Knowledge and Belief

views updated


The nature of knowledge has been a central problem in philosophy from the earliest times. One of Plato's most brilliant dialogues, the Theaetetus, is an attempt to arrive at a satisfactory definition of the concept, and Plato's dualistic ontologya real world of eternal Forms contrasted with a less real world of changing sensible particularsrests on epistemological foundations.

The problem of knowledge occupies an important place in most major philosophical systems. If philosophy is conceived as an ontological undertaking, as an endeavor to describe the ultimate nature of reality or to say what there really is, it requires a preliminary investigation of the scope and validity of knowledge. Only that can reasonably be said to exist which can be known to exist. If, on the other hand, philosophy is conceived as a critical inquiry, as a second-order discipline concerned with the claims of various concrete forms of intellectual activity, it must consider the extent to which these activities issue in knowledge.

In modern philosophy in the widest sense of the phrasethat is, philosophy since the Renaissancetheory of knowledge has usually been the primary field of philosophical investigation. René Descartes and John Locke, David Hume and Immanuel Kant, were all, in the first instance, epistemologists. Epistemological considerations played an important part in the work of Arthur Schopenhauer, but they were less central in G. W. F. Hegel and Friedrich Nietzsche, who were more occupied with the nature of the human mind in general and with the institutions within which it is exercised than with its more narrowly cognitive aspects. With Søren Kierkegaard and his existentialist descendants the focus of interest was man's will rather than his intellect. Anglo-Saxon philosophy, however, has remained epistemological. J. S. Mill, Bertrand Russell, and the analytic philosophers of the twentieth century continued to work in the area marked out by Locke and Hume. Even the British Hegelians of the late nineteenth century, the school of Thomas Hill Green and F. H. Bradley, were led into far-reaching epistemological studies by the character of the native tradition they were seeking to overthrow.

Belief has had less attention from philosophers. It has generally been taken to be a more or less unproblematic inner state, accessible to introspection. But there has been disagreement about whether it is active or passive, Descartes having contended that assent is a matter of will, Hume that it is an emotional condition in which one finds oneself. Alexander Bain urged that belief should be interpreted in terms of the tendencies to action with which it is associated, and Charles Sanders Peirce took the view that it is an unobstructed habit of action that, like health, comes to our notice only when we have lost it. Faith, especially religious faith, and probability, the logic of rational belief, have been thoroughly examined, but belief itself has received surprisingly cursory treatment.

The Definition of Knowledge

According to the most widely accepted definition, knowledge is justified true belief. That it is a kind of belief is supported by the fact that both knowledge and belief can have the same objects (thus, half an hour ago I believed I had left my raincoat in the garage; now I know that I have) and that what is true of someone who believes something to be the case is also true, among other things, of one who knows it. One who comes to know what he formerly believed does not lose the conviction he formerly had.

It is obvious and generally admitted that we can have knowledge only of what is true. If I admit that p is false, I must admit that I did not know it and that no one else did, although I may have thought and said so. It is urged, on the ground that beliefs that merely happen to be true cannot be regarded as knowledge, that knowledge must be justified. I may draw a true conclusion by invalid means from false premises or believe a truth on the strength of a dream or the misremembered testimony of a notorious liar. In such cases as these I do not really know the things I believe, although what I believe is true. There are, however, objections to all three parts of the definition of knowledge as justified true belief.


It has been suggested that the requirement that what is known be true is excessively stringent. Complete certainty of a statement's truth is not to be had; the best we can achieve is very strong grounds for thinking it true. Thus, if knowledge entails truth, we can never attain knowledge or, at any rate, never know that we have done so. This objection is misconceived. If I firmly believe that something is true on what I take to be sufficient grounds, I am right to say that I know it. It may be that the grounds are, in fact, insufficient and that what I claim to know is false. In that case my claim is mistaken, but it does not follow that I was wrong to make it in the sense that I had no justification for doing so.

It has also been argued, with a view to showing that knowledge and belief are quite distinct and unrelated, that whereas beliefs can be true or false, knowledge is neither. This argument exploits the fact that we speak of a belief but not of a knowledge, only of a piece or item of knowledge. Furthermore, since all items or pieces of knowledge are by definition true, we never need to speak of them as true items or pieces in order to distinguish them from false ones.


It is often objected that knowledge cannot be a kind of belief, even though they can have the same objects, because they exclude each other. If I know that p, it would be wrong for me to say that I believe it, since this would suggest that I do not know it. If, knowing p, I am asked "Do you believe that p ?," I should reply "No, 1 know it." This is hardly a serious argument. I should mislead people if I described my wife as the woman I live with, and I might say, "No, she's my wife," if I were asked whether she is the woman I live with. Nevertheless, my wife is the woman I live with. What is true is that I do not merely live with her. Likewise, if I know that p, I do not merely believe it, but I do believe it all the same. It is often wrong or misleading in certain circumstances to say something that is unquestionably true. The boy who, having taken two jam tarts, answers the question "How many have you had?" by saying "One" has told the truth but not the whole truth.

