Chapter 1

C. D. Broad, Perception, Physics, and Reality, 1914

CHAPTER I

ON THE ARGUMENTS AGAINST NAÏF REALISM INDEPENDENT OF THE CAUSAL THEORY OF PERCEPTION

    In the present chapter we are going to begin from the position of naïf realism. It is true that our everyday view of the world is not quite naively realistic, but that is what it would like to be. Common-sense is naively realistic wherever it does not think that there is some positive reason why it should cease to be so. And this is so in the vast majority of its perceptions. When we see a tree we think that it is really green and really waving about in precisely the same way as it appears to be. We do not think of our object of perception being 'like' the real tree, we think that what we perceive is the tree, and that it is just the same at a given moment whether it be perceived or not, except that what we perceive may be only a part of the real tree.
    But, as everyone knows, we do not stay in this happy condition of innocence for long. We have perceptions which are believed to be illusory by which it is meant that their objects only exist when they are perceived. Common-sense does not believe in the reality of what is perceived in dreams or delirium. It applies certain arguments which it thinks are fatal to these. But as soon as it has rejected these objects it needs some account of the way in which the perceptions of them are produced. This leads it to a causal theory of perception, and when this is worked out it leads further still away from naïf realism. We shall discuss the causal theory and the arguments based upon it in a later chapter. Here we want to consider those that are independent of it.
    Although in a sense naïf realism stands at the opposite pole of thought to phenomenalism, there is another sense in which it resembles it much more closely than any intermediate philosophic position about the reality of objects of perception. The point of difference of course is that the naïf realist maintains that in perception we are directly aware of what would equally exist and be unaltered in any of its qualities at the moment at which we perceive it even if we did not perceive it, whilst the pure phenomenalist holds that what can be perceived must be perceived, and that nothing exists except mental states and the objects of those of them which are perceptions so long as they remain objects of them. The point of resemblance is that neither the phenomenalist nor the naïf realist can admit appearance or illusion. For both everything is precisely as it appears; the only difference is that a phenomenalist holds that all that exists must appear and that it ceases to exist when it ceases to be perceived.
    Before we consider the particular arguments which are supposed to make it inevitable to desert naïf realism we must consider what precisely it is that those arguments are supposed to prove. They want to prove that some at least of the objects of which we are immediately aware cease to exist when we cease to be aware of them. naïf realism is the denial of this proposition. There are degrees to which this belief is dropped and we want to see how far, if at all, arguments show us that it ought to be dropped. Thus common-sense would say that a mirror-image of a pin only exists when someone sees it, if, by the mirror-image of a pin you mean something that looks exactly like the sort of pins that we can feel and with which we can scratch our fingers. On the other hand it would say that the pin whose reflexion is the mirror-image exists whether we perceive it or not precisely as we do perceive it when we do so. But natural science would say that what we perceive in this latter case even is not what exists when we do not perceive it, nor even particularly like it. There is no reason, it would hold to believe that it has a colour or a temperature, but most likely it consists of little hard colourless things of uncertain shape vibrating rapidly and never getting far away from each other. Idealism again would say that what is real is not in the least like what we see, but that it is a thought in God's mind, or a community of spirits of low intelligence.
    Now there is one very important point in all this, which is often overlooked. This is that whatever else may or may not exist it is quite certain that what we perceive exists and has the qualities that it is perceived to have. The worst that can be said of it is that it is not also real, i.e. that it does not exist when it is not the object of someone's perception, not that it does not exist at all. When I see a pin that of which I am immediately aware is neither colourless atoms nor a community of spirits; and this is a matter of simple inspection. But it is also quite certain that the objects of existent perceptions exist at least so long as the perception of them does so. Hence on any tenable view of the world there exist things in it that are coloured and hot and extended. The only further points of interest about these qualities are

whether they can ever exist except when someone is immediately aware of them, since it is quite certain that they do when people are aware of them and that many folk are aware of them from time to time; and
whether there is anything in the nature and quality of these objects of immediate awarenesses which justifies a belief in the existence of other things of which no one is immediately aware, which differ more or less from those objects, and yet have a peculiar relation to them which I shall at present denote by the purposely vague phrase 'correspondence.'

    We saw that common-sense accepts naïf realism wherever it is not forced by arguments to abandon it, and we can now see some of the general conditions to which all special arguments must conform. In the first place we must be able to assert that, if p be any quality the attachment of which to any subject A is a mark that the latter cannot exist unperceived, it must not also be fatal to A's existence when perceived. p must only be fatal to the unperceived existence, i.e. to the reality of A and not to A as an object of perception; since it is solely by examining the perceived A which exists that it has the quality p in question. It is not valid to say that, in reply to this, we might argue: Perceived A has not the quality p, but if A were unperceived it would have p, and therefore A, cannot exist unperceived. For, if it has a new property when unperceived it is not A the possibility of whose unperceived existence is denied but something else A' which differs from A in another respect beside that of being unperceived.
    I do not know whether this line of argument will be considered shocking or truistic. I think it is merely the latter; but it is certainly worth insisting upon, since in these matters truisms are disregarded as the necessity for Mr Moore's 'Refutation of Idealism' showed. The great service of that article is to insist on the truism that when you perceive you perceive something, and that what you do perceive cannot be the same as the perception of it. And Mr Moore was undoubtedly right in saying that the recognition of that truism is a complete refutation of most of the current arguments for idealism.
    Let us then apply this general reasoning without undue nervousness. I will take an example. We all perceive extended things in spatial relations, and these perceptions give rise to a series of judgments about extension. The hundred and one arguments against the reality of space independent of the causal theory of perception consist of attempts to show that when you reflect on these judgments they are found to be in contradiction. Now these arguments may be true or false: -- they are all so far as I know false. But let us suppose that one at least of them is true. Then we must admit that nothing extended exists unperceived. But, since it is notorious that we do perceive extended objects, and since it was only by reflexion on the qualities and relations of such objects that we arrived at the judgments that are now proved to contradict each other -- we must also admit that, if our argument be correct something must exist -- at any rate as the object of an existent perception -- which has characteristics that lead to contradictory propositions, even if it be granted that such things could not exist except as objects of perception. In fact we must not say, that a perceptible quality that leads to mutually incompatible judgments about itself or that of which it is a quality cannot exist, but that it cannot be real, i.e. cannot exist otherwise than as an object of perception.
    If such cases really arose we should have to come to some compromise between three very strongly held beliefs:

that we cannot possibly be mistaken in our belief that the objects of some of our perceptions have that peculiar characteristic which we call extension,
that we cannot possibly be mistaken in the propositions about extension which we come to believe in reflecting on such objects of perception, and
that contradictory propositions cannot be true even of something that only exists when perceived.
It seems to me that (1) and (3) must be accepted, but that it is only under certain special circumstances which we shall mention later that we are obliged to accept (2). Hence if it were true (which it is not) that the standard arguments proved that propositions which we believe about extension are mutually incompatible the right course would be neither to reject the reality of extended objects, nor to hold that objects which give rise to incompatible propositions about themselves can exist when perceived though not otherwise, but to conclude that some of the beliefs which we reached by reflexion on such objects were mistaken.
    It is obvious that the main general difficulty of the arguments that we are going to discuss in this chapter will be to find propositions on which every one will agree that if they be true of an object of perception then it cannot exist except when perceived, propositions which at the same time are admittedly true of what we do perceive. The fact that the simple point which I have laboured perhaps too much above has seldom been grasped or explicitly stated is the cause, I think, of the unsatisfactoriness of most of the arguments against secondary qualities. After reading the arguments one is left with an uneasy feeling of puzzlement. What precisely is it that they do prove? Do they prove or only render probable! And must they not assume many more axioms than their users make explicit to prove as much as they seem to be supposed to do? Our best plan then will be to go over the main arguments seriatim and to try to make explicit those propositions that must be assumed if they are to be logically valid grounds for believing that their conclusions are certainly or probably true.

    Before we discuss the arguments in detail it will be well to define the meanings that we intend to attach to certain terms which we shall constantly use. The word object is ambiguous. When I perceive a rainbow or a chair in waking life and when I perceive anything in a dream all that I perceive can equally be called objects. I intend to use the word in the general sense of anything that may be perceived, regardless of the question whether it can exist if it be not perceived. There is a narrower sense of the word which would confine it to objects that are supposed to exist whether they are perceived or not. Thus the dream chairs and tables would not be called objects at all, nor would the colours of the rainbow, but the chair perceived in waking life and believed to exist whether perceived or not would be called an object. In the case of the rainbow it would be quite in accordance with usage to say that the object of the perception of the rainbow was little drops of water with light shining on them. This use is very common where the object of perception in the wider sense is believed to be very much like the object in this narrower one and to have qualities which vary continuously as we move about and which, in some limiting case, coincide with those of what exists whether we perceive it or not. Thus suppose there is a real circle. This will be an object in the narrower sense, since it will continue to exist and be circular whether we perceive it or not according to the usual opinion of common-sense. As we move sideways however what we really perceive is a series of ellipses of continuously varying eccentricity. But if a man standing sideways to the figure were asked: What object do you perceive? he would almost certainly answer: A circle; because he believes that what exists independently of his perception has the quality of circularity. Hence it is not uncommon to call the object of perception in such cases not the object that is actually perceived but the object that is inferred from what is perceived to exist whether it be perceived or not.
    This confusion has been responsible for much error and it is necessary to be clear about it before we can go on. We shall continue to use 'object' in the wider sense to cover, anything that actually is sometimes immediately perceived, whether it exists when we cease to perceive it or not. We shall divide objects into appearances, which only exist when they are perceived, and realities, which exist unchanged whether perceived or not. All appearances are objects, but it does not follow that all realities are objects. For we might have grounds for believing in the existence of realities which we never could directly perceive. Such realities would be a light-wave or an electron. Thus we must accept the possibility of real non-objects, though there cannot be apparent non-objects. When I do not want to beg the question of whether we are dealing with an object or a non-object I shall call it a term, but we shall not often want the word.
[Table added by A.C.]

terms

objects
(in wider sense) non-objects

appearances realities non-realities

    We shall now allow common-sense to assume that naïf realism is true in all cases where there is no reason to believe that it is false, and consider the reasons other than the causal theory of perception that have led people to suppose that in many cases it must be rejected. This will be the easier as the type of argument is exemplified in the rejection of secondary qualities by Locke and Descartes, and of primaries by Berkeley and Bradley. The reasons that have been offered for the rejection of the reality of certain sense-qualities have been numerous. The following list makes no claim to be exhaustive, but I do not think that it misses any argument of importance.

Spatio-temporal incompatibility (e.g. the hands in water experiment).
Conflicting testimony of two different senses in the same person.
Conflicting testimony between the same sense in two different people.
Perceptions in dreams and morbid bodily states.
Qualities like temperature merging into feelings of pleasure-pain.
Relativity of what is perceived to the position and past history and present structure of a bodily organ and to the use of instruments of precision like microscopes.
We can really group (1), (2) and (3) together as far as the main principle on which they rest is concerned, though their differences will demand separate discussion. The point of agreement is that in all the senses testify to the existence of combinations of characteristics in one spatiotemporal point which are known to be synthetically incompatible. (4) and (6) are in the main independent of each other but not entirely so. Morbid bodily states may be produced by drugs and are then supposed to cause illusions; but what is perceived by means of instruments of precision is also caused by extraneous means acting on the body, yet this is supposed to add to our knowledge of the real. Again (4) and (6), and more especially (6), are chiefly concerned with causation of perceptions. As such they will be largely reserved for a later chapter. We can then put our final list of arguments in the following scheme.

