3. EPISTEMOLOGICAL FUNCTIONS OF THE OSTENSIBLE PHYSICAL OBJECT

(a) The Sign Function

I believe that there are two such functions, one of them perhaps more strictly psychological than epistemological. The first of these, the psychological function traditionally attributed to sense-data in the knowing process, is that of somehow determining, at least in part, the nature of the physical object which the perceiver in some sense "believes" to exist at the moment of perception. It is impossible to describe this function more precisely except in terms of some particular form of the Sense-datum Theory. Thus Berkeley, who accepted what I have called the "discursiv inference version" of that theory, was willing to say that the sense-datum given in perceptual experience serves as a "sign" of the existence of a certain kink of physical object (for Berkeley, of course, other "ideas"). Other philosophers who have accepted the Discursive Inference Theory have said that the sense-datum "causes" us to "make a judgement about" a physical object. Such forms of expression are intended to imply that perceptual consciousness is discursive. Philosophers and psychologists who accept the more common Sensory Core Theory, however, have often found it more difficult to describe the relationship between a sense-datum and perceptual beliefs. Words like "sign", "clue", and "cause", do not seem appropriate for a sensory core which is conceived as occurring simultaneously with the perceptual beliefs. "

The best analogy we can offer", says Broad, ". . . is to be found in the case of reading a book. . . . If there were no print we should cognise no meaning, and if the print were different in certain specific ways we should cognise a different meaning".68 This analogy is not very good, however, for the print on the physical page transcends the entire state of perception and might therefore be said partially to "cause" the perception, whereas awareness of the sense-datum is regarded as a constituent of the state. and could not be said, in the same sense, to cause the beliefs which accompany it. Perhaps those who accept the Sensory Core Theory should restrict themselves, as Broad suggests, to the statement that the sense-datum and beliefs are functionally related in such a way that "if this sensum had different properties we should ascribe different properties to the physical ob ject".69

Or perhaps they might say, just a little more specifically, that sense-data and the accompanying beliefs are both caused simultaneously by certain physical events within and without the perceiving organism. Now the Percent Theory, as we have seen, does not admit the existence in ordinary perception, of a temporally distinct sensuous constituent; ostensiblt physical objects, therefore, cannot fulfil the function of a perceptual "sign" as conceived by Berkeley and other exponents of the Discursive Inference Theory. This is probably only a matter of academic interest to most contemporary epistemologists, however, for they have apparently rejected the Discursive Inference Theory in favour of the Sensory Core Theory.

But if it is true, as those who accept this latter theory maintain, that there is some functional relationship between the sensuous constituents of perceptual consciousness and the beliefs which accompany them, this relationship can quite consistently be recognised by supporters of the Percept Theory. In whatever sense it is true, in other words, that consciousness of a sensuously clothed ostensible physical object is accompanied by beliefs about the existence of a real physical object of a certain kind, in that sense it is meaningful to say that the beliefs are function ally related to the characteristics of the ostensible physical object.

And thus we could even say, directly paraphrasing Broad's statement about sensa, that perceptual beliefs are "based upon" ostensible physical objects in such a way that if a particular ostensible physical object had different properties we should ascribe different properties to the physical object. So far as the traditional psychological function of sense-data is concerned, therefore, the Percept Theory gives rise to no problems which are avoided by the Sensory Core Theory; and to carry the discussion beyond this point is unnecessary for the present purpose.

(b) The Function of the Ostensible Physical Object in Confirmation

The second. and more strictly epistemological, function traditionally attributed to sense-data, is that of serving as an important part of the evidence to which we must appeal in any attempt to justify our beliefs about the physical world. The statements which express these beliefs, according to one of the most familiar strains of traditional empiricism, can be divided into two groups: the first includes only singular statements about physical objects or events, and the second includes all other statements about the physical world.

Statements in the second group, according to this theory, can be justified only by an argument the premisses of which include statements in the first group. And statements in the first group can in turn be justified ultimately only by an argument the premisses of which include statements about sense-data. Thus sense-data have been regarded by many philosophers as the very foundation stones of empirical knowledge, without which a rational construction of physical science would not be even theoretically possible. For present purposes it is unnecessary to consider the points of disagreement within this well-known school of empirical thought.

There has been disagreement about the proper analysis of the statements which refer to sense-data, about the necessity for additional premisses, about the number and variety of principles of inference required to draw conclusions about physical objects, and about the epistemological status of such principles of inference; but these differences may be disregarded without prejudice to questions about the general function traditionally attributed to sense-data. I shall say, for convenience, that all members of this empirical school believe that sense-data are "epistemologicaily basic"; and the problem which immediately concerns us, therefore, is whether the Percept Theory has any implications for the theory that sense-data are epistemologically basic.

