During the past fifty or sixty years the traditional concept of sense-datum, which has been referred to, frequently because of epistemological or ontological considerations by many other names ("impression", "idea", "quale", "image", "sensum", "phenomenon", etc.) has been subjected to a type of phenomenological criticism which seems to threaten the foundations of a number of contemporary philosophical systems. Considering the fact that this criticism has been ably developed, and formulated by such distinguished men as William James, Edmund Husserl, John Dewey, and the leading psychologists of the Gestalt School, it is rather surprising to discover how much of the current literature on epistemological problems is entirely unaffected by it. Such lack of concern with vital phenomenological issues may be merely a reflexion of ignorance on the part of epistemologists, but it is probably better construed as a manifestation of the widespread belief that epistemological problems, if they are truly epistemological, not only can but should be stated and solved in abstraction from all issues which might be classified as "psychological". Although this lack of interest in the phenomenology of perception seems to be quite widespread among philosophers, there is a small but respectable group of epistemologists who have taken a very different stand.

They have maintained, in effect, that the traditional epistemological and ontological distinctions between sense-data and physical objects have been so completely annihilated by the criticism of James, Husserl, Dewey, the Gestalt Psychologists, and others, that most of the epistemology of the last three centuries is now entirely outdated. Some of them have asserted, as I shall show later, that it is no longer possible even to believe that there are any sense- data in the traditional meaning of the term; others have said, perhaps more conservatively, that although sense-data do indeed exist, it is no longer possible to distinguish their epistemological status from that of physical objects. Despite such important differences of opinion concerning the precise implications of the new phenomenology of perception, however, epistemologists who belong to this second school of thought are in complete agreement that these implications (whatever they may be) are of revolutionary importance for theory of knowledge. It may be presumptuous to attempt to reconcile two schools of thought which have existed side by side for so many years and which have so long resisted the various forces which might have been expected to increase mutual understanding and appreciation. But the attempt is surely worth the effort, and there are grounds for believing that the differences are to a large extent the result of terminological confusions. On the one hand the critics of the traditional concept of sense-datum have frequently expressed themselves in an esoteric vocabulary which is either quite misleading or quite incomprehensible to the epistemologists.

Many of the latter, on the other hand, firmly convinced that the traditional phenomenology of perception is completely adequate for the formulation and solution of philosophical problems, have not taken the trouble to seek for truth in statements of their critics which they correctly recognise to be either meaningless or absurd when interpreted in terms of the traditional vocabulary of epistemology. In view of the nature of these obstacles to mutual understanding, I shall undertake two tasks in this paper. I shall attempt in the first part to state as clearly as possible the phenomenological theory of perception which has served as a basis for most of the recent criticism of the traditional concept of sense-datum. I shall refer to this theory as the "Percept Theory of Perception" to distinguish it from the traditional "Sense-datum Theory", and I shall limit my description of it to what I take to be the bare essentials that distinguish it from the Sense-datum Theory. To overcome the linguistic obstacles I shall make an effort to describe the Percept Theory in terms of the concepts and vocabularies of contemporary epistemologists who do not accept it, and I shall similarly illustrate the theory, when possible, by examples drawn from the writings of these same epistemologists. I shall then attempt, in the second part of this paper, to evaluate the claims of some of the philosophers who believe that the Percept Theory is of revolutionary importance for epistemology. 1

1. THE NATURE OF SENSE-DATA

(a) The Definition of "Sense Datum"

To understand the position of those who reject the Sense-datum Theory in favour of the Percept Theory, it is necessary to consider, at least briefly, the manner in which the term "sense- datum" is customarily defined. It must be recognised, first of all, that in order to define this term philosophers have always found it necessary to refer to a certain kind of perception or awareness. Sometimes, for example, sense-data are defined as the objects of direct perception or of immediate perception. Thus at the outset of Berkeley's Three Dialogues, Philonous defines what he calls "sensible things" as "those only which are immediately perceived by sense"2 Broad defines sensa as objects of which we are "directly aware" in a perceptual situation".3 Price defines sense-data as those things "directly present to consciousness" in perception. 4 And Moore defines sense-data as the objects of "direct apprehension", citing as an example of such apprehension the having of an after-image. 5 If, however, a philosopher wishes to speak without contradiction of unsensed sense-data, he may define sense-data as entities which could be directly or immediately observed. And if he wants to distinguish between a sense-datum and sense-field, he may define sense-data as the distinguishable parts of whatever could be observed in this manner. But in any case he makes some reference to a particular kind of observation or awareness, which he usually describes as "direct" or "immediate".

