{"collection":"philosophy-ethics","full_text_available":true,"graph_url":"https://unchartedknowledge.5gfusion.net/v2/api/page/gutenberg_72981/graph","html_url":"https://unchartedknowledge.5gfusion.net/v2/page/gutenberg_72981","id":"gutenberg_72981","license":"Public Domain (US)","rights_basis":"gutenberg_metadata_copyright_false","source":"gutenberg","source_url":"https://www.gutenberg.org/ebooks/72981","summary":null,"tags":["Category: Philosophy & Ethics","Nobel Prizes in Literature","Philosophy","Physics -- Philosophy","Psychology"],"text":"Transcriber\u2019s Note: Italics are enclosed in _underscores_. Additional\nnotes will be found near the end of this ebook.\n\n\n\n\nPHILOSOPHY\n\n\n\n\n                               PHILOSOPHY\n\n                                  _By_\n                            Bertrand Russell\n\n\n                             [Illustration]\n\n\n                                NEW YORK\n                     W \u00b7 W \u00b7 NORTON & COMPANY, INC.\n                              _Publishers_\n\n\n\n\n                            Copyright, 1927,\n                            BERTRAND RUSSELL\n\n Published in Great Britain under the title \u201cAn Outline of Philosophy\u201d\n\n\n                PRINTED IN THE UNITED STATES OF AMERICA\n                FOR THE PUBLISHERS BY THE VAN REES PRESS\n\n\n\n\nCONTENTS\n\n\n  CHAPTER                                                           PAGE\n      I.  PHILOSOPHIC DOUBTS                                           1\n\n\n  PART I\n\n  MAN FROM WITHOUT\n\n     II.  MAN AND HIS ENVIRONMENT                                     16\n\n    III.  THE PROCESS OF LEARNING IN ANIMALS AND INFANTS              29\n\n     IV.  LANGUAGE                                                    43\n\n      V.  PERCEPTION OBJECTIVELY REGARDED                             58\n\n     VI.  MEMORY OBJECTIVELY REGARDED                                 70\n\n    VII.  INFERENCE AS A HABIT                                        79\n\n   VIII.  KNOWLEDGE BEHAVIOURISTICALLY CONSIDERED                     88\n\n\n  PART II\n\n  THE PHYSICAL WORLD\n\n     IX.  THE STRUCTURE OF THE ATOM                                   97\n\n      X.  RELATIVITY                                                 107\n\n     XI.  CAUSAL LAWS IN PHYSICS                                     114\n\n    XII.  PHYSICS AND PERCEPTION                                     123\n\n   XIII.  PHYSICAL AND PERCEPTUAL SPACE                              137\n\n    XIV.  PERCEPTION AND PHYSICAL CAUSAL LAWS                        144\n\n     XV.  THE NATURE OF OUR KNOWLEDGE OF PHYSICS                     151\n\n\n  PART III\n\n  MAN FROM WITHIN\n\n    XVI.  SELF-OBSERVATION                                           161\n\n   XVII.  IMAGES                                                     176\n\n  XVIII.  IMAGINATION AND MEMORY                                     187\n\n    XIX.  THE INTROSPECTIVE ANALYSIS OF PERCEPTION                   201\n\n     XX.  CONSCIOUSNESS?                                             210\n\n    XXI.  EMOTION, DESIRE, AND WILL                                  218\n\n   XXII.  ETHICS                                                     225\n\n\n  PART IV\n\n  THE UNIVERSE\n\n  XXIII.  SOME GREAT PHILOSOPHIES OF THE PAST                        236\n\n   XXIV.  TRUTH AND FALSEHOOD                                        254\n\n    XXV.  THE VALIDITY OF INFERENCE                                  266\n\n   XXVI.  EVENTS, MATTER, AND MIND                                   276\n\n  XXVII.  MAN\u2019S PLACE IN THE UNIVERSE                                292\n\n\n\n\nPHILOSOPHY\n\n\n\n\nCHAPTER I\n\nPHILOSOPHIC DOUBTS\n\n\nPerhaps it might be expected that I should begin with a definition\nof \u201cphilosophy\u201d, but, rightly or wrongly, I do not propose to do so.\nThe definition of \u201cphilosophy\u201d will vary according to the philosophy\nwe adopt; all that we can say to begin with is that there are certain\nproblems, which certain people find interesting, and which do not, at\nleast at present, belong to any of the special sciences. These problems\nare all such as to raise doubts concerning what commonly passes for\nknowledge; and if the doubts are to be answered, it can only be by\nmeans of a special study, to which we give the name \u201cphilosophy\u201d.\nTherefore the first step in defining \u201cphilosophy\u201d is the indication of\nthese problems and doubts, which is also the first step in the actual\nstudy of philosophy. There are some among the traditional problems of\nphilosophy that do not seem to me to lend themselves to intellectual\ntreatment, because they transcend our cognitive powers; such problems I\nshall not deal with. There are others, however, as to which, even if a\nfinal solution is not possible at present, yet much can be done to show\nthe direction in which a solution is to be sought, and the _kind_ of\nsolution that may in time prove possible.\n\nPhilosophy arises from an unusually obstinate attempt to arrive at real\nknowledge. What passes for knowledge in ordinary life suffers from\nthree defects: it is cocksure, vague, and self-contradictory. The first\nstep towards philosophy consists in becoming aware of these defects,\nnot in order to rest content with a lazy scepticism, but in order to\nsubstitute an amended kind of knowledge which shall be tentative,\nprecise, and self-consistent. There is of course another quality\nwhich we wish our knowledge to possess, namely comprehensiveness: we\nwish the area of our knowledge to be as wide as possible. But this\nis the business of science rather than of philosophy. A man does\nnot necessarily become a better philosopher through knowing more\nscientific facts; it is principles and methods and general conceptions\nthat he should learn from science if philosophy is what interests\nhim. The philosopher\u2019s work is, so to speak, at the second remove\nfrom crude fact. Science tries to collect facts into bundles by means\nof scientific laws; these laws, rather than the original facts, are\nthe raw material of philosophy. Philosophy involves a criticism of\nscientific knowledge, not from a point of view ultimately different\nfrom that of science, but from a point of view less concerned with\ndetails and more concerned with the harmony of the whole body of\nspecial sciences.\n\nThe special sciences have all grown up by the use of notions derived\nfrom common sense, such as things and their qualities, space, time, and\ncausation. Science itself has shown that none of these common-sense\nnotions will quite serve for the explanation of the world; but it\nis hardly the province of any special science to undertake the\nnecessary reconstruction of fundamentals. This must be the business\nof philosophy. I want to say, to begin with, that I believe it to be\na business of very great importance. I believe that the philosophical\nerrors in common-sense beliefs not only produce confusion in science,\nbut also do harm in ethics and politics, in social institutions, and\nin the conduct of everyday life. It will be no part of my business, in\nthis volume, to point out these practical effects of a bad philosophy:\nmy business will be purely intellectual. But if I am right, the\nintellectual adventures which lie before us have effects in many\ndirections which seem, at first sight, quite remote from our theme. The\neffect of our passions upon our beliefs forms a favourite subject of\nmodern psychologists; but the converse effect, that of our beliefs upon\nour passions, also exists, though it is not such as an old-fashioned\nintellectualist psychology would have supposed. Although I shall not\ndiscuss it, we shall do well to bear it in mind, in order to realise\nthat our discussions may have bearings upon matters lying outside the\nsphere of pure intellect.\n\nI mentioned a moment ago three defects in common beliefs, namely, that\nthey are cocksure, vague, and self-contradictory. It is the business of\nphilosophy to correct these defects so far as it can, without throwing\nover knowledge altogether. To be a good philosopher, a man must have a\nstrong desire to know, combined with great caution in believing that\nhe knows; he must also have logical acumen and the habit of exact\nthinking. All these, of course, are a matter of degree. Vagueness,\nin particular, belongs, in some degree, to all human thinking; we\ncan diminish it indefinitely, but we can never abolish it wholly.\nPhilosophy, accordingly, is a continuing activity, not something\nin which we can achieve final perfection once for all. In this\nrespect, philosophy has suffered from its association with theology.\nTheological dogmas are fixed, and are regarded by the orthodox as\nincapable of improvement. Philosophers have too often tried to produce\nsimilarly final systems: they have not been content with the gradual\napproximations that satisfied men of science. In this they seem to me\nto have been mistaken. Philosophy should be piecemeal and provisional\nlike science; final truth belongs to heaven, not to this world.\n\nThe three defects which I have mentioned are interconnected, and by\nbecoming aware of any one we may be led to recognise the other two. I\nwill illustrate all three by a few examples.\n\nLet us take first the belief in common objects, such as tables and\nchairs and trees. We all feel quite sure about these in ordinary\nlife, and yet our reasons for confidence are really very inadequate.\nNaive common sense supposes that they are what they appear to be, but\nthat is impossible, since they do not appear exactly alike to any two\nsimultaneous observers; at least, it is impossible if the object is\na single thing, the same for all observers. If we are going to admit\nthat the object is not what we see, we can no longer feel the same\nassurance that there is an object; this is the first intrusion of\ndoubt. However, we shall speedily recover from this set-back, and say\nthat of course the object is \u201creally\u201d what physics says it is.[1] Now\nphysics says that a table or a chair is \u201creally\u201d an incredibly vast\nsystem of electrons and protons in rapid motion, with empty space in\nbetween. This is all very well. But the physicist, like the ordinary\nman, is dependent upon his senses for the existence of the physical\nworld. If you go up to him solemnly and say, \u201cwould you be so kind as\nto tell me, as a physicist, what a chair really is\u201d, you will get a\nlearned answer. But if you say, without preamble: \u201cIs there a chair\nthere?\u201d he will say: \u201cOf course there is; can\u2019t you see it?\u201d To this\nyou ought to reply in the negative. You ought to say, \u201cNo, I see\ncertain patches of colour, but I don\u2019t see any electrons or protons,\nand you tell me that they are what a chair consists of\u201d. He may reply:\n\u201cYes, but a large number of electrons and protons close together look\nlike a patch of colour\u201d. What do you mean by \u201clook like\u201d? you will\nthen ask. He is ready with an answer. He means that light-waves start\nfrom the electrons and protons (or, more probably, are reflected by\nthem from a source of light), reach the eye, have a series of effects\nupon the rods and cones, the optic nerve, and the brain, and finally\nproduce a sensation. But he has never seen an eye or an optic nerve or\na brain, any more than he has seen a chair; he has only seen patches of\ncolour which, he says, are what eyes \u201clook like.\u201d That is to say, he\nthinks that the sensation you have when (as you think) you see a chair,\nhas a series of causes, physical and psychological, but all of them,\non his own showing, lie essentially and forever outside experience.\nNevertheless, he pretends to base his science upon observation.\nObviously there is here a problem for the logician, a problem belonging\nnot to physics, but to quite another kind of study. This is a first\nexample of the way in which the pursuit of precision destroys certainty.\n\n      [1] I am not thinking here of the elementary physics to be\n          found in a school text-book; I am thinking of modern\n          theoretical physics, more particularly as regards the\n          structure of atoms, as to which I shall have more to say\n          in later chapters.\n\nThe physicist believes that he infers his electrons and protons from\nwhat he perceives. But the inference is never clearly set forth in\na logical chain, and, if it were, it might not look sufficiently\nplausible to warrant much confidence. In actual fact, the whole\ndevelopment from common-sense objects to electrons and protons has\nbeen governed by certain beliefs, seldom conscious, but existing in\nevery natural man. These beliefs are not unalterable, but they grow\nand develop like a tree. We start by thinking that a chair is as it\nappears to be, and is still there when we are not looking. But we find,\nby a little reflection, that these two beliefs are incompatible. If\nthe chair is to persist independently of being seen by us, it must\nbe something other than the patch of colour we see, because this is\nfound to depend upon conditions extraneous to the chair, such as how\nthe light falls, whether we are wearing blue spectacles, and so on.\nThis forces the man of science to regard the \u201creal\u201d chair as the cause\n(or an indispensable part of the cause) of our sensations when we see\nthe chair. Thus we are committed to causation as an _a priori_ belief\nwithout which we should have no reason for supposing that there is a\n\u201creal\u201d chair at all. Also, for the sake of permanence we bring in the\nnotion of substance: the \u201creal\u201d chair is a substance, or collection of\nsubstances, possessed of permanence and the power to cause sensations.\nThis metaphysical belief has operated, more or less unconsciously, in\nthe inference from sensations to electrons and protons. The philosopher\nmust drag such beliefs into the light of day, and see whether they\nstill survive. Often it will be found that they die on exposure.\n\nLet us now take up another point. The evidence for a physical law, or\nfor any scientific law, always involves both memory and testimony. We\nhave to rely both upon what we remember to have observed on former\noccasions, and on what others say they have observed. In the very\nbeginnings of science, it may have been possible sometimes to dispense\nwith testimony; but very soon every scientific investigation began\nto be built upon previously ascertained results, and thus to depend\nupon what others had recorded. In fact, without the corroboration of\ntestimony we should hardly have had much confidence in the existence\nof physical objects. Sometimes people suffer from hallucinations, that\nis to say, they think they perceive physical objects, but are not\nconfirmed in this belief by the testimony of others. In such cases,\nwe decide that they are mistaken. It is the similarity between the\nperceptions of different people in similar situations that makes us\nfeel confident of the external causation of our perceptions; but for\nthis, whatever naive beliefs we might have had in physical objects\nwould have been dissipated long ago. Thus memory and testimony are\nessential to science. Nevertheless, each of these is open to criticism\nby the sceptic. Even if we succeed, more or less, in meeting his\ncriticism, we shall, if we are rational, be left with a less complete\nconfidence in our original beliefs than we had before. Once more, we\nshall become less cocksure as we become more accurate.\n\nBoth memory and testimony lead us into the sphere of psychology. I\nshall not at this stage discuss either beyond the point at which it is\nclear that there are genuine philosophical problems to be solved. I\nshall begin with memory.\n\nMemory is a word which has a variety of meanings. The kind that I am\nconcerned with at the moment is the recollection of past occurrences.\nThis is so notoriously fallible that every experimenter makes a record\nof the result of his experiment at the earliest possible moment: he\nconsiders the inference from written words to past events less likely\nto be mistaken than the direct beliefs which constitute memory. But\nsome time, though perhaps only a few seconds, must elapse between\nthe observation and the making of the record, unless the record is\nso fragmentary that memory is needed to interpret it. Thus we do not\nescape from the need of trusting memory to some degree. Moreover,\nwithout memory we should not think of interpreting records as applying\nto the past, because we should not know that there was any past. Now,\napart from arguments as to the proved fallibility of memory, there is\none awkward consideration which the sceptic may urge. Remembering,\nwhich occurs now, cannot possibly--he may say--prove that what is\nremembered occurred at some other time, because the world might have\nsprung into being five minutes ago, exactly as it then was, full of\nacts of remembering which were entirely misleading. Opponents of\nDarwin, such as Edmund Gosse\u2019s father, urged a very similar argument\nagainst evolution. The world, they said, was created in 4004 B.C.,\ncomplete with fossils, which were inserted to try our faith. The world\nwas created suddenly, but was made such as it would have been if it\nhad evolved. There is no logical impossibility about this view. And\nsimilarly there is no logical impossibility in the view that the world\nwas created five minutes ago, complete with memories and records. This\nmay seem an improbable hypothesis, but it is not logically refutable.\n\nApart from this argument, which may be thought fantastic, there are\nreasons of detail for being more or less distrustful of memory. It\nis obvious that no _direct_ confirmation of a belief about a past\noccurrence is possible, because we cannot make the past recur. We can\nfind confirmation of an indirect kind in the revelations of others and\nin contemporary records. The latter, as we have seen, involve some\ndegree of memory, but they may involve very little, for instance when\na shorthand report of a conversation or speech has been made at the\ntime. But even then, we do not escape wholly from the need of memory\nextending over a longer stretch of time. Suppose a wholly imaginary\nconversation were produced for some criminal purpose, we should depend\nupon the memories of witnesses to establish its fictitious character in\na law-court. And all memory which extends over a long period of time is\nvery apt to be mistaken; this is shown by the errors invariably found\nin autobiographies. Any man who comes across letters which he wrote\nmany years ago can verify the manner in which his memory has falsified\npast events. For these reasons, the fact that we cannot free ourselves\nfrom dependence upon memory in building up knowledge is, _prima facie_,\na reason for regarding what passes for knowledge as not quite certain.\nThe whole of this subject of memory will be considered more carefully\nin later chapters.\n\nTestimony raises even more awkward problems. What makes them so awkward\nis the fact that testimony is involved in building up our knowledge\nof physics, and that, conversely, physics is required in establishing\nthe trustworthiness of testimony. Moreover, testimony raises all the\nproblems connected with the relation of mind and matter. Some eminent\nphilosophers, _e.g._ Leibniz, have constructed systems according to\nwhich there would be no such thing as testimony, and yet have accepted\nas true many things which cannot be known without it. I do not think\nphilosophy has quite done justice to this problem, but a few words\nwill, I think, show its gravity.\n\nFor our purposes, we may define testimony as noises heard, or shapes\nseen, analogous to those which we should make if we wished to convey\nan assertion, and believed by the hearer or seer to be due to someone\nelse\u2019s desire to convey an assertion. Let us take a concrete instance:\nI ask a policeman the way, and he says, \u201cFourth turn to the right,\nthird to the left.\u201d That is to say, I hear these sounds, and perhaps I\nsee what I interpret as his lips moving. I assume that he has a mind\nmore or less like my own, and has uttered these sounds with the same\nintention as I should have had if I had uttered them, namely to convey\ninformation. In ordinary life, all this is not, in any proper sense,\nan inference; it is a belief which arises in us on the appropriate\noccasion. But if we are challenged, we have to substitute inference for\nspontaneous belief, and the more the inference is examined the more\nshaky it looks.\n\nThe inference that has to be made has two steps, one physical and one\npsychological. The physical inference is of the sort we considered a\nmoment ago, in which we pass from a sensation to a physical occurrence.\nWe hear noises, and think they proceed from the policeman\u2019s body. We\nsee moving shapes, and interpret them as physical motions of his\nlips. This inference, as we saw earlier, is in part justified by\ntestimony; yet now we find that it has to be made before we can have\nreason to believe that there is any such thing as testimony. And this\ninference is certainly sometimes mistaken. Lunatics hear voices which\nother people do not hear; instead of crediting them with abnormally\nacute hearing, we lock them up. But if we sometimes hear sentences\nwhich have not proceeded from a body, why should this not always be\nthe case? Perhaps our imagination has conjured up all the things that\nwe think others have said to us. But this is part of the general\nproblem of inferring physical objects from sensations, which, difficult\nas it is, is not the most difficult part of the logical puzzles\nconcerning testimony. The most difficult part is the inference from\nthe policeman\u2019s body to his mind. I do not mean any special insult to\npolicemen; I would say the same of politicians and even of philosophers.\n\nThe inference to the policeman\u2019s mind certainly _may_ be wrong. It is\nclear that a maker of wax-works could make a life-like policeman and\nput a gramophone inside him, which would cause him periodically to\ntell visitors the way to the most interesting part of the exhibition\nat the entrance to which he would stand. They would have just the sort\nof evidence of his being alive that is found convincing in the case of\nother policemen. Descartes believed that animals have no minds, but are\nmerely complicated automata. Eighteenth-century materialists extended\nthis doctrine to men. But I am not now concerned with materialism; my\nproblem is a different one. Even a materialist must admit that, when he\ntalks, he means to convey something, that is to say, he uses words as\nsigns, not as mere noises. It may be difficult to decide exactly what\nis meant by this statement, but it is clear that it means something,\nand that it is true of one\u2019s own remarks. The question is: Are we sure\nthat it is true of the remarks we hear, as well as of those we make?\nOr are the remarks we hear perhaps just like other noises, merely\nmeaningless disturbances of the air? The chief argument against this\nis analogy: the remarks we hear are so like those we make that we think\nthey must have similar causes. But although we cannot dispense with\nanalogy as a form of inference, it is by no means demonstrative, and\nnot infrequently leads us astray. We are therefore left, once more,\nwith a _prima facie_ reason for uncertainty and doubt.\n\nThis question of what we mean ourselves when we speak brings me to\nanother problem, that of introspection. Many philosophers have held\nthat introspection gave the most indubitable of all knowledge; others\nhave held that there is no such thing as introspection. Descartes,\nafter trying to doubt everything, arrived at \u201cI think, therefore I\nam\u201d, as a basis for the rest of knowledge. Dr. John B. Watson the\nbehaviourist holds, on the contrary, that we do not think, but only\ntalk. Dr. Watson, in real life, gives as much evidence of thinking as\nanyone does, so if _he_ is not convinced that he thinks, we are all\nin a bad way. At any rate, the mere existence of such an opinion as\nhis, on the part of a competent philosopher, must suffice to show that\nintrospection is not so certain as some people have thought. But let us\nexamine this question a little more closely.\n\nThe difference between introspection and what we call perception of\nexternal objects seems to me to be connected, not with what is primary\nin our knowledge, but with what is inferred. We think, at one time,\nthat we are seeing a chair; at another, that we are thinking about\nphilosophy. The first we call perception of an external object; the\nsecond we call introspection. Now we have already found reason to doubt\nexternal perception, in the full-blooded sense in which common-sense\naccepts it. I shall consider later what there is that is indubitable\nand primitive in perception; for the moment, I shall anticipate by\nsaying that what is indubitable in \u201cseeing a chair\u201d is the occurrence\nof a certain pattern of colours. But this occurrence, we shall find,\nis connected with me just as much as with the chair; no one except\nmyself can see exactly the pattern that I see. There is thus something\nsubjective and private about what we take to be external perception,\nbut this is concealed by precarious extensions into the physical world.\nI think introspection, on the contrary, involves precarious extensions\ninto the mental world: shorn of these, it is not very different from\nexternal perception shorn of its extensions. To make this clear, I\nshall try to show what we know to be occurring when, as we say, we\nthink about philosophy.\n\nSuppose, as the result of introspection, you arrive at a belief which\nyou express in the words: \u201cI am now believing that mind is different\nfrom matter\u201d. What do you know, apart from inferences, in such a case?\nFirst of all, you must cut out the word \u201cI\u201d: the person who believes\nis an inference, not part of what you know immediately. In the second\nplace, you must be careful about the word \u201cbelieving\u201d: I am not now\nconcerned with what this word should mean in logic or theory of\nknowledge; I am concerned with what it can mean when used to describe\na direct experience. In such a case, it would seem that it can only\ndescribe a certain kind of feeling. And as for the proposition you\nthink you are believing, namely, \u201cmind is different from matter\u201d,\nit is very difficult to say what is really occurring when you think\nyou believe it. It may be mere words, pronounced, visualised, or in\nauditory or motor images. It may be images of what the words \u201cmean\u201d,\nbut in that case it will not be at all an accurate representation of\nthe logical content of the proposition. You may have an image of a\nstatue of Newton \u201cvoyaging through strange seas of thought alone\u201d, and\nanother image of a stone rolling downhill, combined with the words \u201chow\ndifferent!\u201d Or you may think of the difference between composing a\nlecture and eating your dinner. It is only when you come to expressing\nyour thought in words that you approach logical precision.\n\nBoth in introspection and in external perception, we try to express\nwhat we know in WORDS.\n\nWe come here, as in the question of testimony, upon the social aspect\nof knowledge. The purpose of words is to give the same kind of\npublicity to thought as is claimed for physical objects. A number of\npeople can hear a spoken word or see a written word, because each is a\nphysical occurrence. If I say to you, \u201cmind is different from matter\u201d,\nthere may be only a very slight resemblance between the thought that I\nam trying to express and the thought which is aroused in you, but these\ntwo thoughts have just this in common, that they can be expressed by\nthe same words. Similarly, there may be great differences between what\nyou and I see when, as we say, we look at the same chair; nevertheless\nwe can both express our perceptions by the same words.\n\nA thought and a perception are thus not so very different in their own\nnature. If physics is true, they are different in their correlations:\nwhen I see a chair, others have more or less similar perceptions, and\nit is thought that these are all connected with light-waves coming from\nthe chair, whereas, when I think a thought, others may not be thinking\nanything similar. But this applies also to feeling a toothache, which\nwould not usually be regarded as a case of introspection. On the whole,\ntherefore, there seems no reason to regard introspection as a different\n_kind_ of knowledge from external perception. But this whole question\nwill concern us again at a later stage.\n\nAs for the _trustworthiness_ of introspection, there is again a\ncomplete parallelism with the case of external perception. The actual\ndatum, in each case, is unimpeachable, but the extensions which we make\ninstinctively are questionable. Instead of saying, \u201cI am believing\nthat mind is different from matter\u201d, you ought to say, \u201ccertain images\nare occurring in a certain relation to each other, accompanied by a\ncertain feeling\u201d. No words exist for describing the actual occurrence\nin all its particularity; all words, even proper names, are general,\nwith the possible exception of \u201cthis\u201d, which is ambiguous. When you\ntranslate the occurrence into words, you are making generalisations and\ninferences, just as you are when you say \u201cthere is a chair\u201d. There is\nreally no vital difference between the two cases. In each case, what is\nreally a datum is unutterable, and what can be put into words involves\ninferences which may be mistaken.\n\nWhen I say that \u201cinferences\u201d are involved, I am saying something not\nquite accurate unless carefully interpreted. In \u201cseeing a chair\u201d,\nfor instance, we do not first apprehend a coloured pattern, and then\nproceed to infer a chair: belief in the chair arises spontaneously\nwhen we see the coloured pattern. But this belief has causes not only\nin the present physical stimulus, but also partly in past experience,\npartly in reflexes. In animals, reflexes play a very large part; in\nhuman beings, experience is more important. The infant learns slowly to\ncorrelate touch and sight, and to expect others to see what he sees.\nThe habits which are thus formed are essential to our adult notion\nof an object such as a chair. The perception of a chair by means of\nsight has a physical stimulus which affects only sight directly, but\nstimulates ideas of solidity and so on through early experience. The\ninference might be called \u201cphysiological\u201d. An inference of this sort is\nevidence of past correlations, for instance between touch and sight,\nbut may be mistaken in the present instance; you may, for instance,\nmistake a reflection in a large mirror for another room. Similarly in\ndreams we make mistaken physiological inferences. We cannot therefore\nfeel certainty in regard to things which are in this sense inferred,\nbecause, when we try to accept as many of them as possible, we are\nnevertheless compelled to reject some for the sake of self-consistency.\n\nWe arrived a moment ago at what we called \u201cphysiological inference\u201d\nas an essential ingredient in the common-sense notion of a physical\nobject. Physiological inference, in its simplest form, means this:\ngiven a stimulus S, to which, by a reflex, we react by a bodily\nmovement R, and a stimulus S\u2032 with a reaction R\u2032, if the two stimuli\nare frequently experienced together, S will in time produce R\u2032.[2]\nThat is to say, the body will act as if S\u2032 were present. Physiological\ninference is important in theory of knowledge, and I shall have much\nto say about it at a later stage. For the present, I have mentioned it\npartly to prevent it from being confused with logical inference, and\npartly in order to introduce the problem of _induction_, about which we\nmust say a few preliminary words at this stage.\n\n      [2] _E.g._ if you hear a sharp noise and see a bright light\n          simultaneously often, in time the noise without the light\n          will cause your pupils to contract.\n\nInduction raises perhaps the most difficult problem in the whole theory\nof knowledge. Every scientific law is established by its means, and yet\nit is difficult to see why we should believe it to be a valid logical\nprocess. Induction, in its bare essence, consists of the argument that,\nbecause A and B have been often found together and never found apart,\ntherefore, when A is found again, B will probably also be found. This\nexists first as a \u201cphysiological inference\u201d, and as such is practised\nby animals. When we first begin to reflect, we find ourselves making\ninductions in the physiological sense, for instance, expecting the\nfood we see to have a certain kind of taste. Often we only become\naware of this expectation through having it disappointed, for instance\nif we take salt thinking it is sugar. When mankind took to science,\nthey tried to formulate logical principles justifying this kind of\ninference. I shall discuss these attempts in later chapters; for the\npresent, I will only say that they seem to me very unsuccessful. I\nam convinced that induction must have validity of some kind in some\ndegree, but the problem of showing how or why it can be valid remains\nunsolved. Until it is solved, the rational man will doubt whether his\nfood will nourish him, and whether the sun will rise tomorrow. I am not\na rational man in this sense, but for the moment I shall pretend to be.\nAnd even if we cannot be completely rational, we should probably all\nbe the better for becoming somewhat more rational than we are. At the\nlowest estimate, it will be an interesting adventure to see whither\nreason will lead us.\n\nThe problems we have been raising are none of them new, but they\nsuffice to show that our everyday views of the world and of our\nrelations to it are unsatisfactory. We have been asking whether we\nknow this or that, but we have not yet asked what \u201cknowing\u201d is. Perhaps\nwe shall find that we have had wrong ideas as to knowing, and that our\ndifficulties grow less when we have more correct ideas on this point. I\nthink we shall do well to begin our philosophical journey by an attempt\nto understand knowing considered as part of the relation of man to his\nenvironment, forgetting, for the moment, the fundamental doubts with\nwhich we have been concerned. Perhaps modern science may enable us to\nsee philosophical problems in a new light. In that hope, let us examine\nthe relation of man to his environment with a view to arriving at a\nscientific view as to what constitutes knowledge.\n\n\n\n\nPART I\n\nMAN FROM WITHOUT\n\n\n\n\nCHAPTER II\n\nMAN AND HIS ENVIRONMENT\n\n\nIf our scientific knowledge were full and complete, we should\nunderstand ourselves and the world and our relation to the world. As\nit is, our understanding of all three is fragmentary. For the present,\nit is the third question, that of our relation to the world, that I\nwish to consider, because this brings us nearest to the problems of\nphilosophy. We shall find that it will lead us back to the other two\nquestions, as to the world and as to ourselves, but that we shall\nunderstand both these better if we have considered first how the world\nacts upon us and how we act upon the world.\n\nThere are a number of sciences which deal with Man. We may deal with\nhim in natural history, as one among the animals, having a certain\nplace in evolution, and related to other animals in ascertainable\nways. We may deal with him in physiology, as a structure capable of\nperforming certain functions, and reacting to the environment in ways\nof which some, at least, can be explained by chemistry. We may study\nhim in sociology, as a unit in various organisms, such as the family\nand the state. And we may study him, in psychology, as he appears to\nhimself. This last gives what we may call an internal view of man, as\nopposed to the other three, which give an external view. That is to\nsay, in psychology we use data which can only be obtained when the\nobserver and the observed are the same person, whereas in the other\nways of studying Man all our data can be obtained by observing other\npeople. There are different ways of interpreting this distinction,\nand different views of its importance, but there can be no doubt\nthat there is such a distinction. We can remember our own dreams,\nwhereas we cannot know the dreams of others unless they tell us about\nthem. We know when we have toothache, when our food tastes too salty,\nwhen we are remembering some past occurrence, and so on. All these\nevents in our lives other people cannot know in the same direct way.\nIn this sense, we all have an inner life, open to our own inspection\nbut to no one else\u2019s. This is no doubt the source of the traditional\ndistinction of mind and body: the body was supposed to be that part of\nus which others could observe, and the mind that part which was private\nto ourselves. The importance of the distinction has been called in\nquestion in recent times, and I do not myself believe that it has any\nfundamental philosophical significance. But historically it has played\na dominant part in determining the conceptions from which men set out\nwhen they began to philosophise, and on this account, if on no other,\nit deserves to be borne in mind.\n\nKnowledge, traditionally, has been viewed from within, as something\nwhich we observe in ourselves rather than as something which we can see\nothers displaying. When I say that it has been so viewed, I mean that\nthis has been the practice of philosophers; in ordinary life, people\nhave been more objective. In ordinary life, knowledge is something\nwhich can be tested by examinations, that is to say, it consists in a\ncertain kind of response to a certain kind of stimulus. This objective\nway of viewing knowledge is, to my mind, much more fruitful than the\nway which has been customary in philosophy. I mean that, if we wish\nto give a definition of \u201cknowing\u201d, we ought to define it as a manner\nof reacting to the environment, not as involving something (a \u201cstate\nof mind\u201d) which only the person who has the knowledge can observe.\nIt is because I hold this view that I think it best to begin with\nMan and his environment, rather than with those matters in which the\nobserver and the observed must be the same person. Knowing, as I view\nit, is a characteristic which may be displayed in our reactions to our\nenvironment; it is therefore necessary first of all to consider the\nnature of these reactions as they appear in science.\n\nLet us take some everyday situation. Suppose you are watching a race,\nand at the appropriate moment you say, \u201cthey\u2019re off\u201d. This exclamation\nis a reaction to the environment, and is taken to show knowledge if it\nis made at the same time as others make it. Now let us consider what\nhas been really happening, according to science. The complication of\nwhat has happened is almost incredible. It may conveniently be divided\ninto four stages: first, what happened in the outside world between the\nrunners and your eyes; secondly, what happened in your body from your\neyes to your brain; thirdly, what happened in your brain; fourthly,\nwhat happened in your body from your brain to the movements of your\nthroat and tongue which constituted your exclamation. Of these four\nstages, the first belongs to physics, and is dealt with in the main\nby the theory of light; the second and fourth belong to physiology;\nthe third, though it should theoretically also belong to physiology,\nbelongs in fact rather to psychology, owing to our lack of knowledge\nas to the brain. The third stage embodies the results of experience\nand learning. It is responsible for the fact that you speak, which an\nanimal would not do, and that you speak English, which a Frenchman\nwould not do. This immensely complicated occurrence is, nevertheless,\nabout the simplest example of knowledge that could possibly be given.\n\nFor the moment, let us leave on one side the part of this process which\nhappens in the outside world and belongs to physics. I shall have much\nto say about it later, but what has to be said is not altogether easy,\nand we will take less abstruse matters first. I will merely observe\nthat the event which we are said to perceive, namely the runners\nstarting, is separated by a longer or shorter chain of events from the\nevent which happens at the surface of our eyes. It is this last that\nis what is called the \u201cstimulus\u201d. Thus the event that we are said to\nperceive when we see is not the stimulus, but an anterior event related\nto it in a way that requires investigation. The same applies to hearing\nand smell, but not to touch or to perception of states of our own\nbody. In these cases, the first of the above four stages is absent.\nIt is clear that, in the case of sight, hearing and smell, there must\nbe a certain relation between the stimulus and the event said to be\nperceived, but we will not now consider what this relation must be. We\nwill consider, rather, the second, third, and fourth stages in an act\nof perceptive knowledge. This is the more legitimate as these stages\nalways exist, whereas the first is confined to certain senses.\n\nThe second stage is that which proceeds from the sense-organ to the\nbrain. It is not necessary for our purposes to consider exactly\nwhat goes on during this journey. A purely physical event--the\nstimulus--happens at the boundary of the body, and has a series\nof effects which travel along the afferent nerves to the brain.\nIf the stimulus is light, it must fall on the eye to produce the\ncharacteristic effects; no doubt light falling on other parts of the\nbody has effects, but they are not those that distinguish vision.\nSimilarly, if the stimulus is sound, it must fall on the ear. A\nsense-organ, like a photographic plate, is responsive to stimuli\nof a certain sort: light falling on the eye has effects which are\ndifferent for different wave-lengths, intensities, and directions. When\nthe events in the eye due to incident light have taken place, they\nare followed by events in the optic nerve, leading at last to some\noccurrence in the brain--an occurrence which varies with the stimulus.\nThe occurrence in the brain must be different for different stimuli\nin all cases where we can _perceive_ differences. Red and yellow, for\ninstance, are distinguishable in perception; therefore the occurrences\nalong the optic nerve and in the brain must have a different character\nwhen caused by red light from what they have when caused by yellow\nlight. But when two shades of colour are so similar that they can only\nbe distinguished by delicate instruments, not by perception, we cannot\nbe sure that they cause occurrences of different characters in the\noptic nerve and brain.\n\nWhen the disturbance has reached the brain, it may or may not cause a\ncharacteristic set of events in the brain. If it does not, we shall\nnot be what is called \u201cconscious\u201d of it. For to be \u201cconscious\u201d of\nseeing yellow, whatever else it may be, must certainly involve some\nkind of cerebral reaction to the message brought by the optic nerve.\nIt may be assumed that the great majority of messages brought to the\nbrain by the afferent nerves never secure any attention at all--they\nare like letters to a government office which remain unanswered. The\nthings in the margin of the field of vision, unless they are in some\nway interesting, are usually unnoticed; if they are noticed, they\nare brought into the centre of the field of vision unless we make a\ndeliberate effort to prevent this from occurring. These things are\nvisible, in the sense that we could be aware of them if we chose,\nwithout any change in our physical environment or in our sense-organs;\nthat is to say, only a cerebral change is required to enable them to\ncause a reaction. But usually they do not provoke any reaction; life\nwould be altogether too wearing if we had to be always reacting to\neverything in the field of vision. Where there is no reaction, the\nsecond stage completes the process, and the third and fourth stages do\nnot arise. In that case, there has been nothing that could be called\n\u201cperception\u201d connected with the stimulus in question.\n\nTo us, however, the interesting case is that in which the process\ncontinues. In this case there is first a process in the brain, of\nwhich the nature is as yet conjectural, which travels from the centre\nappropriate to the sense in question to a motor centre. From these\nthere is a process which travels along an efferent nerve, and finally\nresults in a muscular event causing some bodily movement. In our\nillustration of the man watching the beginning of a race, a process\ntravels from the part of the brain concerned with sight to the part\nconcerned with speech; this is what we called the third stage. Then\na process travels along the efferent nerves and brings about the\nmovements which constitute saying \u201cthey\u2019re off\u201d; this is what we called\nthe fourth stage.\n\nUnless all four stages exist, there is nothing that can be called\n\u201cknowledge\u201d. And even when they are all present, various further\nconditions must be satisfied if there is to be \u201cknowledge\u201d. But these\nobservations are premature, and we must return to the analysis of our\nthird and fourth stages.\n\nThe third stage is of two sorts, according as we are concerned with\na reflex or with a \u201clearned reaction\u201d, as Dr. Watson calls it. In\nthe case of a reflex, if it is complete at birth, a new-born infant\nor animal has a brain so constituted that, without the need of any\nprevious experience, there is a connection between a certain process in\nthe afferent nerves and a certain other process in the efferent nerves.\nA good example of a reflex is sneezing. A certain kind of tickling in\nthe nose produces a fairly violent movement having a very definite\ncharacter, and this connection exists already in the youngest infants.\nLearned reactions, on the other hand, are such as only occur because of\nthe effect of previous occurrences in the brain. One might illustrate\nby an analogy which, however, would be misleading if pressed. Imagine\na desert in which no rain has ever fallen, and suppose that at last\na thunderstorm occurs in it; then the course taken by the water will\ncorrespond to a reflex. But if rain continues to fall frequently, it\nwill form watercourses and river valleys; when this has occurred, the\nwater runs away along pre-formed channels, which are attributable to\nthe past \u201cexperience\u201d of the region. This corresponds to \u201clearned\nreactions\u201d. One of the most notable examples of learned reactions\nis speech: we speak because we have learned a certain language, not\nbecause our brain had originally any tendency to react in just that\nway. Perhaps all knowledge, certainly nearly all, is dependent upon\nlearned reactions, _i.e._, upon connections in the brain which are not\npart of man\u2019s congenital equipment but are the result of events which\nhave happened to him.\n\nTo distinguish between learned and unlearned responses is not always an\neasy task. It cannot be assumed that responses which are absent during\nthe first weeks of life are all learned. To take the most obvious\ninstance; sexual responses change their character to a greater or less\nextent at puberty, as a result of changes in the ductless glands, not\nas a result of experience. But this instance does not stand alone:\nas the body grows and develops, new modes of response come into play,\nmodified, no doubt, by experience, but not wholly due to it. For\nexample: a new-born baby cannot run, and therefore does not run away\nfrom what is terrifying, as an older child does. The older child has\nlearned to run, but has not necessarily learned to run _away_; the\nstimulus in learning to run may have never been a terrifying object. It\nwould therefore be a fallacy to suppose that we can distinguish between\nlearned and unlearned responses by observing what a new-born infant\ndoes, since reflexes may come into play at a later stage. Conversely,\nsome things which a child does at birth may have been learned, when\nthey are such as it could have done in the womb--for example, a certain\namount of kicking and stretching. The whole distinction between learned\nand unlearned responses, therefore, is not so definite as we could\nwish. At the two extremes we get clear cases, such as sneezing on the\none hand and speaking on the other; but there are intermediate forms of\nbehaviour which are more difficult to classify.\n\nThis is not denied even by those who attach most importance to the\ndistinction between learned and unlearned responses. In Dr. Watson\u2019s\n_Behaviorism_ (p. 103) there is a \u201cSummary of Unlearned Equipment\u201d,\nwhich ends with the following paragraph:\n\n\u201cOther activities appear at a later stage--such as blinking, reaching,\nhandling, handedness, crawling, standing, sitting-up, walking, running,\njumping. _In the great majority of these later activities it is\ndifficult to say how much of the act as a whole is due to training\nor conditioning. A considerable part is unquestionably due to the\ngrowth changes in structure, and the remainder is due to training and\nconditioning._\u201d (Watson\u2019s italics.)\n\nIt is not possible to make a logically sharp distinction in this\nmatter; in certain cases we have to be satisfied with something less\nexact. For example, we might say that those developments which are\nmerely due to normal growth are to count as unlearned, while those\nwhich depend upon special circumstances in the individual biography are\nto count as learned. But take, say, muscular development: this will\nnot take place normally unless the muscles are used, and if they are\nused they are bound to learn some of the skill which is appropriate to\nthem. And some things which must certainly count as learned, such as\nfocussing with the eyes, depend upon circumstances which are normal\nand must be present in the case of every child who is not blind. The\nwhole distinction, therefore, is one of degree rather than of kind;\nnevertheless it is valuable.\n\nThe value of the distinction between learned and unlearned reactions\nis connected with the laws of learning, to which we shall come in the\nnext chapter. Experience modifies behaviour according to certain laws,\nand we may say that a learned reaction is one in the formation of which\nthese laws have played a part. For example: children are frightened of\nloud noises from birth, but are not at first frightened of dogs; after\nthey have heard a dog barking loudly, they may become frightened of\ndogs, which is a learned reaction. If we knew enough about the brain,\nwe could make the distinction precise, by saying that learned reactions\nare those depending upon modifications of the brain other than mere\ngrowth. But as it is, we have to judge by observations of bodily\nbehaviour, and the accompanying modifications in the brain are assumed\non a basis of theory rather than actually observed.\n\nThe essential points, for our purposes, are comparatively simple.\nMan or any other animal, at birth, is such as to respond to certain\nstimuli in certain specific ways, _i.e._ by certain kinds of bodily\nmovements; as he grows, these ways of responding change, partly as the\nmere result of developing structure, partly in consequence of events in\nhis biography. The latter influence proceeds according to certain laws,\nwhich we shall consider, since they have much to do with the genesis of\n\u201cknowledge\u201d.\n\nBut--the indignant reader may be exclaiming--knowing something is not\na bodily movement, but a state of mind, and yet you talk to us about\nsneezing and such matters. I must ask the indignant reader\u2019s patience.\nHe \u201cknows\u201d that he has states of mind, and that his knowing is itself\na state of mind. I do not deny that he has states of mind, but I ask\ntwo questions: First, what sort of thing are they? Secondly, what\nevidence can he give me that he knows about them? The first question\nhe may find very difficult; and if he wants, in his answer, to show\nthat states of mind are something of a sort totally different from\nbodily movements, he will have to tell me also what bodily movements\nare, which will plunge him into the most abstruse part of physics. All\nthis I propose to consider later on, and then I hope the indignant\nreader will be appeased. As to the second question, namely, what\nevidence of his knowledge another man can give me, it is clear that he\nmust depend upon speech or writing, _i.e._ in either case upon bodily\nmovements. Therefore whatever knowledge may be to the knower, as a\nsocial phenomenon it is something displayed in bodily movements. For\nthe present I am deliberately postponing the question of what knowledge\nis to the knower, and confining myself to what it is for the external\nobserver. And for him, necessarily, it is something shown by bodily\nmovements made in answer to stimuli--more specifically, to examination\nquestions. What else it may be I shall consider at a later stage.\n\nHowever we may subsequently add to our present account by considering\nhow knowledge appears to the knower, that will not invalidate anything\nthat we may arrive at by considering how knowledge appears to the\nexternal observer. And there is something which it is important\nto realise, namely, that we are concerned with a process in which\nthe environment first acts upon a man, and then he reacts upon the\nenvironment. This process has to be considered as a whole if we are\nto discuss what knowledge is. The older view would have been that the\neffect of the environment upon us might constitute a certain kind\nof knowledge (perception), while our reaction to the environment\nconstituted volition. These were, in each case, \u201cmental\u201d occurrences,\nand their connection with nerves and brain remained entirely\nmysterious. I think the mystery can be eliminated, and the subject\nremoved from the realm of guesswork, by starting with the whole\ncycle from stimulus to bodily movement. In this way, knowing becomes\nsomething active, not something contemplative. Knowing and willing, in\nfact, are merely aspects of the one cycle, which must be considered in\nits entirety if it is to be rightly understood.\n\nA few words must be said about the human body as a mechanism. It is\nan inconceivably complicated mechanism, and some men of science think\nthat it is not explicable in terms of physics and chemistry, but is\nregulated by some \u201cvital principle\u201d which makes its laws different\nfrom those of dead matter. These men are called \u201cvitalists\u201d. I do not\nmyself see any reason to accept their view, but at the same time our\nknowledge is not sufficient to enable us to reject it definitely. What\nwe can say is that their case is not proved, and that the opposite view\nis, scientifically, a more fruitful working hypothesis. It is better\nto look for physical and chemical explanations where we can, since\nwe know of many processes in the human body which can be accounted\nfor in this way, and of none which _certainly_ cannot. To invoke a\n\u201cvital principle\u201d is to give an excuse for laziness, when perhaps more\ndiligent research would have enabled us to do without it. I shall\ntherefore assume, as a working hypothesis, that the human body acts\naccording to the same laws of physics and chemistry as those which\ngovern dead matter, and that it differs from dead matter, not by its\nlaws, but by the extraordinary complexity of its structure.\n\nThe movements of the human body may, none the less, be divided into two\nclasses, which we may call respectively \u201cmechanical\u201d and \u201cvital\u201d. As\nan example of the former, I should give the movement of a man falling\nfrom a cliff into the sea. To explain this, in its broad features, it\nis not necessary to take account of the fact that the man is alive; his\ncentre of gravity moves exactly as that of a stone would move. But when\na man climbs up a cliff, he does something that dead matter of the\nsame shape and weight would never do; this is a \u201cvital\u201d movement. There\nis in the human body a lot of stored chemical energy in more or less\nunstable equilibrium; a very small stimulus can release this energy,\nand cause a considerable amount of bodily movement. The situation is\nanalogous to that of a large rock delicately balanced on the top of a\nconical mountain; a tiny shove may send it thundering down into the\nvalley, in one direction or another according to the direction of the\nshove. So if you say to a man \u201cyour house is on fire\u201d, he will start\nrunning; although the stimulus contained very little energy, his\nexpenditure of energy may be tremendous. He increases the available\nenergy by panting, which makes his body burn up faster and increases\nthe energy due to combustion; this is just like opening the draft in a\nfurnace. \u201cVital\u201d movements are those that use up this energy which is\nin unstable equilibrium. It is they alone that concern the bio-chemist,\nthe physiologist, and the psychologist. The others, being just like\nthe movements of dead matter, may be ignored when we are specially\nconcerned with the study of Man.\n\nVital movements have a stimulus which may be inside or outside the\nbody, or both at once. Hunger is a stimulus inside the body, but\nhunger combined with the sight of good food is a double stimulus, both\ninternal and external. The effect of a stimulus may be, in theory,\naccording to the laws of physics and chemistry, but in most cases\nthis is, at present, no more than a pious opinion. What we know from\nobservation is that behaviour is modified by experience, that is to\nsay, that if similar stimuli are repeated at intervals they produce\ngradually changing reactions. When a bus conductor says, \u201cFares,\nplease\u201d, a very young child has no reaction, an older child gradually\nlearns to look for pennies, and, if a male, ultimately acquires the\npower of producing the requisite sum on demand without conscious\neffort. The way in which our reactions change with experience is a\ndistinctive characteristic of animals; moreover it is more marked in\nthe higher than in the lower animals, and most marked of all in Man.\nIt is a matter intimately connected with \u201cintelligence\u201d, and must be\ninvestigated before we can understand what constitutes knowledge from\nthe standpoint of the external observer; we shall be concerned with it\nat length in the next chapter.\n\nSpeaking broadly, the actions of all living things are such as tend to\nbiological survival, _i.e._ to the leaving of a numerous progeny. But\nwhen we descend to the lowest organisms, which have hardly anything\nthat can be called individuality, and reproduce themselves by fission,\nit is possible to take a simpler view. Living matter, within limits,\nhas the chemical peculiarity of being self-perpetuating, and of\nconferring its own chemical composition upon other matter composed of\nthe right elements. One spore falling into a stagnant pond may produce\nmillions of minute vegetable organisms; these, in turn, enable one\nsmall animal to have myriads of descendants living on the small plants;\nthese, in turn, provide life for larger animals, newts, tadpoles,\nfishes, etc. In the end there is enormously more protoplasm in that\nregion than there was to begin with. This is no doubt explicable as a\nresult of the chemical constitution of living matter. But this purely\nchemical self-preservation and collective growth is at the bottom of\neverything else that characterises the behaviour of living things.\nEvery living thing is a sort of imperialist, seeking to transform as\nmuch as possible of its environment into itself and its seed. The\ndistinction between self and posterity is one which does not exist in\na developed form in asexual unicellular organisms; many things, even\nin human life, can only be completely understood by forgetting it.\nWe may regard the whole of evolution as flowing from this \u201cchemical\nimperialism\u201d of living matter. Of this, Man is only the last example\n(so far). He transforms the surface of the globe by irrigation,\ncultivation, mining, quarrying, making canals and railways, breeding\ncertain animals, and destroying others; and when we ask ourselves,\nfrom the standpoint of an outside observer, what is the end achieved\nby all these activities, we find that it can be summed up in one\nvery simple formula: to transform as much as possible of the matter\non the earth\u2019s surface into human bodies. Domestication of animals,\nagriculture, commerce, industrialism have been stages in this process.\nWhen we compare the human population of the globe with that of\nother large animals and also with that of former times, we see that\n\u201cchemical imperialism\u201d has been, in fact, the main end to which human\nintelligence has been devoted. Perhaps intelligence is reaching the\npoint where it can conceive worthier ends, concerned with the quality\nrather than the quantity of human life. But as yet such intelligence\nis confined to minorities, and does not control the great movements of\nhuman affairs. Whether this will ever be changed I do not venture to\npredict. And in pursuing the simple purpose of maximising the amount of\nhuman life, we have at any rate the consolation of feeling at one with\nthe whole movement of living things from their earliest origin on this\nplanet.\n\n\n\n\nCHAPTER III\n\nTHE PROCESS OF LEARNING IN ANIMALS AND INFANTS\n\n\nIn the present chapter I wish to consider the processes by which,\nand the laws according to which, an animal\u2019s original repertoire of\nreflexes is changed into a quite different set of habits as a result\nof events that happen to it. A dog learns to follow his master in\npreference to anyone else; a horse learns to know his own stall in\nthe stable; a cow learns to come to the cow-shed at milking time.\nAll these are acquired habits, not reflexes; they depend upon the\ncircumstances of the animals concerned, not merely upon the congenital\ncharacteristics of the species. When I speak of an animal \u201clearning\u201d\nsomething, I shall include all cases of acquired habits, whether or not\nthey are useful to the animal. I have known horses in Italy \u201clearn\u201d to\ndrink wine, which I cannot believe to have been a desirable habit. A\ndog may \u201clearn\u201d to fly at a man who has ill-treated it, and may do so\nwith such regularity and ferocity as to lead to its being killed. I do\nnot use learning in any sense involving praise, but merely to denote\nmodification of behaviour as the result of experience.\n\nThe manner in which animals learn has been much studied in recent\nyears, with a great deal of patient observation and experiment. Certain\nresults have been obtained as regards the kinds of problems that have\nbeen investigated, but on general principles there is still much\ncontroversy. One may say broadly that all the animals that have been\ncarefully observed have behaved so as to confirm the philosophy in\nwhich the observer believed before his observations began. Nay, more,\nthey have all displayed the national characteristics of the observer.