This problem is sufficiently important to deserve careful consideration even by the many philosophers who are convinced that beliefs about physical objects cannot be justified except in some relative fashion by reference to other beliefs about physical objects. Now among the singular statements which express our beliefs about physical objects, there are some which express perceptual beliefs, i. e., beliefs which are entertained at a time when we are presented with an ostensible physical object, and which are, as we have said, in some sense "based on" the ostensible physical object; such beliefs are usually expressed by statements similar in form to the statement "This (or that) is a tomato". Perceptual beliefs could probably be described dispositionally by reference to tendencies towards bodily behaviour or towards conscious experiences, or both; though perhaps, as we have already observed, some philosophers might wish also to make some reference to phenomenal events occurring simultaneously with the presentation of the ostensible physical object.

It is convenient, in considering the epistemological implications of the Percept Theory, to begin by paying special attention to those singular statements which express non-perceptual beliefs about physical objects but which are commonly supposed to be justifiable by reference to past perceptual experience. Thus we might consider, for examp]e, the statement "There is a tomato behind me", with the understanding that it expresses the present belief of someone who has recently been presented with an ostensible tomato but who is now looking in another direction. Such beliefs, of course, constitute a considerable proportion of all our beliefs about the physical world.

Most philosophers who accept the Sense-datum Theory and who also believe that sense-data are epistemologically basic, would probably maintain that in such a case the belief expressed by the statement "There is a tomato behind me", could be justified to some extent by means of an argument based on the present memory of past sense-data. Indeed they must defend a position of this sort if they are to bring their theory even roughly into line with common-sense, for the fact that someone has recently had a perceptual experience of a kind that he associates with the existence of a tomato, is commonly regarded as epistemologically relevant to his present non-perceptual belief in the existence of a tomato. But if the Percept Theory is true, this non-perceptual belief about the tomato could almost certainly not be justified in this way.

For assuming that the recent perception of the tomato had not been perceptually reduced for psychological or aesthetic purposes, it would simply be false to say that the observer had been aware of any sense-datum at all. The only sensuous constituent of that past perception would have been an ostensible tomato, which, as we have seen, is quite different from a sense-datum as traditionally conceived.

If the present belief is to be justified by reference to anything that has been sensuously given, therefore, it must be justified by reference to the ostensible tomato. And it is just this phenomenal thing, indeed, that the normal observer would remember under such circumstances; he would remember that he had been presented with a full-bodied tomato with all its sensuous qualities of redness, smoothness, warmth, and sweetness, and not that he had been aware of a round red patch, or perhaps, depending on the lighting, a patch of some quite different colour. The traditional analysis of justification by reference to sense-data is somewhat different, however, when the sense-data in question are conceived as future rather than past objects of direct awareness; and because of this difference it is not sufficient for those who support the Percept Theory to point out that sense-data do not occur in ordinary perceptual experience.

According to the philosophers who believe that sense-data are epistemologically basic, the statement "There is a tomato behind me" could be justified to some extent not only by reference to past experience, but also by turning around and inspecting the new sense-datum which is thereupon presented. And in a procedure of this kind it might indeed be possible, by adopting the reducing attitude, to produce a pure state (or approximately pure state) of direct awareness. And the sense-datum which is thus presented might possibly be used as the basis of an argument to justify the original non-perceptual belief about the tomato.

It would be presumptuous indeed, for those who support the Percept Theory to maintain that such a procedure is never followed in an attempt to justify a belief about a physical object. It would be quite unrealistic, on the other hand, to maintain that such a procedure is usually or even frequently followed. The procedure which is usually regarded as confirming a belief about a physical object involves various operations (e. g., manipulation of the object) but it does not involve the operation of introspective reduction. The usual procedure, as Price has aptly described it, is that of "specifying the unspecified". We look at the tomato, for example, from several points of view, turn it over in our hands, squeeze it, etc., and by these means produce a series of perceptual states. The ostensible physical object remains a tomato throughout the entire process, but the tomato becomes progressively more determinate, more specified, in each succeeding perception.

