ArticlePDF Available

Conservatism, Basic Beliefs, and the Diachronic and Social Nature of Epistemic Justification



Discussions of conservatism in epistemology often fail to demonstrate that the principle of conservatism is supported by epistemic considerations. In this paper, I hope to show two things. First, there is a defensible version of the principle of conservatism, a version that applies only to what I will call our basic beliefs. Those who deny that conservatism is supported by epistemic considerations do so because they fail to take into account the necessarily social, diachronic and self-correcting nature of our epistemic practice. Second, I will attempt to show how our basic beliefs are justified via this principle of conservatism.
EPISTEME 2006 203
Discussions of conservatism in epistemology often fail to demonstrate that the principle of
conservatism is supported by epistemic considerations. In this paper, I hope to show two things.
First, there is a defensible version of the principle of conservatism, a version that applies only to
what I will call our basic beliefs. Those who deny that conservatism is supported by epistemic
considerations do so because they fail to take into account the necessarily social, diachronic and
self-correcting nature of our epistemic practice. Second, I will attempt to show how our basic
beliefs are justifi ed via this principle of conservatism.
Traditional epistemic conservatism is roughly the
position that the fact that a belief is held provides
prima facie justifi cation for that belief. Discussions
of conservatism in epistemology often fail to
demonstrate that the principle of conservatism is
supported by epistemic considerations; that is, such
discussions have failed to show that beliefs justifi ed
by conservativism are thereby likely to be true.
Indeed, criticisms of, for example, conservatism
generally point out that there is no epistemic
justifi cation for a principle of conservatism, and
that such a principle therefore has no place in our
epistemic practice. In this paper, I hope to show
two things. First, I hope to show that there is a
defensible version of the principle of conservatism,
a version that applies only to what I will call
our basic beliefs. According to this version of
conservatism, these basic beliefs are justifi ed by
their history, and so an epistemic agent is entitled
to these beliefs even if she is not able to provide
any inferential justifi cation for them. Those who
deny that conservatism is supported by epistemic
considerations do so because they fail to take into
account the necessarily social, diachronic and self-
correcting nature of our epistemic practice. It will
emerge that justifi cation is essentially historical in
nature, and that the history an epistemic practice
must have in order to be justifi cation-conferring
is a generations-long history. Thus, this historical
requirement on justifi cation makes justifi cation
social as well. The second conclusion I will attempt
to establish is that this principle of conservatism
explains how our basic beliefs are justifi ed. The
principle of conservatism which is defended in
this paper will have several qualifi cations, but it
would be premature to introduce them now; let us
see how they emerge dialectically as the paper
In this paper, I will be primarily interested in
giving an account of justifi cation or justifi cation
attributions. I am not sure that all our uses of the
verb ‘to know’ have anything in common, or that
a unifi ed theory of knowledge can be given.
Although I will occasionally speak of knowledge,
this paper will focus primarily on the more
modest task of showing some of justifi cation’s
underpinnings. Two terminological notes are in
order. First, I will use ‘justifi ed’ and ‘rational’ as
synonymous. This claim is not intended to carry
any theoretical weight; it is merely a stipulation
for the sake of clarity and convenience. Second,
the phrase ‘basic belief’ is not used in the way
that, e.g., Plantinga uses this term, to mean non-
inferential (e.g., observational) beliefs. I am using
the phrase to refer to beliefs which are so basic to
our system of beliefs that it is diffi cult or impossible
to provide any inferential justifi cation for them at
all. The notion is borrowed from Wittgenstein’s
notion of framework beliefs in On Certainty, those
beliefs which cannot be justifi ed inferentially. I will
give a fuller account of basic beliefs as the paper
Epist2_3_06_Koons.indd 203Epist2_3_06_Koons.indd 203 15/8/06 16:52:3815/8/06 16:52:38
204 EPISTEME 2006
Jeremy Koons
I. Conservatism and Diachronic Justifi cation
It has been argued that conservatism is an
unavoidable part of justifi cation. For example,
Lawrence Sklar has argued that “conservatism
lies at the very basis of any possible structure
for justifying beliefs at all.”1 Sklar starts with the
seemingly undeniable premise that “all epistemic
justifi cation is relative to an assumed background
of believed theory.”2 From this starting point, Sklar
argues as follows:
[W]e must realize that all justifi cation is
“local.” We justify the beliefs we take to be in
need of justifi cation “one at a time,” using all
the resources of our unchallenged background
belief in the process. Such “local” justifi cations
are the only justifi cations of which we can
make sense, for all justifi cation requires a
body of unchallenged background belief, and
we never could justify our totality of beliefs “all
at once.”3
Thus, all local justifi cation occurs in the context
of a set of background beliefs that is merely
accepted, and cannot be argued for. This is not
a new position. In On Certainty, Wittgenstein
writes, “I did not get my picture of the world by
satisfying myself of its correctness; nor do I have
it because I am satisfi ed of its correctness. No:
it is the inherited background against which I
distinguish between true and false.”4 To give an
example, when I get out of bed in the morning,
I don’t have to consider the hypotheses that
putting my feet on the fl oor will cause them to
explode, or that turning off my alarm clock will
cause demons to rain out of the heavens. Without
a background theory of the world ruling out such
hypotheses, uncertainty would rule, and no action
would be possible. Of course, I don’t explicitly
consider these bizarre hypotheses, and judge
them unworthy of further consideration. I merely
behave as if they were false, and they never enter
my mind. This is why, I think, Wittgenstein says
our knowledge is grounded in a certain way of
acting:5 our knowing something is in many cases
not a matter of considering and rejecting rival
hypotheses; instead, it is in large part a matter of
acting in a way that is incompatible with belief in
such rival hypotheses. We have a way of acting,
a way of ignoring certain hypotheses, that guides
our empirical inquiry, and shapes how we explore
the world. Normally, we don’t need to respond
to challenges to this way of acting; I don’t need
to justify ignoring these rival hypotheses. Rather
than standing in need of justifi cation, this way
of behaving provides the background against
which our other beliefs are justifi ed. But our way
of acting as if certain hypotheses are false forms
the riverbed in which the water of our empirical
knowledge fl ows, to use Wittgenstein’s analogy.
A critic, however, will no doubt say that this
picture of justifi cation is a picture of dogmatism, a
system that allows a background system to be held
conservatively, without (inferential) justifi cation.
The way of acting needs no justifi cation, indeed!
And indeed this would be a picture of dogmatism
if there were no way of providing a justifi cation
for our “unchallenged background belief[s]”. But
there is a way in which we can transcend mere
local justifi cation, and give a justifi cation that
encompasses Sklar’s “assumed background of
believed theory.” Let us see how this more global
justifi cation is to be provided.
The picture of justifi cation just outlined only
looks dogmatic if you look at it as a static system,
with an immovable foundation. However, our
background theories that tell us which hypotheses
to ignore are themselves revisable. The riverbed
moves over time. Consider the proposition, “The
earth is the center of the universe.” In the 1920s,
the astronomer Edwin Hubble made an interesting
observation: in every direction you look, galaxies
are receding from the earth. Furthermore, the
farther away a galaxy is, the faster it is receding.
How did scientists explain this observation?
