Two Types of Foundationalism

Author(s): William P. Alston

Source: The Journal of Philosophy , Apr. 8, 1976, Vol. 73, No. 7 (Apr. 8, 1976), pp. 165-
185
Published by: Journal of Philosophy, Inc.

Stable URL: https://www.jstor.org/stable/2025920

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The Journal of
Philosophy

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 *4- -4 -

THE JOURNAL OF PHILOSOPHY

VOLUME LXXIII, NO. 7, APRIL 8, I976

TWO TYPES OF FOUNDATIONALISM

F OUNDATIONALISM is often stated as the doctrine that

knowledge constitutes a structure the foundations of which
support all the rest but themselves need no support. To make
this less metaphorical we need to specify the mode of support
involved. In contemporary discussions of foundationalism knowl-
edge is thought of in terms of true-justified-belief (with or without
further conditions); thus the mode of support involved is justifica-
tion, and what gets supported a belief., The sense in which a foun-
dation needs no support is that it is not justified by its relation to
other justified beliefs; in that sense it does not "rest on" on other
beliefs. Thus we may formulate foundationalism as follows:

I. Our justified beliefs form a structure, in that some beliefs (the

foundations) are justified by something other than their relation
to other justified beliefs; beliefs that are justified by their relation
to other beliefs all depend for their justification on the foundations.

Notice that nothing is said about knowledge in this formulation.

Since the structure alleged by foundationalism is a structure of the
justification of belief, the doctrine can be stated in terms of that
component of knowledge alone. Indeed, one who thinks that
knowledge has nothing to do with justified belief is still faced
with the question of whether foundationalism is a correct view
about the structure of epistemic justification.
Two emendations will render this formulation more perspicuous.
First, a useful bit of terminology. Where what justifies a belief in-

' Contemporary writers on foundationalism do not seem to notice that Des-

cartes and Locke have a quite different view of knowledge and, hence, that, if
they hold that knowledge rests on foundations, this will mean something rather
different. See below, p. 181, for a translation of a bit of Descartes into current
foundationalist idiom.

I65

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
i66 THE JOURNAL OF PHILOSOPHY

cludes 2 the believer's having certain other justified beliefs, so re-

lated to the first belief as to embody reasons or grounds for it, we
may speak of indirectly (mediately) justified belief. And, where
what justifies a belief does not include any such constituent, we
may speak of directly (immediately) justified belief. Correspond-
ingly, a case of knowledge in which the justification requirement is
satisfied by indirect (mediate) justification will be called indirect
(mediate) knowledge; and a case in which the justification require-
ment is satisfied by direct (immediate) justification will be called
direct (immediate) knowledge.
Second, we should make more explicit how mediate justification
is thought to rest on immediately justified belief. The idea is that,
although the other beliefs involved in the mediate justification of a
given belief may themselves be mediately justified, if we continue
determining at each stage how the supporting beliefs are justified,
we will arrive, sooner or later, at directly justified beliefs. This
will not, in general, be a single line of descent; typically the belief
with which we start will rest on several beliefs, each of which in
turn will rest on several beliefs. So the general picture is that of
multiple branching from the original belief.
With this background we may reformulate foundationalism as fol-
lows (turning the "foundation" metaphor on its head):

II. Every mediately justified belief stands at the origin of a (more or

less) multiply branching tree structure at the tip of each branch
of which is an immediately justified belief.

II can be read as purely hypothetical (if there are any mediately

justified beliefs, then ...) or with existential import (There are
mediately justified beliefs, and . . .). Foundationalists typically make
the latter claim, and I shall understand the doctrine to carry existen-
tial import.
II can usefully be divided into two claims:

(A) There are directly justified beliefs.

(B) A given person has a stock of directly justified beliefs sufficient to

generate chains of justification that terminate in whatever indi-
rectly justified beliefs he has.

In other words, (A) there are foundations, and (B) they suffice to
hold up the building.

2 Only 'includes', because other requirements are also commonly imposed for
mediate justification, e.g., that the first belief be "based" on the others, and,
by some epistemologists, that the believer realize that the other beliefs do
constitute adequate grounds for the first.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM i67

In this paper we shall restrict our attention to A. More spe-

cifically, we shall be concerned with a certain issue over what it
takes for a belief to serve as a foundation.
I. THE SECOND-LEVEL ARGUMENT

Let's approach this issue by confronting foundationalism with a

certain criticism, a recent version of which can be found in Bruce
Aune.3

The line of reasoning behind the empiricist's assumption is, again,

that while intra-language rules may validly take us from premise to
conclusion, they cannot themselves establish empirical truth. If the
premises you start with are false, you will have no guarantee that the
conclusions you reach are not false either. Hence, to attain knowl-
ledge of the actual world, you must ultimately have premises whose
truth is acceptable independently of any inference and whose status is
accordingly indubitable. Only by having such premises can you gain a
starting point that would make inference worthwhile. For con-
venience, these indispensable basic premises may be called "intrin-
sically acceptable." The possibility of empirical knowledge may then
be said to depend on the availability of intrinsically acceptable
premises.
If this line of thought is sound, it follows that utter scepticism
can be ruled out only if one can locate basic empirical premises that
are intrinsically acceptable. Although philosophers who attack scep-
ticism in accordance with this approach generally think they are de-
fending common sense, it is crucial to observe that they cannot actually
be doing so. The reason for this is that, from the point of view of
common experience, there is no plausibility at all in the idea that
intrinsically acceptable premises, as so defined, ever exist. Philosophers
defending such premises fail to see this because they always ignore the
complexity of the situation in which an empirical claim is evaluated.
I have already given arguments to show that introspective claims
are not, in themselves, intrinsically infallible, they may be regarded as
virtually certain if produced by a reliable (sane, clear-headed) ob-
server, but their truth is not a consequence of the mere fact that they
are confidently made. To establish a similar conclusion regarding
the observation claims of everyday life only the sketchiest arguments
are needed. Obviously the mere fact that such a claim is made does
not assure us of its truth. If we know that the observer is reliable,
made his observation in good light, was reasonably close to the object,
and so on, then we may immediately regard it as acceptable. But its
acceptability is not intrinsic to the claim itself ... I would venture to
say that any spontaneous claim, observational or introspective, carries
almost no presumption of truth, when considered entirely by itself. If