Synthetic Incompatibility

Testified by one sense in one person.
Testified by different senses.
Testified by one sense in different people.

Mainly causal

Relativity to organs of perception and the use of instruments of precision or drugs.
Occurrence in dreams or morbid bodily states.
Merging into feelings.

    I shall begin by trying to show that I(a) which has generally been accepted and is almost the only argument used by Locke rests on a confusion and is incapable by itself of proving or rendering probable what it has been supposed to do. For this purpose I shall take Locke's example of the two hands in water which is supposed to disprove the reality of temperature.
    As everyone knows in this experiment the two hands have been treated differently, the one having been plunged into a vessel of cold water and the other into one of hot water. They are then put into a third vessel of water at temperature intermediate between the other two, and the water in this appears to the cold hand to be hot and to the hot hand to be cold. Hence it is argued that the water cannot really be either hot or cold, for if so it must be both hot and cold at the same time, and this is held to be impossible. As it stands then this is an argument from synthetic incompatibility and is an example of I(a). But in as far as what is perceived depends on the past history of the organ it is an example of II.
    The moment we consider it carefully we shall see that its weakness is deplorable.
    (i) In the first place it is to be noted that it cannot make it even probable that two bodies at least -- viz. my two hands -- are not really hot or cold. It is evidently not true that I cannot have contemporary perceptions of different temperatures. It is notorious that I do, and that is indeed one of the data on which the argument is based. The argument rests on the proposition that no body can have different temperatures at the same time and place. But how could this proposition have been discovered or proved if as a matter of fact no bodies have any temperature at all? It would have to be held that we are immediately certain of the following proposition: If (per impossibile) bodies could have real temperatures they could not have two different ones at the same place and time. The fact is however that the argument assumes what is no doubt true that I can never feel the different temperatures at the same time and place. But I localize the temperatures that I feel on my skin, and, since I never do feel the same part of my body to be at different temperatures at the same time, no experiment can give me any reason to think that my own body, at any rate, cannot have temperature.¹ Thus the argument gives no reason for thinking that my two hands in Locke's experiment are not really hot and cold respectively as I perceive them to be, and that caeteris paribus they might not continue to be so when I cease to perceive their temperatures.
    (ii) It is clear then that the argument cannot apply to the temperatures that I directly perceive, viz. those of my hands, but only at best to inferences from these to the temperature of the surrounding medium. Now, if I make the particular assumption that the temperature of the water is that of the hand with which it is in contact, it might be thought that the argument will be valid as a proof that the water has no real temperature. But once more the assumptions made are inadequate, and that in several respects.

If two propositions together lead to a false conclusion you have no right to reject one rather than the other on these grounds alone. Hence if the experiment + the assumption that water has temperature + the assumption that the temperature is of the same degree as that in the hand did lead to the impossible conclusion that the water has different temperatures at the same time and place, this cannot be said to disprove the assumption that it has temperature rather than the assumption as to the magnitude of the temperature if it has it. One assumption would have to go, but the experiment throws no light on which it should be.
But, as a matter of fact, neither assumption need go, and that for a very general if empirical reason which the supporters of this kind of argument seem to have overlooked. That reason is that the conditions which the experiment requires to be satisfied never are and never can be fulfilled. Either the fingers touch or they do not. If they touch the water is not felt but the fingers. If they do not then they are not at exactly the same point in the water. Nobody in the world ever has had his two hands at the same time in the same place, and we may say with some confidence that nobody ever will performed that feat. But, failing it, the whole argument falls to the ground. The experiment + the assumptions can now only prove that at the moment at which it was tried the water had different temperatures at two points very close together. And there is no objection empirical or à priori to such a state of affairs.

    (iii) Supposing that all these difficulties could be set aside (which they cannot), how could the right conclusion be that the bodies have no temperature? We have seen that some bodies, viz. our own, must have temperatures at any rate while we perceive them. For we do perceive them and we do perceive them on our skins. And there is nothing to lead us to suppose that the temperatures of our bodies might not exist even when not perceived. Hence it seems much more reasonable to conclude from the experiment that, though other bodies have temperatures, we cannot tell precisely what their degree is from that of the felt temperature of our hands, than that other bodies have no temperature at all and our own none except when we perceive it.
    Most of the arguments that Berkeley employs are vitiated by a certain fallacy which would prevent him from taking the alternative that we have suggested as the more reasonable one in view of the experimental results and the other facts that are known. This fallacious argument is the following one. He says, what is perfectly true, that those sensible qualities that have intensive magnitude, for instance, cannot exist in general but can only do so with some definite degree of that magnitude. But his ridiculous doctrine about abstract ideas leads him to conclude from this that it is impossible to hold that a certain quality exists with some intensive magnitude although one does not know what the particular degree may be. There is of course nothing impossible or unreasonable in combining the belief that such qualities can only exist in definite degrees with the belief in the existence of such a quality whose definite degree we are unable to determine although we know that it must be determinate.
    We may then sum up the objections to I(a) in as far as it applies to qualities which are perceived as localised on our bodies and extended.

You cannot directly perceive synthetic incompatibilities. Hence the perceived qualities with which the argument starts cannot be synthetically incompatible. There is therefore no reason why they should not be real qualities of your body at any rate. Hence synthetic incompatibility with regard to the deliveries of one sense whose appropriate sense- quality is localised on the body cannot possibly prove that what is directly perceived there might not continue to exist even if it ceased to be perceived. For there can be no reason to suppose that what was synthetically compatible while perceived, will suddenly become synthetically incompatible when you cease to perceive it.
The argument can therefore only apply to what is inferred to exist in other bodies from what you perceive to exist in your own. But then the argument alone will not tell you whether you ought to reject the belief that the quality exists unperceived in these other bodies, or the belief that its magnitude is related in such and such a way to that of the perceived quality in your own body.
The Berkleian objection to the possibility of holding that a quality can be known to exist with a definite intensive magnitude whilst we cannot be sure as to the precise degree of that magnitude is invalid.
Finally, owing to the fact that our organs of sense are extended and incompenetrable, the common assumption that, in the case of those that act by contact, the quality is the same at the point touched as at the touching point of the organ can never be strictly refuted by this line of argument.

    I have assumed in the discussion of the above experiment that the quality of temperature is directly localised in our skins and then inferred to exist elsewhere. I believe this to be true; but, for a general criticism of this line of argument, we want to see what will happen if we remove this restriction. I do not think it will make any essential difference. It is necessary that the qualities to which the reasoning from synthetic incompatibility is to be applied shall be directly localised somewhere, for otherwise their spatiotemporal incompatibility is meaningless. And at that point, whether it be in our organ of perception or in a foreign body, it is impossible that there should be synthetically incompatible qualities, since the qualities there are directly perceived, and you cannot perceive incompatible qualities at the same time and place. Hence as before it will always be a question of arguing from perceived and therefore compatible qualities to unperceived ones that are supposed to be incompatible and, wherever the perceived qualities that are the data of the argument be localised, the same criticisms will apply as we have enumerated above.
    I think it will be worth while to say a word or two about localisation before leaving this subject. It is obviously all- important in questions that depend on synthetic incompatibility, and it is less simple than it seems. What precisely do people mean by saying that different degrees of temperature (e.g.) cannot coexist at the same point? Of course the hands in water experiment with its usual carelessness asserts the obvious falsehood that the same body cannot be both hot and cold at the same time. In the previous discussion we made this precise by replacing it by 'different degrees of temperature cannot coexist at the same point.'
    This statement seems obvious enough, but it is by no means free from difficulty. When we talk of 'the same point' in what space precisely do we suppose the point to be? The spaces revealed by our various senses cannot be immediately identified, and what we call Space is a synthesis and construction with these as data and elements. When we say that two different degrees of temperature cannot coexist at the same point, do we mean 'at the same point in the space of the sense (viz. touch) by which we perceive temperature' or 'at the same point in the supposed common Space'? And we have two further alternatives before us. Either the proposition is supposed to be founded by induction on experience, or else it is à priori in the sense that it is based on the mere contemplation of the nature of temperature. If the proof be inductive and it takes the form that we never have experienced two different degrees of temperature at the same time in the same point of tactual space, and therefore they cannot coexist there, it is singularly weak. It may be a good enough argument to render it probable that we never shall experience different temperatures together at the same point, but it cannot possibly prove that they do not coexist. For an alternative explanation is that we are only capable of perceiving one of them at a time; for instance it might be always the highest of several temperatures coexisting at the same point that we perceive. For if the principle is to be used against people who believe that unperceived temperatures may exist, it must not assume that they are wrong in order to prove itself.
    But I think that a good many people would say that their belief is not founded on induction from experienced lack of coexistence, but is à priori in Meinong'a sense, i.e. although it is in perception that we become acquainted with temperatures, yet, once acquainted with them, we can see that from their very nature it is impossible that different degrees should exist contemporaneously at the same point. This position is at any rate much stronger than the one which we have just discussed. We saw that that would not even prove the proposition for tactual space, a fortiori experienced lack of coexistence in tactual space will not prove absence of coexistence in an assumed common Space. But the à priori certainty, if it exists at all, may be very general; it may even take the form that² quantities with different intensive magnitudes of the same kind which are also extended cannot exist contemporaneously at any point of any space.
    So far as I can see the statement in terms of points, like points themselves, is the child of abstraction and intellectual construction; it is convenient but may not correspond with anything that is real, and ought for the present purpose to be avoided. The more concrete form of statement is that no volume in any space can be wholly occupied at the same moment by each of two quantities possessed of different intensive magnitudes of the same kind. Of course, if this principle be really à priori there is nothing further to be said; but is it?
    We must remember that our principle is supposed to apply not merely to intensive magnitudes but also to shades of colour which are not quantities that differ in intensive magnitude, but are capable of arrangement in a continuous order. It will be best for the moment to confine ourselves to colour because there is no question here of anything corresponding to physical conduction as in temperature. Suppose we have two closed surfaces, one red and one blue, and that they are not initially in contact. Now I do not pretend to know exactly what common-sense means by a surface, but two possibilities are open. Either there are or there are³ not outermost points to physical surfaces. There is no means of deciding this question; but we can at least say that it is arbitrary to suppose that some surfaces have and others have not outer-most points. Now let the two surfaces be brought into contact. I do not know exactly what common-sense means by this, but I take it to mean that a point of contact is either an outermost point of one surface or an outermost point of the other. If then neither has outermost points they cannot be in contact. But we suggested that it is reasonable to suppose that if either has outermost points both will have them. Hence if there are points of contact they must be outermost points of both. But the points of one surface are red and of the other blue; hence at the points of Space occupied by the points of contact red and blue will coexist.⁴
    If this argument be sound common-sense must reject or modify its alleged à priori principle, or deny that coloured bodies are ever really in contact, or introduce a new physical law ad hoc, or reject the reality of colours. With regard to the last three alternatives a few words must be said. There is no evidence that bodies ever are in contact with each other in physical Space unless we suppose that the denial of actio in distans is another à priori axiom. The latter seems to me to have no self-evidence, while the axiom about synthetic incompatibility has considerable plausibility. And, if we admit -- what seems reasonable -- that in the object of a present perception there may be distinctions that are not perceived, there would be a meaning in the statement that no bodies are in contact in perceptual space too. By the introduction of a physical law ad hoe I mean that we might suppose that it is a law of nature that as two coloured outermost points approach each other each modifies the other's colour so that they become more and more alike, and at last when they coincide their colours are identical. You may say that this is not a very probable law on the ground that it seems to treat outermost points in a different way from others, no matter how near them. But there is no need to assume this. When these coloured bodies approach each other all points in the one may influence the colours of all points in the other; but if the influence decreases very quickly with the distance everything will appear unchanged. With these two alternatives open to us it would be foolish to deny the reality of colours on the ground of the principle and the apparent fact that differently coloured bodies can come into contact.
    I suspect however that common-sense would be inclined to keep the possibility of contact and to introduce no physical law, but to modify the principle in the direction that it is no matter that different colours should coexist at the same point of physical Space so long as they are the colours of different bodies. But, until common-sense can give some account of what it means by the same and different bodies which shall render plausible the distinction that it is now trying to introduce, it is hardly worth while to discuss this modification.
    If the above discussion has accomplished nothing else it has perhaps served to show that the alleged à priori principle is by no means so simple as it looks. I do not wish to deny that it may be à priori, but I wish to suggest that it can hardly be used with confidence as a metaphysical criterion of what may and what cannot be real. For an à priori judgment is one where, given the terms, we are supposed, by contemplating their nature and without further appeal to experience, to be able to make assertions about some of their relations. And we cannot place much trust in such judgments when a little discussion shows that our minds are in a most nebulous state as to the nature of the terms and the exact meaning of the assertions about them.
    To conclude with some rather more positive considerations about the alleged à priority of the principle of synthetic incompatibility, I must confess that when I ask myself why I do not believe that two different temperatures or two different colours can contemporaneously fill the whole of any one volume, I am not contented (as I ought to be if the judgment were à priori) to refer merely to the nature of temperature or colour or intensive magnitude as such. I rather say to myself: 'If the different temperatures ever did exist they would at once be equalised by conduction,' and 'if I put one pigment on top of another they mix to a new colour.' But these are empirical laws which may be either laws of the physical world or of the limitations of our perceptive faculties. In either case they may have exceptions, and in the second case, even if there are no exceptions there is no proof that different colours and temperatures do not really coexist. Thus if there actually were experiences which, on the assumption that temperatures or colours are real, would force us to conclude that different temperatures or different colours can fill at one time a single volume, it is not clear that they would necessitate the rejection of the reality of colours or temperatures, since they might be interpreted as examples by reasoning from which we could prove that an alleged physical law has exceptions or that our perceptive faculties have limitations. And neither of these results would be so startling as to compel the rejection of the hypothesis on which they are based.
    Since the argument from synthetic incompatibility thus assumes that the qualities to which it applies are extended and localised there is only a certain number of qualities to which it does apply. These are temperatures, pressures on the skin, colours, and flavours. It does not apply to sounds, because these are not generally localised in the way that we might at first suppose, and moreover are not synthetically incompatible. When we hear a bell tolling we might be thought to localise the sound in the bell, but this is not really comparable with localising temperatures on our skins or colours in definite places. It is impossible to feel temperature without feeling a definite hot (or cold) surface, or to perceive colour except by perceiving a definite coloured surface: -- this is even so when we press the eye and see stars or rings. But it is perfectly possible to perceive a sound without localising it in that body that causes it. When we do think of extension and position in respect to sounds it is not of a volume that has the quality of sound but of one that is in a state of 'sounding,' i.e. of being the cause of the sound or of the perception of it. But we never think of the surfaces that we perceive as coloured, as being not really coloured, but only the causes of the colours. For, even if we adopt the ordinary natural scientific view about colours, it is quite clear

