1oggers that might
arrant an inference from the information that you have about !im and Harry to
the conclusion that they are friends again, or even to the conclusion that, probably
6or ith a certain probability9, they are friends again. 3hat leads you to the
conclusion, and hat according to a considerable number of philosophers may also
arrant this conclusion, is precisely the fact that !im and Harry5s being friends
again ould, if true, best explain the fact that they have 1ust been seen 1ogging
together. 6!he proviso that a hypothesis be true if it is to explain anything is ta/en
as read from here on.9 )imilar remar/s apply to the other to examples. !he type
of inference exhibited here is called abduction or, somehat more commonly
noadays, Inference to the Best Explanation.
1.1 Deduction, induction, abduction
Abduction is normally thought of as being one of three ma1or types of inference,
the other to being deduction and induction. !he distinction beteen deduction, on
the one hand, and induction and abduction, on the other hand, corresponds to the
distinction beteen necessary and non2necessary inferences. In deductive
inferences, hat is inferred is necessarily true if the premises from hich it is
inferred are true" that is, the truth of the premises guarantees the truth of the
conclusion. A familiar type of example is inferences instantiating the schema
All s are Bs.
a is an .
Hence, a is a B.
But not all inferences are of this variety. *onsider, for instance, the inference of
;<ohn is rich= from ;<ohn lives in *helsea= and ;Most people living in *helsea are
rich.= Here, the truth of the first sentence is not guaranteed 6but only made li/ely9
by the 1oint truth of the second and third sentences. &ifferently put, it is not
necessarily the case that if the premises are true, then so is the conclusion$ it is
logically compatible ith the truth of the premises that <ohn is a member of the
minority of non2rich inhabitants of *helsea. !he case is similar regarding your
inference to the conclusion that !im and Harry are friends again on the basis of the
information that they have been seen 1ogging together. 4erhaps !im and Harry are
former business partners ho still had some financial matters to discuss, hoever
much they ould have li/ed to avoid this, and decided to combine this ith their
daily exercise" this is compatible ith their being firmly decided never to ma/e up.
)ince *harles )anders 4eirce, it is standard practice to group non2necessary
inferences into inducti!e and abducti!e ones>see the
)upplement$ 4eirce on Abduction.
Inductive inferences form a somehat heterogeneous class, but for present
purposes they may be characteri?ed as those inferences that are based purely on
statistical data, such as observed frequencies of occurrences of a particular feature
in a given population. An example of such an inference ould be this$
7@ per cent of the :lemish college students spea/ both &utch and :rench.
Aouise is a :lemish college student.
Hence, Aouise spea/s both &utch and :rench.
Hoever, the relevant statistical information may also be more vaguely given, as in
the premise, ;Most people living in *helsea are rich.= 6!here is much discussion
about hether the conclusion of an inductive argument can be stated in purely
qualitative terms or hether it should be a quantitative one>for instance, that it
holds ith a probability of .7@ that Aouise spea/s both &utch and :rench>or
hether it can so"eti"es be stated in qualitative terms>for instance, if the
probability that it is true is high enough>and sometimes not. ,n these and other
issues related to induction, see Byburg #77C 6*h. +9. It should also be mentioned
that Harman 6#7@D9 conceives induction as a special type of abduction.9
!he mere fact that an inference is based on statistical data is not enough to classify
it as an inductive one. .ou may have observed many gray elephants and no non2
gray ones, and infer from this that all elephants are gray, because that
#ould pro!ide the best explanation for #hy you ha!e obser!ed so "any gray
elephants and no non$gray ones. !his ould be an instance of an abductive
inference. It suggests that the best ay to distinguish beteen induction and
abduction is this$ both are a"pliati!e, meaning that the conclusion goes beyond
hat is 6logically9 contained in the premises 6hich is hy they are non2necessary
inferences9, but in abduction there is an implicit or explicit appeal to explanatory
considerations, hereas in induction there is not" in induction, there is only an
appeal to observed frequencies or statistics. 6I emphasi?e ;only,= because in
abduction there may also be an appeal to frequencies or statistics, as the example
about the elephants exhibits.9
A noteorthy feature of abduction, hich it shares ith induction but not ith
deduction, is that it violates "onotonicity, meaning that it may be possible to infer
abductively certain conclusions from a subset of a set % of premises hich cannot
be inferred abductively from % as a hole. :or instance, adding the premise that
!im and Harry are former business partners ho still have some financial matters
to discuss, to the premises that they had a terrible ro some time ago and that they
ere 1ust seen 1ogging together may no longer arrant you to infer that they are
friends again, even if>let us suppose>the last to premises alone do arrant that
inference. !he reason is that hat counts as the best explanation of !im and Harry5s
1ogging together in light of the original premises may no longer do so once the
information has been added that they are former business partners ith financial
matters to discuss.
1.2 The ubiquity of abduction
!he type of inference exemplified in the cases described at the beginning of this
entry ill stri/e most as entirely familiar. 4hilosophers as ell as psychologists
tend to agree that abduction is frequently employed in everyday reasoning.
)ometimes our reliance on abductive reasoning is quite obvious and explicit. But in
some daily practices, it may be so routine and automatic that it easily goes
unnoticed. A case in point may be our trust in other people5s testimony, hich has
been said to rest on abductive reasoning" see Harman #7@D, Adler #77+, :ric/er
#77+, and Aipton #778 for defenses of this claim. :or instance, according to
<onathan Adler 6#77+, 'E+f9, ;FtGhe best explanation for hy the informant asserts
that & is normally that H he believes it for duly responsible reasons and H he
intends that I shall believe it too,= hich is hy e are normally 1ustified in
trusting the informant5s testimony. !his may ell be correct, even though in
coming to trust a person5s testimony one does not normally seem to be aare of any
abductive reasoning going on in one5s mind. )imilar remar/s may apply to hat
some hold to be a further, possibly even more fundamental, role of abduction in
linguistic practice, to it, its role in determining hat a spea/er means by an
utterance. )pecifically, it has been argued that decoding utterances is a matter of
inferring the best explanation of hy someone said hat he or she said in the
context in hich the utterance as made. Even more specifically, authors or/ing
in the field of pragmatics have suggested that hearers invo/e the %ricean maxims
of conversation to help them or/ out the best explanation of a spea/er5s utterance
henever the semantic content of the utterance is insufficiently informative for the
purposes of the conversation, or is too informative, or off2topic, or implausible, or
otherise odd or inappropriate" see, for instance, Bach and Harnish #7E7 67'f9,
&ascal #7E7 6#@E9, and Hobbs 'CC+. As in cases of reliance on spea/er testimony,
the requisite abductive reasoning ould normally seem to ta/e place at a
subconscious level.
Abductive reasoning is not limited to everyday contexts. Iuite the contrary$
philosophers of science have argued that abduction is a cornerstone of scientific
methodology" see, for instance, Boyd #78#, #78+, HarrJ #78@, #788, Aipton #77#,
'CC+, and 4sillos #777. Ernan McMullin 6#77'9 even goes so far as to call
abduction ;the inference that ma/es science.= !o illustrate the use of abduction in
science, e consider to examples.
