EMPIRICISM is the theory that experience rather than reason is the source of knowledge, and in this sense it is opposed to rationalism. This general thesis, however, can receive different emphases and refinements; hence, those philosophers who have been labeled empiricists are united only in their general tendency and may differ in various ways. The word "empiricism" is derived from the Greek empeiria (empeiria), the Latin translation of which is experientia, from which in turn we derive the word "experience." Aristotle conceived of experience as the as yet unorganized product of sense perception and memory; this is a common philosophical conception of the notion. Memory is required so that what is perceived may be retained in the mind. To say that we have learned something from experience is to say that we have come to know of it by the use of our senses. We have experience when we are sufficiently aware of what we have discovered in this way. There is another, perhaps connected, sense of the term "experience" in which sensations, feelings, etc., are experiences and in which to perceive something involves having sense experiences. These are experiences because awareness of them is something that happens to us. Indeed, the suggestion of passivity is common to uses of the word. To go into refinements here would not be relevant; one need only appreciate that the statement that experience is the source of knowledge means that knowledge depends ultimately on the use of the senses and on what is discovered through them. Sense experience may be nec-essary for the attainment of experience, but for present purposes that is unimportant.

The weakest form of empiricism is the doctrine that the senses do provide us with "knowledge" in some sense of the word. This could be denied only by one who had so elevated a conception of knowledge that the senses cannot attain to it. Plato, for example, held at one stage that because of the changeability of the world of sense, sense knowledge lacks the certainty and infallibility that true knowledge must possess. Hence, knowledge cannot be derived from the senses, but only from some other kind of awareness of what he called Forms. The most that sense perception could do would be to remind us of this genuine knowledge. This conception of knowledge demands an infallibility that sense perception cannot provide. Normally, we do not demand such high standards of knowledge, nor do we succumb to this kind of skepticism about sense perception. The commonsense view is that the senses do provide us with knowledge of some sort, and most people, when philosophizing, adopt this kind of empiricist view.

This weak form of empiricism can be generalized into the thesis that all knowledge comes from experience. The extreme form of this thesis would be the claim that no source other than experience provides knowledge at all. But this formulation is ambiguous, because there could be various reasons why all that we know might be dependent in some way upon experience. One reason might be that every proposition that we know is either a direct report on experience or a report whose truth is inferred from experience. A prima facie exception to such a thesis is provided by the propositions of mathematics; they have usually been thought to be a priori, not a posteriori-that is, we can know their truth independently of experience. There have, however, been philosophers who have denied the a priori nature of mathematical propositions. J. S. Mill, for example, maintained that the propositions of mathematics are merely very highly confirmed generalizations from experience and, consequently, all propositions are either reports on experience or generalizations from experience. This view has not been widely accepted.

A second reason for maintaining that all knowledge is dependent on experience would be that we can have no ideas or concepts which are not derived from experience, that is, that all concepts are a posteriori, whether or not the truths which can be asserted by means of these concepts are themselves a posteriori. It may be that we know some propositions without having to resort immediately to expe-rience for their validation; for their truth may depend solely on the logical relations between the ideas involved. Yet these ideas may themselves be derived from experience. If all our ideas are so derived, then knowledge of any sort must be dependent on sense experience in some way. According to this thesis, not all knowledge is derived immediately from experience, but all knowledge is dependent on experience at least in the sense that all the materials for knowledge are ultimately derived from experience. St. Thomas Aquinas was an empiricist in this sense. He thought that all our concepts are derived from experience, in that there is "nothing in the intellect which was not previously in the senses" (a doctrine supposedly derived from Aristotle). He did not think, however, that all knowl-edge either consists of sense experience or is inferred inductively from experience. Similarly, Locke held and tried to show that all our ideas are derived from experience, either directly or by way of reflection on ideas of sense. He did not hold, however, that all knowledge was sense knowledge.

