The analysis of matter

IN this Part, the subject with which we are concerned is the evidence for the truth of physics—not of this or that special result in physics, but of the general structure of the science. It is to be expected that the evidence will not be such as to give certainty, but at best such as to give probability; it is to be expected, also, that this probability may be increased by a suitable interpretation of physics, where "interpretation" is understood in the sense considered in Chapter I. We shall find it desirable to divide our problem into several parts, each of which will have an importance not confined to physics. There is need, first, to be clear as to what we mean by an empirical science, and what is the degree of certainty to be expected of it at the best. There is need to discuss what can be meant by "data," and to distinguish inferences, theories and hypotheses. We shall then discuss the causal theory of perception, and at the same time the philosophy called "phenomenalism." From these topics we shall pass to general discussion, first of cause, then of substance. This will lead us to the epistemological grounds for interpreting physics in accordance with neutral monism, and to the paramount importance of structure in scientific inference. We shall conclude with a definition of perception considered as affording the empirical data for physics, and with the consideration of phenomena analogous to perception in the non-mental world. But first of all it will be well to examine the historical development by means of which our problem has assumed its present form—both the pre-scientific development leading to common sense, and the scientific development leading from common sense to physics.

Common sense consists of a set of beliefs, or at least habits, which work well in practice except in situations which rarely occur. A savage may be puzzled by a box containing an unseen gyroscope, or by rails carrying an electric current; common sense has not prepared him for oddities of this sort. But a little familiarity enables a man to fit them into his common-sense world, and a mechanic soon learns their ways if he has occasion to do so. This illustrates the fact that there is no sharp line between science and common sense: both involve expectations, but those resulting from science are more accurate. It is possible to pursue science practically without any fundamental change from the metaphysic of common sense. But when theoretical science is taken seriously, it is found to involve a quite changed metaphysic, whose relation to that of common sense demands investigation. This will form the topic of the next chapter; in the present chapter, I shall consider the genesis of common sense, not in the race, since that is undiscoverable, but in the individual.

In studying infants, as in studying animals, we are compelled to confine ourselves to behaviouristic methods, whatever our views may be on the subject of behaviourism as a general principle in psychology. We can observe the bodily acts of young infants, but they cannot tell us their thoughts. At a low mental level, however, it is hardly profitable to distinguish between a belief and a habit of action. Beliefs, in the psychological sense, seem to emerge out of previously existing habits, and to be, at first, little more than verbal representations of habits formed before words could be uttered. There is therefore no great loss in being confined to behaviouristic methods when we are considering infants before the age at which they can speak.

It is of course obvious and generally recognized that very young infants do not possess the common-sense notion of an "object." This is by no means obvious with the young of some other kinds of animals—with chickens, for example. They possess, as instincts, useful ways of behaviour which in the human young are only learnt by experience; for example, they can pick up a grain which they see on the ground. The human infant has no such innate skill; for several months, it makes no attempt to touch what it sees. The "hand-eye co-ordination" comes as a result of experience. Some native aptitudes, of course, a new-born child does possess; for example, it can turn its eyes towards a bright light, though not very quickly or accurately. It has a reflex connected with sucking, but not a very intelligent one; indeed, it hardly amounts to more than the practice of trying to suck anything that comes in contact with the lips. Even in this respect, the human infant is inferior to the young of other mammals. We can say that certain stimuli rouse certain reflexes, but these are only just sufficient to keep the infant alive with the help of maternal care.

In this primitive condition, the infant obviously has no conception of an "object." An "object," for common sense, is something having a certain degree of permanence, and connected with several kinds of sensation. This involves something like memory, to give rise to the idea of permanence, or rather, at first, to the feeling of recognition; and it involves experience, to give to one sensory stimulus a reaction originally associated with another. In infants, the most important factor in forming the common-sense notion of an object is the hand-eye co-ordination, the discovery that it is possible, often, to grasp what is seen. In this way, visual and tactual spaces become correlated, which is one of the most important steps in the mental growth of an infant.