At the beginning of the nineteenth century, it as discovered that the orbit of
Kranus, one of the seven planets /non at the time, departed from the orbit as
predicted on the basis of Isaac 0eton5s theory of universal gravitation and the
auxiliary assumption that there ere no further planets in the solar system. ,ne
possible explanation as, of course, that 0eton5s theory is false. %iven its great
empirical successes for 6then9 more than to centuries, that did not appear to be a
very good explanation. !o astronomers, <ohn *ouch Adams and Krbain
Aeverrier, instead suggested 6independently of each other but almost
simultaneously9 that there as an eighth, as yet undiscovered planet in the solar
system" that, they thought, provided the best explanation of Kranus5 deviating orbit.
0ot much later, this planet, hich is no /non as ;0eptune,= as discovered.
!he second example concerns hat is no commonly regarded to have been the
discovery of the electron by the English physicist <oseph <ohn !homson. !homson
had conducted experiments on cathode rays in order to determine hether they are
streams of charged particles. He concluded that they are indeed, reasoning as
As the cathode rays carry a charge of negative electricity, are deflected by an
electrostatic force as if they ere negatively electrified, and are acted on by a
magnetic force in 1ust the ay in hich this force ould act on a negatively
electrified body moving along the path of these rays, I can see no escape from the
conclusion that they are charges of negative electricity carried by particles of
matter. 6!homson, cited in Achinstein 'CC#, #E9
!he conclusion that cathode rays consist of negatively charged particles does not
follo logically from the reported experimental results, nor could !homson dra
on any relevant statistical data. !hat nevertheless he could ;see no escape from the
conclusion= is, e may safely assume, because the conclusion is the best>in this
case presumably even the only plausible>explanation of his results that he could
thin/ of.
Many other examples of scientific uses of abduction have been discussed in the
literature" see, for instance, HarrJ #78@, #788 and Aipton #77#, 'CC+. Abduction is
also said to be the predominant mode of reasoning in medical diagnosis$ physicians
tend to go for the hypothesis that best explains the patient5s symptoms 6see
<osephson and <osephson 6eds.9 #77+, 7L#'9.
Aast but not least, abduction plays a central role in some important philosophical
debates. Arguably, its most notable role is in ob1ections to so2called
underdetermination arguments. Knderdetermination arguments generally start from
the premise that a number of given hypotheses are empirically equivalent, hich
their authors ta/e to mean that the evidence>indeed, any evidence e might ever
come to possess>is unable to favor one of them over the others. :rom this, e are
supposed to conclude that one can never be arranted in believing any particular
one of the hypotheses. 6!his is rough, but it ill do for present purposes" see
&ouven 'CC8, and )tanford 'CC7, for more detailed accounts of underdetermination
arguments.9 A famous instance of this type of argument is the *artesian argument
for global s/epticism, according to hich the hypothesis that reality is more or less
the ay e customarily deem it to be is empirically equivalent to a variety of so2
called s/eptical hypotheses 6such as that e are beguiled by an evil demon, or that
e are brains in a vat, connected to a supercomputer9. )imilar arguments have been
given in support of scientific antirealism, according to hich it ill never be
arranted for us to choose beteen empirically equivalent rivals concerning hat
underlies the observable part of reality.
-esponses to these arguments typically point to the fact that the notion of empirical
equivalence at play unduly neglects explanatory considerations, for instance, by
defining the notion strictly in terms of hypotheses5 ma/ing the same predictions.
!hose responding then argue that even if some hypotheses ma/e exactly the same
predictions, one of them may still be a better explanation of the phenomena
predicted. !hus, if explanatory considerations have a role in determining hich
inferences e are licensed to ma/e>as according to defenders of abduction they
have>then e might still be arranted in believing in the truth 6or probable truth,
or some such, depending>as ill be seen belo>on the version of abduction one
assumes9 of one of a number of hypotheses that all ma/e the same predictions.
:olloing Bertrand -ussell 6#7#', *h. '9, many epistemologists have invo/ed
abduction in arguing against *artesian s/epticism, their /ey claim being that even
though, by construction, the s/eptical hypotheses ma/e the same predictions as the
hypothesis that reality is more or less the ay e ordinarily ta/e it to be, they are
not equally good explanations of hat they predict" in particular, the s/eptical
hypotheses have been said to be considerably less simple than the ;ordinary orld=
hypothesis" see, among many others, Harman #7E( 6*hs. 8 and ##9, %oldman #788
6'CD9, Moser #787 6#@#9, and Mogel #77C, 'CCD. )imilarly, philosophers of science
have argued that e are arranted to believe in )pecial -elativity !heory as
opposed to Aorent?5s version of the Nther theory. :or even though these theories
ma/e the same predictions, the former is explanatorily superior to the latter. 6Most
arguments that have been given for this claim come don to the contention that
)pecial -elativity !heory is ontologically more parsimonious than its competitor,
hich postulates the existence of an Nther. )ee <anssen 'CC' for an excellent
discussion of the various reasons philosophers of science have adduced for
preferring Einstein5s theory to Aorent?5s.9
2. Explicating Abduction
4recise statements of hat abduction amounts to are rare in the literature on
abduction. 64eirce did propose an at least fairly precise statement" but, as explained
in the supplement to this entry, it does not capture hat most noadays understand
by abduction.9 Its core idea is often said to be that explanatory considerations have
confirmation2theoretic import, or that explanatory success is a 6not necessarily
unfailing9 mar/ of truth. *learly, hoever, these formulations are slogans at best,
and it ta/es little effort to see that they can be cashed out in a great variety of prima
facie plausible ays. Here e ill consider a number of such possible explications,
starting ith hat one might term the ;textboo/ version of abduction,= hich, as
ill be seen, is manifestly defective, and then going on to consider various possible
refinements of it. 3hat those versions have in common>unsurprisingly>is that
they are all inference rules, requiring premises encompassing explanatory
considerations and yielding a conclusion that ma/es some statement about the truth
of a hypothesis. !he differences concern the premises that are required, or hat
exactly e are alloed to infer from them 6or both9.
In textboo/s on epistemology or the philosophy of science, one often encounters
something li/e the folloing as a formulation of abduction$
%iven evidence E and candidate explanations '
,H, '
of E, infer the truth
of that '
hich best explains E.