3 Knowledge, Mind and Nature (New York: Random House, 1967).

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
i68 THE JOURNAL OF PHILOSOPHY

we accept such a claim as true, it is only because of our confidence

that a complex body of background assumptions-concerning observ-
ers, standing conditions, the kind of object in question-and, often, a
complex mass of further observations all point to the conclusion that
it is true.
Given these prosaic considerations, it is not necessary to cite experi-
mental evidence illustrating the delusions easily brought about by, for
example, hypnosis to see that no spontaneous claim is acceptable
wholly on its own merits. On the contrary, common experience is
entirely adequate to show that clear-headed men never accept a claim
merely because it is made, without regard to the peculiarities of the
agent and of the conditions under which it is produced. For such
men, the acceptability of every claim is always determined by infer-
ence. If we are prepared to take these standards of acceptability seri-
ously, we must accordingly admit that the traditional search for intrin-
sically acceptable empirical premises is completely misguided (41-43).

Now the target of Aune's critique differs in several important

respects from the foundationalism defined above. First and most
obviously, Aune supposes that any "intrinsically acceptable prem-
ises" will be infallible and indubitable, and some of his arguments
are directed specifically against these features.4 Second, there is an
ambiguity in the term 'intrinsically acceptable'. Aune introduces it
to mean "whose truth is acceptable independently of any inference,"
this looks roughly equivalent to our 'directly justified'. However in
arguing against the supposition that the "observation claims of
everyday" are intrinsically acceptable, he says that "the mere fact
that such a claim is made does not assure us of its truth," thereby
implying that to be intrinsically acceptable a claim would have to
be justified just by virtue of being made. Now it is clear that a
belief (claim) of which this is true is directly justified, but the
converse does not hold. A perceptual belief will also be directly
justified, as that term was explained above, if what justifies it is the
fact that the perceiver "is reliable, made his observation in good
light, was reasonably close to the object, and so on," provided it is
not also required that the he be justified in believing that these con-
ditions are satisfied. Thus this argument of Aune's has no tendency
to show that perceptual beliefs cannot be directly justified, but only
that they cannot enjoy that special sort of direct justification which
we may term "self-justification." 5

4See the distinctions between infallibility, indubitability, and immediacy in

my "Varieties of Privileged Access," American Philosophical Quarterly, viI, 3
(July 1971): 223-241.
5 In "Varieties of Privileged Access" I use the term 'self-warrant' for a belief
that is justified by virtue of being a belief of a certain sort.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM i69

However some of Aune's arguments would seem to be directed

against any immediate justification, and a consideration of these
will reveal a third and more subtle discrepancy between Aune's
target(s) and my version of foundationalism. Near the end of the
passage Aune says:

If we accept such a claim [observational or introspective] as true, it is

only because of our confidence that a complex of background as-
sumptions ... all point to the conclusion that it is true.
And again:
For such men [clear-headed men], the acceptability of every claim is
always determined by inference.
It certainly looks as if Aune is arguing that whenever a claim (be-
lief) is justified it is justified by inference (by relation to other
justified beliefs); and that would be the denial of 'There are directly
justified beliefs'. But look more closely. Aune is discussing not what
would justify the issuer of an introspective or observational claim
in his belief, but rather what it would take to justify "us" in accept-
ing his claim; he is arguing from a third-person perspective. Now it
does seem clear that I cannot be immediately justified in accepting
your introspective or observational claim as true. If I am so justified
it is because I am justified in supposing that you issued a claim of
that sort, that you are in a normal condition and know the lan-
guage, and (if it is an observational claim) that conditions were
favorable for your accurately perceiving that sort of thing. But that
is only because I, in contrast to you, am justified in believing that p
(where what you claimed is that p, and where I have no inde-
pendent access to p) only if I am justified in supposing that you
are justified in believing that p. My access to p is through your
access. It is just because my justification in believing that p pre-
supposes my being justified in believing that you are justified, that
my justification has to be indirect. That is why I have to look
into such matters as conditions of observation, and your normality.
Thus what Aune is really pointing to is the necessity for "inferential"
backing for any higher-level belief to the effect that someone is
justified in believing that p. (I shall call such higher-level beliefs
epistemic beliefs). His argument, if it shows anything, shows that no
epistemic belief can be immediately justified. But it does nothing to
show that the original observer's or introspector's belief that p was
not immediately justified. Hence his argument is quite compatible
with the view that an introspective belief is self-justified and with
the view that an observational belief is justified just by being
formed in favorable circumstances.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
170 THE JOURNAL OF PHILOSOPHY

As a basis for further discussion I should like to present my own

version of an argument against the possibility of immediate justi-
fication for epistemic beliefs-what I shall call the second-level
argument:

Al Where S's belief that p is mediately justified, any jurisdiction for

the belief that S is justified in believing that p is obviously mediate.
For one could not be justified in this latter belief unless it were based
on a justified belief that S is justified in accepting the grounds on
which his belief that p is based. But even where S is immediately justi-
fied in believing that p, the higher-level belief will still be mediately
justified, if at all. For in taking a belief to be justified, we are evaluat-
ing it in a certain way.6 And, like any evaluative property, epistemic
justification is a supervenient property, the application of which is
based on more fundamental properties. A belief is justified because it
possesses what Roderick Firth has called "warrant-increasing proper-
ties." 7 Hence in order for me to be justified in believing that S's belief
that p is justified, I must be justified in certain other beliefs, viz.,
that S's belief that p possesses a certain property, Q. and that Q
renders its possessor justified. (Another way of formulating this last
belief is: a belief that there is a valid epistemic principle to the effect
that any belief that is Q is justified.) Hence in no case can an epistemic
belief that S is justified in believing that p, itself be immediately
justified.

