There is a controversy in contemporary philosophy over the question whether or not knowledge must have a foundation. On one side are the foundationalists, who do accept the metaphor and find the foundation in sensory experience or the like. The coherentists, on the other side, reject the foundations metaphor and consider our body of knowledge a coherent whole floating free of any foundations. This controversy grew rapidly with the rise of idealism many years ago, and it is prominent today not only in epistemology proper but also in philosophy of science and even in ethics.

The discussion of this issue has been lamentably hampered by confusion and misunderstanding, and in this neither side is wholly innocent. The reflections that follow are not meant to settle the controversy, but only to help us understand it more clearly.

Foundationalism is often defended as the only acceptable alternative to an infinite regress of justification. It is now in fashion to reply that an infinite regress is not objectionable so long as we do not require that all members of the regress be actually held, actually justified beliefs, but only that each member be justifiable by reference to its successor. Actual justification will proceed only so far as the occasion demands, but it could always proceed further, if further doubts were pressed (as we are told by Wittgenstein, e.g., early in the Blue Book).

The theory of knowledge has two sides — epistemology and methodology — and a bridge to join them: that a belief is justified if and only if obtained by appropriate use of an adequate organon — a principle of theoretical epistemology requiring an organon or manual of practical methodology. Such organon justification is internalist. (How could one ever miss one's source for it?) But it leads briskly to skepticism on pain of regress or circularity — or so it is argued in Section 1. In Section 2 we consider the epistemology embodied in the Socratic elenchus, which provides a new angle on how methodology relates to epistemology and to science and metaphysics. Thus are we made to face once again the organon account of justification: our internalist bridge principle. That account proves in Section 3 to be a special case of a more general argumentative account of justification, which in turn agrees with intuitions so powerful as to be enshrined already in our dictionaries. Good rhetoric suggests therefore that “justification” and its cognates be yielded to the argumentative account; in which case justification must likely fall from its status as principal concept of epistemology. Justification is hence unlikely to be all that is required in general for knowledge, nor is it likely to be what is always required in any premises of use to justify anything.

INTELLECTUAL VIRTUES

Intellectual virtues might be viewed as ways of coping that are cognitively effective, a view however that would invite the question of just what might make a way of coping “cognitively effective.” According to my dictionary, “cognition” means “the act or process of knowing … also: a product of this act.” As for “effective,” it is said to mean “producing or capable of producing a result,” with an emphasis on “the actual production of or the power to produce an effect <effective thinking>.” Putting all this together, what makes a way of coping “cognitively effective” is its power to produce effects relating to or involving knowledge. But now look where that leaves us:

Thus we start with knowledge and return to it in a narrow circle.

For a more illuminating account we need to escape the circle.

Despair of knowing what knowledge is dates back to Plato's Theaetetus. Most recently, the trinitarian view of knowledge as justified true belief has been refuted, and a multitude of problems has appeared. Progress on this question is perhaps fated to be asymptotic. But such progress as can now be made depends, in my opinion, on a careful study of the conditions within which a correctly believed proposition is a bit of knowledge. In what follows I hope to enhance our knowledge of knowledge by contributing to such a study.

An accepted truth is knowledge only if evident. What then is it for something to be evident? One short answer is this: a proposition is evident to someone provided he is (theoretically) justified in believing it. But under what further circumstances is the truth of a proposition evident to someone? This is our first main question.

To begin with, there are two general situations where it is evident to someone S that p. First, there is the situation where it is self-evident to S that p, i.e., where from the fact that S correctly believes that p we may infer that it is evident to S that p. Our inference here cannot be logically valid as it stands, however, since logic alone will not enable us to infer that anything is evident just from the fact that it is correctly believed.

Polyfacetic epistemology would answer the skeptic, provide how-to-think manuals, explain how we know, and more. To some it is the project of assuring oneself, of validating one's knowledge or supposed knowledge, turning it into real and assured knowledge, thus defeating the skeptic. To others it is a set of rules or instructions, a guide to the perplexed, a manual for conducting the intellect. To others yet it is a meta-discipline, but one whose purpose is not nearly so much guidance as understanding, understanding of what gives us the knowledge we do have, of what factors serve to justify so many of our beliefs well enough to make them knowledge.

What follows is epistemology as understanding, an attempt to understand the relation between epistemic coherence and intellectual virtue at the foundation of epistemology: between the comprehensive coherence prized in the thirst for understanding and the “reliability” that makes a faculty or procedure intellectually virtuous.

SUPERVENIENCE

The central concept of epistemology is justification: not the practical justification of action, nor even such justification of belief as may come of its being expedient or generous or charitable; but rather the cognitive justification required to distinguish belief that is knowledge from what is little more than a lucky guess.