An observation that is frequently made about this rule, and that points to
a potential problem for it, is that it presupposes the notions of candidate
explanation and best explanation, neither of hich has a straightforard
interpretation. 3hile some still hope that the former can be spelled out in
purely logical, or at least purely formal, terms, it is often said that the
latter must appeal to the so2called theoretical virtues, li/e simplicity,
generality, and coherence ith ell2established theories" the best
explanation ould then be the hypothesis hich, on balance, does best
ith respect to these virtues. 6)ee, for instance, !hagard #7E8 and
McMullin #77@.9 !he problem is that none of the said virtues is presently
particularly ell understood. 6%iere, in *allebaut 6ed.9 #77( 6'('9, even
ma/es the radical claim that the theoretical virtues lac/ real content and
play no more than a rhetorical role in science. In vie of recent formal
or/ both on simplicity and on coherence>for instance, :orster and
)ober #77+, and Ai and Mitanyi #77E, on simplicity and Bovens and
Hartmann 'CC(, on coherence>the first part of this claim has become
hard to maintain. 4sychological evidence casts doubt on the second part
of the claim" see, for instance, Aombro?o 'CCE, on the role of simplicity
in people5s assessments of explanatory goodness and Boslos/i et al.
'CC8, on the role of coherence ith bac/ground /noledge in those
:urthermore, many of those ho thin/ AB&# is headed along the right
lines believe that it is too strong. )ome thin/ that abduction arrants an
inference only to the probable truth of the best explanation, others that it
arrants an inference only to the approxi"ate truth of the best
explanation, and still others that it arrants an inference only to
the probableapproxi"ate truth.
!he real problem ith AB&# runs deeper than this, hoever. Because
abduction is ampliative>as explained earlier>it ill not be a sound rule
of inference in the strict logical sense, hoever abduction is explicated
exactly. It can still be reliable in that it mostly leads to a true conclusion
henever the premises are true. An obvious necessary condition for
AB&# to be reliable in this sense is that, "ostly, hen it is true
that ' best explains E, and E is true, then ' is true as ell 6or ' is
approximately true, or probably true, or probably approximately true9.
But this ould not be enough for AB&# to be reliable. :or AB&# ta/es
as its premise only that some hypothesis is the best explanation of the
evidence as co"pared to other hypotheses in a gi!en set. !hus, if the rule
is to be reliable, it must hold that, at least typically, the best explanation
relative to the set of hypotheses that e consider ould also come out as
being best in comparison ith any other hypotheses that e might have
conceived 6but for lac/ of time or ingenuity, or for some other reason, did
not conceive9. In other ords, it must hold that at least typically
the absolutely best explanation of the evidence is to be found among the
candidate explanations e have come up ith, for else AB&# may ell
lead us to believe ;the best of a bad lot= 6van :raassen #787, #+(9.
Ho reasonable is it to suppose that this extra requirement is usually
fulfilledO 0ot at all, presumably. !o believe otherise, e must assume
some sort of privilege on our part to the effect that hen e consider
possible explanations of the data, e are someho predisposed to hit,
inter alia, upon the absolutely best explanation of those data. After all,
hardly ever ill e have considered, or ill it even be possible to
consider, all potential explanations. As van :raassen 6#787, #++9 points
out, it is a priori rather implausible to hold that e are thus privileged.
In response to this, one might argue that the challenge to sho that the
best explanation is alays or mostly among the hypotheses considered
can be met ithout having to assume some form of privilege. :or given
the hypotheses e have managed to come up ith, e can alays
generate a set of hypotheses hich 1ointly exhaust logical space.
)uppose '
are the candidate explanations e have so far been able
to conceive. !hen simply define '
$Q R'
and add this
ne hypothesis as a further candidate explanation to the ones e already
have. ,bviously, the set S'
T is exhaustive, in that one of its
elements must be true. :olloing this in itself simple procedure ould
seem enough to ma/e sure that e never miss out on the absolutely best
explanation. 6)ee Aipton #77(, for a proposal along these lines.9
Alas, there is a catch. :or even though there may be many
hypotheses '
that imply '
and, had they been formulated, ould have
been evaluated as being a better explanation for the data than the best
explanation among the candidate explanations e started out
ith, '
itself ill in general be hardly informative" in fact, in general it
ill not even be clear hat its empirical consequences are. )uppose, for
instance, e have as competing explanations )pecial -elativity !heory
and Aorent?5s version of the Nther theory. !hen, folloing the above
proposal, e may add to our candidate explanations that neither of these
to theories is true. But surely this further hypothesis ill be ran/ed
quite lo )ua explanation>if it ill be ran/ed at all, hich seems
doubtful, given that it is holly unclear hat its empirical consequences
are. !his is not to say that the suggested procedure may never or/. !he
point is that in general it ill give little assurance that the best
explanation is among the candidate explanations e consider.
A more promising response to the above ;argument of the bad lot= begins
ith the observation that the argument capitali?es on a peculiar
asymmetry or incongruence in AB&#. !he rule gives license to an
absolute conclusion>that a given hypothesis is true>on the basis of a
comparative premise, namely, that that particular hypothesis is the best
explanation of the evidence relative to the other hypotheses available 6see
Buipers 'CCC, #E#9. !his incongruence is not avoided by replacing
;truth= ith ;probable truth= or ;approximate truth.= In order to avoid it,
one has to general options.
!he first option is to modify the rule so as to have it require an absolute
premise. :or instance, folloing Alan Musgrave 6#7889 or 4eter Aipton
6#77(9, one may require the hypothesis hose truth is inferred to be not
only the best of the available potential explanations, but also to
be satisfactory 6Musgrave9 or good enough 6Aipton9, yielding the
folloing variant of AB&#$
%iven evidence E and candidate explanations '
,H, '
of E, infer the truth
of that '
hich explains E best, provided '
is satisfactoryUgood
enough )ua explanation.
0eedless to say, AB&' needs supplementing by a criterion for the satisfactoriness
of explanations, or their being good enough, hich, hoever, e are still lac/ing.
)econdly, one can formulate a symmetric or congruous version of abduction by
having it sanction, given a comparative premise, only a comparative conclusion"
this option, too, can in turn be reali?ed in more than one ay. Here is one ay to do
it, hich has been proposed and defended in the or/ of !heo Buipers 6e.g.,
Buipers #78+, #77', 'CCC9.
%iven evidence E and candidate explanations '
,H, '
of E, if '
explains E better
than any of the other hypotheses, infer that '
is closer to the truth than any of the
other hypotheses.
*learly, AB&( requires an account of closeness to the truth, but many such
accounts are on offer today.
,ne noteorthy feature of the congruous versions of abduction considered here is
that they do not rely on the assumption of an implausible privilege on the reasoner5s
part that, e sa, AB&# implicitly relies on. Another is that if one can be certain
that, hoever many candidate explanations for the data one may have missed, none
equals the best of those one hasthought of, then the congruous versions license
exactly the same inference as AB&# does 6supposing that one ould not be certain
that no potential explanation is as good as the best explanation one has thought of if
the latter is not even satisfactory or sufficiently good9.