It is possible to argue an even more complex thesis. It may be held that while there are ideas which are not de-rived from experience-a priori ideas-and while there are a priori truths which may or may not involve a priori ideas, such ideas and truths only have application on the precondition that there is experience. That is to say that -- for human beings at any rate -- reason can function only by way of some kind of connection with experience; "pure" reason is impossible. This was, in effect, Kant's position, and although he did not call himself an empiricist simpliciter, he was certainly opposed to what he called dogmatic rationalism. He held that there is no place for forms of knowledge of reality which are derived from pure reason alone.

It is possible, then, to maintain a general empiricist thesis that all knowledge is derived from experience on the grounds either that (1) all that we know is directly concerned with sense experience or derived from it by strictly experiential means, that is, learning, association, or inductive inference; or (2) all that we know is dependent on sense experience in that all the materials for knowledge are directly derived from sense experience; or (3) all that we know is dependent on sense perception in that even though we can know some things a priori, this is only in a relative sense, since the having of experience is a general precondition for being said to have such knowledge. None of these theses demand any more than the ordinary con-ception of knowledge. They do not demand that the knowledge in question should possess absolute infallibility so that the possibility of error is logically excluded. For none of the theses in question is essentially designed to be an answer to skepticism.

Empiricism and skepticism. Some forms of rationalism, for example, the Platonic theory already referred to, are meant to be answers to skepticism. They presuppose that an adequate reply to philosophical skepticism can be given only by showing that reason can provide forms of knowl-edge where error is logically excluded. The search for certainty, so intimately associated with seventeenth-century rationalism in general and Descartes in particular, aimed at showing that knowledge is possible because there are some things about which we cannot be wrong. Empiricism can be a rival to rationalism, not just in the sense already noted-that it may reject the supposition that reason by itself, without reference to sense perception, can provide knowledge -- but also in the sense that it proposes an alternate way of arriving at certainty. Empiricism, in this sense, is the thesis that the certainty required to answer the skeptic is to be found in the deliverances of the senses themselves and not in the deliverances of reason. Rationalism and empiricism, in this sense, are agreed that some such certainty must be found if skepticism is to be answered. They disagree about the sources of that cer-tainty and about the method by which the rest of what we ordinarily call knowledge is to be derived from the pri-mary certainties. Whereas rationalism seeks to derive knowledge in general from certain primary axioms (the truth of which is indubitable) by means of strictly deduc-tive procedures, empiricism seeks to build up or construct knowledge from certain basic elements which are, again, indubitable. The clearest expression of this point of view is probably to be found in twentieth-century empiricism, especially that associated with the logical positivist movement. This point of view is also found in the British empiricists of the seventeenth and eighteenth centuries, Locke, Berkeley, and Hume, but in their case it is over-laden with other elements and other forms of empiricism, some of which have already been noted. A short historical survey may serve to pinpoint the main issues.



It is often said that, in one sense, Aristotle was the founder of empiricism. Certainly Aquinas believed that he had Aristotle's authority for the view that there is nothing in the intellect which was not previously in the senses. It is not clear, however, that Aristotle ever raised this ques-tion. When he spoke of the relations between reason and the senses, he was concerned with issues in the philoso-phy of mind rather than with epistemology. Certainly Aristotle seems to have believed that knowledge is possible outside the immediate sphere of the senses and that reason can and does furnish us with necessary truths about the world. Aristotle's place in the development of empiricism, then, remains unclear.