that we perceive colours as extended, and
, granted that this perception is caused by surfaces and volumes as sound is believed to be, it cannot be by these surface and volumes that we do perceive, for they at least are coloured whilst science denies that the causes of our perceptions of colour are coloured.
If sound as such be a quality it is perceptually and immediately localized in the space between our ears and the sounding object, and there is no synthetic incompatibility, for that space can be contemporaneously filled by all kinds of different sounds.
    I think we may now claim to have shown that arguments of the type I(a), i.e. arguments from synthetic incompatibility alleged to be testified by one sense can never by themselves prove or render probable the unreality of the qualities against which they are directed. We can pass therefore to

    I (b), i.e. to the denial that a quality can exist unperceived owing to the synthetic incompatibility of the testimonies of two different senses. There is only one set of sensible qualities about which more than one sense is held to be capable of giving direct information. Those are the geometrical qualities like shape and extension, which are perceived by both sight and touch. Now, as everyone is aware, sight and touch can be very discordant in their deliveries, and that in more than one way. For instance, two cases of what looks cubic to sight are sometimes found on being touched to give widely different tactual and muscular sensations. For example, what we should commonly call a 'real cube' will give tactual perceptions of eight corners, twelve edges, etc., whilst the picture of a cube which may look exactly like the former will merely be felt as a flat surface. If then it be held that in sight and touch we become aware of the same reality, then the reality of which we become aware cannot in general be the common object of sight and touch, since we have seen that there is often not a, common object to the two perceptions, as distinctions can be found in the one that cannot be found in the other. We are sure that the same object cannot be both a plane surface and a three-dimensional figure, and therefore, even though sight and touch be indications of the same object, that object cannot be what is directly perceived in both of them and may not be what is directly perceived in either. In the same way, of course, you can get tactual perceptions of three-dimensional objects to which visual perceptions of two-dimensional ones correspond, as when you look straight on to one side of a cube and merely see a square surface, while you can still feel the corners and edge and other faces.
    Now what precisely do these facts taken by themselves prove? It has always been recognized that the perception of extended shapes in definite positions is a much more complicated affair than the perception of the objects of other senses. By this I mean that, although there is of course a large perceptual element in it, there is also a great deal of judgment, association, analysis, and subsequent synthesis involved which are not present to anything like the same extent in what are called the 'perceptions' of the other senses. We want to separate off the perceptual element from that of judgment in order to determine precisely where the common element in perceptions of touch and of sight lies.
    It is sometimes held that perception cannot be defined except by reference to its supposed causes. This, I think, would be a dangerous doctrine. Of course there is in it an element that is indefinable, viz. that of direct awareness of an object. But, then, many people think that there is direct awareness of something in all kinds of cognition, and they none the less recognise the distinction between perceptions and the other sorts of cognition. Hence even if direct awareness as a matter of fact only happens in perception, there is some reason to believe that there must be other marks that distinguish perceptions from judgments. Supposing that there is direct awareness in all forms of cognition, I think we may still be able to find certain common qualities that objects must have for the direct awareness of them to be called perceptions. To begin with, I think there is one negative thing that we can say about perception. Granted that there is direct awareness in all kinds of cognition, still the object of the awareness in the other kinds stands in a peculiar relation to something else, whilst the object of direct awareness in perception does not. When I think of the 47th proposition of the first book of Euclid it is extremely difficult to say of what it is that I am directly aware; it probably differs from person to person, and in one person from moment to moment. But it always has a peculiar relation to the squares on the sides of a right-angled triangle and to their metrical properties such that the whole cognition is said to be 'about' these properties. Now in perception pure and simple there seems to be nothing to correspond to this 'about.' The stage of pure perception of a coloured surface corresponds, it would seem, to the state of a man who just reads the enumeration or sees the figure of the 47th proposition, but to whom the words and figure convey no meaning. Now neither in the case of sense perception nor in the mere apprehension of words are we forced to remain at this point, or do we generally remain there. We generally do proceed to other cognitive acts that introduce the notion of 'about,' but perception differs widely from other kinds of cognition in the acts to which it commonly gives rise. The peculiarity of perception is (a) that it may give rise to cognitions that are about its own object; and (b) that it is not merely held to give rise to them as a matter of psychological fact, but is also the sole and sufficient guarantee of their truth. The cognitions to which a direct perception gives rise and which are 'about' its object are held to be indubitably true. This not purely psychological criterion is, I should say, a sufficient one for a perception. Such propositions have a certainty even greater than that of memory. We recognise in the abstract that memory may be wrong in particular cases, and the nature of the certainty here is that it is reasonable to regard it as right in any particular case until you can prove that it is wrong, whilst any disproof in a particular case will only be possible by means of some other memories, and therefore on the general assumption that memory is true. But we do not even contemplate it as possible that the judgments about the object of a perception to which that perception gives rise should be false.
    The matter will probably become clearer if I give some examples of what I mean. Suppose I assert the proposition that Socrates is mortal. I shall undoubtedly be directly aware of something, but, whatever else it may be, it will not be Socrates. If I happen to be a visualiser probably one of the things of which I shall be directly aware when I make the assertion will be an image which is rather like busts of Socrates that I have seen. But no one supposes either that I am saying that this image is mortal, or that my immediate awareness of that image is any justification for the belief that Socrates is mortal. But suppose I judge that my mental image has a snub-nose. This is a judgment 'about' the object of which I am directly aware, and its truth is justified by it in such a way that it would be madness to doubt it. My immediate awareness of the image then was a perception on our definition. Now it is, of course, true that perceptions are commonly held to justify propositions about other things than their own objects, but we must note

that it is always by a process of inference or comparison that they do so;
that these inferences and comparisons will never have that infallible certainly which the propositions that perceptions justify about their own objects possess; and
if it really be a case of perception there will always be some propositions of this latter kind which can be asserted, even though they will not be the whole of the propositions that the perception is held to justify.