This does not mean, of course, that sense-data cannot be defined without using the word "observation" or the word "awareness"; in fact some philosophers are content to define sense-data as entities which are  (or could be) sensed, or even as entities given to sense, and these definitions are merely verbal analyses of the term "sense-datum". The important point is simply that sense-data are defined not by an enumeration of their kinds but rather by reference to the manner in which we become conscious of them. We do not say that sense-data are patches of colour, rough things and smooth things, hot things and cold things, etc., for we could never be sure of exhausting the denotation of "sense- datum" in this way. Moreover, according to some theories, the surfaces of physical objects can likewise be described as "patches of colour", "rough", "smooth", etc., and the question whether or not some sense-data are surfaces of physical objects should not be prejudiced or confused by our definitions. Sense-data must be defined, therefore, by reference to the manner in which we become conscious of them: they are what we feel, sense, intuit, or immediately observe, or they are what is given to us, or what we are directly aware of, in perception. And once we understand the meaning of "sense-data" as so defined, we can presumably decide to some extent by empirical observation just what kinds of entities are properly called "sense- data".

(b) The Denotation of "Sense-Datum"

Nevertheless -- and here we come to a matter of the greatest importance in understanding and evaluating criticisms of the Sense-datum Theory -- philosophers have always found it impossible to explain the meaning of such terms as "direct awareness" and "immediate perception" without mentioning at least a few examples of the objects of such awareness or perception, namely, sense-data. This fact has been noticed and emphasised by Ayer and Moore. In The Foundations of Empirical Knowledge, Ayer points out that the terms "direct awareness" and "sense-datum" are correlative and that "since each of them is being used in a special, technical sense, it is not satisfactory merely to define one in terms of the other". "In order to show how one or the other of them is to be understood", therefore, it is necessary to use some other method of definition, "such as the method of giving examples".6

Moore makes the same point. That special sense of the word "see", he says, "which is the visual variety of what Berkeley called 'direct perception' . . . can only be explained by giving examples of cases where 'see' is used in that sense"7 It follows, therefore, that in order to understand what philosophers mean by the term "sense-datum", we must supplement our analysis of whatever explicit statements they may have made on the subject, by a careful examination of the examples which they have given.

Now such an examination of the examples which contemporary philosophers have given to illustrate the meaning of the words "sense-datum" and "direct awareness", will make it quite clear that all of them who are using these words in anything like the traditional way, are in agreement on two important points. They agree, in the first place, that the sense-data directly observable by any one sense are quite limited in their qualities. With respect to visual perception, for example, they agree with Berkeley that it is false to say that "we immediately perceive by sight anything beside light, and colours, and figures".8 Thus our sense-datum when we look at a dog, according to Russell, is "a canoid patch of colour".9 And when we look at a penny stamp, according to Broad, our sensum is "a red patch of approximately square shape''.10 And when we look at an apple, according to Lewis, what is given is a "round, ruddy . . . somewhat''.11 And when we look at a tomato, according to Price, our sense-datum is "a red patch of a round and somewhat bully shape".12

Thus it seems to be agreed by all these philosophers that when we gaze, for example, from a warm room at a distant, snow-capped mountain, our awareness of whiteness may properly be described as "direct", whereas our awareness of coldness may not. One of our sense-data is a white patch shaped like a mountain peak but our sensation of temperature, if we are aware of any at all, is one of warmth rather than coldness. In colloquial English, to be sure, we might say that the mountain "looks cold" or "appears cold", just as we might say that it "looks white" or "appears white", but such language is generally supposed to be unsatisfactory for theory of knowledge because it obscures the fact that the manner in which we are conscious of whiteness in such a case is very different from the manner in which we are conscious of coldness. The distinction in question is the very one that has traditionally been drawn by the use of such pairs of words as "impression" and "idea", "sensation" and "perception", "the given" and "the conceptual", "sense-datum" and "image", etc., and philosophers who use the term "direct awareness" in the traditional way must agree, therefore that the sense-data directly observable by any one sense are quite limited in their qualities.