In the following pages five cases of abnormality are examined. It is argued that each reveals presuppositions that apparently underlie much of our commonplace empirical knowledge. And it is also argued that together they support a certain account of such knowledge, one that yields the following ideas: (a) that you know only when you believe correctly what is evident to you; (b) that for something to be evident to you, you must believe it reasonably, your belief must be reasonable; and (c) that the framework of beliefs, assumptions, experiences, or what not, if any, that supports your reasonable belief must not be flawed by a falsehood, i.e., must not involve as an essential component any false presupposition, assumption, presumption, claim, judgment, belief, or the like.

RECOGNITION AND CLASSIFICATION

The strange case of the masked burglar

It may well be thought that the framework (explicit or implicit) that underlies the recognitional judgment that here again is N before me is approximately this: (a) on past occasions when it has looked N-like, it has been due to the fact that it was N himself before me with what, in the circumstances, was his natural look; (b) here again something looks N-like in circumstances like those in the past when N has presented to me his natural N-like look; (c) if so, then here again it looks N-like because once again it is N himself before me, with his natural look in the circumstances.

Contemporary epistemology must choose between the solid security of the ancient foundationalist pyramid and the risky adventure of the new coherentist raft. Our main objective will be to understand, as deeply as we can, the nature of the controversy and the reasons for and against each of the two options. But first of all we take note of two underlying assumptions.

TWO ASSUMPTIONS

1. (A1) Not everything believed is known, but nothing can be known without being at least believed (or accepted, presumed, taken for granted, or the like) in some broad sense. What additional requirements must a belief fill in order to be knowledge? There are surely at least the following two: (a) it must be true, and (b) it must be justified (or warranted, reasonable, correct, or the like).

2. (A2) Let us assume, moreover, with respect to the second condition A1(b): first, that it involves a normative or evaluative property; and, second, that the relevant sort of justification is that which pertains to knowledge: epistemic (or theoretical) justification. Someone seriously ill may have two sorts of justification for believing he will recover: the practical justification that derives from the contribution such belief will make to his recovery and the theoretical justification provided by the lab results, the doctor's diagnosis and prognosis, and so on. Only the latter is relevant to the question whether he knows.

THE “ABSURDITY” OF FOUNDATIONS

Foundationalism has drawn fire repeatedly of late and stands multiply accused of itself resting on no better foundation than a “myth of the given,” and of requiring the evident absurdity of a “test” that compares our beliefs with reality itself in the absence of any intermediary concepts or beliefs. A recent salvo was fired by Donald Davidson in an important essay. Let us now examine the damage.

Davidson's “Coherence Theory of Truth and Knowledge” argues both against foundations and in favor of coherence. It attacks foundationalism for requiring a confrontation intrinsically absurd, and it objects to specific foundational “sources of justification” outside the scope of our beliefs. That is on the negative. On the affirmative, it presents and defends a coherentist alternative.

An allegedly foundationalist idea, that of “confrontation between what we believe and reality” is first argued to be “absurd,” thus opening the way for coherentism, subsequently offered as the alternative.

In explanation and support we are referred to Rorty, who claims that “nothing counts as justification unless by reference to what we already accept, and there is no way to get outside our beliefs and our language so as to find some test other than coherence.”

The traditional conception of knowledge as justified true belief has collapsed under weighty objections. Some of these are well known; but others, though equally weighty and puzzling, have attracted comparatively little attention.

It is sometimes objected that if we require that knowers always be able to ground what they know, provided it is not self-evident, then what ordinarily passes for knowledge would be little of the sort. I happen to know, for example, that Alamogordo is north of El Paso, but I doubt that I could establish it from memory: I cannot cite any authoritative map or encyclopaedia where I saw or read it, etc. In spite of this, however, it seems to me that even the traditional account can be defended here. For even though I may not remember what specific evidence I had in coming to know the relative position of the two towns, I can still remember that I once had such evidence, and perhaps what type of evidence it was, and that I have not since then found any contrary evidence. And this, I think, is enough to justify my continuing to believe. Indeed, even just the fact that one seems to remember may be enough for those of us with a good memory.

The method of reflective equilibrium aims to maximize two factors in one's beliefs: harmonious coherence, and plausibility of content. Analytic philosophy has long paid deference to these factors, for instance in its use of the counterexample, which attacks a principle as incoherent with the plausible (by one's lights). A critique of this tradition has recently appeared, and it shall be my main objective here to assess its merits. An appendix will apply our results to issues of moral relativism and rationality.

EPISTEMOLOGIES IN CONFLICT

Radical methodism is opposed to narrow reflective equilibrium. Karl Popper champions such methodism, Nelson Goodman such equilibrium. This opposition leads to a more moderate equilibrium, wide rather than narrow, in a sense to be explained.

Methodism

“Science is justified by means of induction on the basis of observation and experience.” This has been widely accepted and is widely accepted today. According to Popper, nevertheless, to think thus is a grave error, for as Hume showed, no supposed inductive reasoning is of any value. Popper reasons that “ … theories can never be inferred from observation statements, or rationally justified by them.” For “ … induction cannot be logically justified” (Popper, 42). That is to say, the fact that a theory T has been induced from certain data does not justify accepting T, since it does not ensure that T is true.

On Edmund Gettier's interpretation, the Ayer and Chisholm analyses of the concept of knowledge are sufficiently similar to analysis A, below, to be called the same.

Analysis A:

A person S has knowledge that p iff

1. (i) p is true;

2. (ii) S believes that p;

3. (iii) S is justified in believing that p.

Gettier presents us with two counter-examples to this view. I will now briefly set forth the principle of the second, which is both simpler than the first and not essentially different from it qua counter-example to A.

Suppose S has good evidence for his belief that p, from which in turn he deduces that p ∨ q. But, unknown to S, (∼p) & q. So, all three conditions for knowledge specified in the view under examination are fulfilled; but we still do not want to say that S knows that p ∨ q.

Here is a proposed analysis of the concept of knowledge, proposed as a solution to the Gettier problem:

1. sj1: There is a set of statements, e1, e2, …, en, each of which S believes to be true.

2. […]

Testimony is important both practically and intellectually. We rely on it for our grasp of history, geography, science, and more. We stake our time and fortune, and even our lives, on our beliefs. Which plane to board, what to eat or drink, what instrument readings to accept — all decided through testimony.

If we are largely justified in accepting testimony, how so? We might appeal to a principle like this:

1. (T) Testimony is correct more often than not.

But how to justify acceptance of T? There is so much testimony, past, present, and future! There are so many cultures, and cultures so diverse! How can one be sure about anything so strong as T?

Perhaps nothing so strong as T is needed; maybe it's enough to accept this:

1. (T′) From the sort of people I have dealt with in the sort of circumstances now present, testimony is normally correct.

Some have despaired of justifying any general claim about the correctness of testimony. H. H. Price, for example, prefers to postulate a policy of accepting testimony, in sharp contrast to any substantive belief in the likes of T or T′. Because policies need justification, moreover, for his testimonial policy Price offers the pragmatic justification that if we did not adopt it we would forfeit the rich supply of knowledge brought by testimony.

An intellectual virtue is a quality bound to help maximize one's surplus of truth over error; or so let us assume for now, though a more just conception may include as desiderata also generality, coherence, and explanatory power, unless the value of these is itself explained as derivative from the character of their contribution precisely to one's surplus of truth over error. This last is an issue I mention in order to lay it aside. Here we assume only a teleological conception of intellectual virtue, the relevant end being a proper relation to the truth, exact requirements of such propriety not here fully specified.

Whatever exactly the end may be, the virtue of a virtue derives not simply from leading us to it, perhaps accidentally, but from leading us to it reliably: e.g., “in a way bound to maximize one's surplus of truth over error.” Rationalist intuition and deduction are thus prime candidates, since they would always lead us aright. But it is not so clearly virtuous to admit no other faculties, seeing the narrow limits beyond which intuition and deduction will never lead us. What other faculties might one admit?

There are faculties of two broad sorts: those that lead to beliefs from beliefs already formed, and those that lead to beliefs but not from beliefs.

PHENOMENALISM

According to the main tradition, knowledge is either direct or indirect: direct when it intuits some perfectly obvious fact of introspection or a priori necessity; indirect when based on deductive proof stemming ultimately from intuited premises. Simple and compelling though it is, this Cartesian conception of knowledge must be surmounted to avoid skepticism. Seeing that the straight and narrow of deductive proof leads nowhere, C. I. Lewis wisely opts for a highroad of probabilistic inference. But how can one arrive at a realm inaccessible through direct knowledge having set out from one thus accessible? How could probabilistic inference offer any help? There are two different answers to these questions in Lewis's writings, and he moves from one to the other under pressure of well-known objections from perceptual relativity. Our action divides into three acts, which we now review in turn.

Act One: According to Lewis a statement about one's surroundings, such as P (that there is a sheet of paper before one), has analytic implications for one's experience, these being statements estimating the probability of outcomes for one's experience yielded by various action-cum-experience combinations. Thus our statement P is said by Lewis to have as one of its analytic consequences that if s1 (a visual sheet of paper presentation is given to one) and a1 (one seems to oneself to move one's eyes), then in all probability e1 (a seen displacement of the presentation follows for one).

Externalism and reliabilism go back at least to the writings of Frank Ramsey early in this century. The generic view has been developed in diverse ways by David Armstrong, Fred Dretske, Alvin Goldman, Robert Nozick, and Marshall Swain.

GENERIC RELIABILISM

Generic reliabilism might be put simply as follows:

That simple statement of the view is subject to three main problems: the generality problem, the new evil-demon problem, and the meta-incoherence problem (to give it a label). Let us consider these in turn.

The generality problem for such reliabilism is that of how to avoid processes which are too specific or too generic. Thus we must avoid a process with only one output ever, or one artificially selected so that if a belief were the output of such a process it would indeed be true; for every true belief is presumably the outcome of some such too-specific processes, so that if such processes are allowed, then every true belief would result from a reliable process and would be justified.