    Of course it ought to be noted that as soon as such propositions are put into words they cease to be absolutely infallible. The meaning that is expressed to me by the statement 'my mental image has a snub-nose' is infallible, and everyone would admit that I could hardly be mistaken as to the shape of what I perceive. But when I put the judgment into words it is assumed that I mean by 'snub-nosed' what other people do and that I have rightly subsumed the present case. But this involves comparisons that carry me outside the object of my present perception, and in them I have no infallible guarantee of correctness.
    The above discussion is relevant to the question of the perception of figure and extension. The point is that different kinds of objects of perception guarantee propositions about themselves which vary very much in number and importance, and that, in the case of spatial qualities and relations, the judgments are made so instinctively and are so much the most important part of what is commonly lumped together under the name of 'spatial perception' that it is much more difficult here than in any other case to decide what is really the object of perception as such. We can perceive a colour without thinking of any propositions about it, and the number of those that it will justify is small, whilst they are not of great practical importance. But absolutely the reverse is the case with objects of spatial perception.
    We can now try to decide what weight ought to be given to the discrepancies between sight and touch. In the first place, the purely perceptual element in these two senses is clearly not the same. When I see a cube, and when I feel corners, edges, and faces, it is obvious that what I perceive in the two cases differs. I see extended colours with boundaries; I feel temperatures and texture which also have a quality that I call extension, and I feel certain discontinuities in these objects which I call boundaries. Of course when I have identified tactual and visual extension I call the seen and the felt extensions by the same name, and so too the seen and felt boundaries within which there are not these sudden discontinuities. But the objects perceived by the two senses are never the same, for if they were the sole difference between seeing and feeling would be that one is done with the eye and the other with the skin. But, it will be said, no doubt there are peculiar features in the objects of each sense, viz. colours in sight and temperatures, texture, etc., in touch, but still there is something that is definitely common to the two, and that is extension and figure, which is what interests us now.
    Of course it is clear that there is some analogy, but it is very important and not very easy to be clear as to its exact nature. The analogy seems to be that our sensations of sight whose objects are colours give rise to and justify judgments which analyse out of them the quality of being extended, and of having definite boundaries, and of having spatial relations. Our perceptions of temperature and texture can also have a peculiar quality analysed out of them; they too have objects in which there are discontinuities which constitute boundaries; and they too have certain kinds of relations. But, even so, I do not think that the tactual extension, figures, and relations that can thus be analysed out of the objects of tactual perception can be immediately identified with the visual extensions, figures, and relations that can be analysed out of seen colours. For instance, so far from agreeing with Berkeley that distance cannot be perceived visually, it seems to me quite obvious that it is perceived visually and that it cannot be perceived tactually. Whatever may be the cause or the history of my perception it is surely clear that I actually perceive the opposite side of the Great Court of Trinity as further away than the side in which my own rooms are. It may be perfectly true that my judgment as to how far it is from my side may depend on previous experiences of walking across the court. But feelings of accommodation and convergence associated with revived muscular sensations are not the perceptions of visual distance which the least introspection shows that we possess. Hence I am not at all confident that the extension and figure and relations analysed from objects of tactual perception can at once be identified with those that are analysed from visually perceived objects. What seems to be more true is that in visual extension we can analyse out elements and relations which form a spatial order of the same type as that which we also find on reflection to be constituted by the relations and elements that we can analyse out of the objects of visual perception. We do not perceive an elaborate spatial order, but when we come to analyse and reflect upon what we perceive both by sight and by touch we are led to construct spatial orders of the same type. Of course in people who are neither blind nor anaesthetic these orders are not constructed independently of each other, and so there are not two similar orders left standing side by side, but one which is supposed to include both. It must be noted that when I talk of 'constructing' a spatial order I do not hold, as so many people seem to do, that this implies that the order so reached cannot be that of the real world. There is not the least reason to suppose that the view tacitly assumed in so many philosophical works is true that as soon as you have analysed the steps by which you have come to believe that what is real has the quality or relation, X, it is perfectly obvious that this belief must be false.
    For our purpose we are mainly concerned with the belief that the objects of visual and tactual perceptions are such that reflexion upon the products of an analysis of them lead to the knowledge of a common spatial order. Let us take the cube that I both see and feel. Why do I suppose it to be the same cube, and what precisely do I mean by that statement? In the first place under what conditions do I pronounce that A, which I feel, and B, which I see, are the same thing? I can see an object B and at the same time part of my total object of vision may be my hand touching B. I see the hand moving over B, and, at the same time I feel the different movements of my joints and muscles and the different feelings of contact at edges, corners, etc. There is a correspondence as my eye follows my hand between the different muscular and joint sensations and the different convergence of the eye and between the sudden changes of direction in the boundary lines that my eye sees and my hand feels. Now we localise colours where we see them, and temperatures, as I have already suggested, are commonly supposed to be the same in the body felt as in the hand that feels them. Hence the temperatures are localised within the same space as the colours although the eye cannot perceive the one nor the hand the other. We thus say: There is a cold red cube in a certain place because (a) both temperatures and colours and textures are found on analysis to yield qualities and relations which on reflexion are seen to form wholes of the same general type, and (b) to each relation and discontinuity that can be discovered with regard to these qualities in the object of touch there is found to be one in the object of sight when tho eye perceives the hand to be in contact with the object of visual perception. Of course people do not set up an inductive principle that there will always be this correlation. But there is an association which causes an expectation of such correlation between the deliveries of the two senses, and when, as in the case of the painted picture of a cube in perspective, the correspondence fails, we feel disturbed.
    But it is easy to see that these facts taken by themselves yield no reason whatever for rejecting the belief that the objects of both senses continue to exist when they cease to be perceived. It merely tells us that to the geometrical distinctions that we discover in the object of vision there will generally correspond felt distinctions in the object that is felt when the finger is seen to be in contact with the seen object. Accordingly when we see an object with the geometrical distinctions that constitute a cube and then find that we cannot feel any discontinuity when the fingers are visibly in contact with a visible edge or corner we feel surprised. But then, as far as perception goes, there were two different objects. One was coloured and from it could be analysed the distinctions corresponding to faces, edges, and corners. The other was smooth and of some temperature, but it had no felt boundaries. We experienced by touch in fact what we generally do when we see that our fingers are in contact with a single indefinitely extended surface. We could move our fingers along the seen boundaries and yet we felt none of the stretching of the arm that usually accompanies such action, and none of that peculiar contact sensation that generally happens when we touch a sharp discontinuity. Well, it will be said, here is the eye assuring us of a sharp discontinuity and also that the finger is touching it, whilst the finger feels nothing of it. Surely nothing can be sharply discontinuous and perfectly continuous at the same time and place?
    We have already argued that we only perceive distances from the body with the eye, though we may judge of them by muscular sensation. Hence the fact that in the present case of the painted picture of a cube we see the figure in three dimensions, but do not feel any sensation of stretching the arm in visibly following with the finger the visible edges, that go out from the body is not really a case of synthetic incompatibility. It need only show that sensations of movement are not infallible signs by which to judge of distances in directions seen as going out from the body. Still it is right to remember that this is not merely a question of more or less in regard to the judgment of distance. Sight tells you that the object does extend away from you for some distance, whilst touch leads to the judgment that it does not extend at all in that direction. We might in fact put the argument in this way: the seen change of direction of the surface and the felt continuity are held to be at the same place because most of our visual and tactual experiences can best be interpreted on the assumption that muscular sensation is a sign of visual distance. But in a certain number of cases, like this one of the picture of a cube, if you proceed on the same principle you will be forced to hold that at the same place as judged both by muscular and visual tests the surface bends round as judged by sight and remains flat as judged by touch. Now this bending round is not a mere question of judgment. There is an appropriate tactual experience for this kind of discontinuity which is generally found to go together with the visual discontinuity of direction.
    Thus the real ground based on synthetic incompatibility of the deliveries of two senses for the rejection of naïf realism is pretty complicated. In the first place you must have identified the objects of the two senses in some way before the argument from the synthetic incompatibility of their deliveries need trouble you. It might be said: either the objects of your two senses are the same or they differ. If they be the same they cannot be incompatible, but if they differ then it does not matter whether they are compatible or not. The way in which the employers of the argument get over the difficulty is as follows. In ninety-nine cases out or a hundred to every distinction of a geometrical character in the seen object it is found that one will correspond in what is felt by a finger visibly in contact with the visible object. Hence, as we have seen, the geometrical qualities that can be analysed out of both sorts of objects are held to be identical and within that common boundary it is held that there exist those other qualities which it is the peculiar function of each sense to enable us to perceive. Hence the objects of sight and of touch when certain conditions are fulfilled are identified. And there is of course so far no reason for departing from naïf realism with regard to the objects of these senses. But now cases arise where this identification breaks down, and where, if we insist on identifying the geometrical qualities that can be analysed out of the objects of visual perception with those that can be analysed out of the objects of tactual perception we shall get to sheer synthetic incompatibilities, such as surfaces at the same place both having edges and going straightly and smoothly on in their old direction. Under these circumstances the first thing to do is clearly to deny the invariable identity of visually and tactually perceived geometrical qualities and relations under circumstances under which they had always before been held to be identical. This however is not by itself any good reason for holding that either the visually perceived qualities or the tactually perceived ones cease to exist when they cease to be perceived; for, as soon as you have ceased to believe that they must be identical, the fact that in a given case they prove not to be so is no objection to either of them. It is quite impossible to prove that the coloured surface may not bend round but that its bent part is intangible, whilst the hard, smooth, cold surface goes on without discontinuity. That is not the view that common-sense takes however. It has found that it never gets a tangible surface without also being able under suitable circumstances to perceive a coloured one, and it rejects as appearance coloured surfaces that are not tangible. There is no objection to such a procedure, but I think we have now seen that the argument from synthetic incompatibility of sight- and touch- deliveries is no conclusive reason for doing so, since it is not incompatible with perfectly naïf realism about the objects of sight and of touch.
    ⁵ We may enquire at this point why common-sense should reject as appearance the object of sight and keep that of touch where the identification of the two would lead to synthetic incompatibility.
    I think the usual argument that it would put forward in justification of itself would be the following: sight and touch agree in bearing witness to extension and figure, but they differ as to the figure. Also experience has taught us that spatial characteristics perceived by sight will in general have felt characteristics corresponding to them. Now the sole evidence for their existence is that they are perceived. Hence when, as in the present case, it is certain that the spatial relations perceived by two senses differ in part, it is reasonable to hold that those for which we have the witness of two senses are the ones that continue to exist when we cease to perceive them, and that those for which we have only the evidence of one are those that are mere appearances. Now sight and touch agree in bearing witness to the existence of an extended surface, but touch shows no discontinuity in it, whilst sight tells us that it is bent round so as to form a closed surface in three dimensions. Hence it is reasonable to suppose that what is real is that which has both senses for a witness, viz. the flat continuous surface.
    This is the familiar and trivial line of argument which holds that when several senses bear witness to an object it is more likely to be real than when only one does so. It is liable to be confused with the argument that carried Gibbon back from the Roman Church and back into the last will and testament of an affectionate but Protestant father. I want now to point out that it is really very paradoxical and extremely weak if taken by itself
    It is surely clear that the evidence of' two senses does not add the least certainty to the existence and qualities of an object of perception of either of them so long as that object continues to be perceived. For that certainty, as we have already seen, is the highest that we can have, and therefore no evidence can hope to increase it. Thus it is equally certain that what I see bends round and forms a closed surface in three dimensions so long as I do see it, whether, as is usually the case, I can feel the corners and edges, or, as is the case with the picture of the cube, I cannot. Hence the argument of common-sense is this rather curious one: although, whilst an object is perceived, agreement of more than one sense does not add in the least to the certainty of propositions about its existence and qualities which each sense separately justifies, yet such agreement does make it more probable that what exists when we cease to perceive it has those properties about which, when it was perceived, the senses agreed.
    Now, taken by itself, this is a very odd argument indeed, and I do not think that anyone will find it plausible. It gains any plausibility that it may have from an argument to prove a quite different conclusion. If you are basing a belief in the existence of something that you do not perceive on the fact that if it really exists it will probably produce such and such effects, then the greater the number of such effects observed the more likely it will be that this unperceived object exists and vice versa. Thus Gibbon's argument by which he ceased to believe in the Real Presence was valid enough if we take him to have meant that, if the Body and Blood had been really there, he would have had certain perceptions which he did not have. But the same sort of argument is used to distinguish appearance from reality when the appearance is admittedly there; and here it is quite useless. All that it can prove is that it may not be appropriate to call the object of my perception by the name by which I should if other perceptions were there as well, and not that an object with the qualities that we do perceive with one sense is any less likely to be real than one whose qualities we perceive with two senses. Thus, if the sacramental wine after appropriate manipulation had looked to Gibbon like blood but had tasted like wine, he would not have been justified in thinking it less likely that such a liquid existed when he ceased to perceive it than if it had both looked and tasted like blood. All that he would have been justified in saying would have been: If this liquid which looks like blood and tastes like wine exists when I cease to see and taste it, it will not be safe to call it either blood or wine, since each name will convey wrong impressions to people.
    I have discussed here one of the grounds which common-sense would allege for holding that that on which sight and touch agreed was what was real, when it has separated reality and appearance owing to the discrepancy that we have been considering, because I wanted to show the weakness of this common argument. Better reasons for preferring touch to sight will be discussed in the chapter on the Causal Theory of Perception.
    We have now examined the argument I(b) which attempts to disprove naïf realism from the synthetic incompatibility of the deliveries of two different senses. We have seen that it rests on even more assumptions than I(a), and that, like it, it does not by itself furnish any conclusive ground for the denial of naïf realism with regard to both senses. If naïf realism is to be kept however that identification of tactual and visual spatial characteristics which is found to account so well for the facts over a very wide area must be dropped. And, if we separate the two, we shall be left with this remarkable parallelism still on hand and calling for explanation. The actual procedure of common- sense is to hold to the deliveries of touch as real and to make the deliveries of sight, in as far as they conflict with those under circumstances where they might have been expected to agree, appearances. We saw that the argument that what is testified by more than one sense is more likely to be real than what is only testified by one is quite baseless; and we found what precisely such an argument does allow us to infer. So far then we have seen that there is very little reason why common-sense should desert naïf realism just where it does, and no reason as yet why it should make that part of what it perceives real and that part appearance which it is actually found to do.