\nAnimals studied by Americans rush about frantically, with an incredible\ndisplay of hustle and pep, and at last achieve the desired result by\nchance. Animals observed by Germans sit still and think, and at last\nevolve the solution out of their inner consciousness. To the plain\nman, such as the present writer, this situation is discouraging. I\nobserve, however, that the type of problem which a man naturally sets\nto an animal depends upon his own philosophy, and that this probably\naccounts for the differences in the results. The animal responds to\none type of problem in one way and to another in another; therefore\nthe results obtained by different investigators, though different, are\nnot incompatible. But it remains necessary to remember that no one\ninvestigator is to be trusted to give a survey of the whole field.\n\nThe matters with which we shall be concerned in this chapter belong to\nbehaviourist psychology, and in part to pure physiology. Nevertheless,\nthey seem to me vital to a proper understanding of philosophy, since\nthey are necessary for an objective study of knowledge and inference. I\nmean by an \u201cobjective\u201d study one in which the observer and the observed\nneed not be the same person; when they must be identical, I call the\nstudy \u201csubjective.\u201d For the present we are concerned with what is\nrequired for understanding \u201cknowledge\u201d as an objective phenomenon. We\nshall take up the question of the subjective study of knowledge at a\nlater stage.\n\nThe scientific study of learning in animals is a very recent growth;\nit may almost be regarded as beginning with Thorndike\u2019s _Animal\nIntelligence_, which was published in 1911. Thorndike invented the\nmethod which has been adopted by practically all subsequent American\ninvestigators. In this method an animal is separated from food, which\nhe can see or smell, by an obstacle which he may overcome by chance. A\ncat, say, is put in a cage having a door with a handle which he may by\nchance push open with his nose. At first the cat makes entirely random\nmovements, until he gets his result by a mere fluke. On the second\noccasion, in the same cage, he still makes some random movements, but\nnot so many as on the first occasion. On the third occasion he does\nstill better, and before long he makes no useless movements. Nowadays\nit has become customary to employ rats instead of cats, and to put them\nin a model of the Hampton Court maze rather than in a cage. They take\nall sorts of wrong turnings at first, but after a time they learn to\nrun straight out without making any mistake. Dr. Watson gives averages\nfor nineteen rats, each of which was put into the maze repeatedly,\nwith food outside where the rat could smell it. In all the experiments\ncare was taken to make sure that the animal was very hungry. Dr.\nWatson says: \u201cThe first trial required on the average over seventeen\nminutes. During this time the rat was running around the maze, into\nblind alleys, running back to the starting point, starting for the food\nagain, biting at the wires around him, scratching himself, smelling\nthis spot and that on the floor. Finally he got to the food. He was\nallowed only a bite. Again he was put back into the maze. The taste of\nthe food made him almost frantic in his activity. He dashed about more\nrapidly. The average time for the group on the second trial is only a\nlittle over seven minutes; on the fourth trial not quite three minutes;\nfrom this point to the twenty-third trial the improvement is very\ngradual.\u201d On the thirtieth trial the time required, on the average,\nwas about thirty seconds.[3] This set of experiments may be taken as\ntypical of the whole group of studies to which it belongs.\n\n      [3] Watson, _Behaviorism_, pp. 169\u201370.\n\nThorndike, as a result of experiments with cages and mazes, formulated\ntwo \u201cprovisional laws,\u201d which are as follows:\n\n\u201cThe Law of Effect is that: of several responses made to the same\nsituation, those which are accompanied or closely followed by\nsatisfaction to the animal will, other things being equal, be more\nfirmly connected with the situation, so that, when it recurs, they\nwill be more likely to recur; those which are accompanied or closely\nfollowed by dissatisfaction to the animal will, other things being\nequal, have their connections with that situation weakened, so that,\nwhen it recurs, they will be less likely to recur. The greater the\nsatisfaction or discomfort, the greater the strengthening or weakening\nof the bond.\n\n\u201cThe Law of Exercise is that: Any response to a situation will, other\nthings being equal, be more strongly connected with the situation in\nproportion to the number of times it has been connected with that\nsituation and to the average vigour and duration of the connections.\u201d\n\nWe may sum up these two laws, roughly, in the two statements: First, an\nanimal tends to repeat what has brought it pleasure; second, an animal\ntends to repeat what it has often done before. Neither of these laws is\nat all surprising, but, as we shall see, there are difficulties in the\ntheory that they are adequate to account for the process of learning in\nanimals.\n\nBefore going further there is a theoretical point to be cleared up.\nThorndike, in his first law, speaks of satisfaction and discomfort,\nwhich are terms belonging to subjective psychology. We cannot observe\nwhether an animal _feels_ satisfaction or _feels_ discomfort; we can\nonly observe that it behaves in ways that we have become accustomed to\ninterpret as signs of these feelings. Thorndike\u2019s law, as it stands,\ndoes not belong to objective psychology, and is not capable of being\nexperimentally tested. This, however, is not so serious an objection\nas it looks. Instead of speaking of a result that brings satisfaction,\nwe can merely enumerate the results which, in fact, have the character\nwhich Thorndike mentions, namely, that the animal tends to behave so\nas to make them recur. The rat in the maze behaves so as to get the\ncheese, and when an act has led him to the cheese once, he tends to\nrepeat it. We may say that this is what we mean when we say that the\ncheese \u201cgives satisfaction\u201d, or that the rat \u201cdesires\u201d the cheese.\nThat is to say, we may use Thorndike\u2019s \u201cLaw of Effect\u201d to give us an\nobjective definition of desire, satisfaction, and discomfort. The law\nshould then say: there are situations such that animals tend to repeat\nacts which have led to them; these are the situations which the animal\nis said to \u201cdesire\u201d and in which it is said to \u201cfind satisfaction\u201d.\nThis objection to Thorndike\u2019s first law is, therefore, not very\nserious, and need not further trouble us.\n\nDr. Watson considers one principle alone sufficient to account for\nall animal and human learning, namely, the principle of \u201clearned\nreactions.\u201d This principle may be stated as follows:\n\n_When the body of an animal or human being has been exposed\nsufficiently often to two roughly simultaneous stimuli, the earlier of\nthem alone tends to call out the response previously called out by the\nother._\n\nAlthough I do not agree with Dr. Watson in thinking this principle\nalone sufficient, I do agree that it is a principle of very great\nimportance. It is the modern form of the principle of \u201cassociation\u201d.\nThe \u201cassociation of ideas\u201d has played a great part in philosophy,\nparticularly in British philosophy. But it now appears that this is\na consequence of a wider and more primitive principle, namely, the\nassociation of bodily processes. It is this wider principle that is\nasserted above. Let us see what is the nature of the evidence in its\nfavour.\n\nOur principle becomes verifiable over a much larger field than the\nolder principle owing to the fact that it is movements, not \u201cideas\u201d,\nthat are to be associated. Where animals are concerned, ideas are\nhypothetical, but movements can be observed; even with men, many\nmovements are involuntary and unconscious. Yet animal movements and\nunconscious involuntary human movements are just as much subject to\nthe law of association as the most conscious ideas. Take, _e.g._ the\nfollowing example (Watson, p. 33). The pupil of the eye expands in\ndarkness and contracts in bright light; this is an involuntary and\nunconscious action of which we only become aware by observing others.\nNow take some person and repeatedly expose him to bright light at\nthe same moment that you ring an electric bell. After a time the\nelectric bell alone will cause his pupils to contract. As far as can be\ndiscovered, all muscles behave in this way. So do glands where they can\nbe tested. It is said that a brass band can be reduced to silence by\nsucking a lemon in front of it, owing to the effect upon the salivary\nglands of its members; I confess that I have never verified this\nstatement. But you will find the exact scientific analogue for dogs in\nWatson, p. 26. You arrange a tube in a dog\u2019s mouth so that saliva drops\nout at a measurable rate. When you give the dog food it stimulates the\nflow of saliva. At the same moment you touch his left thigh. After a\ncertain length of time the touch on the left thigh will produce just\nas much saliva without the food as with it. The same sort of thing\napplies to emotions, which depend upon the ductless glands. Children\nat birth are afraid of loud noises, but not of animals. Watson took a\nchild eleven months old, who was fond of a certain white rat; twice at\nthe moment when the child touched the rat, a sudden noise was made just\nbehind the child\u2019s head. This was enough to cause fear of the rat on\nsubsequent occasions, no doubt owing to the fact that the adrenal gland\nwas now stimulated by the substitute stimulus, just like the salivary\nglands in the dog or the trumpet player. The above illustrations show\nthat \u201cideas\u201d are not the essential units in association. It seems that\nnot merely is \u201cmind\u201d irrelevant, but even the brain is less important\nthan was formerly supposed. At any rate, what is known experimentally\nis that the glands and muscles (both striped and unstriped) of the\nhigher animals exhibit the law of transfer of response, _i.e._ when\ntwo stimuli have often been applied together, one will ultimately call\nout the response which formerly the other called out. This law is one\nof the chief bases of habit. It is also obviously essential to our\nunderstanding of language: the sight of a dog calls up the word \u201cdog\u201d,\nand the word \u201cdog\u201d calls up some of the responses appropriate to a real\ndog.\n\nThere is, however, another element in learning, besides mere habit.\nThis is the element dealt with by Thorndike\u2019s \u201cLaw of Effect.\u201d Animals\ntend to repeat acts which have pleasant consequences, and to avoid\nsuch as have unpleasant consequences. But, as we saw a moment ago,\n\u201cpleasant\u201d and \u201cunpleasant\u201d are words which we cannot verify by\nobjective observation. What we can verify by observation is that\nan animal seeks situations which in fact have had certain results,\nand avoids situations which in fact have had certain other results.\nMoreover, broadly speaking, the animal seeks results which tend to\nsurvival of itself or its offspring, and avoids results which tend in\nthe opposite direction. This, however, is not invariable. Moths seek\nflames and men seek drink, though neither is biologically useful. It\nis only approximately, in situations long common, that animals are so\nadjusted to their environment as to act in a way which is advantageous\nfrom a biological standpoint. In fact, biological utility must\nnever be employed as an explanation, but only noticed as a frequent\ncharacteristic, of the ways in which animals behave.\n\nDr. Watson is of the opinion that Thorndike\u2019s \u201cLaw of Effect\u201d is\nunnecessary. He first suggests that only two factors are called for in\nthe explanation of habit, namely, _frequency_ and _recency_. Frequency\nis covered by Thorndike\u2019s \u201cLaw of Exercise\u201d, but recency, which is\nalmost certainly a genuine factor, is not covered by Thorndike\u2019s two\nlaws. That is to say, when a number of random movements have finally\nresulted in success, the more recent of these movements are likely to\nbe repeated earlier, on a second trial, than the earlier ones. But Dr.\nWatson finally abandons this method of dealing with habit-formation in\nfavour of the one law of \u201cconditioned reflexes\u201d or \u201clearned reactions\u201d.\nHe says (_Behaviorism_, p. 166):\n\n\u201cOnly a few psychologists have been interested in the problem. Most\nof the psychologists, it is to be regretted, have even failed to see\nthat there is a problem. They believe habit formation is implanted by\nkind fairies. For example, Thorndike speaks of pleasure stamping in\nthe successful movement and displeasure stamping out the unsuccessful\nmovements. Most of the psychologists talk, too, quite volubly about the\nformation of new pathways in the brain, as though there were a group\nof tiny servants of Vulcan there who run through the nervous system\nwith hammer and chisel digging new trenches and deepening old ones.\nI am not sure that the problem when phrased in this way is a soluble\none. I feel that there must come some simpler way of envisaging the\nwhole process of habit formation or else it may remain insoluble.\nSince the advent of the conditioned reflex hypothesis in psychology\nwith all of the simplifications (and I am often fearful that it may be\nan oversimplification!) I have had my own laryngeal processes [_i.e._\nwhat others call \u201cthoughts\u201d] stimulated to work upon this problem from\nanother angle.\u201d\n\nI agree with Dr. Watson that the explanations of habit-formation which\nare usually given are very inadequate, and that few psychologists have\nrealised either the importance or the difficulty of the problem. I\nagree also that a great many cases are covered by his formula of the\nconditioned reflex. He relates a case of a child who once touched a\nhot radiator, and afterward avoided it for two years. He adds: \u201cIf we\nshould keep our old habit terminology, we should have in this example\na habit formed by a single trial. There can be then in this case no\n\u2018stamping in of the successful movement\u2019 and \u2018no stamping out of the\nunsuccessful movement.\u2019\u201d On the basis of such examples, he believes\nthat the whole of habit-formation can be derived from the principle of\nthe conditioned reflex, which he formulates as follows (p. 168):\n\n_Stimulus X will not now call out reaction R; stimulus Y will call out\nreaction R (unconditioned reflex); but when stimulus X is presented\nfirst and then Y (which does call out R) shortly thereafter, X will\nthereafter call out R. In other words, stimulus X becomes ever\nthereafter substituted for Y._\n\nThis law is so simple, so important, and so widely true that there\nis a danger lest its scope should be exaggerated, just as, in the\neighteenth century, physicists tried to explain everything by means of\ngravitation. But when considered as covering all the ground, it seems\nto me to suffer from two opposite defects. In the first place, there\nare cases where no habit is set up, although by the law it should be.\nIn the second place, there are habits which, so far as we can see at\npresent, have a different genesis.\n\nTo take the first point first: the word \u201cpepper\u201d does not make people\nsneeze, though according to the law it should.[4] Words which describe\nsucculent foods will make the mouth water; voluptuous words will have\nsome of the effect that would be produced by the situations they\nsuggest; but no words will produce sneezes or the reactions appropriate\nto tickling. In the diagram given by Dr. Watson (p. 106), there are\nfour reflexes which appear to be not sources of conditioned reflexes,\nnamely sneezing, hiccoughing, blinking, and the Babinski reflex;\nof these, however, blinking, it is suggested (p. 99) may be really\nitself a conditioned reflex. There may be some quite straightforward\nexplanation of the fact that some reactions can be produced by\nsubstitute stimuli while others cannot, but none is offered. Therefore\nthe law of the conditioned reflex, as formulated, is too wide, and it\nis not clear what is the principle according to which its scope should\nbe restricted.\n\n      [4] Dr. Watson apparently entertains hopes of teaching babies\n          to sneeze when they see the pepper box, but he has not\n          yet done so. See _Behaviorism_, p. 90.\n\nThe second objection to Dr. Watson\u2019s law of habit, if valid, is more\nimportant than the first; but its validity is more open to question. It\nis contended that the acts by which solutions of problems are obtained\nare, in cases of a certain kind, not random acts leading to success by\nmere chance, but acts proceeding from \u201cinsight\u201d, involving a \u201cmental\u201d\nsolution of the problem as a preliminary to the physical solution.\nThis is especially the view of those who advocate _Gestaltpsychologie_\nor the psychology of configuration. We may take, as typical of their\nattitude on the subject of learning, K\u00f6hler\u2019s _Mentality of Apes_.\nK\u00f6hler went to Tenerife with certain chimpanzees in the year 1913;\nowing to the war he was compelled to remain with them until 1917, so\nthat his opportunities for study were extensive. He complains of the\nmaze and cage problems set by American investigators that they are such\nas _cannot_ be solved by intelligence. Sir Isaac Newton himself could\nnot have got out of the Hampton Court maze by any method except trial\nand error. K\u00f6hler, on the other hand, set his apes problems which could\nbe solved by what he calls \u201cinsight\u201d. He would hang up a banana[5] out\nof reach, and leave boxes in the neighbourhood so that by standing\non the boxes the chimpanzees could reach the fruit. Sometimes they\nhad to pile three or even four boxes on top of each other before they\ncould achieve success. Then he would put the banana outside the bars\nof the cage, leaving a stick inside, and the ape would get the banana\nby reaching for it with the stick. On one occasion, one of them, named\nSultan, had two bamboo sticks, each too short to reach the banana;\nafter vain efforts followed by a period of silent thought, he fitted\nthe smaller into the hollow of the other, and so manufactured one stick\nwhich was long enough. It seems, however, from the account, that he\nfirst fitted the two together more or less accidentally, and only then\nrealised that he had found a solution. Nevertheless, his behaviour when\nhe had once realised that one stick could be made by joining the two\nwas scarcely Watsonian: there was no longer anything tentative, but a\ndefinite triumph, first in anticipation and then in action. He was so\npleased with his new trick that he drew a number of bananas into his\ncage before eating any of them. He behaved, in fact, as capitalists\nhave behaved with machinery.\n\n      [5] Called by K\u00f6hler \u201cthe objective,\u201d because the word\n          \u201cbanana\u201d is too humble for a learned work. The pictures\n          disclose the fact that \u201cthe objective\u201d was a mere banana.\n\nK\u00f6hler says: \u201cWe can, from our own experience, distinguish sharply\nbetween the kind of conduct which, from the very beginning, arises out\nof a consideration of the characteristics of a situation, and one that\ndoes not. Only in the former case do we speak of insight, and only\nthat behaviour of animals definitely appears to us intelligent which\ntakes account from the beginning of the lie of the land, and proceeds\nto deal with it in a smooth continuous course. Hence follows this\ncharacteristic: _to set up as the criterion of insight, the appearance\nof a complete solution with reference to the whole lay-out of the\nfield_.\u201d\n\nGenuine solutions of problems, K\u00f6hler says, do not improve by\nrepetition; they are perfect on the first occasion, and, if anything,\ngrow worse by repetition, when the excitement of discovery has worn\noff. The whole account that K\u00f6hler gives of the efforts of his\nchimpanzees makes a totally different impression from that of the\nrats in mazes, and one is forced to conclude that the American work\nis somewhat vitiated by confining itself to one type of problem,\nand drawing from that one type conclusions which it believes to be\napplicable to all problems of animal learning. It seems that there\nare two ways of learning, one by experience, and the other by what\nK\u00f6hler calls \u201cinsight\u201d. Learning by experience is possible to most\nvertebrates, though rarely, so far as is known, to invertebrates.\nLearning by \u201cinsight\u201d, on the contrary, is not known to exist in any\nanimals lower than the anthropoid apes, though it would be extremely\nrash to assert that it will not be revealed by further observations on\ndogs or rats. Unfortunately, some animals--for instance, elephants--may\nbe extremely intelligent, but the practical difficulty and expense of\nexperimentation with them is so great that we are not likely to know\nmuch about them for some time to come. However, the real problem is\nalready sufficiently definite in K\u00f6hler\u2019s book: it is the analysis of\n\u201cinsight\u201d as opposed to the method of the conditioned reflex.\n\nLet us first be clear as to the nature of the problem, when described\nsolely in terms of behaviour. A hungry monkey, if sufficiently near\nto a banana, will perform acts such as, in circumstances to which it\nhas been accustomed, have previously enabled it to obtain bananas.\nThis fits well with either Watson or Thorndike, so far. But if these\nfamiliar acts fail, the animal will, if it has been long without food,\nis in good health, and is not too tired, proceed to other acts which\nhave never hitherto produced bananas. One may suppose, if one wishes to\nfollow Watson, that these new acts are composed of a number of parts,\neach of which, on some former occasion, has occurred in a series which\nended with the obtaining of the banana. Or one may suppose--as I think\nThorndike does--that the acts of the baffled animal are random acts,\nso that the solution emerges by pure chance. But even in the first\nhypothesis, the element of chance is considerable. Let us suppose that\nthe acts A, B, C, D, E, have each, on a former occasion, been part of\na series ending with success, but that now for the first time it is\nnecessary to perform them all, and in the right order. It is obvious\nthat, if they are only combined by chance, the animal will be lucky if\nit performs them all in the right order before dying of hunger.\n\nBut K\u00f6hler maintains that to anyone watching his chimpanzees it was\nobvious they did not obtain \u201ca composition of the solution out of\nchance parts\u201d. He says (pp. 199\u2013200):\n\n\u201cIt is certainly not a characteristic of the chimpanzee, when he is\nbrought into an experimental situation, to make any chance movements,\nout of which, among other things, a non-genuine solution could arise.\nHe is very seldom seen to attempt anything which would have to be\nconsidered accidental in relation to the situation (excepting, of\ncourse, if his interest is turned away from the objective to other\nthings). As long as his efforts are directed to the objective, all\ndistinguishable stages of his behaviour (as with human beings in\nsimilar situations) tend to appear as complete attempts at solutions,\n_none_ of which appears as the product of accidentally arrayed\nparts. This is true, most of all, of the solution which is finally\nsuccessful. Certainly it often follows upon a period of perplexity or\nquiet (often a period of survey), but in real and convincing cases,\nthe solution never appears in a disorder of blind impulses. It is\none continuous smooth action, which can be resolved into its parts\n_only by the imagination_ of the onlooker; in reality they do _not_\nappear independently. But that in so many \u2018genuine\u2019 cases as have been\ndescribed, these solutions as wholes should have arisen from mere\nchance, is an entirely inadmissible supposition.\u201d\n\nThus we may take it as an observed fact that, so far as overt behaviour\nis concerned, there are two objections to the type of theory with\nwhich we began, when considered as covering the whole field. The first\nobjection is that in cases of a certain kind, the solution appears\nsooner than it should according to the doctrine of chances; the second\nis that it appears as a whole, _i.e._ that the animal, after a period\nof quiescence, suddenly goes through the right series of actions\nsmoothly, and without hesitation.\n\nWhere human beings are concerned, it is difficult to obtain such\ngood data as in the case of animals. Human mothers will not allow\ntheir children to be starved, and then shut up in a room containing\na banana which can only be reached by putting a chair on the table\nand a footstool on the chair, and then climbing up without breaking\nany bones. Nor will they permit them to be put into the middle of a\nHampton Court maze, with their dinner getting cold outside. Perhaps\nin time the State will perform these experiments with the children of\npolitical prisoners, but as yet, perhaps fortunately, the authorities\nare not sufficiently interested in science. One can observe, however,\nthat human learning seems to be of both sorts, namely that described\nby Watson and that described by K\u00f6hler. I am persuaded that speech is\nlearnt by the Watsonian method, so long as it is confined to single\nwords: often the trial and error, in later stages, proceeds _sotto\nvoce_, but it takes place overtly at first, and in some children until\ntheir speech is quite correct. The speaking of sentences, however, is\nalready more difficult to explain without bringing in the apprehension\nof wholes which is the thing upon which _Gestaltpsychologie_\nlays stress. In the later stages of learning, the sort of sudden\nillumination which came to K\u00f6hler\u2019s chimpanzees is a phenomenon with\nwhich every serious student must be familiar. One day, after a period\nof groping bewilderment, the schoolboy knows what algebra is all\nabout. In writing a book, my own experience--which I know is fairly\ncommon, though by no means universal--is that for a time I fumble and\nhesitate, and then suddenly I see the book as a whole, and have only\nto write it down as if I were copying a completed manuscript.\n\nIf these phenomena are to be brought within the scope of behaviourist\npsychology, it must be by means of \u201cimplicit\u201d behaviour. Watson makes\nmuch use of this in the form of talking to oneself, but in apes it\ncannot take quite this form. And it is necessary to have some theory\nto explain the success of \u201cimplicit\u201d behaviour, whether we call it\n\u201cthought\u201d or not. Perhaps such a theory can be constructed on Watson\u2019s\nlines, but it has certainly not yet been constructed. Until the\nbehaviourists have satisfactorily explained the kind of discovery which\nappears in K\u00f6hler\u2019s observations, we cannot say that their thesis is\nproved. This is a matter which will occupy us again at a later stage;\nfor the present let us preserve an open mind.\n\n\n\n\nCHAPTER IV\n\nLANGUAGE\n\n\nThe subject of language is one which has not been studied with\nsufficient care in traditional philosophy. It was taken for granted\nthat words exist to express \u201cthoughts,\u201d and generally also that\n\u201cthoughts\u201d have \u201cobjects\u201d which are what the words \u201cmean\u201d. It was\nthought that, by means of language, we could deal directly with what\nit \u201cmeans\u201d, and that we need not analyse with any care either of the\ntwo supposed properties of words, namely that of \u201cexpressing\u201d thoughts\nand that of \u201cmeaning\u201d things. Often when philosophers intended to be\nconsidering the objects meant by words they were in fact considering\nonly the words, and when they were considering words they made the\nmistake of supposing, more or less unconsciously, that a word is a\nsingle entity, not, as it really is, a set of more or less similar\nevents. The failure to consider language explicitly has been a cause\nof much that was bad in traditional philosophy. I think myself that\n\u201cmeaning\u201d can only be understood if we treat language as a bodily\nhabit, which is learnt just as we learn football or bicycling. The only\nsatisfactory way to treat language, to my mind, is to treat it in this\nway, as Dr. Watson does. Indeed, I should regard the theory of language\nas one of the strongest points in favour of behaviourism.\n\nMan has various advantages over the beasts, for example, fire,\nclothing, agriculture, and tools--not the possession of domestic\nanimals, for ants have them. But more important than any of these\nis language. It is not known how or when language arose, nor why\nchimpanzees do not speak. I doubt if it is even known whether writing\nor speech is the older form of language. The pictures made in caves by\nthe Cro-Magnon men may have been intended to convey a meaning, and may\nhave been a form of writing. It is known that writing developed out of\npictures, for that happened in historical times; but it is not known\nto what extent pictures had been used in pre-historic times as a means\nof giving information or commands. As for spoken language, it differs\nfrom the cries of animals in being not merely an expression of emotion.\nAnimals have cries of fear, cries expressing pleasure in the discovery\nof food, and so on, and by means of these cries they influence each\nother\u2019s actions. But they do not appear to have any means of expressing\nanything except emotions, and then only emotions which they are\nactually feeling. There is no evidence that they possess anything\nanalogous to narrative. We may say, therefore, without exaggeration,\nthat language is a human prerogative, and probably the chief habit in\nwhich we are superior to the \u201cdumb\u201d animals.\n\nThere are three matters to be considered in beginning the study of\nlanguage. First: what words are, regarded as physical occurrences;\nsecondly, what are the circumstances that lead us to use a given word;\nthirdly, what are the effects of our hearing or seeing a given word.\nBut as regards the second and third of these questions, we shall find\nourselves led on from words to sentences and thus confronted with fresh\nproblems perhaps demanding rather the methods of _Gestaltpsychologie_.\n\nOrdinary words are of four kinds: spoken, heard, written, and read. It\nis of course largely a matter of convention that we do not use words of\nother kinds. There is the deaf-and-dumb language; a Frenchman\u2019s shrug\nof the shoulders is a word; in fact, any kind of externally perceptible\nbodily movement may become a word, if social usage so ordains. But the\nconvention which has given the supremacy to speaking is one which has\na good ground, since there is no other way of producing a number of\nperceptibly different bodily movements so quickly or with so little\nmuscular effort. Public speaking would be very tedious if statesmen\nhad to use the deaf-and-dumb language, and very exhausting if all\nwords involved as much muscular effort as a shrug of the shoulders. I\nshall ignore all forms of language except speaking, hearing, writing,\nand reading, since the others are relatively unimportant and raise no\nspecial psychological problems.\n\nA spoken word consists of a series of movements in the larynx and\nthe mouth, combined with breath. Two closely similar series of such\nmovements may be instances of the same words, though they may also\nnot be, since two words with different meanings may sound alike; but\ntwo such series which are not closely similar cannot be instances of\nthe same word. (I am confining myself to one language.) Thus a single\nspoken word, say \u201cdog,\u201d is a certain set of closely similar series of\nbodily movements, the set having as many members as there are occasions\nwhen the word \u201cdog\u201d is pronounced. The degree of similarity required\nin order that the occurrence should be an instance of the word \u201cdog\u201d\ncannot be specified exactly. Some people say \u201cdawg\u201d, and this must\ncertainly be admitted. A German might say \u201ctok\u201d, and then we should\nbegin to be doubtful. In marginal cases, we cannot be sure whether a\nword has been pronounced or not. A spoken word is a form of bodily\nbehaviour without sharp boundaries, like jumping or hopping or running.\nIs a man running or walking? In a walking-race the umpire may have\ngreat difficulty in deciding. Similarly there may be cases where it\ncannot be decided whether a man has said \u201cdog\u201d or \u201cdock\u201d. A spoken word\nis thus at once general and somewhat vague.\n\nWe usually take for granted the relation between a word spoken and a\nword heard. \u201cCan you hear what I say?\u201d we ask, and the person addressed\nsays \u201cyes\u201d. This is of course a delusion, a part of the naive realism\nof our unreflective outlook on the world. We never hear what is said;\nwe hear something having a complicated causal connection with what is\nsaid. There is first the purely physical process of sound-waves from\nthe mouth of the speaker to the ear of the hearer, then a complicated\nprocess in the ear and nerves, and then an event in the brain, which\nis related to our hearing of the sound in a manner to be investigated\nlater, but is at any rate simultaneous with our hearing of the sound.\nThis gives the physical causal connection between the word spoken and\nthe word heard. There is, however, also another connection of a more\npsychological sort. When a man utters a word, he also hears it himself,\nand so that the word spoken and the word heard become intimately\nassociated for anyone who knows how to speak. And a man who knows how\nto speak can also utter any word he hears in his own language, so that\nthe association works equally well both ways. It is because of the\nintimacy of this association that the plain man identifies the word\nspoken with the word heard, although in fact the two are separated by a\nwide gulf.\n\nIn order that speech may serve its purpose, it is not necessary, as it\nis not possible, that heard and spoken words should be identical, but\nit is necessary that when a man utters different words the heard words\nshould be different, and when he utters the same word on two occasions\nthe heard word should be approximately the same on the two occasions.\nThe first of these depends upon the sensitiveness of the ear and its\ndistance from the speaker; we cannot distinguish between two rather\nsimilar words if we are too far off from the man who utters them. The\nsecond condition depends upon uniformity in the physical conditions,\nand is realised in all ordinary circumstances. But if the speaker were\nsurrounded by instruments which were resonant to certain notes but not\nto certain others, some tones of voice might carry and others might\nbe lost. In that case, if he uttered the same word with two different\nintonations, the hearer might be quite unable to recognise the\nsameness. Thus the efficacy of speech depends upon a number of physical\nconditions. These, however, we will take for granted, in order to come\nas soon as possible to the more psychological parts of our topic.\n\nWritten words differ from spoken words in being material structures. A\nspoken word is a process in the physical world, having an essential\ntime-order; a written word is a series of pieces of matter, having\nan essential space-order. As to what we mean by \u201cmatter\u201d, that is a\nquestion with which we shall have to deal at length at a later stage.\nFor the present it is enough to observe that the material structures\nwhich constitute written words, unlike the processes that constitute\nspoken words, are capable of enduring for a long time--sometimes\nfor thousands of years. Moreover, they are not confined to one\nneighbourhood, but can be made to travel about the world. These are the\ntwo great advantages of writing over speech. This, at least, has been\nthe case until recently. But with the coming of radio writing it has\nbegun to lose its pre-eminence: one man can now speak to multitudes\nspread over a whole country. Even in the matter of permanence, speech\nmay become the equal of writing. Perhaps, instead of legal documents,\nwe shall have gramophone records, with voice signatures by the parties\nto the contract. Perhaps, as in Wells\u2019s _When the Sleeper Awakes_,\nbooks will no longer be printed but merely arranged for the gramophone.\nIn that case the need for writing may almost cease to exist. However,\nlet us return from these speculations to the world of the present day.\n\nThe word read, as opposed to the written or printed word, is just\nas evanescent as the word spoken or heard. Whenever a written word,\nexposed to light, is in a suitable spatial relation to a normal eye, it\nproduces a certain complicated effect upon the eye; the part of this\nprocess which occurs outside the eye is investigated by the science of\nlight, whereas the part that occurs in the eye belongs to physiological\noptics. There is then a further process, first in the optic nerve and\nafterwards in the brain; the process in the brain is simultaneous\nwith vision. What further relation it has to vision is a question\nas to which there has been much philosophical controversy; we shall\nreturn to it at a later stage. The essence of the matter, as regards\nthe causal efficacy of writing, is that the act of writing produces\nquasi-permanent material structures which, throughout the whole of\ntheir duration, produce closely similar results upon all suitably\nplaced normal eyes; and as in the case of speaking, different written\nwords lead to different read words, and the same word written twice\nleads to the same read word--again with obvious limitations.\n\nSo much for the physical side of language, which is often unduly\nneglected. I come now to the psychological side, which is what really\nconcerns us in this chapter.\n\nThe two questions we have to answer, apart from the problems raised\nby sentences as opposed to words, are: First, what sort of behaviour\nis stimulated by hearing a word? And secondly, what sort of occasion\nstimulates us to the behaviour that consists in pronouncing a word? I\nput the questions in this order because children learn to react to the\nwords of others before they learn to use words themselves. It might be\nobjected that, in the history of the race, the first spoken word must\nhave preceded the first heard word, at least by a fraction of a second.\nBut this is not very relevant, nor is it certainly true. A noise may\nhave meaning to the hearer, but not to the utterer; in that case it\nis a heard word but not a spoken word. (I shall explain what I mean\nby \u201cmeaning\u201d shortly.) Friday\u2019s footprint had \u201cmeaning\u201d for Robinson\nCrusoe but not for Friday. However that may be, we shall do better\nto avoid the very hypothetical parts of anthropology that would be\ninvolved, and take up the learning of language as it can be observed in\nthe human infant of the present day. And in the human infant as we know\nhim, definite reactions to the words of others come much earlier than\nthe power of uttering words himself.\n\nA child learns to understand words exactly as he learns any other\nprocess of bodily association. If you always say \u201cbottle\u201d when you give\na child his bottle, he presently reacts to the word \u201cbottle\u201d, within\nlimits, as he formerly reacted to the bottle. This is merely an example\nof the law of association which we considered in the preceding chapter.\nWhen the association has been established, parents say that the child\n\u201cunderstands\u201d the word \u201cbottle\u201d, or knows what the word \u201cmeans\u201d. Of\ncourse the word does not have _all_ the effects that the actual bottle\nhas. It does not exert gravitation, it does not nourish, it cannot\nbump on to the child\u2019s head. The effects which are shared by the word\nand the thing are those which depend upon the law of association or\n\u201cconditional reflexes\u201d or \u201clearned reactions\u201d. These may be called\n\u201cassociative\u201d effects or \u201cmnemic\u201d effects--the latter name being\nderived from Semon\u2019s book _Mneme_,[6] in which he traces all phenomena\nanalogous to memory to a law which is, in effect, not very different\nfrom the law of association or \u201cconditioned reflexes\u201d.\n\n      [6] London: George Allen & Unwin, Ltd.\n\nIt is possible to be a little more precise as to the class of effects\nconcerned. A physical object is a centre from which a variety of\ncausal chains emanate. If the object is visible to John Smith, one\nof the causal chains emanating from it consists first of light-waves\n(or light-quanta) which travel from the object to John Smith\u2019s eye,\nthen of events in his eye and optic nerve, then of events in his\nbrain, and then (perhaps) of a reaction on his part. Now mnemic\neffects belong only to events in living tissue; therefore only those\neffects of the bottle which happen either inside John Smith\u2019s body,\nor as a result of his reaction to the bottle, can become associated\nwith his hearing the word \u201cbottle\u201d. And even then only certain events\ncan be associated: nourishment happens in the body, yet the word\n\u201cbottle\u201d cannot nourish. The law of conditioned reflexes is subject\nto ascertainable limitations, but within its limits it supplies what\nis wanted to explain the understanding of words. The child becomes\nexcited when he sees the bottle; this is already a conditioned reflex,\ndue to experience that this sight precedes a meal. One further stage\nin conditioning makes the child grow excited when he hears the word\n\u201cbottle\u201d. He is then said to \u201cunderstand\u201d the word.\n\nWe may say, then, that a person understands a word which he hears if,\nso far as the law of conditioned reflexes is applicable, the effects of\nthe word are the same as those of what it is said to \u201cmean\u201d. This of\ncourse only applies to words like \u201cbottle\u201d, which denote some concrete\nobject or some class of concrete objects. To understand a word such\nas \u201creciprocity\u201d or \u201crepublicanism\u201d is a more complicated matter, and\ncannot be considered until we have dealt with sentences. But before\nconsidering sentences we have to examine the circumstances which make\nus use a word, as opposed to the consequences of hearing it used.\n\nSaying a word is more difficult than using it, except in the case of\na few simple sounds which infants make before they know that they are\nwords, such as \u201cma-ma\u201d and \u201cda-da.\u201d These two are among the many random\nsounds that all babies make. When a child says \u201cma-ma\u201d in the presence\nof his mother by chance she thinks he knows what this noise means, and\nshe shows pleasure in ways that are agreeable to the infant. Gradually,\nin accordance with Thorndike\u2019s law of effect, he acquires the habit\nof making this noise in the presence of his mother, because in these\ncircumstances the consequences are pleasant. But it is only a very\nsmall number of words that are acquired in this way. The great majority\nof words are acquired by imitation, combined with the association\nbetween thing and word which the parents deliberately establish in the\nearly stages (after the very first stage). It is obvious that using\nwords oneself involves something over and above the association between\nthe _sound_ of the word and its meaning. Dogs understand many words,\nand infants understand far more than they can say. The infant has to\ndiscover that it is possible and profitable to make noises like those\nwhich he hears. (This statement must not be taken quite literally, or\nit would be too intellectualistic.) He would never discover this if\nhe did not make noises at random, without the intention of talking.\nHe then gradually finds that he can make noises like those which he\nhears, and in general the consequences of doing so are pleasant.\nParents are pleased, desired objects can be obtained, and--perhaps most\nimportant of all--there is a sense of power in making intended instead\nof accidental noises. But in this whole process there is nothing\nessentially different from the learning of mazes by rats. It resembles\nthis form of learning, rather than that of K\u00f6hler apes, because no\namount of intelligence could enable the child to find out the names of\nthings--as in the case of the mazes, experience is the only possible\nguide.\n\nWhen a person knows how to speak, the conditioning proceeds in the\nopposite direction to that which operates in understanding what others\nsay. The reaction of a person who knows how to speak, when he notices\na cat, is naturally to utter the word \u201ccat\u201d; he may not actually do\nso, but he will have a reaction leading towards this act, even if for\nsome reason the overt act does not take place. It is true that he may\nutter the word \u201ccat\u201d because he is \u201cthinking\u201d about a cat, not actually\nseeing one. This, however, as we shall see in a moment, is merely one\nfurther stage in the process of conditioning. The use of single words,\nas opposed to sentences, is wholly explicable, so far as I can see, by\nthe principles which apply to animals in mazes.\n\nCertain philosophers who have a prejudice against analysis contend that\nthe sentence comes first and the single word later. In this connection\nthey always allude to the language of the Patagonians, which their\nopponents, of course, do not know. We are given to understand that a\nPatagonian can understand you if you say \u201cI am going to fish in the\nlake behind the western hill\u201d, but that he cannot understand the word\n\u201cfish\u201d by itself. (This instance is imaginary, but it represents the\nsort of thing that is asserted.) Now it may be that Patagonians are\npeculiar--indeed they must be, or they would not choose to live in\nPatagonia. But certainly infants in civilized countries do not behave\nin this way, with the exception of Thomas Carlyle and Lord Macaulay.\nThe former never spoke before the age of three, when, hearing his\nyounger brother cry, he said, \u201cWhat ails wee Jock?\u201d Lord Macaulay\n\u201clearned in suffering what he taught in song\u201d, for, having spilt a cup\nof hot tea over himself at a party, he began his career as a talker\nby saying to his hostess, after a time, \u201cThank you, Madam, the agony\nis abated\u201d. These, however, are facts about biographers, not about\nthe beginnings of speech in infancy. In all children that have been\ncarefully observed, sentences come much later than single words.\n\nChildren, at first, are limited as to their power of producing sounds,\nand also by the paucity of their learned associations. I am sure the\nreason why \u201cma-ma\u201d and \u201cda-da\u201d have the meaning they have is that they\nare sounds which infants make spontaneously at an early age, and are\ntherefore convenient as sounds to which the elders can attach meaning.\nIn the very beginning of speech there is not imitation of grownups, but\nthe discovery that sounds made spontaneously have agreeable results.\nImitation comes later, after the child has discovered that sounds\ncan have this quality of \u201cmeaning\u201d. The type of skill involved is\nthroughout exactly similar to that involved in learning to play a game\nor ride a bicycle.\n\nWe may sum up this theory of meaning in a simple formula. When through\nthe law of conditioned reflexes, A has come to be a cause of C, we will\ncall A an \u201cassociative\u201d cause of C, and C an \u201cassociative\u201d effect of\nA. We shall say that, to a given person, the word A, when he hears it,\n\u201cmeans\u201d C, if the associative effects of A are closely similar to those\nof C; and we shall say that the word A, when he utters it, \u201cmeans\u201d C,\nif the utterance of A is an associative effect of C, or of something\npreviously associated with C. To put the matter more concretely, the\nword \u201cPeter\u201d means a certain person if the associated effects of\nhearing the word \u201cPeter\u201d are closely similar to those of seeing Peter,\nand the associative causes of uttering the word \u201cPeter\u201d are occurrences\npreviously associated with Peter. Of course as our experience increases\nin complexity this simple schema becomes obscured and overlaid, but I\nthink it remains fundamentally true.\n\nThere is an interesting and valuable book by Messrs. C.\u00a0K. Ogden and\nI.\u00a0A. Richards, called _The Meaning of Meaning_. This book, owing to\nthe fact that it concentrates on the causes of uttering words, not on\nthe effects of hearing them, gives only half the above theory, and that\nin a somewhat incomplete form. It says that a word and its meaning have\nthe same causes. I should distinguish between _active_ meaning, that\nof the man uttering the word, and _passive_ meaning, that of the man\nhearing the word. In active meaning the word is associatively caused by\nwhat it means or something associated with this; in passive meaning,\nthe associative effects of the word are approximately the same as those\nof what it means.\n\nOn behaviourist lines, there is no important difference between proper\nnames and what are called \u201cabstract\u201d or \u201cgeneric\u201d words. A child learns\nto use the word \u201ccat\u201d, which is general, just as he learns to use\nthe word \u201cPeter\u201d, which is a proper name. But in actual fact \u201cPeter\u201d\nreally covers a number of different occurrences, and is in a sense\ngeneral. Peter may be near or far, walking or standing or sitting,\nlaughing or frowning. All these produce different stimuli, but the\nstimuli have enough in common to produce the reaction consisting\nof the word \u201cPeter\u201d. Thus there is no essential difference, from a\nbehaviourist point of view, between \u201cPeter\u201d and \u201cman\u201d. There are\nmore resemblances between the various stimuli to the word \u201cPeter\u201d\nthan between those to the word \u201cman\u201d, but this is only a difference\nof degree. We have not names for the fleeting particular occurrences\nwhich make up the several appearances of Peter, because they are not\nof much practical importance; their importance, in fact, is purely\ntheoretic and philosophical. As such, we shall have a good deal to say\nabout them at a later stage. For the present, we notice that there are\nmany occurrences of Peter, and many occurrences of the word \u201cPeter\u201d;\neach, to the man who sees Peter, is a set of events having certain\nsimilarities. More exactly, the occurrences of Peter are _causally_\nconnected, whereas the occurrences of the word \u201cPeter\u201d are connected by\nsimilarity. But this is a distinction which need not concern us yet.\n\nGeneral words such as \u201cman\u201d or \u201ccat\u201d or \u201ctriangle\u201d are said to denote\n\u201cuniversals\u201d, concerning which, from the time of Plato to the present\nday, philosophers have never ceased to debate. Whether there are\nuniversals, and, if so, in what sense, is a metaphysical question,\nwhich need not be raised in connection with the use of language. The\nonly point about universals that needs to be raised at this point is\nthat the correct use of general words is no evidence that a man can\nthink about universals. It has often been supposed that, because we can\nuse a word like \u201cman\u201d correctly, we must be capable of a corresponding\n\u201cabstract\u201d idea of man, but this is quite a mistake. Some reactions\nare appropriate to one man, some to another, but all have certain\nelements in common. If the word \u201cman\u201d produces in us the reactions\nwhich are common but no others, we may be said to understand the\nword \u201cman\u201d. In learning geometry, one acquires the habit of avoiding\nspecial interpretations of such a word as \u201ctriangle\u201d. We know that,\nwhen we have a proposition about triangles in general, we must not\nthink specially of a right-angled triangle or any one kind of triangle.\nThis is essentially the process of learning to associate with the word\nwhat is associated with _all_ triangles; when we have learnt this,\nwe understand the word \u201ctriangle\u201d. Consequently there is no need to\nsuppose that we ever apprehend universals, although we use general\nwords correctly.\n\nHitherto we have spoken of single words, and among these we have\nconsidered only those that can naturally be employed singly. A child\nuses single words of a certain kind before constructing sentences; but\nsome words presuppose sentences. No one would use the word \u201cpaternity\u201d\nuntil after using such sentences as \u201cJohn is the father of James\u201d; no\none would use the word \u201ccausality\u201d until after using such sentences as\n\u201cthe fire makes me warm\u201d. Sentences introduce new considerations, and\nare not quite so easily explained on behaviourist lines. Philosophy,\nhowever, imperatively demands an understanding of sentences, and we\nmust therefore consider them.\n\nAs we found earlier, all infants outside Patagonia begin with single\nwords, and only achieve sentences later. But they differ enormously in\nthe speed with which they advance from the one to the other. My own two\nchildren adopted entirely different methods. My son first practised\nsingle letters, then single words, and only achieved correct sentences\nof more than three or four words at the age of two and three months.\nMy daughter, on the contrary, advanced very quickly to sentences, in\nwhich there was hardly ever an error. At the age of eighteen months,\nwhen supposed to be sleeping, she was overheard saying to herself:\n\u201cLast year I used to dive off the diving-board, I _did_.\u201d Of course\n\u201clast year\u201d was merely a phrase repeated without understanding. And\nno doubt the first sentences used by children are always repetitions,\nunchanged, of sentences they have heard used by others. Such cases\nraise no new principle not involved in the learning of words. What does\nraise a new principle is the power of putting together known words into\na sentence which has never been heard, but which expresses correctly\nwhat the infant wishes to convey. This involves the power to manipulate\nform and structure. It does not of course involve the apprehension\nof form or structure in the abstract, any more than the use of the\nword \u201cman\u201d involves apprehension of a universal. But it does involve\na causal connection between the form of the stimulus and the form of\nthe reaction. An infant very soon learns to be differently affected by\nthe statement \u201ccats eat mice\u201d from the way he would be affected by the\nstatement \u201cmice eat cats\u201d; and not much later he learns to make one of\nthese statements rather than the other. In such a case, the cause (in\nhearing) or the effect (in speaking) is a whole sentence. It may be\nthat one part of the environment is sufficient to cause one word, while\nanother is sufficient to cause another, but it is only the two parts\nin their relation that can cause the whole sentence. Thus wherever\nsentences come in we have a causal relation between two complex facts,\nnamely the fact asserted and the sentence asserting it; the facts\nas wholes enter into the cause-and-effect relation, which cannot be\nexplained wholly as compounded of relations between their parts.\nMoreover, as soon as the child has learned to use correctly relational\nwords, such as \u201ceat\u201d, he has become capable of being causally affected\nby a relational feature of the environment, which involves a new degree\nof complexity not required for the use of ordinary nouns.\n\nThus the correct use of relational words, _i.e._ of sentences, involves\nwhat may be correctly termed \u201cperception of form\u201d, _i.e._ it involves a\ndefinite reaction to a stimulus which is a form. Suppose, for example,\nthat a child has learnt to say that one thing is \u201cabove\u201d another when\nthis is in fact the case. The stimulus to the use of the word \u201cabove\u201d\nis a relational feature of the environment, and we may say that this\nfeature is \u201cperceived\u201d since it produces a definite reaction. It may\nbe said that the relation _above_ is not very like the word \u201cabove\u201d.\nThat is true; but the same is true of ordinary physical objects. A\nstone, according to the physicists, is not at all like what we see\nwhen we look at it, and yet we may be correctly said to \u201cperceive\u201d it.\nThis, however, is to anticipate. The definite point which has emerged\nis that, when a person can use sentences correctly, that is a proof of\nsensitiveness to formal or relational stimuli.\n\nThe structure of a sentence asserting some relational fact, such as\n\u201cthis is above that\u201d, or \u201cBrutus killed C\u00e6sar\u201d, differs in an important\nrespect from the structure of the fact which it asserts. _Above_ is\na relation which holds between the two terms \u201cthis\u201d and \u201cthat\u201d; but\nthe _word_ \u201cabove\u201d is not a relation. In the sentence the relation is\nthe temporal order of the words (or the spatial order, if they are\nwritten), but the word for the relation is itself as substantial as\nthe other words. In inflected languages, such as Latin, the order of\nthe words is not necessary to show the \u201csense\u201d of the relation; but in\nuninflected languages this is the only way of distinguishing between\n\u201cBrutus killed C\u00e6sar\u201d and \u201cC\u00e6sar killed Brutus\u201d. Words are physical\nphenomena, having spatial and temporal relations; we make use of these\nrelations in our verbal symbolisation of other relations, chiefly to\nshow the \u201csense\u201d of the relation, _i.e._ whether it goes from A to B or\nfrom B to A.\n\nA great deal of the confusion about relations which has prevailed in\npractically all philosophies comes from the fact, which we noticed\njust now, that relations are indicated, not by other relations,\nbut by words which, in themselves, are just like other words.\nConsequently, in thinking about relations, we constantly hover between\nthe unsubstantiality of the relation itself and the substantiality of\nthe word. Take, say, the fact that lightning precedes thunder. If we\nwere to express this by a language closely reproducing the structure\nof the fact, we should have to say simply: \u201clightning, thunder\u201d, where\nthe fact that the first word precedes the second means that what the\nfirst word means precedes what the second word means. But even if we\nadopted this method for temporal order, we should still need words\nfor all other relations, because we could not without intolerable\nambiguity symbolise them also by the order of our words. All this will\nbe important to remember when we come to consider the structure of the\nworld, since nothing but a preliminary study of language will preserve\nus from being misled by language in our metaphysical speculations.\n\nThroughout this chapter I have said nothing about the narrative and\nimaginative uses of words; I have dealt with words in connection\nwith an immediate sensible stimulus closely connected with what they\nmean. The other uses of words are difficult to discuss until we have\nconsidered memory and imagination. In the present chapter I have\nconfined myself to a behaviouristic explanation of the effects of\nwords heard as stimuli, and the causes of words spoken when the words\napply to something sensibly present. I think we shall find that other\nuses of words, such as the narrative and imaginative, involve only new\napplications of the law of association. But we cannot develop this\ntheme until we have discussed several further psychological questions.\n\n\n\n\nCHAPTER V\n\nPERCEPTION OBJECTIVELY REGARDED\n\n\nIt will be remembered that the task upon which we are at present\nengaged is the definition of \u201cknowledge\u201d as a phenomenon discoverable\nby an outside observer. When we have said what we can from this\nobjective standpoint, we will ask ourselves whether anything further,\nand if so what, is to be learnt from the subjective standpoint, in\nwhich we take account of facts which can only be discovered when the\nobserver and the observed are the same person. But for the present we\nwill resolutely confine ourselves to those facts about a human being\nwhich another human being can observe, together with such inferences as\ncan be drawn from these facts.\n\nThe word \u201cknowledge\u201d is very ambiguous. We say that Watson\u2019s rats\n\u201cknow\u201d how to get out of mazes, that a child of three \u201cknows\u201d how to\ntalk, that a man \u201cknows\u201d the people with whom he is acquainted, that\nhe \u201cknows\u201d what he had for breakfast this morning, and that he \u201cknows\u201d\nwhen Columbus first crossed the ocean. French and German are less\nambiguous, since each has two words for different kinds of \u201cknowing\u201d,\nwhich we tend to confuse in our thoughts because we confuse them in our\nlanguage. I shall not attempt as yet to deal with knowledge in general,\nbut rather with certain less general concepts which would ordinarily\nbe included under \u201cknowledge\u201d. And first of all I will deal with\nperception--not as it appears to the perceiver, but as it can be tested\nby an outside observer.\n\nLet us try, first, to get a rough preliminary view of the sort of\nthing we are going to mean by \u201cperception\u201d. One may say that a man\n\u201cperceives\u201d anything that he notices through his senses. This is not\na question of the sense-organs alone, though they are a necessary\ncondition. No man can perceive by sight what is not in his field of\nvision, but he may look straight at a thing without perceiving it. I\nhave frequently had the experience--supposed to be characteristic of\nphilosophers--of looking everywhere for my spectacles although they\nwere before my eyes when my search began. We cannot therefore tell\nwhat a man is perceiving by observing his sense-organs alone, though\nthey may enable us to know that he is _not_ perceiving something.\nThe observer can only know that a man is perceiving something if the\nman reacts in some appropriate manner. If I say to a man \u201cplease\npass the mustard\u201d and he thereupon passes it, it is highly probable\nthat he perceived what I said, although it _may_ of course be a mere\ncoincidence that he passed it at that moment. But if I say to him \u201cthe\ntelephone number you want is 2467\u201d and he proceeds to call that number,\nthe odds against his doing so by mere chance are very great--roughly\n10,000 to 1. And if a man reads aloud out of a book, and I look over\nhis shoulder and perceive the same words, it becomes quite fantastic to\nsuppose that he does not perceive the words he is uttering. We can thus\nin many cases achieve practical certainty as to some of the things that\nother people are perceiving.