A more powerful argument against the definition of knowledge in terms of belief is that people can, it seems, know something to be the case and yet refuse, or be unable to bring themselves, to believe it. A woman told by wholly reliable witnesses with a wealth of circumstantial detail that her husband has been killed in an accident might be in this position. One way of getting around this objection is to say that she believes both that her husband is dead and that he is not. It is possible and not uncommon to believe something and its contradictory. It is not possible both to believe something and to not believe it at the same time, and what she will say is, "I don't believe it," although what she means is that she believes it is false. Another possibility is to say that although she has conclusive grounds for believing that her husband is dead, she does not, in fact, believe it and does not know it either. To have conclusive grounds is one thing; to recognize that they are conclusive is another.

It should be noted that where knowledge and belief overlap, the kind of knowledge involved is propositional knowledge, or what Gilbert Ryle called "knowing that." There is also "knowing how" (to skate, tie a reef knot, do long division), where there are no propositions to be true or false and where knowledge can vary in degree. The two kinds of knowledge are connected in that both are the outcome of learning. Belief is always propositional or believing that; there is no believing how that serves as a defective version of knowing how to do something.


We often express unreasonable hunches or intuitions by saying, "I know," and if they turn out, to our gratified amazement, to be correct, we rejoice by saying, "I knew it." Does this show that true belief can be knowledge even without justification? The emphasis we put on the verb when we use it in such a case suggests that it is an abnormal or marginal use. It is generally accepted that lucky guesses should not count as knowledge.

An important difficulty arises from the requirement that true belief must be justified if it is to be knowledge. What is it for a belief to be justified? One obvious answer is that my belief in q is justified if there is some other belief p that entails or supports it. It is clearly not enough that this further belief p should merely exist. It must also be a belief of mine; I must know it to be true, and I must know that it justifies q. But if this is a definition of justification, the original definition of knowledge is rendered circular and generates a regress. It has the consequence that before any belief can be justified, an infinite series of justifications must already have taken place.

How can such a regress be halted? A natural step is to ask whether all justification has to be of this propositional or inferential kind. As Russell has observed, we can define derivative knowledge in this way but must add an account of intuitive or uninferred knowledge. Philosophers have fastened on two forms of intuitive knowledge that, by standing as the uninferred first premises of all inference, can terminate the regress of justification. First, there are self-evident necessary truths, and, second, there are basic contingent statements, immediately justified by the experiences they report and not dependent on the support of any further statable items of knowledge.

In the first group are the axioms of logic and mathematics, such as the law of excluded middle and the principle of the commutativity of addition (a + b = b + a ), and statements that correspond to familiar verbal definitions, such as that kittens are young cats. Some philosophers hold that such intuitive, necessary truths record the results of intellectual intuition, the direct inspection of the relations of timeless universals; others, that their truth is essentially verbal in character, that one must accept them in order to be regarded as understanding the ordinary meaning of the words they contain. To accept an intuitive, necessary truth is to be ready to draw inferences in accordance with it. If I understand and accept the truth of "If (if p, then q ), then (if not-q, then not-p )," I must regard the deduction of "If he's not over twenty-one, he's not eligible" from "If he's eligible, he's over twenty-one" as valid. By applying such rules of inference to intuitive necessary premises, further demonstrative necessary truths are arrived at.

Intuitive contingent truths have been held to be those that describe the immediate objects of perceptual or introspective experiencefor example, "There is a green patch in the middle of my visual field" or "There appears to me to be a green flag here" and "I am in pain" or "I want to go to sleep." Basic statements like these are said to be incorrigible in the sense that they are wholly certified by the experiences they report and are logically immune from falsification by the results of any further experience. There may be no green flag here, but whatever may happen, there does now appear to be one. I may find it impossible to go to sleep once I get into bed, but I still want to go to sleep now. A statement is incorrigible if its truth follows from the fact that it is believed by the person to whom it refers. Thus, although I can make such a statement falsely, I must know that the statement is false when I do so. I cannot be honestly mistaken about my pains or the contents of my visual field.

It has sometimes been denied that there are any contingent, empirical statements that are basic and incorrigible in this sense. Coherence theories of knowledge have been propounded by the absolute idealists of the late nineteenth century and by C. S. Peirce, Karl R. Popper, and W. V. Quine in more empiricist forms in which beliefs are seen as justifying one another but none as in any sense self-justifying. To overcome the apparent circularity of the doctrine, it has been argued that some beliefs are relatively basic in that they can be accepted as true by some kind of convention or posited for the time being but that the element of dogmatism involved is only provisional and is open to revision.

plato's theaetetus

Several of the points raised concerning truth, belief, and justification were first made in the Theaetetus, that most modern in spirit of Plato's dialogues. In it three definitions of knowledge are examined, and in the end all are rejected. The three are that knowledge is (1) perception or sensation, (2) true belief, and (3) true belief meta logou, translated by John Burnet as "accompanied by a rational account of itself or ground." Against the view that knowledge is true belief Plato made the point that lawyers can persuade juries to accept beliefs that are, in fact, true by using rhetorical devices but cannot be said to provide them with knowledge by doing so. Against the third definition, which, in effect, takes knowledge to be justified true belief, he pointed out that it is circular and regressive.

There is an obvious objection to the definition of knowledge as perception. Perception itself must be defined in terms of knowledgenamely, as the acquisition of knowledge about the external world by means of the senses. Plato's meaning here is perhaps better rendered by understanding his first definition to equate knowledge and sensation. Certainly this makes more plausible Plato's identification of this definition with Protagoras's thesis that man is the measure of all things (or that the truth for each man is simply what appears to him to be the case). In fact, Protagoras's thesis would be more accurately interpreted as the view that knowledge and belief are one and the same. This contention has obviously contradictory implications, as Plato pointed out. We all believe some beliefs of others to be truer than our own, and most people believe that Protagoras's theory is false. Something like that theory persists, however, in the view, to which we shall later return, that the foundations of empirical knowledge consist of incorrigible statements about immediate experience. According to this view, what we believe about our current sensations or experiences, whatever we may choose to say about them, is true. If it is also correct that such sensations are self-intimating, in the sense that they cannot occur without our knowing them to occur, it follows that every sensation is an item of knowledge though not that every item of knowledge is a sensation.

In his discussion of knowledge as true belief Plato raised the problem of false belief. How can we believe falsely that X is Y since if the belief is false, there is no X that is Y to form a belief, true or false, about? A false belief, it seems, is no belief at all. A perhaps oversimple solution to the problem is that we can know a thing X well enough to be able to identify it as a subject of discourse without knowing everything about it (whether, for instance, it is Y or not-Y ). This draws attention to the point that the objects of knowledge are not always propositional, that not all knowledge is knowledge that. In addition to the knowledge how emphasized by Ryle, there is knowledge with a direct object, or knowledge of, claimed in such remarks as "I know Jones" or "I know Paris."

A claim to know a person can be intended and understood in two main ways. In saying that I know Jones, I may mean that I have met him and that I could not recognize him (and, usually, that we have had enough to do with one another for him to remember me). On the other hand, I may mean that I know what his character is like, what sort of things he is likely to do. According to the first interpretation, very little knowing that is involved, although I should be expected to be capable of giving some description of Jones's appearance; according to the second, some knowledge that relating to his character is implied, but none about his past history, health, occupation, and so on is.

A claim to know a place is ordinarily a claim to knowledge how, to an ability to find one's way about in it. It is not enough simply to have been there. Among other individual objects of knowledge are games, languages, and works of art. The last of these kinds of knowledge can be treated in much the same way as knowledge of persons; the others, as cases of knowing how, as claims to the possession of a skill. In general, knowledge of can be reduced to varying mixtures of knowing how and knowing that, though by no single recipe. It never involves a claim to knowledge that of all the facts involving the individual in question. A further point against Plato is that I can know enough about an individual or a thing to be able to refer significantly and successfully to him or it without being in a position to say that I know him or it simpliciter. I know enough about Samarqand to refer to it as a city in Uzbekistan and to ascribe to it a degree of beauty, historical interest, and size, but I do not know Samarqand at all, for I have never been there and could not find my way about in it.

is knowledge definable?

The English philosopher John Cook Wilson (18491915), closely followed in this by his disciple H. A. Prichard (18711947), strenuously maintained that the concept of knowledge is primitive and indefinable. Against such idealist logicians as F. H. Bradley and Bernard Bosanquet, they argued that judgment is not a genus of which knowledge, belief, and opinion are species. A judgment, said Wilson, is the conclusion of an inference, but some knowledge must be uninferred. Nor is knowledge a kind or species of thinking or a species of belief, for belief rests on knowledge in that it requires that there should be both some known evidence for it and the knowledge that this evidence is insufficient. No doubt, belief usually does rest on evidence or what is taken to be evidence, but it is not, as Wilson supposed, necessary that it should do so. I may believe a woman to be married because I take her to be wearing a wedding ring. The fact that it is not a wedding ring that she is wearing does not in the least imply that I do not really believe what I infer from my mistake.

According to Prichard, knowledge is completely sui generis and cannot, as he put it, "be explained." We cannot, he said, derive knowledge from what is not knowledge. This observation, if it is relevant at all, is simply a dogmatic assertion of the indefinability of knowledge. We can certainly define some things in terms of what they are not; for instance, not all cats are kittens, and not all young things are kittens, but a kitten is by definition a young cat. Knowledge and belief, Prichard held, are utterly distinct and cannot be mistaken for each other. We know directly and infallibly whether our state of mind is one of knowledge or belief. If so, knowledge and belief could not be related as genus and species, although they could still be different species of the same genus, another possibility that Prichard ruled out. His view that the two cannot be mistaken for each other seems clearly mistaken. We often claim with complete sincerity to know things that turn out to be false in the end. In so doing, we have taken a belief, mistakenly, to be knowledge.

Is the opposite possibility ever realized? Do we ever take to be mere belief something that, in fact, we really know? Is there a difference between knowing something and knowing that we know it? Benedict de Spinoza held that there is not. "He who has a true idea, knows at that same time that he has a true idea, nor can he doubt concerning the truth of the thing" (Ethics, Part 2, Proposition 43). As Spinoza expressed it, the doctrine is plainly false. I can perfectly well have very little confidence in a belief that is really true if, for example, it has been communicated to me by a notoriously unreliable informant. In other words, I can have a belief that is really true without knowing that it is true. But can I know that something is the case without knowing that I know it? I can certainly have a justified true belief without knowing that that is what it is, for I may not realize that the grounds I have for believing it really do justify it. The question deserves a more thorough investigation than it can be given here.

rationalist theory of knowledge

Plato's distinction between knowledge and belief has had a greater influence on the subsequent course of philosophy than his penetrating but unsuccessful attempts to find a definition of the concept. His essential point was that knowledge and belief are not only distinct attitudes but that they also have distinct and proprietary objects. Knowledge can be only of what is eternal and unchanging, of Forms, Ideas or universals; belief has for its objects the changing sensible particulars that make up the temporal world. Plato's reflections on mathematics seem to have led him to this conclusion. The propositions of geometry are preeminently objects of knowledge in that they can be established as conclusively true, once and for all, by demonstrative reasoning. Our beliefs about matters of temporal fact, on the other hand, are much more liable to illusion and error. The sensible objects of perceptual belief are infected with contradiction; they undergo change and have contrary properties at different times. But the objects of mathematical knowledge are wholly different. The circles and triangles studied by geometers are exact and perfect; they are ideals that the circular and triangular things we perceive with the senses approximate but always fall short of.

There are three ways in which a circular concrete thing may not be really circular. It may be circular at one time and elliptical at another; it may be other things (for example, green, cold, and sweet) as well as circular; and as concrete and sensible, it may not be strictly or perfectly circular. From these facts Plato concluded that such a thing is not wholly real in the way that the ideal circle of the geometer is. The ideal circle is a genuine object of knowledge, and only such wholly knowable things can be wholly real. From the distinction between knowledge and belief, then, Plato derived a distinction between two sorts of object, each sort constituting a separate world of its ownthe abstract world of eternal Forms, which is the knowable reality, and the concrete world of changing particulars, which is only appearance, not nonexistent but not wholly real either, and of which one can have not knowledge but only belief.

Plato's arguments for the unknowability and unreality of concrete, sensible things are not very persuasive. If this once circular mat is now elliptical, it does not follow that it was not really circular before. If this circular object is also green and cold, that does not in any way detract from its circularity. Finally, even if it is not perfectly circular, it may be quite definitely green. In general, there would seem to be many propositions that are known by some people but only believed by others; a mathematician will know the truth of a proposition he has proved, whereas another person will simply believe it on his authority. Some things I now know I used only to believefor instance, that I should be writing this here today; some things I now only believe I once used to knowfor instance, where I bought my raincoat. These considerations show that the objects of knowledge and belief are not wholly mutually exclusive. But it may still be true that there are some things that can be only believed, whereas others can be both believed and known.

At the center of Plato's thinking about this subject is a principle that defines one important sense of the word rationalism the principle that only necessary truths, established by a priori reasoning, can really be known. Something like this principle was accepted by Aristotle, although he rejected Plato's doctrine that Forms or universals occupied a separate abstract world of their own beyond time and space. Aristotle agreed that only the form of things could be known and that the matter that individuated or particularized them was beyond the reach of knowledge. For him true knowledge was to be attained by a process of intuitive induction that discerned the necessary connections between the forms present in concrete things. A science or ordered body of knowledge must consist of propositions deduced from self-evident first principles of this kind.

Descartes's rationalism was inspired by the reflection that ordinary claims to knowledge often prove mistaken. True knowledge, he insisted, must be objectively certain and impossible to doubt. His methodical endeavors to doubt everything were brought up short by the celebrated "I think, therefore I exist." I cannot doubt that I doubt, for in the act of doubting it I prove it to be true; if I doubt, I think; and if I think, I exist. What, he then inquired, is so special about cogito and sum ? What makes them so indubitably certain? His unhelpful conclusion is that they are clearly and distinctly perceived to be true. What he meant by this weakly formulated criterion of certainty can best be discovered by seeing what, in practice, he took it to certify. It appears that two sorts of proposition are clearly and distinctly perceived to be true: (1) necessary truths whose denial is self-evidently contradictory and (2) the immediate deliverances of sensation and introspection about one's own current mental state. Premises of both kinds figure in his first proof of God's existence:

Every event must have an adequate cause.
I have a clear and distinct idea of God.
God alone is an adequate cause for my idea of him.
Therefore, God exists.

In fact, cogito, I think, is not a clear instance of either, let alone both, of these two kinds of knowable, and even if it were, it would not follow from its being, on one hand, necessary and immediate and, on the other, certain that anything else that was necessary and immediate was also certain. Descartes's primary certainty was perhaps first thought of on a Thursday, but it does not follow that anything first thought of on a Thursday either by him or by anyone else is certain, too. It is not a necessary truth that I think or exist, for I might not be awake and might never have existed. If this is the case, the facts in question could not, of course, have been expressed in the first person singular.

Locke, despite his justly recognized position as a founding father of empiricism, reached much the same rationalist conclusion as Descartes, although by a very different route. He defined knowledge as "the perception of the agreement or disagreement of two ideas" (Essay concerning Human Understanding, Book 4, Ch. 1, Sec. 2). He went on to distinguish three kinds of knowledge: (1) intuitive knowledge of such things as the fact that red is not green and the fact of one's own existence; (2) demonstrative knowledge, which includes mathematics, morality, and the existence of God; and (3) sensitive knowledge, which is concerned with "the particular existence of finite beings without us." The third type of knowledge does not conform to his general definition, as he admitted. To become aware of a finite being outside us, we have to infer the existence of something that is not an idea from the ideas of sensation we take it to cause, and in part, to resemble. Locke's definition, as he understood it, restricts knowledge to the domain of a priori necessary truths. In intuition and demonstration there is a direct or indirect awareness of the connection between ideas present to the mind. But in the third case a connection is asserted between an idea of sensation and a physical thing that is not and cannot be directly present to the mind.

Locke did not introduce a special category to accommodate our knowledge of the ideas we passively experience but remitted them in passing to the category of intuitive knowledge. This sort of knowledge is quite unlike his exemplary cases of intuition, being contingent and empirical where the exemplary cases are necessary and a priori, and he might well have introduced a special category of reflective knowledge to accommodate it. It would comprise assertions of the connection of particular ideas, whereas intuition and demonstration would cover the connections of abstract, general ideas. Thus, although Locke's official definition of knowledge confines its application to necessary truths, it could, with a little modification, have been extended to cover a person's awareness of the present contents of his mind. But it could not, by any contortions, have been made to cover sensitive knowledge of real existence, that empirical knowledge par excellence which it was Locke's avowed purpose to justify and explain.


The indestructible vitality of the rationalist theory that necessary truths alone or necessary truths and reports of immediate experience are really knowledge was proved by its wide acceptance among empirically minded philosophers of the twentieth centuryfor example, Russell, C. I. Lewis, and A. J. Ayer. In support of it a powerful battery of arguments was produced, designed to show that despite the subjective certainty we feel in many kinds of belief, they cannot count as knowledge because they are not objectively certain.

Russell contended that all the sources of what we ordinarily regard as common knowledge of fact are in some degree untrustworthy. Perception is tainted by illusions, hallucinations, and dreams. Memory is notoriously fallible. Testimony, which plays such a large part in building up the social fabric of belief, presupposes an inference to other minds that is inevitably shaky and conjectural. Induction never certifies its conclusions, imparting at best only a measure of probability to them. Even introspection, if it is held to convey information about the self as a continuing personality, goes beyond what is directly present to the mind. Only what is directly present to itcurrently occurring thoughts and feelingsis the object of certain, infallible, and indubitable belief.

Lewis generalized Russell's position by distinguishing expressive judgments that report current states of mind from all other empirical propositions on the ground that they alone are wholly nonpredictive and have no implications about future observable happenings by whose failure to occur they might be refuted. Ayer, at one time, went even further. He held that all contingent, empirical propositions whatsoever, including reports of immediate experience, are uncertain on the ground that every such proposition involves the application of a general predicative term to its subject and thus makes a comparison with previous and perhaps faultily remembered instances of the term's application.

This kind of fallibilism about empirical belief was doggedly resisted by G. E. Moore and, after him, by Ludwig Wittgenstein, J. L. Austin, and Norman Malcolm. Moore's main point was that the word certain is learned and thus acquires its meaning from such situations as that in which a man holds up his hand and makes the perceptual judgment "I know for certain that this is a hand." Some rather subtler arguments are sketched in his book Philosophical Papers. Their general upshot is that the rationalists and fallibilists have been working with an unconsidered and excessively stringent concept of certainty. They have simply taken it for granted that for a belief to be certain, it must be impossible to doubt it. Russell, for example, began his search for certain knowledge with the question "Is there any knowledge in the world which is so certain that no reasonable man could doubt it?"

There are at least four senses in which it may be held that a belief cannot be doubted. The first is psychological; a man cannot doubt a belief if he cannot, in fact, bring himself to suspend judgment about it. This kind of certainty will vary from person to person and is of no direct philosophical interest. The second sense is logical. Here "doubt" is taken to mean "suppose false" and "can" to mean "can without logical inconsistency." This yields the strict rationalist view, since only necessary truths cannot be supposed false without inconsistency. A third sense identifies certainty with incorrigibility. According to it, a belief cannot be doubted if its truth follows from the fact that it is believed. Anyone who doubts an incorrigible belief shows that he does not understand the words that express it. The favorite examples of incorrigible beliefs are reports of immediate experience, such as "I am in pain" or "It seems to me now that there is a table here." But the notion would also apply to the more elementary and intuitive kind of necessary truth, such as the law of contradiction. Finally, there is the concept of certainty that, say Moore and his adherents, we actually employ in common speech where it means what cannot reasonably be doubted or supposed false. That people make all sorts of mistakes is not, according to this view, a reason for doubting the truth of a particular proposition. What is required to justify doubt is that propositions just like this, made in circumstances just like these and resting on just this kind of evidence, have in the past turned out to be mistaken. In this sense of certainty many beliefs based on perception, memory, testimony, and induction are objectively certain and thus properly regarded as items of knowledge. This view has the merit of allowing that many propositions that are, in fact, necessary truths are or once were less than certain, and it does not require the theory that there are any incorrigible propositions to be accepted. A further point in its favor is that such surprising theses as the one that no factual belief is certain can surprise us and escape triviality only if they are taken in this sense.

some modern views

In the mid-twentieth century, philosophical discussions of knowledge were much concerned with three distinctions drawn by Russell, Ryle, and Austin that must be briefly mentioned.

Acquaintance and description

In Russell's early writings he drew a distinction between knowledge of things and knowledge of truths, between knowledge of and knowledge that, a distinction marked in French by the verbs connaître and savoir. Within each kind he also discerned a distinction between an immediate and a derived form. Immediate knowledge of truths is conveyed in intuitive statementsfor example, basic judgments of perception and the axioms of logic and mathematics; derivative knowledge of truths, in demonstrable necessary propositions and inferred empirical statements. Parallel to this on the side of knowledge of things is the distinction between knowledge by acquaintance and knowledge by description.

Acquaintance, as Russell defined it, is the converse of presentation; it is the direct and infallible apprehension of some sort of object. But objects of description, unlike those of acquaintance, can fail to exist. Russell held that we are acquainted with present and past particulars and also with universals. This doctrine has led to a good deal of confusion. Certainly we do know things, persons, and places by acquaintance, but to do so is generally to know that something is true of them and is at least to know how to recognize them. The words with which we refer to things we are not acquainted with can be defined or explained in terms of those connected with objects of acquaintance. But this produces understanding rather than knowledge, understanding of singular terms (whether what they purport to refer to exists or not) and of general terms (whether or not there is anything they apply to). Russell's principle of acquaintance ("Every proposition which we can understand must be composed wholly of constituents with which we are acquainted") is really a version of the empiricist theory of meaning. Asserted without qualification, it is highly unplausible. We are not acquainted with anything corresponding to the "if" that occurs in the verbal expression of a hypothetical proposition although we understand the word. In general, to become acquainted with things is to acquire some intuitive knowledge of truths in which they figure, particular objects of acquaintance being the subjects of such truths and universal objects of acquaintance their predicates. In other words, knowledge of things cannot be separated from and regarded as prior to knowledge of truths in the way Russell supposed.

Knowing how and knowing that

Ryle's distinction between knowing how and knowing that has already been mentioned. There is a parallel distinction between remembering how and remembering that (there is also memory of past events). Ryle is anxious to correct the intellectualist bias of theorists of knowledge and to draw attention to the dispositional nature of all kinds of knowledge and belief; we speak, after all, of the knowledge and beliefs of those who are fast asleep. He tends to suggest that knowing that is a special, verbal form of knowing how, that it consists in having learned how to answer certain questions and now being ready to answer them.

Performative and descriptive verbs

John Austin's work on performative utterances has interested many philosophers in that class of verbs that are used in the first person present to do things rather than to describe what is being done. Examples of such performative verbs are "promise," "swear," "take thee, X, to be my wedded wife," and "name this ship Y." A verb ϕ is performative if it follows that I ϕ from the fact that I say, "I ϕ." Austin appears to have thought, wrongly, that "know" is a verb of this kind and that its function is to guarantee or authorize the acceptance of the piece of information that followed it. It is true that to prefix "I know" to a statement of fact does not add much to its content. But p and "I know that p " are not equivalent, since the former may be true when the latter is false. Austin was right in denying that knowledge is a state of assurance stronger than the most assured belief, though it is not clear that anyone ever supposed that it was. But the correctness of this denial, although it entails that it is not some describable psychological feature of the knower's state of mind that differentiates knowledge from belief, does not entail that the difference is not at all describable and lies, rather, in some nondescriptive function that the word performs.

The Nature of Belief

Most philosophers who have in any way adverted to the nature of belief have assumed that belief is an inner state of mind, directly accessible to introspection and distinct from, though causally related to, the believer's behavior. In The Emotions and the Will (1859) the Scottish philosopher Alexander Bain proposed that belief should be defined in terms of behavior: "Belief has no meaning except in reference to our actions no mere conception that does not directly or indirectly implicate our voluntary exertions can ever amount to the state in question." In support of Bain's theory is the fact that not only can others check our claims to believe by considering whether we behave appropriately but we ourselves may also take the results of such a test to overrule claims to believe that we have sincerely made.

Careful statements of the opposing doctrines were given by H. H. Price and R. B. Braithwaite. Price's mentalist definition of belief equates it with entertainment of a proposition together with assent. To entertain a proposition is to understand and attend to its meaning; when it occurs by itself, it is neutral and uncommitted as regards the proposition's truth or falsehood. Price breaks assent down into a volitional and an emotional part. He describes the volitional element as a mental act of preferring a proposition to any incompatible alternatives that have occurred to one; the emotional element is a feeling of conviction or assurance and may vary in degree. Braithwaite identifies belief in a proposition with its entertainment together with a dispositional readiness to act as if it were true. "Being ready to act as if p were true" has at first sight a suggestion of circularity, for it seems to mean being ready to act as if one believed p. But this can be avoided. I act as if p were true if I act in a way that would satisfy my desires if p were in fact true.

Against both theories it should be said that "entertainment" is dispensable if the normal sense of "believe" is in question, for we attend consciously to the propositions we believe only at rare intervals. As regards Price, what is to be understood by an act of preferring as opposed to an emotion of preference? It looks very like the silent assertion of the proposition itself, an inner rehearsal of a piece of outward verbal behavior. Second, feelings of conviction do not always attend even the beliefs we consciously entertain. Unless our confident beliefs are actually challenged, our state would seem to be one of easy and unemotional taking for granted.

Against the view of Bain and Braithwaite it has been urged by Mill, Franz Brentano, and Russell that if a belief has behavioral effects different from mere entertainment, it must differ in its intrinsic mental character. This is a misunderstanding. For a behaviorist there is a difference in the dispositions of one who believes and of one who merely entertains a proposition. A more serious difficulty is presented by beliefs that have negligible practical consequences, such as those about remote historical or astronomical events. But even here there is a disposition to verbal behavior, and, again, a disposition can exist without being actualized. There is also the difficulty that my claims about what I believe become, according to this theory, inductive conjectures about what I should do if certain circumstances arose. One reply is that not all inductive conjectures are conjectural to that degree. I need not, for example, feel very hesitant about what would happen if this iron table were dropped on that china teapot. Braithwaite adds that his theory has the merit of making possible rather precise measurements of subjective probability or degree of belief. The numerical probability I attach to a belief can be regarded as the least favorable odds I should accept on its turning out to be correct. Thus, unless I accept an odds-on bet, I do not believe something more than I believe its denial.

There is an interesting and extreme opposition in the history of philosophy between Descartes, who held that assent is a matter of will that can be freely given or withheld, and Hume, who represented us as largely passive in belief, which he conceived as a feeling that we find ourselves with and must put up with whether we like it or not, much as we find ourselves equipped with desires and aversions. Descartes's activism is shown first in his proposal that the philosopher should undertake a course of methodical doubt, suspending judgment about all the beliefs he has hitherto taken for granted. It reaches its fullest development in his attempt to solve the theological problem of error or intellectual evil, to reconcile the fact, on which his whole philosophy depends, that many of our beliefs are false with the goodness of God. The solution he offered is that God has fitted us out with limited intellects, appropriate to our earthly needs, but in his own image, with unrestricted freedom of will. When we make mistakes it is because we have culpably given free assent to propositions beyond the effective reach of our limited intellects.

In Descartes's favor is the fact that we do assess beliefs as more or less reasonable, a practice whose theory is logic and methodology. And the ethics of belief has not always been confined to distinguishing logically reasonable beliefs from others. It has often been held that some beliefsin the existence of God, for exampleare morally obligatory, and some beliefs are often recommended as prudent or useful. Hume himself propounded rules for judging causes and effects whose acceptance, he maintained, will enable us to advance science and avoid superstition. On Hume's side is the fact that it seems no more possible to resolve to believe something one actually does not believe than it is to increase one's height or eradicate one's distaste for endives by a simple effort of will. What one can do is to fortify or undermine one's belief in a proposition indirectly by voluntarily concentrating one's attention on the evidence for or against it.

It is quite commonly said that belief must rest on evidence and sometimes, especially by those who hold knowledge to be indefinable, that it must rest on knowledge. It is certainly usual for belief to rest on something the believer regards as evidence, whether or not it is true and whether or not it lends any support to the belief in question. But a wildly dogmatic or superstitious belief, maintained in the teeth of all the evidence, is still a belief, however unreasonable it may be.


There is some point to the malicious definition of faith as firm belief in something for which there is no evidence, for faith does involve a measure of risk, a voluntary decision to repose more confidence in a proposition, person, or institution than the statable grounds for doing so would, if neutrally considered, justify. Locke defined faith as resting on authoritative testimony, "the assent to any proposition, not thus made out by the deductions of reason, but upon the credit of the proposer." This applies well enough to the religious faith of traditional Christianity, but it is too narrow to cover the general use of the concept. It is often said that science rests on faith in the uniformity and intelligibility of nature as much as religion does on an undemonstrable conviction that the world is under the direction of a wise and benevolent intelligence. Certainly, science would be wholly sterilized if men were not prepared to consider adventurous and unjustified hypotheses. But it is not obvious that these adventurous conjectures have to be believed by their propounders. The austere maxim of W. K. Clifford"It is wrong, everywhere and for anyone, to believe anything upon insufficient evidence"is not strictly incompatible with intellectual enterprise. Yet even Popper, who of all theorists of knowledge is most insistent on the conjectural and fallible nature of science, admits that "our guesses are guided by the unscientific, the metaphysical (though biologically explicable) faith in laws, in regularities which we can uncover."

See also A Priori and A Posteriori; Evans, Gareth; Kant, Immanuel; Knowledge, A Priori; Kripke, Saul; Meaning; Plantinga, Alvin; Propositions; Putnam, Hilary; Reference.


classic discussions

Bain, Alexander. The Emotions and the Will. London: J.W. Parker, 1859.

Descartes, René. Meditations.

Locke, John. Essay concerning Human Understanding. Book 4.

Plato. Republic. Books 57.

Plato. Theaetetus.

modern accounts

Armstrong, D. M. Belief, Truth, and Knowledge. Cambridge, U.K.: Cambridge University Press, 1973.

Audi, Robert. Epistemology: A Contemporary Introduction to the Theory of Knowledge. 2nd ed. New York: Routledge, 2003.

Austin, John. "Other Minds." In his Philosophical Papers. London: Oxford University Press, 1962.

Ayer, A. J. The Foundations of Empirical Knowledge. London: Macmillan, 1940.

Ayer, A. J. The Problem of Knowledge. London: Macmillan, 1956. Chs. 12.

BonJour, Lawrence. The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press, 1985.

Brewer, Bill. Perception and Reason. Oxford: Clarendon Press, 2002.

Chisholm, Roderick. The Foundations of Knowing. Minneapolis: University of Minnesota Press, 1982.

Chisholm, Roderick M. Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press, 1957. Part I.

Dretske, Fred. Seeing and Knowing. London: Routledge and Kegan Paul, 1969.

Foley, Richard. Working without a Net. New York: Oxford University Press, 1993.

Goldman, Alvin. Epistemology and Cognition. Cambridge, MA: Harvard University Press, 1986.

Harman, Gil. Thought. Princeton, NJ: Princeton University Press, 1973.

Hintikka, Jaakko. Knowledge and Belief. Ithaca, NY: Cornell University Press, 1962.

Klein, Peter. Certainty: A Refutation of Scepticism. Minneapolis: University of Minnesota Press, 1981.

Lehrer, Keith. Theory of Knowledge. Boulder, CO: Westview Press, 1990.

Lewis, C. I. Mind and the World-Order. New York: Scribners, 1929. Ch. 9.

Malcolm, Norman. Knowledge and Certainty. Englewood Cliffs, NJ: Prentice Hall, 1963.

Moore, G. E. Philosophical Papers. London: Allen and Unwin, 1959.

Plantinga, Alvin. Warrant and Proper Function. Oxford: Oxford University Press, 1993.

Popper, Karl R. Conjectures and Refutations. London: Routledge, 1963. See the introduction.

Prichard, H. A. Knowledge and Perception. Oxford: Clarendon Press, 1950.

Russell, Bertrand. Problems of Philosophy. London: Williams and Norgate, 1912. Chs. 5, 13.

Ryle, Gilbert. The Concept of Mind. London: Hutchinson, 1949. Ch. 2.

Sosa, Ernest. Knowledge in Perspective: Selected Essays in Epistemology. Cambridge, U.K.: Cambridge University Press, 1991.

Wilson, John Cook. Statement and Inference. Oxford: Clarendon Press, 1926. Part 1, Chs. 2, 4; Part II, Chs. 13.

Woozley, A. D. Theory of Knowledge: An Introduction. London: Hutchinson, 1949. Ch. 8.

Anthony Quinton (1967)

Bibliography updated by Benjamin Fiedor (2005)