    We pass then to
    I(c). This is the argument from the incompatible deliveries of the same sense in different people. Here again we are faced by a problem which is partly new but in many ways like that which met us as to what was meant by the identification of the objects of visual and tactual perception. This is the question: What do we mean by two people perceiving the same thing? On the naively realistic view from which we start the argument is faced by the old dilemma. Either what A and B perceive is the same or different. If it be the same there is no problem, if different, then, since the qualities of the objects of their perceptions are qualities of different objects, what matters it that they are incompatible? Why should not both objects exist quite comfortably unperceived by either A or B?
    Another question that arises is this: In what sense is the knowledge that each of us has about the objects of his perception communicable to other people so that we can really know that A and B (of whom one may be oneself) have at the same time objects of perception which have incompatible qualities? We will discuss these two questions in order.
    (i) The Same Object. I have already pointed out the ambiguity of the word object and it is very important here to bear in mind what has been said about it on pp. 7 et seq. What then do we mean by two people perceiving the same object in the sense that the direct object of their perceptions is the same and not merely that something that might be inferred from them is the same? What the present argument wants is that A and B should be obliged to attribute incompatible qualities to the same reality if they suppose that the objects of their perceptions are real. Since realities cannot have incompatible qualities it will then follow that the object of the perception of one or both of them must be merely an appearance. This however seems to land us in the old dilemma that, if the objects of their perceptions are the same, there is no difficulty; and, if they differ, then by supposing them to be realities we shall merely have two different realities with qualities which, if they came together in the same object would be incompatible -- a fact which is not relevant to existing circumstances. This difficulty may be met however by supposing that x and y the two objects are as wholes incompatible with each other and yet as objects of A's and B's perceptions respectively occupy the same space at the same time. Thus the important point is to determine when A and B hold that the objects that they perceive are in the same place. If they do this, and the objects perceived have incompatible qualities, it will follow that the two objects or at any rate one of them must be appearance, whether they be identical or not.
A and B say that they perceive the same object when

if B takes up the same position as A has just had and also a like attitude to A's, what he perceives is indistinguishable from what A remembers that he perceived when he occupied this position; and
as A and B change their positions and attitudes what they perceive changes in general continuously with their changes of position and attitude; and
the objects perceived by both and the successive objects perceived by each are localised at the same position of space.
Assuming that A and B can be aware of these three sets of facts then they will say that they perceive the same object. By this they really mean in our phraseology that, at any given moment they perceive different objects, but that the differences are mere appearance, and either what is common is reality, or the appearances are data from which we can argue to the existence qualities of a reality which is not as such perceived in any position of either. But, as we have already said, the really vital point is that the successive objects should be localised at the same point of space and that A and B can know that their contemporary objects of perception which conform to the rules (a) and (b) are also localised at the same position. We have yet to discuss how this is possible. Assuming this to be possible we can see where the synthetic incompatibility comes in. Let us take the case of a tangible sphere. A and B, standing in various places and turning in various ways, will see portions of ellipsoidal surfaces of slightly different eccentricities and rather different distributions of colour. All the conditions that I mentioned above will be fulfilled in this case if they ever are fulfilled. Suppose A can inform B that he locates the centre of his ellipsoid at the same point as B does that of his. Then, if the ellipsoids that A and B perceive be both realities, there will be -- independently of whether they be perceived or not -- two ellipsoids of rather different eccentricities and distributions of colour having the same centre. Hence we should get the result that not only is neither of these ellipsoids tangible -- for all that can be felt is a sphere but also that each is invisible to one of the observers. This is of course not by itself fatal, since the sphere, which commonsense generally holds to be the reality, is not visible to anyone and never can be. But if we are going to introduce invisible reals at all we shall have to do it on a very lavish scale indeed. We must end by holding nothing less than that there are as many real ellipsoids as can be seen by A or B from any position and that each is visible from one position and one only (or, remembering that we only perceive parts of ellipsoids and counting parts of ellipsoids of the same eccentricity as the same, from one set of positions and one only). This argument is of course not conclusive. The real world might contain all this host of real objects which are intangible and only visible from certain definite positions. But this is not a view that commends itself to most people, and, if naïf realism involves it, they will prefer to leave naïf realism. There is not, so far as I can see, any real difficulty from synthetic incompatybility on the assumption that all these ellipsoids do coexist and have their centres at one point. It will involve the additional assumption that the presence of parts of the surface of one of them does not render the parts of the surface of others that are inside it invisible from the appropriate position. The argument then does not in the end rest on the alleged synthetic incompatibility but on the terrible complications of the naively realistic hypothesis. It therefore gives no disproof on naïf realism, but merely a very strong incentive to seek another and simpler explanation of the facts. That explanation and its consequences, which are far-reaching, is discussed in a later chapter on the Causal Theory of Perception.
    But before we leave I(c) we must discuss the remaining question:
    (ii) Communicability of Perceived Qualities. For we have assumed all along that A and B can know that the objects of their perceptions are like each other and that they are in the same place. We must now ask: How precisely is this possible? On the view of naïf realism or of phenomenalism there is of course no difficulty in two people being aware of the same object at the same moment. For it merely involves that X shall be at the same time a constituent of two awarenesses p_x and q_x, one of which is a part of the complex of related awarenesses that constitutes A's mind and the other of which is a part of the complex which is B's mind. But the present problem is different. It is: Granted that A and B can as a matter of fact be aware of the same object at the same moment, can A know that B is aware of the same object as himself? or, more generally; Can A know what are the qualities of the object of B's awareness? Of course we always assume that we can know facts of this kind with more or less accuracy, and our present task is to find the presuppositions of that opinion.
    In the first place the most usual way for us to find out what B perceives is for him to tell us. Assuming that he is not lying a comparison of what he says he perceives with what we know ourselves to perceive will tell us how like are the objects of our respective perceptions. But this simple method is a good deal more complicated than it seems. Suppose B says: 'What I perceive is red,' what information exactly does he convey? What I directly perceive is a mere noise. It is however of course a noise with meaning. When I make noises of that kind I mean to convey information as to what I perceive, and presumably B does also. The first thing then is to recognise that the noise that B made is the same sort of noise as I should have made had I wished B to understand that the object that I perceive has this indefinable quality called red. But why do I suppose that hearing that noise will make B think of the quality that I perceive ?
    Because B has been taught English; it will be said. But how was he taught English in this matter? Presumably his parents or guardians in his youth showed him a number of objects which they believed to agree in being red, and made the noise in question. The objects of course all agreed in other respects, e.g. in that of being extended; but they got over this difficulty by showing him a number of other things which agreed with the red ones in being extended and differed from them in colour, and by then making a different noise. Now what were the presuppositions of this method of teaching? In the first place his teachers thought that if B were put in an appropriate position he would perceive the same or nearly the same distinctions as they did. They thought that, when they perceived differences, he would not perceive identity, and, when they perceived identity that he would not perceive difference. If B had perceived no difference in objects in some of which his teachers perceived one quality (say green) and in others of which they perceived another quality (say red) B would use the words 'red' or 'green' indifferently for objects which they considered red and for those which they held to be green. If, on the other hand, his teachers saw no difference in the colours of objects in which be did see a difference, B would have argued: 'Evidently "red" is a general name for two different colours'; and he would think it odd that his teachers should have no different names for its two species.
    This might indeed land B in difficulties when he came to reflect. Suppose he perceives colour differences where his friends perceive none. Call the two colours that he perceives x and y, and suppose that his friends, seeing no difference between the two, call both x. Then, as we saw B will think that x is a general term covering both x and y. But then it is very likely that be will see nothing in common between x an y except that they are both colours. In that case the term x will stand in his mind for colour in general. But colour is an universal that extends to other things beside x and y, things that his friends can distinguish. It will then seem very strange to B that they only apply the name x to x and y and not to the other colours. This may lead to the very odd result that, precisely because B can distinguish more colours than his friends, he may be led to think that they can distinguish something that he cannot, viz. an universal that is common to x and y but not to all colours.
    Still, in general, B could find out and convey to his friends that he perceived a difference of colour where they perceived none. But he could give them no information whatever about the new colour which they cannot perceive except that it is a colour. Thus what seems to be communicable with regard to perceptions is whether one person can always find a difference and a likeness in what he perceives where the other perceives differences or likenesses. A man cannot be certain that the object of his perception is identical with what another person perceives at the same time; but he can be sure whether he can or cannot find a distinction within his object of perception when the other person finds one in his. Suppose for instance that an angel could see that when I use the word 'green' another man always perceives what I do when I use the word 'red' and vice versa. Still neither I nor the other man could ever find this out, for we should always agree in our use of the words 'red' and 'green.' We cannot then communicate the objects of our perceptions to each other, and what we mean by saying that A and B perceive the same object is that A can discover no distinction in the object that he perceives to which B cannot discover a corresponding one in his own object.
    I have used the words distinctions and agreements; and it will be well to consider them for a moment. When A and B agree in discovering an agreement in their respective objects of perception, we can never be sure that the agreements that they discover are identical with each other. Thus A and B may agree in discovering the universal 'colour,' i.e. they will always apply the term 'coloured' to the same things, but we cannot be at all sure that, when they see the things about which they agree in finding an universal the universals that they find are qualitatively identical with each other. Still it is perfectly unimportant whether they are identical or not so long as they have precisely the same extension. Moreover, with regard to each such universal there can be a further agreement which further determines it. For instance A and B may agree in discovering the universals 'red' and 'sweet,' and although we can never know that A's red and sweet as universals immediately analysable out of his objects of perception are the same as B's red and sweet which are analysed out of his, yet it will still be the case that A and B will agree in holding that the two universals differ from each other and from numerous others that they agree in the present sense in discovering. Hence the ultimate fact that the qualities that we perceive and the universals that we find in comparing them are incommunicable, which I suppose is what Kant meant by saying that 'intuitions without conceptions are blind' is no hindrance to knowledge.
    There remains one more question. How do we come to be able to communicate these differences and agreements? I show a man two surfaces, one red and the other green. Suppose he is colour-blind. Then he and I agree in recognising an universal, viz. colour, but I also recognise two species of it, and he does not. How do I find out that he agrees with me in discovering the universal and differs in seeing only one species of it? In the first instance I believe that he finds an universal because he tells me so, and I believe for the same reason that he does not find the difference. This however assumes that he and I agree in what we mean by agreement and difference. But it might be said: You grant that with a normal man it is perfectly possible that what he perceives when you both use the word green is what you perceive when you both use the word red and vice versa, and the difference can never be discovered. How can you be sure then that when you both use the word distinction he does not mean what you do when you both use the word identity? How do you know in fact that when the man says: 'The objects of my perception have something in common,' he does not mean to indicate the same sort of experience as you have when you perceive that red and sweet differ; or, on the other band, when he says 'I perceive no difference in colour between this and that,' that his experience is not precisely like yours when you experience both red and green? The answer is as follows: Let us take the colour-blind man and suppose that he is being taught to speak as a child. He is put into circumstances under which we perceive red and green and the two names are told to him. His teacher says: 'This (pointing to one) is green, that is red.' This being so, when his teacher says: 'Pick out the green one,' he can of course do so as long as he remembers which one had that name given to it. But suppose that we try the experiment again, and the things are quite alike in all respects except colour, then the only way to distinguish the one called green and the one called red is for the colour-blind man to remember whether it was the third or fourth thing from the left (say) to which the teacher pointed when he used the word. Hence if the articles be mixed up out of his sight he will be just as likely to choose a red one as a green one when asked to choose one of the latter. In fact he will be seen to hesitate hopelessly, trying and failing to remember any characteristic apart from the original position which marks those called green from those called red. His actions are thus precisely like those that we should make if someone showed us two red or two green things exactly alike in all respects and said: 'One is A and the other B,' and then shuffled them about while we were away, and, when we returned, said: 'Now pick out A.' Now we have assumed all along that the other man is a creature whose mind and body work in the main like ours; unless we had this general conviction we should not have been so silly as to try to communicate with him at all. Hence if, under certain circumstances in which we know that we perceive a distinction, we see that he acts in precisely the same way as we do when we do not perceive one, it is perfectly reasonable to suppose that the state of his mind is like that of ours when we do not perceive a distinction. Hence it is owing to the fact that we have reason to believe that, under certain test conditions, the presence or absence of the recognition of a distinction will vitally affect our outward actions in definite ways that we have very strong ground for holding that we can know that, when a man who has been taught to speak in the usual way says that he perceives a difference, something very similar is going on in his mind to what goes on in ours when we use the same phrase. It is the absence of any such effect on outward action in the case of the matter of our sensations that makes it impossible to decide whether we do really perceive the same object when we agree about our use of the names of such objects. If I perceive red, and, whenever I do so A always perceives something exactly like what I perceive when I use the word green and vice versa, I can never find this out by a consideration of A's actions and mine under test conditions. For self-knowledge tells me that I do not act in any two different ways when I perceive red and when I perceive green that can be directly connected with the sensuous peculiarities of the two objects of perception as such. And I must assume that the same is the case with A . Hence it is only when he calls red or green both what I call red and what I call green that I can find any difference and reasonably come to the conclusion that I perceive something that he cannot.
    Thus the assertion that two people perceive the same thing in the same place can never be shown to mean that there is a thing really at a certain place to which the minds of two people are simultaneously in direct perceptual relation. We have reason to believe that, as far as concerns the objects of visual perception, they are never even precisely alike in the sense which is communicable at the same time for two different people. All that the statement really means is that there are certain agreements and differences which people agree or disagree in discovering in their objects of perception. It means, to put it more accurately,