\n\nPerception is a species of a wider genus, namely _sensitivity_.\nSensitivity is not confined to living things; in fact it is best\nexemplified by scientific instruments. A material object is said to\nbe \u201csensitive\u201d to such and such a stimulus, if, when the stimulus is\npresent, it behaves in a way noticeably different from that in which\nit behaves in the absence of the stimulus. A photographic plate is\nsensitive to light, a barometer is sensitive to pressure, a thermometer\nto temperature, a galvanometer to electric current, and so on. In all\nthese cases, we might say, in a certain metaphorical sense, that an\ninstrument \u201cperceives\u201d the stimulus to which it is sensitive. We do not\nin fact say so; we feel that perception involves something more than we\nfind in scientific instruments. What is this something more?\n\nThe traditional answer would be: consciousness. But this answer,\nright or wrong, is not what we are seeking at the moment, because we\nare considering the percipient as he appears to an outside observer,\nto whom his \u201cconsciousness\u201d is only an inference. Is there anything\nin perception as viewed from without that distinguishes it from the\nsensitivity of a scientific instrument?\n\nThere is, of course, the fact that human beings are sensitive to\na greater variety of stimuli than any instrument. Each separate\nsense-organ can be surpassed by something made artificially sensitive\nto its particular stimulus. Photographic plates can photograph stars\nthat we cannot see; clinical thermometers register differences of\ntemperature that we cannot feel; and so on. But there is no way of\ncombining a microscope, a microphone, a thermometer, a galvanometer,\nand so on, into a single organism which will react in an integral\nmanner to the combination of all the different stimuli that affect its\ndifferent \u201csense-organs\u201d. This, however, is perhaps only a proof that\nour mechanical skill is not so great as it may in time become. It is\ncertainly not enough to define the difference between a dead instrument\nand a living body.\n\nThe chief difference--perhaps the only one from our present point of\nview--is that living bodies are subject to the law of association or\nof the \u201cconditioned reflex\u201d. Consider, for instance, an automatic\nmachine. It has a reflex which makes it sensitive to pennies, in\nresponse to which it gives up chocolate. But it never learns to give\nup chocolate on merely seeing a penny, or hearing the word \u201cpenny\u201d.\nIf you kept it in your house, and said \u201cAbracadabra\u201d to it every time\nyou inserted a penny, it would not in the end be moved to action by\nthe mere word \u201cAbracadabra\u201d. Its reflexes remain unconditioned, as\ndo some of ours, such as sneezing. But with us sneezing is peculiar\nin this respect--hence its unimportance. Most of our reflexes can be\nconditioned, and the conditioned reflex can in turn be conditioned\nafresh, and so on without limit. This is what makes the reactions of\nthe higher animals, and especially of man, so much more interesting\nand complicated than the reactions of machines. Let us see whether\nthis one law will suffice to distinguish perception from other forms of\nsensitivity.\n\nThe variability in a human being\u2019s responses to a given stimulus\nhas given rise to the traditional distinction between cognition and\nvolition. When one\u2019s rich uncle comes for a visit, smiles are the\nnatural response; after he has lost his money, a colder demeanour\nresults from the new conditioning. Thus the reaction to the stimulus\nhas come to be divided into two parts, one purely receptive and\nsensory, the other active and motor. Perception, as traditionally\nconceived, is, so to speak, the end term of the receptive-sensory part\nof the reaction, while volition (in its widest sense) is the first term\nof the active-motor part of the reaction. It was possible to suppose\nthat the receptive part of the reaction would be always the same for\nthe same stimulus, and that the difference due to experience would\nonly arise in the motor part. The last term of the passive part, as it\nappears to the person concerned, was called \u201csensation\u201d. But in fact\nthe influence of the law of conditioned reflexes goes much deeper than\nthis theory supposed. As we saw, the contraction of the pupil, which is\nnormally due to bright light, can be conditioned so as to result from\na loud noise. What we see depends largely upon muscular adjustments\nof the eyes, which we make quite unconsciously. But apart from the\ncontraction of the pupil only one of them is a true reflex, namely\nturning the eyes towards a bright light. This is a movement which\nchildren can perform on the day of their birth; I know this, not merely\nfrom personal observation, but also, what is more, from the text-books.\nBut new-born infants cannot follow a moving light with their eyes, nor\ncan they focus or accommodate. As a consequence, the purely receptive\npart of their reaction to visual objects, in so far as this reaction is\nvisual, is different from that of adults or older children, whose eye\nmuscles adjust themselves so as to see clearly.\n\nBut here again all sorts of factors enter in. Innumerable objects are\nin our field of vision, but only some (at most) are interesting to us.\nIf some one says \u201clook, there\u2019s a snake\u201d, we adjust our eyes afresh\nand obtain a new \u201csensation\u201d. Then, when the purely visual part is\nfinished, there are stimulations, by association, of other centres\nin the brain. There are pictures, in K\u00f6hler\u2019s book, of apes watching\nother apes on the top of insecure piles of boxes, and the spectators\nhave their arms raised in sympathetic balancing movements. Any one\nwho watches gymnastics or skilful dancing is liable to experience\nsympathetic muscular contractions. Any visual object that we might be\ntouching will stimulate incipient touch reactions, but the sun, moon,\nand stars do not.\n\nConversely, visual reactions may be stimulated through association with\nother stimuli. When motor-cars were still uncommon, I was walking one\nday with a friend when a tire punctured in our neighbourhood with a\nloud report. He thought it was a revolver, and averred that he had seen\nthe flash. In dreams, this sort of mechanism operates uncontrolled.\nSome stimulus--say the noise of the maid knocking at the door--becomes\ninterpreted in fantastic ways which are governed by association. I\nremember once dreaming that I was in an inn in the country in Germany\nand was wakened by a choir singing outside my window. Finally I\nreally woke, and found that a spring shower was making a very musical\nnoise on the roof. At least, I heard a very musical noise, and now\nre-interpreted it as a shower on the roof. This hypothesis I confirmed\nby looking out of the window. In waking life we are critical of the\ninterpretative hypotheses that occur to us, and therefore do not make\nsuch wild mistakes as in dreams. But the creative, as opposed to the\ncritical, mechanism is the same in waking life as it is in dreams:\nthere is always far more richness in the experience than the sensory\nstimulus alone would warrant. All adaptation to environment acquired\nduring the life of the individual might be regarded as learning to\ndream dreams that succeed rather than dreams that fail. The dreams we\nhave when we are asleep usually end in a surprise: the dreams we have\nin waking life are less apt to do so. Sometimes they do, as when pride\ngoes before a fall; but in that case they are regarded as showing\nmaladjustment, unless there is some large external cause, such as\nan earthquake. One might say that a person properly adapted to his\nenvironment is one whose dreams never end in the sort of surprise that\nwould wake him up. In that case, he will think that his dreams are\nobjective reality. But if modern physics is to be believed, the dreams\nwe call waking perceptions have only a very little more resemblance to\nobjective reality than the fantastic dreams of sleep. They have some\ntruth, but only just so much as is required to make them useful.\n\nUntil we begin to reflect, we unhesitatingly assume that what we\nsee is really \u201cthere\u201d in the outside world, except in such cases as\nreflections in mirrors. Physics and the theory of the way in which\nperceptions are caused show that this naive belief cannot be quite\ntrue. Perception may, and I think does, enable us to know something of\nthe outer world, but it is not the direct revelation that we naturally\nsuppose it to be. We cannot go into this question adequately until we\nhave considered what the philosopher has to learn from physics; I am\nmerely giving, by anticipation, the reasons for regarding perception\nas a form of reaction to the environment, displayed in some bodily\nmovement, rather than as a form of knowledge. When we have considered\nfurther what constitutes knowledge, we may find that perception is,\nafter all, a form of knowledge, but only because knowledge is not\nquite what we naturally suppose it to be. For the present, let us\nstick to the view of perception that can be obtained by the external\nobserver, _i.e._ as something displayed in the manner of reacting to\nthe environment.\n\nFrom the point of view of the external observer, perception is\nestablished just like any other causal correlation. We observe that,\nwhenever a certain object stands in a certain spatial relation to\na man\u2019s body, the man\u2019s body makes a certain movement or set of\nmovements; we shall then say that the man \u201cperceives\u201d the object. So\nthe new-born baby turns its eyes slowly towards a bright light which\nis not in the centre of the field of vision; this entitles us to say\nthat the baby \u201cperceives\u201d the light. If he is blind, his eyes do not\nmove in this way. A bird flying about in a wood does not bump into the\nbranches, whereas in a room it will bump into the glass of the window.\nThis entitles us to say that the bird perceives the branches but not\nthe glass. Do we \u201cperceive\u201d the glass or do we merely know that it\nis there? This question introduces us to the complications produced\nby association. We know by experience, from the sense of touch, that\nthere is usually glass in window-frames; thus it makes us react to\nthe window-frames as if we could see the glass. But sometimes there\nis no glass, and still we shall perhaps behave as if there were. If\nthis can happen, it shows that we do not perceive the glass, since our\nreaction is the same whether there is glass or not. If, however, the\nglass is coloured, or slightly distorting, or not perfectly clean,\na person accustomed to glass will be able to distinguish a frame\ncontaining glass from one which has none. In that case it is more\ndifficult to decide whether we are to say that he \u201cperceives\u201d the\nglass or not. It is certain that perception is affected by experience.\nA person who can read perceives print where another would not. A\nmusician perceives differences between notes which to an untrained ear\nare indistinguishable. People unaccustomed to the telephone cannot\nunderstand what they hear in it; but this is perhaps not really a case\nin point.\n\nThe difficulty we are considering arises from the fact that a human\nbody, unlike a scientific instrument, is perpetually changing its\nreaction to a given stimulus, under the influence of the law of\nassociation. Moreover, the human body is always doing something.\nHow, then, are we to know whether what it is doing is the result of\na given stimulus or not? In most cases, however, this difficulty is\nnot very serious, particularly when we are dealing with people old\nenough to speak. When you go to the oculist he asks you to read a\nnumber of letters growing gradually smaller; at some point you fail.\nWhere you have succeeded, he knows that you have perceived enough to\nmake out what letter it is. Or you take a pair of compasses and press\nthe points into a man\u2019s back, asking him if he feels two pricks or\nonly one. He may say one when the two points are near together; if he\nis on his guard against this error he may say two when in fact there\nis only one. But if the points are sufficiently far apart he will\nnever make a mistake. That is to say, the bodily movement consisting\nin pronouncing the word \u201ctwo\u201d will invariably result from a certain\nstimulus. (Invariably, I mean, for a given subject on a given day.)\nThis entitles us to say that the man can perceive that there are two\npoints provided they are not too near together. Or you say: \u201cWhat can\nyou see on the horizon?\u201d One man says, \u201cI see a ship\u201d. Another says, \u201cI\nsee a steamer with two funnels\u201d. A third says, \u201cI see a Cunarder going\nfrom Southampton to New York\u201d. How much of what these three people say\nis to count as perception? They may all three be perfectly right in\nwhat they say, and yet we should not concede that a man can \u201cperceive\u201d\nthat the ship is going from Southampton to New York. This, we should\nsay, is inference. But it is by no means easy to draw a line; some\nthings which are, in an important sense, inferential, must be admitted\nto be perceptions. The man who says \u201cI see a ship\u201d is using inference.\nApart from experience, he only sees a queerly shaped dark dot on a blue\nbackground. Experience has taught him that that sort of dot \u201cmeans\u201d a\nship; that is to say, he has a conditioned reflex which causes him to\nutter, aloud or to himself, the word \u201cship\u201d when his eye is stimulated\nin a certain way. To disentangle what is due to experience, and what\nnot, in the perceptions of an adult, is a hopeless task. Practically,\nif a word comes without previous verbal intermediaries, the ordinary\nman would include what the word means in the perception, while he would\nnot do so if the man arrives at the word after verbal preliminaries,\novert or internal. But this is itself a question of familiarity. Show\na child a pentagon, and he will have to count the sides to know how\nmany there are; but after a little experience of geometrical figures,\nthe word \u201cpentagon\u201d will arise without any previous words. And in any\ncase such a criterion is theoretically worthless. The whole affair is a\nmatter of degree, and we cannot draw any sharp line between perception\nand inference. As soon as this is realised, our difficulties are seen\nto be purely verbal and therefore unimportant.\n\nIt will be observed that we are not attempting at present to say what\nconstitutes perception, but only what kind of behaviour on the part\nof a person whom we are observing will justify us in saying that he\nhas perceived this or that feature of his environment. I suggest that\nwe are justified in saying that a man \u201cperceives\u201d such a feature if,\nthroughout some such period as a day, there is some bodily act which\nhe performs whenever that feature is present, but not at any other\ntime. This condition is clearly sufficient, but not necessary--that\nis to say, there may be perception even when it is not fulfilled. A\nman\u2019s reaction may change through conditioning, even in so short a\nperiod as a day. Again, there may be a reaction, but one which is too\nslight to be observable; in this case the criterion of perception\nis theoretically satisfied, but not practically, since no one can\nknow that it is. We often have evidence later on that something was\nperceived, although at the moment there was no discoverable reaction.\nI have frequently known children repeat afterwards some remark which,\nat the time, they seemed not to have heard. This sort of case affords\nanother kind of evidence of perception, namely, the evidence afforded\nby a delayed response. Some people will sit silent and impassive in a\ncompany of talkers, giving no evidence that they are listening; yet\nthey may go home and write down the conversation verbatim in their\njournals. These are the typical writers of memoirs. More remarkable\nstill, I know one man--a man of genius, it is true--who talks\nincessantly, who yet, after meeting a total stranger, knows exactly\nwhat the stranger would have said if he had been given the chance.\nHow this is managed, I do not know; but such a man is rightly called\n\u201cperceptive\u201d.\n\nObviously, in dealing with human beings old enough to talk, words\nafford the best evidence of perception. A man\u2019s verbal responses to\nperceptive situations do not change much after the first few years of\nlife. If you see a kingfisher, and at the same moment your companion\nsays \u201cthere\u2019s a kingfisher\u201d, that is pretty conclusive evidence that\nhe saw it. But, as this case illustrates, our evidence that some one\nelse has perceived something always depends upon our own perceptions.\nAnd our own perceptions are known to us in a different way from that\nin which the perceptions of others are known to us. This is one of\nthe weak spots in the attempt at a philosophy from the objective\nstandpoint. Such a philosophy really assumes knowledge as a going\nconcern, and takes for granted the world which a man derives from\nhis own perceptions. We cannot tackle all our philosophical problems\nby the objective method, but it is worth while to proceed with it as\nfar as it will take us. This whole question of perception will have\nto be attacked afresh from a different angle, and we shall then find\nreason to regard the behaviouristic standpoint as inadequate, though\nvalid so far as it goes. We have still, however, a long road to go\nbefore we shall be driven to consider the subjective standpoint;\nmore particularly, we have to define \u201cknowledge\u201d and \u201cinference\u201d\nbehaviouristically, and then, making a new start, to consider what\nmodern physics makes of \u201cmatter\u201d. But for the moment there are still\nsome things to be said about perception from the objective standpoint.\n\nIt will be seen that, according to our criterion of perception, an\nobject perceived need not be in contact with the percipient\u2019s body. The\nsun, moon, and stars are perceived according to the above criterion.\nIn order, however, that an object not in contact with the body should\nbe perceived, there are physical as well as physiological conditions\nto be fulfilled. There must be some physical process which takes place\nat the surface of the body when the object in question is suitably\nsituated, but not otherwise; and there must be sense-organs capable of\nbeing affected by such a process. There are, as we know from physics,\nmany processes which fulfil the necessary physical conditions, but\nfail to affect us through the inadequacy of our sense-organs. Waves of\na certain sort make sound, but waves of exactly the same sort become\ninaudible if they are too short. Waves of a certain sort make light,\nbut if they are too long or too short they are invisible. The waves\nused in wireless are of the same sort as those that make light, but are\ntoo long. There is no reason _a priori_ why we should not be aware of\nwireless messages through our senses, without the need of instruments.\nX-rays are also of the same sort as those that make light, but in this\ncase they are too short to be seen. They might render the objects from\nwhich they come visible, if we had a different sort of eye. We are not\nsensitive to magnetism, unless it is enormously powerful; but if we\nhad more iron in our bodies, we might have no need of the mariner\u2019s\ncompass. Our senses are a haphazard selection of those that the nature\nof physical processes renders possible; one may suppose that they have\nresulted from chance variation and the struggle for existence.\n\nIt is important to observe that our perceptions are very largely\nconcerned with form or shape or structure. This is the point emphasised\nby what is called \u201c_Gestaltpsychologie_\u201d, or psychology of form.\nReading is a case in point. Whether we read black letters on white\npaper or white letters on a blackboard is a matter which we hardly\nnotice; it is the forms of the letters that affect us, not their colour\nor their size (so long as they remain legible). In this matter, the\nsense of sight is pre-eminent, although blind men (and others to a less\ndegree) can acquire a good knowledge of form by the sense of touch.\n\nAnother point of importance about our perceptions is that they give us,\nwithin limits, a knowledge of temporal sequence. If you say to a man\n\u201cBrutus killed C\u00e6sar\u201d, and then \u201cC\u00e6sar killed Brutus\u201d, the difference\nbetween the two statements is likely to be perceived by him if he\nis listening; in the one case he will say \u201cof course\u201d, in the other\n\u201cnonsense\u201d, which is evidence of his having different perceptions in\nthe two cases, according to our definition. Further, if you ask him\nwhat the difference is, he can tell you that it is a difference in the\norder of the words. Thus time-order within a short period of time is\nclearly perceptible.\n\nThe objective method, which we have been applying in this chapter, is\nthe only possible one in studying the perceptions of animals or of\ninfants before they can talk. Many animals too low in the scale of\nevolution to have eyes are yet sensitive to light, in the sense that\nthey move towards it or move away from it. Such animals, according to\nour criterion, perceive light, though there is no reason to suppose\nthat they perceive colour or visual form or anything beyond the bare\npresence of light. We can perceive the bare presence of light when our\neyes are shut; perhaps one may imagine their sensitiveness to be more\nor less analogous in its limitations.\n\nIt is not to be supposed, in any case, that \u201cperceiving\u201d an object\ninvolves knowing what it is like. That is quite another matter.\nWe shall see later that certain inferences, of a highly abstract\ncharacter, can be drawn from our perceptions to the objects perceived;\nbut these inferences are at once difficult and not quite certain. The\nidea that perception, in itself, reveals the character of objects,\nis a fond delusion, and one, moreover, which it is very necessary to\novercome if our philosophy is to be anything more than a pleasant\nfairy-tale.\n\n\n\n\nCHAPTER VI\n\nMEMORY OBJECTIVELY REGARDED\n\n\nWe are concerned in these chapters with what we can know about other\nmen by merely observing their behaviour. In this chapter, I propose\nto consider everything that would commonly be called \u201cmemory\u201d, in so\nfar as it can be made a matter of external observation. And perhaps it\nmay be as well, at this point, to state my own view of the question\nof \u201cbehaviourism\u201d. This philosophy, of which the chief protagonist is\nDr. John B. Watson, holds that everything that can be known about man\nis discoverable by the method of external observation, _i.e._ that\nnone of our knowledge depends, essentially and necessarily, upon data\nin which the observer and the observed are the same person. I do not\nfundamentally agree with this view, but I think it contains much more\ntruth than most people suppose, and I regard it as desirable to develop\nthe behaviourist method to the fullest possible extent. I believe that\nthe knowledge to be obtained by this method, so long as we take physics\nfor granted, is self-contained, and need not, at any point, appeal\nto data derived from introspection, _i.e._ from observations which a\nman can make upon himself but not upon any one else. Nevertheless,\nI hold that there are such observations and that there is knowledge\nwhich depends upon introspection. What is more, I hold that data of\nthis kind are required for a critical exposition of physics, which\nbehaviourism takes for granted. I shall, therefore, after setting forth\nthe behaviourist view of man, proceed to a scrutiny of our knowledge of\nphysics, returning thence to man, but now as viewed from within. Then,\nfinally, I shall attempt to draw conclusions as to what we know of the\nuniverse in general.\n\nThe word \u201cmemory\u201d or \u201cremembering\u201d is commonly used in a number\nof different senses, which it is important to distinguish. More\nespecially, there is a broad sense, in which the word applies to the\npower of repeating any habitual act previously learnt, and a narrow\nsense, in which it applies only to recollection of past events. It is\nin the broad sense that people speak of a dog remembering his master or\nhis name, and that Sir Francis Darwin spoke of memory in plants. Samuel\nButler used to attribute the sort of behaviour that would usually be\ncalled instinctive to memory of ancestral experience, and evidently he\nwas using the word \u201cmemory\u201d in its widest possible sense. Bergson, on\nthe contrary, dismisses \u201chabit-memory\u201d as not true memory at all. True\nmemory, for him, is confined to the recollection of a past occurrence,\nwhich, he maintains, cannot be a habit, since the event remembered\nonly occurred once. The behaviourist maintains that this contention is\nmistaken, and that all memory consists in the retention of a habit. For\nhim, therefore, memory is not something requiring special study, but\nis merged into the study of habit. Dr. Watson says: \u201cThe behaviourist\nnever uses the term \u2018memory\u2019. He believes that it has no place in an\nobjective psychology.\u201d He proceeds to give instances, beginning with a\nwhite rat in a maze. On the first occasion, he says, it took this rat\nforty minutes to get out of the maze, but after thirty-five trials he\nlearnt to get out in six seconds, without taking any wrong turnings. He\nwas then kept away from the maze for six months, and on being put in it\nagain he got out in two minutes, with six mistakes. He was just as good\nas he had been before at the twentieth trial. We have here a measure of\nthe extent to which the habit of the maze had been retained. A similar\nexperiment with a monkey showed even more retentiveness. He was put\ninto a problem box which at first took him twenty minutes to open, but\nat the twentieth trial he opened it in two seconds. He was then kept\naway from it for six months, and on being put back in it he opened it\nin four seconds.\n\nWith human beings, we know that many of the habits we learn are\nretained through long periods of disuse--skating, bicycling, swimming,\ngolf, etc., are familiar instances. Perhaps Dr. Watson goes a trifle\ntoo far when he says: \u201cIf a poor shot or an inexpert golfer tells you\nthat he was good five years ago but that lack of practice has made him\npoor, don\u2019t believe him; he never was good!\u201d At any rate, this is not\nthe belief of violinists and pianists, who consider it essential to\npractise every day. But even if it be somewhat of an exaggeration, it\nis certainly true that we retain bodily habits pretty well. Some, such\nas swimming, seem to be more completely retained than others. The power\nof talking a foreign language, for example, is one which is greatly\nimpaired by disuse. The whole matter is quantitative, and easily tested\nby experiment.\n\nBut memory in the sense of recollection of past events, if it can be\nexplained as a habit, will have, one might suppose, to be a _verbal_\nhabit. As to this, Dr. Watson says:\n\n\u201cWhat the man on the street ordinarily means by an exhibition of memory\nis what occurs in some such situation as this: An old friend comes to\nsee him, after many years\u2019 absence. The moment he sees this friend,\nhe says: \u2018Upon my life! Addison Sims of Seattle! I haven\u2019t seen you\nsince the World\u2019s Fair in Chicago. Do you remember the gay parties we\nused to have in the Wilderness Hotel? Do you remember the Midway? Do\nyou remember ... etc.,\u2019 _ad infinitum_. The psychology of this process\nis so simple that it seems almost an insult to your intelligence to\ndiscuss it, and yet a good many of the behaviourists\u2019 kindly critics\nhave said that behaviourism cannot adequately explain memory. Let us\nsee if this is a fact.\u201d\n\nHe goes on to say that during the period, long ago, when the man on\nthe street was seeing Mr. Sims, they formed verbal and manual habits\ntowards one another, so that \u201cfinally, just the sight of man, even\nafter months of absence, would call out not only the old verbal\nhabits, but many other types of bodily and visceral responses.\u201d\n\nHe sums up: \u201cBy \u2018memory\u2019, then, we mean nothing except the\nfact that when we meet a stimulus again after an absence, we\ndo the old habitual thing (say the old words and show the old\nvisceral--emotional--behaviour) that we learned to do when we were in\nthe presence of that stimulus in the first place.\u201d\n\nThis theory is preferable to ordinary psychological theories in many\nways. In the first place, it is not an attempt to treat memory as\nsome sort of mystical \u201cfaculty\u201d, and does not suppose that we are\nalways remembering everything that we should remember if a suitable\nstimulus were applied. It is concerned with the causation of specific\nacts of remembering, these acts being all externally observable. I do\nnot see any good reason to question it. Bergson\u2019s contention that the\nrecollection of a unique occurrence cannot be explained by habit is\nclearly fallacious. There are many instances, both with animals and\nwith human beings, of a habit becoming firmly established through one\nexperience. It is, therefore, quite possible that a stimulus associated\nwith a previous occurrence should set going a train of bodily events\nwhich, in turn, produce words describing that occurrence. There is\nhere, however, a difficulty. The memory of a past occurrence cannot\nbe a _verbal_ habit, except when the occurrence has been frequently\nrelated. When Watson\u2019s man on the street says \u201cDo you remember the\nMidway\u201d, he is not using words that have become habitual; very likely\nhe never used these words before. He is using words which a verbal\nhabit associates with an event that is now happening in him, and the\nevent is called up by a habit associated with Mr. Sims. So at least\nwe must suppose, if we accept Watson\u2019s view. But this diminishes the\nplausibility and the verifiability of his view. It is not our actual\nlanguage that can be regarded as habitual, but only what our words\nexpress. In repeating a poem we have learned by heart, the language is\nhabitual, but not so when we recount a past incident in words we never\nused before. In this case, it is not the actual words that we repeat,\nbut only their meaning. The habitual element, therefore, if it really\naccounts for the recollection, must not be sought in words.\n\nThis is something of a difficulty in the Watsonian theory of language.\nWhen a rat learns a maze, it learns certain definite bodily movements;\nso do we when we learn by heart. But I may say to one person, \u201cI met\nMr. Jones in the train to-day\u201d, and to another \u201cJoseph was in the\n9.35 this morning.\u201d With the exception of the words \u201cin the\u201d, these\ntwo sentences have nothing verbally in common, yet they may relate\nthe same fact, and I may use either indifferently when I recall the\nfact. Thus my recollection is certainly _not_ a definite verbal\nhabit. Yet words are the only overt bodily movements by which I make\nknown my recollections to other people. If the behaviourist tells me\nthat my recollection is bodily habit, and begins by telling me that\nit is a _verbal_ habit, he can be driven by such instances to the\nadmission that it must be some other kind of habit. If he says this,\nhe is abandoning the region of observable fact, and taking refuge in\nhypothetical bodily movements invoked to save a theory. But these are\nhardly any better than \u201cthoughts.\u201d\n\nThis question is more general than the problem of memory. Many\ndifferent forms of words may be used to express the same \u201cmeaning\u201d, and\nthere seems no reason in mere habit to account for the fact that we\nsometimes use one form of words and sometimes another when we \u201cthink\u201d\nof that which all the various forms of words express. The association\nseems to go, not direct from stimulus to words, but from stimulus to\n\u201cmeaning\u201d and thence to words expressing the \u201cmeaning\u201d. You may, for\ninstance, be quite unable to recollect whether you were told \u201cJacob is\nolder than Joseph\u201d, or \u201cJoseph is younger than Jacob\u201d, though you may\nremember quite definitely that you were told the fact which both these\nforms of words express. Again, if you are learning, say, a proof of a\nmathematical theorem, you do not learn by heart what the book says,\nunless you are a very bad mathematician; you learn, as people say, to\n\u201cunderstand\u201d the proof, and then you can reproduce it in symbols quite\ndifferent from those in the book. It is such facts, among others, that\nmake it difficult to explain the mechanism of association, whether in\nmemory or in \u201cthought\u201d in general, if we assume that words, or even\nsentences, are the terms associated.\n\nPerhaps, however, the theory as to the \u201cmeaning\u201d of words which we\ndeveloped in an earlier chapter may help us out of the difficulty.\nWe defined the \u201cmeaning\u201d of a word by means of its associations;\ntherefore, if two words are synonyms, they have the same associations;\nand any stimulus which calls up one may also call up the other. The\nquestion which of two synonyms we use will then depend upon some\nextraneous circumstance.\n\nThis is all very well so far as single words are concerned; it would\naccount satisfactorily, for instance, for the fact that I call a man\nsometimes by his surname and sometimes by his Christian name. But it\nis hardly so adequate when we come to the question of sentences. To\nrevert to the illustration of a moment ago, in response to the stimulus\n\u201cDid anything happen on your journey?\u201d you may say either \u201cI met Mr.\nJones in the train to-day\u201d, or \u201cJoseph was in the 9.35 this morning\u201d,\nor any one of an indefinite number of other sentences expressive of the\nsame occurrence. Are we to suppose that, while you were in the train,\nyou were rehearsing all these different sentences to yourself, so that\neach of them became firmly associated with the words \u201cjourney to-day\u201d?\nClearly such a supposition would be absurd. Yet all the separate words\nof your sentence have many other associations; it is only the sentence\nas a whole that is associated with your journey. You have met other\npeople besides Mr. Jones; you have had other contacts with Mr. Jones\nbesides meeting him this morning; \u201ctrain\u201d and \u201cto-day\u201d equally are\nappropriate to other occurrences that you might relate. Thus it has\nto be the whole sentence that is the associative unit, and yet the\nsentence may never have been in your head before. It seems clear that\nit is possible to state in words something that you remember, although\nyou never put it into words before. Suppose I say \u201cWhat did you have\nfor breakfast to-day?\u201d Probably you will be able to tell me, though\nit is very likely that you have not given names to the things you ate\nuntil this moment.\n\nThis whole matter is connected with the distinction between sentences\nand single words, which we found important when we were discussing\nlanguage. But even when we confine ourselves to single words, there\nare difficulties in Dr. Watson\u2019s view. Cases are alleged in which\nchildren, after learning to speak, can recall incidents which occurred\nbefore they could speak, and describe them in correct words. This would\nshow that the memory had persisted in a non-verbal form throughout\nthe period before they learned to speak, and had only subsequently\nfound verbal expression. Such extreme incidents are rare and might be\nquestioned, but in a less extreme form it ought not to be difficult to\nobtain examples of the same sort of thing. Suppose, for example, that a\nyoung child hurt his wrist badly before he knew the word \u201cwrist\u201d, and\nthat some time afterwards he learnt it; I should not be surprised if he\ncould relate that he had hurt his wrist. Such instances, however, would\nnot refute the essence of Watson\u2019s theory. He would allow \u201cvisceral\u201d\nmemory, for example, and the association with the word \u201cwrist\u201d might\nbe grafted on to this. The real difficulty in Dr. Watson\u2019s view, to my\nmind, is the fact that our sentence may vary verbally as much as it\nlikes so long as it retains the same \u201cmeaning\u201d, and that we clearly do\nnot rehearse to ourselves beforehand all the possible sentences having\nthe \u201cmeaning\u201d in question.\n\nIt should be realised that behaviourism loses much of its\nattractiveness if it is compelled to postulate movements that no one\ncan observe and that there is no other reason to assume. Dr. Broad,\nin his book on _The Mind and its Place in Nature_, distinguishes\nbetween \u201cmolar\u201d and \u201cmolecular\u201d behaviourism: the former assumes only\nsuch bodily movements as can be observed, while the latter allows\nand utilises hypothetical minute movements, more especially in the\nbrain. Now here we must make a distinction. Physics believes in a large\nnumber of phenomena which are too minute to be observed even with the\nstrongest microscope, and if physics is at all correct, there must be\nminute movements in all parts of a human body, of a sort which we can\nnever hope to see. We cannot reasonably demand of the behaviourist that\nhe should abstain from an hypothesis which physics asserts for very\ngood reasons. And in the process which leads from stimulus to reaction\nthere are bound to be small occurrences in the brain, which, though\nthey cannot be observed, are essential to the physiological explanation\nof what occurs. But when the behaviourist assumes small occurrences\nfor which there is no ground in physics, and which are needed solely\nin order to safeguard his theory, he is in a less strong position.\nDr. Watson asserts, for instance, that whenever we \u201cthink\u201d there are\nsmall movements in the larynx which are beginnings of the movements we\nshould make as if we spoke words out loud. It may be that this is true;\ncertainly I am not prepared to deny it. But I am not prepared to say\nthat it _must_ be true merely because, if it were not, behaviourism\nwould be false. We do not know in advance that behaviourism is true;\nwe have to find out whether it will explain observed facts. Whenever\nit has to postulate something unobserved merely in order to avoid a\nrefutation, it weakens its case. And if it maintains, as, from Dr.\nWatson\u2019s language, it seems to do, that we only remember an occurrence\nby forming a verbal habit in connection with it, then it is obliged to\npostulate much implicit use of words of which we have no evidence.\n\nTo sum up this discussion. While it is quite possible, by behaviourist\nmethods, to ascertain whether a person remembers a past occurrence or\nnot, unless he is deliberately obstructing the observer, and while\nmuch memory can be quite adequately explained as habit, there do seem\nto be great difficulties in the view that memory consists _entirely_\nof habit, at least in the case of the recollection of an event. These\ndifficulties seem insuperable if we suppose memory to be _essentially_\na verbal habit. They are not insuperable if we postulate sufficient\nminute unobservable bodily movements. We have not considered whether\nthey can be overcome by introducing data derived from introspection,\nsince we wish, for the present, to maintain a strictly objective\nattitude to human behaviour. The introspective discussion of memory\nwill be taken up at a later stage.\n\n\n\n\nCHAPTER VII\n\nINFERENCE AS A HABIT\n\n\nIn this chapter, we are concerned with inference as it can be observed\nwhen practised by some one else. Inference is supposed to be a mark of\nintelligence and to show the superiority of men to machines. At the\nsame time, the treatment of inference in traditional logic is so stupid\nas to throw doubt on this claim, and syllogistic inference, which was\ntaken as the type from Aristotle to Bacon (exclusive), is just the sort\nof thing that a calculating machine could do better than a professor.\nIn syllogistic inference, you are supposed to know already that all\nmen are mortal and that Socrates is a man; hence you deduce, what you\nnever suspected before, that Socrates is mortal. This form of inference\ndoes actually occur, though very rarely. The only instance I have ever\nheard of was supplied by Dr. F.\u00a0C.\u00a0S. Schiller. He once produced a\ncomic number of the philosophical periodical _Mind_, and sent copies to\nvarious philosophers, among others to a certain German, who was much\npuzzled by the advertisements. But at last he argued: \u201cEverything in\nthis book is a joke, therefore the advertisements are jokes\u201d. I have\nnever come across any other case of new knowledge obtained by means of\na syllogism. It must be admitted that, for a method which dominated\nlogic for two thousand years, this contribution to the world\u2019s stock of\ninformation cannot be considered very weighty.\n\nThe inferences that we actually make in daily life differ from those\nof syllogistic logic in two respects, namely, that they are important\nand precarious, instead of being trivial and safe. The syllogism may be\nregarded as a monument to academic timidity: if an inference might be\nwrong, it was dangerous to draw it. So the medi\u00e6val monks, in their\nthinking as in their lives, sought safety at the expense of fertility.\n\nWith the Renaissance, a more adventurous spirit came into the world,\nbut at first in philosophy, it only took the form of following Greeks\nother than Aristotle, and more especially Plato. It is only with Bacon\nand Galileo that the inductive method arrived at due recognition:\nwith Bacon as a programme which was largely faulty, but with Galileo\nas something which actually led to brilliant results, namely, the\nfoundation of modern mathematical physics. Unfortunately, when the\npedants got hold of induction, they set to work to make it as tame and\nscholastic as deduction had been. They searched for a way of making\nit _always_ lead to true results, and in so doing robbed it of its\nadventurous character. Hume turned upon them with sceptical arguments,\nproving quite conclusively that if an induction is worth making it may\nbe wrong. Thereupon Kant deluged the philosophic world with muddle and\nmystery, from which it is only now beginning to emerge. Kant has the\nreputation of being the greatest of modern philosophers, but to my mind\nhe was a mere misfortune.\n\nInduction, as it appears in the text-books, consists, roughly\nspeaking, in the inference that, because A and B have been found often\ntogether and never apart, therefore they are probably always together,\nand either may be taken as a sign of the other. I do not wish, at\nthis stage, to examine the logical justification of this form of\nargumentation; for the present, I am considering it as a _practice_,\nwhich we can observe in the habits of men and animals.\n\nAs a practice, induction is nothing but our old friend, the law of\nconditioned reflexes or of association. A child touches a knob that\ngives him an electric shock; after that, he avoids touching the knob.\nIf he is old enough to speak he may state that the knob hurts when it\nis touched; he has made an induction based on a single instance. But\nthe induction will exist as a bodily habit even if he is too young\nto speak, and it occurs equally among animals, provided they are not\ntoo low in the scale. The theories of induction in logic are what\nFreudians call a \u201crationalisation\u201d; that is to say, they consist of\nreasons invented afterwards to prove that what we have been doing is\nsensible. It does not follow that they are bad reasons; in view of the\nfact that we and our ancestors have managed to exist since the origin\nof life, our behaviour and theirs must have been fairly sensible, even\nif we and they were unable to prove that it was. This, however, is not\nthe point that concerns us at present. What concerns us at present is\nthe fact that verbal induction is a late development of induction in\nbehaviour, which is nothing more or less than the principle of \u201clearned\nreactions\u201d.\n\nThis principle, as the reader will remember, states that, if a\ncertain event calls out a certain response, and if another event is\nexperienced just before it, or at the same moment, in time that other\nevent will tend to call out the response which, originally, only\nthe first event would call out. This applies both to muscles and to\nglands; it is because it applies to glands that words are capable of\ncausing emotions. Moreover, we cannot set limits to the length of the\nchain of associations that may be established. If you hold an infant\u2019s\nlimbs, you call out a rage reaction; this appears to be an \u201cunlearned\nreaction\u201d. If you, and no one else, repeatedly hold an infant\u2019s limbs,\nthe mere sight of you will call out a rage reaction after a time.\nWhen the infant learns to talk your name may have the same effect.\nIf, later, he learns that you are an optician, he may come to hate\nall opticians; this may lead him to hate Spinoza because he made\nspectacles, and thence he may come to hate metaphysicians and Jews.\nFor doing so he will no doubt have the most admirable reasons, which\nwill seem to him to be his real ones; he will never suspect the process\nof conditioning by which he has in fact arrived at his enthusiasm for\nthe Ku Klux Klan. This is an example of conditioning in the emotional\nsphere; but it is rather in the muscular sphere that we must seek the\norigin of the practice of induction.\n\nDomestic animals which are habitually fed by a certain person will run\ntowards that person as soon as they see him. We say that they expect\nfood, and in fact their behaviour is very like what it would be if\nthey saw food. But really we have only an example of \u201cconditioning\u201d:\nthey have often seen first the farmer and then the food, so that in\ntime they react to the farmer as they originally reacted to the food.\nInfants soon learn to react to the sight of the bottle, although at\nfirst they only react to the touch of it. When they can speak, the same\nlaw makes them say \u201cdinner\u201d when they hear the dinner-bell. It is quite\nunnecessary to suppose that they first think \u201cthat bell means dinner\u201d,\nand then say \u201cdinner\u201d. The sight of dinner (by previous \u201clearned\nreaction\u201d) causes the word \u201cdinner\u201d: the bell frequently precedes the\nsight of dinner; therefore in time the bell produces the word \u201cdinner\u201d.\nIt is only subsequent reflection, probably at a much later age, that\nmakes the child say \u201cI knew dinner was ready because I heard the bell\u201d.\nLong before he can say this, he is acting as if he knew it. And there\nis no good reason for denying that he knows it, when he acts as if he\ndid. If knowledge is to be displayed by behaviour, there is no reason\nto confine ourselves to _verbal_ behaviour as the sole kind by which\nknowledge can manifest itself.\n\nThe situation, stated abstractly, is as follows. Originally, stimulus\nA produced reaction C; now stimulus B produces it, as a result of\nassociation. Thus B has become a \u201csign\u201d of A, in the sense that it\ncauses the behaviour appropriate to A. All sorts of things may be signs\nof other things, but with human beings words are the supreme example of\nsigns. All signs depend upon some practical induction. Whenever we read\nor hear a statement, its effect upon us depends upon induction in this\nsense, since the words are signs of what they mean, in the sense that\nwe react to them, in certain respects, as we should to what they stand\nfor. If some one says to you \u201cyour house is on fire\u201d, the effect upon\nyou is practically the same as if you saw the conflagration. You may,\nof course, be the victim of a hoax, and in that case your behaviour\nwill not be such as achieves any purpose you have in view. This risk\nof error exists always, since the fact that two things have occurred\ntogether in the past cannot prove conclusively that they will occur\ntogether in the future.\n\nScientific induction is an attempt to regularise the above process,\nwhich we may call \u201cphysiological induction\u201d. It is obvious that, as\npractised by animals, infants, and savages, physiological induction is\na frequent source of error. There is Dr. Watson\u2019s infant who induced,\nfrom two examples, that whenever he saw a certain rat there would be\na loud noise. There is Edmund Burke, who induced from one example\n(Cromwell) that revolutions lead to military tyrannies. There are\nsavages who argue, from one bad season, that the arrival of a white\nman causes bad crops. The inhabitants of Siena, in 1348, thought\nthat the Black Death was a punishment for their pride in starting to\nbuild too large a cathedral. Of such examples there is no end. It is\nvery necessary, therefore, if possible, to find some method by which\ninduction can be practised so as to lead, in general, to correct\nresults. But this is a problem of scientific method, with which we will\nnot yet concern ourselves.\n\nWhat does concern us at present is the fact that all inference, of\nthe sort that really occurs, is a development of this one principle\nof conditioning. In practice, inference is of two kinds, one typified\nby induction, the other by mathematical reasoning. The former is by\nfar the more important, since, as we have seen, it covers all use of\nsigns and all empirical generalisations as well as the habits of which\nthey are the verbal expression. I know that, from the traditional\nstandpoint, it seems absurd to talk of inference in most cases of this\nsort. For example, you find it stated in the paper that such and such a\nhorse has won the Derby. According to my own use of words, you practise\nan induction when you arrive thence at the belief that that horse has\nwon. The stimulus consists of certain black marks on white paper--or\nperhaps on pink paper. This stimulus is only connected with horses\nand the Derby by association, yet your reaction is one appropriate\nto the Derby. Traditionally, there was only inference where there\nwas a \u201cmental process\u201d, which, after dwelling upon the \u201cpremisses\u201d,\nwas led to assert the \u201cconclusion\u201d by means of insight into their\nlogical connection. I am not saying that the process which such words\nas the above are intended to describe never takes place; it certainly\ndoes. What I am saying is that, genetically and causally, there is no\nimportant difference between the most elaborate induction and the most\nelementary \u201clearned reaction\u201d. The one is merely a developed form of\nthe other, not something radically different. And our determination to\nbelieve in the results of inductions, even if, as logicians, we see\nno reason to do so, is really due to the potency of the principle of\nassociation; it is an example--perhaps the most important example--of\nwhat Dr. Santayana calls \u201canimal faith\u201d.\n\nThe question of mathematical reasoning is more difficult. I think we\nmay lay it down that, in mathematics, the conclusion always asserts\nmerely the whole or part of the premisses, though usually in new\nlanguage. The difficulty of mathematics consists in seeing that\nthis is so in particular cases. In practice, the mathematician has\na set of rules according to which his symbols can be manipulated,\nand he acquires technical skill in working according to the rules\nin the same sort of way as a billiard-player does. But there is a\ndifference between mathematics and billiards: the rules of billiards\nare arbitrary, whereas in mathematics some at least are in some sense\n\u201ctrue\u201d. A man cannot be said to understand mathematics unless he\nhas \u201cseen\u201d that these rules are right. Now what does this consist\nof? I think it is only a more complicated example of the process of\nunderstanding that \u201cNapoleon\u201d and \u201cBonaparte\u201d refer to the same person.\nTo explain this, however, we must revert to what was said, in the\nchapter on \u201cLanguage\u201d, about the understanding of form.\n\nHuman beings possess the power of reacting to form. No doubt some of\nthe higher animals also possess it, though to nothing like the same\nextent as men do; and all animals, except a few of the most intelligent\nspecies, appear to be nearly devoid of it. Among human beings, it\ndiffers greatly from one individual to another, and increases, as a\nrule, up to adolescence. I should take it as what chiefly characterises\n\u201cintellect\u201d. But let us see, first, in what the power consists.\n\nWhen a child is being taught to read, he learns to recognise a given\nletter, say H, whether it is large or small, black or white or red.\nHowever it may vary in these respects his reaction is the same: he\nsays \u201cH\u201d. That is to say, the essential feature in the stimulus is its\n_form_. When my boy, at the age of just under three, was about to eat a\nthree-cornered piece of bread and butter, I told him it was a triangle.\n(His slices were generally rectangular.) Next day, unprompted, he\npointed to triangular bits in the pavement of the Albert Memorial, and\ncalled them \u201ctriangles\u201d. Thus the form of the bread and butter, as\nopposed to its edibility, its softness, its colour, etc., was what had\nimpressed him. This sort of thing constitutes the most elementary kind\nof reaction to form.\n\nNow \u201cmatter\u201d and \u201cform\u201d can be placed, as in the Aristotelian\nphilosophy, in a hierarchy. From a triangle we can advance to a\npolygon, thence to a figure, thence to a manifold of points. Then we\ncan go on and turn \u201cpoint\u201d into a formal concept, meaning \u201csomething\nthat has relations which resemble spatial relations in certain formal\nrespects\u201d. Each of these is a step away from \u201cmatter\u201d and further into\nthe region of \u201cform\u201d. At each stage the difficulty increases. The\ndifficulty consists in having a uniform reaction (other than boredom)\nto a stimulus of this kind. When we \u201cunderstand\u201d a mathematical\nexpression, that means that we can react to it in an appropriate\nmanner, in fact, that it has \u201cmeaning\u201d for us. This is also what we\nmean by \u201cunderstanding\u201d the word \u201ccat\u201d. But it is easier to understand\nthe word \u201ccat\u201d, because the resemblances between different cats are\nof a sort which causes even animals to have a uniform reaction to all\ncats. When we come to algebra, and have to operate with _x_ and _y_,\nthere is a natural desire to know what _x_ and _y_ really are. That,\nat least, was my feeling: I always thought the teacher knew what they\nreally were, but would not tell me. To \u201cunderstand\u201d even the simplest\nformula in algebra, say (x + y)\u00b2 = x\u00b2 + 2xy + y\u00b2, is to be able to\nreact to two sets of symbols in virtue of the form which they express,\nand to perceive that the form is the same in both cases. This is a\nvery elaborate business, and it is no wonder that boys and girls find\nalgebra a bugbear. But there is no novelty _in principle_ after the\nfirst elementary perceptions of form. And perception of form consists\nmerely in reacting alike to two stimuli which are alike in form but\nvery different in other respects. For, when we can do that, we can say,\non the appropriate occasion, \u201cthat is a triangle\u201d; and this is enough\nto satisfy the examiner that we know what a triangle is, unless he is\nso old-fashioned as to expect us to reproduce the verbal definition,\nwhich is of course a far easier matter, in which, with patience, we\nmight teach even a parrot to succeed.\n\nThe meanings of complex mathematical symbols are always fixed by rules\nin relation to the meaning of simpler symbols; thus their meanings are\nanalogous to those of sentences, not to those of single words. What was\nsaid earlier about the understanding of sentences applies, therefore,\nto any group of symbols which, in mathematics, will be declared to have\nthe same meaning as another group, or part of that meaning.\n\nWe may sum up this discussion by saying that mathematical inference\nconsists in attaching the same reactions to two different groups of\nsigns, whose meanings are fixed by convention in relation to their\nconstituent parts, whereas induction consists, first, in taking\nsomething as a sign of something else, and later, when we have learned\nto take A as a sign of B, in taking A as also a sign of C. Thus the\nusual cases of induction and deduction are distinguished by the fact\nthat, in the former, the inference consists in taking one sign as\na sign of two different things, while in the latter the inference\nconsists in taking two different signs as signs of the same thing.\nThis statement is a little too antithetical to be taken as an exact\nexpression of the whole truth in the matter. What is true, however, is\nthat both kinds of inferences are concerned with the relation of a sign\nto what it signifies, and therefore come within the scope of the law of\nassociation.\n\n\n\n\nCHAPTER VIII\n\nKNOWLEDGE BEHAVIOURISTICALLY CONSIDERED\n\n\nThe word \u201cknowledge\u201d, like the word \u201cmemory\u201d, is avoided by the\nbehaviourist. Nevertheless there is a phenomenon commonly called\n\u201cknowledge\u201d, which is tested behaviouristically in examinations. I want\nto consider this phenomenon in this chapter, with a view to deciding\nwhether there is anything in it that the behaviourist cannot deal with\nadequately.\n\nIt will be remembered that, in Chapter II, we were led to the view that\nknowledge is a characteristic of the complete process from stimulus to\nreaction, or even, in the cases of sight and hearing, from an external\nobject to a reaction, the external object being connected with the\nstimulus by a chain of physical causation in the outer world. Let us,\nfor the moment, leave on one side such cases as sight and hearing, and\nconfine ourselves, for the sake of definiteness, to knowledge derived\nfrom touch.\n\nWe can observe touch producing reactions in quite humble animals, such\nas worms and sea anemones. Are we to say that they have \u201cknowledge\u201d\nof what they touch? In some sense, yes. Knowledge is a matter of\ndegree. When it is regarded in a purely behaviouristic manner, we\nshall have to concede that it exists, in some degree, wherever there\nis a characteristic reaction to a stimulus of a certain kind, and this\nreaction does not occur in the absence of the right kind of stimulus.\nIn this sense, \u201cknowledge\u201d is indistinguishable from \u201csensitivity\u201d,\nwhich we considered in connection with perception. We might say that\na thermometer \u201cknows\u201d the temperature, and that a compass \u201cknows\u201d the\ndirection of the magnetic north. This is the only sense in which, on\ngrounds of observation, we can attribute knowledge to animals that\nare low in the scale. Many animals, for example, hide themselves when\nexposed to light, but as a rule not otherwise. In this, however, they\ndo not differ from a radiometer. No doubt the mechanism is different,\nbut the observed molar motion has similar characteristics. Wherever\nthere is a reflex, an animal may be said, in a sense, to \u201cknow\u201d the\nstimulus. This is, no doubt, not the usual sense of \u201cknowledge\u201d, but it\nis the germ out of which knowledge in the usual sense has grown, and\nwithout which no knowledge would be possible.\n\nKnowledge in any more advanced sense is only possible as a result of\nlearning, in the sense considered in Chapter III. The rat that has\nlearned the maze \u201cknows\u201d the way out of it; the boy who has learned\ncertain verbal reactions \u201cknows\u201d the multiplication table. Between\nthese two cases there is no important difference. In both cases, we\nsay that the subject \u201cknows\u201d something because he reacts in a manner\nadvantageous to himself, in which he could not react before he had had\ncertain experiences. I do not think, however, that we ought to use such\na notion as \u201cadvantageous\u201d in connection with knowledge. What we can\nobserve, for instance, with the rat in the maze, is violent activity\nuntil the food is reached, followed by eating when it is reached; also\na gradual elimination of acts which do not lead to the food. Where this\nsort of behaviour is observed, we may say that it is directed towards\nthe food, and that the animal \u201cknows\u201d the way to the food when he gets\nto it by the shortest possible route.\n\nBut if this view is right, we cannot define any knowledge acquired\nby learning except with reference to circumstances toward which an\nanimal\u2019s activity is directed. We should say, popularly, that the\nanimal \u201cdesires\u201d such circumstances. \u201cDesire\u201d, like \u201cknowledge\u201d,\nis capable of a behaviouristic definition, and it would seem that\nthe two are correlative. Let us, then, spend a little time on the\nbehaviouristic treatment of \u201cdesire\u201d.\n\nThe best example of desire, from this point of view, is hunger. The\nstimulus to hunger is a certain well-ascertained bodily condition.\nWhen in this condition, an animal moves restlessly; if he sees or\nsmells food, he moves in a manner which, in conditions to which he is\naccustomed, would bring him to the food; if he reaches it, he eats it,\nand if the quantity is sufficient he then becomes quiescent. This kind\nof behaviour may be summarised by saying that a hungry animal \u201cdesires\u201d\nfood. It is behaviour which is in various ways different from that of\ninanimate matter, because restless movements persist until a certain\ncondition is realised. These movements may or may not be the best\nadapted to realising the condition in question. Every one knows about\nthe pike that was put on one side of a glass partition, with minnows\non the other side. He continually bumped his nose on the glass, and\nafter six weeks gave up the attempt to catch them. When, after this,\nthe partition was removed, he still refrained from pursuing them. I do\nnot know whether the experiment was tried of leaving a possibility of\ngetting to the minnows by a roundabout route. To have learned to take a\nroundabout route would perhaps have required a degree of intelligence\nbeyond the capacity of fishes; this is a matter, however, which offers\nlittle difficulty to dogs or monkeys.\n\nWhat applies to hunger applies equally to other forms of \u201cdesire\u201d.