And as the relatively unspecified tomato becomes more and more specified, so we at the same time become more and more convinced that our belief has been justified. There is obviously no need to refer to sense-data in describing this process, and if we were to formalise this common method of confirmation we should have to treat the series of ostensible physical objects, and not sense-data, as epistemologically basic. There is also another fact which shows that if the Percept Theory is true, sense-data cannot be regarded as epistemologically basic without doing violence to commonsense.

This is the fact that we are so often surprised, after we adopt the reducing attitude for psychological or aesthetic purposes, at the characteristics of the sense-data which are thereupon presented to us. The fact of the matter is that most of us are simply not prepared, in many cases, to predict the characteristics of the sense-data which we can produce by perceptual reduction; we are not prepared to say, for example, what the colour of our sense-datum will turn out to be if we are looking at a field of green grass on a cloudy day, or in the late afternoon when it is lighted by the rays of the setting sun.

We are simply not familiar enough with the relationship between the physical stimulus and the conditions of observation on the one hand, and the sense-data which are the products of perceptual reduction on the other hand. Because of the psychological fact of object- constancy, however, the qualities of ostensible physical objects are more easily predicted; the ostensible grass, for example, is likely to be green whether the sky is clear or cloudy.

However the epistemological relationship between physical objects and ostensible physical objects may be conceived, therefore, it is apparently better understood on the practical level than the relationship between physical objects and the relatively rare objects of direct awareness. This in itself seems to be a good reason for putting greater faith in a process of confirmation in which ostensible physical objects, rather than sense-data, are treated as epistemologically basic.

If the philosophers who believe that sense-data are epistemologically basic are not interested in bringing their theory into line with common practice in the justification of belief, they might still insist, to be sure, that perceptual reduction is necessary for "proper" confirmation of beliefs about the physical world. But it is difficult to see what could be said in support of such a position, unless an appeal were made to some ontological theory about the constitution of physical objects, in particular to a theory that physical objects are in some more or less literal sense "composed of" sense-data.

Such theories, however, would lose whatever plausibility they may have if they were not themselves supported by epistemological considerations, and if the Percept Theory is true, indeed, these theories must themselves be revised if they are to retain the epistemological advantages traditionally attributed to them. To many philosophers this fact will undoubtedly seem to be one of the most important consequences of the Percept Theory, and we ought to consider it briefly before concluding our examination of the epistemological implications of the Percept Theory. Let us now do so.

4. THE PERCEPT THEORY AND EPISTEMOLOGICAL MONISM

Epistemological monists from Berkeley to many contemporary realists have used the theory that sense-data are epistemologically basic as a premiss in their attacks on epistemological dualism. Knowledge of physical objects is possible, they have said, only if statements about physical objects can be construed, in some more or less complex manner, as statements about sense-data -- or only, as some of them have preferred to say, if physical objects are somehow "composed of" sense-data. Their analysis of physical statements has not usually been based entirely on epistemological considerations, but in most cases their analysis was at least suggested by these considerations. If the Percept Theory is true, however, the epistemological advantages traditionally attributed to monism can be retained only by reinterpreting statements about the physical world in terms of ostensible physical objects and events instead of sense-data.

For one effect of the Percept Theory, as we have seen, is to change the denotation of the term "epistemologically basic". In whatever sense, therefore, a philosopher wishes to maintain that physical objects are "composed of" things that are epistemologically basic, in that sense he must say, if he accepts the Percept Theory, that physical objects are "composed of" ostensible physical objects. I do not propose to evaluate the results of such a reinterpretation but only to indicate what form it must take, and in particular to point out that ostensible physical objects can vary in ways that sense-data cannot, so that a new problem is uncovered as soon as we attempt such a reinterpretation. To state this problem clearly let us say that according to the traditional forms of epistemological monism every physical object is a "family" of sense-data. 70

If we do not specify the ontological status of sense- data -- whether, for example, they can exist unsensed and whether they may be "hypothetical" -- this statement can be interpreted as expressing a point of agreement among epistemological monists whether they be called "idealists", "realists", or "phenomenalists". For a family of sense-data would simply be the class of all those sense-data, actual or possible, which would have to be mentioned in making a complete ontological analysis of a particular physical object.

Now according to the traditional Sense-datum Theory, as we have seen, the sense-data observable by any one sense are quite limited in their qualities; visual sense- data, for example, may vary only in shape and colour. If the shape and colour of a visual sense-datum remain unaltered throughout a given period of time, therefore, changes in the attitude of the observer during that period cannot be said to affect the sense-datum at all. In the case of ostensible physical objects, however, the case is quite different; the colour and shape of two ostensible tomatoes may be exactly similar although the ostensible tomatoes, because of changes in the attitude of the observer, are quite different in other respects. To a hungry man the tomato may be presented as warm and sweet and edible whereas to someone looking for a missile it may be presented as soft and juicy and just about as heavy as a baseball.