Signifi cantly, nobody suggested the hypothesis
that the earth is at the center of the universe;
that hypothesis was not one that even merited
discussion. Scientists eventually settled on the
hypothesis that space itself is expanding. It was a
revisionary hypothesis, but no one was willing to
postulate a geocentric universe. Notice, though,
that 1,000 years ago, the proposition “The earth
is at the center of the universe” was not merely
a hypothesis that needed to be taken seriously;
cosmological hypotheses which confl icted with
this belief were immediately rejected. Thus, the
hypothesis that the earth is at the center of the
universe started out as a proposition used to test
Epist2_3_06_Koons.indd 204Epist2_3_06_Koons.indd 204 15/8/06 16:52:3815/8/06 16:52:38
hypotheses, and ended up as a hypothesis to be
discarded without serious consideration.6 It is for
this reason that Wilfrid Sellars writes:
Above all, the [traditional picture of knowledge]
is misleading because of its static character.
One seems forced to choose between the
picture of an elephant which rests on a tortoise
(What supports the tortoise?) and the picture
of a great Hegelian serpent of knowledge
with its tail in its mouth (Where does it begin?).
Neither will do. For empirical knowledge, like
its sophisticated extension, science, is rational,
not because it has a foundation but because
it is a self-correcting enterprise which can
put any claim in jeopardy, though not all at
As our system of empirical knowledge evolves
through revision (which revision occurs through
argument and the “tribunal of experience,” as
Quine calls it), we sometimes realize that what
were groundless beliefs, in no need of justifi cation,
are false and need to be discarded. That is why
our system of knowledge is rational: it is rational
because no belief has been in principle immune to
revision. We are justifi ed in ruling out hypotheses
such as “Demons will fall from the heavens if I
turn off my alarm clock” because our theory of
the world which dictates that such hypotheses be
ignored has survived the tribunal of experience,
and because the theory has been revisable in
light of evidence that it is false. Revisability, not a
foundation, is the source of justifi cation.8
So judging requires that certain background
beliefs merely be accepted without being argued
for. Recall Wittgenstein’s quote, which was cited
earlier: “I did not get my picture of the world by
satisfying myself of its correctness; nor do I have
it because I am satisfi ed of its correctness. No:
it is the inherited background against which I
distinguish between true and false.”9 But this is not
dogmatism, because these background beliefs
evolve as our empirical knowledge grows.10 It
is through the evolutionary pressures imposed
on theory by experience that old background
beliefs get rejected and new ones introduced.
And it is this evolution over time—this revision of
background beliefs over time, to better accord
with experience—that allows us (correctly) to
regard these background beliefs as adequate.11
For even if we didn’t choose these background
beliefs ourselves, they are not arbitrary: they are the
product of millennia of empirical inquiry. This fact
is what makes our body of empirical knowledge
rational. Our critic, then, misunderstands the
nature of justifi cation: she thinks that for a system
to be justifi ed at time t, it must be possible at time
t to give an explicit justifi cation for any belief in
the system. But this picture of justifi cation ignores
the fact that only some beliefs are justifi ed this
way; others (namely our basic beliefs) are justifi ed
purely by their history. Thus, the critic ignores the
temporal element crucial to an understanding of
justifi cation: the basic beliefs in the system, the ones
for which we can offer no inferential justifi cation,
are justifi ed—and thereby fi t to serve a justifi catory
role—because they themselves have withstood the
test of time and evidence, because these beliefs
are the product of epistemic evolution, because
the system has been allowed to evolve over time.
A system that became immune to revision would
before long cease to be justifi ed. The system,
then, is rational, not because of its structure at
time t, but because no belief in the system has
always been de jure immune to revision.
Let me put this point another way: having
made a judgment, you may go on to justify the
judgment, but this justifi cation will, of course, rest
on “assumed background of believed theory,” as
Sklar put it. One might, if the dialectical situation
requires it, go on to justify some or all of these
assumed background beliefs; but of course,
such a justifi cation will itself rely on an assumed
background of theory. You might suppose that at
some point, we will reach a set of beliefs that
we cannot justify; there are no beliefs more
basic than these that we could use to justify these
“foundational” beliefs. As Wittgenstein says, at
some point “I have exhausted the justifi cation, I
have reached bedrock and my spade is turned.
Then I am inclined to say, ‘This is simply what I
do.’”12 Does that mean these “bedrock” beliefs
are arational? No;13 to think so is to think that a
belief is only justifi ed if we can present an explicit,
inferential justifi cation for that belief. Some beliefs
are justifi ed that way, but our “foundational”
beliefs are justifi ed for a different reason: they are
justifi ed because they are the result of a millennia-
long inquiry of the world, because they are the
Epist2_3_06_Koons.indd 205Epist2_3_06_Koons.indd 205 15/8/06 16:52:3815/8/06 16:52:38
206 EPISTEME 2006
Jeremy Koons
product of epistemic evolution. It follows that if
we were to declare a set of basic beliefs de jure
unrevisable, then in short order these beliefs (and
all that they support) would cease to be justifi ed.
The reason is that justifi cation for our basic beliefs
relies in large part on their having faced the
tribunal of inquiry and survived, and declaring
a set of basic beliefs de jure unrevisable is to
remove them from before this tribunal.14,15 But
if we haven’t treated our background beliefs as
de jure unrevisable,16 then our system of beliefs
is justifi ed. Thus, although many philosophers
(including Wittgenstein) have argued that
justifi cation requires foundations, the correct view
is that justifi cation requires revisability. It is only
because of their historical revisability that our so-
called “foundations” (the basic propositions) are
justifi ed.
One might object that some basic beliefs, such
as logical beliefs, haven’t been revisable, but
are nevertheless justifi ed. There are two ways of
thinking about this issue. One way (which I take to
be Wittgenstein’s view) is that such beliefs haven’t
been revisable, and therefore aren’t justifi ed—but
these beliefs form the framework of our language,
the framework within which justifi cation takes
place. The other way of thinking about this
issue is more Quinean: such beliefs may never
be revised, but what makes them justifi ed is that
they have been open to revision. If a belief (such
as a belief in the law of the excluded middle)
has been open to revision for centuries, but so
far no compelling reason has risen to revise it,
then the belief has earned its epistemic status in
virtue of having survived this centuries-long test.
Thus, justifi cation doesn’t necessarily require that
the belief is the product of revision; justifi cation
requires only that all of our basic beliefs must have
been subject to revision, as a condition of their
being justifi ed. As Mark Lance and John O’Leary-
Hawthorne put the point,
Treating a whole bunch of claims as de jure
unchallengeable … seems …constitutive of
dogmatism…Meanwhile, the recognition
that some claims may turn out to be de facto
unchallengeable (and even necessarily so)
runs no similar cognitive risks.17
It should be clear that this paper adopts the
latter Quinean approach. The reason it is not
dogmatic to judge by our framework propositions
is that these framework propositions (not just
laws of logic, but more malleable framework
propositions, such as the belief that no human
has ever set foot on Mars) have not been immune
to revision.
Despite the defense offered above, one might
nevertheless think that the present account suffers
from a kind of dogmatism. The objection goes
as follows:
You say our background beliefs are justifi ed
because they derive from a history of revision.
But any history will start with arbitrarily chosen
beliefs. Our current beliefs are determined
by an arbitrary initial choice, and there is no
reason to prefer one choice to another. So our
current background beliefs are arbitrary, and
holding them is dogmatic.18
This objection overstates the infl uence of
our starting place on our current theory. It is
plausible to think that our current theory depends
more on the revisionary pressures our theories
have encountered over the years rather than on
our starting place in the distant past. Consider
an analogy from evolution. Richard Dawkins
writes that “eyes have evolved no fewer than
forty times, and probably more than sixty times,
independently in various parts of the animal
kingdom.”19 So similar environmental inputs led
various lineages to develop the same feature.