Perhaps the first declared empiricist was Epicurus, who maintained that the senses are the only source of knowl-edge. Epicurus was an extreme atomist and held that sense perception comes about as a result of contact between the atoms of the soul and films of atoms issuing from the bod-ies around us. By this means phantasiae (appearances) are set up. These are all veridical. All sensations are true, and there is no standard other than sensation to which we may refer our judgments about the world. Sensations are set up in the soul by external stimuli, and for this reason Epicu-rus takes them to be "given." They constitute phantasiae when they occur in bulk. There is no further evidence that can be adduced in order that their veridicality may be assessed, either from other sensations or from reason. This is not to say that we cannot be in error concerning objects of perception; the films of atoms may become distorted in transit or the phantasiae caused by them may be fitted to the wrong prolepsis (conception). The last is a kind of abstract idea built up from successive sensations; the fitting of a phantasia to a prolepsis is what corresponds to judgment in Epicurus. It would appear that what Epicurus meant by his assertion that all sensations are true was that since they are caused in us, we can go no further in seeking information; they may not make us have true knowledge of objects, but in themselves they are incorri-gible. Precisely how all knowledge was to be built up from these sensations is not clear, and it has often been remarked that the axioms on which Epicurus' metaphysical system rests are far from the data of sense and are often based on more or less a priori arguments. Nevertheless, Epicurus' ideal of knowledge is one which not only depends on experience for its materials but is based on basic truths of experience.

A theory of knowledge similar in many ways to that of Epicurus may be found in St. Thomas Aquinas, although the main sources of Aquinas' philosophy are to be found in Aristotle. Aquinas was not a complete empiricist, for he did not think that all knowledge was derived from truths of experience. Knowledge of God, for example, could be obtained in other ways, and his existence could be proved by logical argument. Yet Aquinas did think that the ma-terials for knowledge must be derived from sense experience, and he gave an account of the mechanism by which this comes about. Roughly, when the sense organs are stimulated, there also results a change in the soul, which is the form of the body; this is a phantasm, a kind of sensory image. In order for sense perception to occur, the universal character of the phantasm must be seen as such. For this purpose, Aquinas resorted to Aristotle's distinction between an active and a passive reason. The active reason has to make possible the acquisition by the passive reason of the sensible form of the object of perception by a process which Aquinas-probably adapting an analogy used by Aristotle -- described as the illuminating of the phantasm. The active reason reveals the sensible form of the object by abstraction from the phantasm. This form is imposed upon the passive reason, which produces a species expressa, or verbal concept, which in turn is used in judgment. This process is called the conversio ad phantasmata; all concepts are arrived at in this way, by abstraction from phantasms. Hence, in applying them to entities which cannot be objects of perception, we must do so by means of analogies of various kinds with sensible objects. Aquinas' empiricism is, therefore, limited to concepts, and it is only in this limited sense that he held "there is nothing in the intellect which was not previously in the senses."



When thinking of empiricism, one tends to think, above all, of the British empiricists of the seventeenth and eighteenth centuries.

Locke. John Locke was an empiricist in roughly the same sense that Aquinas was, and he set the tone for his successors. His "new way of ideas," as it was called, had as its purpose "to inquire into the original, certainty, and extent of human knowledge, together with the grounds and degrees of belief, opinion, and assent." The reference to certainty makes it appear that he was concerned with skepticism or with skeptical arguments similar to Des-cartes' method of doubt. Locke's solution to this problem, however, was by no means consistently empiricist. His main target for attack was the doctrine of innate ideas, the doctrine that there may be ideas with which we are born or, at any rate, which we do not have to derive from sense experience. The first book of his Essay Concerning Human Understanding is devoted to a biting attack on this doctrine. In the rest of the book he sets out a positive account of the way in which ideas are built up, explaining that by "idea" he means that which the mind "is applied about whilst thinking." Ideas may be either of sensation or of reflection upon those of sensation; there is no other source. Ideas are also classified as simple or complex, the latter being built up out of the former. The mind has a certain freedom in this process, which may lead to error. (Locke later admitted ideas of relation and general ideas alongside the simple and complex.) The second book of the Essay is an exhaustive account of the way in which all objects of the mind are built up from ideas of sense. In this respect, then, Locke's philosophy may be considered an attempt to show in detail the truth of the kind of view which Aquinas had embraced, without accepting the same view of the mechanism whereby ideas come into being.