that when A stood in the position in his perceptual space in which he judges that B now stands and in the same attitude he discovered no distinctions in the object of his perception to which B does not now find corresponding ones in the object of his. And we saw that agreement or disagreement on this point is capable of being elicited with fair certainty on the assumption that B is a being whose mind works in substantially the same way as A 's.
That A judges that the object of his present perception is much like that which he perceived when he stood in the same position and attitude as that in which he now judges to be standing. And
that A found that as he moved from one position to another there was continuous variation of those qualities of his successive objects of perception that changed at all.
This is what is meant when A and B hold that they 'perceive the same object'; it is all that is communicable; and, as we have now seen, it is perfectly compatible with extreme qualitative difference between the objects of A 's and B's perceptions even when each judges that he is in the same position as has been vacated by the other.
    I think then that we may hold that this consideration of what is meant by two people 'perceiving the same object' and of one person 'perceiving the same object from different positions' is a serious stumbling block to naïf realism. Not only is it certain that the immediate objects of their perceptions at the same time are not the same and that the objects of the successive perception of each are not the same; but also a complete agreement between them in all that is communicable is not incompatible with extreme differences in the objects of their perception. This, whilst it prevents an argument from synthetic incompatibility from being valid, makes naïf realism so complicated and artificial that one is almost forced to the theory of a common cause of the perceptions of each person and to the degradation of the objects of most if not of all of these perceptions to the level of appearance. But this alternative needs a separate chapter.

    But there remain other alleged arguments against naïf realism which can be discussed independently of a causal view of perception, and to these we must now turn. We begin with the argument that I have called (III). This deals with the alleged disproof of naïf realism based on the happening of dreams and hallucinations. What we perceive in dreams has most of the qualities of what we perceive in waking-life. On the other hand it is not supposed to be real. Hence the naïf realist is told that this fact is enough to disprove his contention that the objects of perception exist as such whether they are perceived or not. Since to be real, for us, just means to be capable of existing independently of being perceived the question turns upon what justification there is for believing that the dream world as such only exists when it is perceived.
    It is often rejected from reality on the ground of internal incoherence. But I do not think that this position can be maintained. The inconsistency that is noticed in dreams is not internal inconsistency, but merely lack of coherence with what happens in waking-life, and this is not really relevant unless you assume that, for the dream-world to be real at all, it must be part of the waking-world. When the Lord Chancellor in 'Iolanthe' suffered from nightmare his dream seems to us to be wholly incoherent because, in it, he crossed the Channel from Harwich in a vehicle slightly larger than a bathing-machine which was boarded at Sloane Square and South Kensington stations by a party of friends to whom he supplied a collation consisting of cold meat and penny ices. Now it is perfectly true that this is a concatenation of circumstances which it is most unlikely for any Law Lord in waking-life to experience, but that does not make it intrinsically absurd. Hence, if he intended to convey that the Channel, the friends, and the stations are those which he respectively crosses, meets, and uses in daily life, we should be justified in thinking (even though a respect for the English legal system prevented us from saying) that he was mistaken. But this by itself is not any good ground for denying that, in another world which he experiences when asleep and which continues to exist when he wakes and ceases to perceive it, there are objects sufficiently like the Channel, his friends, and the stations on the Underground between Victoria and Glo'ster Road to be mistaken for them, which stand in such relations to each other as it would be unlikely or impossible for the things which resemble them in waking-life to stand in. The only argument against this suggestion is that it is unlikely that, within each of two sets of objects corresponding one to one by likeness so close as to lead to mistakes of one for the other, there should yet be two such very different sets of relations. And this is certainly not a strong argument taken by itself
    The next argument against the reality of dream-objects is from the discontinuity of dreams. Taking this objection merely in the form that there are intervals between our dreams in which we experience a quite different set of objects this would of course cut both ways. But the argument is not so weak as this. It says that when dreams interrupt the course of our waking-life we can easily pick up the thread of it again. What we experienced before we slept is continuous with what we experience when we wake once more. But what we dreamt before we wake is scarcely ever continuous with what we dream after we go to sleep again. This is of course in general true, though, in certain abnormal cases, it is not and the successive dreams form as continuous a life as the successive waking experiences. A good example of this is discussed by Prof. Flournoy in his book Des Indes au Planète Mars. But in taking normal dreaming we must be careful that the continuity in waking-life shall be judged from an analogous standpoint to the discontinuity of the dream-life. The continuity in waking-life is judged in waking-life. But the discontinuity of dream-life is also judged in waking-life, and this makes the comparison hardly fair. In a fair comparison, either I ought in my present dream to judge that its details are not continuous with those of the dream that I had last night, or else I ought to judge in a dream that what I perceived before I went to sleep is continuous with what I perceived at some other time when I was awake. As far as I am aware we never do make such judgments in dreams, and so the comparison never is quite in pari materiâ. The fact is that, when we wake after a dream, we do find a discontinuity, and this troubles us so that we have to consider closely for a little while before persuading ourselves that our dream experience really is discontinuous with the present one whilst the latter is continuous with that we experienced before we fell asleep. In the dream however we rarely do have this state of discomfort which makes us reflect. Hence it remains a question whether, if we did reflect in dreams, we might not find them continuous with past dreams. In this case there would not be real discontinuity, but merely a temporary abeyance of our critical and comparative powers. It is not a fact that in one dream we remember others and positively see that there are incompatibilities between them and the present one, but that we simply do not in general remember and compare dreams whilst in the dreaming state. In waking-life we do to some extent remember dreams, and, on comparison, find them much less connected than our interrupted waking experiences; but then our memory of dreams is notoriously very bad and fleeting, and so this does not really prove very much. Hence I think it is fair to conclude that the evidence for the discontinuity in character between our successive dreams is neither strictly comparable with nor nearly so strong as the evidence for the essential continuity of our successive waking experiences.
    Suppose however that the evidence were quite as good, I still do not think that it would furnish any valid ground when taken by itself for denying that the objects of our dream-experiences are as likely to be real as those of waking-life. If I go to sleep and my body be moved, without waking me, from my bedroom to a strange field there will be as much discontinuity between the objects perceived before going to sleep and after waking as there is between successive dreams. We should indeed be very puzzled at first, but we should not believe that this proved that the waking- world was unreal. We should consider what explanation our general knowledge of the waking-world rendered probable and should no doubt come to the right conclusion. In the dream, as I have argued, we do not have much memory, our critical faculties are dormant, and we feel no puzzlement that might set us to work in using them. But, supposing that there be a real world with which we become acquainted in sleep, there is no reason whatever why we should always come into cognitive relations with the same parts of it every time that we enter it at all. We do not know enough of its character to have learnt the general laws that may hold in it and, if known, might explain why at different times we do enter into cognitive relations with different parts of it, such as are the rules about the motion of our bodies in the waking-world. But then we never try to find those laws in our dreams, since the philosopher and the scientist rarely remain philosophic or scientific when they are asleep, and so the fact that we do not find them is no proof that they do not exist. If it were our custom to walk in our sleep instead of it being an exception we should find as much discontinuity between successive portions of our waking-life as we now do in our successive dreams; and if it were our custom to seek for the laws of the dream-world while we were in it we might be able to explain its discontinuities as comfortably as we could those in the waking-life of the somnambulist.
    I do not think then that the standard arguments from incoherence and discontinuity supply by themselves much reason for denying that the dream-world is real though different from that with which we come in contact in our waking-life.
    We pass then to other arguments. It is urged that what people perceive in waking-life under similar circumstances agrees pretty well as a rule, but that this is not so in dreams. To this argument there is more than one answer.
    (a) In hallucinations and visions produced in definite ways there is frequently substantial agreement. The experiences of most persons who have delirium tremens seem to be very similar, so too those produced by certain drugs like Indian hemp. To these agreements ought certainly to be added those of mystics. On the whole there is a remarkable agreement in essentials between the visions of different mystics, as may be gathered from reading a book like the Va rieties of Religious Experience. The differences come in, as might be expected, in the philosophical and theological interpretations of their experiences that various mystics offer. I understand that the agreement is still more substantial in the case of Indian mystics who undergo a definite and common form of initiation. Of course all the agreements can be explained, hypothetically at any rate, by a causal theory; but we are purposely leaving that aside in the present chapter. What I wish to point out here is that there is another possible explanation, viz. that all people who do certain things do come into perceptual contact with a real common world which those who do not do these things cannot perceive, and that it is a little hard to follow those people who hold it to be certain that the microscopic structure of nuclei is real because if students follow prescribed rules they will perceive it, and yet hold that the fact that the mystics have equally concordant experiences is not merely explained but explained away as far as the reality of the objects of the latter is concerned by pointing to the fact that they have undergone a common course of initiation. But I also quarrel with religious mysticism for a precisely similar inconsistency. How can it argue that the agreement of mystics who have had a common training tends to prove that they perceive a reality and deny that the agreement of the persons who have drunk themselves into delirium tremens points to the reality of pink rats? The mystic's God and the drunkard's pink rats must, at this stage, stand or fall together.
    (b) No doubt though, in normal sleep at any rate, there is great disagreement. The disagreement between sleepers and wakers seems to me to prove nothing; it is that between sleepers and sleepers that is important. But the difficulty is: How can you tell whether two sleepers are in the same relevant conditions? We know pretty well what conditions are relevant to what is perceived in waking-life, but there is no reason to suppose that they would be the same, with regard to the world which we experience in dreams. It is perfectly absurd to argue that, because two people in the same bed experience much the same objects when both are awake, they ought to do so when both are asleep if there be a real dream-world. To argue so would be as reasonable as to argue that, if two people happened to dream that they were in the Great Court at the same time, the fact that, when they woke up, they found themselves in their respective bedrooms throws doubt on the reality of the waking-world.
    I do not think then that agreement and disagreement taken by themselves are any good argument against the reality of the world of dreams, of delirium, and of mystical vision.
    ⁶ The sole remaining argument with which I am acquainted involves causality and must be deferred. This argument in its crude form would hold that for anything to be real it must be perceived by means of the appropriate organs. Now we see things in sleep with our eyes shut, and therefore it is argued that what we see in dreams cannot be real. Here I have only to indicate the following points.