In the second place, all philosophers who use the term "direct awareness" in the traditional way will agree on a still more important point, viz, that we are never directly aware of physical objects. It may seem, on first thought, that philosophers who accept the theory of perception called "direct realism", or some other more or less sophisticated variation of naive realism, are exceptions to this rule. Closer examination of their positions will probably show, however, that what these philosophers actually maintain is that some visual and tactual sense-data -- though not, of course, data of the other senses -- are literally the surfaces of physical objects. But these "surfaces" it should be noted, are not themselves physical objects: they are surfaces, and differ from physical objects in that they do not occupy a volume of space. And since these direct realists admit that it is only the surfaces of physical objects which we can perceive directly (i. e., that our sense-data are surfaces and not physical objects) we may conclude that their theory is not distinguished by any special propositions concerning the psychology of perception.

To emphasize the fact that physical objects are not accessible to direct observation, it has long been customary among philosophers and psychologists to reserve the verb "to perceive" for those cases in which the observation in question is not direct. According to this convention, which I shall adopt, the observing of physical objects is called "perceiving". Thus this second point of agreement among philosophers who use the correlative terms "sense-datum" and "direct awareness" in their traditional meanings, may be stated as follows: Physical objects are perceived but they are never the objects of direct awareness.

(c) Criticism of the Traditional Concept

Now in view of the necessity for defining the term "sense-datum" by the method of giving examples, it is clear that not only the truth but the very meaningfulness of the traditional Sense-datum Theory depends on the possibility of making the distinctions involved in these two points of agreement just formulated. Yet it is precisely these distinctions which have been denied by philosophers who accept the Percept Theory. They have sometimes developed their criticism in a rather haphazard manner, but I believe that their rejection of the Sense-datum Theory has always been based on objections to one or both of these two points of agreement.

The first objection consists in denying that there is any discoverable kind of observation or awareness which is present in every perception, and which takes as its objects only the kinds of things which have traditionally been offered as examples of sense-data. And this is not a trivial objection, for most advocates of the Percept Theory would go so far as to say that the experience of a man looking at a distant mountain from a warm room might comprise both whiteness and coldness, each in precisely the same manner, and neither in any other manner -- a statement which, as I have pointed out, has been either explicitly or implicitly denied by all philosophers who use the term "sense-datum" in its traditional meaning.

The second objection to the Sense-datum Theory is one which is not entailed by the first but which many psychologists and philosophers regard as an essential part of the Percept Theory. This objection consists in maintaining that in fact physical objects themselves are observed as directly as patches of colour, odours, tastes, and other so-called "sense-data". The direct and immediate experience of anyone who looks at the world about him, according to this interpretation of the Percept Theory, always consists of a number of full-bodied physical objects. And this, of course, is flatly to deny the distinction between perception and direct awareness which is essential to the Sense- datum Theory.

Now even the first of these two objections, if valid, is sufficient to necessitate a reformulation of most of the epistemological theories in the history of modern philosophy. Just how radical that reformulation would have to be, is a question which I shall discuss later. But the second objection to the Sense- datum Theory has implications which are even more serious, especially for those theories which maintain that physical objects are all, in some more or less literal sense, "composed of" sense-data (or of possible sense-data). Not only Berkeley and other subjectivists, but many more modern philosophers including Bergson, James, Russell, the new realists and many of the pragmatists and logical positivists, have supported the view that physical objects are knowable just because they are reducible to objects of direct awareness. But if sense-data are defined as the objects of direct awareness, and if, as some advocates of the Percept Theory have maintained, the objects of direct awareness may be physical objects, then physical objects are merely a subclass of sense-data. And the theory that physical objects are in some sense "composed of" sense-data is either false or tautological, of course, if it is understood that physical objects are themselves sense-data.