Before proceeding I shall make two comments on this argument

and its conclusion.
(1) It may appear that the conclusion of the argument is incom-
patible with the thesis that one cannot be justified in believing that
p without also being justified in believing that one is justified in
believing that p. For if being immediately justified in believing that
p necessarily carried with it being justified in believing that I am
justified in believing that p, it would seem that this latter justifi-
cation would be equally immediate. I would not shirk from such an
incompatibility, since I feel confident in rejecting that thesis. It is
not clear, however, that there is any such incompatibility. It all
depends on how we construe the necessity. If, e.g., it is that my
being justified in believing that p necessarily puts me into pos-
session of the grounds I need for being justified in the higher-level

6 For one attempt to explain the distinctively epistemic dimension of evalua-

tion, see R. M. Chisholm, "On the Nature of Empirical Evidence," in Chisholm
and R. J. Swartz, eds., Empirical Knowledge (Englewood Cliffs, N.J.: Prentice-
Hall, 1973), pp. 225-230.
7 In "Coherence, Certainty, and Epistemic Priority," this JOURNAL, LXI, 19
(Oct. 15, 1964): 545-557.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM 171

belief, then that is quite compatible with our conclusion that the
latter can only be mediately justified.
(2) The conclusion should not be taken to imply that one must
perform any conscious inference to be justified in an epistemic
belief, or even that one must be explicitly aware that the lower-level
belief has an appropriate warrant-increasing property. Here as in
other areas, one's grounds can be possessed more or less implicitly.
Otherwise we would have precious little mediate knowledge.
I have already suggested that the second-level argument is not
really directed against II. To be vulnerable to this argument, a
foundationalist thesis would have to require of foundations not
only that they be immediately justified, but also that the believer
be immediately justified in believing that they are immediately
justified. A position that does require this we may call iterative
foundationalism, and we may distinguish it from the earlier form
(simple foundationalism) as follows (so far as concerns the status
of the foundations):

Simple Foundationalism: For any epistemic subject, S. there are p's

such that S is immediately justified in believing that p.

Iterative Foundationalism: For any epistemic subject, S. there are p's

such that S is immediately justified in believing that p and S is im-
mediately justified in believing that he is immedately justified in be-
lieving that p.'s

It would not take much historical research to show that both posi-
tions have been taken. What I want to investigate here is which of
them there is most reason to take. Since the classic support for foun-
dationalism has been the regress argument, I shall concentrate on
determining which form emerges from that line of reasoning.
II. THE REGRESS ARGUMENT

The regress argument seeks to show that the only alternatives to

admitting epistemic foundations are circularity of justification or an
equally unpalatable infinite regress of justification. It may be formu-

8 One should not confuse the respect in which Iterative is stronger than
Simple Foundationalism with other ways in which one version of the position
may be stronger than another. These include at least the following: (1) whether
it is required of foundations that they be infallible, indubitable, or incorrigible;
(2) whether foundations have to be self-justified, or whether some weaker form
of direct justification is sufficient; (3) how strongly the foundations support
various portions of the superstructure. I am convinced that none of these modes
of strength requires any of the others, but I will not have time to argue that
here. Note too that our version of the regress argument (to be presented in a
moment) does nothing to support the demand for foundations that are strong
in any of these respects,

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 172 THE JOURNAL OF PHILOSOPHY

lated as follows:

A2 Suppose we are trying to determine whether S is mediately justified in

believing that p. To be so justified he has to be justified in believing
certain other propositions, q, r, ... that are suitably related to p (so as
to constitute adequate grounds for p). Let's say we have identified a set
of such propositions each of which S believes. Then he is justified in
believing that p only if he is justified in believing each of those proposi-
tions.9 And, for each of these propositions q, r, ... that he is not im-
mediately justified in believing, he is justified in believing it only if he
is justified in believing some other propositions that are suitably re-
lated to it. And for each of these latter propositions ...
Thus in attempting to give a definitive answer to the original ques-
tion we are led to construct a more or less extensive true structure, in
which the original belief and every other putatively mediately justified
belief form nodes from which one or more branches issue, in such a
way that every branch is a part of some branch that issues from the
original belief. Now the question is: what form must be assumed by the
structure in order that S be mediately justified in believing that p?
There are the following conceivable forms for a given branch:
A. It terminates in an immediately justified belief.

B. It terminates in an unjustified belief.

C. The belief that p occurs at some point (past the origin), so that the
branch forms a loop.

D. The branch continues infinitely.

Of course some branches might assume one form and others another.
The argument is that the original belief will be mediately justified
only if every branch assumes form A. Positively, it is argued that on this
condition the relevant necessary condition for the original belief's being
mediately justified is satisfied, and, negatively, it is argued that if any
branch assumes any of the other forms, is not.

A. Where every branch has form A, this necessary condition is satisfied

for every belief in the structure. Since each branch terminates in an
immediately justified belief that is justified without necessity for
further justified beliefs, the regress is ended along each branch.
Hence justification is transferred along each branch right back -to
the original belief.

B. For any branch that exhibits form B, no element, even the origin, is
justified, at least by this structure. Since the terminus is not justified,
the prior element, which is justified only if the terminus is, is not
justified. And, since it is not justified, its predecessor, which is justi-

9 am adopting the simplifying assumption that, for each mediately justified

belief, there is only one set of adequate grounds that S justifiably believes. The
argument can be formulated so as to allow for "overjustification," but at the
price of further complexity.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM I73

fied only if it is, is not justified either. And so on, right back to the
origin, which therefore itself fails to be justified.