As mentioned, there is idespread agreement that people frequently rely on
abductive reasoning. 3hich of the above rules exactly is it that people rely onO ,r
might it be still some further rule that they rely onO ,r might they in some contexts
rely on one version, and in others on anotherO 4hilosophical argumentation is
unable to anser these questions. And hile experimental psychologists have
started paying attention to the role humans give to explanatory considerations in
reasoning, so far there is nothing to be found in the literature that gives any
indication as to hat the ansers should be. 3ith respect to the normative question
of hich of the above rules e ought to rely on 6if e ought to rely on any form of
abduction9, here philosophical argumentation should be able to help, the situation
is hardly any better. In vie of the argument of the bad lot, AB&# does not loo/
very good. ,ther arguments against abduction are claimed to be independent of the
exact explication of the rule" belo, these arguments ill be found anting. ,n the
other hand, arguments that have been given in favor of abduction>some of hich
ill also be discussed belo>do not discern beteen specific versions. )o,
supposing people do indeed commonly rely on abduction, it must be considered an
open question as to hich version6s9 of abduction they rely on. Equally, supposing
it is rational for people to rely on abduction, it must be considered an open question
as to hich version, or perhaps versions, of abduction they ought to, or are at least
permitted to, rely on.
. The !tatu" of Abduction
Even if it is true that e routinely rely on abductive reasoning, it may still be as/ed
hether this practice is rational. :or instance, experimental studies have shon that
hen people are able to thin/ of an explanation for some possible event, they tend
to overestimate the li/elihood that this event ill actually occur. 6)ee Boehler
#77#, for a survey of some of these studies" see also Brem and -ips 'CCC.9 More
telling still, !ania Aombro?o 6'CCE9 shos that, in some situations, people tend to
grossly overrate the probability of simpler explanations compared to more
complicated ones. Although these studies are not directly concerned ith abduction
in any of the forms discussed so far, they nevertheless suggest that ta/ing into
account explanatory considerations in one5s reasoning may not alays be for the
better. 6It is to be noted that Aombro?o5s experiments are directly concerned ith
some proposals that have been made for explicating abduction in a Bayesian
frameor/" see )ection +.9 Hoever, the most pertinent remar/s about the
normative status of abduction are so far to be found in the philosophical literature.
!his section discusses the main criticisms that have been levelled against
abduction, as ell as the strongest arguments that have been given in its defense.
.1 #ritici"$"
3e have already encountered the so2called argument of the bad lot, hich, e sa,
is valid as a criticism of AB&# but poerless against various 6hat e called9
congruous rules of abduction. 3e here consider to ob1ections that are meant to be
more general. !he first even purports to challenge the core idea underlying
abduction" the second is not quite as general, but it is still meant to undermine a
broad class of candidate explications of abduction. Both ob1ections are due to Bas
van :raassen.
!he first ob1ection has as a premise that it is part of the meaning of ;explanation=
that if one theory is more explanatory than another, the former must be more
informative than the latter 6see, e.g., van :raassen #78(, )ect. '9. !he alleged
problem then is that it is ;an elementary logical point that a more informative
theory cannot be more li/ely to be true Fand thusG attempts to describe inductive or
evidential support through features that require information 6such as VInference to
the Best ExplanationW9 must either contradict themselves or equivocate= 6van
:raassen #787, #7'9. !he elementary logical point is supposed to be ;most
FobviousG H in the paradigm case in hich one theory is an extension of another$
clearly the extension has more ays of being false= 6van :raassen #78D, '8C9.
It is important to note, hoever, that in any other /ind of case than the ;paradigm=
one, the putative elementary point is not obvious at all. :or instance, it is entirely
unclear in hat sense )pecial -elativity !heory ;has more ays of being false=
than Aorent?5s version of the Nther theory, given that they ma/e the same
predictions. And yet the former is generally regarded as being
superior, )ua explanation, to the latter. 6If van :raassen ere to ob1ect that the
former is not really more informative than the latter, or at any rate not more
informative in the appropriate sense>hatever that is>then e should certainly
refuse to grant the premise that in order to be more explanatory a theory must be
more informative.9
!he second ob1ection, proffered in van :raassen #787 6*h. @9, is levelled at
probabilistic versions of abduction. !he ob1ection is that such rules must either
amount to Bayes5 rule, and thus be redundant, or be at variance ith it but then, on
the grounds of Aeis5 dynamic &utch boo/ argument 6as reported in !eller #7E(9,
be probabilistically incoherent, meaning that they may lead one to assess as fair a
number of bets hich together ensure a financial loss, come hat may" and, van
:raassen argues, it ould be irrational to follo a rule that has this feature.
Hoever, this ob1ection fares no better than the first. :or one thing, as 4atric/
Maher 6#77'9 and Brian )/yrms 6#77(9 have pointed out, a loss in one respect may
be outeighed by a benefit in another. It might be, for instance, that some
probabilistic version of abduction does much better, at least in our orld, than
Bayes5 rule, in that, on average, it approaches the truth faster in the sense that it is
faster in assigning a high probability 6understood as probability above a certain
threshold value9 to the true hypothesis. If it does, then folloing that rule instead of
Bayes5 rule may have advantages hich perhaps are not so readily expressed in
terms of money yet hich should arguably be ta/en into account hen deciding
hich rule to go by. It is, in short, not so clear hether folloing a probabilistically
incoherent rule must be irrational.
:or another thing, Igor &ouven 6#7779 argues that the question of hether a
probabilistic rule is coherent is not one that can be settled independently of
considering hich other epistemic and decision2theoretic rules are deployed along
ith it" coherence should be understood as a property of pac/ages of both epistemic
and decision2theoretic rules, not of epistemic rules 6such as probabilistic rules for
belief change9 in isolation. In the same paper, a coherent pac/age of rules is
described hich includes a probabilistic version of abduction. 6)ee Bvanvig #77+,
Harman #77E, Aeplin #77E, 0iiniluoto #777, and ,/asha 'CCC, for different
responses to van :raassen5s critique of probabilistic versions of abduction.9
.2 Defen"e"
Hardly anyone noadays ould ant to subscribe to a conception of truth that
posits a necessary connection beteen explanatory force and truth>for instance,
because it stipulates explanatory superiority to be necessary for truth. As a result, a
priori defenses of abduction seem out of the question. Indeed, all defenses that have
been given so far are of an empirical nature in that they appeal to data that
supposedly support the claim that 6in some form9 abduction is a reliable rule of
!he best2/non argument of this sort as developed by -ichard Boyd in the #78Cs
6see Boyd #78#, #78+, #78D9. It starts by underlining the theory2dependency of
scientific methodology, hich comprises methods for designing experiments, for
assessing data, for choosing beteen rival hypotheses, and so on. :or instance, in
considering possible confounding factors from hich an experimental setup has to
be shielded, scientists dra heavily on already accepted theories. !he argument
next calls attention to the apparent reliability of this methodology, hich, after all,
has yielded, and continues to yield, impressively accurate theories. In particular, by
relying on this methodology, scientists have for some time no been able to find
ever more instrumentally adequate theories. Boyd then argues that the reliability of
scientific methodology is best explained by assuming that the theories on hich it
relies are at least approximately true. :rom this and from the fact that these theories
ere mostly arrived at by abductive reasoning, he concludes that abduction must be
a reliable rule of inference.