In recent years, moreover, the view that physical objects can be observed as directly as the entities which have traditionally been called "sense-data", has been used by a number of philosophers as a basis for criticizing one or more of these very epistemological positions. Wild, for example, has maintained in an article entitled "The Concept of the Given in Contemporary Philosophy", that what is actually given in perception is a "world of things". He quotes with approval a statement of Lewis that "it is indeed the thick experience of the world of things . . . which constitutes the datum for philosophical reflection", that "we do not see patches of colour, but trees and houses; we hear not indescribable sounds, but voices and violins". But then he goes on to criticize Lewis for abandoning this "classic view of the given" for the more restricted one of Berkeley and other modern empiricists. Modern empiricism, Wild asserts, "abandons the aim of classic philosophy to describe the thick experience of the world of things as it is given. Instead of this, it singles out a certain portion of the given as peculiarly accessible or given in some special sense". 13

Reichenbach, in his Experience and Prediction, has also declared that physical objects are immediately given in perception and has used this as an argument against positivistic theories of "reduction". Reichenbach's position, however, is much more extreme than Wild's. According to Wild, those things that are called "sense-data" by modern epistemologists are part of what is given; what he objects to is the view that "the immediately given alone is given". According to Reichenbach, however, such sense-data (what he calls "impressions") are not given at all. "What I observe", he says "are things, not impressions. I see tables, and houses, and thermometers, and trees, and men, and the sun, and many other things in the sphere of crude physical objects; but I have never seen my impressions of these things".14

Many statements of this kind have appeared in philosophical literature in recent years, and in most cases they appear to be based on the Percept Theory. The central thesis of this theory now seems to be accepted by most psychologists who are interested in the phenomenology of perception, although there are, as we shall see, differences of opinion concerning the implications of the theory. The central thesis was stated by William James in his Principles of Psychology as concisely, I believe, as it has ever been stated. A perception, he said, "is one state of mind or nothing"; if does not contain a sensation.

"We certainly ought not to say what is usually said by psychologists, and treat the perception as a sum of distinct psychic entities, the present sensation namely, plus a lot of images from the past, all 'integrated' together in a way impossible to describe. The perception is one state of mind or nothing."15

We may look at a physical object in such a way, James admitted, that what we apprehend approaches "sensational nudity"; thus by turning a painting upside down, or looking at it with a purely aesthetic attitude, "we lose much of its meaning, but, to compensate for the loss, we feel more freshly the value of the mere tints and shadings, and become aware of the lack of purely sensible harmony or balance that it may show".16 Nevertheless, the fact remains that sensations do not occur as constituents of perceptions, but at most only as complete and independent states of mind.

--------------------------------------------------------------------------------

Notes

2 Berkeley, Three Dialogues, in Works, Oxford, 1891, Vol. I, p. 381.

3 C. D. Broad, Scientific Thought, p. 239, Kegan Paul, London, 1923.

4 H. H. Price, Perception, p. 3, McBride, New York, 1933.

5 G. E. Moore, Philosophical Studies, p. 173 et seq., Harcourt Brace, New York, 1922, and "A Reply to My Critics", The Philosophy of G. E. Moore, p. 629, Northwestern U. Press, Chicago, 1942.

6 A. J. Ayer, The Foundations of Empirical Knowledge, p. 61, Macmillan, New York, 1940.

7 "A Reply to My Critics", in Philosophy of G. E. Moore, p. 628.

8 Principles, in Works, vol. I, p. 282.

9 Bertrand Russell, Inquiry into Meaning and Truth, p. 174, Norton, New York 1940. Quine has pointed out that Russell's word "canoid" means not "dog-shaped" but "basket-shaped". (Review of Russell's Inquiry into Meaning and Truth, p 10 Scientific Thought, p. 119

11 C. I. Lewis, Mind and the World-Order, p. 119, Scribners, New York, 1929.

12 Perception, p. 3. We may overlook for the moment the disagreement among these philosophers concerning the number of spatial dimensions possessed by visual sense-data.

13 Ja.

13 John Wild, Philosophy and Phenomenological Research, pp. 70-71, September, 1940.

14 Hans Reichenbach, Experience and Prediction, p. 164, U. of Chicago Press, Chicago, 1938. It is interesting to observe that in The Unity of Science, Kegan Paul, London,

1934, pp. 45-48, Carnap himself had questioned what Reichenbach calls "the positivist dogma" that impressions are given. There is a view, Carnap said, that "material things are elements of the given", and although "it is not often held to-day, it is . . . more plausible than it appears and deserves more detailed investigation".

2. THE SENSE-DATUM THEORY

This description of perceptual consciousness differs sharply, of course, from the traditional Sense-datum Theory, which is based on a supposed distinction between two constituents of perceptual consciousness. (1) direct awareness of a sense-datum and (2) mediated "perception" of a physical object. There are, however, two versions, of the Sense-datum Theory itself which must be distinguished in order to understand precisely what is asserted and denied by the Percept Theory: I shall refer to these two versions as the "Discursive Inference Theory" and the "Sensory Core Theory".

(a) The Discursive Inference Theory

The Discursive Inference Theory is most easily illustrated by turning to some of the great epistemological works of the British empirical school. In the Essay concerning Human Understanding, for example, Locke seems to maintain that perception is a discursive process which begins with awareness of a sense-datum and ends with the "idea" of a physical object. According to this analysis the perception of a physical object always involves a sensation and a subsequent act of judgement; every perception, therefore, includes awareness of a sense-datum as a temporally distinct act or state of consciousness. When we look at an alabaster globe, for example, the idea thereby imprinted on our mind is that of a flat circle. But knowing from experience that the cause of this appearance is a convex body, "judgment frames to itself the perception of a convex figure".17 Locke admits that the transiting from sense-datum to judgement "in many cases by a settled habit . . . is performed so constantly and so quick that we take that for the perception of our sensation which is an idea formed by our judgment; so that the one, viz. that of sensation, serves only to excite the other and is scarce taken notice of itself".18 But he does not doubt that both the sensation and the idea of judgment always occur when we perceive a physical object and that they always occur one after the other. Berkeley's analog as of perception in his New Theory of Vision is almost identical with Locke's. Perception is described as a process of discursive inference in which a sensation "suggests" a physical object to the observer. The mind no sooner perceives a sensation, Berkeley says, ". . . but it withal perceives the different idea of distance which was wont to be connected with that sensation". Thus, "having of a long time experienced certain ideas perceivable by touch . . . to have been connected with certain ideas of sight, I do, upon perceiving these ideas of sight, forthwith conclude what tangible ideas are like to follow".19 Berkeley recognises that there are times when "we find it difficult to discriminate between the immediate and mediate objects of sight. . . . They are, as it were, most closely twisted, blended, and incorporated together".20 But he does not seem to doubt that in every act of perception there are two successive events: the occurrence of a sense-datum and the occurrence of an idea which it suggests. Like Locke, in short, he maintains the Discursive Inference Theory, although frankly admitting that the successive components of perception may sometimes be hard to distinguish.

(b) The Sensory Core Theory

Almost all contemporary epistemologists who accept the Sense-datum Theory, however, have rejected the discursive inference version in favour of the Sensory Core Theory. Whereas Locke and Berkeley found it merely difficult to distinguish a temporally distinct state of direct awareness in every perception, most contemporary psychologists and epistemologists have found it quite impossible. In fact many of them have concluded that perceptual consciousness is never a discursive process involving a preliminary state of direct awareness.

An observer might report, to be sure, that on a certain occasion he was aware of a mere noise and then subsequently judged it to be an air-raid warning; but his report would probably be more accurate if he said that he first heard (in the sense of "perceived") a siren or "some sort of whistle" and then subsequently refined his judgement. The fact that a series of perceptions may become increasingly refined or determinate, in short, does not constitute proof of the existence of separate states of direct awareness. "If the content of perception is first given and then, in a later moment interpreted", says Lewis, "we have no consciousness of such a first state of intuition unqualified by thought, though we do observe alteration and extension of interpretation of a given content as a psychological temporal process".21 The many philosophers who support the Sensory Core Theory, therefore, do so because they believe that direct awareness of a sense-datum is a constituent of perceptual consciousness even though perceptual consciousness is not a discursive process. They believe that perceptual consciousness is a twofold state consisting of (1) direct awareness of a sense-datum and (2) an element of interpretation (variously described as "belief", "acceptance", "expectation", "judgment", etc.) and they believe that these two parts exist simultaneously. In perceiving an apple, for example, the sense-datum -- perhaps a round, red patch -- is one part of what is before our minds; the element of interpretation which distinguishes the perception of an apple from the perception of a tomato, is the other. The distinctive feature of this theory, in short, is that it regards awareness of a sense-datum as literallv a part of perceptual consciousness, but not as a part temporally distinct.

I have called this theory of perception the "Sensory Core Theory" because it asserts that there is, in some more or less literal sense, a core of sense-data in every perception. Psychologists of the Titchenerian School are sometimes said to have believed quite literally that sense-data form a core or nucleus within every perception, 22 but it is possible to accept the Sensory Core Theory, as I have defined it, without committing oneself to any such topographical analysis as Titchener's. Thus Price nowhere suggests that perceptual consciousness is strictly a nucleus of sensation surrounded by a fringe of images, but he does explicitly endorse the Sensory Core Theory. Perception involves no inference, he says, nor any discursive process whatsoever: "The two states of mind, the acquaintance with a sense datum and the perceptual consciousness [of the object] just arise together."23 Broad also accepts the Sensory Core Theory, for with certain important qualifications concerning the nature of perceptual belief, Broad is willing to say that "in a perceptual situation we are acquainted with an objective constituent which sensuously manifests certain qualities, and that this acquaintance gives rise to and is accompanied by a belief that the constituent is part of a larger spatio-temporal whole of a specific kind".24 Lewis has also endorsed the Sensory Core Theory by emphasing the fact that awareness of a sense-datum does not precede but accompanies the other constituent of perception. "Immediate awareness", he says, "is an element in knowledge rather than a state of mind occurring by itself or preceding conceptual interpretation"25

All these philosophers, and indeed the vast majority of contemporary epistemologists, believe that sense-data are distinguishable constituents of perception, and this, of course, is the view that is specifically rejected by James and other advocates of the Percept Theory. It must be clearly understood that both the Percept Theory and the Sensory Core Theory are theories about the nature of ordinary perceptual states -- in which we are in some sense "conscious" of physical objects. Neither of these theories implies anything whatsoever concerning the existence of pure states of direct awareness -- states in which we are directly aware of sense-data but not conscious of physical objects in the manner characteristic of ordinary perception. Contemporary philosophers seem to disagree about the frequency and even the possibility of such non-perceptual sensory states, but their opinions on this subject seem to be independent of their conclusions concerning the validity of the Percept Theory and the Sensory Core Theory. Lewis calls such states "states of pure esthesis", and doubts whether there are any. James says that "pure sensations", which he defines as the objects of direct acquaintance, "can only be realised in the earliest days of life. They are all but impossible to adults with memories and stores of association acquired."26 Price believes that it is possible on rare occasions only, "in a moment of intense intellectual preoccupation", to "pass over into the state of pure sensing, where there is not even the vaguest and most inattentive acceptance of anything material at all".27 Other philosophers, and many psychologists, however, seem to believe that pure states of sense-datum awareness are more easily obtainable, and have even said that they must be obtained for certain psychological and epistemological purposes. 28 Since the disagreements may be partly verbal, and since the issue is in any case not strictly relevant to an analysis of ordinary perceptual consciousness, I shall henceforth speak as though it were agreed that pure states of direct awareness are obtainable, but with the understanding, that "pure" may be interpreted to mean "approximately pure".

Notes
15 William James, The Principles of Psychology, Vol. II, p. 80, Holt, New York, 1896. Italics mine except for "plus".
16 Ibid., Vol. II, p. 81.
17 John Locke, An Essay Concerning Human Understanding, Vol. I, pp. 185-186, Oxford, 1894.
18 Ibid., p. 186.
19 New Theory of Vision, in Works, Vol. I, p. 148.
20 Ibid., p. 151.
21 Mind and the World-Order, p. 66.
22 "A typical perception", Titchener said, "resolves to begin with unto a number of sensations . . . -- the part that we may conveniently call its core or nucleus". Around this nucleus, is the context which carries the meaning, "the fringe of related processes that gathers about the central group of sensations or images" (A Beginner's Psychology, pp. 114, 118, Macmillan, New York, 1922.)
23 Perception, p. 151. Italics mine.
24 C. D. Broad, The Mind and Its Place in Nature, p. 153, Harcourt Brace, New York, 1929. Italics mine.
25 Mind and the World-Order, p. 276.
26 The Principles of Psychology, Vol. II, p. 7.
27 Perception, p. 165.
28 This point is discussed more fully in Part II of this paper..