\nEvery animal has a certain congenital apparatus of \u201cdesires\u201d; that\nis to say, in certain bodily conditions he is stimulated to restless\nactivities which tend towards the performance of some reflex, and\nif a given situation is often repeated the animal arrives more and\nmore quickly at the performance of the reflex. This last, however, is\nonly true of the higher animals; in the lower, the whole process from\nbeginning to end is reflex, and can therefore only succeed in normal\ncircumstances. The higher animals, and more especially men, have a\nlarger proportion of learning and a smaller proportion of reflexes in\ntheir behaviour, and are therefore better able to adapt themselves\nto new circumstances. The helplessness of infants is a necessary\ncondition for the adaptability of adults; infants have fewer useful\nreflexes than the young of animals, but have far more power of forming\nuseful habits, which can be adapted to circumstances and are not\nfatally fixed from birth. This fact is intimately connected with the\nsuperiority of the human intelligence above that of the brutes.\n\nDesire is extremely subject to \u201cconditioning\u201d. If A is a primitive\ndesire and B has on many occasions been a means to A, B comes to be\ndesired in the same sense in which A was previously desired. It may\neven happen, as in misers, that the desire for B completely displaces\nthe desire for A, so that B, when attained, is no longer used as a\nmeans to A. This, however, is more or less exceptional. In general, the\ndesire for A persists, although the desire for B has a more or less\nindependent life.\n\nThe \u201cconditioning\u201d of primitive desires in human beings is the source\nof much that distinguishes our life from that of animals. Most animals\nonly seek food when they are hungry; they may, then, die of starvation\nbefore they find it. Men, on the contrary, must have early acquired\npleasure in hunting as an art, and must have set out on hunting\nexpeditions before they were actually hungry. A further stage in the\nconditioning of hunger came with domestic animals; a still further\nstage with agriculture. Nowadays, when a man sets to work to earn his\nliving, his activity is still connected, though not very directly,\nwith hunger and the other primitive desires that can be gratified by\nmeans of money. These primitive desires are still, so to speak, the\npower station, though their energy is widely distributed to all sorts\nof undertakings that seem, at first sight, to have no connection with\nthem. Consider \u201cfreedom\u201d and the political activities it inspires; this\nis derivable, by \u201cconditioning\u201d, from the emotion of rage which Dr.\nWatson observed in infants whose limbs are not \u201cfree\u201d. Again we speak\nof the \u201cfall\u201d of empires and of \u201cfallen\u201d women; this is connected with\nthe fear which infants display when left without support.\n\nAfter this excursion into the realm of desire, we can now return to\n\u201cknowledge\u201d, which, as we saw, is a term correlative to \u201cdesire\u201d, and\napplicable to another feature of the same kind of activity. We may say,\nbroadly, that a response to a stimulus of the kind involving desire\nin the above sense shows \u201cknowledge\u201d if it leads by the quickest or\neasiest route to the state of affairs which, in the above sense, is\nbehaviouristically the object of desire. Knowledge is thus a matter\nof degree: the rat, during its progressive improvements in the maze,\nis gradually acquiring more and more knowledge. Its \u201cintelligence\nquotient\u201d, so far as that particular task is concerned, will be the\nratio of the time it took on the first trial to the time it takes now\nto get out of the maze. Another point, if our definition of knowledge\nis accepted, is, that there is no such thing as purely contemplative\nknowledge: knowledge exists only in relation to the satisfaction of\ndesire, or, as we say, in the capacity to choose the right means to\nachieve our ends.\n\nBut can such a definition as the above really stand? Does it represent\nat all the sort of thing that would commonly be called knowledge? I\nthink it does in the main, but there is need of some discussion to make\nthis clear.\n\nIn some cases, the definition is obviously applicable. These are the\ncases that are analogous to the rat in the maze, the consideration of\nwhich led us to our definition. Do you \u201cknow\u201d the way from Trafalgar\nSquare to St. Pancras? Yes, if you can walk it without taking any wrong\nturnings. In practice, you can give verbal proof of such knowledge,\nwithout actually having to walk the distance; but that depends upon\nthe correlation of names with streets, and is part of the process of\nsubstituting words for things. There may, it is true, come doubtful\ncases. I was once on a bus in Whitehall, and my neighbour asked \u201cWhat\nstreet is this?\u201d I answered him, not without surprise at his ignorance.\nHe then said, \u201cWhat building is that?\u201d and I replied \u201cThe Foreign\nOffice\u201d. To this he retorted, \u201cbut I thought the Foreign Office was in\nDowning Street\u201d. This time, it was his knowledge that surprised me.\nShould we say that he knew where the Foreign Office is? The answer is\nyes or no according to his purpose. From the point of view of sending a\nletter to it, he knew; from the point of view of walking to it, he did\nnot know. He had, in fact, been a British Consul in South America, and\nwas in London for the first time.\n\nBut now let us come to cases less obviously within the scope of our\ndefinition. The reader \u201cknows\u201d that Columbus crossed the ocean in\n1492. What do we mean by saying that he \u201cknows\u201d this? We mean, no\ndoubt, primarily that writing down this statement is the way to pass\nexaminations, which is just as useful to us as getting out of the\nmaze is to the rat. But we do not mean only this. There is historical\nevidence of the fact, at least I suppose there is. The historical\nevidence consists of printed books and manuscripts. Certain rules have\nbeen developed by historians as to the conditions in which statements\nin books or manuscripts may be accepted as true, and the evidence in\nour case is (I presume) in accordance with these rules. Historical\nfacts often have importance in the present; for example, wills, or\nlaws not yet repealed. The rules for weighing historical evidence\nare such as will, in general bring out self-consistent results. Two\nresults are self-consistent when, in relation to a desire to which both\nare relevant, they enjoin the same action, or actions which can form\npart of the one movement towards the goal. At Coton, near Cambridge,\nthere is (or was in my time) a signpost with two arms pointing in\ndiametrically opposite directions, and each arm said \u201cTo Cambridge\u201d.\nThis was a perfect example of self-contradiction, since the two arms\nmade statements representing exactly opposite actions. And this case\nillustrates why self-contradiction is to be avoided. But the avoidance\nof self-contradiction makes great demands upon us; Hegel and Bradley\nimagined that we could know the nature of the universe by means of\nthis principle alone. In this they were pretty certainly mistaken,\nbut nevertheless a great deal of our \u201cknowledge\u201d depends upon this\nprinciple to a greater or less extent.\n\nMost of our knowledge is like that in a cookery book, maxims to be\nfollowed when occasion arises, but not useful at every moment of every\nday. Since knowledge may be useful at any time, we get gradually,\nthrough conditioning, a general desire for knowledge. The learned man\nwho is helpless in practical affairs is analogous to the miser, in that\nhe has become absorbed in a means. It should be observed, also, that\nknowledge is neutral as among different purposes. If you know that\narsenic is a poison, that enables you equally to avoid it if you wish\nto remain in health, and to take it if you wish to commit suicide. You\ncannot judge from a man\u2019s conduct in relation to arsenic whether he\nknows that it is a poison or not, unless you know his desires. He may\nbe tired of life, but avoid arsenic because he has been told that it\nis a good medicine; in this case, his avoidance of it is evidence of\n_lack_ of knowledge.\n\nBut to return to Columbus: surely, the reader will say, Columbus really\ndid cross the Atlantic in 1492, and that is why we call this statement\n\u201cknowledge\u201d. This is the definition of \u201ctruth\u201d as \u201ccorrespondence with\nfact\u201d. I think there is an important element of correctness in this\ndefinition, but it is an element to be elicited at a later stage,\nafter we have discussed the physical world. And it has the defect--as\npragmatists have urged--that there seems no way of getting at \u201cfacts\u201d\nand comparing them with our beliefs: all that we ever reach consists of\nother beliefs. I do not offer our present behaviouristic and pragmatic\ndefinition of \u201cknowledge\u201d as the only possible one, but I offer it as\nthe one to which we are led if we wish to regard knowledge as something\ncausally important, to be exemplified in our reactions to stimuli. This\nis the appropriate point of view when we are studying man from without,\nas we have been doing hitherto.\n\nThere is, however, within the behaviourist philosophy, one important\naddition to be made to our definition. We began this chapter with\nsensitivity, but we went on to the consideration of learned reactions,\nwhere the learning depended upon association. But there is another\nsort of learning--at least it is _prima facie_ another sort--which\nconsists of increase of sensitivity. All sensitivity in animals and\nhuman beings must count as a sort of knowledge; that is to say, if an\nanimal behaves, in the presence of a stimulus of a certain kind, as it\nwould not behave in the absence of that stimulus then, in an important\nsense, it has \u201cknowledge\u201d as regards the stimulus. Now it appears that\npractice--_e.g._ in music--very greatly increases sensitivity. We\nlearn to react differently to stimuli which only differ sightly; what\nis more, we learn to react to differences. A violin-player can react\nwith great precision to an interval of a fifth; if the interval is very\nslightly greater or less, his behaviour in tuning is influenced by the\ndifference from a fifth. And as we have already had occasion to notice,\nwe become, through practice, increasingly sensitive to form. All this\nincreased sensitivity must count as increase of knowledge.\n\nBut in saying this we are not saying anything inconsistent with our\nearlier definition of knowledge. Sensitivity is essential to choosing\nthe right reaction in many cases. To take the cookery-book again; when\nit says \u201ctake a pinch of salt\u201d, a good cook knows how much to take,\nwhich is an instance of sensitivity. Accurate scientific observation,\nwhich is of great practical importance, depends upon sensitivity. And\nso do many of our practical dealings with other people: if we cannot\n\u201cfeel\u201d their moods, we shall be always getting at cross purposes.\n\nThe extent to which sensitivity is improved by practice is astonishing.\nTown-bred people do not know whether the weather is warm or cold until\nthey read the weather reports in the paper. An entomologist perceives\nvastly more beetles in the course of a country walk than other people\ndo. The subtlety with which connoisseurs can distinguish among wines\nand cigars is the despair of youths who wish to become men of the\nworld. Whether this increase of sensitivity can be accounted for by\nthe law of association, I do not know. In many cases, probably, it can,\nbut I think sensitiveness to form, which is the essential element in\nthe more difficult forms of abstract thought as well as in many other\nmatters, cannot be regarded as derivative from the law of association,\nbut is more analogous to the development of a new sense. I should\ntherefore include improvement in sensitivity as an independent element\nin the advancement of knowledge. But I do so with some hesitation.\n\nThe above discussion does not pretend to cover the whole of the ground\nthat has to be covered in discussing the definition of \u201cknowledge\u201d.\nThere are other points of view, which are also necessary to a complete\nconsideration of the question. But these must wait until, after\nconsidering the physical world, we come to the discussion of man as\nviewed from within.\n\n\n\n\nPART II\n\nTHE PHYSICAL WORLD\n\n\n\n\nCHAPTER IX\n\nTHE STRUCTURE OF THE ATOM\n\n\nIn all that we have said hitherto on the subject of man from without,\nwe have taken a common-sense view of the material world. We have not\nasked ourselves: what _is_ matter? Is there such a thing, or is the\noutside world composed of stuff of a different kind? And what light\ndoes a correct theory of the physical world throw upon the process of\nperception? These are questions which we must attempt to answer in the\nfollowing chapters. And in doing so the science upon which we must\ndepend is physics. Modern physics, however, is very abstract, and by\nno means easy to explain in simple language. I shall do my best, but\nthe reader must not blame me too severely if, here and there, he finds\nsome slight difficulty or obscurity. The physical world, both through\nthe theory of relativity and through the most recent doctrines as to\nthe structure of the atom, has become very different from the world\nof everyday life, and also from that of scientific materialism of the\neighteenth-century variety. No philosophy can ignore the revolutionary\nchanges in our physical ideas that the men of science have found\nnecessary; indeed it may be said that all traditional philosophies have\nto be discarded, and we have to start afresh with as little respect\nas possible for the systems of the past. Our age has penetrated more\ndeeply into the nature of things than any earlier age, and it would be\na false modesty to over-estimate what can still be learned from the\nmetaphysicians of the seventeenth, eighteenth and nineteenth centuries.\n\nWhat physics has to say about matter, and the physical world generally,\nfrom the standpoint of the philosopher, comes under two main heads:\nfirst, the structure of the atom; secondly, the theory of relativity.\nThe former was, until recently, the less revolutionary philosophically,\nthough the more revolutionary in physics. Until 1925, theories of the\nstructure of the atom were based upon the old conception of matter as\nindestructible substance, although this was already regarded as no\nmore than a convenience. Now, owing chiefly to two German physicists,\nHeisenberg and Schr\u00f6dinger, the last vestiges of the old solid atom\nhave melted away, and matter has become as ghostly as anything in\na spiritualist seance. But before tackling these newer views, it\nis necessary to understand the much simpler theory which they have\ndisplaced. This theory does not, except here and there, take account\nof the new doctrines on fundamentals that have been introduced by\nEinstein, and it is much easier to understand than relativity. It\nexplains so much of the facts that, whatever may happen, it must remain\na stepping-stone to a complete theory of the structure of the atom;\nindeed, the newer theories have grown directly out of it, and could\nhardly have arisen in any other way. We must therefore spend a little\ntime in giving a bare outline, which is the less to be regretted as the\ntheory is in itself fascinating.\n\nThe theory that matter consists of \u201catoms\u201d, _i.e._ of little bits\nthat cannot be divided, is due to the Greeks, but with them it was\nonly a speculation. The evidence for what is called the atomic\ntheory was derived from chemistry, and the theory itself, in its\nnineteenth-century form, was mainly due to Dalton. It was found that\nthere were a number of \u201celements\u201d, and that other substances were\ncompounds of these elements. Compound substances were found to be\ncomposed of \u201cmolecules\u201d, each molecule being composed of \u201catoms\u201d of one\nsubstance combined with \u201catoms\u201d of another or of the same. A molecule\nof water consists of two atoms of hydrogen and one atom of oxygen; they\ncan be separated by electrolysis. It was supposed, until radio-activity\nwas discovered, that atoms were indestructible and unchangeable.\nSubstances which were not compounds were called \u201celements\u201d. The\nRussian chemist Mendeleev discovered that the elements can be arranged\nin a series by means of progressive changes in their properties; in his\ntime, there were gaps in this series, but most of them have since been\nfilled by the discovery of new elements. If all the gaps were filled,\nthere would be 92 elements; actually the number known is 87, or,\nincluding three about which there is still some doubt, 90. The place of\nan element in this series is called its \u201catomic number\u201d. Hydrogen is\nthe first, and has the atomic number 1; helium is the second, and has\nthe atomic number 2; uranium is the last, and has the atomic number 92.\nPerhaps in the stars there are elements with higher atomic numbers, but\nso far none has been actually observed.\n\nThe discovery of radio-activity necessitated new views as to \u201catoms\u201d.\nIt was found that an atom of one radio-active element can break up into\nan atom of another element and an atom of helium, and that there is\nalso another way in which it can change. It was found also that there\ncan be different elements having the same place in the series; these\nare called \u201cisotopes\u201d. For example, when radium disintegrates it gives\nrise, in the end, to a kind of lead, but this is somewhat different\nfrom the lead found in lead-mines. A great many \u201celements\u201d have been\nshown by Dr. F.\u00a0W. Aston to be really mixtures of isotopes, which can\nbe sorted out by ingenious methods. All this, but more especially the\ntransmutation of elements in radio-activity, led to the conclusion that\nwhat had been called \u201catoms\u201d were really complex structures, which\ncould change into atoms of a different sort by losing a part. After\nvarious attempts to imagine the structure of an atom, physicists were\nled to accept the view of Sir Ernest Rutherford, which was further\ndeveloped by Niels Bohr.\n\nIn this theory, which, in spite of recent developments, remains\nsubstantially correct, all matter is composed of two sorts of units,\nelectrons and protons. All electrons are exactly alike, and all protons\nare exactly alike. All protons carry a certain amount of positive\nelectricity, and all electrons carry an equal amount of negative\nelectricity. But the mass of a proton is about 1835 times that of an\nelectron: it takes 1835 electrons to weigh as much as one proton.\nProtons repel each other, and electrons repel each other, but an\nelectron and a proton attract each other. Every atom is a structure\nconsisting of electrons and protons. The hydrogen atom, which is the\nsimplest, consists of one proton with one electron going round it as\na planet goes round the sun. The electron may be lost, and the proton\nleft alone; the atom is then positively electrified. But when it has\nits electron, it is, as a whole, electrically neutral, since the\npositive electricity of the proton is exactly balanced by the negative\nelectricity of the electron.\n\nThe second element, helium, has already a much more complicated\nstructure. It has a nucleus, consisting of four protons, and two\nelectrons very close together, and in its normal state it has two\nplanetary electrons going round the nucleus. But it may lose either or\nboth of these, and it is then positively electrified.\n\nAll the latter elements consist, like helium, of a nucleus composed of\nprotons and electrons, and a number of planetary electrons going round\nthe nucleus. There are more protons than electrons in the nucleus,\nbut the excess is balanced by the planetary electrons when the atom\nis unelectrified. The number of protons in the nucleus gives the\n\u201catomic weight\u201d of the element: the excess of protons over electrons\nin the nucleus gives the \u201catomic number\u201d, which is also the number of\nplanetary electrons when the atom is unelectrified. Uranium, the last\nelement, has 238 protons and 146 electrons in the nucleus, and when\nunelectrified it has 92 planetary electrons. The arrangement of the\nplanetary electrons in atoms other than hydrogen is not accurately\nknown, but it is clear that, in some sense, they form different rings,\nthose in the outer rings being more easily lost than those nearer the\nnucleus.\n\nI come now to what Bohr added to the theory of atoms as developed\nby Rutherford. This was a most curious discovery, introducing, in a\nnew field, a certain type of discontinuity which was already known\nto be exhibited by some other natural processes. No adage had seemed\nmore respectable in philosophy than \u201cnatura non facit saltum\u201d, Nature\nmakes no jumps. But if there is one thing more than another that the\nexperience of a long life has taught me, it is that Latin tags always\nexpress falsehoods; and so it has proved in this case. Apparently\nNature does make jumps, not only now and then, but whenever a body\nemits light, as well as on certain other occasions. The German\nphysicist Planck was the first to demonstrate the necessity of jumps.\nHe was considering how bodies radiate heat when they are warmer\nthan their surroundings. Heat, as has long been known, consists of\nvibrations, which are distinguished by their \u201cfrequency\u201d, _i.e._ by the\nnumber of vibrations per second. Planck showed that, for vibrations\nhaving a given frequency, not all amounts of energy are possible, but\nonly those having to the frequency a ratio which is a certain quantity\n_h_ multiplied by 1 or 2 or 3 or some other whole number, in practice\nalways a small whole number. The quantity _h_ is known as \u201cPlanck\u2019s\nconstant\u201d; it has turned out to be involved practically everywhere\nwhere measurement is delicate enough to know whether it is involved or\nnot. It is such a small quantity that, except where measurement can\nreach a very high degree of accuracy, the departure from continuity is\nnot appreciable.[7]\n\n      [7] The dimensions of _h_ are those of \u201caction\u201d, _i.e._\n          energy multiplied by time, or moment of momentum, or\n          mass multiplied by length multiplied by velocity. Its\n          magnitude is about 6.55 \u00d7 10.27 erg secs.\n\nBohr\u2019s great discovery was that this same quantity _h_ is involved\nin the orbits of the planetary electrons in atoms, and that it\nlimits the possible orbits in ways for which nothing in Newtonian\ndynamics had prepared us, and for which so far, there is nothing in\nrelativity-dynamics to account. According to Newtonian principles, an\nelectron ought to be able to go round the nucleus in any circle with\nthe nucleus in the centre, or in any ellipse with the nucleus in a\nfocus; among possible orbits, it would select one or another according\nto its direction and velocity. But in fact only certain out of all\nthese orbits occur. Those that occur are among those that are possible\non Newtonian principles, but are only an infinitesimal selection from\namong these. It will simplify the explanation if we confine ourselves,\nas Bohr did at first, to circular orbits; moreover we will consider\nonly the hydrogen atom, which has one planetary electron and a nucleus\nconsisting of one proton. To define the circular orbits that are\nfound to be possible, we proceed as follows: multiply the mass of the\nelectron by the circumference of its orbit, and this again by the\nvelocity of the electron; the result will always be _h_ or 2_h_, or\n3_h_, or some other small exact multiple of _h_, where _h_, as before,\nis \u201cPlanck\u2019s constant\u201d. There is thus a smallest possible orbit, in\nwhich the above product is _h_; the radius of the next orbit, in which\nthe above produce is 2_h_, will have a length four times this minimum;\nthe next, nine times; the next, sixteen times; and so on through the\n\u201csquare numbers\u201d (_i.e._ those got by multiplying a number by itself).\nApparently no other circular orbits than these are possible in the\nhydrogen atom. Elliptic orbits are possible, and these again introduce\nexact multiples of _h_: but we need not, for our purposes, concern\nourselves with them.\n\nWhen a hydrogen atom is left to itself, if the electron is in the\nminimum orbit it will continue to rotate in that orbit so long as\nnothing from outside disturbs it; but if the electron is in any of\nthe larger possible orbits, it may sooner or later jump suddenly to a\nsmaller orbit, either the minimum or one of the intermediate possible\norbits. So long as the electron does not change its orbit, the atom\ndoes not radiate energy, but when the electron jumps to a smaller\norbit, the atom loses energy, which is radiated out in the form of a\nlight-wave. This light-wave is always such that its energy divided by\nits frequency is exactly _h_. The atom may absorb energy from without,\nand it does so by the electron jumping to a larger orbit. It may then\nafterwards, when the external source of energy is removed, jump back to\nthe smaller orbit; this is the cause of fluorescence, since, in doing\nso, the atom gives out energy in the form of light.\n\nThe same principles, with greater mathematical complications, apply\nto the other elements. There is, however, with some of the latest\nelements, a phenomenon which cannot have any analogue in hydrogen, and\nthat is radio-activity. When an atom is radio-active, it emits rays of\nthree kinds, called respectively \u03b1-rays, \u03b2-rays, and \u03b3-rays. Of these,\nthe \u03b3-rays are analogous to light, but of much higher frequencies,\nor shorter wave-lengths; we need not further concern ourselves with\nthem. The \u03b1-rays and \u03b2-rays, on the contrary, are important as our\nchief source of knowledge concerning the nuclei of atoms. It is found\nthat the \u03b1-rays consist of helium nuclei, while the \u03b2-rays consist\nof electrons. Both come out of the nucleus, since the atom after\nradio-activity disruption is a different element from what it was\nbefore. But no one knows just why the nucleus disintegrates when it\ndoes, nor why, in a piece of radium, for example, some atoms break down\nwhile others do not.\n\nThe three principal sources of our knowledge concerning atoms have been\nthe light they emit, X-rays and radio-activity. As everyone knows,\nwhen the light emitted by a glowing gas is passed through a prism, it\nis found to consist of well-defined lines of different colours, which\nare characteristic for each element, and constitute what is called its\n\u201cspectrum\u201d. The spectrum extends beyond the range of visible light,\nboth into the infra-red and into the ultra-violet. In the latter\ndirection, it extends right into the region of X-rays, which are only\nultra-ultra-violet light. By means of crystals, it has been found\npossible to study X-ray spectra as exactly as those of ordinary light.\nThe great merit of Bohr\u2019s theory was that it explained why elements\nhave the spectra they do have, which had, before, been a complete\nmystery. In the cases of hydrogen and positively electrified helium,\nthe explanation, particularly as extended by the German physicist\nSommerfeld, gave the most minute numerical agreement between theory\nand observation; in other cases, mathematical difficulties made\nthis completeness impossible, but there was every reason to think\nthat the same principles were adequate. This was the main reason for\naccepting Bohr\u2019s theory; and certainly it was a very strong one. It\nwas found that visible light enabled us to study the outer rings of\nplanetary electrons, X-rays enabled us to study the inner rings,\nand radio-activity enabled us to study the nucleus. For the latter\npurpose, there are also other methods, more particularly Rutherford\u2019s\n\u201cbombardment\u201d, which aims at breaking up nuclei by firing projectiles\nat them, and sometimes succeeds in making a hit in spite of the\nsmallness of the target.\n\nThe theory of atomic structure that has just been outlined, like\neverything in theoretical physics, is capable of expression in\nmathematical formul\u00e6; but like many things in theoretical physics, it\nis also capable of expression in the form of an imaginative picture.\nBut here, as always, it is necessary to distinguish sharply between\nthe mathematical symbols and the pictorial words. The symbols are\npretty sure to be right, or nearly so; the imaginative picture, on\nthe other hand, should not be taken too seriously. When we consider\nthe nature of the evidence upon which the above theory of the atom is\nbased, we can see that the attempt to make a picture of what goes on\nhas led us to be far more concrete than we have any right to be. If\nwe want to assert only what we have good reason to believe, we shall\nhave to abandon the attempt to be concrete about what goes on in the\natom, and say merely something like this: An atom with its electrons\nis a system characterised by certain integers, all small, and all\ncapable of changing independently. These integers are the multiples of\n_h_ involved. When any of them changes to a smaller integer, energy\nof a definite amount is emitted, and its frequency will be obtained\nby dividing the energy of _h_. When any of the integers concerned\nchanges to a larger integer, energy is absorbed, and again the amount\nabsorbed is definite. But we cannot know what goes on when the atom is\nneither absorbing nor radiating energy, since then it has no effects in\nsurrounding regions; consequently all evidence as to atoms is as to\ntheir changes, not as to their steady states.\n\nThe point is not that the facts do not fit with the hypothesis of\nthe atom as a planetary system. There are, it is true, certain\ndifficulties which afford empirical grounds for the newer theory which\nhas superseded Bohr\u2019s, and which we shall shortly consider. But even\nif no such grounds existed, it would be obvious that Bohr\u2019s theory\nstates more than we have a right to infer from what we can observe. Of\ntheories that state so much, there must be an infinite number that are\ncompatible with what is known, and it is only what all of these have in\ncommon that we are really entitled to assert. Suppose your knowledge\nof Great Britain were entirely confined to observing the people and\ngoods that enter and leave the ports; you could, in that case, invent\nmany theories as to the interior of Great Britain, all of which would\nagree with all known facts. This is an exact analogy. If you delimit\nin the physical universe any region, large or small, not containing a\nscientific observer, all scientific observers will have exactly the\nsame experiences whatever happens inside this region, provided it\ndoes not affect the flow of energy across the boundary of the region.\nAnd so, if the region contains one atom, any two theories which give\nthe same results as to the energy that the atom radiates or absorbs\nare empirically indistinguishable, and there can be no reason except\nsimplicity for preferring one of them to the other. On this ground,\neven if on no other, prudence compels us to seek a more abstract theory\nof the atom than that which we owe to Rutherford and Bohr.\n\nThe newer theory has been put forward mainly by two physicists already\nmentioned, Heisenberg and Schr\u00f6dinger, in forms which look different,\nbut are in fact mathematically equivalent. It is as yet an impossible\ntask to describe this theory in simple language, but something can be\nsaid to show its philosophical bearing. Broadly speaking, it describes\nthe atom by means of the radiations that come out of it. In Bohr\u2019s\ntheory, the planetary electrons are supposed to describe orbits over\nand over again while the atom is not radiating; in the newer theory,\nwe say nothing at all as to what happens at these times. The aim\nis to confine the theory to what is empirically verifiable, namely\nradiations; as to what there is where the radiations come from, we\ncannot tell, and it is scientifically unnecessary to speculate. The\ntheory requires modifications in our conception of space, of a sort not\nyet quite clear. It also has the consequence that we cannot identify an\nelectron at one time with an electron at another, if in the interval,\nthe atom has radiated energy. The electron ceases altogether to have\nthe properties of a \u201cthing\u201d as conceived by common sense; it is merely\na region from which energy may radiate.\n\nOn the subject of discontinuity, there is disagreement between\nSchr\u00f6dinger and other physicists. Most of them maintain that quantum\nchanges--_i.e._ the changes that occur in an atom when it radiates or\nabsorbs energy--must be discontinuous. Schr\u00f6dinger thinks otherwise.\nThis is a matter in debate among experts, as to which it would be rash\nto venture an opinion. Probably it will be decided one way or other\nbefore very long.\n\nThe main point for the philosopher in the modern theory is the\ndisappearance of matter as a \u201cthing\u201d. It has been replaced by\nemanations from a locality--the sort of influences that characterise\nhaunted rooms in ghost stories. As we shall see in the next chapter,\nthe theory of relativity leads to a similar destruction of the solidity\nof matter, by a different line of argument. All sorts of events happen\nin the physical world, but tables and chairs, the sun and moon,\nand even our daily bread, have become pale abstractions, mere laws\nexhibited in the successions of events which radiate from certain\nregions.\n\n\n\n\nCHAPTER X\n\nRELATIVITY\n\n\nWe have seen that the world of the atom is a world of revolution rather\nthan evolution: the electron which has been moving in one orbit hops\nquite suddenly into another, so that the motion is what is called\n\u201cdiscontinuous\u201d, that is to say, the electron is first in one place and\nthen in another, without having passed over any intermediate places.\nThis sounds like magic, and there may be some way of avoiding such a\ndisconcerting hypothesis. At any rate, nothing of the sort seems to\nhappen in the regions where there are no electrons and protons. In\nthese regions, so far as we can discover, there is continuity, that\nis to say, everything goes by gradual transitions, not by jumps. The\nregions in which there are no electrons and protons may be called\n\u201c\u00e6ther\u201d or \u201cempty space\u201d as you prefer: the difference is only verbal.\nThe theory of relativity is especially concerned with what goes on in\nthese regions, as opposed to what goes on where there are electrons\nand protons. Apart from the theory of relativity, what we know about\nthese regions is that waves travel across them, and that these waves,\nwhen they are waves of light or electromagnetism (which are identical),\nbehave in a certain fashion set forth by Maxwell in certain formul\u00e6\ncalled \u201cMaxwell\u2019s equations\u201d. When I say we \u201cknow\u201d this, I am saying\nmore than is strictly correct, because all we know is what happens\nwhen the waves reach our bodies. It is as if we could not see the sea,\nbut could only see the people disembarking at Dover, and inferred the\nwaves from the fact that the people looked green. It is obvious, in any\ncase, that we can only know so much about the waves as is involved in\ntheir having such-and-such causes at one end and such-and-such effects\nat the other. What can be inferred in this way will be, at best,\nsomething wholly expressible in terms of mathematical structure. We\nmust not think of the waves as being necessarily \u201cin\u201d the \u00e6ther or \u201cin\u201d\nanything else; they are to be thought of merely as progressive periodic\nprocesses, whose laws are more or less known, but whose intrinsic\ncharacter is not known and never can be.\n\nThe theory of relativity has arisen from the study of what goes on in\nthe regions where there are no electrons and protons. While the study\nof the atom has led us to discontinuities, relativity has produced\na completely continuous theory of the intervening medium--far more\ncontinuous than any theory formerly imagined. At the moment, these\ntwo points of view stand more or less opposed to each other, but no\ndoubt before long they will be reconciled. There is not, even now, any\nlogical contradiction between them; there is only a fairly complete\nlack of connection.\n\nFor philosophy, far the most important thing about the theory of\nrelativity is the abolition of the one cosmic time and the one\npersistent space, and the substitution of space-time in place of both.\nThis is a change of quite enormous importance, because it alters\nfundamentally our notion of the structure of the physical world, and\nhas, I think, repercussions in psychology. It would be useless, in our\nday, to talk about philosophy without explaining this matter. Therefore\nI shall make the attempt, in spite of some difficulty.\n\nCommon-sense and pre-relativity physicists believed that, if two events\nhappen in different places, there must always be a definite answer, in\ntheory, to the question whether they were simultaneous. This is found\nto be a mistake. Let us suppose two persons A and B a long way apart,\neach provided with a mirror and means of sending out light-signals. The\nevents that happen to A still have a perfectly definite time-order, and\nso have those that happen to B; the difficulty comes in connecting A\u2019s\ntime with B\u2019s. Suppose A sends a flash to B, B\u2019s mirror reflects it,\nand it returns to A after a certain time. If A is on the earth and B\non the sun, the time will be about sixteen minutes. We shall naturally\nsay that the time when B received the light-signal is half way between\nthe times when A sent it out and received it back. But this definition\nturns out to be not unambiguous; it will depend upon how A and B are\nmoving relatively to each other. The more this difficulty is examined,\nthe more insuperable it is seen to be. Anything that happens to A after\nhe sends out the flash and before he gets it back is neither definitely\nbefore nor definitely after nor definitely simultaneous with the\narrival of the flash at B. To this extent, there is no unambiguous way\nof correlating times in different places.\n\nThe notion of a \u201cplace\u201d is also quite vague. Is London a \u201cplace\u201d? But\nthe earth is rotating. Is the earth a \u201cplace\u201d? But it is going round\nthe sun. Is the sun a \u201cplace\u201d? But it is moving relatively to the\nstars. At best you could talk of a place at a given time; but then it\nis ambiguous what is a given time, unless you confine yourself to one\nplace. So the notion of \u201cplace\u201d evaporates.\n\nWe naturally think of the universe as being in one state at one time\nand in another at another. This is a mistake. There is no cosmic time,\nand so we cannot speak of the state of the universe at a given time.\nAnd similarly we cannot speak unambiguously of the distance between\ntwo bodies at a given time. If we take the time appropriate to one of\nthe two bodies, we shall get one estimate; if the time of the other,\nanother. This makes the Newtonian law of gravitation ambiguous, and\nshows that it needs restatement, independently of empirical evidence.\n\nGeometry also goes wrong. A straight line, for example, is supposed to\nbe a certain track in space whose parts all exist simultaneously. We\nshall now find that what is a straight line for one observer is not a\nstraight line for another. Therefore geometry ceases to be separable\nfrom physics.\n\nThe \u201cobserver\u201d need not be a mind, but may be a photographic plate.\nThe peculiarities of the \u201cobserver\u201d in this region belong to physics,\nnot to psychology.\n\nSo long as we continue to think in terms of bodies moving, and try to\nadjust this way of thinking to the new ideas by successive corrections,\nwe shall only get more and more confused. The only way to get clear is\nto make a fresh start, with _events_ instead of bodies. In physics, an\n\u201cevent\u201d is anything which, according to the old notions, would be said\nto have both a date and a place. An explosion, a flash of lightning,\nthe starting of a light-wave from an atom, the arrival of the\nlight-wave at some other body, any of these would be an \u201cevent\u201d. Some\nstrings of events make up what we regard as the history of one body;\nsome make up the course of one light-wave; and so on. The unity of a\nbody is a unity of history--it is like the unity of a tune, which takes\ntime to play, and does not exist whole in any one moment. What exists\nat any one moment is only what we call an \u201cevent\u201d. It may be that the\nword \u201cevent\u201d, as used in physics, cannot be quite identified with the\nsame word as used in psychology; for the present we are concerned\nwith \u201cevents\u201d as the constituents of physical processes, and need not\ntrouble ourselves about \u201cevents\u201d in psychology.\n\nThe events in the physical world have relations to each other which\nare of the sort that have led to the notions of space and time. They\nhave relations of order, so that we can say that one event is nearer\nto a second than to a third. In this way we can arrive at the notion\nof the \u201cneighbourhood\u201d of an event: it will consist roughly speaking\nof all the events that are very near the given event. When we say that\nneighbouring events have a certain relation, we shall mean that the\nnearer two events are to each other, the more nearly they have this\nrelation, and that they approximate to having it without limit as they\nare taken nearer and nearer together.\n\nTwo neighbouring events have a measurable quantitative relation\ncalled \u201cinterval\u201d, which is sometimes analogous to distance in space,\nsometimes to lapse of time. In the former case it is called space-like,\nin the latter time-like. The interval between two events is time-like\nwhen one body might be present at both--for example, when both are\nparts of the history of your body. The interval is space-like in the\ncontrary case. In the marginal case between the two, the interval is\nzero; this happens when both are parts of one light-ray.\n\nThe interval between two neighbouring events is something objective,\nin the sense that any two careful observers will arrive at the same\nestimate of it. They will not arrive at the same estimate for the\ndistance in space or the lapse of time between the two events, but the\ninterval is a genuine physical fact, the same for all. If a body can\ntravel freely from one event to the other, the interval between the two\nevents will be the same as the time between them as measured by a clock\ntravelling with the body. If such a journey is physically impossible,\nthe interval will be the same as the distance as estimated by an\nobserver to whom the two events are simultaneous. But the interval is\nonly definite when the two events are very near together; otherwise\nthe interval depends upon the route chosen for travelling from the one\nevent to the other.\n\nFour numbers are needed to fix the position of an event in the world;\nthese correspond to the time and the three dimensions of space in the\nold reckoning. These four numbers are called the co-ordinates of the\nevent. They may be assigned on any principle which gives neighbouring\nco-ordinates to neighbouring events; subject to this condition, they\nare merely conventional. For example, suppose an aeroplane has had an\naccident. You can fix the position of the accident by four numbers:\nlatitude, longitude, altitude above sea-level, and Greenwich Mean Time.\nBut you cannot fix the position of the explosion in space-time by means\nof less than four numbers.\n\nEverything in relativity-theory goes (in a sense) from next to\nnext; there are no direct relations between distant events, such as\ndistance in time or space. And of course there are no forces acting\nat a distance; in fact, except as a convenient fiction, there are\nno \u201cforces\u201d at all. Bodies take the course which is easiest at each\nmoment, according to the character of space-time in the particular\nregion where they are; this course is called a _geodesic_.\n\nNow it will be observed that I have been speaking freely of bodies and\nmotion, although I said that bodies were merely certain strings of\nevents. That being so, it is of course necessary to say what strings of\nevents constitute bodies, since not all continuous strings of events\ndo so, not even all geodesics. Until we have defined the sort of thing\nthat makes a body, we cannot legitimately speak of motion, since this\ninvolves the presence of one body on different occasions. We must\ntherefore set to work to define what we mean by the persistence of a\nbody, and how a string of events constituting a body differs from one\nwhich does not. This topic will occupy the next chapter.\n\nBut it may be useful, as a preliminary, to teach our imagination to\nwork in accordance with the new ideas. We must give up what Whitehead\nadmirably calls the \u201cpushiness\u201d of matter. We naturally think of an\natom as being like a billiard-ball; we should do better to think of it\nas like a ghost, which has no \u201cpushiness\u201d and yet can make you fly. We\nhave to change our notions both of substance and of cause. To say that\nan atom persists is like saying that a tune persists. If a tune takes\nfive minutes to play, we do not conceive of it as a single thing which\nexists throughout that time, but as a series of notes, so related as\nto form a unity. In the case of the tune, the unity is \u00e6sthetic; in\nthe case of the atom, it is causal. But when I say \u201ccausal\u201d I do not\nmean exactly what the word naturally conveys. There must be no idea of\ncompulsion or \u201cforce\u201d, neither the force of contact which we imagine\nwe see between billiard-balls nor the action at a distance which\nwas formerly supposed to constitute gravitation. There is merely an\nobserved law of succession from next to next. An event at one moment\nis succeeded by an event at a neighbouring moment, which, to the first\norder of small quantities, can be calculated from the earlier event.\nThis enables us to construct a string of events, each, approximately,\ngrowing out of a slightly earlier event according to an intrinsic law.\nOutside influences only affect the second order of small quantities. A\nstring of events connected, in this way, by an approximate intrinsic\nlaw of development is called one piece of matter. This is what I mean\nby saying that the unity of a piece of matter is causal. I shall\nexplain this notion more fully in later chapters.\n\n\n\n\nCHAPTER XI\n\nCAUSAL LAWS IN PHYSICS\n\n\nIn the last chapter we spoke about the substitution of space-time for\nspace and time, and the effect which this has had in substituting\nstrings of events for \u201cthings\u201d conceived as substances. In this chapter\nwe will deal with cause and effect as they appear in the light of\nmodern science. It is at least as difficult to purge our imagination\nof irrelevances in this matter as in regard to substance. The\nold-fashioned notion of cause appeared in dynamics as \u201cforce\u201d. We still\nspeak of forces just as we still speak of the sunrise, but we recognise\nthat this is nothing but a convenient way of speaking, in the one case\nas in the other.\n\nCausation is deeply embedded in language and common sense. We say that\npeople build houses or make roads: to \u201cbuild\u201d and to \u201cmake\u201d are both\nnotions involving causality. We say that a man is \u201cpowerful\u201d, meaning\nthat his volitions are causes over a wide range. Some examples of\ncausation seem to us quite natural, others less so. It seems natural\nthat our muscles should obey our will, and only reflection makes us\nperceive the necessity of finding an explanation of this phenomenon. It\nseems natural that when you hit a billiard-ball with a cue it moves.\nWhen we see a horse pulling a cart, or a heavy object being dragged by\na rope, we _feel_ as if we understood all about it. It is events of\nthis sort that have given rise to the common-sense belief in causes and\nforces.\n\nBut as a matter of fact the world is incredibly more complicated than\nit seems to common sense. When we think we understand a process--I mean\nby \u201cwe\u201d the non-reflective part in each of us--what really happens\nis that there is some sequence of events so familiar through past\nexperience that at each stage we expect the next stage. The whole\nprocess seems to us peculiarly intelligible when human desires enter\nin, for example, in watching a game: what the ball does and what the\nplayers do seem \u201cnatural\u201d, and we feel as if we quite understood how\nthe stages succeed each other. We thus arrive at the notion of what is\ncalled \u201cnecessary\u201d sequence. The text-books say that A is the cause of\nB if A is \u201cnecessarily\u201d followed by B. This notion of \u201cnecessity\u201d seems\nto be purely anthropomorphic, and not based upon anything that is a\ndiscoverable feature of the world. Things happen according to certain\nrules; the rules can be generalised, but in the end remain brute facts.\nUnless the rules are concealed conventions or definitions, no reason\ncan be given why they should not be completely different.\n\nTo say that A is \u201cnecessarily\u201d followed by B is thus to say no more\nthan that there is some general rule, exemplified in a very large\nnumber of observed instances, and falsified in none, according to which\nevents such as A are followed by events such as B. We must not have any\nnotion of \u201ccompulsion\u201d, as if the cause _forced_ the effect to happen.\nA good test for the imagination in this respect is the reversibility\nof causal laws. We can just as often infer backwards as forwards.\nWhen you get a letter, you are justified in inferring that somebody\nwrote it, but you do not feel that your receiving it _compelled_\nthe sender to write it. The notion of compulsion is just as little\napplicable to effects as to causes. To say that causes compel effects\nis as misleading as to say that effects compel causes. Compulsion is\nanthropomorphic: a man is compelled to do something when he wishes to\ndo the opposite, but except where human or animal wishes come in the\nnotion of compulsion is inapplicable. Science is concerned merely with\nwhat happens, not with what _must_ happen.\n\nWhen we look for invariable rules of sequence in nature, we find that\nthey are not such as common sense sets up. Common sense says: thunder\nfollows lightning, waves at sea follow wind, and so on. Rules of\nthis sort are indispensable in practical life, but in science they\nare all only approximate. If there is any finite interval of time,\nhowever short, between the cause and the effect, something may happen\nto prevent the effect from occurring. Scientific laws can only be\nexpressed in differential equations. This means that, although you\ncannot tell what may happen after a finite time, you can say that, if\nyou make the time shorter and shorter, what will happen will be more\nand more nearly according to such-and-such a rule. To take a very\nsimple case: I am now in this room; you cannot tell where I shall be\nin another second, because a bomb may explode and blow me sky-high,\nbut if you take any two small fragments of my body which are now very\nclose together, you can be sure that, after _some_ very short finite\ntime, they will still be very close together. If a second is not short\nenough, you must take a shorter time; you cannot tell in advance how\nshort a time you may have to take, but you may feel fairly certain that\nthere is a short enough time.\n\nThe laws of sequence in physics, apart from quantum phenomena, are of\ntwo sorts, which appeared in traditional dynamics as laws of velocity\nand laws of acceleration. In a very short time, the velocity of a\nbody alters very little, and if the time is taken short enough, the\nchange of velocity diminishes without limit. This is what, in the last\nchapter, we called an \u201cintrinsic\u201d causal law. Then there is the effect\nof the outer world, as it appeared in traditional dynamics, which\nis shown in acceleration. The small change which does occur in the\nvelocity in a short time is attributed to surrounding bodies, because\nit is found to vary as they vary, and to vary according to ascertained\nlaws. Thus we think of surrounding bodies as exerting an influence,\nwhich we call \u201cforce\u201d, though this remains as mysterious as the\ninfluence of the stars in astrology.\n\nEinstein\u2019s theory of gravitation has done away with this conception in\nso far as gravitational forces are concerned. In this theory, a planet\nmoving round the sun is moving in the nearest approach to a straight\nline that the neighbourhood permits. The neighbourhood is supposed to\nbe non-Euclidean, that is to say, to contain no straight lines such\nas Euclid imagined. If a body is moving freely, as the planets do, it\nobserves a certain rule. Perhaps the simplest way to state this rule\nis as follows: Suppose you take any two events which happen on the\nearth, and you measure the time between them by ideally accurate clocks\nwhich move with the earth. Suppose some traveller on a magic carpet had\nmeanwhile cruised about the universe, leaving the earth at the time of\nthe first event and returning at the time of the second. By his clocks\nthe period elapsed will be less than by the terrestial clocks. This is\nwhat is meant by saying that the earth moves in a \u201cgeodesic\u201d, which\nis the nearest approach to a straight line to be found in the region\nin which we live. All this is, so to speak, geometrical, and involves\nno \u201cforces\u201d. It is not the sun that makes the earth go round, but the\nnature of space-time where the earth is.\n\nEven this is not quite correct. Space-time does not make the earth go\nround the sun; it makes us _say_ the earth goes round the sun. That is\nto say, it makes this the shortest way of describing what occurs. We\ncould describe it in other language, which would be equally correct,\nbut less convenient.\n\nThe abolition of \u201cforce\u201d in astronomy is perhaps connected with the\nfact that astronomy depends only upon the sense of sight. On the\nearth, we push and pull, we touch things, and we experience muscular\nstrains. This all gives us a notion of \u201cforce\u201d, but this notion is\nanthropomorphic. To imagine the laws of motion of heavenly bodies,\nthink of the motions of objects in a mirror; they may move very fast,\nalthough in the mirror world there are no forces.\n\nWhat we really have to substitute for force is laws of correlation.\nEvents can be collected in groups by their correlations. This is\nall that is true in the old notion of causality. And this is not a\n\u201cpostulate\u201d or \u201ccategory\u201d, but an observed fact--lucky, not necessary.\n\nAs we suggested before, it is these correlations of events that lead\nto the definition of permanent \u201cthings\u201d. There is no essential\ndifference, as regards substantiality, between an electron and a\nlight-ray. Each is really a string of events or of sets of events. In\nthe case of the light-ray, we have no temptation to think otherwise.\nBut in the case of the electron, we think of it as a single persistent\nentity. There _may_ be such an entity, but we can have no evidence that\nthere is. What we can discover is (_a_) a group of events spreading\noutwards from a centre--say, for definiteness, the events constituting\na wave of light--and attributed, hypothetically, to a \u201ccause\u201d in that\ncentre; (_b_) more or less similar groups of events at other times,\nconnected with the first group according to the laws of physics, and\ntherefore attributed to the same hypothetical cause at other times. But\nall that we ought to assume is series of groups of events, connected by\ndiscoverable laws. These series we may _define_ as \u201cmatter\u201d. Whether\nthere is matter in any other sense, no one can tell.\n\nWhat is true in the old notion of causality is the fact that events at\ndifferent times are connected by laws (differential equations). When\nthere is a law connecting an event A with an event B, the two have a\ndefinite unambiguous time-order. But if the events are such that a ray\nof light starting from A would arrive at any body which was present at\nB after B had occurred, and _vice versa_, then there is no definite\ntrue order, and no possible causal law connecting A and B. A and B must\nthen be regarded as separate facts of geography.\n\nPerhaps the scope and purpose of this and the foregoing chapters may\nbe made clearer by showing their bearing upon certain popular beliefs\nwhich may seem self-evident but are really in my opinion either false\nor likely to lead to falsehood. I shall confine myself to objections\nwhich have actually been made to me when trying to explain the\nphilosophical outcome of modern physics.[8]\n\n      [8] These objections are quoted (with kind permission) from a\n          letter written to me by a well-known engineer, Mr. Percy\n          Griffith, who is also a writer on philosophical subjects.\n\n\u201cWe cannot conceive of movement apart from some _thing_ as moving.\u201d\nThis is, in a sense, a truism; but in the sense in which it is usually\nmeant, it is a falsehood. We speak of the \u201cmovement\u201d of a drama or\npiece of music, although we do not conceive either as a \u201cthing\u201d which\nexists complete at every moment of the performance. This is the sort of\npicture we must have in our minds when we try to conceive the physical\nworld. We must think of a string of events, connected together by\ncertain causal connections, and having enough unity to deserve a single\nname. We then begin to imagine that the single name denotes a single\n\u201cthing\u201d, and if the events concerned are not all in the same place, we\nsay the \u201cthing\u201d has \u201cmoved.\u201d But this is only a convenient shorthand.\nIn the cinema, we seem to see a man falling off a skyscraper, catching\nhold of the telegraph wires, and reaching the ground none the worse.\nWe know that, in fact, there are a number of different photographs,\nand the appearance of a single \u201cthing\u201d moving is deceptive. In this\nrespect, the real world resembles the cinema.\n\nIn connection with motion one needs to emphasise the very difficult\ndistinction between _experience_ and _prejudice_. Experience, roughly,\nis what you see, and prejudice is what you only think you see.\nPrejudice tells you that you see the _same_ table on two different\noccasions; you _think_ that experience tells you this. If it really\nwere experience, you could not be mistaken; yet a similar table may be\nsubstituted without altering the experience. If you look at a table\non two different occasions, you have very similar sensations, and\nmemory tells you that they are similar; but there is nothing to show\nthat one identical entity causes the two sensations. If the table is\nin a cinema, you know that there is not such an entity, even though\nyou can watch it changing with apparent continuity. The experience is\njust like that with a \u201creal\u201d table; so in the case of a \u201creal\u201d table\nalso, there is nothing in the actual experience to show whether there\nis a persistent entity or not. I say, therefore: I do not know whether\nthere is a persistent entity, but I do know that my experiences can be\nexplained without assuming that there is. Therefore it can be no part\nof legitimate science to assert or deny the persistent entity; if it\ndoes either, it goes beyond the warrant of experience.\n\nThe following is a verbally cited passage in the letter referred to\nobjecting to what was said above about \u201cforce\u201d:\n\n\u201cThe concept of Force is not of physical but of psychological origin.\nRightly or wrongly it arises in the most impersonal contemplation of\nthe Stellar Universe, where we observe an infinite number of spherical\nbodies revolving on their own axes and gyrating in orbits round each\nother. Rightly or wrongly, we naturally conceive of these as having\nbeen so constituted and so maintained by some Force or Forces.\u201d\n\nWe do not, in fact, \u201cobserve\u201d what it is here said that we observe; all\nthis is _inferred_. What we observe, in astronomy, is a two-dimensional\npattern of points of light, with a few bright surfaces of measurable\nsize when seen through the telescope (the planets), and of course\nthe larger bright surfaces that we call the sun and moon. Most of\nthis pattern (the fixed stars) rotates round the earth once in every\ntwenty-three hours and fifty-six minutes. The sun rotates in varying\nperiods, which average twenty-four hours and never departs very far\nfrom the average. The moon and planets have apparent motions which are\nmore irregular. These are the _observed_ facts. There is no logical\nimpossibility about the formul\u00e6 doctrine of spheres rotating round the\nearth, one for each planet and one for the stars. The modern doctrines\nare simpler, but not one whit more in accordance with observed facts;\nit is our passion for _simple_ laws that has made us adopt them.\n\nThe last sentence of the above quotation raises some further points of\ninterest. \u201cRightly or wrongly\u201d, the writer says, \u201cwe naturally conceive\nof these as having been so constituted and so maintained by some Force\nor Forces.\u201d I do not deny this. It is \u201cnatural\u201d, and it is \u201cright or\nwrong\u201d--more specifically, it is wrong. \u201cForce\u201d is part of our love of\nexplanations. Everyone knows about the Hindu who thought that the world\ndoes not fall because it is supported by an elephant, and the elephant\ndoes not fall because it is supported by a tortoise. When his European\ninterlocutor said \u201cBut how about the tortoise?\u201d he replied that he was\ntired of metaphysics and wanted to change the subject. \u201cForce\u201d, as an\nexplanation, is no better than the elephant and the tortoise. It is\nan attempt to get at the \u201cwhy\u201d of natural processes, not only at the\n\u201chow\u201d. What we observe, to a limited extent, is what happens, and we\ncan arrive at laws according to which observable things happen, but\nwe cannot arrive at a reason for the laws. If we invent a reason, it\nneeds a reason in its turn, and so on. \u201cForce\u201d is a rationalising of\nnatural processes, but a fruitless one since \u201cforce\u201d would have to be\nrationalised also.\n\nWhen it is said, as it often is, that \u201cforce\u201d belongs to the world of\nexperience, we must be careful to understand what can be meant. In\nthe first place, it may be meant that calculations which employ the\nnotion of force work out right in practice. This, broadly speaking,\nis admitted: no one would suggest that the engineer should alter his\nmethods, or should give up working out stresses and strains. But that\ndoes not prove that there are stresses and strains. A British medical\nman renders his accounts in guineas, although they have long since\nceased to exist except as a name; he obtains a real payment, though he\nemploys a fictitious coin. Similarly, the engineer is concerned with\nthe question whether his bridge will stand: the fact of experience is\nthat it stands (or does not stand), and the stresses and strains are\nonly a way of explaining what sort of bridge will stand. They are as\nuseful as guineas, but equally imaginary.