The causal argument commonly ends by proving that there is no reason to suppose that what we do perceive by means of our organs is real. Hence it can hardly prove that the dream-world is less than that of waking-life. It can only prove that both are unreal and that the perception of both is presumably caused by something that is real.
It is a common premise of the causal argument that we perceive everything by means of organs of perception. This, as stated, is simply false in view of dream-perceptions, and so it is hardly possible to use both the argument from relativity to an organ and that from occurrence in dreams to disprove naïf realism, as Bradley quite happily does.
On the other hand the causal theory, here as in most places, is capable of accounting for the relevant facts. It accounts for the respective agreements of the drunkards, the microscopists, and the mystics. But what the common-sense and scientific supporters of it fail to see is that, primâ facie, at any rate, whilst it reduces God to the level of the pink rats as regards reality, it degrades the cell-nuclei to the same position. Now most scientists could bear with fortitude the reduction of the Deity to the level of the rodents; but they resent such an affront to really important things like nuclei.

    It remains now to discuss the argument
    (IV). This attempts to refute naïf realism by pointing to an alleged close approximation of objects of sense-perception to feelings, and the general intimate connexion between the two. This argument is evidently thought to be a strong one. It is used by Berkeley and not despised by Bradley. One of the many humours of philosophy is that it is also used by Locke to refute the solipsist, a person whom that philosopher treats very much de haut en bas. Whilst Philonous tells Hylas that a great heat is a great pain, and proceeds to argue from this premise that temperature is purely subjective, Locke orders the solipsist to put his hand in the fire -- a method of refutation that recalls the 'coxcombs'⁷ and Dr Johnson with their 'grinning' and kicking arguments for realism.
    I will quote Bradley lest we seem to be arguing with thin air. He says: '...The pleasant and disgusting which we boldly locate in the object, how can they be there? Is a thing delightful or sickening really and in itself? Am, even I the constant owner of these wandering adjectives?'⁸ Do we 'boldly locate' the pleasant and the disgusting in the object? In general we have seen that it is not of importance to the reality of a quality where we happen to locate it; it is not until we begin to locate incompatible qualities in the same place that difficulties arise. Now, if we locate these feelings at all, we certainly do not locate them at the same time in the same place, so that this cannot be an argument from synthetic incompatibility. Moreover, we saw in discussing sound that we often do talk as if we located qualities in what as a matter of fact we regard merely as the cause of the perception of them, although, when we come to consider the matter, we see that, if they are to be located at all, they must occupy a different place from that in which verbally we put them. I suppose then that the best way of stating the argument would be as follows: We do often say that objects are pleasant or disgusting, and yet we are sure that these are not real qualities. Ought not this to make us doubt whether other qualities that we attribute to objects and recognise by sense really belong to those objects?
    We must distinguish between the questions of whether a quality be real and whether it really belongs to the object to which it is ascribed. The first simply asks: Does the quality exist unchanged when no one perceives it? The second asks: No matter whether the quality be real or apparent, ought we to localise it where we do? Now the present argument assumes that we know that, with regard to feelings of pleasure and disgust, we can answer both questions in the negative. Pleasures and pains are not believed to exist when not felt, and they are not believed to be localised in external bodies when we come to reflect. And the argument is: If we can know this on reflection about pleasantness which we yet often treat as a real quality of objects, may we not suspect that, if we reflected more carefully, we should see that we make the same mistake about colours and spatial qualities?
    To this the right answer seems to be that the more carefully we reflect the less likely we are to confuse feelings like pleasure and pain with qualities like red and square. For, when we reflect, we see that red and square are objects of certain awarenesses but not qualities of any awareness; since we are persuaded that no awareness is red or square. On the other hand, when we reflect on pleasure and pain, we see that, whilst they may be the objects of the same sort of awarenesses as those of which red and square are objects, yet in general they exist in an entirely different way in which red and square never do exist. And that is as qualities of awarenesses which are themselves not awarenesses of pleasure and pain, but of objects like red and square. If we go too near a fire we may say that the fire is white-hot and the sensation painful; we never call a sensation white-hot or a fire painful unless, in the former case, we confuse a sensation as a mental event with the object that is cognised in it, and, in the latter, mean that the fire causes pain but does not feel it. But, although we have seen that we never really do mean that a material object is pleasant or painful and that we do seem to mean that pleasure and pain are qualities of the cognitions of those objects which we term pleasant or painful, can this latter view be maintained as it stands?
    Do we really mean that the feeling of pain is a quality of our sensation of heat in the same way as the red colour is a quality of the fire? The difficulty is that in the latter case we know pretty well by x being a quality of Y, viz. that x is located in the same place as the other qualities which together with it constitute Y⁹ . But when we call the feeling of pain a quality of a sensation we cannot mean anything like this, for there is no reason to suppose that a sensation is in space. And there is a further difficulty. I have tried to maintain that, in ordinary perception, that of which we are directly aware is full of distinctions which are only made explicit in what are at any rate rudimentary judgments. With regard to the more salient features these judgments are made at once and automatically, but, when there is little interest or the distinctions are hard to discover, the judgments are not in general made. Yet somehow all the distinctions that we can discover by paying attention to our object of perception must have been in it all along, and indeed they are felt in a vague way, whether we distinguish them and discover their relations or not. Now this vague feeling of distinctions and relations that are afterwards made explicit by judgments seems to me to be closely analogous to the way in which we become aware of pleasure and pain. But here arises a difficulty. Just as we can make judgments and find that the vaguely felt distinctions and relations become explicit as the qualities of objects perceived, so we can make our pleasures and pains the objects of cognitions. It would seem that by analogy they ought then to be clearly recognised as qualities of what was previously but vaguely felt, and that the evidence for the view that they are qualities of mental states should be that, on reflexion, they always do turn out to be such. But the trouble is that this is not always the case. Some seem to be as clearly recognised as qualities of parts of the body as any ordinary sense-qualities can be. Hence the evidence for supposing that pleasures and pains differ from sense-qualities in being qualities of our mental states breaks down. Bodily pains like toothache, when they pass out of the stage of mere feeling and come to be cognised, seem to be as clearly localised as red and temperature. And it is hard to see how they can be at once qualities of the body and of mental states. Nor can it be maintained that toothache is merely a quality of the perception of one's tooth which gets its spatial characteristics by a kind of subreption from those of the object of this perception. For we do not generally have any perception of the tooth at all until the pain begins. It is only then that we put our tongue to it, and it is the previous localisation of the pain that tells us where to put our tongue.
    The upshot of our discussion is that

no sense-qualities are qualities of mental events;
though some pains may be so it is difficult to tell precisely what such a statement means, and it is certain that this is not the case with bodily pains.
Some pains, like toothache, seem indistinguishable from sense-qualities, and it seems to be a strictly comparable use of terms to say: 'My tooth is painful' and to say 'My tooth is black and has a hole in it.'
    It is evident that only bodily pains and pleasures are sufficiently like sense-qualities for a belief in their non-existence when unperceived to be any strong argument by analogy for the unreality of sense-qualities. Hence the question turns on whether we are justified in believing that bodily pleasures and pains necessarily cease to exist when they cease to be felt.
    But even before we discuss this question we can see that Bradley's argument is greatly weakened. Pleasantness and disgustingness are not localised in objects, but only the power to produce pleasures and pains. And the pleasures and pains, if localised at all, are localised in our own bodies. It is quite useless to say that it is inconsistent to replace pleasantness by a cause and to keep redness as a quality of a foreign body. You cannot help perceiving red as extended and at a definite place in space, and the close analogy between bodily pleasures and pains and sense-qualities is not an analogy between pleasantness and painfulness and sense-qualities.
    However, it will be worth while to discuss whether bodily pleasures and pains be real or only exist when they are perceived. We might either

be immediately certain of this, or
be certain of it on grounds that depend on the differences between pleasures and pains and sense-qualities, or
on grounds that apply equally to both.
It is evident that if our certainty springs from (b) or (c) the argument based on it is open to the charge of being an ignoratio elenchi in the one case or a pleonasm in the other. If we only assure ourselves that toothache, e.g., ceases to exist when we cease to perceive it on the ground of the difference between toothache and sensible qualities, it is clear that no analogical argument from a premise so proved will give you any reason for supposing that sensible qualities are mere appearances. On the other hand, if it be the characteristics that toothache and sensible qualities have in common that make us think that toothache only exists when perceived, the characteristics of sensible qualities will equally dispose of their reality without any need to drag in the mention of bodily pains. Hence it can only be valid and useful to argue that sensible qualities are appearances on the ground that toothache only exists when we are aware of it, provided our belief in the latter proposition is founded on no grounds at all. Now, I quite grant that most people believe that pleasures and pains only exist when someone is aware of them. The question is: What, if any, are their grounds for that belief?
    The usual grounds that are given are relativity to an organ, and different tastes of different people at the same time and of the same people at different times. But, as we have seen, these arguments are also brought independently against the reality of sense-qualities, and therefore if they be valid of pleasure and pain and these be like enough to sense-qualities for an analogical argument from one to the other to hold, it would still be a ridiculously round-about way of degrading sense-qualities to the rank of appearances. To use the argument by analogy under such circumstances would merely be to use a weaker argument where a stronger one is available and has to be assumed.
    I fancy, however, that most people would say that it is immediately certain that a pain or pleasure cannot exist unfelt, that its whole essence is to be felt, and that no meaning can be attached to an unfelt pleasure or pain. Now, if this be really an immediate certainty, and if sense-qualities be really like enough to pleasures and pains to make an analogical argument from one to the other plausible, it would be a most extraordinary thing if this immediate certainty should only extend to pleasures and pains and not to sense-qualities. Yet it seems quite certain that it does not extend to sense-qualities, since it is always held to be a paradox that needs support by arguments when the latter are held to be mere appearances. But is this quite so certain as it seems at first sight? I think we shall find that it is not. It is indeed felt to be extremely paradoxical to deny the reality of figure and motion, and rather paradoxical to deny that of colour. But as we go lower down in the scale of the senses, by which I mean as we pass to those whose deliveries seem to be less and less capable of scientific elaboration, the paradox of denying the reality of their objects decreases rapidly. We seem to have little difficulty in thinking of an unperceived shape or motion or colour; but what exactly is an unperceived smell or taste? Would not common-sense think it almost as paradoxical to hold that a lump of sugar has a sweet taste when no one is tasting it as to hold that it has not a cubical figure when no one is seeing or touching it? No doubt the perception of a sweet taste is formally just as much analysable into content and object as that of a triangular surface; but, whereas common-sense is certain that in the latter case the object is also existentially separable from the content, it is pretty sure that in the former such independence is impossible. This example brings us in touch with an argument of Berkeley's which, though stated in terms of his erroneous theory about universals, is not essentially dependent on it. May not the belief in the possibility of the separate existence of objects apart from content be the result of a vicious abstraction? It is not a vicious abstraction (as Berkeley would probably have held) to recognise the two elements in any perception, but it is certainly arguable that to believe in the possibility of their separate existence is like believing in the possibility of a separate existence of colours and visible shapes because we can recognise the distinction between the colour and the shape of what we see. It is certainly important to notice two points that favour this view:

It is extremely difficult (if it be really possible at all) to become aware of the content element in the supposed complex; and
it is commonly believed that, whilst many objects can exist apart from contents, no contents can exist apart from objects.

    I think the difference in our attitude towards the reality of shapes and colours and of smells and tastes is expressed in a distinction in popular language. Consider the phrases: That patch looks red; that patch is red; that substance tastes sweet; that substance is sweet. The first two phrases have as a rule distinctly different shades of meaning. When I say 'that patch looks red' I generally convey a suspicion that red is not really the colour that the patch is. But to say that a thing tastes sweet would generally be regarded as equivalent to saying that it is sweet. This linguistic distinction is no doubt a mark of our belief that there are real colours and our doubt whether there are real tastes.
    Granted, then, that there is a difference of attitude, is there any justification for it? Berkeley's argument would be equally valid as it stands against the reality of shapes and motions as against that of colours, and we defer its treatment to a later chapter¹⁰ . But common-sense makes a distinction, and the question here is whether it has a right to do so. What it claims is that by direct inspection of smells and tastes it can see that they are not the kind of things that could exist unperceived. What, it asks, could a taste that no one tasted be? Such a position implies either that common-sense is in possession of some general principle by which it can assert that things with the qualities x cannot be real, whilst things with the qualities y can be, and that tastes have the qualities x and shapes the qualities y; or else that there are immediate judgments about various classes of objects which assert that the objects of one class are real and that those of another are not. The difference is of some importance. In the first case we should say: Tastes are not real because they have the qualities x, and shapes are real because they have the qualities y. In the second we should have to be content with: Tastes (etc.) are unreal and do as a matter of fact all have the qualities x; shapes (etc.) are real and do as a matter of fact have the qualities y. I think it will be admitted that if the second be the actual position of common-sense it is an unsatisfactory one, since it seems arbitrary and insusceptible of further argument or application.
    Let us consider the most important distinctions between tastes or smells and shapes or colours. They are the following:

shapes and colours are perceptible at points in visual space that differ from those where the percipient organ (the eye) is present. Tastes are only perceived on the tongue, whilst smells, in as far as they are localised directly at all, are placed vaguely at the nose. (The statement that a body has such and such a smell is admittedly an inference. It means that we smell the air in our nostrils and that we have reason to believe that its smell is due to odorous particles from a certain body -- or, as we put it, 'the smell comes from this body.')
Colours often have definite shapes, positions, and relations in visual space. Tastes and smells only have a very vague extension without definite shape. If tastes have any spatial relations they are not directly perceived, but reached by correlating them with points of tactual space on the tongue.
Tastes and smells never seem to appear as terms in any causal laws in the physical world. Tastes do not cause other tastes except in the sense that they may influence by contrast, whilst they are never causally connected, as far as we know, with motions or colours. On the other hand, most causal laws are in terms of shapes and motions and spatial relations reached by reflecting on visual and tactual percepts. Of course in this respect colours resemble tastes and smells rather than shapes.

    These seem to be the most important differences between the two kinds of objects. Do they provide any ground for supposing that one kind is more likely to be real than the other? I am inclined to think that they provide at least a good excuse if not a very strong reason for the difference of attitude taken by common-sense. Common-sense says that it finds it almost impossible to conceive of an unsmelled smell or an untasted taste, but finds little difficulty in an unseen colour and none in an unseen figure. And the cause of the difficulty seems to me to be that a taste or a smell has so few qualities and stands in so few relations that it approaches that pure being which Hegelians tell us differs in no way from nothing. There are hardly any propositions that can be asserted about tastes and smells. They are of course determinate things with a unique nature which makes a smell a smell and a taste a taste. But apart from this sensuous uniqueness there is hardly any thing that can be said about them. They can be very roughly localised in the perceptual spaces of other senses, they are vaguely extended, they can be compared in a slight measure in intensity, and they have ill-defined qualitative differences comparable (though on an indefinitely lower scale) with colour differences. Now, when you try to conceive what an object that you perceive would be if you no longer perceived it you can only do this by thinking of propositions which might still be true about it. And we see that there are hardly any such propositions that can be at all definitely formulated. You cannot say much more than that a smell will still be the smell that it was when you smelled it; but you cannot describe this characteristic to yourself or other people in any general terms.
    When, as in shape and motion, and even to a less degree in colour, you have a mass of relations and qualities giving rise to numberless propositions you can conceive where you no longer perceive. For although the qualities and relations were given in perception and still have the sensuous peculiarities which make, e.g., before and after in time something different from before and after on a straight line, still they have general logical characteristics that can be grasped by thought without further help from sense.
    But although this seems to me to account for the difficulty felt by common-sense in conceiving an unsmelled smell or an untasted taste, I do not think that it provides any argument against the possibility of their being real. As smelled and tasted they have a definite nature and exist, and the fact that they have few qualities and stand in few relations, whilst it makes it difficult for us to conceive their unperceived existence, in no way disproves it. There is no reason why whatever is real should stand in many relations and have many qualities, and much in fact may be real which we cannot intellectually describe.
    But however this may be there is one additional characteristic that even bodily pains have in distinction from ordinary sense-data, and that is their relation to volition. All pains as such we desire to remove. We may of course will to endure a pain because we see that its removal would be incompatible with the existence of something that we think very valuable, or would entail a greater pain later. I think, then, we might say that a bodily pain is a sense-quality located in our bodies which we directly desire to remove. This certainly covers all bodily pains; the only question is whether it may not be too wide.
    It might be said that, if one were deformed, that would be a sense- quality localised in the body that anyone would desire to remove, and yet it would not be a pain. This is true; but then it is not a sense-quality that we directly desire to remove like extremely high temperature of the skin. We desire to remove deformity because we think it ugly, or because we feel that it makes us repulsive to others, or because it prevents us doing many things that we should like to do. We do not desire to remove it directly and for its own sake without reference to anything else. But when I consider any bodily sensation which I call painful I always find a definite quality located in my body and a desire to remove it for its own sake, a desire on which of course I need not, on reflexion, act. I am not persuaded that there is any more in bodily pains than their being sense-qualities located in the body and standing in this peculiar relation to volition. But this is not important for us. The point is that a necessary condition for a bodily pain is that there should be this relation to the will.
    Now I suspect that when a man says that he is immediately certain that an unfelt bodily pain cannot exist, what he means is that he is certain that this relation to the will would not exist unless he were aware of the sense-quality. And here I agree with him. An unfelt toothache would not be an ache, because it does not become one till the will is affected, and what the mind does not perceive the will does not trouble about. But this does not seem to me to imply that the sensible quality that is located in the tooth when we do feel toothache cannot exist unperceived. it would, then, merely not affect the will, and so not be a pain.
    Now consider Berkeley's argument about the increase of temperature ending in pain. There is no sudden discontinuity in what we feel as the temperature rises; yet at first most people would say that the skin was really warm and would continue to be so even though we did not feel it, whilst, later on, when it becomes painful, most people would say that the pain could not exist except when felt. Even if Berkeley's solution were right there is a paradox in the change in the person's conviction as to the independent existence of what he perceives as the temperature rises. Berkeley's explanation does not account for this, and I think that ours does. As the temperature rises we still continue to have a sensation whose object is a temperature continuous with what was experienced before, though of greater intensive magnitude. But when this intensive magnitude increases beyond a certain point it begins to affect the will, and we desire its removal directly and for its own sake with constantly growing vehemence. Now the subject is certain that nothing can affect his will of which he is not aware, and therefore he says that the pain could not exist if it were not perceived, because temperature does not become a pain except by affecting his will in this way. But where he and Berkeley are wrong is in supposing that this proves or renders it probable that temperature as such cannot exist unperceived, contrary to the plain man's belief when the temperature was lower. All that he has a right to conclude is that the temperature could not be painful unless it were perceived, not that it might not exist unperceived.
    Thus the argument from bodily pleasures and pains seems to me to have no validity taken by itself. And it is only such feelings that resemble sense-qualities enough to make an analogical argument from them to the case of sense-qualities plausible. I therefore conclude that this argument by itself will not do what it claims. It is largely confused with arguments against naïf realism based on relativity to an organ and disagreement among different subjects. And these can only be fully discussed when we deal with the causal theory of perception.
    We have now discussed all the important arguments against naïf realism that do not depend on causality. They have commonly been accepted by philosophers without criticism and handed down from classical scepticism to Descartes, and from Descartes and Locke to modern thought. We have found that some are entirely futile; that none make naïf realism absolutely impossible; but that to maintain it in face of the discrepancies between sight and touch and of the visual perceptions of the same person in different places and of different people at the same time would be to assume an immensely complicated world of largely imperceptible reals for which there is no evidence except the doctrine of naïf realism itself. These complications are swept away by a causal view; but such a view has its own troubles, and these have yet to be discussed. But before doing this we must digress for a chapter and consider the Law of Causation itself. It is vital to science in general and to the causal view of perception in particular, and it has been severely handled by philosophers. To it we must now turn.

Notes
¹ The only paradox here is that it is the hand whose past treatment might have been expected to have made it the hotter which is actually the colder and vice versa. But nothing is proved by this paradox.
² I use the terms quantity, magnitude, and kind, in the same sense as Mr Russell in his Principles of Mathematics.
³ I need scarcely mention that a finite volume and a definite shape are perfectly compatible with the absence of outermost points.
⁴ Of course other meanings of contact are possible. If bodies have no outermost points we might say that A and B touch at P when every point on one side of P is red and belongs to A and every point on the other side of P is blue and belongs to B, while P has no colour. I merely want to illustrate that common-sense is not clear as to precisely what it means and that an alleged à priori certainty is of little value under such circumstances.
⁵ The superiority of touch after the Causal Theory of Perception has been assumed is further discussed in Chap. IV. p. 253 of et seq.
⁶ Cf. Chap. IV. p. 271 et seq. where this question is finally discussed.
⁷ 'And coxcombs vanquish Berkeley with a grin.'
⁸ Appearance and Reality (Second Edition), p. 12.
⁹ Cf. Chap. II. p. 91 et seq. where this view is more fully discussed.
¹⁰ Chapter III.
Contents -- Chapter 2

terms
objects (in wider sense)		non-objects
appearances	realities		non-realities