At this point, it is important to be clear as to the difference between "space" in psychology and "space" in physics. There is undoubtedly a connection between the two, which it will be part of our business to make clear at a later stage. But the connection is very round-about and inferential. At the outset, it is much more useful to realize the difference between them than the connection, since much confusion of thought arises from supposing the connection to be closer than it is. In physics there is only one space, while in psychology there are several for each individual; these can, it is true, be reduced by manipulation to one for each individual, but they cannot be reduced further without introducing obscurities that it is impossible to dissipate. The space containing my visual objects has no point in common with the space containing yours, since no visual object in my world is precisely identical with one in yours. And the amalgamation of the spaces of my different senses into one space is a piece of early science, performed by the infant at about the age of three months. Dr Whitehead, who is anxious to bridge the gulf between perception and physics, seems to me to make his task too easy where space is concerned. For example, he says:[35]

This passage is so pleasant that I hate to criticize it. But I do not know how else to make clear where I differ from Dr Whitehead. There is first a purely verbal question to be cleared up. Dr Whitehead says it is an error to deduce space from the relations between figures. It is certainly an error to deduce physical space in this way, but with psychological space the matter is different. There certainly are perceived relations between figures, and these perceived relations are part of our perceptual data in physics. Whether they are to be said to constitute a space or not, is a verbal question. Psychologists, as a rule, find it convenient to say so; but the matter is unimportant. When this question has been cleared away, however, there remain others which are vital to an understanding of the relation between physics and perception.

Take, first, the question of the two times. As will appear when we come to the causal theory of perception, the whole of my perceptual world is, from the standpoint of physics, in my head; any two events which I experience together overlap in physical space, and all of them together, in physical space, occupy a volume smaller than my head, since it certainly does not include the hair, skull, teeth, etc. Consequently, on relativity principles, there is no question of two times, since this only arises for events which are spatially separated in physical space.

As for the necessity of distinguishing tactual and visual space: there are perceived relations between objects seen simultaneously, and also between objects touched simultaneously, and these relations are part of the crude material out of which we construct our notion of space. These relations cannot hold between a visual and a tactual percept. But there are other relations which do hold—namely, those of correlation: when I see my hand in contact with a visual object I feel it in contact with a tactual object, and moreover the visual and the tactual object have certain relations to each other—e.g. where we see a corner we get a tactual sensation of sharpness. All this, however, is learnt by experience; that is to say, we learn the laws of the correlation by experience. The infant can be seen learning them. One may call these laws "pre-established harmonies," but they are no more so than any other scientific laws. Unless we are going to say that all laws of nature must be demonstrable by pure logic, which is hardly conceivable nowadays, we must admit that there are co-existences and sequences which we expect on a basis of past experience, in spite of the fact that their failure would not be logically impossible. And the correlation of visual and tactual sensations is a case of this sort.

It is sometimes suggested, in such cases, that the correlated occurrences are merely different manifestations of one and the same entity. This is, in fact, the view of common sense, which holds that it can both see and touch the same object. I have no objection whatever to this way of speaking, and I do not deny that, rightly interpreted, it may express a correct view. But it remains nevertheless true that the entity said to be manifested is inferred from experience of a correlation, and that the percepts correlated are not logically interconnected, but only empirically. We have , a visual percept, and at the same time , a tactual percept. Each rouses appropriate reflexes, and, owing to their frequently occurring together, it happens in time that each rouses also the reflexes appropriate to the other. This practical induction occurs before the child has reflected that the two are correlated; indeed, unless he becomes a learned man he probably never realizes the correlation of and . But as soon as we reflect upon the matter we can see that there is no necessary correlation. It fails with blind men, and with men whose fingers have been anæsthetized. In general, however, the correlation holds good. Common sense explains it by regarding both touch and sight as ways of getting to know an object which is at once tangible and visible. In the language of the causal theory of perception, we say that and have a common cause, in general external to the body. I do not wish to deny this, but only to point out that, when we are considering the grounds of our knowledge, we cannot say that we know of the correlation because we know of the common external cause. The order in knowledge is the opposite: we have evidence for the correlation in our experience, and we infer[36] the common cause from the correlation, so that the common cause cannot have more certainty than the correlation, which is its premiss. From a behaviouristic point of view, the infant "knows" the correlation when either stimulus calls out the response originally appropriate to the other.

We must here guard against a small possible misunderstanding. If and are invariably correlated, it may be said, it is impossible that one should occur without the other, and therefore there can be no means of judging whether one alone would elicit the response belonging to the other. In fact, the matter is not quite so simple as we have been taking it to be. What we learn by infantile experience is not that and are always correlated; it is possible to touch in the dark, or with the eyes shut, and it is possible to see without touching. What we learn is that the correlation can be brought about easily in many cases. Movements of the eye will usually give a visual sensation corresponding to a previously uncorrelated tactual sensation, and movements of the hand (or other part of the body) will, in a certain proportion of cases, give a tactual sensation corresponding to a previously uncorrelated visual sensation. Children practising the hand-eye co-ordination attempt to grasp objects not within their reach; it is only gradually that distance comes to be judged more or less correctly. When objects are not within our grasp, a new correlation comes into play—namely, between the visual sensation and the journey required to bring the object within our reach. Unfamiliar circumstances will cause even adults to make mistakes—for example, that of underestimating the depth of objects under water. Great distances remain permanently beyond the scope of common sense: only science can assure us that the sun is farther off than the moon.

What we can observe the infant learning is the bodily acts which will, in fact, reinforce a percept of one sense by a percept of another; more particularly he learns to touch what he sees—i.e. to procure for himself a correlated pair , , instead of the isolated . Similarly he learns to look round when he hears a voice, and so on. All this implies that he has, so far as action is concerned, the notion of a physical object, as something capable of affecting several senses simultaneously. The element of recognition is logically separable, and arises somewhat earlier.

These motor habits are essential in generating common-sense beliefs, which arise at a much later stage of mental growth. Common sense, in its more primitive form, is hardly aware that there is such an occurrence as perceiving; it is only aware of the perceived object. And by the time that even the most rudimentary reflection begins, each sense calls out responses connected with other senses, so that even when, from the standpoint of external stimulus, only one sense is affected, the experience has the massiveness of something in which several senses are involved. See, for example, the pictures in Kohler's Mentality of Apes: here we see chimpanzees which are watching others with sympathetic movements of the arms that indicate stimulation of bodily feelings connected with balance, although the sole stimulus is visual. This accounts for the fact that common sense can so confidently identify an object touched and not seen with an object seen but not touched—e.g. the cricket-ball now successfully caught and the same ball as it flew through the air. The reason is that the experience is always richer than the sensory stimulus alone would warrant: it contains always responses arising from physiological experience of past correlations. If an adult were to hear a donkey's bray for the first time, without having previously known that there was an animal which made that noise, his experience would be amazingly unlike that of a normal adult in the same circumstances.

Common sense does not initially distinguish as sharply as civilized nations do between persons, animals, and things. Primitive religion affords abundant evidence of this. A thing, like an animal, has a sort of power residing within it: it may fall on your head, roll over in the wind, and so on. It is only gradually that inanimate objects become sharply separated from people, through the observation that their actions have no purpose. But animals are not separable from people on this ground, and are in fact thought by savages to be much more intelligent than they are.

Common sense is, in most respects, naively realistic: it believes that, as a rule, our perceptions show us objects as they really are. It is able to hold this view because of the mass of experience which, in each individual, precedes the common-sense outlook. We do not think a distant person smaller than a person near at hand; we do not judge circular objects seen sideways to be elliptic; and so on. All this is, for common sense, part of the perception; it may be doubted whether it is not so also for psychology. But it is certainly not part of the infant's initial perceptive apparatus: it is something which the infant has to learn. Some of it is learnt after the beginnings of speech have been acquired—particularly a right judgment as to the size of distant objects. But at any rate by the time a child is three years old he has acquired the common-sense outlook. That is to say, his immediate reaction to a sensory stimulus involves a great deal of previous experience, and is such as to enable him to arrive, without any mental process, at a far more objective view of what he perceives than was possible at birth. I mean here by "objective" not anything metaphysical, but merely "agreeing with the testimony of others." It would be a complete mistake to suppose that, in an adult, there is first an experience corresponding to the bare sensory stimulus, and then an inference to that of which it is a sign. This may occur in certain cases, for example, if we watch a man drawing a face in an apparently haphazard manner, and do not realize till the last moment that a face is being intended. But such an experience is quite unlike normal perception, where the "inference," in the only sense in which it can be said to exist, is physiological, or at any rate not discoverable by introspection. It is because the sensory stimulus is able to lead us, without any mental intermediary, to an object practically identical with that perceived by others in our neighbourhood, that we are able to adopt the common-sense belief that we actually perceive external objects.

The notion of cause is part of the apparatus of common sense. I do not think it would be true to say that common sense regards objects as the causes of our perceptions; it would not, unless challenged, think of bringing in causation in this connection. It looks for causes when it is surprised, not when an occurrence seems perfectly natural. It demands causes for a mirage, a reflexion, a dream, an earthquake, a plague, and so on, but not for the ordinary course of nature. And the cause which it looks for, wherever the event concerned has great emotional interest, is pretty sure to be animistic: the anger of the gods, or something analogous. The idea of universal causation, and of causation divorced from purpose, belongs to a later stage of mental development, and marks the beginnings of philosophy and science.

Substance is a category which comes naturally to common sense, though without the attribute of indestructibility added by the metaphysicians—but as to this perhaps diverse opinions are possible. One would be inclined to suppose that common sense regards fire as destroying what it burns; but the Chinese, when they had made a solemn covenant, used to burn it, in order that the gods might take cognisance of it through the smoke. (A copy was kept for terrestrial purposes.) And races that practise cremation do not, as a rule, suppose that they are totally destroying the body. On the other hand, there has existed a religious prejudice against cremation which implied the belief that the body was thereby totally annihilated. I think one must conclude, therefore, that the attitude of common sense as to the indestructibility of substance is vacillating; on the whole, the success of physics in providing immortal material units represents a triumph of the philosopher over the plain man.

Substance, whether indestructible or not, is of great importance in primitive thought, and dominates syntax, through which it has dominated philosophy down to our own day. At a primitive stage, there is no distinction between "substance" and "thing"; both express, first in language and then in thought, the emotion of recognition. To an infant, recognition is a very strong emotion, particularly when connected with something agreeable or disagreeable. When the infant begins to use words, it applies the same word to percepts on two occasions, if the second rouses the emotion of recognition associated with memory of the first, or perhaps merely with the word which was learnt in presence of the first. (When I say that the infant uses the "same" word, I mean that he makes closely similar noises.) Using a given word as a response to stimuli of a certain kind is a motor habit, like reaching for the bottle. Two percepts to which the same word applies are thought to be identical, unless both can be present at once; this characteristic distinguishes general names from proper names. The basis of this whole process is the emotion of recognition. When the process, as a learning of motor habits, is complete, and reflection upon it begins, identity of name is taken to indicate identity of substance—in one sense in the case of proper names, in another sense in the case of names applicable to two or more simultaneous percepts—i.e. general names (Platonic ideas, universals). Throughout, language comes first and thought follows in its footsteps. And language is governed largely by physiological causation.

A substance or thing is supposed to be identical at different times, although its properties may change. John Jones is the same person throughout his life, although he grows from childhood to manhood, is sometimes pleased and sometimes cross, sometimes awake and sometimes asleep. Primarily, he is considered to be the same person because he has the same name. But the name, like the person, is not exactly the same on different occasions; it may be spoken loud or soft, quickly or slowly. These differences, however, are too slight to prevent recognition, except on rare occasions—e.g. when the name is pronounced very badly by a foreigner; one of the merits of names is that they change less than the person named.

The conception of substantial identity with varying properties is embedded in language, in common sense, and in metaphysic. To my mind, it is useful in practice, but harmful in theory. It is harmful, I mean, if taken as metaphysically ultimate: what appears as one substance with changing states should, I maintain, be conceived as a series of occurrences linked together in some important way. I will not yet argue this view. It would have been utterly foreign to physics until the substitution of space-time for space and time, with the corresponding substitution of a four-dimensional continuum of events for the older conception of persistent material units moving in a three-dimensional space. But the older conception still appears the natural one to apply to electrons and protons, so that physics may be said to have, at the moment, two different points of view on this issue. For the present, I am not concerned to criticize the notion of substance, but only to show its genesis, which I take to be derived from the pre-human emotion which we reflectively call "recognition," though it has not, originally, the definite cognitive character attached to the word when applied to the mental processes of an adult human being.

Induction, like substance, plays a large part in common sense, and has a basis which is primarily physiological. I am not at present discussing the validity of induction, but the cause of the practice of induction among animals, children, and savages. Of course the validity of induction is really assumed in such a discussion, since, without it, causes cannot be discovered. But we do not assume the validity of the primitive inductions which we are discussing; we assume only that there is some valid form of induction. Throughout genetic psychology we assume the validity of ordinary scientific procedure. If this assumption were to lead us to views on genetic psychology which threw doubt on the validity of scientific procedure, that would constitute a reductio ad absurdum, which would destroy genetic psychology along with the rest. Therefore, whenever some obviously invalid process is said to be the psychological source of a method essential to science, we must suppose, unless we are to embrace complete scepticism, that there is some valid process which, in most of the cases to which the invalid process is applied by unscientific people, gives rather similar results. All this has perhaps only a pragmatic justification, but whether this is the case cannot be decided ab initio. The real utility of investigating crude primitive forms of inference is that the contrast between them and current scientific inference may suggest directions in which the latter is capable of still further improvement. The direct logical importance of investigations into the origins of our mental processes is nil, but the importance as a means of stimulating imagination in the formation of hypotheses may be considerable. It is for this reason that the topics of the present chapter form a useful introduction to those which form our proper subject-matter.

The source of induction, speaking historically, is the general law of what Dr J. B. Watson calls "learned reactions." In its schematic simplicity, this law is as follows: If a stimulus to a living body of an animal produces a reaction , and a stimulus produces a reaction , then if and are applied together, there is a tendency for alone, afterwards, to produce as well as . E.g. if you expose a person frequently to a certain loud noise and a bright light simultaneously, after a while the loud noise alone will cause his pupils to contract. It is obvious that the practice of induction is simply the application of this law to cognitive reactions. If you have frequently heard the words "there's Jones" when you could see Jones, these words will in the end cause you to believe that Jones is present even if, for the moment, you do not see him. This form of induction is involved in understanding speech. It is obvious that, in its cruder forms, induction may give rise to false beliefs as well as to true ones; scientific methodology has to seek a form of induction which shall make false inferences much rarer than true ones. If such a form can be found, a man may train himself, in his professional activity, to abstain from the more primitive forms. But as an ordinary mortal he could not survive for a day if he refused to trust to what we may call physiological induction, which stores up in the body the lessons of past experience. In practice, a nearly instantaneous method of inference which is right nine times out of ten is preferable to a slow method which is always right. A man who subjected all his food to chemical analysis before eating it would avoid being poisoned, but would also fail to be adequately nourished.

Throughout the development of theory, great intellectual changes have been repeatedly necessitated by errors which were very small from the standpoint of practice. The theory of relativity is a remarkable instance of this: an immense reconstruction has been made to meet discrepancies which could only be detected by the most delicate measurements. The further science advances, the more minute become the facts which it cannot yet assimilate. Common sense does well enough for most of the needs of a pre-industrial community, but not for the construction of a dynamo or a wireless station. For these, we have to advance to the standpoint of pre-relativity physics. Machines involving relativity physics do not yet exist, but presumably they will some day. This, however, is beside the point. The point is, that a small discrepancy between theory and observation may indicate a large error in theory. Take, e.g., naive realism and the velocity of light, the latter from a pre-relativity point of view. The supposition of common sense and naive realism, that we see the actual physical object, is very hard to reconcile with the scientific view that our perception occurs somewhat later than the emission of light by the object; and this difficulty is not overcome by the fact that the time involved, like the notorious baby, is a very little one. We cannot therefore argue from the practical success of common sense to its approximate theoretical accuracy, but only to a certain rough correspondence between its commoner inferences and those permitted by a correct theory. If physics has had to desert common sense, that is no reason for finding fault with physics.