\n\nBut when it is said that force is a fact of experience, there is\nsomething quite different that may be meant. It may be meant that we\nexperience force when we experience such things as pressure or muscular\nexertion. We cannot discuss this contention adequately without going\ninto the relation of physics to psychology, which is a topic we shall\nconsider at length at a later stage. But we may say this much: if\nyou press your finger-tip upon a hard object, you have an experience\nwhich you attribute to your finger-tip, but there is a long chain of\nintermediate causes in nerves and brain. If your finger were amputated\nyou could still have the same experience by a suitable operation on\nthe nerves that formerly connected the finger with the brain, so that\nthe force between the finger-tip and the hard object, as a fact of\nexperience, may exist when there is no finger-tip. This shows that\nforce, in this sense, cannot be what concerns physics.\n\nAs the above example illustrates, we do not, in fact, experience\nmany things that we think we experience. This makes it necessary to\nask, without too much assurance, in what sense physics can be based\nupon experience, and what must be the nature of its entities and its\ninferences if it is to make good its claim to be empirically grounded.\nWe shall begin this inquiry in the next chapter.\n\n\n\n\nCHAPTER XII\n\nPHYSICS AND PERCEPTION\n\n\nIt will be remembered that we regarded perception, in Chapter V, as\na species of \u201csensitivity\u201d. Sensitivity to a given feature of the\nenvironment we defined as consisting in some characteristic reaction\nwhich is exhibited whenever that feature is present, but not otherwise;\nthis property is possessed more perfectly, in given directions, by\nscientific instruments than by living bodies, though scientific\ninstruments are more selective as to the stimuli to which they will\nrespond. We decided that what, from the standpoint of an external\nobserver, distinguishes perception from other forms of sensitivity is\nthe law of association or conditioned reflexes. But we also found that\nthis purely external treatment of perception presupposes our knowledge\nof the physical world as a going concern. We have now to investigate\nthis presupposition, and to consider how we come to know about physics,\nand how much we really do know.\n\nAccording to the theory of Chapter V, it is possible to perceive things\nthat are not in a spatial contact with the body. There must be a\nreaction to a feature of the environment, but that feature may be at a\ngreater or less distance from the body of the percipient; we can even\nperceive the sun and stars, within the limits of the definition. All\nthat is necessary is that our reaction should depend upon the spatial\nrelation between our body and the feature of the environment. When our\nback is towards the sun, we do not see it; when our face is towards it,\nwe do.\n\nWhen we consider perception--visual or auditory--of an external event,\nthere are three different matters to be examined. There is first the\nprocess in the outside world, from the event to the percipient\u2019s body;\nthere is next the process in his body, in so far as this can be known\nby an outside observer; lastly, there is the question, which must be\nfaced sooner or later, whether the percipient can perceive something of\nthe process in his body which no other observer could perceive. We will\ntake these points in order.\n\nIf it is to be possible to \u201cperceive\u201d an event not in the percipient\u2019s\nbody, there must be a physical process in the outer world such that,\nwhen a certain event occurs, it produces a stimulus of a certain kind\nat the surface of the percipient\u2019s body. Suppose, for example, that\npictures of different animals are exhibited on a magic lantern to a\nclass of children, and all the children are asked to say the name of\neach animal in turn. We may assume that the children are sufficiently\nfamiliar with animals to say \u201ccat\u201d, \u201cdog\u201d, \u201cgiraffe\u201d, \u201chippopotamus\u201d,\netc., at the right moments. We must then suppose--taking the physical\nworld for granted--that some process travels from each picture to the\neyes of the various children, retaining throughout these journeys such\npeculiarities that, when the process reaches their eyes, it can in one\ncase stimulate the word \u201ccat\u201d and in another the word \u201cdog\u201d. All this\nthe physical theory of light provides for. But there is one interesting\npoint about language that should be noticed in this connection. If the\nusual physical theory of light is correct, the various children will\nreceive stimuli which differ greatly according to their distance and\ndirection from the picture, and according to the way the light falls.\nThere are also differences in their reactions, for, though they all\nutter the word \u201ccat\u201d, some say it loud, others soft, some in a soprano\nvoice, some in a contralto. But the differences in their reactions are\nmuch less than the differences in the stimuli. This is still more the\ncase if we consider various different pictures of cats, to all of which\nthey respond with the word \u201ccat\u201d. Thus language is a means of producing\nresponses which differ less than the stimuli do, in cases where the\nresemblances between the stimuli are more important to us than the\ndifferences. This fact makes us apt to overlook the differences\nbetween stimuli which produce nearly identical responses.\n\nAs appears from the above, when a number of people simultaneously\nperceive a picture of a cat, there are differences between the stimuli\nto their various perceptions, and these differences must obviously\ninvolve differences in their reactions. The _verbal_ responses may\ndiffer very little, but even the verbal responses could be made to\ndiffer by putting more complicated questions than merely \u201cWhat animal\nis that?\u201d One could ask: \u201cCan the picture be covered by your thumb-nail\nheld at arm\u2019s length?\u201d Then the answer would be different according\nas the percipient was near the picture or far off. But the normal\npercipient, if left to himself, will not notice such differences,\nthat is to say, his verbal response will be the same in spite of the\ndifferences in the stimuli.\n\nThe fact that it is possible for a number of people to perceive the\nsame noise or the same coloured pattern obviously depends upon the fact\nthat a physical process can travel outward from a centre and retain\ncertain of its characteristics unchanged, or very little changed. The\nmost notable of such characteristics is frequency in a wave-motion.\nThat, no doubt, affords a biological reason for the fact that our most\ndelicate senses, sight and hearing, are sensitive to frequencies, which\ndetermine colour in what we see and pitch in what we hear. If there\nwere not, in the physical world, processes spreading out from centres\nand retaining certain characters practically unchanged, it would be\nimpossible for different percipients to perceive the same object from\ndifferent points of view, and we should not have been able to discover\nthat we all live in a common world.\n\nWe come now to the process in the percipient\u2019s body, in so far as\nthis can be perceived by an outside observer. This raises no new\nphilosophical problems, because we are still concerned, as before, with\nthe perception of events outside the observer\u2019s body. The observer,\nnow, is supposed to be a physiologist, observing, say, what goes on\nin the eye when light falls upon it. His means of knowing are, in\nprinciple, exactly the same as in the observation of dead matter.\nAn event in an eye upon which light is falling causes light-waves\nto travel in a certain manner until they reach the eye of the\nphysiologist. They there cause a process in the physiologist\u2019s eye\nand optic nerve and brain, which ends in what he calls \u201cseeing what\nhappens in the eye he is observing\u201d. But this event, which happens in\nthe physiologist, is not what happened in the eye he was observing;\nit is only connected with this by a complicated causal chain. Thus\nour knowledge of physiology is no more direct or intimate than our\nknowledge of processes in dead matter; we do not know any more about\nour eyes than about the trees and fields and clouds that we see by\nmeans of them. The event which happens when a physiologist observes an\neye is an event in him, not on the eye that he is observing.\n\nWe come now at last to the question of self-observation, which we have\nhitherto avoided. I say \u201cself-observation\u201d rather than \u201cintrospection\u201d,\nbecause the latter word has controversial associations that I wish to\navoid. I mean by \u201cself-observation\u201d anything that a man can perceive\nabout himself but that others, however situated, cannot perceive about\nhim. What follows is only preliminary, since the subject will be\ndiscussed at length in Chapter XVI.\n\nNo one can deny that we know things about ourselves which others\ncannot know unless we tell them. We know when we have toothache, when\nwe feel thirsty, what we were dreaming when we woke up, and so on.\nDr. Watson might say that the dentist can know we have toothache by\nobserving a cavity in a tooth. I will not reply that the dentist is\noften mistaken; this may be merely because the art of dentistry has not\nbeen sufficiently perfected. I will concede as possible, in the future,\na state of odontology in which the dentist could always know whether\nI am feeling toothache. But even then his knowledge has a different\ncharacter from mine. His knowledge is an inference, based upon the\ninducive law that people with such-and-such cavities suffer pain of\na certain kind. But this law cannot be established by observation of\ncavities alone; it requires that, where these are observed, the people\nwho have them should tell us that they feel toothache. And, more than\nthat, they must be speaking the truth. Purely external observation can\ndiscover that people with cavities _say_ they have toothache, but not\nthat they have it. Saying one has toothache is a different thing from\nhaving it; if not we could cure toothache by not talking about it, and\nso save our dentists\u2019 bills. I am sure the expert opinion of dentists\nwill agree with me that this is impossible.\n\nTo this argument, however, it might be replied that having toothache is\na state of the body, and that knowing I have toothache is a response to\nthis bodily stimulus. It will be said that, theoretically, the state\nof my body when I have toothache can be observed by an outsider, who\ncan then also know that I have toothache. This answer, however, does\nnot really meet the point. When the outside observer knows that I have\ntoothache, not only is his knowledge based upon an inductive inference,\nas we have already seen, but his knowledge of the inferred term,\n\u201ctoothache\u201d, must be based upon personal experience. No knowledge of\ndentistry could enable a man to know what toothache is if he had never\nfelt it. If, then, toothache is really a state of the body--which, at\nthe moment, I neither affirm nor deny--it is a state of the body which\nonly the man himself can perceive. In a word, whoever has experienced\ntoothache and can remember it has knowledge that cannot be possessed by\na man who has never experienced toothache.\n\nTake next our knowledge of our own dreams. Dr. Watson has not, so far\nas I know, ever discussed dreams, but I imagine he would say something\nlike this: In dreams, there are probably small laryngeal movements\nsuch as, if they were greater, would lead to speech; indeed, people\ndo sometimes cry out in dreams. There may also be stimulations of the\nsense-organs, which produce unusual reactions owing to the peculiar\nphysiological condition of the brain during sleep: but all these\nreactions must consist of small movements, which could theoretically be\nseen from outside, say by some elaboration of X-ray apparatus. This is\nall very well, but meantime it is hypothetical, and the dreamer himself\nknows his dreams without all this elaborate inference. Can we say that\nhe really knows these hypothetical small bodily movements, although\nhe thinks he knows something else? That would presumably be Dr.\nWatson\u2019s position, and it must be admitted that, with a definition of\n\u201cknowledge\u201d such as we considered in Chapter VIII, such a view is not\nto be dismissed offhand as obviously impossible. Moreover, if we are to\nsay the perception gives knowledge of the physical world, we shall have\nto admit that what we are perceiving may be quite different from what\nit seems to be. A table does not look like a vast number of electrons\nand protons, nor like trains of waves meeting and clashing. Yet this\nis the sort of thing a table is said to be by modern physicists. If,\nthen, what seems to us to be just a table such as may be seen any day\nis really this odd sort of thing, it is possible that what seems to us\nto be a dream is really a number of movements in the brain.\n\nThis again is all very well, but there is one point which it fails to\nexplain, namely, what is meant by \u201cseeming\u201d. If a dream or a table\n\u201cseems\u201d to be one sort of thing while it is \u201creally\u201d another, we shall\nhave to admit that it _really_ seems, and that what it seems to be has\na reality of its own. Nay more, we only arrive at what it \u201creally\u201d is\nby an inference, valid or invalid, from what it seems to be. If we are\nwrong about the seeming, we must be doubly wrong about the reality,\nsince the sole ground for asserting the table composed of electrons and\nprotons is the table that we see, _i.e._ the \u201cseeming\u201d table. We must\ntherefore treat \u201cseeming\u201d with respect.\n\nLet us consider Dr. Watson watching a rat in a maze. He means to be\nquite objective, and report only what really goes on. Can he succeed?\nIn one sense he can. He can use words about what he sees which are\nthe same as any other scientifically trained observer will use if he\nwatches the same rat at the same time. But Dr. Watson\u2019s objectivity\nemphatically does not consist in using the same words as other people\nuse; his vocabulary is very different from that of most psychologists.\nHe cannot take as the sole test of truth the consensus of mankind.\n\u201cSecurus judicat orbis terrarum\u201d is another example of a Latin tag\nwhich is false, and which certainly Dr. Watson would not consider\ntrue. It has happened again and again in human history that a man who\nsaid something that had never been said before turned out to be right,\nwhile the people who repeated the wise saws of their forefathers were\ntalking nonsense. Therefore, when Dr. Watson endeavours to eliminate\nsubjectivity in observing rats, he does not mean that he says what\neverybody else says. He means that he refrains from inferring anything\nabout the rat beyond its bodily movements. This is all to the good,\nbut I think he fails to realise that almost as long and difficult\nan inference is required to give us knowledge of the rat\u2019s bodily\nmovements as to give us knowledge of its \u201cmind\u201d. And what is more,\nthe data from which we must start in order to get to know the rat\u2019s\nbodily movements are data of just the sort that Dr. Watson wishes to\navoid, namely private data patent to self-observation but not patent to\nanyone except the observer. This is the point at which, in my opinion,\nbehaviourism _as a final philosophy_ breaks down.\n\nWhen several people simultaneously watch a rat in a maze, or any other\nexample of what we should naturally regard as matter in motion, there\nis by no means complete identity between the physical events which\nhappen at the surface of their eyes and constitute the stimuli to\ntheir perceptions. There are differences of perspective, of light and\nshade, of apparent size, and so on, all of which will be reproduced\nin photographs taken from the places where the eyes of the several\nobservers are. These differences produce differences in the reactions\nof the observers--differences which a quite unthinking person may\noverlook, but which are familiar to every artist. Now it is contrary\nto all scientific canons to suppose that the object perceived, in\naddition to affecting us in the way of stimulus and reaction, also\naffects us directly by some mystical epiphany; certainly it is not what\nany behaviourist would care to assert. Our knowledge of the physical\nworld, therefore, must be contained in our reaction to the stimulus\nwhich reaches us across the intervening medium; and it seems hardly\npossible that our reaction should have a more intimate relation to the\nobject than the stimulus has. Since the stimulus differs for different\nobservers, the reaction also differs; consequently, in all our\nperceptions of physical processes there is an element of subjectivity.\nIf, therefore, physics is true in its broad outlines (as the above\nargument supposes), what we call \u201cperceiving\u201d a physical process is\nsomething private and subjective, at least in part, and is yet the only\npossible starting-point for our knowledge of the physical world.\n\nThere is an objection to the above argument which might naturally be\nmade, but it would be in fact invalid. It may be said that we do not in\nfact proceed to _infer_ the physical world from our perceptions, but\nthat we begin at once with a rough-and-ready knowledge of the physical\nworld, and only at a late stage of sophistication compel ourselves to\nregard our knowledge of the physical world as an inference. What is\nvalid in this statement is the fact that our knowledge of the physical\nworld is not at first inferential, but that is only because we take our\npercepts to _be_ the physical world. Sophistication and philosophy come\nin at the stage at which we realise that the physical world cannot be\nidentified with our percepts. When my boy was three years old, I showed\nhim Jupiter, and told him that Jupiter was larger than the earth. He\ninsisted that I must be speaking of some other Jupiter, because, as he\npatiently explained, the one he was seeing was obviously quite small.\nAfter some efforts, I had to give it up and leave him unconvinced. In\nthe case of the heavenly bodies, adults have got used to the idea that\nwhat is really there can only be _inferred_ from what they see; but\nwhere rats in mazes are concerned, they still tend to think that they\nare seeing what is happening in the physical world. The difference,\nhowever, is only one of degree, and naive realism is as untenable in\nthe one case as in the other. There are differences in the perceptions\nof two persons observing the same process; there are sometimes no\ndiscoverable differences between two perceptions of the same persons\nobserving different processes, _e.g._ pure water and water full of\nbacilli. The subjectivity of our perceptions is thus of practical as\nwell as theoretical importance.\n\nI am not maintaining that what we primarily know is our own\nperceptions. This is largely a verbal question; but with the definition\nof knowledge given in Chapter VIII, it will be correct to say that\nfrom the first we know external objects, the question is not as to\nwhat are the objects we know, but rather as to how accurately we\nknow them. Our non-inferential knowledge of an object cannot be more\naccurate than our reaction to it, since it is part of that reaction.\nAnd our reaction cannot be more accurate than the stimulus. But what\non earth can you mean by the \u201caccuracy\u201d of a stimulus? I may be\nasked. I mean just the same as by the accuracy of a map or a set of\nstatistics. I mean a certain kind of correspondence. One pattern is\nan accurate representation of another if every element of the one can\nbe taken as the representative of just one element of the other, and\nthe relations that make the one set into a pattern correspond with\nrelations making the other set into a pattern. In this sense, writing\ncan represent speech with a certain degree of accuracy; to every spoken\nword a written word corresponds, and to the time-order of the spoken\nwords the space-order of the written words corresponds. But there are\ninflexions and tones of voice that cannot be represented in writing,\nexcept, to some extent, by musical notation. A gramophone record is\na much more accurate representation of vocal sounds than any writing\ncan be; but even the best gramophone record fails to be completely\naccurate. The impression made upon an observer is very analogous to\na gramophone record or a photograph, but usually less accurate owing\nto the influence of the law of association, and the lack of delicacy\nin our senses. And whatever limitations there are to the accuracy of\nour impressions are limitations to the accuracy of our non-inferential\nknowledge of the external world.\n\nAnother point: If we accept the definition of knowledge given\nin Chapter VIII, which was framed so far as to be as favourable\nas possible to behaviourism, a given reaction may be regarded as\nknowledge of various different occurrences. When we see Jupiter, we\nhave, according to the definition, knowledge of Jupiter, but we also\nhave knowledge of the stimulus at the surface of the eye, and even\nof the process in the optic nerve. For it is arbitrary at what point\nwe start in the process leading to a certain event in the brain:\nthis event, and the consequent bodily action, may be regarded as a\nreaction to a process starting at any earlier point. And the nearer\nour starting-point is to the brain, the more accurate becomes the\nknowledge displayed in our reaction. A lamp at the top of a tall\nbuilding might produce the same visual stimulus as Jupiter, or at any\nrate one practically indistinguishable from that produced by Jupiter.\nA blow on the nose might make us \u201csee stars\u201d. Theoretically, it should\nbe possible to apply a stimulus direct to the optic nerve, which\nshould give us a visual sensation. Thus when we think we see Jupiter,\nwe may be mistaken. We are less likely to be mistaken if we say that\nthe surface of the eye is being stimulated in a certain way, and still\nless likely to be mistaken if we say that the optic nerve is being\nstimulated in a certain way. We do not eliminate the risk of error\ncompletely unless we confine ourselves to saying that an event of a\ncertain sort is happening in the brain; this statement may still be\ntrue if we see Jupiter in a dream.\n\nBut, I shall be asked, what do you know about what is happening in the\nbrain? Surely nothing. Not so, I reply. I know about what is happening\nin the brain exactly what naive realism thinks it knows about what is\nhappening in the outside world. But this needs explaining, and there\nare other matters that must be explained first.\n\nWhen the light from a fixed star reaches me, I see the star if it is\nnight and I am looking in the right direction. The light started years\nago, probably many years ago, but my reaction is primarily to something\nthat is happening _now_. When my eyes are open, I see the star; when\nthey are shut, I do not. Children discover at a fairly early age\nthat they see nothing when their eyes are shut. They are aware of the\ndifference between seeing and not seeing, and also of the difference\nbetween eyes open and eyes shut; gradually they discover that these\ntwo differences are correlated--I mean that they have expectations of\nwhich this is the intellectualist transcription. Again, children learn\nto name the colours, and to state correctly whether a thing is blue\nor red or yellow or what-not. They ought not to be sure that light of\nthe appropriate wave-length started from the object. The sun looks red\nin a London fog, grass looks blue through blue spectacles, everything\nlooks yellow to a person suffering from jaundice. But suppose you ask:\nWhat colour are you seeing? The person who answers, in these cases,\nred for the sun, blue for the grass, and yellow for the sick-room of\nthe jaundiced patient, is answering quite truly. And in each of these\ncases he is stating something that he _knows_. What he knows in such\ncases is what I call a \u201cpercept\u201d. I shall contend later that, from the\nstandpoint of physics, a percept is in the brain; for the present, I am\nonly concerned to say that a percept is what is most indubitable in our\nknowledge of the world.\n\nTo behaviourism as a metaphysic one may put the following dilemma.\nEither physics is valid in its main lines, or it is not. If it is\nnot, we know nothing about the movements of matter; for physics is\nthe result of the most serious and careful study of which the human\nintelligence has hitherto been capable. If, on the other hand, physics\nis valid in its main lines, any physical process starting either inside\nor outside the body will, if it reaches the brain, be different if\nthe intervening medium is different; moreover two persons, initially\nvery different, may become indistinguishable as they spread and grow\nfainter. On both grounds, what happens in the brain is not connected\nquite accurately with what happens elsewhere, and our perceptions\nare therefore infected with subjectivity on purely physical grounds.\nEven, therefore, when we assume the truth of physics, what we know\nmost indubitably through perception is not the movements of matter,\nbut certain events in ourselves which are connected, in a manner not\nquite invariable, with the movements of matter. To be specific, when\nDr. Watson watches rats in mazes, what he knows, apart from difficult\ninferences, are certain events in himself. The behaviour of the rats\ncan only be inferred by the help of physics, and is by no means to be\naccepted as something accurately knowable by direct observation.\n\nI do not in fact entertain any doubts that physics is true in its main\nlines. The interpretation of physical formul\u00e6 is a matter as to which\na considerable degree of uncertainty is possible; but we cannot well\ndoubt that there is an interpretation which is true roughly and in\nthe main. I shall come to the question of interpretation later; for\nthe present, I shall assume that we may accept physics in its broad\noutlines, without troubling to consider how it is to be interpreted.\nOn this basis, the above remarks on perception seem undeniable. We\nare often misled as to what is happening, either by peculiarities of\nthe medium between the object and our bodies, or by unusual states of\nour bodies, or by a temporary or permanent abnormality in the brain.\nBut in all these cases _something_ is really happening, as to which,\nif we turn our attention to it, we can obtain knowledge that is not\nmisleading. At one time when, owing to illness, I had been taking a\ngreat deal of quinine, I became hypersensitive to noise, so that when\nthe nurse rustled the newspaper I thought she was spilling a scuttle\nof coals on the floor. The interpretation was mistaken, but it was\nquite true that I heard a loud noise. It is a commonplace that a man\nwhose leg has been amputated can still feel pains in it; here again,\nhe does really feel the pains, and is only mistaken in his belief that\nthey come from his leg. A percept is an observable event, but its\ninterpretation as knowledge of this or that event in the physical world\nis liable to be mistaken, for reasons which physics and physiology can\nmake fairly clear.\n\nThe subjectivity of percepts is a matter of degree. They are more\nsubjective when people are drunk or asleep than when they are sober\nand awake. They are more subjective in regard to distant objects than\nin regard to such as are near. They may acquire various peculiar kinds\nof subjectivity through injuries to the brain or to the nerves. When\nI speak of a percept as \u201csubjective\u201d I mean that the physiological\ninferences to which it gives rise are mistaken or vague. This is\nalways the case to some extent, but much more so in some circumstances\nthan in others. And the sort of defect that leads to mistakes must be\ndistinguished from the sort that leads to vagueness. If you see a man a\nquarter of a mile away, you can see that it is a man if you have normal\neyesight, but you probably cannot tell who it is, even if in fact\nit is some one you know well. This is vagueness in the percept: the\ninferences you draw are correct so far as they go, but they do not go\nvery far. On the other hand, if you are seeing double and think there\nare two men, you have a case of mistake. Vagueness, to a greater or\nless extent, is universal and inevitable; mistakes, on the other hand,\ncan usually be avoided by taking trouble and by not always trusting\nto physiological inference. Anybody can see double on purpose, by\nfocussing on a distant object and noticing a near one; but this will\nnot cause mistakes, since the man is aware of the subjective element in\nhis double vision. Similarly we are not deceived by after-images, and\nonly dogs are deceived by gramophones.\n\nFrom what has been said in this chapter, it is clear that our knowledge\nof the physical world, if it is to be made as reliable as possible,\nmust start from percepts, and must scrutinize the physiological\ninferences by which percepts are accompanied. Physiological inference\nis inference in the sense that it sometimes leads to error and physics\ngives reason to expect that percepts will, in certain circumstances,\nbe more or less deceptive if taken as signs of something outside the\nbrain. It is these facts that give a subjective cast to the philosophy\nof physics, at any rate in its beginnings. We cannot start cheerfully\nwith a world of matter in motion, as to which any two sane and sober\nobservers must agree. To some extent, each man dreams his own dream,\nand the disentangling of the dream element in our percepts is no easy\nmatter. This is, indeed, the work that scientific physics undertakes to\ndo.\n\n\n\n\nCHAPTER XIII\n\nPHYSICAL AND PERCEPTUAL SPACE\n\n\nPerhaps there is nothing so difficult for the imagination as to teach\nit to feel about space as modern science compels us to think. This is\nthe task which must be attempted in the present chapter.\n\nWe said in Chapter XII that we know about what is happening in the\nbrain exactly what naive realism thinks it knows about what is\nhappening in the world. This remark may have seemed cryptic; it must\nnow be expanded and expounded.\n\nThe gist of the matter is that percepts, which we spoke about at the\nend of last chapter, are in our heads; that percepts are what we can\nknow with most certainty; and that percepts contain what naive realism\nthinks it knows about the world.\n\nBut when I say that my percepts are in my head, I am saying something\nwhich is ambiguous until the different kinds of space have been\nexplained, for the statement is only true in connection with _physical_\nspace. There is also a space in our percepts, and of this space\nthe statement would not be true. When I say that there is space\nin our percepts, I mean nothing at all difficult to understand. I\nmean--to take the sense of sight, which is the most important in this\nconnection--that in what we see at one time there is up and down, right\nand left, inside and outside. If we see, say, a circle on a blackboard,\nall these relations exist within what we see. The circle has a top half\nand a bottom half, a right-hand half and a left-hand half, an inside\nand an outside. Those relations alone are enough to make up a space\nof sorts. But the space of everyday life is filled out with what we\nderive from touch and movement--how a thing feels when we touch it,\nand what movements are necessary in order to grasp it. Other elements\nalso come into the genesis of the space in which everybody believes\nwho has not been troubled by philosophy; but it is unnecessary for\nour purposes to go into this question any more deeply. The point that\nconcerns us is that a man\u2019s percepts are private to himself: what I\nsee, no one else sees; what I hear, no one else hears; what I touch,\nno one else touches; and so on. True, others hear and see something\nvery like what I hear and see, if they are suitably placed; but there\nare always differences. Sounds are less loud at a distance; objects\nchange their visual appearance according to the laws of perspective.\nTherefore it is impossible for two persons at the same time to have\nexactly identical percepts. It follows that the space of percepts, like\nthe percepts, must be private; there are as many perceptual spaces as\nthere are percipients. My percept of a table is outside my percept of\nmy head, in my perceptual space; but it does not follow that it is\noutside my head as a physical object in physical space. Physical space\nis neutral and public: in this space, all my percepts are in my head,\neven the most distant star _as I see it_. Physical and perceptual space\nhave relations, but they are not identical, and failure to grasp the\ndifference between them is a potent source of confusion.\n\nTo say that you see a star when you see the light that has come from it\nis no more correct than to say that you see New Zealand when you see\na New Zealander in London. Your perception when (as we say) you see a\nstar is causally connected, in the first instance, with what happens\nin the brain, the optic nerve, and the eye, then with a light-wave\nwhich, according to physics, can be traced back to the star as its\nsource. Your sensations will be closely similar if the light comes from\na lamp at the top of a mast. The physical space in which you believe\nthe \u201creal\u201d star to be is an elaborate inference; what is given is the\nprivate space in which the speck of light you see is situated. It is\nstill an open question whether the space of sight has depth, or is\nmerely a surface, as Berkeley contended. This does not matter for\nour purposes. Even if we admit that sight alone shows a difference\nbetween an object a few inches from the eyes and an object several feet\ndistant, yet you certainly cannot, by sight alone, see that a cloud\nis less distant than a fixed star, though you may _infer_ that it is,\nbecause it can hide the star. The world of astronomy, from the point of\nview of sight, is a surface. If you were put in a dark room with little\nholes cut in the ceiling in the pattern of the stars letting light come\nthrough, there would be nothing in your immediate visual data to show\nthat you were not \u201cseeing the stars\u201d. This illustrates what I mean by\nsaying that what you see is _not_ \u201cout there\u201d in the sense of physics.\n\nWe learn in infancy that we can sometimes touch objects we see, and\nsometimes not. When we cannot touch them at once, we can sometimes do\nso by walking to them. That is to say, we learn to correlate sensations\nof sight with sensations of touch, and sometimes with sensations of\nmovement followed by sensations of touch. In this way we locate our\nsensations in a three-dimensional world. Those which involve sight\nalone we think of as \u201cexternal\u201d, but there is no justification for this\nview. What you see when you see a star is just as internal as what you\nfeel when you feel a headache. That is to say, it is internal from the\nstandpoint of _physical_ space. It is distant in your private space,\nbecause it is not associated with sensations of touch, and cannot be\nassociated with them by means of any journey you can perform.\n\nYour own body, as known to you through direct experience, is quite\ndifferent from your own body as considered in physics. You know more\nabout your own body than about any other through direct experience,\nbecause your own body can give you a number of sensations that no other\nbody can, for instance all kinds of bodily pains. But you still know\nit only through sensations; apart from inference, it _is_ a bundle of\nsensations, and therefore quite different, _prima facie_, from what\nphysics calls a body.\n\nMost of the things you see are outside what you see when (as one says)\nyou see your own body. That is to say: you see certain other patches of\ncolour, differently situated in visual space, and say you are seeing\nthings outside your body. But from the point of view of physics, all\nthat you see must count as inside your body; what goes on elsewhere\ncan only be inferred. Thus the whole space of your sensible world with\nall its percepts counts as one tiny region from the point of view of\nphysics.\n\nThere is no direct spatial relation between what one person sees and\nwhat another sees, because no two ever see exactly the same object.\nEach person carries about a private space of his own, which can be\nlocated in physical space by indirect methods, but which contains no\nplace in common with another person\u2019s private space. This shows how\nentirely physical space is a matter of inference and construction.\n\nTo make the matter definite, let us suppose that a physiologist is\nobserving a living brain--no longer an impossible supposition, as\nit would have been formerly. It is natural to suppose that what the\nphysiologist sees is in the brain he is observing. But if we are\nspeaking of physical space, what the physiologist sees is in his\nown brain. It is in no sense in the brain that he is observing,\nthough it is in the percept of that brain, which occupies part of\nthe physiologist\u2019s perceptual space. Causal continuity makes the\nmatter perfectly evident: light-waves travel from the brain that is\nbeing observed to the eye of the physiologist, at which they only\narrive after an interval of time, which is finite though short. The\nphysiologist sees what he is observing only after the light-waves have\nreached his eye; therefore the event which constitutes his seeing comes\nat the end of a series of events which travel from the observed brain\ninto the brain of the physiologist. We cannot, without a preposterous\nkind of discontinuity, suppose that the physiologist\u2019s percept,\nwhich comes at the end of this series, is anywhere else but in the\nphysiologist\u2019s head.\n\nThis question is very important, and must be understood if metaphysics\nis ever to be got straight. The traditional dualism of mind and\nmatter, which I regard as mistaken, is intimately connected with\nconfusions on this point. So long as we adhere to the conventional\nnotions of mind and matter, we are condemned to a view of perception\nwhich is miraculous. We suppose that a physical process starts from a\nvisible object, travels to the eye, there changes into another physical\nprocess, causes yet another physical process in the optic nerve,\nfinally produces some effect in the brain, simultaneously with which\nwe see the object from which the process started, the seeing being\nsomething \u201cmental\u201d, totally different in character from the physical\nprocesses which precede and accompany it. This view is so queer\nthat metaphysicians have invented all sorts of theories designed to\nsubstitute something less incredible. But nobody noticed an elementary\nconfusion.\n\nTo return to the physiologist observing another man\u2019s brain: what the\nphysiologist sees is by no means identical with what happens in the\nbrain he is observing, but is a somewhat remote effect. From what he\nsees, therefore, he cannot judge whether what is happening in the brain\nhe is observing is, or is not, the sort of event that he would call\n\u201cmental\u201d. When he says that certain physical events in the brain are\naccompanied by mental events, he is thinking of physical events as if\nthey were what he sees. He does not see a mental event in the brain he\nis observing, and therefore supposes there is in that brain a physical\nprocess which he can observe and a mental process which he cannot.\nThis is a complete mistake. In the strict sense, he cannot observe\nanything in the other brain, but only the percepts which he himself has\nwhen he is suitably related to that brain (eye to microscope, etc.).\nWe first identify physical processes with our percepts, and then,\nsince our percepts are not other people\u2019s thoughts, we argue that the\nphysical processes in their brains are something quite different from\ntheir thoughts. In fact, everything that we can directly observe of\nthe physical world happens inside our heads, and consists of \u201cmental\u201d\nevents in at least one sense of the word \u201cmental\u201d. It also consists\nof events which form part of the physical world. The development of\nthis point of view will lead us to the conclusion that the distinction\nbetween mind and matter is illusory. The stuff of the world may be\ncalled physical or mental or both or neither, as we please; in fact,\nthe words serve no purpose. There is only one definition of the words\nthat is unobjectionable: \u201cphysical\u201d is what is dealt with by physics,\nand \u201cmental\u201d is what is dealt with by psychology. When, accordingly, I\nspeak of \u201cphysical\u201d space, I mean the space that occurs in physics.\n\nIt is extraordinarily difficult to divest ourselves of the belief that\nthe physical world is the world we perceive by sight and touch; even\nif, in our philosophic moments, we are aware that this is an error,\nwe nevertheless fall into it again as soon as we are off our guard.\nThe notion that what we see is \u201cout there\u201d in physical space is one\nwhich cannot survive while we are grasping the difference between what\nphysics supposes to be really happening, and what our senses show us\nas happening; but it is sure to return and plague us when we begin to\nforget the argument. Only long reflection can make a radically new\npoint of view familiar and easy.\n\nOur illustrations hitherto have been taken from the sense of sight;\nlet us now take one from the sense of touch. Suppose that, with your\neyes shut, you let your finger-tip press against a hard table. What\nis really happening? The physicist says that your finger-tip and the\ntable consist, roughly speaking, of vast numbers of electrons and\nprotons; more correctly, each electron and proton is to be thought of\nas a collection of processes of radiation, but we can ignore this for\nour present purposes. Although you think you are touching the table,\nno electron or proton in your finger ever really touches an electron\nor proton in the table, because this would develop an infinite force.\nWhen you press, repulsions are set up between parts of your finger and\nparts of the table. If you try to press upon a liquid or a gas, there\nis room in it for the parts that are repelled to get away. But if you\npress a hard solid, the electrons and protons that try to get away,\nbecause electrical forces from your finger repel them, are unable to\ndo so, because they are crowded close to others which elbow them back\nto more or less their original position, like people in a dense crowd.\nTherefore the more you press the more they repel your finger. The\nrepulsion consists of electrical forces, which set up in the nerves a\ncurrent whose nature is not very definitely known. This current runs\ninto the brain, and there has effects which, so far as the physiologist\nis concerned, are almost wholly conjectural. But there is one effect\nwhich is not conjectural, and that is the sensation of touch. This\neffect, owing to physiological inference or perhaps to a reflex, is\nassociated by us with the finger-tip. But the sensation is the same\nif, by artificial means, the parts of the nerve nearer the brain are\nsuitably stimulated--_e.g._ if your hand has been amputated and the\nright nerves are skilfully manipulated. Thus our confidence that touch\naffords evidence of the existence of bodies at the place which we think\nis being touched is quite misplaced. As a rule we are right, but we can\nbe wrong; there is nothing of the nature of an infallible revelation\nabout the matter. And even in the most favorable case, the perception\nof touch is something very different from the mad dance of electrons\nand protons trying to jazz out of each other\u2019s way, which is what\nphysics maintains is really taking place at your finger-tip. Or, at\nleast, it _seems_ very different. But as we shall see, the knowledge\nwe derive from physics is so abstract that we are not warranted\nin saying that what goes on in the physical world is, or is not,\nintrinsically very different from the events that we know through our\nown experiences.\n\n\n\n\nCHAPTER XIV\n\nPERCEPTION AND PHYSICAL CAUSAL LAWS\n\n\nIn an earlier chapter we saw the inadequacy of the traditional notion\nof cause, without adequately explaining the causal laws which are a\nsubstitute in the practice of science. The time has now come when it\nis possible to remedy this defect, and, in so doing, to fit perception\ninto its place in the chain of physical causation and recapitulate the\nmain points of previous arguments.\n\nThe old view was that an event A will always be followed by a certain\nevent B, and that the problem of discovering causal laws is the\nproblem, given an event B, of finding that event A which is its\ninvariable antecedent or _vice versa_. At an early stage of a science\nthis point of view is useful; it gives laws which are true usually,\nthough probably not always, and it affords the basis for more exact\nlaws. But it has no philosophical validity, and is superseded in\nscience as soon as we arrive at genuine laws. Genuine laws, in advanced\nsciences, are practically always quantitative laws of _tendency_. I\nwill try to illustrate by taking the simplest possible case in physics.\n\nImagine a hydrogen atom, in which the electron is revolving not in\nthe minimum orbit, but in the next, which has four times the minimum\nradius. So long as this state continues, the atom has no external\neffects, apart from its infinitesimal gravitational action; we cannot,\ntherefore, obtain any evidence of its existence except when it changes\nits state. In fact, our knowledge of atoms is like that which a ticket\ncollector has of the population of his town: he knows nothing of those\nwho stay quietly at home. Now at some moment, according to laws of\nwhich we have only statistical knowledge, the electron in our atom\njumps to a smaller orbit, and the energy lost to the atom travels\noutward in a light-wave. We know no causal law as to when the electron\nwill jump, though we know how far it will jump and exactly what will\nhappen in the neighbourhood when it does. At least, when I say we\nknow exactly what will happen, I ought to say that we know exactly\nthe mathematical laws of what will happen. A series of events, having\nquantitative characteristics which obey certain equations, will travel\noutward in all directions from the electron, and will proceed quite\nregularly, like ripples on a pool, until other matter is encountered.\nWe have here one important and apparently fundamental kind of causal\nlaw, the kind regulating the propagation of light _in vacuo_. This is\nsummed up in Maxwell\u2019s equations, which enable us to calculate the\ndiffusion of an electro-magnetic disturbance starting from a source. So\nlong as two such disturbances do not meet, the matter is exceedingly\nsimple; but the equations also tell us what happens when they do\nmeet. We then have, as always in traditional physics, two separate\ntendencies, which have a resultant compounded according to mathematical\nlaws, of which the parallelogram law is the oldest and simplest. That\nis to say, each previous circumstance in the space-time neighbourhood\ncontributes a tendency, and the resulting event is obtained by\ncompounding these tendencies according to a mathematical law.\n\nSo far, we have been considering only electro-magnetic phenomena in\nempty space. We have another set of facts about empty space, namely\nthose upon which gravitation depends. These have to do with the\nstructure of space-time, and show that this structure has singularities\nin the regions where there is matter, which spread with diminishing\nintensity as we get away from these regions. You may conceive the\nstructure of space-time on the analogy of a pond with a fountain\nplaying in it, so that wherever a spray falls from the fountain there\nis a little hill of water which flattens quickly as you get away from\nthe spot where the spray falls. Here again the same sort of thing\napplies: to infer the structure in a small region of space-time from\nthat in the neighbourhood, it will be necessary to superpose a number\nof tendencies according to mathematical rules. Thus philosophically\nthis introduces no novelty.\n\nBut now consider what happens when the wave of light which started\nfrom our hydrogen atom comes in contact with matter. Various things\nmay happen. The matter may absorb all or some of the energy of the\nlight-ray; this is the interesting case from our point of view. The\nabsorption may take the form of causing the electrons to move in larger\norbits, in which case, later, when they return to their previous\norbits, we get the phenomenon of fluorescence. Or the body may become\nheated; or it may visibly move, like a radiometer. The effects upon\nbodies depend upon the bodies as well as the light. Some of them can be\nindividually predicted, others can only be calculated in statistical\naverages; this depends upon whether quantum considerations come in\nor not. Where they do, we can enumerate possibilities, and state the\nrelative frequencies with which they will be realised, but we cannot\ntell which will be realised in any given case.\n\nSo far, we have considered the radiation of energy from matter into\nempty space, its propagation in empty space and its impact on matter\nfrom empty space. We have not considered the history of a given piece\nof matter, or the distinction between matter and empty space.\n\nThe essence of matter appears to be this: We can distinguish series of\nevents in space-time which have a certain kind of close resemblance\nto each other, such that common sense regards them as manifestations\nof one \u201cthing\u201d. But when we look closely at the question, it turns\nout that what physics offers is something more abstract than this.\nTake, _e. g._ the continued existence of a certain electron. This\nmeans to say that events in a certain neighbourhood will be such as\ncan be calculated on the assumption that there is an electric charge\nof a certain standard magnitude in the middle of that neighbourhood;\nand that the neighbourhoods of which this is true form a tube in\nspace-time.\n\nSo long as we stick to the standpoint of pure physics there is a\ncertain air of taking in each other\u2019s washing about the whole business.\nEvents in empty space are only known as regards their abstract\nmathematical characteristics; matter is only an abstract mathematical\ncharacteristic of events in empty space. This seems rather a cold\nworld. But as a matter of fact we know some things that are a little\nmore concrete. We know, _e.g._ what it feels like when we see things.\nFrom the point of view of physics, when our light-wave starts out\nthrough empty space, if it presently reaches our eye we know one link\nin the causal chain, namely the visual sensation, otherwise than as a\nterm in an abstract mathematical formula. And it is this one term which\nforms the basis for our belief in all the rest. Seeing is believing.\n\nAt this point I propose to make a brief digression on the subject of\nour evidence for causal laws. The laws for which we first get evidence\nare such as do not hold always, but only as a general rule. As a\nrule, when you decide to move your arm, it moves: but sometimes it is\nparalysed and remains motionless. As a rule, when you say how-do-you-do\nto an old friend, he says the same to you; but he may have grown blind\nand deaf since you last saw him, and not notice your words or gesture.\nAs a rule, if you put a match to gunpowder, it explodes; but it may\nhave got damp. It is such common but not invariable rules of sequence\nthat we notice first. But science is always seeking to replace them by\nlaws that may have no exceptions. We notice first that heavy bodies\nfall, then that some bodies do not fall. Then we generalise both sets\nof facts into the law of gravitation and the laws of resistance of the\nair. These more general laws do not state that anything will actually\nhappen: they state a tendency, and lead to the conclusion that what\nactually happens is the resultant of a number of tendencies. We cannot\nknow what the resultant will be unless we know a great deal about\nthe neighbourhood concerned. For example, I might, within the next\nfew seconds be hit on the head by a meteorite; to know whether this\nis going to happen, I must know what matter is to be found in the\nneighbourhood of the earth. This illustrates that actual predictions\nbased upon laws which are perfectly valid may always be falsified by\nsome unknown fact of what we may call geography. Moreover, we can\nnever be _sure_ that our scientific laws are quite right; of this the\nEinsteinian modification of the law of gravitation has afforded a\nnotable instance.\n\nLet us now return to the relation between perception and the causal\nlaws of physics.\n\nHaving realised the abstractness of what physics has to say, we no\nlonger have any difficulty in fitting the visual sensation into the\ncausal series. It used to be thought \u201cmysterious\u201d that purely physical\nphenomena should end in something mental. That was because people\nthought they knew a lot about physical phenomena, and were sure they\ndiffered in quality from mental phenomena. We now realise that we know\nnothing of the intrinsic quality of physical phenomena except when\nthey happen to be sensations, and that therefore there is no reason to\nbe surprised that some are sensations, or to suppose that the others\nare totally unlike sensations. The gap between mind and matter has\nbeen filled in, partly by new views of mind, but much more by the\nrealisation that physics tells us nothing as to the intrinsic character\nof matter.\n\nI conceive what happens when we see an object more or less on the\nfollowing lines. For the sake of simplicity, let us take a small\nself-luminous object. In this object, a certain number of atoms are\nlosing energy and radiating it according to the quantum principle.\nThe resulting light-waves become superposed according to the usual\nmathematical principles; each part of each light-wave consists of\nevents in a certain region of space-time. On coming in contact with the\nhuman body, the energy in the light-wave takes new forms, but there\nis still causal continuity. At last it reaches the brain, and there\none of its constituent events is what we call a visual sensation. This\nvisual sensation is popularly called seeing the object from which the\nlight-waves started--or from which they were reflected if the object\nwas not self-luminous.\n\nThus what is called a perception is only connected with its object\nthrough the laws of physics. Its relation to the object is causal and\nmathematical; we cannot say whether or not it resembles the object in\nany intrinsic respect, except that both it and the object are brief\nevents in space-time.\n\nI think we may lay down the following universal characteristics of\ncausal laws in an advanced science. Given any event, there are other\nevents at neighbouring places in space-time which will occur slightly\nlater if no other factors intervene; but in practice other factors\nalmost always do intervene, and, in that case, the event which actually\noccurs at any point of space-time is a mathematical resultant of those\nwhich would have followed the various neighbouring events if they had\nbeen alone concerned. The equations of physics give the rules according\nto which events are connected, but all are of the above sort.\n\nFormerly it was thought that the equations of physics sufficed,\ntheoretically, to determine the course of affairs in the physical\nworld, given all the facts about some finite stretch of time, however\nshort. Now it appears that this is not the case, so far as the known\nequations are concerned. The known equations suffice to determine what\nhappens in empty space, and statistical averages as to what happens to\nmatter; but they do not tell us when an individual atom will absorb or\nradiate energy. Whether there are laws, other than those of statistics,\ngoverning the behaviour of an individual atom in this respect, we do\nnot know.\n\nIt should be observed that there are causal laws of a different sort\nfrom those of pure physics; such are the laws that light-waves \u201ccause\u201d\nvisual sensations and sound-waves \u201ccause\u201d auditory sensations. All the\nempirical evidence for physics rests upon such laws, therefore nothing\nin physics can have a higher degree of certainty than such laws have.\nLet us stop a moment to ask what we mean by \u201ccause\u201d in this connection.\n\nThe connection of light-waves and visual sensations looks a little\ndifferent according as we start with physics or with psychology,\nthough, of course, ultimately the result must be the same. Let us\nfirst start with physics. I say, then, that when a light-wave travels\noutwards from a body there are successive events at successive places,\nand that the corresponding event in a brain behind a normal eye is a\nvisual sensation. This is the only event in the whole series about\nwhich I can say anything not purely abstract and mathematical.\n\nNow let us start from the sensation. I say, then, that this sensation\nis one of a vast series of connected events, travelling out from a\ncentre according to certain mathematical laws, in virtue of which the\nsensation enables me to know a good deal about events elsewhere. That\nis why the sensation is a source of physical knowledge.\n\nIt will be seen that, according to the view I have been advocating,\nthere is no difficulty about interaction between mind and body. A\nsensation is merely one link in a chain of physical causation; when we\nregard the sensation as the end of such a chain, we have what would\nbe regarded as an effect of matter on mind; when as the beginning,\nan effect of mind on matter. But mind is merely a cross-section in a\nstream of physical causation, and there is nothing odd about its being\nboth an effect and a cause in the physical world. Thus physical causal\nlaws are those that are fundamental.\n\nThere seems no reason to regard causation as _a priori_, though this\nquestion is not simple. Given certain very general assumptions as to\nthe structure of space-time, there are bound to be what we have called\ncausal laws. These general assumptions must really replace causality\nas our basic principles. But, general as they are, they cannot be\ntaken as _a priori_; they are the generalisation and abstract epitome\nof the fact that there are causal laws, and this must remain merely\nan empirical fact, which is rendered probable, though not certain, by\ninductive arguments.\n\n\n\n\nCHAPTER XV\n\nTHE NATURE OF OUR KNOWLEDGE OF PHYSICS\n\n\nIn this chapter, we shall seek an answer to two questions: First, _how_\ndo we know about the world dealt with in physics? Secondly, _what_ do\nwe know about it, assuming the truth of modern physics?\n\nFirst: How do we know about the physical world? We have already seen\nthat this question cannot have a simple answer, since the basis of\ninference is something that happens in our own heads, and our knowledge\nof anything outside our own heads must be more or less precarious. For\nthe present, I shall take it for granted that we may accept testimony,\nwith due precautions. In other words, I shall assume that what we hear\nwhen, as we believe, others are speaking to us, does in fact have\n\u201cmeaning\u201d to the speaker, and not only to us; with a corresponding\nassumption as regards writing. This assumption will be examined at a\nlater stage. For the present, I will merely emphasise that it _is_ an\nassumption, and that it may possibly be false, since people seem to\nspeak to us in dreams, and yet, on waking, we become persuaded that\nwe invented the dream. It is impossible to prove, by a demonstrative\nargument, that we are not always dreaming; the best we can hope is a\nproof that this is improbable. But for the present let us leave this\ndiscussion on one side, and assume that the words we hear and read\n\u201cmean\u201d what they would if we spoke or wrote them.\n\nOn this basis, we have reason to know that the worlds of different\npeople are alike in certain respects and different in certain others.\nTake, for example, the audience at a theatre: they all, we say, hear\nthe same words and see the same gestures, which, moreover, are those\nthat the actors wish them to hear and see. But those who are near the\nstage hear the words more loudly than those further off; they also\nhear them somewhat earlier. And those who sit on the right do not see\nquite what is seen by those who sit on the left or in the centre. These\ndifferences are of two sorts: on the one hand, some people can see\nsomething invisible to others; on the other hand, when two people, as\nwe say, see the \u201csame\u201d thing, they see it differently owing to effects\nof perspective and of the way the light is reflected. All this is a\nquestion of physics, not of psychology; for if we place a camera in an\nempty seat in the theatre, the perspective in the resultant photograph\nis intermediate between the perspectives that are seen by persons\nsitting on either side; indeed the whole matter of perspective is\ndetermined by quite simple geometrical laws. These laws show also what\nis common to the shapes that two people see when they see the \u201csame\u201d\nthing from different points of view: what is common is what is studied\nby projective geometry, which is concerned with what is independent of\nmeasurement in geometrical figures. All the differences in appearance\ndue to perspective have to be learned in learning to draw: for this\npurpose, it is necessary to learn to see things as they really seem,\nand not as they seem to seem.\n\nBut, it will be said, what can you mean by how things \u201creally\u201d seem\nand how they \u201cseem\u201d to seem? We come here upon an important fact about\nlearning. When, in early infancy, we are learning to correlate sight\nand touch, we acquire the habit of reacting to a visual stimulus in a\nmanner which is more \u201cobjective\u201d than that in which a camera reacts.\nWhen we see a coin not directly in front of us, we judge it to be\ncircular, although the camera would show it as oval, and a man would\nhave to make it oval in a picture of a scene which contained it. We\nlearn, therefore, to react to a visual shape in a manner corresponding\nto how it would appear if it were in the centre of the field of vision,\nprovided we do not immediately focus upon it, which is what we\nnaturally do when anything visible interests us. As a matter of fact,\nwe are constantly looking in different directions, and, as a rule, only\nnoticing what, at the moment, is in the centre of our field of vision.\nThus our visual world consists rather of a synthesis of things viewed\ndirectly in succession than of things seen simultaneously while the\ncentre of the field is kept fixed. This is one reason why the rules of\nperspective have to be learned, although a picture which ignores them\nmakes an impression of being \u201cwrong\u201d.\n\nAnother reason for the objectivity of the impressions we derive from\nsight is correlation with other senses, especially touch. Through\nthis correlation we soon get to \u201cknow\u201d that a man twenty yards away\nis \u201creally\u201d just as big as a man one yard away. When children are\nlearning to draw, they find it very difficult to make distant objects\nsufficiently small, because they know they are not \u201creally\u201d small. We\nsoon learn to judge the distance of a visual object, and to react to\nit according to the size that it would have if touched--or travelled\nover--in the case of very large objects such as mountains. Our sense\nof size is not derived from sight, but from such sources as touch and\nlocomotion; our metrical judgments, when the stimulus is only visual,\nare a result of previous experience.\n\nBy the time a child can speak well, he has had a great deal of this\nkind of experience. Consequently our verbal reactions contain a great\ndeal more objectivity than they would if they came at an earlier stage\nof infancy. The result is that a number of people can view a scene\nsimultaneously, and use exactly the same words about it. The words\nwhich we naturally use in describing what we see are those describing\nfeatures that will also be evident to others in our neighbourhood. We\nsay, \u201cthere is a man\u201d, not, \u201cthere is a coloured shape whose visual\ndimensions are such-and-such an angle vertically, and such another\nhorizontally\u201d. The inference is a physiological inference, and only\nsubsequent reflection makes us aware that it has taken place. We\ncan, however, become aware of it through occasional mistakes; a dot\non the window-pane may be mistaken for a man in a distant field. In\nthis case, we can discover our error by opening the window, or by\nmoving the head. In general, however, physiological inferences of this\nsort are correct, since they have resulted from correlations which\nare very common, and are likely to be present on a given occasion.\nConsequently our words tend to conceal what is private and peculiar in\nour impressions, and to make us believe that different people live in a\ncommon world to a greater extent than is in fact the case.\n\nWe have been using the word \u201cobjectivity\u201d in the preceding pages,\nand it is time to consider exactly what we mean by it. Suppose some\nscene--say in a theatre--is simultaneously seen by a number of people\nand photographed by a number of cameras. The impression made upon\na person or a camera is in some respects like that made upon other\npersons and cameras, in other respects different. We shall call the\nelements which are alike \u201cobjective\u201d elements in the impression,\nand those which are peculiar we shall call \u201csubjective\u201d. Thus those\nfeatures of shapes which are considered in projective geometry will\nbe objective, whereas those considered in metrical geometry (where\nlengths and angles are measured) cannot be made objective through sight\nalone, but demand the use of other senses. In the photographs, a man\non the stage will be longer if the camera is near the stage than if it\nis far off, assuming all the cameras to be alike. But if four actors\nare standing in a row in one photograph, they will be standing in a\nrow in another; this is an \u201cobjective\u201d feature of the impression. And\nthe differences in the visual impressions of a number of spectators\nwith normal eyesight are exactly analogous to the differences in the\nphotographs; so also are the likenesses. Thus the \u201csubjectivity\u201d that\nwe are speaking about at present is something belonging to the physical\nworld, not to psychology. It marks the fact that the stimulus, whether\nto an eye or to a camera, is not exactly the same wherever the eye or\nthe camera may be placed; there are features of the stimulus which are\nconstant (within limits), but there are others which are different from\nany two different points of view.\n\nThe tendency of our perceptions is to emphasise increasingly the\nobjective elements in an impression, unless we have some special\nreason, as artists have, for doing the opposite. This tendency begins\nbefore speech, is much accentuated after speech has been acquired,\nand is prolonged by scientific physics. The theory of relativity is\nonly the last term, so far, in the elimination of subjective elements\nfrom impressions. But it must not be supposed that the subjective\nelements are any less \u201creal\u201d than the objective elements; they are\nonly less important. They are less important because they do not point\nto anything beyond themselves as the others do. We want our senses to\ngive us information, _i.e._ to tell us about something more than our\nown momentary impression. We acquire information through our senses\nif we attend to the objective elements in the impression and ignore\nthe others; but the subjective elements are just as truly part of the\nactual impression. This is obvious as soon as we realise that the\ncamera is as subjective as we are.\n\nSuch considerations lead irresistibly to the scientific view that,\nwhen an object can be seen or photographed from a number of points\nof view, there is a connected set of events (light-waves) travelling\noutward from a centre; that, moreover, there are some respects in\nwhich all these events are alike, and others in which they differ one\nfrom another. We must not think of a light-wave as a \u201cthing\u201d, but as a\nconnected group of rhythmical events. The mathematical characteristics\nof such a group can be inferred by physics, within limits; but the\nintrinsic character of the component events cannot be inferred. The\nevents constituting light-waves are only known through their effects\nupon our eyes, optic nerves, and brains, and these effects are not\nthemselves light-waves, as is obvious from the fact that nerves and\nbrains are not transparent. Light in the physical world, therefore,\nmust consist of events which are in some way different from the events\nwhich happen when we see; but we cannot say more than this as to the\nintrinsic quality of these external events. Moreover, when a number of\npeople, as we say, \u201csee the same thing\u201d, what we have reason to believe\nis that light-waves emanating from a certain region have reached the\neyes of all these people. As to what is in the region from which the\nlight-waves come, we cannot tell.\n\nBut--so the plain man is tempted to argue--we can tell quite well,\nbecause we can touch objects that we see, and discover that there is\nsomething hard and solid in the place from which the light-waves come.\nOr, again, we may find that there is something there which, though not\nsolid, is very hot, and burns us when we try to touch it. We all feel\nthat touch gives more evidence of \u201creality\u201d than sight; ghosts and\nrainbows can be seen but not touched. One reason for this greater sense\nof reality is that our spatial relation to an object when we touch it\nwith our finger-tips is given, and therefore an object does not give\nsuch different impressions of touch to different people as it does of\nsight. Another reason is that there are a number of objects that can be\nseen but not touched--reflections, smoke, mist, etc.--and that these\nobjects are calculated to surprise the inexperienced. None of these\nfacts, however, justify the plain man in supposing that touch makes him\nknow real things as they are, though we are verbally forced to admit\nthat it brings him into \u201ccontact\u201d with them.\n\nWe have seen on an earlier occasion how complex is the physical and\nphysiological process leading from the object to the brain when we\ntouch something; and we have seen that illusions of touch can be\nproduced artificially. What we experience when we have a sensation\nof touch is, therefore, no more a revelation of the real nature of\nthe object touched than what we experience when we look at it. As a\nmatter of fact, if modern physics is to be believed, sight, prudently\nemployed, gives us a more delicate knowledge concerning objects\nthan touch can ever do. Touch, as compared with sight, is gross and\nmassive. We can photograph the path of an individual electron. We\nperceive colours which indicate the changes happening in atoms. We can\nsee faint stars even though the energy of the light that reaches us\nfrom them is inconceivably minute. Sight may deceive the unwary more\nthan touch, but for accurate scientific knowledge it is incomparably\nsuperior to any of the other senses.\n\nIt is chiefly through ideas derived from sight that physicists have\nbeen led to the modern conception of the atom as a centre from which\nradiations travel. We do not know what happens in the centre. The idea\nthat there is a little hard lump there, which _is_ the electron or\nproton, is an illegitimate intrusion of common-sense notions derived\nfrom touch. For aught we know, the atom may consist entirely of the\nradiations which come out of it. It is useless to argue that radiations\ncannot come out of nothing. We know that they come, and they do not\nbecome any more really intelligible by being supposed to come out of a\nlittle lump.\n\nModern physics, therefore, reduces matter to a set of events which\nproceed outward from a centre. If there is something further in the\ncentre itself, we cannot know about it, and it is irrelevant to\nphysics. The events that take the place of matter in the old sense are\ninferred from their effect on eyes, photographic plates, and other\ninstruments. What we know about them is not their intrinsic character,\nbut their structure and their mathematical laws. Their structure is\ninferred chiefly through the maxim \u201csame cause, same effect\u201d. It\nfollows from this maxim that if the effects are different, the causes\nmust be different; if, therefore, we see red and blue side by side,\nwe are justified in inferring that in the direction where we see\nred something different is happening from what is happening in the\ndirection where we see blue. By extensions of this line of argument\nwe arrive at the mathematical laws of the physical world. Physics is\nmathematical, not because we know so much about the physical world,\nbut because we know so little: it is only its mathematical properties\nthat we can discover. For the rest, our knowledge is negative. In\nplaces where there are no eyes or ears or brains there are no colours\nor sounds, but there are events having certain characteristics which\nlead them to cause colours and sounds in places where there are eyes\nand ears and brains. We cannot find out what the world looks like from\na place where there is nobody, because if we go to look there will be\nsomebody there; the attempt is as hopeless as trying to jump on one\u2019s\nown shadow.\n\nMatter as it appears to common sense, and as it has until recently\nappeared in physics, must be given up. The old idea of matter was\nconnected with the idea of \u201csubstance\u201d, and this, in turn, with a view\nof time that the theory of relativity shows to be untenable. The old\nview was that there is one cosmic time, and that, given any two events\nin any two parts of the universe, either they are simultaneous, or the\nfirst is earlier than the second, or the second earlier than the first.\nIt was thought that the time-order of the two events must always be\nobjectively definite, although _we_ might be unable to determine it.\nWe now find that this is not the case. Events which can be regarded as\nall in one place, or all parts of the history of one piece of matter,\nstill have a definite time-order. So do events in different places if a\nperson situated where the second takes place can see the first before\nthe second happens, or, more exactly, if light can travel from the\nplace of the one to the place of the other so as to reach the other\nplace before the second event. (Here we mean by a \u201cplace\u201d the position\nof a given piece of matter: however the matter may move relatively\nto other matter, it is always in the same \u201cplace\u201d from its own point\nof view.) But if light travelling from the place of the one event to\nthe place of the other event arrives at the place of the other event\nafter the other event has taken place, and conversely, then there is\nno definite objective time-order of the two events, and there is no\nreason for regarding either as earlier than the other; nor yet for\nregarding the two as simultaneous; ideally careful observers will judge\ndifferently according to the way in which they are moving. Thus time\nis not cosmic, but is to some extent individual and personal for each\npiece of matter.\n\nWhat do we mean by a \u201cpiece of matter\u201d in this statement? We do not\nmean something that preserves a simple identity throughout its history,\nnor do we mean something hard and solid, nor even a hypothetical\nthing-in-itself known only through its effects. We mean the \u201ceffects\u201d\nthemselves, only that we no longer invoke an unknowable cause for\nthem. We find that energy in various forms spreads outwards from\nvarious centres; we find also that such centres have a certain degree\nof persistence, though this persistence is not absolute--the modern\nphysicist faces cheerfully the possibility that an electron and a\nproton may mutually annihilate each other, and even suggests that this\nmay be the main source of the radiant energy of the stars, because\nwhen it happens it makes an explosion. What is asserted may be put as\nfollows: When energy radiates from a centre, we can describe the laws\nof its radiation conveniently by imagining something in the centre,\nwhich we will call an electron or a proton according to circumstances,\nand for certain purposes it is convenient to regard this centre as\npersisting, _i.e._ as not a single point in space-time but a series\nof such points, separated from each other by time-like intervals. All\nthis, however, is only a convenient way of describing what happens\nelsewhere, namely the radiation of energy away from the centre. As to\nwhat goes on in the centre itself, if anything, physics is silent.\n\nWhat Dr. Whitehead calls the \u201cpushiness\u201d of matter disappears\naltogether on this view. \u201cMatter\u201d is a convenient formula for\ndescribing what happens where it isn\u2019t. I am talking physics, not\nmetaphysics; when we come to metaphysics, we may be able, tentatively,\nto add something to this statement, but science alone can hardly add\nto it. Materialism as a philosophy becomes hardly tenable in view of\nthis evaporation of matter. But those who would formerly have been\nmaterialists can still adopt a philosophy which comes to much the\nsame thing in many respects. They can say that the type of causation\ndealt with in physics is fundamental, and that all events are subject\nto physical laws. I do not wish, as yet, to consider how far such a\nview should be adopted; I am only suggesting that it must replace\nmaterialism as a view to be seriously examined.\n\n\n\n\nPART III\n\nMAN FROM WITHIN\n\n\n\n\nCHAPTER XVI\n\nSELF-OBSERVATION\n\n\nIt will be remembered that throughout Part I, we agreed to consider\nonly those facts about a man which can be discovered by external\nobservation, and we postponed the question whether this excluded any\ngenuine knowledge or not. The usual view is that we know many things\nwhich could not be known without self-observation, but the behaviourist\nholds that this view is mistaken. I might be inclined to agree wholly\nwith the behaviourist but for the considerations which were forced upon\nus in Part II, when we were examining our knowledge of the physical\nworld. We were then led to the conclusion that, assuming physics to\nbe correct, the data for our knowledge of physics are infected with\nsubjectivity, and it is impossible for two men to observe the same\nphenomenon except in a rough and approximate sense. This undermines the\nsupposed objectivity of the behaviourist method, at least in principle;\nas a matter of degree, it may survive to some extent. Broadly speaking,\nif physics is true and if we accept a behaviourist definition of\nknowledge such as that of Chapter VIII, we ought, as a rule, to know\nmore about things that happen near the brain than about things that\nhappen far from it, and most of all about things that happen in the\nbrain. This _seemed_ untrue because people thought that what happens\nin the brain is what the physiologist sees when he examines it; but\nthis, according to the theory of Chapter XII, happens in the brain of\nthe physiologist. Thus the _a priori_ objection to the view that we\nknow best what happens in our brains is removed, and we are led back\nto self-observation as the most reliable way of obtaining knowledge.\nThis is the thesis which is to be expanded and sustained in the present\nchapter.\n\nAs every one knows, the certainty of self-observation was the basis\nof Descartes\u2019 system, with which modern philosophy began. Descartes,\nbeing anxious to build his metaphysic only upon what was absolutely\ncertain, set to work, as a preliminary, to doubt anything that he could\nmake himself doubt. He succeeded in doubting the whole external world,\nsince there might be a malicious demon who took pleasure in presenting\ndeceitful appearances to him. (For that matter, dreams would have\nsupplied a sufficient argument.) But he could not manage to doubt his\nown existence. For, said he, I am really doubting; whatever else may be\ndoubtful, the fact that I doubt is indubitable. And I could not doubt\nif I did not exist. He summed up the argument in his famous formula:\n_I think, therefore I am_. And having arrived at this certainty, he\nproceeded to build up the world again by successive inferences. Oddly\nenough, it was very like the world in which he had believed before his\nexcursion into scepticism.\n\nIt is instructive to contrast this argument with Dr. Watson\u2019s. Dr.\nWatson, like Descartes, is sceptical of many things which others\naccept without question; and, like Descartes, he believes that there\nare some things so certain that they can be safely used as the basis\nof a startling philosophy. But the things which Dr. Watson regards\nas certain are just those which Descartes regarded as doubtful, and\nthe thing which Dr. Watson most vehemently rejects is just what\nDescartes regarded as absolutely unquestionable. Dr. Watson maintains\nthat there is no such thing as thinking. No doubt he believes in\nhis own existence, but not because he thinks he can think. The\nthings that strike him as absolutely indubitable are rats in mazes,\ntime-measurements, physiological facts about glands and muscles, and so\non. What are we to think when two able men hold such opposite views?\nThe natural inference would be that _everything_ is doubtful. This may\nbe true, but there are degrees of doubtfulness, and we should like to\nknow which of these two philosophers, if either, is right as to the\nregion of minimum doubtfulness.\n\nLet us begin by examining Descartes\u2019 view. \u201cI think, therefore I am\u201d is\nwhat he says, but this won\u2019t do as it stands. What, from his own point\nof view, he should profess to know is not \u201c_I_ think\u201d, but \u201cthere is\nthinking\u201d. He finds doubt going on, and says: There is doubt. Doubt\nis a form of thought, therefore there is thought. To translate this\ninto \u201c_I_ think\u201d is to assume a great deal that a previous exercise\nin scepticism ought to have taught Descartes to call in question. He\nwould say that thoughts imply a thinker. But why should they? Why\nshould not a thinker be simply a certain series of thoughts, connected\nwith each other by causal laws? Descartes believed in \u201csubstance\u201d,\nboth in the mental and in the material world. He thought that there\ncould not be motion unless something moved, nor thinking unless some\none thought. No doubt most people would still hold this view; but in\nfact it springs from a notion--usually unconscious--that the categories\nof grammar are also the categories of reality. We have already seen\nthat \u201cmatter\u201d is merely a name for certain strings of sets of events.\nIt follows that what we call motion of matter really means that the\ncentre of such a set of events at one time does not have the same\nspatial relations to other events as the connected centre at another\ntime has to the connected other events. It does not mean that there is\na definite entity, a piece of matter, which is now in one place and\nnow in another. Similarly, when we say, \u201cI think first this and then\nthat\u201d, we ought not to mean that there is a single entity \u201cI\u201d which\n\u201chas\u201d two successive thoughts. We ought to mean only that there are two\nsuccessive thoughts which have causal relations of the kind that makes\nus call them parts of one biography, in the same sort of way in which\nsuccessive notes may be parts of one tune; and that these thoughts are\nconnected with the body which is speaking in the way (to be further\ninvestigated) in which thoughts and bodies are connected. All this is\nrather complicated, and cannot be admitted as part of any ultimate\ncertainty. What Descartes really felt sure about was a certain\noccurrence, which he described in the words \u201cI think\u201d. But the words\nwere not quite an accurate representation of the occurrence; indeed,\nwords never can escape from certain grammatical and social requirements\nwhich make them say at once more and less than we really mean. I think\nwe ought to admit that Descartes was justified in feeling sure that\nthere was a certain occurrence, concerning which doubt was impossible;\nbut he was not justified in bringing in the word \u201cI\u201d in describing this\noccurrence, and it remains to be considered whether he was justified in\nusing the word \u201cthink\u201d.\n\nIn using a general word such as \u201cthink\u201d, we are obviously going beyond\nthe datum, from a logical point of view. We are subsuming a particular\noccurrence under a heading, and the heading is derived from past\nexperience. Now all words are applicable to many occurrences; therefore\nall words go beyond any possible datum. In this sense, it is impossible\never to convey in words the particularity of a concrete experience; all\nwords are more or less abstract. Such, at least, is a plausible line\nof argument, but I am by no means sure that it is valid. For example,\nthe sight of a particular dog may make the general word \u201cdog\u201d come\ninto your mind; you then know that it is a dog, but you may not notice\nwhat sort of dog it is. In this sense, the knowledge with which we\nstart is abstract and general; that is to say, it consists of a learned\nreaction to a stimulus of a certain sort. The reactions, at any rate\nin so far as they are verbal, are more uniform than the stimuli. A\nwitness might be asked \u201cDid you see a dog?\u201d \u201cYes.\u201d \u201cWhat sort of dog?\u201d\n\u201cOh, just an ordinary dog; I don\u2019t remember more about it.\u201d That is to\nsay, the witness\u2019s reaction consisted of the generalised word \u201cdog\u201d\nand no more. One is almost reminded of quantum phenomena in the atom.\nWhen light falls on a hydrogen atom, it may make the electron jump from\nthe first orbit to the second, or to the third, etc. Each of these\nis a generalised reaction to a stimulus which has no corresponding\ngenerality. So dogs and cats have each their individual peculiarities,\nbut the ordinary inobservant person responds with the generalised\nreaction \u201cdog\u201d or \u201ccat\u201d, and the particularity of the stimulus leads to\nno corresponding particularity in the knowledge-reaction.\n\nTo return to Descartes and his thinking: it is possible, according to\nwhat we have just said, that Descartes knew he was thinking with more\ncertainty than he knew what he was thinking about. This possibility\nrequires that we should ask what he meant by \u201cthinking\u201d. And since, for\nhim, thinking was the primitive certainty, we must not introduce any\nexternal stimulus, since he considered it possible to doubt whether\nanything external existed.\n\nDescartes used the word \u201cthinking\u201d somewhat more widely than we should\ngenerally do nowadays. He included all perception, emotion, and\nvolition, not only what are called \u201cintellectual\u201d processes. We may\nperhaps with advantage concentrate upon perception. Descartes would say\nthat, when he \u201csees the moon\u201d, he is more certain of his visual percept\nthan he is of the outside object. As we have seen, this attitude is\nrational from the standpoint of physics and physiology, because a given\noccurrence in the brain is capable of having a variety of causes, and\nwhere the cause is unusual common sense will be misled. It would be\ntheoretically possible to stimulate the optic nerve artificially in\njust the way in which light coming from the moon stimulates it; in this\ncase, we should have the same experience as when we \u201csee the moon\u201d, but\nshould be deceived as to its external source. Descartes was influenced\nby an argument of this sort, when he brought up the possibility of a\ndeceitful demon. Therefore what he felt certain about was not what\nhe had initially felt certain about, but what remained certain after\nan argument as to the causes of perception. This brings us to a\ndistinction which is important, but difficult to apply; the distinction\nbetween what we in fact do not doubt, and what we should not doubt if\nwe were completely rational. We do not in fact question the existence\nof the sun and moon, though perhaps we might teach ourselves to do\nso by a long course of Cartesian doubt. Even then, according to\nDescartes, we could not doubt that we have the experiences which we\nhave hitherto called \u201cseeing the sun\u201d and \u201cseeing the moon\u201d, although\nwe shall need different words if we are to describe these experiences\ncorrectly.\n\nThe question arises: Why should we not doubt everything? Why should we\nremain convinced that we have these experiences? Might not a deceitful\ndemon perpetually supply us with false memories? When we say \u201ca moment\nago I had the experience which I have hitherto called seeing the sun\u201d,\nperhaps we are deceived. In dreams we often remember things that never\nhappened. At best, therefore, we can be sure of our present momentary\nexperience, not of anything that happened even half a minute ago.\nAnd before we can so fix our momentary experience as to make it the\nbasis of a philosophy, it will be past, and therefore uncertain. When\nDescartes said \u201cI think\u201d, he may have had certainty; but by the time he\nsaid \u201ctherefore I am\u201d, he was relying upon memory, and may have been\ndeceived. This line of argument leads to complete scepticism about\neverything. If we are to avoid such a result, we must have some new\nprinciple.\n\nIn actual fact, we start by feeling certainty about all sorts of\nthings, and we surrender this feeling only where some definite argument\nhas convinced us that it is liable to lead to error. When we find any\nclass of primitive certainties which never leads to error, we retain\nour convictions in regard to this class. That is to say, wherever we\nfeel initial certainty, we require an argument to make us doubt, not an\nargument to make us believe. We may therefore take, as the basis of our\nbeliefs, any class of primitive certainties which cannot be shown to\nlead us into error. This is really what Descartes does, although he is\nnot clear about it himself.\n\nMoreover, when we have found an error in something of which we were\npreviously certain, we do not as a rule abandon entirely the belief\nwhich misled us, but we seek, if we can, to modify it so that it\nshall no longer be demonstrably false. This is what has happened\nwith perception. When we think we see an external object, we may be\ndeceived by a variety of causes. There may be a mirage or a reflection;\nin this case, the source of the error is in the external world, and a\nphotographic plate would be equally deceived. There may be a stimulus\nto the eye of the sort that makes us \u201csee stars\u201d, or we may see little\nblack dots owing to a disordered liver, in which case, the source of\nthe error is in the body but not in the brain. We may have dreams in\nwhich we seem to see all sorts of things; in this case, the source\nof the error is in the brain. Having gradually discovered these\npossibilities of error, people have become somewhat wary as to the\nobjective significance of their perceptions; but they have remained\nconvinced that they really have the perceptions they thought they had,\nalthough the common-sense interpretation of them is sometimes at fault.\nThus, while retaining the conviction that they are sure of something,\nthey have gradually changed their view as to what it is that they are\nsure of. Nothing is known that tends to show error in the view that we\nreally have the percepts we think we have, so long as we are prudent\nin interpreting them as signs of something external. That is the valid\nbasis for Descartes\u2019 view that \u201cthought\u201d is more certain than external\nobjects. When \u201cthought\u201d is taken to mean the experiences which we\nusually regard as percepts of objects, there are sound reasons for\naccepting Descartes\u2019 opinion to this extent.\n\nLet us now take up Dr. Watson\u2019s view. We shall find, if I am not\nmistaken, that his position also is to a very large extent valid. That\nbeing so, we shall seek to find an intermediate opinion, accepting what\nseems valid and rejecting what seems doubtful in the contentions of\nboth protagonists.\n\nDr. Watson\u2019s view as to what is most certain is one which is in entire\naccordance with common sense. All psychological matters the plain man\nregards as more or less open to question, but he has no doubts about\nhis office, his morning train, the tax-collector, the weather, and\nthe other blessings of this life. It may amuse him, in an idle hour,\nto listen to some one playing with the idea that life is a dream, or\nsuggesting that the thoughts of the people in the train are more real\nthan the train. But unless he is a philosophical lecturer, he does not\ncountenance such notions in business hours. Who can imagine a clerk in\nan office conceiving metaphysical doubts as to the existence of his\nboss? Or would any railroad president regard with favour the theory\nthat his railroad is only an idea in the minds of the shareholders?\nSuch a view, he would say, though it is often sound as regards\ngold-mines, is simply silly when it comes to a railroad: anybody can\nsee it, and can get himself run over if he wanders on the tracks under\nthe impression that they do not exist. Belief in the unreality of\nmatter is likely to lead to an untimely death, and that, perhaps, is\nthe reason why this belief is so rare, since those who entertained it\ndied out. We cannot dismiss the common-sense outlook as simply silly,\nsince it succeeds in daily life; if we are going to reject it in part,\nwe must be sure that we do so in favor of something equally tough as a\nmeans of coping with practical problems.\n\nDescartes says: I think, therefore I am. Watson says: There are rats in\nmazes, therefore I don\u2019t think. At least, a parodist might thus sum up\nhis philosophy. What Watson really says is more like this: (1) The most\ncertain facts are those which are public, and can be confirmed by the\ntestimony of a number of observers. Such facts form the basis of the\nphysical sciences: physics, chemistry, biology, anatomy, physiology,\nto mention only those that are relevant to the matter in question. (2)\nThe physical sciences are capable of affording an explanation of all\nthe publicly observable facts about human behaviour. (3) There is no\nreason to suppose that there are any facts about human beings that can\nbe known only in some other way. (4) In particular, \u201cintrospection\u201d, as\na means of discovering by self-observation things that are in principle\nundiscoverable by observation of others, is a pernicious superstition,\nwhich must be swept away before any really sound knowledge of man\nbecomes possible. (5) And, as a corollary, there is no reason to\nbelieve in the existence of \u201cthought\u201d as opposed to speech and other\nbodily behaviour.\n\nI have numbered the above propositions, as it is important to keep them\nseparate. On the whole, (1), (2), and (3) seem to me to be true, but\n(4) and (5) seem to me to be false. Behaviourists, I think, incline to\nthe view that (4) and (5) follow from (1), (2), and (3); but this view\nI attribute to what I should regard as errors concerning the basis of\nphysics. That is why it was necessary to discuss physics before coming\nto a decision on this question of self-observation. But let us examine\neach of the above propositions in turn.\n\n(1) It is true that the facts upon which the physical sciences are\nbased are all of them public, in the sense that many men can observe\nthem. If a phenomenon is photographed, any number of people can inspect\nthe photograph. If a measurement is made, not only may several people\nbe present, but others can repeat the experiment. If the result does\nnot confirm the first observer, the supposed fact is rejected. The\npublicity of physical facts is always regarded as one of the greatest\nassets of physics. On a common-sense basis, therefore, the first of the\npropositions in which we have summed up the behaviourist philosophy\nmust be admitted.\n\nThere are, however, some very important provisos which must be\nmentioned. In the first place, a scientific observer is not expected\nto note his integral reaction to a situation, but only that part of it\nwhich experience leads him to regard as \u201cobjective\u201d, _i.e._ the same as\nthe reaction of any other competent observer. This process of learning\nto note only \u201cobjective\u201d features in our reaction is, as we have seen,\nbegun in infancy; training in science only carries it further. A \u201cgood\u201d\nobserver does not mention what is peculiar to himself in his reaction.\nHe does not say: \u201cA boring speck of light danced about, causing me\neye-fatigue and irritation; finally it settled at such-and-such a\npoint.\u201d He says simply: \u201cThe reading was such-and-such\u201d. All this\nobjectivity is a result of training and experience. One may say, in\nfact, that very few men have the \u201cright\u201d reaction to a scientific\nsituation. Therefore an immense amount of theory is mixed up with what\npasses in science as pure observation. The nature and justification of\nthis theory is a matter requiring investigation.\n\nIn the second place, we must not misinterpret the nature of the\npublicity in the case of physical phenomena. The publicity consists in\nthe fact that a number of people make closely similar reactions at a\ngiven moment. Suppose, for example, that twelve men are told to watch a\nscreen for the appearance of a bright light, and to say \u201cnow\u201d when it\nappears. Suppose the experimenter hears them all just when he himself\nsees the light; then he has good reason to believe that they have each\nhad a stimulus similar to his. But physics compels us to hold that they\nhave had twelve separate stimuli, so that when we say they have all\nseen the same light we can only legitimately mean that their twelve\nstimuli had a common causal origin. In attributing our perceptions to a\nnormal causal origin outside ourselves, we run a certain risk of error,\nsince the origin may be unusual: there may be reflection or refraction\non the way to the eye, there may be an unusual condition of the eye\nor optic nerve or brain. All these considerations give a certain very\nsmall probability that, on a given occasion, there is not such an\noutside cause as we suppose. If, however, a number of people concur\nwith us, _i.e._ simultaneously have reactions which they attribute to\nan outside cause that can be identified with the one we had inferred,\nthen the probability of error is enormously diminished. This is exactly\nthe usual case of concurrent testimony. If twelve men, each of whom\nlies every other time that he speaks, independently testify that some\nevent has occurred, the odds in favour of their all speaking the truth\nare 4095 to 1. The same sort of argument shows that our public senses,\nwhen confirmed by others, are probably speaking the truth, except where\nthere are sources of collective illusion such as mirage or suggestion.\n\nIn this respect, however, there is no _essential_ difference between\nmatters of external observation and matters of self-observation.\nSuppose, for example, that, for the first time in your life, you smell\nassaf\u0153tida. You say to yourself \u201cthat is a most unpleasant smell\u201d. Now\nunpleasantness is a matter of self-observation. It may be _correlated_\nwith physiological conditions which can be observed in others, but it\nis certainly not identical with these, since people knew that things\nwere pleasant and unpleasant before they knew about the physiological\nconditions accompanying pleasure and its opposite. Therefore when\nyou say \u201cthat smell is unpleasant\u201d you are noticing something that\ndoes not come into the world of physics as ordinarily understood. You\nare, however, a reader of psycho-analysis, and you have learned that\nsometimes hate is concealed love and love is concealed hate. You say to\nyourself, therefore: \u201cPerhaps I really like the smell of assaf\u0153tida,\nbut am ashamed of liking it\u201d. You therefore make your friends smell it,\nwith the result that you soon have no friends. You then try children,\nand finally chimpanzees. Friends and children give verbal expression\nto their disgust: chimpanzees are expressive, though not verbal. All\nthese facts lead you to state: \u201cThe smell of assaf\u0153tida is unpleasant\u201d.\nAlthough self-observation is involved, the result has the same kind of\ncertainty, and the same kind of objective verification, as if it were\none of the facts that form the empirical basis of physics.\n\n(2) The second proposition, to the effect that the physical sciences\nare capable of affording an explanation of all the publicly observable\nfacts about human behaviour, is one as to which it is possible to argue\nendlessly. The plain fact is that we do not yet know whether it is true\nor false. There is much to be said in its favour on general scientific\ngrounds, particularly if it is put forward, not as a dogma, but as a\nmethodological precept, a recommendation to scientific investigators\nas to the direction in which they are to seek for solution of their\nproblems. But so long as much of human behaviour remains unexplained\nin terms of physical laws, we cannot assert dogmatically that there is\nno residue which is theoretically inexplicable by this method. We may\nsay that the trend of science, so far, seems to render such a view\nimprobable, but to say even so much is perhaps rash, though, for my\npart, I should regard it as still more rash to say that there certainly\nis such a residue. I propose, therefore, as a matter of argument, to\nadmit the behaviourist position on this point, since my objections to\nbehaviourism as an ultimate philosophy come from quite a different kind\nof considerations.\n\n(3) The proposition we are now to examine may be stated as follows:\n\u201cAll facts that can be known about human beings are known by the\nsame method by which the facts of physics are known.\u201d This I hold to\nbe true, but for a reason exactly opposite to that which influences\nthe behaviourist. I hold that the facts of physics, like those of\npsychology, are obtained by what is really self-observation, although\ncommon sense mistakenly supposes that it is observation of external\nobjects. As we saw in Chapter XIII, your visual, auditory, and other\npercepts are all in your head, from the standpoint of physics.\nTherefore, when you \u201csee the sun\u201d, it is, strictly speaking, an event\nin yourself that you are knowing: the inference to an external cause\nis more or less precarious, and is on occasion mistaken. To revert to\nthe assaf\u0153tida: it is by a number of self-observations that you know\nthat the smell of assaf\u0153tida is unpleasant, and it is by a number of\nself-observations that you know that the sun is bright and warm. There\nis no essential difference between the two cases. One may say that the\ndata of psychology are those private facts which are not very directly\nlinked with facts outside the body, while the data of physics are those\nprivate facts which have a very direct causal connection with facts\noutside the body. Thus physics and psychology have the same method;\nbut this is rather what is commonly taken to be the special method of\npsychology than what is regarded as the method of physics. We differ\nfrom the behaviourist in assimilating physical to psychological method,\nrather than the opposite.\n\n(4) Is there a source of knowledge such as is believed in by those\nwho appeal to \u201cintrospection\u201d? According to what we have just been\nsaying, all knowledge rests upon something which might, in a sense, be\ncalled \u201cintrospection\u201d. Nevertheless, there may be some distinction to\nbe discovered. I think myself that the only distinction of importance\nis in the degree of correlation with events outside the body of the\nobserver. Suppose, for example, that a behaviourist is watching a rat\nin a maze, and that a friend is standing by. He says to the friend\n\u201cDo you see that rat?\u201d If the friend says yes, the behaviourist is\nengaged in his normal occupation of observing physical occurrences.\nBut if the friend says no, the behaviourist exclaims, \u201cI must give up\nthis boot-legged whiskey\u201d. In that case, if his horror still permits\nhim to think clearly, he will be obliged to say that in watching the\nimaginary rat he was engaged in introspection. There was certainly\nsomething happening, and he could still obtain knowledge by observing\nwhat was happening, provided he abstained from supposing that it had\na cause outside his body. But he cannot, without outside testimony\nor some other extraneous information, distinguish between the \u201creal\u201d\nrat and the \u201cimaginary\u201d one. Thus in the case of the \u201creal\u201d rat also,\nhis primary datum ought to be considered introspective, in spite of\nthe fact that it does not seem so; for the datum in the case of the\n\u201cimaginary\u201d rat also does not _seem_ to be merely introspective.\n\nThe real point seems to be this: some events have effects which radiate\nall round them, and can therefore produce reactions in a number of\nobservers; of these, ordinary speech is an illustration. But other\nevents produce effects which travel linearly, not spherically; of\nthese, speech into a telephone from a sound-proof telephone box may\nserve as an illustration. This can be heard by only one person beside\nthe speaker; if instead of a speaker we had an instrument at the\nmouthpiece, only one person could hear the sound, namely the person\nat the other end of the telephone. Events which happen inside a human\nbody are like the noise in the telephone: they have effects, in the\nmain, which travel along nerves to the brain, instead of spreading\nout in all directions equally. Consequently, a man can know a great\ndeal about his own body which another man can only know indirectly.\nAnother man can see the hole in my tooth, but he cannot feel my\ntoothache. If he infers that I feel toothache, he still does not have\nthe very same knowledge that I have; he may use the same words, but\nthe stimulus to his use of them is different from the stimulus to\nmine, and I can be acutely aware of the pain which is the stimulus to\nmy words. In all these ways a man has knowledge concerning his own\nbody which is obtained differently from the way in which he obtains\nknowledge of other bodies. This peculiar knowledge is, in one sense,\n\u201cintrospective\u201d, though not quite in the sense that Dr. Watson denies.\n\n(5) We come now to the real crux of the whole matter, namely to the\nquestion: Do we think? This question is very ambiguous, so long as\n\u201cthinking\u201d has not been clearly defined. Perhaps we may state the\nmatter thus: Do we know events in us which would not be included in\nan absolutely complete knowledge of physics? I mean by a complete\nknowledge of physics a knowledge not only of physical laws, but also\nof what we may call geography, _i.e._ the distribution of energy\nthroughout space-time. If the question is put in this way, I think it\nis quite clear that we do know things not included in physics. A blind\nman could know the whole of physics, but he could not know what things\nlook like to people who can see, nor what is the difference between\nred and blue as seen. He could know all about wave-lengths, but people\nknew the difference between red and blue as seen before they knew\nanything about wave-lengths. The person who knows physics and can see\nknows that a certain wave-length will give him a sensation of red, but\nthis knowledge is not part of physics. Again, we know what we mean by\n\u201cpleasant\u201d and \u201cunpleasant\u201d, and we do not know this any better when\nwe have discovered that pleasant things have one kind of physiological\neffect and unpleasant things have another. If we did not already know\nwhat things are pleasant and what unpleasant, we could never have\ndiscovered this correlation. But the knowledge that certain things are\npleasant and certain others unpleasant is no part of physics.\n\nFinally, we come to imaginations, hallucinations, and dreams. In all\nthese cases, we may suppose that there is an external stimulus, but\nthe cerebral part of the causal chain is unusual, so that there is not\nin the outside world something connected with what we are imagining\nin the same way as in normal perception. Yet in such cases we can\nquite clearly know what is happening to us; we can, for example,\noften remember our dreams. I think dreams must count as \u201cthought\u201d, in\nthe sense that they lie outside physics. They may be accompanied by\nmovements, but knowledge of them is not knowledge of these movements.\nIndeed all knowledge as to movements of matter is inferential, and\nthe knowledge which a scientific man should take as constituting\nhis primary data is more like our knowledge of dreams than like our\nknowledge of the movements of rats or heavenly bodies. To this extent,\nI should say, Descartes is in the right as against Watson. Watson\u2019s\nposition seems to rest upon naive realism as regards the physical\nworld, but naive realism is destroyed by what physics itself has to say\nconcerning physical causation and the antecedents of our perceptions.\nOn these grounds, I hold that self-observation can and does give us\nknowledge which is not part of physics, and that there is no reason to\ndeny the reality of \u201cthought\u201d.\n\n\n\n\nCHAPTER XVII\n\nIMAGES\n\n\nIn this chapter we shall consider the question of images. As the reader\ndoubtless knows, one of the battle-cries of behaviourism is \u201cdeath\nto images\u201d. We cannot discuss this question without a good deal of\nprevious clearing of the ground.\n\nWhat are \u201cimages\u201d as conceived by their supporters? Let us take this\nquestion first in the sense of trying to know some of the phenomena\nintended, and only afterwards in the sense of seeking a formal\ndefinition.\n\nIn the ordinary sense, we have visual images if we shut our eyes and\ncall up pictures of scenery or faces we have known; we have auditory\nimages when we recall a tune without actually humming it; we have\ntactual images when we look at a nice piece of fur and think how\npleasant it would be to stroke it. We may ignore other kinds of images,\nand concentrate upon these, visual, auditory, and tactual. There is no\ndoubt that we have such experiences as I have suggested by the above\nwords; the only question is as to how these experiences ought to be\ndescribed. Then we have another set of experiences, namely dreams,\nwhich feel like sensations at the moment, but do not have the same kind\nof relation to the external world as sensations have. Dreams, also,\nindubitably occur, and again it is a question of analysis whether we\nare to say that they contain \u201cimages\u201d or not.\n\nThe behaviourist does not admit images, but he equally does not\nadmit sensations and perceptions. Although he does not say so quite\ndefinitely, he may be taken to maintain that there is nothing but\nmatter in motion. We cannot, therefore, tackle the question of images\nby contrasting them with sensations or perceptions, unless we have\nfirst clearly proved the existence of these latter and defined their\ncharacteristics. Now it will be remembered that in Chapter V we\nattempted a behaviourist definition of perception, and decided that its\nmost essential feature was \u201csensitivity\u201d. That is to say, if a person\nalways has a reaction of a certain kind B when he has a certain spatial\nrelation to an object of a certain kind A, but not otherwise, then we\nsay that the person is \u201csensitive\u201d to A. In order to obtain from this a\ndefinition of \u201cperception\u201d, it is necessary to take account of the law\nof association; but for the moment we will ignore this complication,\nand say that a person \u201cperceives\u201d any feature of his environment, or\nof his own body, to which he is sensitive. Now, however, as a result\nof the discussion in Chapter XVI, we can include in his reaction, not\nonly what others can observe, but also what he alone can observe.\nThis enlarges the known sphere of perception, practically if not\ntheoretically. But it leaves unchanged the fact that the essence\nof perception is a causal relation to a feature of the environment\nwhich, except in astronomy, is approximately contemporaneous with the\nperception, though always at least slightly earlier, owing to the\ntime taken by light and sound to travel and the interval occupied in\ntransmitting a current along the nerves.\n\nLet us now contrast with this what happens when you sit still with\nyour eyes shut, calling up pictures of places you have seen abroad,\nand perhaps ultimately falling asleep. Dr. Watson, if I understand\nhim aright, maintains that either there is actual stimulation of the\nretina, or your pictures are mere word-pictures, the words being\nrepresented by small actual movements such as would, if magnified and\nprolonged, lead to actual pronunciation of the words. Now if you are\nin the dark with your eyes shut, there is no stimulation of the retina\nfrom without. It may be that, by association, the eye can be affected\nthrough stimuli to other senses; we have already had an example in the\nfact that the pupil can be taught to contract at a loud noise if this\nhad been frequently experienced along with a bright light. We cannot,\ntherefore, dismiss the idea that a stimulus to one sense may, as a\nresult of past events, have an effect upon the organs of another sense.\n\u201cImages\u201d might be definable as effects produced in this way. It may be\nthat, when you see a picture of Napoleon, there is an effect upon your\naural nerves analogous to that of having the word \u201cNapoleon\u201d pronounced\nin your presence, and that that is why, when you see the picture, the\nword \u201cNapoleon\u201d comes into your head. And similarly, when you shut\nyour eyes and call up pictures of foreign scenes, you may actually\npronounce, completely or incipiently, the word \u201cItaly\u201d, and this may,\nthrough association, stimulate the optic nerve in a way more or less\nsimilar to that in which some actual place in Italy stimulated it on\nsome former occasion. Thence association alone may carry you along\nthrough a series of journeys, until at last, when you fall asleep, you\nthink you are actually making them at the moment. All this is quite\npossible, but so far as I know there is no reason to hold that it is\nmore than possible, apart from an _a priori_ theory excluding every\nother explanation.\n\nWhat I think is clearly untenable is the view, sometimes urged by Dr.\nWatson, that when we are, as we think, seeing imaginary pictures with\nthe eyes shut, we are really only using such words as would describe\nthem. It seems to me as certain as anything can be that, when I\nvisualise, something is happening which is connected with the sense of\nsight. For example, I can call up quite clear mental pictures of the\nhouse in which I lived as a child; if I am asked a question as to the\nfurniture of any of the rooms in that house, I can answer it by first\ncalling up an image and then looking to see what the answer is, just\nas I should look to see in an actual room. It is quite clear to me\nthat the picture comes first and that words after; moreover, the words\nneed not come at all. I cannot tell what is happening in my retina or\noptic nerve at these moments of visualisation, but I am quite sure\nthat something is happening which has a connection with the sense of\nsight that it does not have with other senses. And I can say the same\nof aural and tactual images. If this belief were inconsistent with\nanything else that seems to me equally certain, I might be induced to\nabandon it. But so far as I can see, there is no such inconsistency.\n\nIt will be remembered that we decided in favour of perceptions as\nevents distinct from those which they perceive, and only causally\nconnected with them. There is, therefore, no reason why association\nshould not work in this region as well as in the region of muscles\nand glands; in other words, there is no reason to deny what used to\nbe called \u201cassociation of ideas\u201d, in spite of the fact that bodily\nchanges can also be associated. If a physical basis is wanted, it can\nbe assumed to exist in the brain. The state of the brain which causes\nus to hear the word \u201cNapoleon\u201d may become associated with the state of\nbrain which causes us to see a picture of Napoleon, and thus the word\nand the picture will call each other up. The association _may_ be in\nthe sense-organs or nerves, but may equally well be in the brain. So\nfar as I know, there is no conclusive evidence either way, nor even\nthat the association is not purely \u201cmental.\u201d\n\nWhen we try to find a definition of the difference between a sensation\nand an image, it is natural to look first for intrinsic differences.\nBut intrinsic differences between ordinary sensations and ordinary\nimages, for example as to \u201cliveliness\u201d, are found to be subject to\nexceptions, and therefore unsuitable for purposes of definition. Thus\nwe are brought to differences as to causes and effects.\n\nIt is obvious that, in an ordinary case, you perceive a table because\n(in some sense) the table is there. That is to say, there is a causal\nchain leading backwards from your perception to something outside your\nbody. This alone, however, is hardly sufficient as a criterion. Suppose\nyou smell peat smoke and think of Ireland, your thought can equally be\ntraced to a cause outside your body. The only real difference is that\nthe outside cause (peat smoke) would not have had the effect (images\nof Ireland) upon every normal person, but only upon such as had smelt\npeat smoke in Ireland, and not all of them. That is to say, the normal\ncerebral apparatus does not cause the given stimulus to produce the\ngiven effect except where certain previous experiences have occurred.\nThis is a very vital distinction. Part of what occurs in us under the\ninfluence of a stimulus from without depends upon past experience; part\ndoes not. The former part includes images, the latter consists of pure\nsensations. This, however, as we shall see later, is inadequate as a\ndefinition.\n\nMental occurrences which depend upon past experience are called\n\u201cmnemic\u201d occurrences, following Semon. Images are thus to be included\namong mnemic occurrences, at least so far as human experience goes.\nThis, however, does not suffice to define them, since there are others,\n_e.g._ recollections. What further defines them is their similarity\nto sensations. This only applies strictly to simple images; complex\nones may occur without a prototype, though all their parts will have\nprototypes among sensations. Such, at least, is Hume\u2019s principle, and\non the whole it seems to be true. It must not, however, be pressed\nbeyond a point. As a rule, an image is more or less vague, and has a\nnumber of similar sensations as its prototypes. This does not prevent\nthe connection with sensation in general, but makes it a connection\nwith a number of sensations, not with one only.\n\nIt happens that, when a complex of sensations has occurred at some\ntime in a person\u2019s experience, the recurrence of part of the whole\ntends to produce images of the remaining parts or some of them. This is\nassociation, and has much to do with memory.\n\nIt is common to speak of images as \u201ccentrally excited\u201d, as opposed to\nsensations, which are excited by a stimulus to some sense organ. In\nessence this is quite correct, but there is need of some caution in\ninterpreting the phrase. Sensations also have _proximate_ causes in\nthe brain; images also may be due to some excitement of a sense-organ,\nwhen they are roused by a sensation through association. But in such\ncases there is nothing to explain their occurrence except the past\nexperience and its effect on the brain. They will not be aroused by\nthe same stimulus in a person with similar sense-organs but different\npast experience. The connection with past experience is clearly known;\nit is, however, an explanatory hypothesis, not directly verifiable in\nthe present state of knowledge, to suppose that this connection works\nthrough an effect of the past experience on the brain. This hypothesis\nmust be regarded as doubtful, but it will save circumlocution to adopt\nit. I shall therefore not repeat, on each occasion, that we cannot feel\nsure it is true. In general, where the causal connection with past\nexperience is obvious, we call an occurrence \u201cmnemic\u201d, without implying\nthis or that hypothesis as to the explanation of mnemic phenomena.\n\nIt is perhaps worth while to ask how we know that images are like the\nsensations which are their prototypes. The difficulty of this question\narises as follows. Suppose you call up an image of the Brooklyn\nBridge, and you are convinced that it is like what you see when you\nlook at Brooklyn Bridge. It would seem natural to say that you know\nthe likeness because you remember Brooklyn Bridge. But remembering is\noften held to involve, as an essential element, the occurrence of an\nimage which is regarded as referring to a prototype. Unless you can\nremember without images, it is difficult to see how you can be sure\nthat images resemble prototypes. I think that in fact you cannot be\nsure, unless you can find some indirect means of comparison. You might,\nfor example, have photographs of Brooklyn Bridge taken from a given\nplace on two different days, and find them indistinguishable, showing\nthat Brooklyn Bridge has not changed in the interval. You might see\nBrooklyn Bridge on the first of these days, remember it on the second,\nand immediately afterwards look at it. In looking at it, you might find\nevery detail coming to you with a feeling of expectedness, or you might\nfind some details coming with a feeling of surprise. In this case you\nwould say that your image had been wrong as regards the details which\nwere surprising. Or, again, you might make a picture of Brooklyn Bridge\non paper, from memory, and then compare it with the original or a\nphotograph. Or you might content yourself by writing down a description\nof it in words, and verifying its accuracy by direct observation.\nInnumerable methods of this kind can be devised by which you can test\nthe likeness of an image to its prototype. The result is that there is\noften a great likeness, though seldom complete accuracy. The belief in\nthe likeness of an image to its prototype is, of course, not generated\nin this way, but only tested. The belief exists prior to evidence as to\nits correctness, like most of our beliefs. I shall have more to say on\nthis subject in the next chapter, which will be concerned with memory.\nBut I think enough has been said to show that it is not unreasonable\nto regard images as having a greater or less degree of resemblance to\ntheir prototypes. To claim more is hardly justifiable.\n\nWe can now reach a definite conclusion about perception, sensation, and\nimages. Let us imagine a number of people placed, as far as possible,\nin the same environment; we will suppose that they sit successively in\na certain chair in a dark room, in full view of illuminated pictures\nof two eminent politicians of opposite parties whose names are written\nunderneath them. We will suppose that all of them have normal eyesight.\nTheir reactions will be partly similar, partly different. If any of\nthese observers are babies too young to have learnt to focus, they will\nnot see sharp outlines, but a mere blurr, not from an optical defect,\nbut from a lack of cerebral control over muscles. In this respect,\nexperience has an effect even upon what must count as pure sensation.\nBut this difference is really analogous to the difference between\nhaving one\u2019s eyes open and having them shut; the difference is in the\nsense-organ, although it may be due to a difference in the brain. We\nwill therefore assume that all the spectators know how to adjust the\neyes so as to see as well as possible, and all try to see. We shall\nthen say that, if the spectators differ as widely as is possible for\nnormal human beings, what is common to the reactions of all of them\nis sensation, provided it is connected with the sense of sight, or,\nmore correctly, provided it has that quality which we observe to be\ncommon and peculiar to visual objects. But probably all of them, if\nthey are over three months old, will have tactile images while they\nsee the pictures. And if they are more than about a year old, they\nwill interpret them as pictures, which represent three-dimensional\nobjects; before that age, they may see them as coloured patterns,\nnot as representations of faces. Most animals, though not all, are\nincapable of interpreting pictures as representations. But in an adult\nhuman being this interpretation is not deliberate; it has become\nautomatic. It is, I think, mainly a question of tactile images: the\nimages you have in looking at a picture are not those appropriate to a\nsmooth flat surface, but those appropriate to the object represented.\nIf the object represented is a large one, there will also be images of\nmovement--walking round the object, or climbing up it, or what not.\nAll these are obviously a product of experience, and therefore do not\ncount as part of the sensation. This influence of experience is still\nmore obvious when it comes to reading the names of the politicians,\nconsidering whether they are good likenesses, and feeling what a fine\nfellow one of them looks and what unmitigated villainy is stamped upon\nthe features of the other. None of this counts as sensation, yet it is\npart of a man\u2019s spontaneous reaction to an outside stimulus.