*ritics have accused this argument of being circular. )pecifically, it has been said
that the argument rests on a premise>that scientific methodology is informed by
approximately true bac/ground theories>hich in turn rests on an inference to the
best explanation for its plausibility. And the reliability of this type of inference is
precisely hat is at sta/e. 6)ee, for instance, Aaudan #78# and :ine #78+.9
!o this, )tathis 4sillos 6#777, *h. +9 has responded by invo/ing a distinction
credited to -ichard Braithaite, to it, the distinction beteen premise2circularity
and rule2circularity. An argument is premise2circular if its conclusion is amongst its
premises. A rule2circular argument, by contrast, is an argument of hich the
conclusion asserts something about an inferential rule that is used in the very same
argument. As 4sillos urges, Boyd5s argument is rule2circular, but not premise2
circular, and rule2circular arguments, 4sillos contends, need not be viciously
circular 6even though a premise2circular argument is alays viciously circular9. !o
be more precise, in his vie, an argument for the reliability of a given rule * that
essentially relies on * as an inferential principle is not vicious, provided that the
use of * does not guarantee a positive conclusion about *5s reliability. 4sillos
claims that in Boyd5s argument, this proviso is met. :or hile Boyd concludes that
the bac/ground theories on hich scientific methodology relies are approximately
true on the basis of an abductive step, the use of abduction itself does not guarantee
the truth of his conclusion. After all, granting the use of abduction does nothing to
ensure that the best explanation of the success of scientific methodology is the
approximate truth of the relevant bac/ground theories. !hus, 4sillos concludes,
Boyd5s argument still stands.
Even if the use of abduction in Boyd5s argument might have led to the conclusion
that abduction is not reliable, one may still have orries about the argument5s being
rule2circular. :or suppose that some scientific community relied not on abduction
but on a rule that e may dub ;Inference to the 3orst Explanation= 6I3E9, a rule
that sanctions inferring to the #orstexplanation of the available data. 3e may
safely assume that the use of this rule mostly ould lead to the adoption of very
unsuccessful theories. 0evertheless, the said community might 1ustify its use of
I3E by dint of the folloing reasoning$ ;)cientific theories tend to be hugely
unsuccessful. !hese theories ere arrived at by application of I3E. !hat I3E is a
reliable rule of inference>that is, a rule of inference mostly leading from true
premises to true conclusions>is surely the orst explanation of the fact that our
theories are so unsuccessful. Hence, by application of I3E, e may conclude that
I3E is a reliable rule of inference.= 3hile this ould be an utterly absurd
conclusion, the argument leading up to it cannot be convicted of being viciously
circular anymore than Boyd5s argument for the reliability of abduction can 6if
4sillos is right9. It ould appear, then, that there must be something else amiss ith
It is fair to note that for 4sillos, the fact that a rule2circular argument does not
guarantee a positive conclusion about the rule at issue is not sufficient for such an
argument to be valid. A further necessary condition is ;that one should not have
reason to doubt the reliability of the rule>that there is nothing currently available
hich can ma/e one distrust the rule= 64sillos #777, 8D9. And there is plenty of
reason to doubt the reliability of I3E" in fact, the above argument supposes that it
is unreliable. !o questions arise, hoever. :irst, hy should e accept the
additional conditionO )econd, do e really have no reason to doubt the reliability
of abductionO *ertainly so"e of the abductive inferences e ma/e lead us to
acceptfalsehoods. Ho many falsehoods may e accept on the basis of abduction
before e can legitimately begin to distrust this ruleO 0o clear ansers have been
given to these questions.
Be this as it may, even if rule2circularity is neither vicious nor otherise
problematic, one may still onder ho Boyd5s argument is to convert a critic of
abduction, given that it relies on abduction. But 4sillos ma/es it clear that the point
of philosophical argumentation is not alays, and in any case need not be, to
convince an opponent of one5s position. )ometimes the point is, more modestly, to
assure or reassure oneself that the position one endorses, or is tempted to endorse,
is correct. In the case at hand, e need not thin/ of Boyd5s argument as an attempt
to convince the opponent of abduction of its reliability. -ather, it may be thought of
as 1ustifying the rule from ithin the perspective of someone ho is already
sympathetic toards abduction" see 4sillos #777 6879.
!here have also been attempts to argue for abduction in a more straightforard
fashion, to it, via enumerative induction. !he common idea of these attempts is
that every nely recorded successful application of abduction>li/e the discovery
of 0eptune, hose existence had been postulated on explanatory grounds 6see
)ection #.'9>adds further support to the hypothesis that abduction is a reliable rule
of inference, in the ay in hich every nely observed blac/ raven adds some
support to the hypothesis that all ravens are blac/. Because it does not involve
abductive reasoning, this type of argument is more li/ely to also appeal to
disbelievers in abduction. )ee HarrJ #78@, #788, Bird #778 6#@C9, Bitcher 'CC#,
and &ouven 'CC' for suggestions along these lines.
%. Abduction &er"u" 'aye"ian #onfir$ation Theory
In the past decade, Bayesian confirmation theory has firmly established itself as the
dominant vie on confirmation" currently one cannot very ell discuss a
confirmation2theoretic issue ithout ma/ing clear hether, and if so hy, one5s
position on that issue deviates from standard Bayesian thin/ing. Abduction, in
hichever version, assigns a confirmation2theoretic role to explanation$
explanatory considerations contribute to ma/ing some hypotheses more credible,
and others less so. By contrast, Bayesian confirmation theory ma/es no reference at
all to the concept of explanation. &oes this imply that abduction is at loggerheads
ith the prevailing doctrine in confirmation theoryO )everal authors have recently
argued that not only is abduction compatible ith Bayesianism, it is a much2needed
supplement to it. !he so far fullest defense of this vie has been given by Aipton
6'CC+, *h. E9" as he puts it, Bayesians should also be ;explanationists= 6his name
for the advocates of abduction9. 6:or other defenses, see ,/asha 'CCC, Mc%re
'CC(, and 3eisberg 'CC7.9
!his requires some clarification. :or hat could it mean for a Bayesian to be an
explanationistO In order to apply Bayes5 rule and determine the probability
for ' after learning E, the Bayesian agent ill have to determine the probability
of ' conditional on E. :or that, he needs to assign unconditional probabilities
to ' and E as ell as a probability to E given '" the former to are mostly called
;prior probabilities= 6or 1ust ;priors=9 of, respectively, ' and E, the latter the
;li/elihood= of ' on E. 6!his is the official Bayesian story. 0ot all of those ho
sympathi?e ith Bayesianism adhere to that story. :or instance, according to some
it is more reasonable to thin/ that conditional probabilities are basic and that e
derive unconditional probabilities from them" see HX1e/ 'CC(, and references
therein.9 Ho is the Bayesian to determine these valuesO As is ell /non,
probability theory gives us more probabilities once e have some" it does not give
us probabilities from scratch. ,f course, hen ' implies E or the negation of E, or
hen ' is a statistical hypothesis that bestos a certain chance on E, then the
li/elihood follos ;analytically.= 6!his claim assumes some version of Aeis5
6#78C9 4rincipal 4rinciple, and it is controversial hether or not this principle is
analytic" hence the scare quotes.9 But this is not alays the case, and even if it
ere, there ould still be the question of ho to determine the priors. !his is
here, according to Aipton, abduction comes in. In his proposal, Bayesians ought
to determine their prior probabilities and, if applicable, li/elihoods on the basis of
explanatory considerations.
Exactly ho are explanatory considerations to guide one5s choice of priorsO !he
anser to this question is not as simple as one might at first thin/. )uppose you are
considering hat priors to assign to a collection of rival hypotheses and you ish
to follo Aipton5s suggestion. Ho are you to do thisO An obvious>though still
somehat vague>anser may seem to go li/e this$ 3hatever exact priors you are
going to assign, you should assign a higher one to the hypothesis that explains the
available data best than to any of its rivals 6provided there is a best explanation9.
0ote, though, that your neighbor, ho is a Bayesian but thin/s confirmation has
nothing to do ith explanation, may ell assign a prior to the best explanation that
is even higher than the one you assign to that hypothesis. In fact, his priors for best
explanations may even be consistently higher than yours, not because in his vie
explanation is someho related to confirmation>it is not, he thin/s>but, ell,
1ust because. In this context, ;1ust because= is a perfectly legitimate reason, because
any reason for fixing one5s priors counts as legitimate by Bayesian standards.
According to mainstream Bayesian epistemology, priors 6and sometimes
li/elihoods9 are up for grabs, meaning that one assignment of priors is as good as
another, provided both are coherent 6that is, they obey the axioms of probability
theory9. Aipton5s recommendation to the Bayesian to be an explanationist is meant
to be entirely general. But hat should your neighbor do differently if he ants to
follo the recommendationO )hould he give the same prior to any best explanation
that you, his explanationist neighbor, give to it, that is, lo#er his priors for best
explanationsO ,r rather should he give even higher priors to best explanations than
those he already givesO
4erhaps Aipton5s proposal is not intended to address those ho already assign
highest priors to best explanations, even if they do so on grounds that have nothing
to do ith explanation. !he idea might be that, as long as one does assign highest
priors to those hypotheses, everything is fine, or at least finer than if one does not
do so, regardless of one5s reasons for assigning those priors. !he anser to the
question of ho explanatory considerations are to guide one5s choice of priors
ould then presumably be that one ought to assign a higher prior to the best
explanation than to its rivals, if this is not hat one already does. If it is, one should
1ust /eep doing hat one is doing.
6As an aside, it should be noticed that, according to standard Bayesian usage, the
term ;priors= does not necessarily refer to the degrees of belief a person assigns
before the receipt ofany data. If there are already data in, then, clearly, one may
assign higher priors to hypotheses that best explain the then2available data.
Hoever, one can sensibly spea/ of ;best explanations= even before any data are
/non. :or example, one hypothesis may be 1udged to be a better explanation than
any of its rivals because the former requires less complicated mathematics, or
because it is stated in terms of familiar concepts only, hich is not true of the
others. More generally, such 1udgments may be based on hat Bosso 6#77', (C9
callsinternal features of hypotheses or theories, that is, features that ;can be
evaluated ithout having to observe the orld.=9
A more interesting anser to the above question of ho explanation is to guide
one5s choice of priors has been given by <onathan 3eisberg 6'CC79. 3e said that
mainstream Bayesians regard one assignment of prior probabilities as being as
good as any other. )o2called ob1ective Bayesians do not do so, hoever. !hese
Bayesians thin/ priors must obey principles beyond the probability axioms in order
to be admissible. ,b1ective Bayesians are divided among themselves over exactly
hich further principles are to be obeyed, but at least for a hile they agreed that
the 4rinciple of Indifference is among them. -oughly stated, this principle counsels
that, absent a reason to the contrary, e give equal priors to competing hypotheses.
As is ell /non, hoever, in its original form the 4rinciple of Indifference may
lead to inconsistent assignments of probabilities and so can hardly be advertised as
a principle of rationality. !he problem is that there are typically various ays to
partition logical space that appear plausible given the problem at hand, and that not
all of them lead to the same prior probability assignment, even assuming the
4rinciple of Indifference. 3eisberg5s proposal amounts to the claim that
explanatory considerations may favor some of those partitions over others. 4erhaps
e ill not alays end up ith a unique partition to hich the 4rinciple of
Indifference is to be applied, but it ould already be progress if e ended up ith
only a handful of partitions. :or e could then still arrive in a motivated ay at our
prior probabilities, by proceeding in to steps, namely, by first applying the
4rinciple of Indifference to the partitions separately, thereby possibly obtaining
different assignments of priors, and by then ta/ing a eighted average of the thus
obtained priors, here the eights, too, are to depend on explanatory
considerations. !he result ould again be a probability function>the uniquely
correct prior probability function, according to 3eisberg.
!he proposal is intriguing as far as it goes but, as 3eisberg admits, in its current
form, it does not go very far. :or one thing, it is unclear ho exactly explanatory
considerations are to determine the eights required for the second step of the
proposal. :or another, it may be idle to hope that ta/ing explanatory considerations
into account ill in general leave us ith a manageable set of partitions, or that,
even if it does, this ill not be due merely to the fact that e are overloo/ing a
great many prima facie plausible ays of partitioning logical space to begin ith.
6!he latter point echoes the argument of the bad lot, of course.9
Another suggestion about the connection beteen abduction and Bayesian
reasoning>to be found in ,/asha 'CCC, Mc%re 'CC(, and Aipton 'CC+ 6*h. E9>
is that the explanatory considerations may serve as a heuristic to determine, even if
only roughly, priors and li/elihoods in cases in hich e ould otherise be
clueless and could do no better than guessing. !his suggestion is sensitive to the
ell2recogni?ed fact that e are not alays able to assign a prior to every
hypothesis of interest, or to say ho probable a given piece of evidence is
conditional on a given hypothesis. *onsideration of that hypothesis5 explanatory
poer might then help us to figure out, if perhaps only ithin certain bounds, hat
prior to assign to it, or hat li/elihood to assign to it on the given evidence.
Bayesians, especially the more modest ones, might ant to retort that the Bayesian
procedure is to be folloed if, and only if, either 6a9 priors and li/elihoods can be
determined ith some precision and ob1ectivity, or 6b9 li/elihoods can be
determined ith some precision and priors can be expected to ;ash out= as more
and more evidence accumulates, or 6c9 priors and li/elihoods can both be expected
to ash out. In the remaining cases>they might say>e should simply refrain
from applying Bayesian reasoning. A fortiori, then, there is no need for an
abduction2enhanced Bayesianism in these cases. And some incontrovertible
mathematical results indicate that, in the cases that fall under 6a9, 6b9, or 6c9, our
probabilities ill converge to the truth anyho. *onsequently, in those cases there
is no need for the /ind of abductive heuristics that the above2mentioned authors
suggest, either. 63eisberg 'CC7, )ect. (.', raises similar concerns.9
4sillos 6'CCC9 proposes yet another ay in hich abduction might supplement
Bayesian confirmation theory, one that is very much in the spirit of 4eirce5s
conception of abduction. !he idea is that abduction may assist us in selecting
plausible candidates for testing, here the actual testing then is to follo Bayesian
lines. Hoever, 4sillos concedes 6'CC+9 that this proposal assigns a role to
abduction that ill stri/e committed explanationists as being too limited.
:inally, a possibility that has so far not been considered in the literature is that
abduction and Bayesianism do not so much or/ in tandem>as they do on the
above proposals>as operate in different modes of reasoning" the Bayesian and the
explanationist are characters that feature in different plays, so to spea/. It is idely
accepted that sometimes e spea/ and thin/ about our beliefs in a categorical
manner, hile at other times e spea/ and thin/ about them in a graded ay. It is
far from clear ho these different ays of spea/ing and thin/ing about beliefs>the
epistemology of belief and the epistemology of degrees of belief, to use -ichard
:oley5s 6#77'9 terminology>are related to one another. In fact, it is an open
question hether there is any straightforard connection beteen the to, or even
hether there is a connection at all. Be that as it may, given that the distinction is
undeniable, it is a plausible suggestion that, 1ust as there are different ays of
tal/ing and thin/ing about beliefs, there are different ays of tal/ing and thin/ing
about the re!ision of beliefs. In particular, abduction could ell have its home in
the epistemology of belief, and be called upon henever e reason about our
beliefs in a categorical mode, hile at the same time Bayes5 rule could have its
home in the epistemology of degrees of belief. Hard2nosed Bayesians may insist
that hatever reasoning goes on in the categorical mode must eventually be
1ustifiable in Bayesian terms, but this presupposes the existence of bridge principles
connecting the epistemology of belief ith the epistemology of degrees of belief>
and, as mentioned, hether such principles exist is presently unclear.
Achinstein, 4., 'CC#. +he Boo, of E!idence, ,xford$ ,xford Kniversity 4ress.
Adler, <., #77+. ;!estimony, !rust, Bnoing,= -ournal of &hilosophy, 7#$ '@+L'ED.
Bach, B. and Harnish, -., #7E7. .inguistic /o""unication and %peech cts,
*ambridge MA$ MI! 4ress.
Bird, A., #778. &hilosophy of %cience, Aondon$ K*A 4ress.
Bovens, A. and Hartmann, )., 'CC(. ;)olving the -iddle of *oherence,= Mind, ##'$
Boyd, -., #78#. ;)cientific -ealism and 0aturalistic Epistemology,= in 4. Asquith
and -. %iere 6eds.9, &% 1900, 6vol. II9, East Aansing MI$ 4hilosophy of )cience
Association, pp. @#(L@@'.
Boyd, -., #78+. ;!he *urrent )tatus of )cientific -ealism,= in <. Aeplin
6ed.9, %cientific *ealis", Ber/eley *A$ Kniversity of *alifornia 4ress, pp. +#L8'.
Boyd, -., #78D. ;Aex ,randi est Aex *redendi,= in 4. *hurchland and *. Hoo/er
6eds.9, I"ages of %cience, *hicago IA$ Kniversity of *hicago 4ress, pp. (L(+.
Brem, ). and -ips, A. <., 'CCC. ;Explanation and Evidence in Informal
Argument,= /ogniti!e %cience, '+$ DE(L@C+.
*allebaut, 3. 6ed.9, #77(. +a,ing the 1aturalistic +urn, *hicago IA$ Kniversity of
*hicago 4ress.
&ascal, M., #7E7. ;*onversational -elevance,= in A. Margalit 6ed.9, Meaning
and, 2se, &ordrecht$ -eidel, pp. #D(L#E+.
&ouven, I., #777. ;Inference to the Best Explanation Made
*oherent,= &hilosophy, of %cience, @@$ )+'+L)+(D.
&ouven, I., 'CC'. ;!esting Inference to the Best Explanation,= %ynthese, #(C$ (DDL
&ouven, I., 'CC8. ;Knderdetermination,= in ). 4sillos and M. *urd
6eds.9, +he, *outledge /o"panion to the &hilosophy of %cience, Aondon$
-outledge, pp. '7'L(C#.
:ann, B. !., #7EC. &eirce3s +heory of bduction, !he Hague$ Martinus 0i1hoff.
:ine, A., #78+. ;!he 0atural ,ntological Attitude,= in <. Aeplin
6ed.9, %cientific, *ealis", Ber/eley *A$ Kniversity of *alifornia 4ress, pp. 8(L#CE.
:oley, -., #77'. ;!he Epistemology of Belief and the Epistemology of &egrees of
Belief,= "erican &hilosophical 4uarterly, '7$ ###L#'+.
:orster, M. and )ober, E., #77+. ;Ho to !ell hen )impler, More Knified, or
Aess d 'oc, !heories ill 4rovide More Accurate 4redictions,= British -ournal
for, the &hilosophy of %cience, +D$ #L(@.
:ran/furt, H., #7D8. ;4eirce5s 0otion of Abduction,= -ournal of &hilosophy, DD$
:ric/er, E., #77+. ;Against %ullibility,= in B. B. Matilal and A. *ha/rabarti
6eds.9, 5no#ing fro" Words, &ordrecht$ Bluer, pp. #'DL#@#.
%oldman, A., #788. E"pirical 5no#ledge, Ber/eley *A$ Kniversity of *alifornia
HX1e/, A., 'CC(. ;3hat *onditional 4robability *ould 0ot Be,= %ynthese, #(E$
Harman, %., #7@D. ;!he Inference to the Best Explanation,= &hilosophical *e!ie#,
E+$ 88L7D.
Harman, %., #7E(. +hought, 4rinceton 0<$ 4rinceton Kniversity 4ress.
Harman, %., #77E. ;4ragmatism and -easons for Belief,= in *. Bulp
6ed.9, *ealis"6ntirealis" and Episte"ology, !otoa 0<$ -oman and Aittlefield,
pp. #'(L#+E.
HarrJ, -., #78@. 7arieties of *ealis", ,xford$ Blac/ell.
HarrJ, -., #788. ;-ealism and ,ntology,= &hilosophia 1aturalis, 'D$ (8@L(78.
Hobbs, <. -., 'CC+. ;Abduction in 0atural Aanguage Knderstanding,= in A. Horn
and %. 3ard 6eds.9, +he 'andboo, of &rag"atics, ,xford$ Blac/ell, pp. E'+L
<anssen, M., 'CC'. ;-econsidering a )cientific -evolution$ !he *ase of
Einstein !ersus, Aorent?,= &hysics in &erspecti!e, +$ +'#L++@.
<osephson, <. -. and <osephson, ). %. 6eds.9, #77+. bducti!e Inference,
*ambridge$ *ambridge Kniversity 4ress.
Bitcher, 4., 'CC#. ;-eal -ealism$ !he %alilean )trategy,= &hilosophical *e!ie#,
##C$ #D#L#7E.
Boehler, &. <., #77#. ;Explanation, Imagination, and *onfidence in
<udgment,= &sychological Bulletin, ##C$ +77LD#7.
Boslos/i, B., Marasia, <., *helen?a, M., and &ublin, -., 'CC8. ;Information
Becomes Evidence hen an Explanation *an Incorporate it into a *ausal
:rameor/,= /ogniti!e 8e!elop"ent, '($ +E'L+8E.
Bosso, 4., #77'. *eading the Boo, of 1ature, *ambridge$ *ambridge Kniversity
Buipers, !., #78+. ;Approaching the !ruth ith the -ule of
)uccess,= &hilosophia, 1aturalis, '#$ '++L'D(.
Buipers, !., #77'. ;0aive and -efined !ruth Approximation,= %ynthese, 7($ '77L
Buipers, !., 'CCC. Fro" Instru"entalis" to /onstructi!e *ealis", &ordrecht$
Bvanvig, <., #77+. ;A *ritique of van :raassen5s Moluntaristic
Epistemology,= %ynthese, 78$ ('DL(+8.
Byburg <r., H., #77C. %cience and *eason, ,xford$ ,xford Kniversity 4ress.
Aaudan, A., #78#. ;A *onfutation of *onvergent -ealism,= &hilosophy of %cience,
+8$ #7L+7.
Aeis, &., #78C. ;A )ub1ectivist5s %uide to ,b1ective *hance,= in -. <effrey
6ed.9, %tudies in Inducti!e .ogic and &robability, Ber/eley *A$ Kniversity of
*alifornia 4ress, pp. '@(L'7(.
Ai, M. and Mitanyi, 4., #77E. n Introduction to 5ol"ogoro! /o"plexity and
its, pplications, 0e .or/$ )pringer.
Aipton, 4., #77#. Inference to the Best Explanation, Aondon$ -outledge.
Aipton, 4., #77(. ;Is the Best %ood EnoughO= &roceedings of the
ristotelian, %ociety, 7($ 87L#C+.
Aipton, 4., #778. ;!he Epistemology of !estimony,= %tudies in 'istory
and, &hilosophy of %cience, '7$ #L(#.
Aipton, 4., 'CC+. Inference to the Best Explanation, 6'nd ed.9, Aondon$ -outledge.
Aombro?o, !., 'CCE. ;)implicity and 4robability in *ausal
Explanation,= /ogniti!e, &sychology, DD$ '('L'DE.
Maher, 4., #77'. ;&iachronic -ationality,= &hilosophy of %cience, D7$ #'CL#+#.
Mc%re, !., 'CC(. ;*onfirmation, Heuristics, and Explanatory
-easoning,= British, -ournal for the &hilosophy of %cience, D+$ DD(LD@E.
McMullin, E., #77'. +he Inference that Ma,es %cience, Milau/ee 3I$ Marquette
Kniversity 4ress.
McMullin, E., #77@. ;Epistemic Mirtue and !heory Appraisal,= in I. &ouven and A.
Horsten 6eds.9, *ealis" in the %ciences, Aeuven$ Aeuven Kniversity 4ress, pp. #(L
Moore, %. E., #7@'. ;4roof of an External 3orld,= in his &hilosophical &apers,
0e .or/$ *ollier Boo/s, pp. #'@L#+7.
Moser, 4., #787. 5no#ledge and E!idence, *ambridge$ *ambridge Kniversity
Musgrave, A., #788. ;!he Kltimate Argument for )cientific -ealism,= in -. 0ola
6ed.9, *elati!is" and *ealis" in %cience, &ordrecht$ Bluer, pp. ''7L'D'.
0iiniluoto, I., #777. ;&efending Abduction,= &hilosophy of %cience, @@$ )+(@L
,/asha, )., 'CCC. ;Man :raassen5s *ritique of Inference to the Best
Explanation,= %tudies in 'istory and &hilosophy of %cience, (#$ @7#LE#C.
4eirce, *. ). F#(G. /ollected &apers of /harles %anders &eirce, edited by *.
Hartshorne, 4. 3eiss, and A. Bur/s, #7(#L#7D8, *ambridge MA$ Harvard
Kniversity 4ress.
4sillos, )., #777. %cientific *ealis"9 'o# %cience +rac,s +ruth, Aondon$
4sillos, )., 'CCC. ;Abduction$ Beteen *onceptual -ichness and *omputational
*omplexity,= in A. B. Ba/as and 4. :lach 6eds.9, bduction and
Induction9 , Essays on their *elation and Integration, &ordrecht$ Bluer, pp. D7L
4sillos, )., 'CC+. ;Inference to the Best Explanation and Bayesianism,= in :.
)tadler 6ed.9, Induction and 8eduction in the %ciences, &ordrecht$ Bluer, pp. 8(L
4utnam, H., #78#. *eason, +ruth and 'istory, *ambridge$ *ambridge Kniversity
-ussell, B., #7#'. +he &roble"s of &hilosophy, ,xford$ ,xford Kniversity 4ress.
)chur?, %., 'CC8. ;4atterns of Abduction,= %ynthese, #@+$ 'C#L'(+.
)/yrms, B., #77(. ;A Mista/e in &ynamic *oherence ArgumentsO= &hilosophy
of, %cience, @C$ ('CL('8.
)tanford, B., 'CC7. ;Knderdetermination of )cientific !heory,= in %tanford
Encyclopedia of &hilosophy 63inter 'CC7 Edition9, Edard 0. Yalta 6ed.9, K-A Q
!eller, 4., #7E(. ;*onditionali?ation and ,bservation,= %ynthese, '@$ '#8L'D8.
!hagard, 4., #7E8. ;!he Best Explanation$ *riteria for !heory *hoice,= -ournal
of, &hilosophy, ED$ E@L7'.
van :raassen, B., #78(. ;%lymour on Evidence and Explanation,= in <. Earman
6ed.9, +esting %cientific +heories, Minneapolis$ Kniversity of Minnesota 4ress, pp.
van :raassen, B., #78D. ;Empiricism in the 4hilosophy of )cience,= in 4.
*hurchland and *. Hoo/er 6eds.9, I"ages of %cience, *hicago IA$ Kniversity of
*hicago 4ress, pp. '+DL(C8.
van :raassen, B., #787. .a#s and %y""etry, ,xford$ *larendon 4ress.
Mogel, <., #77C. ;*artesian )/epticism and Inference to the Best
Explanation,= -ournal of &hilosophy, 8E$ @D8L@@@.
Mogel, <., 'CCD. ;!he -efutation of )/epticism,= in M. )teup and E. )osa
6eds.9, /onte"porary 8ebates in Episte"ology, ,xford$ Blac/ell 4ublishing, pp.
3eisberg, <., 'CC7. ;Aocating IBE in the Bayesian :rameor/,= %ynthese, #@E$