But Locke wanted to assess the certainty of our knowledge as well as its extent. The mind's freedom in forming complex ideas is a source of error, but in the case of simple ideas the mind, to Locke, was like a great mirror, capable of reflecting only what is set before it. Nevertheless, he did not maintain that all our ideas reflect the exact properties of things nor that all knowledge is of this character. In the fourth book of the Essay he asserts that all knowledge consists of "the perception of the connection of and agree-ment, or disagreement and repugnancy, of any of our ideas," but he goes on to distinguish three degrees of knowledge-intuitive, demonstrative, and sensitive. We can have intuitive knowledge of our own existence, demonstrative knowledge of God's existence, and sensitive knowledge of the existence of particular finite things. Intuition and demonstration bring certainty with them; they provide in effect a priori knowledge. The question of how there can be a priori knowledge of the existence of anything and how this can be a matter of the agreement or disagreement between ideas presents many problems.

These problems become acute in connection with sen-sitive knowledge. Locke tried to argue at one point that knowledge of the existence of particular finite things is a matter of the perception of the agreement of our ideas with that of existence. This will not do; to know that something exists is not to know merely that the idea of it fits in with the idea of existence. Hence, Locke admitted that this knowledge has not the certainty of the other two, although he insisted that it goes beyond mere probability and is commonly thought of as knowledge. He also tried to argue for the claim that we do have knowledge of sensible things, maintaining that simple ideas are caused in us in such a way that the mind is passive in receiving them. Moreover, the senses may cohere in their reports. None of these considerations really show that we do have knowledge of sensible things, and Locke admitted that they did not amount to proof.

Locke did not claim that all our ideas correspond to the properties of things. He felt this claim was true in the case of the so-called primary qualities, for example, bulk, figure, and motion, qualities without which, he maintained, a thing could not exist. It was not true of secondary qualities-for example, color and taste. In this case, the prop-erties of things cause us to have ideas that are not represent-ative of those things; the term "secondary quality" is thus a misnomer. Locke's denial of the real existence of sec-ondary qualities turns on his assimilation of our ideas of them to feelings like pain. (His acceptance of primary qualities was probably influenced by the success of physics in his time and its preoccupation with these properties of things.) As for things themselves, Locke maintained that we have little or no knowledge of their real essence, only of their nominal essence-their nature as determined by the way in which we classify them. This is due to the weakness of our senses. We cannot penetrate to the real essence of things, and our ideas of substances are mostly those of powers-the powers that things have to affect us and each other. It can be seen from all this that Locke was an empiricist in a very limited sense. In his view all the materials for knowledge are provided by sense perception, but the extent and certainty of sensible knowledge is limited, while on the other hand, there is nonempirical a priori knowledge of nonsensible things.

Berkeley. One aim of Berkeley, the second of the Brit-ish empiricists, was to rid Locke's philosophy of those elements which were inconsistent with empiricism, al-though Berkeley's main aim was to produce a metaphysical view which would show the glory of God. According to this view, there is nothing which our understanding cannot grasp, and our perceptions can be regarded as a kind of divine language by which God speaks to us; for God is the cause of our perceptions. The esse of sensible things is percipi -- they consist in being perceived and they have no existence without the mind. There exist, therefore, only sensations or ideas and spirits which are their cause. God is the cause of our sensations, and we ourselves can be the cause of ideas of the imagination.

Berkeley argued against those elements of Locke's phi-losophy which presupposed a physical reality lying behind our ideas. He attacked Locke's conception of substance and the distinction between primary and secondary qualities, pointing out that there was no distinction to be made between them in respect of their dependence on mind. He also attacked the doctrine of abstract ideas which Locke had held, the doctrine that we have general ideas of things abstracted from the conditions of their particular existence -- Locke's theory of universals. This Berkeley did because he believed that Locke's theory might provide a loophole for asserting the existence of an idea of sub-stance. The outcome of this was Berkeley's claim that there are no restrictions on the extent of our knowledge. We have knowledge of the existence of God and ourselves to the extent that we have notions of these spirits. We have knowledge of everything else, since the existence of everything else is a matter of its being perceived. There is nothing further beyond our ken. Even subjects like geom-etry, which might be supposed to involve knowledge of nonempirical matters, had to be limited in scope in order to rule out nonempirical objects of knowledge. Thus, Berkeley maintained that there is a least perceptible size; hence, there can be no ideas of infinitesimals or points.