C. Where we have a branch that forms a closed loop, again nothing on

that branch, even the origin, is justified, so far as its justification
depends on this tree structure. For what the branch "says" is that the
belief that p is justified only if the belief that r is justified, and that
belief is justified only if ...., and the belief that z is justified only if
the belief that p is justified. So what this chain of necessary condi-
tions tells us is that the belief that p is justified only if the belief
that p is justified. True enough, but that still leaves it completely
open whether the belief that p is justified.
D. If there is a branch with no terminus, that means that no matter
how far we extend the branch the last element is still a belief that is
mediately justified if at all. Thus, as far as this structure goes, wher-
ever we stop adding elements we have still not shown that the rele-
vant necessary condition for the mediate justification of the original
belief is satisfied. Thus the structure does not exhibit the original
belief as mediately justified.
Hence the original belief is mediately justified only if every branch in
the tree structure terminates in an immediately justified belief. Hence
every mediately justified belief stands at the origin of a tree structure
at the tip of each branch of which is an immediately justified belief.10
Now this version of the argument, analogues of which occur fre-
quently in the literature," supports only simple foundationalism. It
has no tendency to show that there is immediately justified epistemic
belief. So long as S is directly justified in believing some t for each
branch of the tree, that will be. quite enough to stop the regress; for
all that is needed is that he be justified in believing t without
thereby incurring the need to be justified in believing some further
10 The weakest link in this argument is the rejection of D. So far as I am
aware, this alternative is never adequately explained, and much less is adequate
reason given for its rejection. Usually, I fear, being justified is confused with
exhibiting one's justification, and it is argued (irrelevantly) that one cannot
do the latter for an infinite sequence of propositions. It is interesting in this
connection that in two very recent attacks on foundationalism the infinite
regress rejected by the regress argument is construed as a regress of showing
justification, and in different ways the critics argue that the impossibility of
completing an infinite sequence of such showings does not imply that there may
not be an infinite sequence of mediate justification. See Keith Lehrer, Knowledge
(Oxford: Clarendon Press, 1974), pp. 15/6, and Frederick L. Will, Induction
and Justification (Ithaca, N.Y.: Cornell, 1974), pp. 176-185.
An adequate treatment of the argument would involve looking into the
possibility of an infinite structure of belief and the patterns of justification that
can obtain there. Pending such an examination, the most one can say for
the argument is that it is clear that mediate justification is possible on alterna-
tive A and not clear that it is possible on alternative D.
"'See, e.g., Bertrand Russell, Human Knowledge, Its Scope and Limits (Lon-
don: Allen & Unwin, 1948), p. 171; Anthony Quinton, The Nature of Things
(London: Routledge & Kegan Paul, 1973), p. 119.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
I 74 THE JOURNAL OF PHILOSOPHY

proposition. But perhaps there are other versions that yield the
stronger conclusion. Indeed, in surveying the literature one will dis-
cover versions that differ from A2 is one or both of the following
respects:

1. Their starting points (the conditions of which they seek to establish)

are cases of being justified in believing that one knows (is justified in
believing) that p, rather than, more generally, cases of being justified in
believing that p.

2. They are concerned to establish what is necessary for showing that p,

rather than what is necessary for being justified in believing that p.

Let's consider whether regress arguments with one or the other of

these features will yield iterative foundationalism.
First let's consider an argument that differs from A2 only in the
first respect. In his essay "Theory of Knowledge" in a volume de-
voted to the history of twentieth-century American philosophy, R.
M. Chisholm 12 launches a regress argument as follows:

To the question "What justification do I have for thinking that I

know that a is true?" one may reply: "I know that b is true, and if I
know that b is true then I also know that a is true." And to the ques-
tion "What justification do I have for thinking I know that b is true?"
one may reply: "I know that c is true, and if I know that c is true then
I also know that b is true." Are we thus led, sooner or later, to some-
thing, n, of which one may say "What justifies me in thinking I know
that n is true is simply the fact that n is true"? (263)

Chisholm then supports an affirmative answer to this last question

by excluding other alternatives in a manner similar to that of A2.
Now the crucial question is: why does Chisholm conclude not
just that mediate justification of claims to know requires some im-
mediately justified beliefs, but that it requires immediately justified
epistemic beliefs? Of course, having granted the general position
that any mediately justified belief rests on some immediately justi-
fied belief(s), it is natural to suppose that mediately justified epi-
stemic beliefs will rest on immediately justified epistemic beliefs.
But we should not assume that all cases of mediate knowledge rest
12 Philosophy (Englewood Cliffs, N.J.: Prentice-Hall, 1964). Because of the
ambiguity of the term 'knowledge claim', formulations and criticisms of the
argument are often ambiguous in the present respect. When we ask how a
"knowledge claim" is justified, we may be asking what it takes to justify an
assertion that p or we may be asking what it takes to justify a claim that one
knows that p. Thus, e.g., we find Arthur Danto beginning the argument by
speaking of m being justified in asserting s but then sliding into a consideration
of what it takes to justify "claims to know" [Analytical Philosophy of Knowledge
(New York: Cambridge, 1968), pp. 26-28].