Thus it is clear that a family of ostensible physical objects is even more numerous than a family of sense-data, and that the relations among its members are many times more complicated. As a matter of fact it might even be more appropriate to say that a physical object is nothing less than a nation of sense-data, the nation in its turn comprising as many famines as there are attitudes capable of affecting the content of perceptual experience. Thus the various perspective views of a tomato which are obtained by walking around it, could be said to be members of one family provided that they are all determined by the same attitude.

And by walking around the tomato a second time, but with a different attitude, the ob server could be said to become acquainted with members of a second family. And so on.

A philosopher who wished to maintain such a position, however, might not feel obliged to hold that members of all these families must be mentioned in an ideal translation of every statement about the tomato in question. He might insist, and with considerable plausibility, that what the hungry man means when he uses the word "tomato" is likely to be quite different from what the man in search of a missile means when he uses the same word, and that this difference in meaning must he reflected, in an adequate analysis, by the choice of families to be represented in the translation. Thus there might often be cases in which the meaning of the statement "This is a tomato", if used to express the limited belief of a particular observer, might be adequately translated into statements about ostensible physical objects which belong to very few families within the nation, or perhaps to only one. This is not the place to examine such possibilities more fully; it is sufficient for present purposes to point out that the Percept Theory creates a new problem for the epistemological monist, but that this new problem does not appear insoluble.

5. CONCLUSION

We may finally conclude, therefore, that there are at least four important epistemological implications of the Percept Theory: The traditional psychological distinction between the given and its meaning or interpretation, must usually be construed as a distinction between the ostensible physical object and certain accompanying events, either phenomenal or physiological or both.

There is one traditional meaning of "the given", however, for which there is no simple substitute in terms of the Percept Theory; for this particular meaning a more complex substitute may be provided by means of a "definition in use". The denotation of the term "epistemic" must be understood to be ostensible physical obiects rather than those things which have traditionally been called "sense data". As a result of this fact all the traditional forms of epistemological monism must be reinterpreted to make physical objects "nations" of ostensible physical objects rather than than "families" of sense-data.

On the other hand the Percept Theory has no implications concerning the general epistemological or ontological status of physical objects. It does not imply that some form of epistemological monism must be correct, nor even that there is some reasonable sense in which it would be true to say, in an epistemological context, that physical objects are "directly presented" in perception. Those philosophers are mistaken, therefore, who have inferred from the Percept Theory that the traditional problems concerning the epistemology of perception are pseudo-problems, or that they must be completely recast to make them fit the phenomenological facts on which the Percept Theory is based. The epistemological implications of the Percept Theory, we may conclude, are important but not revolutionary.

AUTHOR'S NOTE

Since this essay was written, the doctrine has become widespread among philosophers that it is a mistake (or, at the very least, misleading) to employ any form of "act-object" terminology to describe sense experience -- any terminology, that is to say, in which sense experience is represented as an act of perceiving, sensing, experiencing, having, or being aware of such "objects" as appearances, images, sensa, or sense-data; and with respect to this doctrine it is of course just as wrong to say that in perception we are aware of percepts or ostensible objects as to say that we are aware of sense-data. It is important to recog nize, therefore, that the phenomenological and epistemological issues discussed in this essay are entirely independent of the act-object terminology. They can be formulated in any terminology which allows us to describe a sensory constituent which may occur in hallucination as well as in "genuine" perception.

If we agree, for example, to use the idiom "It looks as if I am seeing ----" for this purpose, the phenomenological issues discussed in Part I, Section 3 of this essay can be construed as issues concerning the types of words which may properly be inserted in the blank. It may be true, when I gaze at a snow-capped mountain, that it looks as if I am seeing a triangular patch of white. Can it also be true in this same purely sensory use of "looks as if" that it looks as if I am seeing a mountain capped with snow?

Or, as so many philosophers have traditionally maintained, are we confusing the phenomenology of sense experience with an "interpretation" when we use words like "mountain" and "snow", and thus failing to describe our sense experience as it is really given? All the phenomenological and epistemological issues discussed in this essay may be formulated in some analogous way, using whatever idiom is approved by those who want to avoid an act-object terminology.

Notes
70 The term, of course, is the one used by Price for a collection of sense-data unified in a certain way. See Perception, p. 227.