\n\nIt is evidently difficult to avoid a certain artificiality in\ndistinguishing between the effects of experience and the rest in a\nman\u2019s reaction to a stimulus. Perhaps we could tackle the matter in a\nslightly different way. We can distinguish stimuli of different sorts:\nto the eye, the ear, the nose, the palate, etc. We can also distinguish\nelements of different sorts in the reaction: visual elements, auditory\nelements, etc. The latter are defined, not by the stimulus, but by\ntheir intrinsic quality. A visual sensation and a visual image have\na common quality which neither shares with an auditory sensation or\nan auditory image. We may then say: a visual image is an occurrence\nhaving the visual quality but not due to a stimulus to the eye, _i.e._\nnot having as a direct causal antecedent the incidence of light-waves\nupon the retina. Similarly an auditory image will be an occurrence\nhaving the auditory quality but not due to sound-waves reaching the\near, and so on for the other senses. This means a complete abandonment\nof the attempt to distinguish psychologically between sensations and\nimages; the distinction becomes solely one as to physical antecedents.\nIt is true that we can and do arrive at the distinction without\nscientific physics, because we find that certain elements in our\nintegral reactions have the correlations that make us regard them as\ncorresponding to something external while others do not--correlations\nboth with the experience of others and with our own past and future\nexperiences. But when we refine upon this common-sense distinction and\ntry to make it precise, it becomes the distinction in terms of physics\nas stated just now.\n\nWe might therefore conclude that an image is an occurrence having the\nquality associated with stimulation by some sense-organ, but not due\nto such stimulation. In human beings, images seem to depend upon past\nexperience, but perhaps in more instinctive animals they are partly due\nto innate cerebral mechanisms. In any case dependence upon experience\nis not the mark by which they are to be defined. This shows how\nintimate is the dependence of traditional psychology upon physics, and\nhow difficult it is to make psychology into an autonomous science.\n\nThere is, however, still a further refinement necessary. Whatever is\nincluded under our present definition is an image, but some things\nnot included are also images. The sight of an object may bring with\nit a visual image of some other object frequently associated with it.\nThis latter is called an image, not a sensation, because, though also\nvisual, it is not appropriate to the stimulus in a certain sense: it\nwould not appear in a photograph of the scene, or in a photograph\nof the retina. Thus we are forced to say: the sensation element in\nthe reaction to a stimulus is that part which enables you to draw\ninferences as to the nature of the extra-cerebral event (if any)\nwhich was the stimulus;[9] the rest is images. Fortunately, images\nand sensations _usually_ differ in intrinsic quality; this makes it\npossible to get an approximate idea of the external world by using the\nusual intrinsic differences, and to correct it afterwards by means of\nthe strict causal definition. But evidently the matter is difficult\nand complicated, depending upon physics and physiology, not upon pure\npsychology. This is the main thing to be realised about images.\n\n      [9] _I.e._ the immediate stimulus, not the \u201cphysical object\u201d.\n\nThe above discussion has suggested a definition of the word \u201cimage\u201d.\nWe might have called an event an \u201cimage\u201d when it is recognisably of\nthe same kind as a \u201cpercept\u201d, but does not have the stimulus which it\nwould have if it were a percept. But if this definition is to be made\nsatisfactory, it will be necessary to substitute a different word in\nplace of \u201cpercept\u201d. For example, in the percept of a visible object\nit would be usual to include certain associated tactual elements, but\nthese must, from our point of view, count as images. It will be better\nto say, therefore, that an \u201cimage\u201d is an occurrence recognisably visual\n(or auditory, etc., as the case may be), but not caused by a stimulus\nwhich is of the nature of light (or sound etc., as the case may be), or\nat any rate only indirectly so caused as a result of association. With\nthis definition, I do not myself feel any doubt as to the existence of\nimages. It is clear that they constitute most of the material of dreams\nand day-dreams, that they are utilised by composers in making music,\nthat we employ them when we get out of a familiar room in the dark\n(though here the rats in mazes make a different explanation possible),\nand that they account for the shock of surprise we have when we take\nsalt thinking it is sugar or (as happened to me recently) vinegar\nthinking it is coffee. The question of the causation of images--_i.e._\nwhether it is in the brain or in other parts of the body--is not one\nwhich it is necessary to our purposes to decide, which is fortunate,\nsince, so far as I know, there is not at present any adequate evidence\non the point. But the existence of images and their resemblance to\nperception is important, as we shall see in the next chapter.\n\nImages come in various ways, and play various parts. There are\nthose that come as accretions to a case of sensation, which are not\nrecognised as images except by the psychologist; these form, for\nexample, the tactual quality of things we only see, and the visual\nquality of things we only touch. I think dreams belong, in part, to\nthis class of images: some dreams result from misinterpreting some\nordinary stimulus, and in these cases the images are those suggested\nby a sensation, but suggested more uncritically than if we were awake.\nThen there are images which are not attached to a present reality, but\nto one which we locate in the past; these are present in memory, not\nnecessarily always, but sometimes. Then there are images not attached\nto reality at all so far as our feeling about them goes: images which\nmerely float into our heads in reverie or in passionate desire.\nAnd finally there are images which are called up voluntarily, for\nexample, in considering how to decorate a room. This last kind has its\nimportance, but I shall say nothing more about it at present, since we\ncannot profitably discuss it until we have decided what we are to mean\nby the word \u201cvoluntary\u201d. The first kind, which comes as an accretion\nto sensation, and gives to our feeling of objects a certain rotundity\nand full-bloodedness which the stimulus alone would hardly warrant, has\nbeen considered already. Therefore what remains for the present is the\nuse of images in memory and imagination; and of these two I shall begin\nwith memory.\n\n\n\n\nCHAPTER XVIII\n\nIMAGINATION AND MEMORY\n\n\nIn this chapter we have to consider the two topics of imagination and\nmemory. The latter has already been considered in Chapter VI, but there\nwe viewed it from outside. We want now to ask ourselves whether there\nis anything further to be known about it by taking account of what is\nonly perceptible to the person remembering.\n\nAs regards the part played by images, I do not think this is essential.\nSometimes there are memory-images, sometimes not; sometimes when images\ncome in connection with memory, we may nevertheless know that the\nimages are incorrect, showing that we have also some other and more\nreliable source of memory. Memory _may_ depend upon images, as in the\ncase mentioned above, of the house where I lived as a child. But it may\nalso be purely verbal. I am a poor visualiser, except for things I saw\nbefore I was ten years old; when now I meet a man and wish to remember\nhis appearance, I find that the only way is to describe him in words\nwhile I am seeing him, and then remember the words. I say to myself:\n\u201cThis man has blue eyes and a brown beard and a small nose; he is\nshort, with a rounded back and sloping shoulders\u201d. I can remember these\nwords for months, and recognise the man by means of them, unless two\nmen having these characteristics are present at once. In this respect,\na visualiser would have the advantage of me. Nevertheless, if I had\nmade my verbal inventory sufficiently extensive and precise, it would\nhave been pretty sure to answer its purpose. I do not think there is\nanything in memory that absolutely demands images as opposed to words.\nWhether the words we use in \u201cthought\u201d are themselves sometimes images\nof words, or are always incipient movements (as Watson contends), is a\nfurther question, as to which I offer no opinion, since it ought to be\ncapable of being decided experimentally.\n\nThe most important point about memory is one which has nothing to do\nwith images, and is not mentioned in Watson\u2019s brief discussion. I mean\nthe reference to the past. This reference to the past is not involved\nin mere habit memory, _e.g._ in skating or in repeating a poem formerly\nlearned. But it is involved in recollection of a past incident. We do\nnot, in this case, merely repeat what we did before: then, we felt the\nincident as present, but now we feel it as past. This is shown in the\nuse of the past tense. We say to ourselves at the time \u201cI _am having_ a\ngood dinner\u201d, but next day we say \u201cI _did have_ a good dinner\u201d. Thus we\ndo not, like a rat in a maze, repeat our previous performance: we alter\nthe verbal formula. Why do we do so? What constitutes this reference of\na recollection to the past?[10]\n\n     [10] On this subject, cf. Broad, _The Mind and Its Place in\n          Nature_, p. 264 ff., in his chapter on \u201cMemory\u201d.\n\nLet us take up the question first from the point of view of\nsensitivity. The stimulus to a recollection is, no doubt, always\nsomething in the present, but our reaction (or part of it) is more\nintimately related to a certain past event than to the present\nstimulus. This, in itself, can be paralleled in inanimate objects,\nfor example, in a gramophone record. It is not the _likeness_ of our\nreaction to that called forth on a former occasion that concerns us\nat the moment; it is its _un-likeness_, in the fact that now we have\nthe feeling of pastness, which we did not have originally. You cannot\nsing into a dictaphone \u201cI love you\u201d, and have it say five days hence\n\u201cI loved you last Wednesday\u201d; yet that is what we do when we remember.\nI think, however, that this feature of memory is probably connected\nwith a feature of reactions due to association when the association\nis cerebral: I think also that this is connected with the difference\nin quality that exists usually, though not always, between images and\nsensations. It would seem that, in such cases, the reaction aroused\nthrough association is usually different from that which would have\nbeen aroused directly, in certain definite ways. It is fainter, and\nhas, when attended to, the sort of quality that makes us call it\n\u201cimaginary\u201d. In a certain class of cases, we come to know that we can\nmake it \u201creal\u201d if we choose; this applies, _e.g._ to the tactual images\nproduced by visible objects that we can touch. In such cases, the image\nis attached by us to the object, and its \u201cimaginary\u201d character fails to\nbe noticed. These are the cases in which the association is not due to\nsome accident of our experience, but to a collocation which exists in\nnature. In other cases, however, we are perfectly aware, if we reflect,\nthat the association depends upon some circumstance in our private\nlives. We may, for instance, have had a very interesting conversation\nat a certain spot, and always think of this conversation when we find\nourselves in this place. But we know that the conversation does not\nactually take place again when we go back to where it happened. In\nsuch a case, we notice the intrinsic difference between the event as\na sensible fact in the present and the event as merely revived by\nassociation. I think this difference has to do with our feeling of\npastness. The difference which we can directly observe is not, of\ncourse, between our present recollection and the past conversation,\nbut between our present recollection and present sensible facts. This\ndifference, combined with the inconsistency of our recollection with\npresent facts if our recollection were placed in the present, is\nperhaps a cause of our referring memories to the past. But I offer this\nsuggestion with hesitation; and, as we shall find when we have examined\nimagination, it cannot be the whole truth, though it may be part of it.\n\nThere are some facts that tend to support the above view. In dreams,\nwhen our critical faculty is in abeyance, we may live past events\nover again under the impression that they are actually happening; the\nreference of recollections to the past must, therefore, be a matter\ninvolving a somewhat advanced type of mental activity. Conversely,\nwe sometimes have the impression that what is happening now really\nhappened in the past; this is a well-known and much discussed illusion.\nIt happens especially when we are profoundly absorbed in some inward\nstruggle or emotion, so that outer events only penetrate faintly.\nI suggest that, in these circumstances, the quality of sensations\napproximates to that of images, and that this is the source of the\nillusion.\n\nIf this suggestion is right, the feeling of pastness is really complex.\nSomething is suggested by association, but is recognisably different\nfrom a present sensible occurrence. We therefore do not suppose that\nthis something is happening now; and we may be confirmed in this by\nthe fact that it is inconsistent with something that is happening now.\nWe may then either refer the something to the past, in which case we\nhave a recollection, though not necessarily a correct one; or we may\nregard the something as purely imaginary, in which case we have what we\nregard as pure imagination. It remains to inquire why we do sometimes\nthe one and sometimes the other, which brings us to the discussion of\nimagination. I think we shall find that memory is more fundamental\nthan imagination, and that the latter consists merely of memories of\ndifferent dates assembled together. But to support this theory will\ndemand first an analysis of imagination and then, in the light of this\nanalysis, an attempt to give further precision to our theory of memory.\n\nImagination is not, as the word might suggest, essentially connected\nwith images. No doubt images are often, even usually, present when\nwe imagine, but they need not be. A man can improvise on the piano\nwithout first having images of the music he is going to make; a poet\nmight write down a poem without first making it up in his head.\nIn talking, words suggest other words, and a man with sufficient\nverbal associations may be successfully carried along by them for a\nconsiderable time. The art of talking without thinking is particularly\nnecessary to public speakers, who must go on when once they are on\ntheir feet, and gradually acquire the habit of behaving in private\nas they do before an audience. Yet the statements they make must\nbe admitted to be often imaginative. The essence of imagination,\ntherefore, does not lie in images.\n\nThe essence of imagination, I should say, is the absence of belief\ntogether with a novel combination of known elements. In memory, when\nit is correct, the combination of elements is not novel; and whether\ncorrect or not, there is belief. I say that in imagination there is \u201ca\nnovel combination of known elements\u201d, because, if nothing is novel,\nwe have a case of memory, while if the elements, or any of them,\nare novel, we have a case of perception. This last I say because I\naccept Hume\u2019s principle that there is no \u201cidea\u201d without an antecedent\n\u201cimpression\u201d. I do not mean that this is to be applied in a blind and\npedantic manner, where abstract ideas are concerned. I should not\nmaintain that no one can have an idea of liberty until he has seen the\nStatue of Liberty. The principle applies rather to the realm of images.\nI certainly do not think that, in an image, there can be any element\nwhich does not resemble some element in a previous perception, in the\ndistinctive manner of images.\n\nHume made himself an unnecessary difficulty in regard to the theory\nthat images \u201ccopy\u201d impressions. He asked the question: Suppose a\nman has seen all the different shades of colour that go to make up\nthe spectrum, except just one shade. To put the thing in modern\nlanguage, suppose he has never seen light of a certain small range of\nwave-lengths, but has seen light of all other wave-lengths. Will he\nbe able to form an image of the shade he has never seen? Hume thinks\nhe will, although this contradicts the principle. I should say that\nimages are always more or less vague copies of impressions, so that an\nimage might be regarded as a copy of any one of a number of different\nimpressions of slightly different shades. In order to get a test\ncase for Hume\u2019s question, we should have to suppose that there was a\nbroad band of the spectrum that the man had never seen--say the whole\nof the yellow. He would then, one may suppose, be able to form images\nwhich, owing to vagueness, might be applicable to orange-yellow, and\nothers applicable to green-yellow, but none applicable to a yellow\nmidway between orange and green. This is an example of an unreal puzzle\nmanufactured by forgetting vagueness. It is analogous to the following\nprofound problem: A man formerly hairy is now bald; he lost his hairs\none by one; therefore there must have been just one hair that made\nthe difference, so that while he had it he was not bald but when he\nlost it he was. Of course \u201cbaldness\u201d is a vague conception; and so\nis \u201ccopying\u201d, when we are speaking of the way in which images copy\nprototypes.\n\nWhat causes us, in imagination, to put elements together in a new way?\nLet us think first of concrete instances. You read that a ship has\ngone down on a route by which you have lately travelled; very little\nimagination is needed to generate the thought \u201cI might have gone\ndown\u201d. What happens here is obvious: the route is associated both with\nyourself and with shipwreck, and you merely eliminate the middle term.\nLiterary ability is largely an extension of the practice of which the\nabove is a very humble example. Take, say:\n\n    And all our yesterdays have lighted fools\n    The way to dusty death. Out, out, brief candle!\n    Life\u2019s but a walking shadow, a poor player\n    Who struts and frets his hour upon the stage,\n    And then is heard no more. It is a tale\n    Told by an idiot, full of sound and fury\n    Signifying nothing.\n\nI do not pretend to explain all the associations which led Shakespeare\nto think of these lines, but some few are obvious. \u201cDusty death\u201d is\nsuggested by Genesis iii. 19: \u201cDust thou art, and unto dust shalt thou\nreturn\u201d. Having spoken of \u201clighting fools the way\u201d, it is natural\nto think of a \u201ccandle\u201d, and thence of a \u201cwalking shadow\u201d being\nlighted by the candle along the way. From shadows to players was a\nwell-established association in Shakespeare\u2019s mind; thus in _Midsummer\nNight\u2019s Dream_ he says of players: \u201cThe best in this kind are but\nshadows, and the worst are no worse, if imagination amend them\u201d. From\na \u201cpoor player\u201d to a \u201ctale told by an idiot\u201d is no very difficult\ntransition for a theatre-manager; and \u201csound and fury\u201d no doubt often\nformed part of the tales to which he had to listen in spite of their\n\u201csignifying nothing\u201d. If we knew more about Shakespeare, we could\nexplain more of him in this sort of way.\n\nThus exceptional imaginative gifts appear to depend mainly upon\nassociations that are unusual and have an emotional value owing to\nthe fact that there is a certain uniform emotional tone about them.\nMany adjectives are suitable to death: in a mood quite different from\nMacbeth\u2019s, it may be called \u201cnoble, puissant and mighty\u201d. A Chancellor\nof the Exchequer, thinking of the Death Duties, might feel inclined to\nspeak of \u201clucrative death\u201d; nevertheless he would not, like Vaughan,\nspeak of \u201cdear, beauteous death\u201d. Shakespeare also would not have\nspoken of death in such terms, for his view of it was pagan; he speaks\nof \u201cthat churl death\u201d. So a man\u2019s verbal associations may afford a key\nto his emotional reactions, for often what connects two words in his\nmind is the fact that they rouse similar emotions.\n\nThe absence of belief that accompanies imagination is a somewhat\nsophisticated product; it fails in sleep and in strong and emotional\nexcitement. Children invent terrors for fun, and then begin to believe\nin them. The state of entertaining an idea without believing it is one\ninvolving some tension, which demands a certain level of intellectual\ndevelopment. It may be assumed that imagination, at first, always\ninvolved belief, as it still does in dreams. I am not concerned at the\nmoment to define \u201cbelief\u201d, but a criterion is influence on action. If\nI say \u201csuppose there were a tiger outside your front door\u201d, you will\nremain calm; but if I say, with such a manner as to command belief,\n\u201cthere is a tiger outside your front door\u201d, you will stay at home, even\nif it involves missing your train to the office. This illustrates what\nI mean when I say that imagination, in its developed form, involves\nabsence of belief. But this is not true of its primitive forms. And\neven a civilised adult, passing through a churchyard on a dark night,\nmay feel fear if his imagination turns in the direction of ghosts.\n\nWhen imagination passes into belief, it does not, as a rule, become a\nbelief about the past. Generally we place the imagined object in the\npresent, but not where it would be perceptible to our senses. If we\nplace it in the past, it is because the past has some great emotional\nsignificance for us. If a person we love has been in great danger,\nand we do not know whether he has come through safely, imagination\nof his death may lead us to believe that he has been killed. And\noften imagination leads us to believe that something is going to\nhappen. What is common to all such cases is the emotional interest:\nthis first causes us to imagine an event, and then leads us to think\nthat it has happened, is happening, or will happen, according to the\ncircumstances. Hope and fear have this effect equally; wish-fulfilment\nand dread-fulfilment are equally sources of dreams and day-dreams.\nA great many beliefs have a source of this kind. But, in spite of\npsycho-analysis, there are a great many that have a more rational\nfoundation. I believe that Columbus first crossed the ocean in 1492,\nthough 1491 or 1493 would have suited me just as well. I cannot\ndiscover that there is any emotional element in this belief, or in the\nbelief that Semipalatinsk is in Central Asia. The view that all our\nbeliefs are irrational is perhaps somewhat overdone nowadays, though it\nis far more nearly true than the views that it has displaced.\n\nWe must now return to the subject of memory. Memory proper does not,\nlike imagination, involve a re-arrangement of elements derived from\npast experience; on the contrary, it should restore such elements\nin the pattern in which they occurred. This is the vital difference\nbetween memory and imagination; belief, even belief involving reference\nto the past, may, as we have seen, be present in what is really\nimagination though it may not seem to be so to the person concerned.\nThat being so, we still have to consider what constitutes the reference\nto the past, since the view tentatively suggested before we had\nconsidered imagination turns out to be inadequate.\n\nThere is one possible view, suggested, though not definitely adopted,\nby Dr. Broad in his chapter on \u201cMemory\u201d already referred to. According\nto this view, we have to start from temporal succession as perceived\nwithin what is called the \u201cspecious present\u201d, _i.e._ a short period of\ntime such that the events that occur throughout it can be perceived\ntogether. (I shall return to this subject presently.) For example, you\ncan see a quick movement as a whole; you are not merely aware that\nthe object was first in one place and then in another. You can see\nthe movement of the second-hand of a watch, but not of the hour-hand\nor minute-hand. When you see a movement in this sense, you are aware\nthat one part of it is earlier than another. Thus you acquire the idea\n\u201cearlier\u201d, and you can mean by \u201cpast\u201d \u201cearlier than this\u201d, where \u201cthis\u201d\nis what is actually happening. This is a logically possible theory,\nbut it seems nevertheless somewhat difficult to believe. I do not\nknow, however, of any easier theory, and I shall therefore adopt it\nprovisionally while waiting for something better.\n\nFor the understanding of memory, it is a help to consider the links\nconnecting its most developed forms with other occurrences of a less\ncomplex kind. True recollection comes at the end of a series of stages.\nI shall distinguish five stages on the way, so that recollection\nbecomes the sixth in gradual progress. The stages are as follows:\n\n1. _Images._--As we have seen, images, at any rate in their simpler\nparts, in fact copy past sensations more or less vaguely, even when\nthey are not known to do so. Images are \u201cmnemic\u201d phenomena, in the\nsense that they are called up by stimuli formerly associated with their\nprototypes, so that their occurrence is a result of past experience\naccording to the law of association. But obviously an image which in\nfact copies a past occurrence does not constitute a recollection unless\nit is _felt_ to be a copy.\n\n2. _Familiarity._--Images and perceptions may come to us, and so may\nwords or other bodily movements, with more or less of the feeling we\ncall \u201cfamiliarity\u201d. When you recall a tune that you have heard before,\neither by images or by actually singing it, part of what comes to you\nmay feel familiar, part unfamiliar. This may lead you to judge that\nyou have remembered the familiar part rightly and the unfamiliar part\nwrongly, but this judgment belongs to a later stage.\n\n3. _Habit-Memory._--We have already discussed this in Chapter VI.\nPeople say they remember a poem if they can repeat it correctly.\nBut this does not necessarily involve any recollection of a past\noccurrence; you may have quite forgotten when and where you read the\npoem. This sort of memory is mere habit, and is essentially like\nknowing how to walk although you cannot remember learning to walk. This\ndoes not deserve to be called memory in the strict sense.\n\n4. _Recognition._--This has two forms. (_a_) When you see a dog, you\ncan say to yourself \u201cthere is a dog\u201d, without recalling any case in\nwhich you have seen a dog before, and even without reflecting that\nthere have been such cases. This involves no knowledge about the past;\nessentially it is only an associative habit. (_b_) You may know \u201cI\nsaw this before\u201d, though you do not know when or where, and cannot\nrecollect the previous occurrence in any way. In such a case there is\nknowledge about the past, but it is very slight. When you judge: \u201cI saw\n_this_ before\u201d, the word \u201cthis\u201d must be used vaguely, because you did\nnot see exactly what you see now, but only something very like this.\nThus all that you are really knowing is that, on some past occasion,\nyou saw something very like what you are seeing now. This is about the\nminimum of knowledge about the past that actually occurs.\n\n5. _Immediate Memory._--I come now to a region intermediate between\nsensation and true memory, the region of what is sometimes called\n\u201cimmediate memory\u201d. When a sense-organ is stimulated, it does not,\non the cessation of the stimulus, return at once to its unstimulated\ncondition: it goes on (so to speak) vibrating, like a piano-string,\nfor a short time. For example, when you see a flash of lightning, your\nsensation, brief as it is, lasts much longer than the lightning as a\nphysical occurrence. There is a period during which a sensation is\nfading: it is then called an \u201cacoleuthic\u201d sensation. It is owing to\nthis fact that you can see a movement as a whole. As observed before,\nyou cannot see the minute-hand of a watch moving, but you can see\nthe second-hand moving. That is because it is in several appreciably\ndifferent places within the short time that is required for one visual\nsensation to fade, so that you do actually, at one moment, see it in\nseveral places. The fading sensations, however, feel different from\nthose that are fresh, and thus the various positions which are all\nsensibly present are placed in a series by the degree of fading, and\nyou acquire the perception of movement as a process. Exactly the same\nconsiderations apply to hearing a spoken sentence.\n\nThus not only an instant, but a short finite time is sensibly present\nto you at any moment. This short finite time is called the \u201cspecious\npresent\u201d. By the felt degree of fading, you can distinguish earlier and\nlater in the specious present, and thus experience temporal succession\nwithout the need of true memory. If you see me quickly move my arm from\nleft to right, you have an experience which is quite different from\nwhat you would have if you now saw it at the right and remembered that\na little while ago you saw it at the left. The difference is that, in\nthe quick movement, the whole falls within the specious present, so\nthat the entire process is sensible. The knowledge of something as\nin the immediate past, though still sensible, is called \u201cimmediate\nmemory\u201d. It has great importance in connection with our apprehension of\ntemporal processes, but cannot count as a form of true memory.\n\n6. _True Recollection._--We will suppose, for the sake of definiteness,\nthat I am remembering what I had for breakfast this morning. There\nare two questions which we must ask about this occurrence: (_a_) What\nis happening now when I recollect? (_b_) What is the relation of the\npresent happening to the event remembered? As to what is happening now,\nmy recollection may involve either images or words; in the latter case,\nthe words themselves may be merely imagined. I will take the case in\nwhich there are images without words, which must be the more primitive,\nsince we cannot suppose that memory would be impossible without words.\n\nThe first point is one which seems so obvious that I should be ashamed\nto mention it, but for the fact that many distinguished philosophers\nthink otherwise. The point is this: whatever may be happening now, the\nevent remembered is _not_ happening. Memory is often spoken of as if\nit involved the actual persistence of the past which is remembered;\nBergson, _e.g._ speaks of the interpenetration of the present by the\npast. This is mere mythology; the event which occurs when I remember\nis quite different from the event remembered. People who are starving\ncan remember their last meal, but the recollection does not appease\ntheir hunger. There is no mystic survival of the past when we remember;\nmerely a new event having a certain relation to the old one. What this\nrelation is, we shall consider presently.\n\nIt is quite clear that images are not enough to constitute\nrecollection, even when they are accurate copies of a past occurrence.\nOne may, in a dream, live over again a past experience; while one is\ndreaming, one does not seem to be recalling a previous occurrence, but\nliving through a fresh experience. We cannot be said to be remembering,\nin the strict sense, unless we have a belief referring to the past.\nImages which, like those in dreams, feel as if they were sensations,\ndo not constitute recollection. There must be some feeling which makes\nus refer the images to a past prototype. Perhaps familiarity is enough\nto cause us to do so. And perhaps this also explains the experience of\ntrying to remember something and feeling that we are not remembering\nit right. Parts of a complex image may feel more familiar than other\nparts, and we then feel more confidence in the correctness of the\nfamiliar parts than in that of the others. The conviction that the\nimage we are forming of a past event is wrong might _seem_ to imply\nthat we must be knowing the past otherwise than by images, but I do not\nthink this conclusion is really warranted, since degrees of familiarity\nin images suffice to explain this experience.\n\n(_b_) What is the relation of the present happening to the event\nremembered? If we recollect correctly, the several images will have\nthat kind of resemblance of quality which images can have to their\nprototypes, and their structure and relations will be identical with\nthose of their prototypes. Suppose, for instance, you want to remember\nwhether, in a certain room, the window is to the right or left of the\ndoor as viewed from the fireplace. You can observe your image of the\nroom, consisting (_inter alia_) of an image of the door and an image\nof the window standing (if your recollection is correct) in the same\nrelation as when you are actually seeing the room. Memory will consist\nin attaching to this complex image the sort of belief that refers to\nthe past; and the correctness of memory consists of similarity of\nquality and identity of structure between the complex image and a\nprevious perception.\n\nAs for the trustworthiness of memory, there are two things to be said.\nTaken as a whole, memory is one of the independent sources of our\nknowledge; that is to say, there is no way of arriving at the things we\nknow through memory by any argument wholly derived from things known\notherwise. But no single memory is obliged to stand alone, because it\nfits, or does not fit, into a system of knowledge about the past based\nupon the sum-total of memories. When what is remembered is a perception\nby one or more of the public senses, other people may corroborate it.\nEven when it is private, it may be confirmed by other evidence. You\nmay remember that you had a toothache yesterday, and that you saw\nthe dentist to-day; the latter fact may be confirmed by an entry in\nyour diary. All these make a consistent whole, and each increases the\nlikelihood of the other. Thus we can test the truth of any particular\nrecollection, though not of memory as a whole. To say that we cannot\ntest the truth of memory as a whole is not to give a reason for\ndoubting it, but merely to say that it is an independent source of\nknowledge, not wholly replaceable by other sources. We know that our\nmemory is fallible, but we have no reason to distrust it on the whole\nafter sufficient care in verification has been taken.\n\nThe causation of particular acts of recollection seems to be wholly\nassociative. Something in the present is very like something in the\npast, and calls up the context of the past occurrence in the shape of\nimages or words; when attention falls upon this context, we believe\nthat it occurred in the past, not as mere images, and we then have an\nact of recollection.\n\nThere are many difficult problems connected with memory which I have\nnot discussed, because they have an interest which is more purely\npsychological than philosophical. It is memory as a source of knowledge\nthat specially concerns the philosopher.\n\n\n\n\nCHAPTER XIX\n\nTHE INTROSPECTIVE ANALYSIS OF PERCEPTION\n\n\nWe have considered perception already from the behaviourist standpoint,\nand also from that of physics. In the present chapter we are to\nconsider it from the standpoint of self-observation, with a view to\ndiscovering as much as we can about the intrinsic character of the\nevent in us when we perceive. I shall begin with certain traditional\ndoctrines as to mental events, and shall thence pass to the doctrines\nthat I wish to advocate.\n\nThe words \u201cmind\u201d and \u201cmatter\u201d are used glibly, both by ordinary people\nand by philosophers, without any adequate attempt at definition.\nPhilosophers are much to blame for this. My own feeling is that there\nis not a sharp line, but a difference of degree; an oyster is less\nmental than a man, but not wholly un-mental. And I think \u201cmental\u201d is a\ncharacter, like \u201charmonious\u201d or \u201cdiscordant\u201d, that cannot belong to a\nsingle entity in its own right, but only to a system of entities. But\nbefore defending this view, I wish to spend some time on the theories\nthat have been current in the past.\n\nTraditionally, there are two ways of becoming aware that something\nexists, one by the senses, the other by what is called \u201cintrospection\u201d,\nor what Kant called the \u201cinner sense\u201d. By means of introspection, it is\nmaintained, we become aware of occurrences quite different in kind from\nthose perceived through the outer senses. Occurrences known through\nintrospection are traditionally called \u201cmental\u201d, and so are any other\noccurrences which intrinsically resemble them.\n\nMental occurrences are traditionally of three main types, called\nknowing, willing, and feeling. \u201cFeeling\u201d, in this connection, means\npleasure and unpleasure--we do not say \u201cpleasure and pain\u201d, because\n\u201cpain\u201d is an ambiguous word: it may stand for painful sensation, as\nwhen you say \u201cI have a pain in my tooth\u201d, or it may stand for the\nunpleasant character of the sensation. Roughly pleasure is a quality\nwhich makes you want an experience to continue, and unpleasure is the\nopposite quality which makes you want an experience to stop. However, I\nam not concerned to enlarge upon feeling at present.\n\nAs for the other two kinds of mental occurrence, \u201cknowing\u201d and\n\u201cwilling\u201d are recognised as too narrow to describe what is meant.\nPhilosophers wish to include not only knowledge but also error, and\nnot only the sort of knowledge that is expressed in beliefs but also\nthe sort that occurs in perception. The word \u201ccognition\u201d or \u201ccognitive\nstate\u201d is used to cover everything that could possibly be described as\neither knowledge or error; perception is _prima facie_ included, but\npure sensation is more debatable.\n\n\u201cWilling\u201d, again, is too narrow a term. A term is required which will\ninclude desire and aversion, and generally those states of mind which\nlead up to action. These are all included under the head of \u201cconation\u201d,\na technical term invented for this special purpose.\n\nCognition and conation both have, in the orthodox theory, the property\nof being directed to an _object_. What you perceive or believe, what\nyou desire or will, is something different from your state of mind. To\ntake instances: you remember a past event, but your remembering occurs\nnow; therefore your remembering is a different occurrence from what you\nremember. You will to move your arm, but the movement is a physical\noccurrence, and therefore obviously different from your volition. Many\npsychologists have taken this relation to an object as the essential\ncharacteristic of mind--notably the two Austrians Brentano and Meinong.\nSometimes feeling also is regarded as having an object: it is held\nthat we are pleased or displeased _at_ something. This view, however,\nhas never won general acceptance, whereas the view that cognition and\nconation are directed to objects may be regarded as orthodox.\n\nIt is undeniable that this characteristic of being directed to objects\nis, in some sense, a property of cognition and conation, but there\nis room for great difference of opinion as to the proper analysis of\nthe property. I think we cannot hope to understand the word \u201cmental\u201d\nuntil we have undertaken this analysis, and I shall therefore proceed\nto address myself to it. I shall confine myself to cognition, which is\nmore important for our present purposes than conation.\n\nAs regards cognition, though philosophers have disagreed widely, I\nthink that, until recently, most would have assented to at least the\nfollowing paragraph:\n\nCognition is of various sorts. Take, as important kinds, perception,\nmemory, conception, and beliefs involving concepts. Perception is the\nordinary awareness of sensible objects: seeing a table, hearing a\npiano, and so on. Memory is awareness of a past occurrence, when this\nawareness is direct, not inferred or derived from testimony. Conception\nis more difficult to characterise. One may say, as a way of pointing\nout what is intended, that we \u201cconceive\u201d whenever we understand the\nmeaning of an abstract word, or think of that which is in fact the\nmeaning of the word. If you see a white patch of snow, or recall it\nby means of images, you do not have a concept; but if you think about\nwhiteness, you have a concept. Similarly if, after seeing a number of\ncoins, you think about roundness as a common characteristic of all of\nthem, you have a concept. The object of your thought, in such a case,\nis a _universal_ or a Platonic idea. Every sentence must contain at\nleast one word expressing a concept, and therefore every belief that\ncan be expressed in words contains concepts.\n\nEach of these kinds of cognitive attitude involves its own problems.\nIn the present chapter we are concerned with perception. This has to\nbe treated both introspectively and causally; it is the introspective\ntreatment that we have now to undertake.\n\nWhen you have the experience called \u201cseeing a table\u201d, there is a\ncertain amount of difference between your unreflecting judgment and\nwhat careful examination reveals as to the nature of your experience.\nYou judge that the table is rectangular, but the patch of colour in\nyour visual field is not a rectangle; when you learn to draw, you have\nto draw the table as it really seems and not as it seems to seem. You\nhave images of sensations of touch; if you were to try to touch the\ntable and it turned out to be an optical illusion, you would get a\nviolent shock of surprise. You have also expectations of a certain\ndegree of permanence and weight. If you went to lift the table, you\nwould find your muscles quite wrongly adjusted if the table were much\nlighter than it looked. All these elements must be included in the\nperception, though not in the sensation.\n\n\u201cSensation\u201d, as opposed to perception, is more or less hypothetical.\nIt is supposed to be the core, in the perception, which is solely due\nto the stimulus and the sense-organ, not to past experience. When\nyou judge that the table is rectangular, it is past experience that\nenables and compels you to do so; if you had been born blind and just\noperated upon, you could not make this judgment. Nor would you have\nexpectations of hardness, etc. But none of this can be discovered by\nintrospection. From an introspective point of view, the elements due\nto past experience are largely indistinguishable from those due to the\nstimulus alone. One supposes that past experience modifies the brain,\nand thereby modifies the mental occurrence due to the stimulus. The\nnotion of sensation as opposed to perception belongs, therefore, to the\ncausal study of perception, not to the introspective study.\n\nThere is, however, a distinction to be made here. You can discover\nby mere self-observation that visual objects are accompanied by\nexpectations or images of touch; and similarly if you touch an object\nin the dark you will probably be led to form some visual image of\nit. Here you can arrive at a certain degree of analysis of your\nperception through the fact that images, as a rule, _feel_ different\nfrom the immediate results of a sensory stimulus. On the other hand,\nno amount of introspection alone will reveal such things as the blind\nspot. The filling in of a sensation by elements belonging to the same\nsense is much less discoverable by introspection than the filling\nin by associated images belonging to other senses. Thus although\nby introspection alone we could discover _part_ of the influence\nof experience on perception, there is another part which we cannot\ndiscover in this way.\n\nRemaining in the introspective attitude, it is evident that the\ncontents of our minds at any given moment are very complex. Throughout\nour normal waking life we are always seeing, hearing, and touching,\nsometimes smelling and tasting, always having various bodily\nsensations, always feeling pleasant or unpleasant feelings (usually\nboth), always having desires or aversions. We are not normally aware\nof all these items, but we can become aware of any of them by turning\nour attention in the right direction. I am not at present discussing\n\u201cunconscious mental states\u201d, because they, obviously, can only be known\ncausally, and we are now considering what can be known introspectively.\nThere may be any number of perceptions that cannot be known by\nintrospection; the point for us, at the moment, is that those that can\nbe discovered by introspection at any one time are many and various.\n\nI do not wish, just now, to discuss the nature of attention; I wish\nonly to point out that it enables us to take the first steps in\nabstraction. Out of the whole multiplicity of objects of sense, it\nenables us to single out a small selection, which is an indispensable\npreliminary to abstraction. For example, attention will enable us to\ndiscriminate a coloured pattern which we are seeing, and to separate\nit from the other things we see and from images and other objects of\nsense and thoughts which may exist simultaneously. For the sake of\nsimplicity, let us suppose that we discriminate a black and white\npattern in the form of a triangle. Within this pattern we can further\ndiscriminate sides and angles and an inside and outside--of course the\nsides are not mathematical lines nor the angles mathematical points.\n\nWe now come to a question of very great importance, upon which our\nviews of the relations of mind and matter largely depend. The question\nis this:\n\nWhat difference is there between the propositions \u201cthere is a triangle\u201d\nand \u201cI see a triangle\u201d?\n\nBoth these statements seem as certain as any statement can be--at least\nif rightly interpreted. As always happens in such cases, we are quite\ncertain of something, but not quite certain what it is that we are\ncertain of. I want to ask whether this something that we are certain\nof is really different in the above two statements, or whether the\ndifference between them is only as to surroundings of which we are not\ncertain. Most philosophers hold that there is a difference in what we\nare certain of; Mach, James, Dewey, the American realists, and I hold\nthat the difference is in the uncertain context. Let us examine this\nquestion.\n\nThe _suggestions_ of the two statements \u201cI see a triangle\u201d and \u201cthere\nis a triangle\u201d are obviously different. The first states an event in\nmy life, and suggests its possible effects upon me. The second aims at\nstating an event in the world, supposed to be equally discoverable by\nother people. You might say \u201cthere is a triangle\u201d if you had seen it\na moment ago but now had your eyes shut; in this case you would not\nsay \u201cI see a triangle\u201d. On the other hand, one sometimes, under the\ninfluence of indigestion or fatigue, sees little black dots floating in\nthe air; in such circumstances you would say \u201cI see a black dot\u201d, but\nnot \u201cthere is a black dot\u201d. This illustration shows that when you say\n\u201cthere is a black dot\u201d you are making a stronger assertion than when\nyou say \u201cI see a black dot\u201d. In the other case, when you say \u201cthere is\na triangle\u201d because you saw it a moment ago, though not now, you have\nthree stages: First, memory assures you of the proposition \u201cI saw a\ntriangle\u201d, and then you pass on to \u201cthere was a triangle\u201d, and then,\nfurther, to \u201cthere is a triangle, because nothing can have happened to\ndestroy it so quickly.\u201d Here we have obviously passed far beyond the\nregion of immediate certainty.\n\nIt seems clear, therefore, that, of our two statements, the one which\ncomes nearest to expressing the fact of which we are immediately\ncertain is \u201cI see a triangle\u201d, because the other makes inferences to\nsomething public, and thus goes beyond the bare datum. This is on the\nassumption that we should not say \u201cthere is a black dot\u201d when we see\na black dot which we attribute to eye-trouble and therefore suppose\nthat no one else can see. Let us therefore concentrate upon \u201cI see a\ntriangle\u201d, and ask ourselves whether the whole of this, or only part,\ncan be accepted as a primitive certainty.\n\nA moment\u2019s reflection shows that both \u201cI\u201d and \u201csee\u201d are words which\ntake us beyond what the momentary event reveals. Take \u201cI\u201d to begin\nwith. This is a word whose meaning evidently depends upon memory\nand expectation. \u201cI\u201d means the person who had certain remembered\nexperiences and is expected to have certain future experiences. We\nmight say \u201cI see a triangle now and I saw a square a moment ago.\u201d The\nword \u201cI\u201d has exactly the same meaning in its two occurrences in this\nsentence, and therefore evidently has a meaning dependent upon memory.\nNow it is our object to arrive at the contribution to your knowledge\nwhich is made by seeing the triangle at the moment. Therefore, since\nthe word \u201cI\u201d takes you beyond this contribution, we must cut it out\nif we want to find a correct verbal expression for what is added to\nour knowledge by seeing the triangle. We will say \u201ca triangle is being\nseen\u201d. This is at any rate one step nearer to what we are seeking.\n\nBut now we must deal with the word \u201cseen\u201d. As ordinarily used, this is\na causal word, suggesting something dependent upon the eyes. In this\nsense, it obviously involves a mass of previous experience; a new-born\nbaby does not know that what it sees depends upon its eyes. However,\nwe could eliminate this. Obviously all objects of sight have a common\nquality, which no objects of touch or hearing have; a visual object is\ndifferent from an auditory object, and so on. Therefore instead of\nsaying \u201ca triangle is being seen\u201d, we should say \u201cthere is a visual\ntriangle\u201d. Of course the meanings of the words \u201cvisual\u201d and \u201ctriangle\u201d\ncan only be learnt by experience, but they are not _logically_\ndependent upon experience. A being could be imagined which would know\nthe words at birth; such a being could express its datum in the words\n\u201cthere is a visual triangle\u201d. In any case, the problems remaining\nbelong to the study of concepts; we will therefore ignore them at\npresent.\n\nNow in English the words \u201cthere is\u201d are ambiguous. When I used them\nbefore, saying, \u201cthere is a triangle\u201d, I meant them in the sense of\n\u201c_voila_\u201d or \u201c_da ist_\u201d. Now I mean them in the sense of \u201c_il y a_\u201d\nor \u201c_es giebt_\u201d. One might express what is meant by saying \u201ca visual\ntriangle exists\u201d, but the word \u201cexist\u201d has all sorts of metaphysical\nconnotations that I wish to avoid. Perhaps it is best to say \u201coccurs\u201d.\n\nWe have now arrived at something which is just as true when your\nperception is illusory as when it is correct. If you say \u201ca visual\nblack dot is occurring\u201d, you are speaking the truth, if there is one\nin your field of vision. We have eliminated the suggestion that others\ncould see it, or that it could be touched, or that it is composed of\nmatter in the sense of physics. All these suggestions are present when\none says, in ordinary conversation, \u201cthere is a black dot\u201d; they are\nintended to be eliminated by the addition of the word \u201cvisual\u201d and the\nsubstitution of \u201cis occurring\u201d for \u201cthere is\u201d. By these means we have\narrived at what is indubitable and intrinsic in the addition to your\nknowledge derived from a visual datum.\n\nWe must now ask ourselves once more: Is there still a distinction,\nwithin what is immediate and intrinsic, between the occurrence of a\nvisual datum and the cognition of it? Can we say, on the basis of\nimmediate experience, not only \u201ca visual black dot occurs\u201d, but also\n\u201ca visual black dot is cognised\u201d? My feeling is that we cannot. When\nwe say that it is cognised, we seem to me to mean that it is part\nof an experience, that is to say, that it can be remembered, or can\nmodify our habits, or, generally, can have what are called \u201cmnemic\u201d\neffects. All this takes us beyond the immediate experience into the\nrealm of its causal relations. I see no reason to think that there is\nany duality of subject and object in the occurrence itself, or that\nit can properly be described as a case of \u201cknowledge\u201d. It gives rise\nto knowledge, through memory, and through conscious or unconscious\ninferences to the common correlates of such data. But in itself it\nis not knowledge, and has no duality. The datum is a datum equally\nfor physics and for psychology; it is a meeting point of the two. It\nis neither mental nor physical, just as a single name is neither in\nalphabetical order nor in order of precedence; but it is part of the\nraw material of both the mental and the physical worlds. This is the\ntheory which is called \u201cneutral monism\u201d, and is the one that I believe\nto be true.\n\n\n\n\nCHAPTER XX\n\nCONSCIOUSNESS?\n\n\nTwenty-three years have elapsed since William James startled the\nworld with his article entitled \u201cDoes \u2018consciousness\u2019 exist?\u201d In this\narticle, reprinted in the volume called _Essays in Radical Empiricism_,\nhe set out the view that \u201cthere is only one primal stuff or material in\nthe world\u201d, and that the word \u201cconsciousness\u201d stands for a function,\nnot an entity. He holds that there are \u201cthoughts\u201d, which perform the\nfunction of \u201cknowing\u201d, but that thoughts are not made of any different\n\u201cstuff\u201d from that of which material objects are made. He thus laid\nthe foundations for what is called \u201cneutral monism\u201d, a view advocated\nby most American realists. This is the view advocated in the present\nvolume. In this chapter, we have to ask ourselves whether there is\nanything that we can call \u201cconsciousness\u201d in any sense involving a\npeculiar kind of stuff, or whether we can agree with William James that\nthere is no \u201cinner duplicity\u201d in the stuff of the world as we know it,\nand that the separation of it into knowing and what is known does not\nrepresent a fundamental dualism.\n\nThere are two very different meanings attached to the word\n\u201cconsciousness\u201d by those who use it. On the one hand, we are said to be\n\u201cconscious of\u201d something; in this sense, \u201cconsciousness\u201d is a relation.\nOn the other hand, \u201cconsciousness\u201d may be regarded as a quality of\nmental occurrences, not consisting in their relation to other things.\nLet us take the first view first, since, in discussing it, we shall\nfind reasons for rejecting the second view.\n\nWhat is the relation we call being \u201cconscious of\u201d something? Take the\ndifference between a person awake and a person asleep. The former\nreacts to all kinds of stimuli to which the latter does not react; we\ntherefore say that the latter is not \u201cconscious of\u201d what is happening\nin his neighbourhood. But even if the sleeper does react in a fashion,\nfor example, by turning away from the light, such a reaction does not\nfall within what is commonly regarded as \u201cknowledge\u201d or \u201cawareness\u201d; we\nshould say that the sleeper turned over \u201cunconsciously\u201d. If he wakes\nup sufficiently to speak intelligently, for instance to address the\ndisturber by name, we consider him \u201cconscious\u201d. So we do if we find\nthat he remembers the incident next morning. But common sense does\nnot regard any and every bodily movement in response to a stimulus as\nevidence of \u201cconsciousness\u201d. There is no doubt, I think, that common\nsense regards certain kinds of response as evidence of some \u201cmental\u201d\nprocess caused by the stimulus, and regards the \u201cconsciousness\u201d as\nresiding in the inferred \u201cmental\u201d occurrence.\n\nSometimes, however, as in hypnotism and sleep-walking, people refuse\nto admit \u201cconsciousness\u201d even where many of the usual marks of it are\npresent. For this there are certain reasons. One of them is subsequent\nlack of memory; another is lack of intelligence in what is being done.\nIf you offer a hypnotised patient a drink of ink, telling him it is\nport wine, and he drinks it up with every sign of enjoyment, you say\nthat he is not \u201cconscious\u201d, because he does not react normally to\nthe nasty taste. It would seem better, however, to say that he is\nconscious of the hypnotist and what he commands, though not of other\nthings of which he would be conscious in a normal condition. And lack\nof subsequent memory is a very difficult criterion, since we normally\nforget many things that have happened to us, and the sleep-walker\u2019s\nforgetting is only unusually complete. This is obviously a matter of\ndegree. Take next morning\u2019s memories in the case of a man who was drunk\novernight. They become more and more vague as he reviews the later\nhours of the evening, but there is no sharp line where they cease\nabruptly. Thus, if memory is a test, consciousness must be a matter of\ndegree. I think that here, again, common sense regards a certain amount\nof memory as necessary evidence to prove that there were \u201cmental\u201d\nprocesses at the time of the acts in question, acts in sleep being\nregarded as not involving \u201cmind\u201d, and other acts in certain abnormal\nconditions being supposed to resemble those of sleep in this respect.\n\nIt follows that, if we are to find out what is commonly meant by\n\u201cconsciousness\u201d, we must ask ourselves what is meant by a \u201cmental\u201d\noccurrence. Not every mental occurrence, however, is in question.\nThe only kinds concerned are those which seem to have relation to an\n\u201cobject\u201d. A feeling of pleasant drowsiness would commonly count as\n\u201cmental\u201d, but does not involve \u201cconsciousness\u201d of an \u201cobject\u201d. It is\nthis supposed peculiar relation to an \u201cobject\u201d that we have to examine.\n\nWe may take, as the best example, an ordinary act of perception. I\nsee, let us say, a table, and I am convinced that the table is outside\nme, whereas my seeing of it is a \u201cmental\u201d occurrence, which is inside\nme. In such a case I am \u201cconscious\u201d of the table--so at least common\nsense would say. And since I cannot see without seeing something, this\nrelation to an \u201cobject\u201d is of the very essence of seeing. The same\nessential relation to an \u201cobject\u201d, it would be said, is characteristic\nof every kind of consciousness.\n\nBut when we begin to consider this view more closely, all sorts of\ndifficulties arise. We have already seen that, on grounds derived from\nphysics, the table itself, as a physical thing, cannot be regarded as\nthe object of our perception, if the object is something essential to\nthe existence of the perception. In suitable circumstances, we shall\nhave the same perception although there is no table. In fact, there\nis no event outside the brain which _must_ exist whenever we \u201csee\na table\u201d. It seems preposterous to say that when we think we see a\ntable we really see a motion in our own brain. Hence we are led to the\nconclusion that the \u201cobject\u201d which is essential to the existence of\nan act of perception is just as \u201cmental\u201d as the perceiving. In fact,\nso this theory runs, the mental occurrence called \u201cperceiving\u201d is one\nwhich contains within itself the relation of perceiver and perceived,\nboth sides of the relation being equally \u201cmental\u201d.\n\nNow, however, there seems no longer any reason to suppose that there is\nany essentially relational character about what occurs in us when we\nperceive. The original reason for thinking so was the naively realistic\nview that we see the actual table. If what we see is as mental as our\nseeing, why distinguish between the two? The coloured pattern that\nwe see is not really \u201cout there\u201d, as we had supposed; it is in our\nheads, if we are speaking of physical space. True, more than a coloured\npattern occurs when we \u201csee a table\u201d. There are tactual expectations or\nimages: there is probably belief in an external object; and afterwards\nthere may be memory or other \u201cmnemic\u201d effects. All this may be taken as\nrepresenting what the above theory took to be the \u201csubject\u201d side of an\nact of perception, while the coloured pattern is what the theory took\nto be the \u201cobject\u201d side. But both sides are on a level as regards being\n\u201cmental\u201d. And the relation between the two sides is not of such a kind\nthat the existence of the one logically demands the existence of the\nother; on the contrary, the relation between the two sides is causal,\nbeing dependent upon experience and the law of association.\n\nIf this is correct, what really happens when, as common sense would\nsay, we are conscious of a table, is more or less as follows. First\nthere is a physical process external to the body, producing a stimulus\nto the eye which occurs rarely (not never) in the absence of an actual\nphysical table. Then there is a process in the eye, nerves, and brain,\nand finally there is a coloured pattern. This coloured pattern, by the\nlaw of association, gives rise to tactual and other expectations and\nimages; also, perhaps, to memories and other habits. But everything in\nthis whole series consists of a causally continuous chain of events\nin space-time, and we have no reason to assert that the events in us\nare so very different from the events outside us--as to this, we must\nremain ignorant, since the outside events are only known as to their\nabstract mathematical characteristics, which do not show whether these\nevents are like \u201cthoughts\u201d or unlike them.\n\nIt follows that \u201cconsciousness\u201d cannot be defined either as a\npeculiar kind of relation or as an intrinsic character belonging to\ncertain events and not to others. \u201cMental\u201d events are not essentially\nrelational, and we do not know enough of the intrinsic character of\nevents outside us to say whether it does or does not differ from that\nof \u201cmental\u201d events. But what makes us call a certain class of events\n\u201cmental\u201d and distinguish them from other events is the combination\nof sensitivity with associative reproduction. The more markedly this\ncombination exists, the more \u201cmental\u201d are the events concerned; thus\nmentality is a matter of degree.\n\nThere is, however, a further point which must be discussed in this\nconnection, and that is \u201cself-consciousness\u201d, or awareness of our own\n\u201cmental\u201d events. We already had occasion to touch on this in Chapter\nXVI in connection with Descartes\u2019 \u201cI think, therefore I am\u201d. But I want\nto discuss the question afresh in connection with \u201cconsciousness\u201d.\n\nWhen the plain man sees \u201ca table\u201d in the presence of a philosopher, the\nplain man can be driven, by the arguments we have repeatedly brought\nforward, to admit that he cannot have complete certainty as to anything\noutside himself. But if he does not lose his head or his temper, he\nwill remain certain that there is a coloured pattern, which may be in\nhim, but indubitably exists. No argument from logic or physics even\ntends to show that he is mistaken in this; therefore there is no reason\nwhy he should surrender his conviction. The argument about knowledge\nin Chapter VIII showed that, accepting the usual views of physicists\nas to causal laws, our knowledge becomes more certain as the causal\nchain from object to reaction is shortened, and can only be quite\ncertain when the two are in the same place in space-time, or at least\ncontiguous. Thus we should expect that the highest grade of certainty\nwould belong to knowledge as to what happens in our own heads. And this\nis exactly what we have when we are aware of our own \u201cmental\u201d events,\nsuch as the existence of a coloured pattern when we thought we were\nseeing a table.\n\nWe might, therefore, if we were anxious to preserve the word \u201cmental\u201d,\ndefine a \u201cmental\u201d event as one that can be known with the highest\ngrade of certainty, because, in physical space-time, the event and the\nknowing of it are contiguous. Thus \u201cmental\u201d events will be certain of\nthe events that occur in heads that have brains. They will not be _all_\nevents that occur in brains, but only such as cause a reaction of the\nkind that can be called \u201cknowledge\u201d.\n\nThere are, however, still a number of difficult questions, to which, as\nyet, a definitive answer cannot be given. When we \u201cknow\u201d a thought of\nour own, what happens? And do we know the thought in a more intimate\nway than we know anything else? Knowledge of external events, as we\nhave seen, consists of a certain sensitivity to their presence, but\nnot in having in or before our minds anything similar to them, except\nin certain abstract structural respects. Is knowledge of our own minds\nequally abstract and indirect? Or is it something more analogous to\nwhat we ordinarily imagine knowledge to be?\n\nTake first the question: What happens when we \u201cknow\u201d a thought of our\nown? Taking the definition of \u201cknowing\u201d that we adopted in Chapter\nVIII, we shall say: We \u201cknow\u201d a thought of our own when an event in\nour brain causes a characteristic reaction which is present when the\nevent occurs and not otherwise. In this sense, whenever we say, \u201cI see\na table\u201d, we are knowing a thought, since an event in our brain is the\nonly invariable antecedent of such a statement (assuming it to be made\ntruthfully). We may _think_ we are knowing a table, but this is an\nerror.\n\nThus the difference between introspective and other knowledge is only\nin our intention and in the degree of certainty. When we say, \u201cI see a\ntable\u201d, we may intend to know an external object, but if so we may be\nmistaken; we are, however, actually knowing the occurrence of a visual\npercept. When we describe the same occurrence in the words \u201ca certain\ncoloured pattern is occurring\u201d, we have changed our intention and are\nmuch more certain of being right. Thus all that differentiates our\nreaction when it gives introspective knowledge from our reaction when\nit gives knowledge of another kind is the elimination of a possible\nsource of error.\n\nI come now to the question: Do we know our own thoughts in a more\nintimate way than we know anything else? This is a question to which it\nis difficult to give precision; it describes something that one _feels_\nto be a problem without being able to say exactly what the problem\nis. However, some things can be said which may serve to clear up our\nfeelings, if not our ideas.\n\nSuppose you are asked to repeat after a man whatever he says, as a\ntest of your hearing. He says \u201chow do you do?\u201d and you repeat \u201chow\ndo you do?\u201d This is your knowledge-reaction, and you hear yourself\nspeaking. You can perceive that what you hear when you speak is\nclosely similar to what you hear when the other man speaks. This makes\nyou feel that your reaction reproduces accurately what you heard.\nYour knowledge-reaction, in this case, is the cause of an occurrence\nclosely similar to the occurrence that you are knowing. Moreover, our\ninveterate naive realism makes us think that what we said was what we\nheard while we were speaking. This is, of course, an illusion, since\nan elaborate chain of physical and physiological causation intervenes\nbetween speaking and hearing oneself speak; nevertheless, the illusion\nre-enforces our conviction that our knowledge, in such a case, is very\nintimate. And it is, in fact, as intimate as it can hope to be, when\nour knowledge-reaction reproduces the very event we are knowing, or\nat least an event extremely similar to it. This _may_ be the case on\nother occasions, but we can only know, with any certainty, that it\nis the case when what is known is a percept. This accounts for the\nfact that our most indubitable and complete knowledge is concerning\npercepts, not concerning other mental events or events in the external\nworld. Our reaction to a sound can be to make a similar sound, and if\nwe are clever enough we can paint something very like what we see. But\nwe cannot show our knowledge of a pleasure by creating for ourselves\nanother very similar pleasure, nor of a desire by creating a similar\ndesire. Thus percepts are known with more accuracy and certainty than\nanything else either in the outer world or in our own minds.\n\nThe conclusion we have reached in this chapter is that William James\nwas right in his views on \u201cconsciousness\u201d. No mental occurrence has,\nin its own intrinsic nature, that sort of relational character that\nwas implied in the opposition of subject and object, or of knower and\nknown. Nevertheless we can distinguish \u201cmental\u201d events from others,\nand our most indubitable knowledge is concerned with a certain class\nof mental events. We have arrived at this result by following out to\nits logical conclusion the behaviourist definition of knowledge which\nwe gave in Chapter VIII. We have had to modify considerably the point\nof view which originally led us to that definition, the modification\nhaving been forced upon us by the physical knowledge which, starting\nfrom a common-sense realism, has been gradually driven, through the\ncausal theory of perception, to a view of cognition far more subjective\nthan that from which physicists, like the rest of mankind, originally\nset out. But I do not see how there can be any escape from this\ndevelopment.\n\n\n\n\nCHAPTER XXI\n\nEMOTION, DESIRE, AND WILL\n\n\nHitherto, in our investigation of man from within, we have considered\nonly the cognitive aspect, which is, in fact, the most important to\nphilosophy. But now we must turn our attention to the other sides of\nhuman nature. If we treat them more briefly than the cognitive side,\nit is not because they are less important, but because their main\nimportance is practical and our task is theoretical. Let us begin with\nthe emotions.\n\nThe theory of the emotions has been radically transformed by the\ndiscovery of the part played by the ductless glands. Cannon\u2019s _Bodily\nChanges in Pain, Hunger, Fear and Rage_ is a book whose teaching\nhas come to be widely known, though not more so than its importance\nwarrants. It appears that certain secretions from the glands into the\nblood are the essential physiological conditions of the emotions.\nSome people say that the physiological changes correlated with these\nsecretions _are_ the emotions. I think this view must be received with\nsome caution. As everyone knows, the adrenal glands secrete adrenin,\nwhich produces the bodily symptoms of fear or rage. On one occasion my\ndentist injected a considerable amount of this substance into my blood,\nin the course of administering a local an\u00e6sthetic. I turned pale and\ntrembled, and my heart beat violently; the bodily symptoms of fear were\npresent, as the books said they should be, but it was quite obvious to\nme that I was not actually feeling fear. I should have had the same\nbodily symptoms in the presence of a tyrant about to condemn me to\ndeath, but there would have been something extra which was absent when\nI was in the dentist\u2019s chair. What was different was the cognitive\npart: I did not feel fear because I knew there was nothing to be afraid\nof. In normal life, the adrenal glands are stimulated by the perception\nof an object which is frightful or enraging; thus there is already a\ncognitive element present. The fear or rage attaches itself to the\nobject which has stimulated the glands, and the full emotion arises.\nBut when adrenin is artificially administered, this cognitive element\nis absent, and the emotion in its entirety fails to arise. Probably if\nit were administered in sleep it would produce a dream of terror, in\nwhich the dreamer\u2019s imagination would supply an object for fear. The\nsame thing might happen on waking life with animals or young children.\nBut with an adult of average rationality, the knowledge that there is\nnothing to be afraid of inhibits the full development of the emotion.\nFear and rage are both active emotions, demanding a certain kind of\nbehaviour towards an object; when this behaviour is obviously not\ncalled for, it is impossible to feel either emotion fully.\n\nThere are, however, other emotions, such as melancholy, which do not\ndemand an object. These, presumably, can be caused in their entirety\nby administering the proper secretions. A disordered liver may cause\nmelancholy which is not relieved by knowledge of its source. The\nemotions which do not require an object are those which do not call for\nany appropriate line of action.\n\nEmotions are subject to \u201cconditioning\u201d, so that the stimuli which call\nthem out become more various as a result of experience. Dr. Watson has\nfound only two original stimuli to fear in young infants, namely loud\nnoises, and lack of support; but anything associated with either of\nthese may become terrifying.\n\nThe separation of an emotional element in our integral reaction to a\nsituation is more or less artificial. No doubt there is a definite\nphysiological concomitant, namely stimulation of a gland; but fear,\nfor example, involves a mode of action towards an object, for which\nmode of action the secretion of adrenin is helpful. There is,\nhowever, something in common among a number of occasions that have a\ngiven emotional tone; this may be seen from the fact that they are\nassociated. When we are feeling some emotion strongly, we tend to think\nof other occasions when we have had similar feelings. Association by\nmeans of emotional similarity is a characteristic of a great deal of\npoetry. And this accounts for the fact that, if our blood is in a state\nusually associated with terror, we shall, if our critical faculty is in\nabeyance, be very likely to imagine some cause of fear so vividly as to\nbelieve that it is really present:\n\n    In the night, imagining some fear,\n    How easy is a bush supposed a bear.\n\nBut in a rational man, if he is not drunk or sleepy, other associations\nare too strong for this production of imaginary terrors. That is why it\nis possible to show the physical symptoms of fear under the influence\nof adrenin, without actually feeling the emotion.\n\nThe emotions are what makes life interesting, and what makes us feel it\nimportant. From this point of view, they are the most valuable element\nin human existence. But when, as in philosophy, we are trying to\nunderstand the world, they appear rather as a hindrance. They generate\nirrational opinions, since emotional associations seldom correspond\nwith collocations in the external world. They cause us to view the\nuniverse in the mirror of our moods, as now bright, now dim, according\nto the state of the mirror. With the sole exception of curiosity, the\nemotions are on the whole a hindrance to the intellectual life, though\nthe degree of vigour required for successful thinking is likely to be\ncorrelated with a considerable susceptibility to emotion. If I say\nlittle about the emotions in this book, it is not from under-estimating\ntheir human importance, but solely because the task upon which we are\nengaged is theoretical rather than practical: to understand the world,\nnot to change it. And if emotion determines the ends we shall pursue,\nknowledge is what gives us the power to realise them. Even from the\npractical point of view, the advancement of knowledge is more useful\nthan anything else that lies within human power.\n\nI come now to the subject of _desire_, which we considered from a\nbehaviourist standpoint in Chapter III. I want now to ask whether there\nis anything to be added from an introspective point of view.\n\nLet us again remind ourselves that there is an element of artificiality\nin isolating elements within the one process leading from stimulus to\nreaction. Whenever a stimulus produces a reaction, we may consider the\nreaction as the effect of the stimulus, or as the cause of further\neffects. The former is the natural way of viewing the reaction when we\nare concerned with knowledge; the latter is the natural way when we are\nconcerned with desire and will. In desire, we wish to change something\nin ourselves or in our environment or both. The question is: What can\nwe discover introspectively about desire?\n\nI think that here, as in the case of knowledge, the purely\nbehaviouristic account is more important causally than the\nintrospective account, and applies over a much wider range. Desire as\na characteristic of behaviour, as considered in Chapter III, begins\nvery low in the scale of evolution, and remains, even in human beings,\nthe whole of what can be discovered in a large number of instances.\nThe Freudian \u201cunconscious\u201d desires give a formula which is useful as\nexplaining causally a number of acts, but these desires do not exist as\nanything except ways of behaving. Some desires, on the other hand, are\nconscious and explicit. What, exactly, is added in these last that is\nnot present in the others?\n\nLet us take some stock instance, say, Demosthenes desiring to become\na great orator. This was a desire of which he was conscious, and\nin accordance with which he deliberately moulded his actions. One\nmay suppose, to begin with, a merely behaviouristic tendency to do\nsuch things as seemed likely to impress his companions. This is\na practically universal characteristic of human nature, which is\ndisplayed naively by children. Then come attempts, just like those of\nrats in mazes, to reach the goal; wrong turnings, leading to derision;\nright turnings, leading to a brief nibble at the cheese of admiration.\nSelf-observation, still of a behaviourist kind, may lead to the\nformula: I want to be admired. At this point the desire has become\n\u201cconscious\u201d. When this point has been reached, knowledge can be brought\nto bear on the problem of achieving the desired end. By association,\nthe means come to be desired also. And so Demosthenes arrives at the\ndecision to subject himself to a difficult training as an orator, since\nthis seems the best way of achieving his end. The whole development\nis closely analogous to that of explicit knowledge out of mere\nsensitivity; it is, indeed, part of the very same evolution. We cannot,\nin our integral reaction to a situation, separate out one event as\nknowledge and another as desire; both knowledge and desire are features\nwhich characterise the reaction, but do not exist in isolation.\n\nIn explicit conscious desire there is always an object, just as there\nis in explicit conscious perception; we desire some event or some state\nof affairs. But in the primitive condition out of which explicit desire\nis evolved, this is not the case. We have a state of affairs which may\nbe said to involve discomfort, and activities of various sorts until a\ncertain different state of affairs is achieved, or fatigue supervenes,\nor some other interest causes a distraction. These activities will\nbe such as to achieve the new state of affairs quickly if there has\nbeen previous experience of a relevant kind. When we reach the level\nof explicit conscious desire, it seems as if we were being attracted\nto a goal, but we are really still pushed from behind. The attraction\nto the goal is a shorthand way of describing the effects of learning\ntogether with the fact that our efforts will continue till the goal\nis achieved, provided the time required is not too long. There are\nfeelings of various kinds connected with desire, and in the case of\nfamiliar desires, such as hunger, these feelings become associated\nwith what we know will cause the desire to cease. But I see no more\nreason in the case of desire than in the case of knowledge to admit\nan _essentially_ relational occurrence such as many suppose desire to\nbe. Only experience, memory and association--so I should say--confer\nobjects upon desire, which are initially blind tendencies to certain\nkinds of activity.\n\nIt remains to say a few words about \u201cwill\u201d. There is a sense in\nwhich will is an observable phenomena, and another in which it is a\nmetaphysical superstition. It is obvious that I can say, \u201cI will hold\nmy breath for thirty seconds\u201d, and proceed to do so; that I can say, \u201cI\nwill go to America\u201d, and proceed to do so; and so on. In this sense,\nwill is an observable phenomenon. But as a faculty, as a separable\noccurrence, it is, I think, a delusion. To make this clear, it will be\nnecessary to examine the observable phenomenon.\n\nVery young infants do not appear to have anything that could be called\n\u201cwill\u201d. Their movements, at first, are reflexes, and are explicable,\nwhere they first cease to be reflexes, by the law of conditioned\nreflexes. One observes, however, something that looks very like will\nwhen the child learns control over fingers and toes. It seems clear,\nin watching this process, that, after some experience of involuntary\nmovements, the child discovers how to think of a movement first\nand then make the movement, and that this discovery is exceedingly\npleasurable. We know that, in adult life, a deliberate movement is one\nwhich we think of before we make it. Obviously we cannot think of a\nmovement unless we have previously made it; it follows that no movement\ncan be voluntary unless it has previously been involuntary. I think\nthat, as William James suggested, a voluntary movement is merely one\nwhich is preceded by the thought of it, and has the thought of it as an\nessential part of its cause.\n\nWhen I say this, I do not mean to take any particular view as to what\nconstitutes \u201cthinking\u201d. It may consist almost entirely of talking, as\nDr. Watson holds; or it may be something more. That is not the point\nat present. The point is that, whatever philosophy one may adopt, there\ncertainly is an occurrence which is described by ordinary people as\n\u201cthinking of getting up in the morning\u201d, or \u201cthinking of\u201d any other\nbodily movement. Whatever the analysis of this occurrence may be, it is\nan essential part of the cause of any movement which can be attributed\nto the \u201cwill\u201d.\n\nIt is true, of course, that we may think of a movement without\nperforming it. This is analogous to imagining a state of affairs\nwithout believing in it; each is a rather sophisticated and late\ndevelopment. Each will only happen when we think of several things\nat once, and one of them interferes with another. It may, I think,\nbe assumed that, whenever we think of a possible movement, we have a\ntendency to perform it, and are only restrained, if at all, by some\nthought, or other circumstance, having a contrary tendency.\n\nIf this is the case, there is nothing at all mysterious about the will.\nWhatever may constitute \u201cthinking of\u201d a movement, it is certainly\nsomething associated with the movement itself; therefore, by the usual\nlaw of learned reactions we should expect that thinking of a movement\nwould tend to cause it to occur. This, I should say, is the essence of\nwill.\n\nEmphatic cases of volition, where we decide after a period of\ndeliberation, are merely examples of conflicting forces. You may have\nboth pleasant and unpleasant associations with some place that you are\nthinking of going to; this may cause you to hesitate, until one or\nother association proves the stronger. There may be more than this in\nvolition, but I cannot see any good ground for believing that there is.\n\n\n\n\nCHAPTER XXII\n\nETHICS\n\n\nEthics is traditionally a department of philosophy, and that is my\nreason for discussing it. I hardly think myself that it ought to be\nincluded in the domain of philosophy, but to prove this would take as\nlong as to discuss the subject itself, and would be less interesting.\n\nAs a provisional definition, we may take ethics to consist of general\nprinciples which help to determine rules of conduct. It is not the\nbusiness of ethics to say how a person should act in such and such\nspecific circumstances; that is the province of casuistry. The\nword \u201ccasuistry\u201d has acquired bad connotations, as a result of the\nProtestant and Jansenist attacks on the Jesuits. But in its old and\nproper sense it represents a perfectly legitimate study. Take, say,\nthe question: In what circumstances is it right to tell a lie? Some\npeople, unthinkingly, would say: Never! But this answer cannot be\nseriously defended. Everybody admits that you should lie if you meet a\nhomicidal maniac pursuing a man with a view to murdering him, and he\nasks you whether the man has passed your way. It is admitted that lying\nis a legitimate branch of the art of warfare; also that priests may\nlie to guard the secrets of the confessional, and doctors to protect\nthe professional confidences of their patients. All such questions\nbelong to casuistry in the old sense, and it is evident that they are\nquestions deserving to be asked and answered. But they do not belong to\nethics in the sense in which this study has been included in philosophy.\n\nIt is not the business of ethics to arrive at actual rules of conduct,\nsuch as: \u201cThou shalt not steal\u201d. This is the province of morals. Ethics\nis expected to provide a basis from which such rules can be deduced.\nThe rules of morals differ according to the age, the race, and the\ncreed of the community concerned, to an extent that is hardly realised\nby those who have neither travelled nor studied anthropology. Even\nwithin a homogeneous community differences of opinion arise. Should a\nman kill his wife\u2019s lover? The Church says no, the law says no, and\ncommon sense says no; yet many people would say yes, and juries often\nrefuse to condemn. These doubtful cases arise when a moral rule is\nin process of changing. But ethics is concerned with something more\ngeneral than moral rules, and less subject to change. It is true that,\nin a given community, an ethic which does not lead to the moral rules\naccepted by that community is considered immoral. It does not, of\ncourse, follow that such an ethic is in fact false, since the moral\nrules of that community may be undesirable. Some tribes of head-hunters\nhold that no man should marry until he can bring to the wedding the\nhead of an enemy slain by himself. Those who question this moral\nrule are held to be encouraging licence and lowering the standard of\nmanliness. Nevertheless, we should not demand of an ethic that it\nshould justify the moral rules of head-hunters.\n\nPerhaps the best way to approach the subject of ethics is to ask what\nis meant when a person says: \u201cYou _ought_ to do so-and-so\u201d or \u201cI\n_ought_ to do so-and-so\u201d. Primarily, a sentence of this sort has an\nemotional content; it means \u201cthis is the act towards which I feel the\nemotion of approval\u201d. But we do not wish to leave the matter there; we\nwant to find something more objective and systematic and constant than\na personal emotion. The ethical teacher says: \u201cYou ought to approve\nacts of such-and-such kinds\u201d. He generally gives reasons for this view,\nand we have to examine what sorts of reasons are possible. We are here\non very ancient ground. Socrates was concerned mainly with ethics;\nPlato and Aristotle both discussed the subject at length; before their\ntime, Confucius and Buddha had each founded a religion consisting\nalmost entirely of ethical teaching, though in the case of Buddhism\nthere was afterwards a growth of theological doctrine. The views of\nthe ancients on ethics are better worth studying than their views on\n(say) physical science; the subject has not yet proved amenable to\nexact reasoning, and we cannot boast that the moderns have as yet\nrendered their predecessors obsolete.\n\nHistorically, virtue consisted at first of obedience to authority,\nwhether that of the gods, the government, or custom. Those who\ndisobeyed authority suffered obvious penalties. This is still the view\nof Hegel, to whom virtue consists in obedience to the State. There\nare, however, different forms of this theory, and the objections to\nthem are different. In its more primitive form, the theory is unaware\nthat different authorities take different views as to what constitutes\nvirtue, and it therefore universalises the practice of the community\nin which the theoriser lives. When other ages and nations are found to\nhave different customs, these are condemned as abominations. Let us\nconsider this view first.\n\nThe view we are now to examine is the theory that there are certain\nrules of conduct--_e.g._ the Decalogue--which determine virtue in all\nsituations. The person who keeps all the rules is perfectly virtuous;\nthe person who fails in this is wicked in proportion to the frequency\nof his failures. There are several objections to this as the basis\nof ethics. In the first place, the rules can hardly cover the whole\nfield of human conduct; _e.g._ there is nothing in the Decalogue to\nshow whether we ought to have a gold standard or not. Accordingly\nthose who hold this view regard some questions as \u201cmoral issues\u201d,\nwhile others have not this character. That means, in practice, that in\nregard to \u201cmoral issues\u201d we ought to act in a certain way, regardless\nof consequences, while in other matters we ought to consider which\ncourse will do the most good. Thus in effect we are driven to adopt two\ndifferent ethical systems, one where the code has spoken, the other\nwhere it is silent. This is unsatisfactory to a philosopher.\n\nThe second objection to such a view is suggested by the first. We all\nfeel that certain results are desirable, and others undesirable; but\na code of conduct which takes no account of circumstances will have\nsometimes the sort of consequences we think desirable, and sometimes\nthe sort we think undesirable. Take, _e.g._ the precept \u201cThou shalt\nnot kill\u201d. All respectable people hold that this does not apply when\nthe State orders a person to kill; on this ground among others, the\nNew York School Board recently refused to sanction the teaching of the\nDecalogue in schools.\n\nA third objection is that it may be asked how the moral rules are\nknown. The usual answer, historically, is that they are known by\nrevelation and tradition. But these are extra-philosophical sources of\nknowledge. The philosopher cannot but observe that there have been many\nrevelations, and that it is not clear why he should adopt one rather\nthan another. To this it may be replied that conscience is a personal\nrevelation to each individual, and invariably tells him what is right\nand what is wrong. The difficulty of this view is that conscience\nchanges from age to age. Most people nowadays consider it wrong to burn\na man alive for disagreeing with them in metaphysics, but formerly this\nwas held to be a highly meritorious act, provided it was done in the\ninterests of the right metaphysics. No one who has studied the history\nof moral ideas can regard conscience as invariably right. Thus we are\ndriven to abandon the attempt to define virtue by means of a set of\nrules of conduct.\n\nThere is, however, another form of the view that virtue consists in\nobedience to authority. This may be called \u201cthe administrator\u2019s ethic\u201d.\nA Roman or Anglo-Indian pro-consul would define virtue as obedience to\nthe moral code of the community to which a man happens to belong. No\nmatter how moral codes may differ, a man should always obey that of\nhis own time and place and creed. A Mohammedan, for instance, would\nnot be regarded as wicked for practising polygamy, but an Englishman\nwould, even if he lived in a Mohammedan country. This view makes social\nconformity the essence of virtue; or, as with Hegel, regards virtue as\nobedience to the government. The difficulty of such theories is that\nthey make it impossible to apply ethical predicates to authority: we\ncannot find any meaning for the statement that a custom is good or that\nthe government is bad. The view is appropriate to despots and their\nwilling slaves; it cannot survive in a progressive democracy.\n\nWe come a little nearer to a correct view when we define right conduct\nby the motive or state of mind of the agent. According to this theory,\nacts inspired by certain emotions are good, and those inspired by\ncertain other emotions are bad. Mystics hold this view, and have\naccordingly a certain contempt for the letter of the law. Broadly\nspeaking, it would be held that acts inspired by love are good, and\nthose inspired by hate are bad. In practice, I hold this view to be\nright; but philosophically I regard it as deducible from something more\nfundamental.\n\nAll the theories we have hitherto considered are opposed to those which\njudge the rightness or wrongness of conduct by its consequences. Of\nthese the most famous is the utilitarian philosophy, which maintained\nthat happiness is the good, and that we ought to act so as to maximise\nthe balance of happiness over unhappiness in the world. I should not\nmyself regard happiness as an adequate definition of the good, but I\nshould agree that conduct ought to be judged by its consequences. I do\nnot mean, of course, that in every practical exigency of daily life\nwe should attempt to think out the results of this or that line of\nconduct, because, if we did, the opportunity for action would often\nbe past before our calculations were finished. But I do mean that\nthe received moral code, in so far as it is taught in education and\nembodied in public opinion or the criminal law, should be carefully\nexamined in each generation, to see whether it still serves to achieve\ndesirable ends, and, if not, in what respects it needs to be amended.\nThe moral code, in short, like the legal code, should adapt itself to\nchanging circumstances, keeping the public good always as its motive.\nIf so, we have to consider in what the public good consists.\n\nAccording to this view, \u201cright conduct\u201d is not an autonomous concept,\nbut means \u201cconduct calculated to produce desirable results\u201d. It will be\nright, let us say, to act so as to make people happy and intelligent,\nbut wrong to act so as to make them unhappy and stupid. We have to\nask ourselves how we can discover what constitutes the ends of right\nconduct.\n\nThere is a view, advocated, _e.g._ by Dr. G.\u00a0E. Moore, that \u201cgood\u201d is\nan indefinable notion, and that we know _a priori_ certain general\npropositions about the kinds of things that are good on their own\naccount. Such things as happiness, knowledge, appreciation of beauty,\nare known to be good, according to Dr. Moore; it is also known that we\nought to act so as to create what is good and prevent what is bad. I\nformerly held this view myself, but I was led to abandon it, partly by\nMr. Santayana\u2019s _Winds of Doctrine_. I now think that good and bad are\nderivative from desire. I do not mean quite simply that the good is the\ndesired, because men\u2019s desires conflict, and \u201cgood\u201d is, to my mind,\nmainly a social concept, designed to find an issue from this conflict.\nThe conflict, however, is not only between the desires of different\nmen, but between incompatible desires of one man at different times,\nor even at the same time, and even if he is solitary, like Robinson\nCrusoe. Let us see how the concept of \u201cgood\u201d emerges from reflection or\nconflicts of desires.\n\nWe will begin with Robinson Crusoe. In him there will be conflicts,\nfor example, between fatigue and hunger, particularly between fatigue\nat one time and foreseen hunger at another. The effort which he will\nrequire in order to work when he is tired with a view to providing\nfood on another occasion has all the characteristics of what is called\na moral effort: we think better of a man who makes the effort than of\none who does not, and the making of it requires self-control. For some\nreason, this sort of thing is called, not morals, but \u201cmorale\u201d; the\ndistinction, however, seems to me illusory. Robinson Crusoe is bound\nto realise that he has many desires, each of which is stronger at one\ntime than at another, and that, if he acts always upon the one that\nis strongest at the moment, he may defeat others that are stronger in\nthe long run. So far, only intelligence is involved; but one may assume\nthat, with the progress of intelligence, there goes a growing desire\nfor a harmonious life, _i.e._ a life in which action is dominated by\nconsistent quasi-permanent desires. Again: some desires, in addition to\nthe desire for a harmonious life, are more likely to lead to harmony\nthen certain other desires. Intellectual curiosity, _e.g._ affords a\nmild diffused satisfaction, whereas drugs provide ecstasy followed by\ndespair. If we arrive unexpectedly in Robinson Crusoe\u2019s island and find\nhim studying botany, we shall think better of him than if we find him\ndead drunk on his last bottle of whisky. All this belongs to morals,\nalthough it is purely self-regarding.\n\nWhen we come to considering men in society, moral questions become\nboth more important and more difficult, because conflicts between\nthe desires of different persons are harder to resolve than internal\nconflicts among the desires of one person. There are some distinctions\nto be made. First, there is the difference between the point of view\nof the neutral authority contemplating a squabble in which it is not\ninterested, and the point of view of the disputants themselves. Then\nthere is the distinction between what we wish people to _do_, and what\nwe wish them to _feel_ in the way of emotions and desires.\n\nThe view of authority everywhere is that squabbles to which it is not a\nparty are undesirable, but that in the squabbles to which it is a party\nvirtue consists in promoting the victory of authority. In the latter\nrespect, it is acting, not as an authority, but merely as a combination\nof quarrelsome individuals who think it more profitable to quarrel\nwith outsiders than with each other; we will therefore ignore this\naspect of authority, and consider its action only when it is a neutral.\nIn this case, it aims at preventing quarrels by punishing those who\nbegin them, or sometimes by punishing both parties. Monsieur Huc, the\nJesuit missionary who wrote a fascinating account of his travels in\nChina, Tartary, and Tibet about eighty years ago, relates an amusing\nconversation he had with a mandarin. Monsieur Huc had remarked that\nChinese justice was dilatory, expensive, and corrupt. The mandarin\nexplained that it had been made so in obedience to an Imperial edict,\nsetting forth that the subjects of the Son of Heaven had become too\nmuch addicted to litigation, and must be led to abandon this practice.\nThe rescript then proceeded to suggest to magistrates and judges\nthe desirability of the above defects as a means of diminishing the\nnumber of law-suits. It appeared that the Emperor\u2019s commands had been\nfaithfully obeyed in this respect--more so than in some others.\n\nAnother method adopted by public authorities to prevent the impulse\ntowards internal quarrels is the creation of _esprit de corps_, public\nspirit, patriotism, etc., _i.e._ a concentration of quarrelsome\nimpulses or persons outside the group over which it rules. Such a\nmethod, obviously, is partial and external; it would not be open to a\nworld-wide democratic authority, should this ever come into existence.\nSuch an authority would have to adopt better methods of producing\nharmony; it would also have a higher claim to the obedience of citizens\nthan some authorities have at present.\n\nWhat can we say from the point of view of the disputants themselves? It\nis of course obvious that there will be a greater _total_ satisfaction\nwhen two people\u2019s desires harmonise than when they conflict, but that\nis not an argument which can be used to people who in fact hate each\nother. One can argue that the one who is going to be beaten would do\nwell to give way, but each will think that he himself is going to be\nvictorious. One can argue that there is more happiness to be derived\nfrom love than from hate, but people cannot love to order, and there is\nno satisfaction to be derived from an insincere love. Nor is it always\ntrue in an individual case that love brings more happiness than hate.\nDuring and immediately after the war, those who hated the Germans were\nhappier than those who still regarded them as human beings, because\nthey could feel that what was being done served a good purpose. I\nthink, therefore, that certain departments of morals, and those the\nmost important, cannot be inculcated from a personal point of view, but\nonly from the point of view of a neutral authority. That is why I said\nthat ethics is mainly social.\n\nThe attitude of a neutral authority would, it seems to me, be this:\nMen desire all sorts of things, and in themselves all desires,\ntaken singly, are on a level, _i.e._ there is no reason to prefer\nthe satisfaction of one to the satisfaction of another. But when we\nconsider not a single desire but a group of desires, there is this\ndifference, that sometimes all the desires in a group can be satisfied,\nwhereas in other cases the satisfaction of some of the desires in\nthe group is incompatible with that of others. If A and B desire to\nmarry each other, both can have what they want, but if they desire to\nkill each other, at most one can succeed, unless they are Kilkenny\ncats. Therefore the former pair of desires is socially preferable to\nthe latter. Now our desires are a product of three factors: native\ndisposition, education, and present circumstances. The first factor is\ndifficult to deal with at present, for lack of knowledge. The third is\nbrought into operation by means of the criminal law, economic motives,\nand social praise and blame, which make it on the whole to the interest\nof an individual in a community to promote the interests of the\ndominant group in that community. But this is done in an external way,\nnot by creating good desires, but by producing a conflict of greed and\nfear in which it is hoped that fear will win. The really vital method\nis education, in the large sense in which it includes care of the body\nand habit-formation in the first few years. By means of education,\nmen\u2019s desires can be changed, so that they act spontaneously in a\nsocial fashion. To force a man to curb his desires, as we do by the\ncriminal law, is not nearly so satisfactory as to cause him genuinely\nto feel the desires which promote socially harmonious conduct.\n\nAnd this brings me to the last point with which we are concerned,\nnamely, the distinction between feeling and doing. No doubt, from a\nsocial point of view the important thing is what a man does, but it is\nimpossible to cause a man to do the right things consistently unless he\nhas the right desires. And the right desires cannot be produced merely\nby praising them or by desiring to have them; the technique of moral\neducation is not one of exhortation or _explicit_ moral instruction.\n\nWe can now state the ethic at which we have arrived in abstract terms.\nPrimarily, we call something \u201cgood\u201d when we desire it, and \u201cbad\u201d when\nwe have an aversion from it. But our use of words is more constant than\nour desires, and therefore we shall continue to call a thing good even\nat moments when we are not actually desiring it, just as we always\ncall grass green though it sometimes looks yellow. And the laudatory\nassociations of the word \u201cgood\u201d may generate a desire which would not\notherwise exist: we may want to eat caviare merely because we are told\nthat it is good. Moreover the use of words is social, and therefore\nwe learn only to call a thing good, except in rare circumstances, if\nmost of the people we associate with are also willing to call it good.\nThus \u201cgood\u201d comes to apply to things desired by the whole of a social\ngroup. It is evident, therefore, that there can be more good in a world\nwhere the desires of different individuals harmonise than in one where\nthey conflict. The supreme moral rule should, therefore, be: _Act so\nas to produce harmonious rather than discordant desires_. This rule\nwill apply wherever a man\u2019s influence extends: within himself, in his\nfamily, his city, his country, even the world as a whole, if he is able\nto influence it.\n\nThere will be two main methods to this end: first, to produce social\ninstitutions under which the interests of different individuals or\ngroups conflict as little as possible; second, to educate individuals\nin such a way that their desires can be harmonised with each other and\nwith the desires of their neighbours. As to the first method, I shall\nsay nothing further, since the questions that arise belong to politics\nand economics. As to the second, the important period is the formative\nperiod of childhood, during which there should be health, happiness,\nfreedom, and a gradual growth of self-discipline through opportunities\nfor difficult achievement of a sort which is useful and yet satisfies\nthe impulse towards mastery of the environment. The desire for power,\nwhich is present in most people and strongest in the most vigorous,\nshould be directed towards power over things rather than over people.\n\nIt is clear that, if harmonious desires are what we should seek, love\nis better than hate, since, when two people love each other, both\ncan be satisfied, whereas when they hate each other one at most can\nachieve the object of his desire. It is obvious also that desire for\nknowledge is to be encouraged, since the knowledge that a man acquires\nis not obtained by taking it away from some one else; but a desire for\n(say) large landed estates can only be satisfied in a small minority.\nDesire for power over other people is a potent source of conflict,\nand is therefore to be discouraged; a respect for the liberty of\nothers is one of the things that ought to be developed by the right\nkind of education. The impulse towards personal achievement ought to\ngo into such things as artistic creation or scientific discovery or\nthe promotion of useful institutions--in a word, into activities that\nare creative rather than possessive. Knowledge, which may do positive\nharm where men\u2019s desires conflict (for example, by showing how to make\nwar more deadly), will have only good results in a world where men\u2019s\ndesires harmonise, since it tends to show how their common desires are\nto be realised.\n\nThe conclusion may be summed up in a single phrase: _The good life is\none inspired by love and guided by knowledge_.[11]\n\n     [11] Cf. _What I Believe_, by the present author--To-day and\n          To-morrow Series.\n\n\n\n\nPART IV\n\nTHE UNIVERSE\n\n\n\n\nCHAPTER XXIII\n\nSOME GREAT PHILOSOPHIES OF THE PAST\n\n\nOur discussions, hitherto, have been concerned very largely with\nMan, but Man on his own account is not the true subject-matter of\nphilosophy. What concerns philosophy is the universe as a whole; Man\ndemands consideration solely as the instrument by means of which we\nacquire knowledge of the universe. And that is why it is human beings\nas capable of knowledge that have concerned us mainly in past chapters,\nrather than as centres of will or of emotion. We are not in the mood\nproper to philosophy so long as we are interested in the world only as\nit affects human beings; the philosophic spirit demands an interest in\nthe world for its own sake. But since we apprehend the world through\nour own senses, and think about it with our own intellect, the picture\nthat we acquire is inevitably coloured by the personal medium through\nwhich it comes to us. Consequently we have to study this medium,\nnamely ourselves, in order to find out, if we can, what elements in\nour picture of the world are contributed by us, and what elements we\nmay accept as representative of outside fact. Previous chapters have\nstudied cognition, both as an outwardly observable reaction, and as\nit appears to introspection. In the chapters that remain, we shall\nbe concerned with what we can know about the universe, in view of\nthe nature of the instrument that we have to employ. I do not think\nwe can know as much as many philosophers of the past have supposed,\nbut I think it is worth while to have in our minds an outline of\ntheir systems. I shall therefore begin by setting forth a few typical\nphilosophical constructions of earlier centuries.\n\nModern philosophy is generally taken as beginning with Descartes,\nwho flourished in the first half of the seventeenth century. We have\nalready had occasion, in Chapter XVI, to consider his argument \u201cI\nthink, therefore, I am\u201d, but now we will deal with him somewhat more\ngenerally. He inaugurated two movements, one in metaphysics, one in\ntheory of knowledge. In metaphysics, he emphasised the gulf between\nmind and matter, or between soul and body; in theory of knowledge he\nadvocated a critical scrutiny of premises. These two movements had\ndifferent histories, each of them interesting. The science of dynamics\nwas rapidly developing in Descartes\u2019 time, and seemed to show that the\nmotions of matter could be calculated mathematically, given sufficient\ndata. As the motions of matter include our bodily acts, even speaking\nand writing, it seemed as if the consequence must be a materialistic\ntheory of human behaviour. This consequence, however, was distasteful\nto most philosophers, and they therefore invented various ways of\nescaping from it. Descartes himself thought that the will could have\ncertain direct physical effects. He thought that the brain contains a\nfluid called the \u201canimal spirits\u201d, and that the will could influence\nthe direction of its motion, though not the velocity. In this way he\nwas still able to hold that the will is effective in the manner in\nwhich common sense supposes it to be. But this view did not fit in at\nall well with the rest of his philosophy. He held that, apart from the\nSupreme Substance, namely God, there are two created substances, mind\nand matter; that the essence of mind is thought, and that the essence\nof matter is extension. He made these two substances so different\nthat interaction between them became difficult to understand, and his\nfollowers decided that there is never any effect either of mind on\nmatter or of matter on mind.\n\nThe motives for this development were various; perhaps the most\nimportant was the development of physics immediately after Descartes\u2019\ntime. A law was discovered called the \u201cconservation of momentum\u201d. This\nstates that, if a system of bodies is in any sort of motion, and is\nfree from outside influences, the amount of motion _in any direction_\nis constant. This showed that the kind of action of the will on the\n\u201canimal spirits\u201d which Descartes had assumed was contrary to the\nprinciples of dynamics. It seemed to follow that mind cannot influence\nmatter, and it was inferred that matter cannot influence mind, since\nthe two were regarded as co-equal substances. It was held that each\ngoes it own way, according to its own laws. The fact that our arm moves\nwhen we will it to move was regarded as analogous to the fact that two\nperfectly accurate clocks strike at the same moment, though neither\nhas any effect upon the other. The series of mental events and the\nseries of physical events were parallel, each going at the same rate as\nthe other, therefore they continued to synchronise, in spite of their\nindependence of each other.\n\nSpinoza sought to make this parallelism less mysterious by denying\nthat there are two separate substances, mind and matter. He maintained\nthat there is only one substance, of which thought and extension are\nattributes. But there seemed still no good reason why the events\nbelonging to the two attributes should develop along parallel lines.\nSpinoza is in many ways one of the greatest philosophers, but his\ngreatness is rather ethical than metaphysical. Accordingly he was\nregarded by contemporaries as a profound metaphysician but a very\nwicked man.\n\nThe notion of the impossibility of interaction between mind and body\nhas persisted down to our own day. One still hears of \u201cpsychophysical\nparallelism\u201d, according to which to every state of the brain a state of\nmind _corresponds_ and vice versa, without either acting on the other.\nThis whole point of view, though not exactly that of Descartes, derives\nfrom him. It has a number of sources, religious, metaphysical, and\nscientific; but there seems no ground whatever for regarding it as true.\n\nTake, first, the rigid determinism of traditional physics, which was\nto have been avoided. Spinoza rightly perceived that this could not\nbe avoided by such methods, and therefore accepted determinism in\nthe psychical as in the physical realm. If everything we _say_ is\ndetermined by physical causes, our thoughts are only free when we\ntell lies: so long as we say what we think, our thoughts also can be\ninferred from physics. The philosophy which I advocate escapes this\nconsequence in several ways. In the first place, causality does not\ninvolve _compulsion_, but only a law of sequence: if physical and\nmental events run parallel, either may with equal justice be regarded\nas causing the other, and there is no sense in speaking of them as\ncausally independent. Thus the Cartesian dualism does not have the\npleasant consequences which were intended. In the second place, modern\nphysics has become less deterministic than the physics of the past\nfew centuries. We do not know, _e.g._ what makes a radio-active atom\nexplode or an electron jump from a larger to a smaller orbit. In these\nmatters we only know statistical averages.\n\nTake next the view that mind and matter are quite disparate. This we\nhave criticised already. It rests upon a notion that we know much\nmore about matter than we do, and in particular upon the belief that\nthe space of physics can be identified with the space of sensible\nexperience. This belief is absent in Leibniz, who, however, never quite\nrealised what his own view was. It is not absent in Kant, who realised\nthat the space of sensible experience is subjective, and inferred\nthat the space of physics is subjective. Since Kant, no one seems to\nhave thought clearly about space until Einstein and Minkowski. The\nseparation of physical and sensible space, logically carried out, shows\nthe groundlessness of traditional views about mind and matter. This\npart of Descartes\u2019 philosophy, therefore, though it accelerated the\nprogress of physics, must be regarded as metaphysically an aberration.\n\nThe other part of Descartes\u2019 philosophy, namely, the emphasis upon\nmethodical doubt, and consequently upon theory of knowledge, has\nbeen more fruitful. The beginning of a philosophic attitude is the\nrealisation that we do not know as much as we think we do, and to this\nDescartes contributed notably. We have seen that he set to work to\ndoubt all he could, but found he could not doubt his own existence,\nwhich he therefore took as the starting-point of his constructive\nsystem. He supposed that the most certain fact in the world is \u201cI\nthink\u201d. This was unfortunate, since it gave a subjective bias to modern\nphilosophy. As a matter of fact, \u201cI\u201d seems to be only a string of\nevents, each of which separately is more certain than the whole. And\n\u201cthink\u201d is a word which Descartes accepted as indefinable, but which\nreally covers complicated relations between events. When is an event\na \u201cthought\u201d? Is there some intrinsic characteristic which makes it a\nthought? Descartes would say yes, and so would most philosophers. I\nshould say no. Take, _e.g._ a visual and an auditory sensation. Both\nare \u201cthoughts\u201d in Descartes\u2019 sense, but what have they in common? Two\nvisual sensations _have_ an indefinable common quality, viz. that\nwhich makes them visual. Two auditory sensations likewise. But a\nvisual and an auditory sensation have in common, if I am not mistaken,\nno intrinsic property, but a certain capacity for being _known_\nwithout inference. This amounts to saying that they are mnemic causes\nof a certain kind of event, called a cognition, and that they have\nmoreover, a certain formal similarity to the cognition which they\ncause. Therefore, instead of taking the general \u201cI think\u201d as our basis,\nwe ought to take the particular occurrences which are known without\ninference, among which sensations (or rather \u201cperceptions\u201d) will be\nincluded. These occurrences, as we have already seen, may be regarded\nwith equal justice as physical and mental: they are parts of chains of\nphysical causation, and they have mnemic effects which are cognitions.\nThe former fact makes us call them physical, the latter mental, both\nquite truly. It is the particular events which are certain, not the\n\u201cI think\u201d which Descartes made the basis of his philosophy. It is not\ncorrect to regard the ultimate certainties as \u201csubjective\u201d, except in\nthe sense that they are events in that part of space-time in which our\nbody is--and our mind also, I should say.\n\nA new turn was given to the Cartesian type of metaphysics by Leibniz\n(1646\u20131716), who, like Descartes, was supremely eminent both in\nmathematics and in philosophy. Leibniz rejected the view that there\nis only one substance, as Spinoza held, or only two other than God,\nas the orthodox followers of Descartes maintained. He also rejected\nthe dualism of mind and matter, holding that there are innumerable\nsubstances all in a greater or less degree mental, and none in any\ndegree material. He maintained that every substance is immortal, and\nthat there is no interaction between one substance and another--this\nlast being a view derived from the Cartesian independence of mind\nand matter. He also extended to his many substances the belief in\nparallelism which had existed for the two substances of the Cartesians.\nHe called his substances \u201cmonads\u201d, and maintained that every monad\nmirrors the universe, and develops along lines which correspond, point\nby point, with those along which every other monad is developing. A\nman\u2019s soul or mind is a single monad, while his body is a collection of\nmonads, each mental in some degree, but less so than the monad which\nis his soul. Inferior monads mirror the world in a more confused way\nthan higher ones do, but there is some element of confusion in the\nperceptions of even the most superior monads. Every monad mirrors the\nworld from its own point of view, and the difference between points of\nview is compared to a difference of perspective. \u201cMatter\u201d is a confused\nway of perceiving a number of monads; if we perceived clearly, we\nshould see that there is no such thing as matter.\n\nLeibniz\u2019s system had great merits and great demerits. The theory that\n\u201cmatter\u201d is a confused way of perceiving something non-material was\nan advance upon anything to be found in his predecessors. He had,\nthough only semi-consciously, the distinction between physical and\nperceptual space: there is space in each monad\u2019s picture of the world,\nand there is also the assemblage or pattern of \u201cpoints of view\u201d. The\nlatter corresponds to what I have called \u201cphysical space\u201d, the former\nto \u201cperceptual space\u201d. Leibniz maintained, as against Newton, that\nspace and time consists only of relations--a view which has achieved a\ndefinitive triumph in Einstein\u2019s theory of relativity. The weak point\nof his system was what he called the \u201cpre-established harmony\u201d, in\nvirtue of which all the monads (so to speak) kept step, in spite of\nthe fact that they were \u201cwindowless\u201d and never acted upon each other.\nPerception, for Leibniz, was not an effect of the object perceived, but\na modification arising in the perceiving monad and running parallel\nwith what was happening in the perceived object. This view would\nnever have seemed plausible but for the anterior Cartesian theory of\nthe mutual independence of mind and matter. And if Leibniz himself\ndeveloped, as he believed, in complete independence of all other\ncreated things, it is not clear what good reasons he could have had for\nbelieving in the existence of anything except himself, since, by his\nown theory, his experiences would remain unchanged if everything else\nwere annihilated. In fact, he was only able to refute this possibility\nby bringing in theological considerations, which, whether valid or\nnot, are out of place in philosophy. For this reason, his doctrines,\ningenious as they were, found little acceptance in France and England,\nthough in Germany they prevailed, in a modified form, until the time of\nKant.\n\nThe systems of Descartes, Spinoza and Leibniz have one very important\ncharacteristic in common, namely, that they all depend upon the\ncategory of \u201csubstance\u201d. This is a concept which has developed out of\nthe common-sense notion of \u201cthing\u201d. A \u201csubstance\u201d is that which has\nqualities, and is in general supposed to be indestructible, though\nit is difficult to see why. It acquired its hold over metaphysicians\npartly because both matter and the soul were held to be immortal, and\npartly through a hasty transference to reality of ideas derived from\ngrammar. We say \u201cPeter is running\u201d, \u201cPeter is talking\u201d, \u201cPeter is\neating\u201d, and so on. We think that there is one entity, Peter, who does\nall these things, and that none of them could be done unless there were\nsomeone to do them, but that Peter might quite well do none of them.\nSimilarly we assign qualities to Peter: we say he is wise and tall\nand blond and so on. All these qualities, we feel, cannot subsist by\nthemselves in the void, but only when there is a subject to which they\nbelong; but Peter would remain Peter even if he became foolish and\nshort and dyed his hair. Thus Peter, who is regarded as a \u201csubstance\u201d,\nis self-subsistent as compared with his qualities and states, and he\npreserves his substantial identity throughout all sorts of changes.\nSimilarly in the material world an atom is supposed (or rather was\nsupposed until recently) to preserve its identity throughout all time,\nhowever it might move and whatever combinations it might form with\nother atoms. The concept of \u201cmotion\u201d, upon which all physics seemed to\ndepend, was only strictly applicable to a substance which preserves\nits identity while changing its spatial relations to other substances;\nthus \u201csubstance\u201d acquired an even firmer hold upon physics than upon\nmetaphysics.\n\nNevertheless, the notion of \u201csubstance\u201d, at any rate in any sense\ninvolving permanence, must be shut out from our thoughts if we are to\nachieve a philosophy in any way adequate either to modern physics or\nto modern psychology. Modern physics, both in the theory of relativity\nand in the Heisenberg-Schr\u00f6dinger theories of atomic structure, has\nreduced \u201cmatter\u201d to a system of events, each of which lasts only for a\nvery short time. To treat an electron or a proton as a single entity\nhas become as wrong-headed as it would be to treat the population of\nLondon or New York as a single entity. And in psychology, equally,\nthe \u201cego\u201d has disappeared as an ultimate conception, and the unity\nof a personality has become a peculiar causal nexus among a series\nof events. In this respect, grammar and ordinary language have been\nshown to be bad guides to metaphysics. A great book might be written\nshowing the influence of syntax on philosophy; in such a book, the\nauthor could trace in detail the influence of the subject-predicate\nstructure of sentences upon European thought, more particularly in\nthis matter of \u201csubstance\u201d. And it must be understood that the same\nreasons which lead to the rejection of substance lead also to the\nrejection of \u201cthings\u201d and \u201cpersons\u201d as ultimately valid concepts. I\nsay \u201cI sit at my table\u201d, but I ought to say: \u201cOne of a certain string\nof events causally connected in the sort of way that makes the whole\nseries that is called a \u2018person\u2019 has a certain spatial relation to one\nof another string of events causally connected with each other in a\ndifferent way and having a spatial configuration of the sort denoted\nby the word \u2018table\u2019\u201d. I do not say so, because life is too short; but\nthat is what I should say if I were a true philosopher. Apart from\nany other grounds, the inadequacy of the notion of \u201csubstance\u201d would\nlead us to regard the philosophy of Descartes, Spinoza, and Leibniz as\nincompatible with modern science. There is of course in all three, a\ngreat deal that does not depend upon \u201csubstance\u201d, and that still has\nvalue; but \u201csubstance\u201d supplied the framework and a good deal of the\nargumentation, and therefore introduces a fatal defect into these three\ngreat systems.\n\nI come now to the triad of British philosophers, Locke, Berkeley, and\nHume--English, Irish, and Scotch respectively. Perhaps from patriotic\nbias or from community of national temperament, I find more that I\ncan accept, and regard as still important, in the writings of these\nthree than in the philosophy of their continental predecessors. