In this paper I am going to discuss some aspects of the realist view of sense-perception. The term 'realism' in epistemology covers more than one standpoint, but I shall use it to descrive the claim that we sometimes at least perceive things as they really are, that our perception of them is direct and not via the intermediary of images or ideas. Things, or material objects, are understood to be entities which exist whether or not we are perceiving them. They are such that although our perception of them may be said to vary, it does not follow that they too must vary accordingly. Moreover, they must occupy space and be capable of being perceived by more than one sense and more than one observer. This view, which I suppose all but a minute number of people implicitly adhere to, has been considered inadequate by certain philosophers for various reasons. I shall be concerned mainly with some of the arguments which purport to show that things cannot properly be said to have the properties which we attribute to them, and that consequently we cannot justify the claim to know things as they really are, except in the impoverished sense of 'as they really are' equivalent to 'as they are to us'.
The reduction of descriptions of thing as being really hot, white, square etc., to descriptions involving an observer is often supported by an appeal to the variability of our perceptions. Thus Berkeley considered it wrong to attribute a definite property to an object if this property were revealed to us only under certain conditions of the object, the environment and the observer's sense-organs. Water which we should normally call warm feels quite cool if our hands are particularly hot, colours very according to the light in which they are viewed, and so on. The same water, moreover, may be hot to one person, cool to another. Doesn't this all go to show that heat and coolness are in the observer, not in the objects themselves? After all, whilst this involves no contradiction, it is inconstistent to ascribe contrary qualities to any object at any given time. Even if we do ascribe qualities to things, how are we to decide which of the many appearances the object presents is its real colour, say? surely any particular selection from among the numerous shades which the object takes on will be quite arbitrary? Thus a book may look blue in the sunlight and in a yellow light it may look black, and yet we would unhesitatingly say that the book's real colour was blue. But why should we give the appearance of the book as seen in daylight a privileged status, allowing it to be revelatory of the real colour of the book? The selection of sunlight as the standard condition for perception of the real colour of the book is not, however, arbitrary, for the important fact about sunlight as opposed to page 43other forms of sunlight is not that it is the commonest kind of light, but that in it we are capable of the maximum possible discrimination of shades. Thus, although both a black and a blue book appear as black under a yellow light, they can be readily distinguished in sunlight. Of course, if some other light were devised which enabled us to make even more discriminations than we can at present with the aid of sunlight, then on the principle we have just put forward, this new light would supersede that of the sun as standard.
If this principle is considered arbitrary, it should be mentioned that it conforms to an even more important standard, that of maximum predictability of further appearances of the object. For instance, if you saw the two books in a hellow light and could not distinguish them, you would be at a loss to predict how they would appear in sunlight, whereas if you saw them in sunlight, the one blue, the other black, you would, after a few experiments, know that both would appear black in yellow light. Using sunlight as the standard, the other appearances of the object are predictable from its appearance in it; using yellow light as the standard, you cannot predict which, if any, of two books appearing black will turn out to look blue in the sunlight.
And yet, when all this has been said, we may still be inclined to feel that the phrase 'real colour of X' has not been given the sort of application that leaves us satisfied. The reasons for this lingering dissatisfaction are valuable in showing just what sort of an argument we have been considering, so I propose now to examine them. It will have been noticed that the distinction we made between the real and the apparent colour of an object was a distinction within experience. That is, the same type of phenomena were considered in deciding which was to count as real, and the unreel (or apparent) as well as the real were elements of our perceptual experience. Now this fact is for some the strength and for some the weakness of the argument and the source of dissatisfaction with it. The strength of some, since by contrasting other knowable phenomens with those deemed real, content is given to the real-unreal contract. This is important, for if we are to give any significance to the term 'real', we must know what it would be for an appearance to fail in this respect. Its meaningfulness requires that it be opposed to 'unreal', and such an opposition is stipulated in the principles of maximum discrimination and maximum predictability. Those who see the fact of the contrast's being within experience as a weakness in the argument would probably maintain that although content has been given to the real-unreal distinction, in confining this distinction to context within experience, it fails to indicate in any way what the colour of an page 44object is when placed outside experience. What they want is a specification of 'the real colour of X' which prescinds from all situations involving an observer and which is consequently true of X when X is not being observed. We will consider later whether such a demand can be met; however, let us first look at the question of the real shape of a material object. (I am confining my attention to colour and shape,for these seem to be the most important features of objects in such a discussion as this.)
It has been argued — by Rudolf Ziedine — that our selection of the shape of an object as seen from directly above as the real shape is not at all arbitrary, but is essential if we are to apply even the simplest notions of Euclidean geometry in describing the world. For Euclidean geometry allows of only one basic method of application, — that is, from straight on. This is tied to the notion of congruence of measuring apparatus and object measured. Any other attempt to determine the shape of an object is in practice impossible. Another means of determining the real shape of, say, a penny, is by an appeal to tactile as well as visual data. Thus the fact that a penny always feels round supports our decision to select the round shape as the real shape. Other criteria, such as conditions where we are least likely to make a mistake, can also be invoked to show that the choice of the real shape is far from arbitrary.
However, there are further difficulties which the realist must face. For, it may be said, if you define the real shape of a penny as that which it presents when viewed from an angle of ninety degrees, then someone who looks at the penny frcm another angle and sees it as elliptical cannot, on your definition of 'real shape' be in touch with reality. This presents the phenomenalist with a glorious opportunity of advocating his way of describing the situation. 'We must be perceiving something,' he declares, 'but that something cannot be the penny, because it's round, and what we perceive is elliptical.' This something which we do undoubtedly perceive he calls a 'sense-datum', via which we do in this case perceive the penny. This sense-datum terminology he extends to all perceptual situations, for every perceptual experience contains something which can be made to vary independently of the physical object being perceived. If, for example, you are standing before a table, you will find something in your experience whose shape you can change merely by walking around the table. From here you experience something diamond-shaped, from there something trapezoidal, and so on. Presumably the table itself does not change, but something certainly does, namely the sense-datum. The claim that sense-data and not material objectspage 45are given in perception is, to say the least, paradoxical, for it looks as if all forms of perception have been assimilated to one with which they are normally contrested, namely the perception of mental images. Moreover, the status of these sense-data has bever been satisfactorily elucidated, and we are left with a number of notorious problems concerning them. Thus, how is my round sense-datum of the penny related to your elliptical sense-datum of it? Surely cannot be both in the same place. Perhaps one is where the surface of the penny is, as G.E. Moore at times suggested; but if so, where are the others? Somewhere in the space surrounding the penny? Perhaps sense-data are too ethereal to enjoy a special existence at all, but if we relegate them to a mental existence it seems as though we have lost the table altogether, and cannot claim to know anything about it. Do sense-data exist independently of us or are we their creators and the measure of their existence? The phenomenalist may attempt a reply to some of theee objections — for example, he may eay that he has not assimilated all forms of perception to the perception of mental images in a way that blurs the distinction we ordinarily make, for the same criteria for discriminating between (say) veridical and hallucinatory perception still hold, the latter being identifiable by the irregular sense-data which they present and by the fact that confirmation from other observers is lacking. However, he may contend that whatever the difficultly involved in elucidating the status of sense-data, their existence cannot be doubted. The claim that I am seeing something round at the present moment is such that no subsequent findings would lend me to retract it. The sense-data is to be understood as that in virtue of whose existence I am entitled to make such an incorrigible claim. To say that it is an actual material object that is round is to make a judgment whose truth is not quaranteed by the experience I am now having. For to say my perception is of a material object is to commit myself to saying that other people too can or could see it and that its future behaviour will be of such-and-such a kind. The falsity of the many claims that such a statement involves is quite consistent with my having my present perceptual experience, and, therefore, I am not entitled to claim to know the truth of a statement which, because it goes beyond the actual evidence at hand, is such that further evidence might show me to have been mistaken. This support, the demand that we confine our claims to the indubitable, rests, however, on the ambiguity of 'knowledge must be infallible' or 'he who knows cannot be mistaken'. There in no inconsistency in claiming to know one is perceiving a material object, even though the evidence for one's claim does not logically entail its truth, for although the truth of a claim to knowledge precludes the possibility that one is in fact mistaken, the unrealized possibility that one might have been mistaken does not oblige anyone to withdraw his claim to know.page 46
The introduction of sense-data can also be attacked by showing how the various models, those of the 'picture and the 'veil', for example, break down at crucial points. However, even if the phenomenalist's recommendation to adopt the sense-datum terminology is accepted, it soon becomes clear that this does not provide us with a satisfactory languuge for dealing with material-object statements. This can be shown by examining his assention that material objects are logical constructions out of sense-data, or better, that material-object statements can be translated into statements about sense-data, without remainder. Such a translation involved the difficulty of specifying time and place in phneomenalistic terms, and the fact that an endless number of sense-datum statements would be required, but these obstacles are perhaps only of secondary importance for it can be shown quite easily that the equivalence does not hold [unclear: betw] the two types of statement. In the first place, no number of sense-datum statements entails the existence of a physical object. This is so because a material-object statement takes a far [unclear: greater] risk of being falsified than any number of sense-datum statements. This is not in the least surprising, for one of the cnief reasons for introducing sense-datum talk was to avoid taking precisely this risk. Moreover, it could be argued that the other implication of the equivalence, that the existence of a material object entails the occurrence of certain sense-data, does not necessarily hold. Certainly it would be odd for it not to hold, but it is not clear to me that this would be a logically impossible state of affairs. At any rate, whether or not the phenomenalist is right on this point, the equivalence clearly fails for the reason first given.
Fortunately, however, we are not obliged to become phenomenalists, for, instead of using such descriptions as 'from here the penny presents an elliptical sense-datum' we need only say' from here the penny looks elliptical', and similarly the stick in water which allegedly presents a crooked sense-datum is better said simply to look or appear bent. In this way our perceptions can be described with reference only to material objects, and not to any mysterious tertium quids which somehow mediate between the world and our perception of it.
However, let us return to the question proposed just before the phenomenalist intruded to advocate his way of discussing perception. we were in the position of having to say that someone who sees the penny us elliptical is not seeing the penny as it really is. Now some may feel uncomfortable about this way of speaking, for on the proposed definitions of 'real colour' and 'real shape' it would turn out that they were seldom in a page 47position to perceive things as they really are. Consequently, they might reject our recommendations as to the use of 'real' on the grounds that the resultant relegation of the best part of our perceptions to the realm of the unreal is misleading. This is a reasonable reaction, for 'unreal' does suggest entities which exist only in our imagination, and we do not wish to ascribe that status to the appearances of objects viewed from other than straight on, or in light other than that of the sun. After all, this so-called 'real shape' is just the shape of the object as appears in one out of many like perceptual situations. In other words, the difference made here between appearance and reality is merely the difference one particular appearance of the object and the various other appearances from different points of view. But the former is no less an appearance than the latter. Your use of 'real', it is argued, just reflects a preference for certain aspects of the object as more important or convenient; it denotes nothing deeper, and, since 'unreal' has innappropriate connotations, why don't we banish this 'things-as-they-really-are' or 'real shape' and 'real colour' talk, thereby dispensing with the misleading dichotomy. Since the object presents just as much an appearance of a certain sort when viewed from face on as it does when viewed from anywhere else, let us translate 'the penny is really round' into 'when viewed from directly above, the penny looks round.' Then we shall have no reason to elevate this perceptual situation above any others, for here, as in these others, we will be talking about how the penny looks or appears when viewed from a certain position.
This move could be made by a realist, but he might not be too happy with it, for it turns out to be closer to phenomenalism than he would wish. For, if we dispense with talk about real features of objects and restate all descriptions in terms of appearance talk, then it is impossible to make any categorical assertions about unobserved things. That is, if the position of an observer and the state of his sense-organs must be specified — at least by implication — in descriptions of material objects, then categorical statements about material objects will be meaningless, for part of the context essential to the meaningfulness of such descriptions will have been omitted. We will be reduced to saying that such statements are actually hypotheticals masquerading as categoricals. For example, on this view 'this penny is round' becomes 'if you were to look at this penny from directly above, it would appear round.' This kind of translation may be found quite satisfactory by some, but I must confess that I am not at all happy with it. I still feel strongly inclined to say that the penny's actually being round implies that it will appear round when viewed page 48straight on, implies, in fact all the possible appearances of the penny, but not 'that the penny is round' means the same as the suggested translation. That something has gone wrong is, I think, indicated by the difficulties which the exclusive use of the language of appearing gives rise to. Thus if we take 'appears' as a term denoting a relation between observer and observed, it follows that when nobody is around, nobody is appeared to, and hence nothing appears. If 'appears' is not understood as involving a relation, we are still left with the problem of how to reconcile the simultaneous existence of a well-defined appearance of an object's shape and a blurred appearance of the same shape, a perplexity rather like that engendered by the problem of unsensed sense-data.
However, once more I do not think that we are obliged to adopt a troublesome way of speaking, not at least as regards shape. It will be recalled that exclusive use of the appearance language was advocated on the grounds that the inappropriate connotations of the word 'unreal' made for a misleading dichotomy. This objection can perhaps be remedied by a further analysis of what 'unreal' means in this context. In fact, if we simply replace it by the word 'apparent' much of the uneasiness vanishes. However, I suspect that this may be the sort of problem which admits of an appeal to ordinary usage for its solution. Let me try and make this clearer. If someone asks what is the real shape of an object, we give him a description of the shape of the object as viewed from straight on. If he then asks 'But is the object really that shape when unobserved?' the right answer is 'Yes'. This is the right answer because a description of shape is a characterization of certain invariant relatione holding among the various boundaries or points of the object that determine it. We are talking of relations which are intrinsic to the object, and which can be talked about without reference to an observer. If the questioner proposes a new application of the words 'real shape' then he is entitled to frame hie questions in the appropriately different way. But he should remember that words associated with the old use of 'shape' words such as 'length', 'surface', 'distance', etc., will probably need to have their meaning changed accordingly.
Unfortunately, colours do not appear to be susceptible of a similar treatment. For, if the colours we see are functions of a complex situation involving an observer, the condition of his sensory apparatus and the nature of the light source, then to describe unobserved objects as having certain precise colours categorically may be unjustified. I do not particularly want to adopt this view, and perhaps it can be shown that I do not have to, but at present it seems to me that the onlypage 49categorical I can justifiably assert is one to the effect that a given object has a certain invariant property in virtue of which I perceive that object as red, say, rather than some other colour. Perhaps talk about this property is physicist's talk — about unobservables — and perhaps such talk is quite appropriate here, for, as statements about unobservables are statements whose truth or falsity is independent of the presence of observers, such statements would seem to provide suitable categoricals.
The argument I have given looks very much like a plea for Locke's distinction between primary and secondary qualities. If this distinction is meant to discriminate between those perceived qualities which are invariant in all perceptual situations and those which are not, then I have little in common with Locke, for there are no such qualities. If, however, primary qualities are those properties which science attributes to objects, then you can, perhaps, condemn or commend us both together. At any rate, the essence of what I have been saying is this: there can be description on two levels; the first is the common-sense level, which maintains the mind independence of the characteristics or qualities of material objects. However, in the case of certain qualities it seems that such independence does not obtain, and that descriptions involving such properties of unobserved objects can plausibly be construed hypothetically. In these cases we can give a meaning to real colour, smell or taste which is by no means arbitrary, but this is a distinction within experience and allows of no extrapolation beyond the context of experience. On the other hand, there are descriptions of properties of objects which, because they prescind from all observers, are true or false regardless of the presence or absence of any observer.