In addition to claiming unrestricted scope for our knowledge, Berkeley asserted that knowledge is entirely dependent on sensations for all its materials other than the notions we have of God and ourselves. Berkeley claimed that this view "gives certainty to knowledge" and prevents skepticism. At the same time it defends common sense, he argued, because it does not involve the postulation of a reality behind ideas. His view gave certainty, he held, because sensations are by definition free from error; for error can arise only from the wrong use of ideas in judgment. The certainty of our sensations is due to the fact that there can be no question whether they actually represent a reality behind them; and this is the basis of Berkeley's claim to deal with skepticism. In general, all knowledge apart from that of our own existence and of God must, for Berkeley, ultimately be derived from sense perception. With these exceptions, therefore, Berkeley was an empiricist not only in respect of the scope and materials of knowledge but also in respect of its foundations. All truths must be founded on the truths of sense experience. The relations between ideas, which Locke had found a source of knowledge, were, for Berkeley, the result of the mind's own acts.

The mind operates upon the ideas given to it, comparing or contrasting them; it does not merely record what is there. Formal disciplines like mathematics, which might be thought to turn on the relations between ideas, thus depend on the ways in which the mind arbitrarily puts ideas together. Hence, to put the matter in terms more familiar today, mathematics is as much a matter of invention as discovery.

Hume. In respect to relations between ideas Hume perhaps went back to Locke, but in other respects much of Hume's philosophy may be represented as an attempt to rid empiricism of the remaining excrescences of nonempir-icist doctrine in Berkeley. As to the materials for knowledge, Hume tried to improve on his predecessors with attempts at greater precision. He distinguished first be-tween impressions and ideas, the former being the con-tents of the mind in perception, the latter those in imagi-nation, etc. He further subdivided ideas into those of sense and those of reflection, and again, into those which are simple and those which are complex. Like Berkeley, he denied the existence of anything behind impressions, and a cardinal point of his empiricism, to which he returned again and again, was that every simple idea is a copy of a corresponding impression. The understanding is therefore limited to these mental contents. Hume's main method in philosophy was what he called the "experimental method," the reference in all philosophical problems to the discoveries of experience. In effect, the conclusions which he drew from this are the opposite of Berkeley's. They can produce only skepticism. No justification can be given for belief in the existence of the self and an external world, for example. Reason cannot justify such beliefs, for all that we are given is a bundle of impressions and ideas. Only a psychological explanation can be given to account for our having such beliefs. Hume gives such an explanation in terms of the constancy and coherence of our impressions and ideas, and the principles of the associa-tion of ideas.

Hume's theory of knowledge is based on a distinction between two kinds of relations of ideas. In the Treatise of Human Nature he makes the distinction between relations that depend completely on the related ideas and those that can be changed without changing the ideas. The former, in effect, constitute necessary connections, the latter factual ones. In the later Enquiry Concerning Human Under-standing he short-circuited the discussion by distinguishing simply between relations of ideas and matters of fact. Mathematics depends entirely on relations of ideas and is thus concerned with necessary truths, the denial of which involves a contradiction. Matters of fact may rest simply on observation, but in the causal relation Hume finds the only case of a matter-of-fact relation which can take us from one idea to another. He shows that statements of causal connection cannot be logically necessary truths, in spite of the fact that we do attach some necessity to causal connections. After a long discussion he finds the explanation for this in the fact that causes precede their effects, are contiguous to them, and are such that there is a constant conjunction between them. As a result, the mind, through custom, tends to pass from one to the other. The feeling derived from this, which is an impression of reflection, constitutes the feeling of necessity which we find in the causal connection. Hume denied any real connection between cause and effect but tried to explain why we think that there is such. His demonstration that the causal connection is a contingent one is of the utmost importance, but his conclusions about it are skeptical. He held that there can be no real or objective justification for inference from cause to effect. He did allow, it is true, that certain rules can be provided which, when followed, will give some kind of probability to those inductive inferences which we actually do make. The aim of these rules is to make custom reliable and to avoid superstition. Hume has really no right, according to his own principles, to allow so much, and in doing so, he deserts skepticism in favor of a reductionist positivism, which seeks only to deny any necessary connection among things, while retaining belief in inductive inference. The concept of causal connection is thus in effect reduced to that of constant association of events contiguous in space and closely related in time. This is a position incompatible with his general skepti-cism. Apart from this, Hume's philosophy is of a piece. In Hume, then, extreme empiricism led to skepticism. Apart from relations of ideas, he held, the only knowledge we can have is of what we can directly observe, and any attempt to palliate this conclusion can produce only inconsistency.

In British empiricism, therefore, the gradual weeding out of anything inconsistent with empiricism, either in the form of the claim that the materials for knowledge must be derived from experience or in the form of the claim that knowledge cannot go beyond experience in its objects, resulted in skepticism about most of the things which we ordinarily claim to know. Kant proposed a reconciliation between this thesis and rationalism, maintaining that the rationalist claim of a priori knowledge about reality must be restricted to its application to experience. There is no room for a priori knowledge of anything which is not an object of experience. Pure reason can provide no real knowledge, despite the claims of rationalist metaphysi-cians. Such nonanalytic propositions as we do know a priori constitute principles that lay down the conditions to which experience must conform if it is to be objectively valid and not just a product of the imagination. A priori truths other than mere analytic truths have validity only in reference to experience; hence, while all knowledge is based on experience, it is not all derived from experience. This is scarcely empiricism in any recognized form, nor did Kant claim that it was; but it is a thesis that gives an important role to experience in knowledge.

One final point may he made about the British empiricists: They all employed a common method of trying to build up the body of knowledge from simple building blocks. The model for this method may have been the empirical science of the day. (Hume claimed to derive his experimental method from Newton.) The rationalists claimed more for reason and sought to reveal sources for knowledge and its materials other than experience; but they were also opposed to the empiricists in their choice of method, finding their inspiration in the method of axiomatic geometry.

John Stuart Mill. J. S. Mill, the main figure in nineteenth-century empiricism, followed directly in the tradition of Hume. Mill's account of our knowledge of the ex-ternal world, for example, was in part phenomenalist in character; it maintained that things are merely permanent possibilities of sensation. But it was mainly an account of the way in which we come to believe in such a thing as an external world and thus followed Hume in its psychological character. In one respect, however, Mill was more radical than Hume. He was so impressed by the possibili-ties of the use of induction that he found inductive infer-ence in places where we should not ordinarily expect to find it. In particular, he claimed that mathematical truths were merely very highly confirmed generalizations from experience; mathematical inference, generally conceived as deductive in nature, he set down as founded on induction. Thus, in Mill's philosophy there was no real place for knowledge based on relations of ideas. In his view logi-cal and mathematical necessity is psychological; we are merely unable to conceive any other possibilities than those which logical and mathematical propositions assert. This is perhaps the most extreme version of empiricism known, but it has not found many defenders.



Empiricists in the twentieth century have generally reverted to the radical distinction between necessary truths, as found in logic and mathematics, and empirical truths, as found elsewhere. Necessity is confined by them, however, to logic and mathematics, and all other truths are held to be merely contingent. Partly for this reason and partly because it has been held that the apparatus of mod-ern logic may be relevant to philosophical problems, twentieth-century empiricists have tended to call them-selves "Logical Empiricists" (at least those who have been connected in one way or another with logical positivism). On the other hand, Bertrand Russell, who has derived something from the positivists, but who owes equally much to the British empiricists, has always claimed that there are limits to empiricism, on the grounds that the principles of inductive inference cannot themselves be justified by reference to experience.

In general twentieth-century empiricists have been less interested in the question of the materials for knowledge than in that of the empirical basis for knowledge. Insofar as they have considered the former question, the tendency has been, as in other matters, to eschew psychological considerations and to raise the problem in connection with meaning. All descriptive symbols, it is maintained, should he definable in terms of other symbols, except that ultimately one must come to expressions which are definable ostensively only. That is, there must ultimately be terms which can be cashed by direct reference to experience and to it alone; ostensive definition consists of giving the term together with some direct act of pointing, such that no other understanding of meaning is required. In regard to nondescriptive terms the situation is less clear, but the general tendency is to assume that the only possible source of ideas which might be called a priori is logic and mathematics. Following Russell, twentieth-century em-piricists assumed that mathematical notions can be reduced to logical ones or can at least involve similar fea-tures and that logical notions are concerned only with relations between symbols and can be defined accordingly. Russell, it is true, has suggested that terms like "or" might also be defined ostensively, for example, by reference to feelings of hesitation, but this suggestion has not been generally accepted.

If the views on the question of the materials for knowledge are not clear-cut, there has not been the same indefiniteness over the basis of knowledge. Although some positivists, the so-called physicalists, have maintained that the language of physics should be taken as providing the basic truths, most philosophers of positivist persuasion have gone to direct experience for the truths on which knowledge is taken to rest. These truths are to be found in sense-datum propositions -- propositions which are a direct record of experience and which are for this reason incorrigible, consisting of ostensively definable terms, that is, names of sense data. It is not clear what would constitute an example of this. (Russell, for example, suggested "Red here now," where every expression is what he called a "logically proper name," such that its reference is guaran-teed.) Nevertheless, it has been assumed that all proposi-tions except logical ones must be reducible to these "basic propositions," which are about sense data. However, propositions about physical objects are not incorrigible. Yet to suppose that such propositions deal with entities which lie behind the immediate data of the senses and which can only be inferred from those data would be to suppose that there is a gap between us and physical objects, the crossing of which is problematical. This would allow an opening for the skeptic. An alternative view is phenomenalism, the doctrine that the meaning of our statements about physical objects can be analyzed in terms of propositions about sense data. Physical objects are logical constructions out of sense data ("logical" because the issue concerns the correct logical analysis of propositions about physical objects and not the question of how, as a matter of psychological fact, we construct our ideas of physical objects). In general, according to positivists, all propositions other than those which are logically necessary must be verifiable by reduction, either directly or indirectly, to propositions about sense data. Anything which is not so reducible is nonsense. In epistemological terms, any contingent truth which we can be said to know must be founded on and reducible to propositions concerning sense experience. Necessary truths, it is generally held, are true by conven-tion or in virtue of the meaning of the words involved. They tell us nothing about the world as such.

This program has run into difficulties of two main kinds. First, there have been difficulties in actually carrying out the analysis demanded. It would be almost universally agreed that propositions about physical objects cannot be analyzed in terms of propositions about actual and possible sense data, since the analysis would have to be infinitely long. This is an objection of principle. Second, the crite-rion of verifiability tends to exclude some kinds of propositions which we ordinarily think that we understand. There have been difficulties in this respect, for example, over propositions of natural law, as well as propositions of ethics, etc. There has been widespread dissatisfaction with attempts to justify empiricism of this sort.

It should now be possible to offer some assessment of empiricism. As an answer to skepticism it claims that the certainty and incorrigibility that knowledge demands can (apart from logical truths) be found only in immediate experience and that the rest of knowledge must be built upon this. In this sense, the theory is misguided as well as unsuccessful in carrying out its program. The lack of success can be seen in the fact that eighteenth-century em-piricism led to skepticism, while the twentieth-century program of reduction has been very widely admitted as a failure. The attempt was misguided in that knowledge does not require this kind of certainty and incorrigibility. Skepticism is not to be answered by providing absolutely certain truths, but by examining the grounds of skepticism itself. According to our ordinary conception of knowledge, what we claim to know must be true and based on the best of reasons. But by the best of reasons is not meant proof. Experience certainly provides justification for belief in, for example, physical objects, but if this belief is to amount to knowledge, it is not necessary that the justification should amount to proof. It is futile to argue whether experience or reason alone can provide proof of what we ordinarily claim to know. No one could have knowledge of the world un-less he had experiences and could reason, but this does not mean that either experience or reason by themselves could provide the kind of absolute certainty which would constitute proof. Nor is it required that they should provide proof in order that knowledge may be possible.

What of the thesis that, whether or not experience can provide certainty, all knowledge is derived from experience? In Mill's sense, that all truths, of whatever kind, receive their validation from experience, the thesis is obviously false and need be considered no further. The thesis that all the materials for knowledge are derived from experience may seem more plausible. Yet, despite the number of philosophers who have maintained this thesis, it is not altogether clear what it means. The version of the doctrine held by Locke and Aquinas looks like a psychological account of the origin of our ideas; in logical dress it amounts to the view that all our concepts or all the words which we use are definable in terms of those which are ostensively definable. Whether or not there are any a priori notions outside logic and mathematics, it certainly seems implausible to say that logical and mathematical notions may ultimately be definable ostensively. More important, the notion of ostensive definition is itself suspect. How could one understand what was going on when a noise was made, accompanied by a pointing to something, unless one knew the kind of thing which was being indicated and, more important perhaps, was aware that it was language that was being used? In other words, much has to be understood before this kind of definition can even begin. The notion that words can be cashed in terms of direct experi-ence without further presuppositions is, thus, highly suspect. This is not to say that there are no distinctions to be made between different kinds of concepts or words, but merely that the distinctions in question cannot be made by means of any simple distinction between empiricism and rationalism.

There remains the Kantian point that the having of ex-perience is a condition for any further knowledge. This would certainly be the case for creatures of our kind of sensibility, as Kant would put it. Yet the logical possibility of the possession of knowledge by nonsensitive creatures remains, whether or not any such creatures exist in fact.



Zeller, Eduard, Stoics, Epicureans and Sceptics, translated by 0. J. Reichel. London, 1892.

Bailey, Cyril, Epicurus, the Extant Remains. Oxford, 1926. In Greek, with English translation.

Bailey, Cyril, The Greek Atomists and Epicurus. Oxford, 1928.



Aquinas, St. Thomas, Summa Theologica, Ia, 78 ff., in Vol. IV of the English translation by the Fathers of the English Dominican Province. London, 1922.

Copleston, F. C., Aquinas. London, 1955.



Ayer, A. J., and Winch, Raymond, eds., British Empirical Phi-losophers. London, 1952. A collection of writings by the British empiricists.

Mill, J. S., System of Logic, 8th ed. London, 1872.

See also Immanuel Kant, Critique of Pure Reason, the transla-tion by Norman Kemp Smith. London, 1953.



Anderson, John, Studies in Empirical Philosophy. Sydney, Australia, 1962.

Ayer, A. J., Foundations of Empirical Knowledge. London, 1940.

Ayer, A. J., Language, Truth and Logic, 2d ed. London, 1946.

Ayer, A. J., Philosophical Essays. London, 1954.

Ayer, A. J., Problem of Knowledge. London, 1956.

Ayer, A. J., ed., Logical Positivism. Glencoe, III., 1959.

Lewis, C. I., Analysis of Knowledge and Valuation. La Salle, Ill., 1946.

Price, H. H., Thinking and Experience. London, 1953.

Russell, Bertrand, Inquiry into Meaning and Truth. London, 1940.

Russell, Bertrand, Human Knowledge. London, 1948.