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISNM 175

on foundations that are similar in content. On the contrary, every

version of foundationalism holds that from a certain set of basic
beliefs one erects a superstructure that is vastly different from these
foundations. From knowledge of sense data one derives knowledge
of public physical objects, from knowledge of present occurrences
one derives knowledge of the past and future, and so on. So why
suppose that if mediate epistemic beliefs rest on foundations, those
foundations will be epistemic beliefs? We would need some special
reason for this. And neither Chisholm nor, to my knowledge, any-
one else has given any such reason. All rely on essentially the same
argument as A2, which at most yields the weaker conclusion. They
seem to have just assumed uncritically that the foundations on
which epistemic beliefs rest are themselves epistemic.13
Thus, altering the regress argument in the first way does not pro-
vide grounds for iterative foundationalism. Let's turn to the second
modification. In order to maximize our chances, let's combine it
with the first, and consider what it would take to show, for some p,
that I am justified in believing that p.14 It is easy to see how one

13 Lest this assumption still seem obvious to some of my readers, let me take
a moment to indicate how mediate epistemic knowledge might conceivably be
derived from nonepistemic foundations. Let's begin the regress with Chisholm
and follow the line of the first ground he mentions: that I justifiably believe that
b. (To simplify this exposition I am replacing 'know' with 'justifiably believe'
throughout.) By continuing to raise the same question we will at last arrive at
a c such that I have immediate justification for believing that c. Here my
justification (for believing that I justifiably believe that c) will shift from one or
more other justified beliefs to the appropriate "warrant-increasing" property.
What is then required at the next stage is a justification for supposing the
belief that c to have this property, and for supposing that this property does
confer warrant. It is highly controversial just how claims like these are to be
justified, but, in any event, at this point we have exited from the arena of ex-
plicit claims to being justified in a certain belief; what needs justification from
here on are beliefs as to what is in fact the case, and beliefs as to what principles
of evaluation are valid, not beliefs as to my epistemic relation to these matters,
And, without attempting to go into the details, it seems plausible that, if a
foundationalist view is tenable at all, these sorts of beliefs will rest on the same
sort of foundation as other factual and evaluative beliefs.
14 I have not located a clear-cut example of a regress argument with this start-
ing point and with the conclusion in question. Nevertheless, the prospect seems
tempting enough to be worth deflating. Moreover, it forces us to raise interesting
questions concerning the concept of showing.
Just as the ambiguity of 'knowledge claim' led to versions' of the regress argu-
ment being indeterminate with respect to the earlier feature, so the process-
product ambiguity of terms like 'justification' and 'justified' often make it uncer-
tain whether a philosopher is talking about what it takes for a belief to be justi-
fied or about what it takes to justify a belief in the sense of showing it to be
justified. See, e.g., C. I. Lewis, An Analysis of Knowledge and Valuation (La
Salle, Ill.: Open Court, 1946), p. 187; Leonard Nelson, "The Impossibility of the
'Theory of Knowledge'," in Chisholm & Swartz, op. cit., p. 8.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
176 THE JOURNAL OF PHILOSOPHY

might be led into this. One who accepted the previous argument
might still feel dissatisfied with simple foundationalism. "You
have shown," he might say, "that it is possible to be justified in
believing that p without having any immediately justified epistemic
belief. But are we in fact justified in believing any p? To answer
that question you will have to show, for some p, that you are justi-
fied in believing it. And the question is, what is required for that?
Is it possible to do that without immediately justified epistemic
belief?"
Now if we are to show, via a regress argument, that immediately
justified epistemic belief is necessary for showing that I am justified
in believing any p, it must be because some requirement for show-
ing sets up a regress that can only be stopped if we have such beliefs.
What could that requirement be? Let's see what is required for
showing that p. Clearly, to show that p I must adduce some other
(possibly compound) proposition, q. What restrictions must be put
on a q and my relations thereto?

1. It is true that q.15

2. q constitutes adequate grounds for p.

These requirements give rise to no regress, or at least none that is

vicious. Even if no proposition can be true without some other
proposition's being true, there is nothing repugnant about the no-
tion of an infinity of true propositions. Hence we may pass on.

3. I am justified in believing that q.16

This requirement clearly does give rise to a regress, viz., that al-
ready brought out in A2. We have seen that immediately justified
epistemic belief is not required to end that regress; so again we
may pass on.

4. I am justified in believing that I am justified in believing that q.

I am not prepared to admit this requirement, my reasons being

closely connected with the point that one may be justified in be-
15 It may also be required that p be true, on the ground that it makes no
sense to speak of my having shown what is not the case. ('Show' is a success con-
cept.) I neglect this point since it has no bearing on our present problem.
16 One may contest this requirement on the grounds that, if I have produced
what is in fact a true adequate ground, that is all that should be demanded.
And it may be that there is some "objective" concept of showing of which this
is true. Nevertheless where we are interested in whether Jones has shown that p
(rather than just whether "it has been shown that p," where perhaps all we are
interested in is whether there are true adequate grounds), it seems that we
must adopt this requirement in order to exclude wildly accidental cases in
which Jones is asserting propositions at random and just happens to hit the mark.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM 177

lieving that q without even believing that one is so justified, much

less being justified in believing that one is so justified. However, it is
not necessary to discuss that issue here. Even if 4 is required, it will
simply set up a regress of the sort exemplified by Chisholm's argu-
ment, an argument we have seen to have no stronger conclusion
than simple foundationalism.

5. I am able to show that q.

This looks more promising. Clearly this requirement gives rise to a

regress that is different from that of A2. If I can show that p by
citing q only if I am able to show that q, and if, in turn, I am able
to show that q by citing r only if I am able to show that r, it is clear
that we will be able to avoid our familiar alternatives of circularity
and infinite regress, only if at some point I arrive at a proposition
that I can show to be correct without appealing to some other
proposition. In deciding whether this argument provides support
for iterative foundationalism, we must consider first whether re-
quirement 5 is justified and, second, whether immediately justified
epistemic belief would stop the regress so generated.
The requirement looks plausible. For, if I cannot show that q,
then it looks as if I won't be able to settle whether or not it is
the case that q, and in that case how can I claim to have settled
the question about p? But this plausibility is specious, stemming
from one of the protean forms assumed by that confusion of levels
typified by the confusion of knowing that p with knowing that one
knows that p. It's quite true that an inability to show that q will
prevent me from showing that I have shown that p; for to do the
latter I have to show that the grounds I have cited for p are correct.
But why suppose that it also prevents me from showing that p?
Can't I prove a theorem in logic without being able to prove that
I have proved it? The former requires only an ability to wield the
machinery of first-order logic, which one may possess without the
mastery of metalogic required for the second. Similarly, it would
seem that I can show that p, by adducing true adequate grounds I
am justified in accepting, without being able to show that those
grounds are true.
This conclusion is reinforced by the point that it is all too possi-
ble to have adequate grounds for a belief without being able to
articulate them. Having observed Jones for a while, I may have
adequate reasons for supposing him to be unsure of himself, with-
out being able to specify just what features of his bearing and be-
havior provide those reasons. A philosophically unsophisticated per-