Now it may be pointed out that all of these
different lineages have a common descent, but
it is not their common descent which explains
their common evolution of the eye. Rather, similar
environmental and evolutionary pressures caused
the eye to evolve again and again; common
descent plays little or no role in the explanation of
this. Plausibly, the same applies to the evolution
of our belief systems over time. Consider the
situation of different communities characterized
by different theoretical starting points. As long as
they experience epistemic evolution characterized
by progress (which we will endeavor to defi ne in
the following few paragraphs), the fact that these
communities have different starting points will be
less important than the fact that they live in the
same world and are hence subject to the same
Epist2_3_06_Koons.indd 206Epist2_3_06_Koons.indd 206 15/8/06 16:52:3915/8/06 16:52:39
causal infl uences and pressures on their scientifi c
theories. Just as external pressures proved more
important than starting place in the case of
evolution, so will external epistemic pressures
prove more important than starting place in the
evolution of our theories.
So far, we have used an evolutionary analogy
to discuss the type of change our theories
undergo over the generations. If researchers allow
themselves to be guided by goals such as a desire
to reach the truth, it seems likely that epistemic
evolution will result in progress. As Karl Popper
What characterizes the empirical method is its
manner of exposing to falsifi cation, in every
conceivable way, the system to be tested. Its
aim is not to save the lives of untenable systems
but, on the contrary, to select the one which is
by comparison the fi ttest, by exposing them all
to the fi ercest struggle for survival.20
In a locus classicus of evolutionary
epistemology, Donald Campbell writes that “a
blind-varation-and-selective-retention process is
fundamental to all inductive achievements, to all
genuine increases in knowledge, to all increases
in fi t of system to environment.”21
It is important to note that the change that
confers justifi cation on basic beliefs must be a
certain sort of change. We have talked about the
importance of epistemic evolution in the justifi cation
of our background beliefs. But presumably, not
all change is positive, and hence not all change
is justifi cation-conferring. What is needed is that
we revise our background theories in a way that
ts experience. That is, what is needed is not
just change, but progress. But of course, not all
researchers are guided by such pure motives, and
one might worry, then, that not all theory change
represents progress toward better fi t between
theory and environment. Thus, it is necessary to
have some tool in hand for distinguishing progress
from mere change (or even regress). Such a tool
has been developed by Philip Kitcher in The
Advancement of Science. Kitcher notes that
progress is not one-dimensional; rather, there are
different types of progress, which Kitcher identifi es
as follows. First is practical progress, which is an
increase in our ability to control the world. Then
there are varieties of cognitive progress. First is
conceptual progress, which Kitcher defi nes as
Conceptual progress is made when we adjust
the boundaries of our categories to conform
to kinds and when we are able to provide
more adequate specifi cations of our referents.
Striking examples come from the history of
all sciences: ‘planet,’ ‘electrical attraction,’
‘molecule,’ ‘acid,’ ‘gene,’ ‘homology,’ ‘Down’s
syndrome,’ are all terms for which faulty modes
of reference have been improved.22
Another type of cognitive progress is
explanatory progress. Kitcher writes that
Explanatory progress consists in improving
our view of the dependencies of phenomena.
Scientists typically recognize some phenomena
as prior, others as dependent. For example,
ever since Dalton, chemists have regarded
molecular arrangements and rearrangements
as prior to the macroscopic phenomena of
chemical reactions, and, since the 1960s,
geologists have viewed interactions among
plates as prior to facts about mountain building
and earthquakes.23
Given these specifi c varieties of progress,
Kitcher defi nes progressive change in a practice
as follows:
…let us say that the sequence of practices
P1,…, Pn is broadly progressive just in case
for every pair of adjacent members there is a
component of practice with respect to which
the change from the earlier to the later is
progressive and the change from P1 to Pn is
progressive with respect to every component
of practice.24
Kitcher argues that science in general does
progress, and outlines in chapter 6 the conditions
(which he thinks normally obtain) under which
science advances. Thus, using tools of the sort
supplied by Kitcher, we can defi ne the sort of
progressive change which confers justifi cation on
our background theories.
Before moving on to draw some intermediate
Epist2_3_06_Koons.indd 207Epist2_3_06_Koons.indd 207 15/8/06 16:52:3915/8/06 16:52:39
208 EPISTEME 2006
Jeremy Koons
conclusions from the above discussion, let
me pause to make some comments about the
epistemic values that are at work here. In keeping
with the evolutionary analogy, the primary notion
we are working with is one of fi t: beliefs that do
not fi t with their environment are discarded and
replaced with those that do fi t. But a creature’s
environment has two elements: fi rst, there are the
physical, non-living elements of the environment.
Second, there are other organisms. Both of these
environmental features create pressures on a
particular creature; and to survive, a creature
must exhibit fi tness with respect to both elements
of its environment. The analogy can be applied,
with only a little strain, to the evolutionary view of
epistemology under discussion here: evolutionary
forces work through confl icts with inputs from
the physical world as well as with other beliefs
(‘creatures’). Confl ict with either environmental
feature can cause a belief to ‘die out’ and be
replaced by a fi tter specimen. And thus, we see
that the emphasis on fi tness makes coherence with
other beliefs and coherence with empirical inputs
to be important considerations.
There are two reasons, however, for regarding
environmental inputs as in an important sense prior
to coherence with other beliefs. First, incoherence
is most often introduced into a belief system through
observation: observation introduces a new belief
into the system which creates incoherence, and
thus a ‘struggle for survival’ among the beliefs
present in the system. The exponential growth of
knowledge over the past centuries has mostly been
the growth of empirical knowledge—this is the
primary source of new inputs into the system, and
so will understandably be the source of most of
the inconsistencies. Ideally, the inconsistency will
then be reconciled by the familiar considerations
of simplicity and so forth.
The second reason for the emphasis on
observation and experience relates to the issue
discussed above: namely, some beliefs are not
revised, and are yet justifi ed. As I indicated
above, such beliefs are justifi ed because they have
survived so long without needing revision. But for
their survival to be epistemically meaningful, they
must have survived constant testing. As we noted
above, the most torrential source of new information
in our system is observation; it is observation that
will be placing the most evolutionary pressures on
a given belief. If this fl ow of new information, and
the exponential increase in knowledge that we
have become accustomed to, does not dislodge
a particular belief, then this is a powerful argument
that the belief is true: it has survived such a long
and severe test at the hands of ever-increasing
empirical data.
Some intermediate conclusions
We are now in a position to draw some important
(1) Conservatism is an unavoidable part of
justifi cation. As Sklar and Wittgenstein recognize,
all inferential justifi cation takes place against a
background of beliefs which the individual is not
capable of justifying. There are certain principles
we must merely accept.
(2) This conservatism is not anti-epistemic,
because of the diachronic nature of justifi cation.25
We are led to the following, interesting conclusion:
our basic beliefs’ justifi cation is essentially
diachronic, and their justifi cation requires
revisability. If these requirements are not met—if
a system of basic beliefs is not revisable—then
our conservatism becomes mere dogmatism. Our
system of background beliefs becomes unjustifi ed
(since evolutionary forces are not allowed to work
on them); we are no longer epistemically justifi ed
in ruling out remote hypotheses (since this practice
depends on having a justifi ed background theory,
and the justifi cation of this background theory
can only be understood in terms of revisability
and diachronic evolution). Justifi cation has an
essentially diachronic element. We will come
back to this point in a few pages, to clear up any
potential misunderstandings that might arise.