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
I78 THE JOURNAL OF PHILOSOPHY

son (and many of the philosophically sophisticated as well) may be

amply justified in believing that there is a tree in front of his
wide-open eyes, but not be able to show that he is so justified. I
may be justified in believing that Louis IX reigned in the thirteenth
century, since I acquired that belief on excellent authority, but not
now be able to specify that authority, much less show that it is
reliable. Of course in the case under discussion, I am able to articu-
late my grounds for p, for ex hypothesi I have adduced adequate
grounds for p. But to suppose that it is reasonable to require that I
be able to show that those grounds are true, and the grounds of
these grounds, and ... is to ignore the elementary point that a
person may have adequate grounds for q and so be in an epi-
stemically sound position vis-4-vis q, without being able to articu-
late those grounds. The latter ability is the exception rather than
the rule with mediately justified belief.
But even if requirement 5 were justified and the show-regress
were launched, immediately justified epistemic beliefs would be
powerless to stop it. Let's say that I originally set out to show that I
am justified in believing that a, and in the regress of showings
thus generated I eventually cite as a ground that I am immediately
justified in believing that z (call this higher-level proposition "Z"),
where I am in fact immediately justified in believing that Z. How
will this latter fact enable me to show that Z? As a result of being
immediately justified in believing that Z, I may have no doubt
about the matter; I may feel no need to show myself that Z. But of
course that doesn't imply that I have shown that Z. However im-
mediate my justification for accepting Z, I haven't shown that Z un-
less I adduce grounds for it that meet the appropriate conditions.
And once I do that we are off to the races again. The regress has
not been stopped. In the nature of the case it cannot be stopped.
In this it differs from the original regress of being justified. Show-
ing by its very nature requires the exhibition of grounds. Further-
more, grounds must be different from the proposition to be shown.
(This latter follows from the "pragmatic" aspect of the concept of
showing. To show that p is to present grounds that one can justifi-
ably accept without already accepting p. Otherwise showing would
lack the point that goes toward making it what it is.) Hence, there
are no conceivable conditions under which I could show that p
without citing other propositions that, by requirement 5, I must be
able to show. If we accept requirement 5, if an infinite structure
of abilities to show is 'ruled out, and if circularity is unacceptable,
it follows that it is impossible ever to show anything. (That would

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM I79

seem to be an additional reason for rejecting 5.) Since immediately

justified epistemic belief would do nothing to stop the regress, this
kind of regress argument can provide no support for iterative
foundationalism.
III. FUNCTIONS OF FOUNDATIONALISM

Thus, although simple foundationalism is strongly supported by

A2, we have failed to find any argument that supports iterative
foundationalism. And the second-level argument strikes at the latter
but not the former. Hence it would seem that foundationalism has a
chance of working only in its simple form. This being the case,
it is of some interest to determine the extent to which simple foun-
dationalism satisfies the demands and aspirations that foundation-
alism is designed to satisfy, other than stopping the regress of justi-
fication. I shall consider two such demands.

Answering Skepticism. Skepticism assumes various forms, many of

which no sort of foundationalism could sensibly be expected to
answer. For example, the extreme skeptic who refuses to accept any-
thing until it has been shown to be true and who will not allow
his opponent any premises to use for this purpose, obviously can-
not be answered whatever one's position. Talking with him is a los-
ing game. Again there are more limited skepticisms in which one
sort of knowledge is questioned (e.g., knowledge of the conscious
states of other persons) but others are left unquestioned (e.g., knowl-
edge of the physical environment). Here the answering will be
done, if at all, by finding some way of deriving knowledge of the
questioned sort from knowledge of the unquestioned sort. The
role of a general theory of knowledge will be limited to laying down
criteria for success in the derivation, and differences over what is
required for foundations would seem to make no difference to such
criteria.
The kind of "answer to skepticism" that one might suppose to be
affected by our difference is that in which the skeptic doubts that
we have any knowledge, a successful answer being a demonstration
there is some. One may think that the possession of immediate
epistemic knowledge will put us in a better position to do that job.
Whether it does, and if so how, depends on what it takes to show
that one knows something. The discussion of showing in section II
yielded the following conditions for S's showing that p:
1. It is true that p.

2. S cites in support of p a certain proposition q such that:

A. It is true that q.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
i8o THE JOURNAL OF PHILOSOPHY

B. q is an adequate ground for p.

C. S is justified in believing q.

We rejected the further condition that S be able to show that q.

However, since we are here concerned with showing something
to a skeptic, it may be that some further requirement should be
imposed. After all, we could hardly expect a skeptic to abandon
his doubt just on the chance that his interlocuter is correct in
the grounds he gives. The skeptic will want to be given some
reason for supposing those grounds to be correct; and this does
not seem unreasonable. But we can't go back to the unqualified re-
quirement that every ground adduced be established or even estab-
lishable, without automatically making showing impossible. For-
tunately there is an intermediate requirement that might satisfy
a reasonable skeptic while not rendering all showing impossible.
Let's require that S be able to show that r, for any r among his
grounds concerning which his audience has any real doubt. This
differs from the unqualified requirement in leaving open the possi-
bility that there will be grounds concerning which no reasonable
person who has reflected on the matter will have any doubt; and
if there be such it may still be possible for S to succeed in showing
that p. Thus we may add to our list of conditions:

D. If there is real doubt about q, S is able to show that q.

Now when p is 'S knows that a', the question is whether one or
more of these conditions is satisfiable only if S has immediately justi-
fied epistemic beliefs. Let's consider the conditions in turn. As for
1, S can in fact know that a without having any directly justified
epistemic belief, even if it should be the case that one can't know
that a without knowing that one knows that a. For, as we saw in sec-
tion II, there is no reason to doubt that all justified beliefs that one
knows or is justified in believing something are themselves mediately
justified. As for 2A and 2B, there should be no temptation to sup-
pose that they depend on iterative foundationalism. Surely the
grounds I adduce for the claim to know that a can be true and
adequate without my having any immediately justified epistemic
beliefs. Even if one or more of the grounds should themselves be
claims to knowledge, the question of what is required for their
truth can be handled in the same way as requirement 1. And
adequacy, being a matter of relations between propositions, can-
not depend on what sort of justification S has for one or another

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM i8i

belief. As for 2C, the discussion in sections i and ii failed to turn up

any reasons for supposing that immediately justified epistemic
belief is required for my being justified in believing anything.
That leaves 2D. But this has already been covered. To satisfy 2D I
have to be able to show that (some of) my grounds are true. But
that will not require conditions that are different in kind from
those already discussed. Hence we may conclude that iterative
foundationalism is not a presupposition of our showing that we do
have knowledge. Of course it remains an open question whether
we are in fact capable of showing that we know something. But
if we are incapable, it is not because of the lack of immediately
justified-epistemic belief.

Self-consciously Reconstructing Knowledge from the Foundations.

What I have in mind here is the enterprise classically exemplified by
Descartes in the Meditations. There Descartes first sets out to iden-
tify those items for which there could not be any grounds for
doubt. Having done so, he seeks to use these items as a basis for
showing that other items are known as well. Now we cannot assimi-
late Descartes to our scheme without some adjustments. For one
thing, Descartes required indubitability and infallibility of his foun-
dations. For another, he was not working with a true-justified-belief
conception of knowledge. Translating Descartes into the conception
of knowledge we are using and ignoring the extra demands of in-
dubitability and infallibility, it is clear that Descartes takes his
foundational beliefs to be immediately justified. I am justified in
believing that I exist or that I am presently thinking about epistem-
ology, regardless of what else I may be justified in believing. I am
so justified just by the fact that the belief "records" the content of
a clear and distinct intuition of the fact that makes the belief true.
Hence, in order to identify a belief, B, as foundational Descartes
must be justified in the higher-level belief that B is immediately
justified. And if he is to perform this identification at the outset of
his reconstruction, when nothing is recognized as mediately justi-
fied, this justification must be immediate, since he lacks a suitable
body of other beliefs on which to base it.17 Hence this enterprise is
possible only if one can be immediately justified in taking a certain

17 To be sure, this short treatment leaves open the abstract possibility that the
first such higher-level belief might be justified by some of the lower-level beliefs
among the current foundations (if indeed the rules of the game permit their use
in justification without first having been justifiably recognized as immediately
justified). But it is clear that the foundational beliefs Descartes recognizes are
radically unsuitable for this employment.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
i82 THE JOURNAL OF PHILOSOPHY

belief to be immediately justified. Here, then, is a point at which

iterative foundationalism is genuinely needed.18
If iterative foundationalism is both without strong support and
subject to crushing objections, it looks as if we will have to do with-
out a self-conscious reconstruction of knowledge. How grievous a
loss is this? Why should anyone want to carry out such a reconstruc-
tion? Well, if knowledge does have a foundational structure it
seems intolerable that we should be unable to spell this out. And
it may seem that such a spelling out would have to take the present
form. But that would be an illusion. If there are foundations, one
can certainly identify them and determine how other sorts of
knowledge are based on them without first taking on the highly
artificial stance assumed by Descartes in the Meditations. One can
approach this problem, as one approaches any other, making use
of whatever relevant knowledge or justified belief one already
possesses. In that case immediate epistemic knowledge is by no
means required, just as we have seen it is not required to show that
one is justified in holding certain beliefs.
The Cartesian program has been branded as unrealistic on more
grounds than one. And if I am right in holding that the simple
form of foundationalism is the most we can have, I have provided
one more ground. If iterative foundationalism is false, we can still
have as much epistemic knowledge as you like, but only after we
have acquired quite a lot of first-level knowledge. And why should
that not satisfy any epistemic aspirations that are fitting for the
human condition?
IV. ENVOI

As we have seen, the main reason for adopting foundationalism is

the seeming impossibility of a belief's being mediately justified with-
out resting ultimately on immediately justified belief. And the
main reason for rejecting it (at least the main antecedent reason,
apart from the difficulties of working it out) is that reason one
version of which we found in the quotation from Aune. That is, it
appears that the foundationalist is committed to adopting beliefs
in the absence of any reasons for regarding them as acceptable. And
this would appear to be the sheerest dogmatism. It is the aversion to

18 Descartes apparently felt that he was required not only to identify his
foundations as such before building anything on them, but also to show at that
stage that each of the foundations had the required status. And not even itera-
tive foundationalism could help him with that. In the attempt to show that he
immediately knows that, e.g., 2 plus 2 equals 4, he is inevitably and notoriously
led to make use of premises the knowledge of which needs to be shown just as
much or as little as the proposition with which he begins.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 TWO TYPES OF FOUNDATIONALISM i83

dogmatism, to the apparent arbitrariness of putative foundations,

that leads many philosophers to embrace some form of coherence
or contextualist theory, in which no belief is deemed acceptable
unless backed by sound reasons.
The main burden of this paper is that with simple foundational-
ism one can have the best of both arguments; one can stop the re-
gress of justification without falling into dogmatism. We have
already seen that Aune's form of the dogmatism argument does not
touch Simple Foundationalism. For that form of the argument at-
tacks only the ungrounded acceptance of claims to knowledge or
justification; and simple foundationalism is not committed to the
immediate justification of any such higher-level claims. But one
may seek to apply the same argument to lower-level beliefs. Even
simple foundationalism, the critic may say, must allow that some be-
liefs may be accepted in the absence of any reasons for supposing
them to be true. And this is still arbitrary dogmatism. But the
simple foundationalist has an answer. His position does not require
anyone to accept any belief without having a reason for doing so.
Where a person is immediately justified in believing that p, he may
find adequate reasons for the higher-level belief that he is im-
mediately justified in believing that p. And if he has adequate
reasons for accepting this epistemic proposition, it surely is not
arbitrary of him to accept the proposition that p. What better rea-
son could he have for accepting it?
Lest the reader dismiss this answer as a contemptible piece of
sleight-of-hand, let me be more explicit about what is involved.
Though the simple foundationalist requires some immediately
justified beliefs in order to terminate the regress of justification, his
position permits him to recognize that all epistemic beliefs require
mediate justification. Therefore, for any belief that one is immedi-
ately justified in believing, one may find adequate reasons for ac-
cepting the proposition that one is so justified. The curse (of
dogmatism) is taken off immediate justification at the lower level,
just by virtue of the fact that propositions at the higher level are
acceptable only on the basis of reasons. A foundational belief, b,
is immediately justified just because some valid epistemic principle
lays down conditions for its being justified which do not include the
believer's having certain other justified beliefs. But the believer will
be justified in believing that he is immediately justified in holding
b only if he has reasons for regarding that principle as valid and
for regarding b as falling under that principle. And if he does

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
184 THE JOURNAL OF PHILOSOPHY

have such reasons he certainly cannot be accused of arbitrariness or

dogmatism in accepting b. The absence of reasons for b is "com-
pensated" for by the reasons for the correlated higher-level belief.
Or, better, the sense in which one can have reasons for accepting
an immediately justified belief is necessarily different from that
in which one can have reasons for accepting a mediately justified
belief. Reasons in the former case are necessarily "meta" in char-
acter; they have to do with reasons for regarding the belief as justi-
fied. Whereas in the latter case, though one may move up a level
and find reasons for the higher-level belief that the original belief
is mediately justified, it is also required that one have adequate
reasons for the lower-level belief itself.
We should guard against two possible misunderstandings of the
above argument. First, neither simple foundationalism nor any
other epistemology can guarantee that one will, or can, find ade-
quate reasons for a given epistemic proposition, or for any other
proposition. The point rather is that there is nothing in the posi-
tion that rules out the possibility that, for any immediately justified
belief that one has, one can find adequate reasons for the proposi-
tion that one is so justified. Second, we should not take the critic
to be denying the obvious point that people are often well advised,
in the press of everyday life, to adopt beliefs for which they do
not have adequate reasons. We should interpret him as requiring
only that an ideal epistemic subject will adopt beliefs only for
good and sufficient reason. Hence he insists that our epistemology
must make room for this possibility. And, as just pointed out,
Simple Foundationalism does so.
The dogmatism argument may be urged with respect to showing
that p, as well as with respect to accepting the proposition that p.
That is, the critic may argue that foundationalism is committed to
the view that "foundations cannot be argued for." Suppose that in
trying to show that p I adduce some grounds, and, the grounds
being challenged, I try to show that they are true, and ... in this
regress I finally arrive at some foundation f. Here, according to the
critic, the foundationalist must hold that the most I can (properly)
do is simply assert f, several times if necessary, and with increasing
volume. And again this is dogmatism. But again Simple Founda-
tionalism is committed to no such thing. It leaves something for the
arguer to do even here, viz., try to establish the higher-level proposi-
tion that he is immediately justified in believing that f. And, if he
succeeds in doing this, what more could we ask? Unless someone
demands that he go on to establish the grounds appealed to in that

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms
 BOOK REVIEWS 185

argument-to which again the simple foundationalist has no objec-

tion in principle. Of course, as we saw earlier, the demand that one
establish every ground in a demonstration is a self-defeating de-
mand. But the point is that the simple foundationalist need not,
any more than the coherence theorist, mark out certain points at
which the regress of showing must come to an end. He allows the
possibility of one's giving reasons for an assertion whenever it is
appropriate to do so, even if that assertion is of a foundation.
But, like many positions that give us the best of both worlds, this
one may be too good to be true. Although I am convinced that
simple foundationalism is the most defensible form of foundation-
alism, especially if it also divests itself of other gratuitous claims for
foundations, such as infallibility and incorrigibility,'9 I do not
claim that it can actually be made to work. Though it escapes the
main antecedent objection, it still faces all the difficulties involved
in finding enough immediately justified beliefs to ground all our
mediately justified beliefs. And on this rock I suspect it will founder.
Meanwhile, pending a final decision on that question, it is the
version on which both constructive and critical endeavors should
be concentrated.
WILLIAM P. ALSTON

Douglass College, Rutgers University

BOOK REVIEWS

Knowledge and Justification. JOHN L. POLLOCK. Princeton, N.J.: Uni-

versity Press, 1974. xii, 348 p. $15.00.
In spite of its title, this book is about justification and only very in-
directly about knowledge. Pollock accepts the view that knowledge
is justified true belief plus whatever is required to circumvent
the well-known counterexamples, but his book makes no attempt to
determine what is required. What the reader will find is an ex-
tensive, closely argued attempt to spell out when it is that our
beliefs are justified, not when it is that our justified beliefs consti-
tute knowledge. An extensive intellectual debt to Roderick Chisholm
is acknowledged, and it seems fair to say that Pollock has accom-
plished a very professional and original update of Chisholmian epi-
stemological analysis, an update which deals with such topics
as the reidentification of physical things and persons, memory,

19 For a position that approximates this, see Anthony Quinton, The Nature
of Things (London: Routledge & Kegan Paul, 1973), pt. II.

This content downloaded from

88.254.0.77 on Sun, 27 Oct 2024 10:45:11 UTC
All use subject to https://about.jstor.org/terms