(3) The type of conservatism that is defended
here is very different from the version of
conservatism discussed by most commentators.
Most commentators are concerned with the
version of conservatism that says, roughly, that a
belief acquires some degree of justifi cation merely
by being believed.26 But this simple version of
the principle is almost certainly false. Rather, the
principle of conservatism applies only to certain
propositions, propositions we will call basic
beliefs. Further, which propositions fi t this category
is determined by the diachronic progress of our
particular social practice. Finally, such beliefs are
Epist2_3_06_Koons.indd 208Epist2_3_06_Koons.indd 208 15/8/06 16:52:3915/8/06 16:52:39
justifi ed by their history, rather than by other sorts
of reasons we might adduce in support of them.
(We will discuss this last point in the following
paragraphs). Let us further explain and defi ne the
version of conservatism defended here.
(a) Basic beliefs are propositions that are so
basic that we fi nd it diffi cult, if not impossible,
to provide any sort of inferential justifi cation for
them. Propositions that might fi t into this category
include “Most humans have two hands;” “The sky
is usually blue;” “The earth is not at the center
of the universe;” “Demons do not interfere with
scientifi c experiments;” and so on. If asked
to provide an inferential justifi cation for such a
belief, we might fi nd ourselves unsure of what to
say. In many cases, we simply don’t know how
to go about inferentially justifying these beliefs;
they simply seem too basic. Another type of case
is illustrated by one of Wittgenstein’s famous
examples: “My not having been on the moon is
as sure a thing for me as any grounds I could give
for it.”27 I could provide an inferential justifi cation
for the claim, “I have two hands” or “I have never
been on the moon,” but it is pointless to do so,
as the considerations I would cite as support are
not more basic than the claim they are supposed
to support.
(b) These basic beliefs are justifi ed by the
diachronic progress of our process of inquiry.
Even though the average person cannot provide
an inferential justifi cation for such beliefs, these
beliefs are (as I argued earlier) justifi ed by the fact
that they are the product of centuries of epistemic
evolution, and have so far withstood the challenges
which are part of the progress of knowledge
and the exponential increase in knowledge and
information which accompanies such progress.
Thus, the difference between a basic belief and
an inferential belief is that a basic belief needs
no justifi cation beyond its history, whereas an
inferentially justifi ed belief is one that is ultimately
justifi ed on the basis of some basic beliefs.28
c) A third feature of basic beliefs follows from
other features of basic beliefs, namely, that basic
beliefs are justifi ed, and yet we cannot provide
any (inferential or non-inferential justifi cation) for
them. It follows from this that when a basic belief is
challenged, the burden of proof is on the challenger
to defend his or her position. The holder of the
basic belief cannot be expected to defend his
belief (since part of the defi nition of ‘basic belief’
is that such a defense is not possible), and yet the
belief is justifi ed all the same; and so a challenger
to this belief must herself shoulder the burden of
proof. If the challenger can not give us evidence
suggesting that the proposition in question is false,
then we are justifi ed in continuing to believe that
the proposition is true. Indeed, it virtually follows
from any defi nition of conservatism that the burden
of proof is on the challenger: the holder of the
basic belief is prima facie entitled to this belief,
and so it is up to the challenger to demonstrate
that this entitlement does not hold. This feature of
conservatism (that the burden of proof is on the
one who wishes to revise the basic belief) will
become important when we turn our discussion to
the issue of radical skepticism.
(d) To say that basic beliefs are prima facie
justifi ed in no way entails that they are immune
from revision. Consider an example of a basic
belief offered by Wittgenstein: “No one has ever
been on the moon.” Wittgenstein writes,
What we believe depends on what we
learn. We all believe that it isn’t possible to
get to the moon; but there might be people
who believe that that is possible and that it
sometimes happens. We say: these people do
not know a lot that we know. And, let them be
never so sure of their belief - they are wrong
and we know it. If we compare our system of
knowledge with theirs then theirs is evidently
the poorer one by far.29
And yet someone who now denied that
humans had ever set foot on the moon would
be dismissed as ignorant or crazy. And so
basic propositions are revisable. With the basic
proposition of no one having ever been on the
moon, the process of revision was speedy, as the
event was televised. But with other basic beliefs
(such as the belief that the earth is at the center
of the universe), conservatism will often entail that
their revision will be a lengthy (and sometimes
painful) process. And indeed, as I argued above,
such basic beliefs must be revisable; it is their
revisability which is ultimately the source of their
(e) It is important to emphasize the sense in
which what I am calling conservatism here really is
Epist2_3_06_Koons.indd 209Epist2_3_06_Koons.indd 209 15/8/06 16:52:3915/8/06 16:52:39
210 EPISTEME 2006
Jeremy Koons
a version of conservatism. From one perspective,
it isn’t conservatism at all: the basic propositions
are justifi ed not merely because they are believed
by us, but because they are the process of a
millennia-long process of empirical inquiry. That
is, they are justifi ed by their history. So there is
something justifying these basic propositions,
something beyond the mere fact that they are
believed by us. So from a sort of external view,
taking our theory of the world as our object of
study, conservatism is not conservatism at all.
However, from the perspective of individual
agents, the basic propositions are conservatively
held.30 Most agents are not capable of telling a
story about what justifi es these basic propositions;
they hold them (and are entitled to hold them)
for reasons that are conservative in nature.
They represent a starting point from which the
individual can reason.31 Thus, conservatism really
is conservatism—at least from the perspective of
the individual.
(4) Our fi nal intermediate conclusion is
that justifi cation is social in character, in that it
presupposes a community with a history of inquiry,
with the results of this inquiry (and the evolving
epistemic background governing this inquiry)
passed down through generations of inquirers.
Remember the course of our epistemic argument
for the justifi cation of basic beliefs: our so-called
basic beliefs are justifi ed by the generations-
long history of inquiry which produced them.
The argument is this: certain beliefs (such as a
belief that demons don’t interfere with scientifi c
experiments) are justifi ed because they have
endured generations of inquiry, practiced by
countless researchers, and they have survived this
history without refutation. That our basic beliefs
survived so long, without becoming encumbered
by ad hoc epicycles, open to revision and
refutation but not having succumbed, is what
lends epistemic weight to our background theory.
Thus, justifi cation is social in that local justifi cation
requires that our basic beliefs are justifi ed, and
these basic beliefs have their justifi cation in the
history of inquiry which has produced them.
As this history spreads over generations and
countless inquirers, there is a social dimension to
justifi cation. I will return to this point in a moment,
when addressing objections, but the key thing to
note is that this history is crucial to the justifi cation
of our basic beliefs; a community lacking such a
history would have theories built on sand, lacking
justifi cation.
Before moving on, I want to clear up one
potential misunderstanding. I am only attempting
to establish that revision and diachronicity are
crucial to the justifi cation of basic beliefs
those beliefs which are conservatively held. As
I noted earlier in the paper, some beliefs are
inferentially justifi ed, and others seem to defy
such inferential justifi cation. It is the latter beliefs
whose justifi cation essentially relies on their place
in our history of inquiry, whose justifi cation is
essentially diachronic and essentially relies on
openness to revision. The former are inferentially
justifi ed, and so their justifi cation need not rely
on the conservatism defended above. Thus,
when explaining how an ordinary belief (such as
“Pelé lead Brazil to three World Cup victories”)
is justifi ed, we will appeal not to conservatism,
diachronicity and revision, but instead to memory,
authority, or some other recognized justifi er.32
The justifi cation of such beliefs might ultimately
trace back to the justifi cation of our basic beliefs,
and so the justifi cation of ordinary belief-claims
might ultimately rely on revision, etc. But this
would only show that a belief such as “Pelé lead
Brazil to three World Cup victories” indirectly
relies on revision and diachronicity (in the same
way that a foundationalist would say that an
inferentially-justifi ed belief indirectly relies on
some foundational belief); it is not directly justifi ed
by such considerations.
Objections and Replies
Let us now turn our attention to answering some
objections. First, one might object as follows: “An
appeal to revisability and the long history of human
inquiry cannot be part of an argument vindicating
those theories, because the appeal makes use
of those theories. Therefore, the argument is
circular.”33 This objection fails, because I am not
trying to justify those theories constituting the history
of inquiry. I am trying to justify the current theory of
the world, held by an epistemic community. That
theory is justifi ed because the previous theories
led, through a more or less objective process of
inquiry, to the current theory. The current theory
represents the culmination of centuries of inquiry,
Epist2_3_06_Koons.indd 210Epist2_3_06_Koons.indd 210 15/8/06 16:52:4015/8/06 16:52:40
and its authority rests on the process that led up
to it. The current theory isn’t justifi ed by itself; it is
justifi ed by the history that preceded it.
Of course, this is not to say that those previous
theories weren’t justifi ed in their time. They were,
and for similar reasons: they were justifi ed by the
history of inquiry that preceded them. At any time,
the theory justifi ed at that time owes its justifi cation
to what went before. This is part of what it means
to say that epistemic justifi cation is diachronic.
There seem to be, however, several classes
of counterexamples to the theory of justifi cation
developed here. Consider the following:
A community that springs into existence (say,
Adam and Eve) presumably have justifi ed
beliefs, even though there is no history of
inquiry in this community. (For future reference,
we will call this community the Ruritanians.)
Beliefs about goblins, etc., are not justifi ed,
even though they may be part of the
fundamental assumptions in a community with
a long history of inquiry.
A victim of an evil demon is justifi ed in her
beliefs, even though she is part of no such
Initially, these seem like daunting
counterexamples. However, I think they can
be defused. These examples exploit internalist
intuitions. I will argue that the diachronic
requirement defended in this paper serves as a
quasi-externalist criterion, and will argue that we
must see our beliefs as subject to evaluation by
such criteria.
The importance of recognizing an external
constraint such as the presence of a diachronic
history can be shown by the following train of
thought. Traditional theories of epistemology have
placed certain structural constraints on a person’s
belief system. For foundationalism, beliefs in the
system are only justifi ed if they stand in the proper
inferential relation to various foundational beliefs.
For coherentism, beliefs are only justifi ed to the
extent that they belong to a belief system which
is itself coherent. But notice that these structural
features cannot be internalist constraints on belief,
for the simple fact that it is impossible for an
individual to ascertain whether these constraints
are met. Internalist constraints can only demand
of the agent things which can be reasonably
expected of human cognizers. Well, what can
be reasonably expected of humans, epistemically
speaking? We can’t expect them to calculate
conditional probabilities for all their beliefs and
have a belief set that conforms to Bayesian
axioms. Perhaps it would be ideal if they could—
but it would also perhaps be ideal if humans could
withstand any amount of torture without betraying
their principles and ideals. What would be ideal
is not relevant to what counts as good in the actual
world, given our actual abilities.34 Can humans
be expected to have a belief system that satisfi es
coherentist constraints? Probably not; our belief
systems are massive things, and examination of
whether the system is coherent or not can only
proceed piecemeal.
One might reply to this by saying, “So much
the worse for structural constraints on belief.” But
clearly there is a sense in which (for example)
a coherent belief system is, ceteris paribus,
rationally superior to an incoherent one. Suppose
a person possess a belief system which is
riven with inconsistencies, ad hoc epicycles
and beliefs, and so forth. This system is clearly
rationally inferior to a system which is coherent.
But this superiority is not something that can simply
be seen or detected by normal human cognizers;
and so coherence must function in this case as a
quasi-externalist constraint. And this is the sense in
which the diachronic requirement functions in the
alleged counterexamples to the current theory.
Other authors have argued that our evaluation
of epistemic agents must be relativized to agents’
actual abilities. For example, Alvin Goldman
Advice in matters intellectual, as in other
matters, should take account of the agent’s
capacities. There is no point in recommending
procedures that cognizers cannot follow
or prescribing results that cognizers cannot
attain. As in the ethical sphere, ‘ought’ implies
And so if we are evaluating individual
performance or blameworthiness (as an internalist
can be taken to do), then we must take into account
the actual epistemic abilities of agents. Thus,
requirements that transcend such abilities must be
regarded as structural, externalist standards, not
Epist2_3_06_Koons.indd 211Epist2_3_06_Koons.indd 211 15/8/06 16:52:4015/8/06 16:52:40
212 EPISTEME 2006
Jeremy Koons
personal, internalist standards.
This distinction allows us to answer some of
the above objections. The belief system of the
Ruritanians is fl awed because it fails to satisfy
a rather externalist requirement, one that is
not obviously recognizable from the internal
perspective. The victim of the evil demon has a
similarly fl awed belief system, although again,
this fl aw is not one that is recognizable from the
internalist perspective. The goblin case is a bit
trickier. As I argued above, the history of epistemic
inquiry in a community must be characterized in
part, at least, by some degree of responsible
revisability. If, for example, beliefs about goblins
have been treated for some time as de jure
unrevisable in the community in question, then these
beliefs are not rationally held. But if the community
has been responsible in revising its beliefs, and
belief in goblins has (so far) survived, then why
not say that these beliefs are rational? There is no
reason, in principle, why a false belief cannot be
rationally-held; to deny this is to confl ate truth and
justifi cation. Just as in our community it was no
doubt rational to believe in caloric or phlogiston,
this community is entitled to its belief in goblins.
So for this last example, we must recognize that
having a long history of inquiry is a necessary but
not suffi cient condition for justifi cation: the way in
which this history went, and the way in which the
community went about revising its commitments in
order to reach its current point, are relevant too.
But at the risk of confl ating justifi cation and truth,
we must recognize that there are cases where
this history can go correctly, yet still result in some
false (yet justifi ed) beliefs.
No doubt many readers will feel some
dissatisfaction at this solution; they will want to insist
that there is a sense in which the Ruritanians (and
the victim of the evil demon) are justifi ed in their
beliefs. To insist on this is to insist on the applicability
of internalist standards of justifi cation to the cases
in question. There is no reason to disagree with
this insistence, unless you somehow think that a
cognizer cannot be evaluated according to both
internalist and externalist criteria. But why think
this? Indeed, I think there is a way to satisfy both
internalist intuitions and the conclusion that belief
is answerable to externalist constraints. The way
to achieve this reconciliation is to recognize that
there are two types of justifi cation attribution one
might make. Consider the following sentence:
C: Astrology as a system has been soundly
refuted, but given her upbringing and available
evidence, she is perfectly justifi ed in believing
in astrology.36
Notice that both clauses in this sentence
make epistemic claims, and that there is a sense
in which these claims oppose each other. Let us
briefl y examine these two types of claims.
The claim expressed by the second clause
of C expresses what we shall call subjective (or
personal) justifi cation. In evaluating an agent’s
subjective justifi cation, we are evaluating the
agent’s performance and beliefs. We ask whether
the agent performed well, epistemically, given
the evidence and epistemic resources available
to her. Thus, subjective justifi cation is internalist
in spirit. But remember, as we argued above,
internalist constraints can only demand of the
agent things which can be reasonably expected
of human cognizers, and human cognizers
cannot reasonably be expected to evaluate
their community’s epistemic history, the structural
features of their theory, and so forth. Thus, in
addition to subjective justifi cation, we must also
recognize the relevance of objective justifi cation
attributions, which is the sort of claim made by
the fi rst clause of C. With objective justifi cation
attributions, one takes an externalist or third-
person view, with a view of features of justifi cation
to which individual performance cannot be held
accountable (including structural and historical
features of the individual’s and community’s
belief system). As a more externalist standard of
justifi cation, the diachronicity requirement belongs
to objective justifi cation.
An agent can be subjectively justifi ed without
being objectively justifi ed. That is, an agent
can have performed well, epistemically (or, if
you believe in epistemic responsibility, can be
epistemically free of blame) even if an external
observer can see that the belief or theory the
agent holds as a result of this performance is not
the best justifi ed one, or that it has structural fl aws,
etc.37 That is to say, an agent can be subjectively
justifi ed without her theory being objectively
justifi ed. In C above, we are saying that she has
not performed badly relative to her epistemic
Epist2_3_06_Koons.indd 212Epist2_3_06_Koons.indd 212 15/8/06 16:52:4015/8/06 16:52:40
circumstances, but that her theory is not objectively
justifi ed: perhaps there is evidence available in
her community, not widely disseminated (so she
is not held accountable for failing to obtain it),
that astrology is false; or perhaps the history of
astrological inquiry in her community has not been
characterized by suffi cient openness to refutation,
or so on.38 Alvin Goldman makes the following
comment on this case: we are inclined to say that
there is a sense in which she is justifi ed because of
“the cultural plight of our believer…Our believer
has good reasons to trust his cultural peers on
many matters, and lacks decisive reasons for
distrusting their confi dence in astrology.”39 On the
other hand, there is a sense in which we want to
say this person’s belief, which results from reading
zodiacal signs, is not justifi ed because the method
of astrology “looks improper and inadequate.”40
Thus, in the same case, a belief can be justifi ed
in one sense, but not in another.
So there is a simple answer to how we
reconcile internalist and externalist constraints.
Consider the fi rst example, of a community that
is created ex nihilo, with a ready set of beliefs
about the world. We can admit, for example,
that the Ruritanians are subjectively justifi ed in their
beliefs—they satisfy relevant internalist constraints.
But they lack objective justifi cation because they
fail to satisfy the relevant diachronicity constraint.
This brief account gives the rough idea of
the solution considered here, but there are
complications, depending on how the Ruritanians
came to hold their beliefs in the fi rst place. Let
us examine the case in more detail. As I noted
in the previous paragraph, it is perhaps true
that the Ruritanians are subjectively justifi ed in
forming beliefs about the world, and perhaps
in maintaining the beliefs they already have.41
To put things in terms of a deontic conception
of subjective justifi cation, the Ruritanians are not
epistemically to be blamed for believing as they
do. But when we shift perspectives, and view the
situation through the lens of objective justifi cation,
matters are different. It depends, fi rst, on how
the Ruritanians acquired their beliefs. Suppose
Ruritania was created by a race of super-beings,
who gave the new community many of the beliefs
that the race of super-beings themselves held. In
this case, the Ruritanians might well be objectively
justifi ed—but the objective justifi cation of their
beliefs is parasitic on the objective justifi cation of
the beliefs of the race of super-beings. Because
the super-beings’ beliefs are objectively justifi ed
(because of their history of inquiry), the Ruritanians
beliefs are also justifi ed, since the super-beings
endowed the new creatures with these same
beliefs. But suppose, in the creation of Ruritania,
there was no conscious or intelligent control over
the beliefs of the new community. It is not clear
why we should attribute objective justifi cation to
the Ruritanians. If the new beliefs of the Ruritanians
were formed not via intelligent control, but by
some non-intelligent process (due to some feature
of the ex nihilo formation process, the Ruritanians
just happened to wind up with this set of beliefs),
then it is not clear at all why we should say that
these beliefs are objectively justifi ed. Again, the
Ruritanians may be subjectively justifi ed—no
blame accrues to them for believing as they
do—but the beliefs they have are not worthy of
epistemic respect, as there is no particular reason
to suppose the process by which these creatures
were endowed with this particular set of beliefs
would give them any true beliefs at all. (I suppose
it is open to one to argue that a community of
sentient, sapient believers could come into
existence, ex nihilo, with a set of beliefs, and the
process of belief-endowment, though not controlled
by any intelligent force, somehow guarantees that
the beliefs are true, and that the beliefs of this
community are therefore objectively justifi ed. But
actually, on second thought, it is not really open to
someone to argue this absurd position.)
In any case, it is not clear what relevance
science-fi ction examples like this one have to our
actual conception of justifi cation. If our epistemic
circumstances were radically different than they
are, then perhaps a different conception of
justifi cation would be appropriate. Perhaps, for
example, the Ruritanians would at fi rst be entitled
only to claim a very weak type of justifi cation
(similar, say, to a notion of permission: they are
permitted to believe as they do, but not justifi ed in
a strong sense). As the community develops, and
acquires a history of inquiry, only then do they
become entitled to claim for themselves (or are we
entitled to attribute to them) a stronger notion of
justifi cation. Thus, I am not forced to concede that
the fi rst counterexample is a threat to the theory of
justifi cation developed in this paper.
Epist2_3_06_Koons.indd 213Epist2_3_06_Koons.indd 213 15/8/06 16:52:4015/8/06 16:52:40
214 EPISTEME 2006
Jeremy Koons
Let us consider the second alleged
counterexample, the community in which belief
in (say) goblins forms part of the fundamental
assumptions of the community. First, we must ask
about the epistemic evolution of the community.
As I noted above, the history of epistemic inquiry
in the community must be characterized in part, at
least, by some degree of responsible revisability.
Thus, if beliefs about goblins have long been de
jure unrevisable in the community in question, then
in fact these beliefs are not objectively justifi ed
(although again, the members of the community
may be subjectively justifi ed in believing in
But suppose the epistemic history of this
community meets these constraints, and thus the
community is somehow “entitled” to its belief in
goblins. One might then be able to say that at
this particular stage in the community’s epistemic
evolution, belief in goblins is objectively justifi ed.
Of course, future developments might cause the
community to quit believing in goblins, but in the
meantime, their belief is objectively justifi ed—just
as our epistemic community might previously have
been objectively justifi ed in believing in caloric or
By now, the response to the fi nal counterexample
(the victim of the evil demon) should be clear. The
victim is subjectively justifi ed in her beliefs—she
hasn’t gone wrong in thinking there is an external
world, etc., given the evidence at her disposal—
but her beliefs are not objectively justifi ed. In
this way, the fi nal counterexample can also be
One interesting consequence of an insistence on
the relevance of the internalist perspective is that it
gives us more reasons in favor of conservatism and
the importance of revision. Consider conservatism
rst: when one is trained into a set of beliefs (as
a child, or as a novice scientist or philosopher,
etc.), one is given a starting point that is (from
the perspective of the child or novice scientist)
entirely arbitrary.42 We do not choose which
theory of the world we are dogmatically trained
into as children; this is beyond our control. Thus,
one’s epistemic performance cannot be judged
by the theory one holds initially; rather, the theory
one is trained into is one of the circumstances to
which epistemic evaluation is relativized. One’s
epistemic performance cannot be judged by the
theory one holds because, as I argued earlier,
humans are not capable of evaluating their belief
systems tout court. One is forced to take one’s
belief system as a given, and revise it piecemeal.
Thus, consideration of our epistemic capacities
(and of the fact that internalist justifi cation must
be relative to what is humanly possible) supports
conservatism: one’s epistemic performance
cannot be faulted on the grounds that one cannot
evaluate (and then accept or reject) one’s entire
belief system tout court. One must accept one’s
belief system as given—one must accept it
conservatively—and use this system as the basis
for further revisions.
This emphasis on conservatism in turn supports
giving revision an important role in internalist
epistemic evaluation. Since each of us is given
a more or less arbitrary starting place, it seems
contrary to the spirit of internalism to evaluate a
person’s epistemic performance based on his or
her starting place. After all, we do not choose
which theory of the world we are dogmatically
trained into as children; this is beyond our control
(and hence, we are not to be epistemically praised
or blamed for holding this system). Similarly, as
we noted in our discussion of Kuhn, an aspiring
scientist being trained into the currently-held
scientifi c theory doesn’t have the knowledge to
evaluate whether this theory is correct or not;
only once the scientist has been trained into
the theory and given its epistemic resources
can the scientist then go about trying to prove,
disprove, or revise the theory. We must merely
accept an arbitrary starting place, and then use
that starting place as a platform for revising our
theory, always using (subject to their revision) the
tools that are initially given to us by this starting
place. This strongly suggests that our evaluation of
a person’s epistemic performance should depend
not so much on the person’s starting place (and
whether one has evidence for one’s “starting
theory”), but rather on how one revises one’s
beliefs over time in light of new evidence and
so forth. So it is not epistemically irresponsible or
unsound to hold a belief for which one has no
evidence (we cannot help but do that, since we
are simply dogmatically trained into a particular
practice); rather, it is epistemically unsound for an
agent to revise her theory without being directed
by evidence, or failing to revise the theory in
Epist2_3_06_Koons.indd 214Epist2_3_06_Koons.indd 214 15/8/06 16:52:4115/8/06 16:52:41
the face of new evidence.43 Thus, it seems that
conservatism is supported both as an internal and
an external constraint on justifi cation.
Summary and Conclusion
To sum up, we have seen that conservatism is
an unavoidable part of justifi cation. However, the
conservatism defended in this paper is different
from most versions of conservatism detailed in
the literature. The chief difference is that it only
applies to a certain category of beliefs, those
we have characterized as basic beliefs. Further,
this conservatism is not dogmatic, as these basic
beliefs are justifi ed by the diachronic progress
of our epistemic inquiry. Finally, in response to
a set of potential counterexamples, we saw that
the diachronic requirement on justifi cation served
as an external constraint, which can be viewed
as supplementing a variety of familiar internalist
Adler, Jonathan E. 1990. “Conservatism and Tacit Confi rmation,” Mind 99:396, pp. 559-570.
Adler, Jonathan E.1996. “An overlooked argument for epistemic conservatism,” Analysis 56:2, pp.
Audi, Robert 1993. The Structure of Justifi cation. Cambridge, UK, and New York: Cambridge
University Press.
Brandom, Robert 1997. “Study Guide,” in Wilfrid Sellars, Empiricism and the Philosophy of Mind.
Cambridge, MA: Harvard University Press.
Campbell, Donald 1974. “Evolutionary Epistemology,” in Paul Arthur Schilpp (ed.), The Philosophy
of Karl Popper, The Library of Living Philosophers Volume XIV Book 1. La Salle, Illinois: Open
Court, pp. 413-463.
Christensen, David 1994. “Conservatism in Epistemology,” Noûs 28:1, pp. 69-89.
Dawkins, Richard 1996. Climbing Mount Improbable. New York: W.W. Norton and Company.
Engel, Mylan, Jr. 1992. “Personal and Doxastic Justifi cation in Epistemology”, Philosophical Studies
67:2, pp. 133-150.
Foley, Richard 1982. “Epistemic Conservatism,” Philosophical Studies 43:2, pp. 165-182.
Goldman, Alvin I. 1978. “Epistemics: The Regulative Theory of Cognition,” The Journal of
Philosophy 75:10, pp. 509-523.
Goldman, Alvin I.1986. Epistemology and Cognition. Cambridge, MA: Harvard University Press.
Goldman, Alvin I.1988. “Strong and Weak Justifi cation,” in James E. Tomberlin (ed.), Philosophical
Perspectives 2: Epistemology. Atascadero, California: Ridgeview Publishing Company, pp. 51-
Goldstick, Daniel 1971. “Methodological Conservatism,” American Philosophical Quarterly 8:2,
pp. 186-191.
Goldstick, Daniel 1976. “More on Methodological Conservatism,” Philosophical Studies 30, pp.
Harman, Gilbert 1986. Change in View. Cambridge, MA: MIT Press.
Kitcher, Philip 1993. The Advancement of Science. New York: Oxford University Press.
Kuhn, Thomas S. 1959. “The Essential Tension: Tradition and Innovation in Scientifi c Research,” in
C.W. Taylor, ed., The Third (1959) University of Utah Research Conference on the Identifi cation
of Scientifi c Talent. Salt Lake City: University of Utah Press, pp. 162-74. Reprinted in Kuhn 1977,
pp. 225-39.
Kuhn, Thomas S.1970. The Structure of Scientifi c Revolutions, 2nd edn. Chicago, IL: The University
of Chicago Press.
Lance, Mark 2000. “The Best is the Enemy of the Good: Bayesian epistemology as a case study
in unhelpful idealization,” in Niall Shanks and Robert B. Gardner (eds.) Logic, Probability, and
Science. Amsterdam and Atlanta: Rodopi.
Epist2_3_06_Koons.indd 215Epist2_3_06_Koons.indd 215 15/8/06 16:52:4115/8/06 16:52:41
216 EPISTEME 2006
Jeremy Koons
Lance, Mark and John O’Leary-Hawthorne 1997. The Grammar of Meaning. Cambridge:
Cambridge University Press.
Levi, Isaac 1991. The Fixation of Belief and Its Undoing: Changing Beliefs Through Inquiry. New
York: Cambridge University Press.
Mill, John Stuart 1978 (1859). On Liberty, ed. Elizabeth Rapaport. Indianapolis, IN: Hackett
Publishing Company, Inc.
Quine, W. V. O. 1951. “Two Dogmas of Empiricism,” in The Philosophical Review 60:1 (January),
pp. 20-43.
Sellars, Wilfrid 1956. “Empiricism and the Philosophy of Mind,” in Herbert Feigl and Michael
Scriven, eds., Minnesota Studies in the Philosophy of Science, Vol. I, The Foundations of Science
and the Concepts of Psychology and Psychoanalysis. Minneapolis, MN: University of Minnesota
Press. Reprinted in Sellars 1991, pp. 127-96.
Sellars, Wilfrid 1991: Science, Perception and Reality. Atascadero, CA: Ridgeview Publishing
Sklar, Lawrence 1975. “Methodological Conservatism,” The Philosophical Review 84:3, pp. 374-
Wittgenstein, Ludwig 1958. Philosophical Investigations, 3rd edition, G.E.M. Anscombe, transl.
New York: Macmillan Publishing Company, Inc.
Wittgenstein, Ludwig 1969. On Certainty, G.E.M. Anscombe and G.H. von Wright, eds., Denis
Paul and G.E.M. Anscombe, transl. New York: Harper and Row.
1 Sklar (1975), p. 375.
2 Sklar (1975), p. 396.
3 Sklar (1975), pp. 396-7.
4 Wittgenstein (1969), §94.
5 See, e.g, On Certainty, §110.
6 On Certainty, §98: “the same proposition may get treated at one time as something to test by
experience, at another as a rule of testing.”
7 Sellars 1956, §38.
8 The importance of revisability has been advocated not just by Sellars (1956), but also by Levi
(1991), Brandom (1997), Lance and O’Leary-Hawthorne (1997) and others.
9 On Certainty, §94
10 This account is not intended as an exposition of Wittgenstein. I am arguing that our basic beliefs
are rational; Wittgenstein, on the other hand, held them to be neither rational nor irrational.
11 The evolutionary analogy will go only so far—after all, evolution does not approach any
goal, whereas our empirical inquiry has the goal of knowledge and explanation of the world.
Many have taken evolutionary analogies like this more literally, and used this as fodder for
instrumentalism in the philosophy of science. Discussion of this aspect of the realism/antirealism
debate would, however, take us too far afi eld.
12 Wittgenstein (1958), §217.
13 Again, Wittgenstein thinks these foundational beliefs are arational and not justifi ed, but I think he
is mistaken (for reasons we are now exploring).
14 This is why Mill (1978) writes that we can only know we are right because we allow free inquiry.
In On Liberty, he writes, “There is the greatest difference between presuming an opinion to be true
because, with every opportunity for contesting it, it has not been refuted, and assuming its truth for
the purpose of not permitting its refutation. Complete liberty of contradicting and disproving our
opinion is the very condition which justifi ed us in assuming its truth for purposes of action; and on
no other terms can a being with human faculties have any rational assurance of being right” (18).
Epist2_3_06_Koons.indd 216Epist2_3_06_Koons.indd 216 15/8/06 16:52:4115/8/06 16:52:41
If we don’t allow challenges to orthodoxy, then orthodoxy ceases to be rational. It is only rational
as long as it is before the tribunal of experience; when a belief is given permanent reprieve from
challenges, then it ceases to be justifi ed.
15 If a system became immune to challenge and revision, individuals subscribing to that system might
be perfectly well justifi ed in continuing to have the beliefs dictated by the system in question, and
thinking that the system is rational. But the system itself ceases to be rational. We will discuss this
further later in the paper.
16 Actually, the accuracy of this statement depends on the scope of the ‘we.’ Fideists treat belief
in God as de jure unrevisable; for the Catholic Church, certain moral teachings are de jure
unrevisable. For this reason, such fi deists and Catholics are not justifi ed in holding these beliefs.
17 Lance and O’Leary-Hawthorne (1997), p. 120.
18 This objection was raised by an anonymous associate editor for Episteme.
19 Dawkins (1996), p. 139.
20 Popper (1959), p. 41. Quoted in Campbell (1974), p. 415.
21 Campbell (1974), p. 421.
22 Kitcher (1993), pp. 95-6.
23 Kitcher (1993), p. 105.
24 Kitcher (1993), p. 92, italics in original.
25 Some philosophers have argued, though, that conservatism is supported by pragmatic reasons.
Kuhn (1959) argues that conservatism speeds the progress of science. If these philosophers are
right, then conservatism is supported by both pragmatic and epistemic principles. I will not deny
this, but I will insist that conservatism is at least supported by the latter type of reasons.
26 See, for example, Goldstick (1971) [“…independently of any such empirical grounds the bare
fact that some proposition has been believed by us up to the present should be a consideration
in its favor” (p. 186)]; Goldstick (1976); Foley (1982) [“… a proposition acquires a favorable
epistemic status for a person simply by being believed by him” (p. 165)]; Christensen (1994)
[“…an agent is in some measure justifi ed in maintaining a belief simply in virtue of the fact that
the agent has that belief” (p. 69)]; Adler (1996) [“…believing that p is a reason for belief or
continued belief that p” (p. 80)]. The above quotes merely represent the authors’ characterization
of epistemic conservatism; most of the authors cited in fact reject conservatism as an epistemic
principle. Harman (1986) is one philosopher who supports the principle of conservatism,
27 Wittgenstein (1969), §111.
28 An associate editor for Episteme helped in the formulation of this point.
29 On Certainty, §286.
30 I will discuss these two different perspectives on epistemic justifi cation later in the paper.
31 I argue at the end of the paper that given the fact that an individual cannot evaluate the
rationality of these basic propositions, we should judge individual rationality not on the basis of
the structure of an individual’s set of beliefs, but instead based on how the agent revises his or her
32 Some have argued that revision is crucial to all beliefs, not just basic ones. Mark Lance and John
O’Leary-Hawthorne write, “A practice could in effect adopt the positivist proposal of treating a
whole bunch of claims as de jure unchallengeable…But we would not be tempted to adopt such
a practice ourselves. Such a practice seems to encourage—even be constitutive of—dogmatism,
preclude dialogue, induce cognitive sterility, and all at no obvious gain” (1997, p. 120). I will
not pursue this issue here, but will instead confi ne my argument to the importance of revisability
for basic beliefs.
33 I owe this objection to an anonymous referee.
34 Lance (2000) addresses some of these issues. Bayesians often claim that their account is offered
as a regulative ideal, and that an agent is not necessarily irrational for failing to adhere to
Epist2_3_06_Koons.indd 217Epist2_3_06_Koons.indd 217 15/8/06 16:52:4115/8/06 16:52:41
218 EPISTEME 2006
Jeremy Koons
Bayesian constraints. But this is just an admission that Bayesianism does not offer a complete
theory of rationality, and must be supplemented with something else.
35 Goldman (1978), p. 510. Goldman (1986, pp. 279-283) argues further that standards of
individual rationality should not be tied to what an ‘ideally logical being’ is capable of.
36 A similar example, illustrating a similar distinction between different types of justifi cation, is offered
in Goldman (1988).
37 Other philosophers have also distinguished between different types of justifi cation. Audi (1993)
distinguishes between personal and impersonal justifi cation; Engel (1992) also argues that
there are two different types of justifi cation, corresponding to internalism and externalism. These
distinctions are somewhat different from the one I am drawing here.
38 This distinction between subjective and objective justifi cation is very similar to Goldman’s (1988)
distinction between weak and strong justifi cation.
39 Goldman (1988), p. 52.
40 Goldman (1988), p. 52.
41 I am not entirely sure this is the case. If the refl ective members of the community are aware that
they recently came into existence, and were created with these beliefs, they might question the
source of their beliefs about the world and come, legitimately, to doubt them.
42 See Kuhn (1959).
43 Although Kuhn has pointed out (1959, 1970) that we should not be too quick to revise our
theories in the light of evidence that confl icts with our theory.
Jeremy Koons is an associate professor of philosophy at the American University of Beirut. He
has published several articles, primarily in the areas of epistemology and metaethics.
Epist2_3_06_Koons.indd 218Epist2_3_06_Koons.indd 218 15/8/06 16:52:4215/8/06 16:52:42
ResearchGate has not been able to resolve any citations for this publication.
L'A. critique le principe du conservatisme epistemique, qui amene de facon illegitime a justifier ses propres convictions et a les maintenir dans une attitude dogmatique en introduisant un critere d'evidence dans les mecanismes cognitifs
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies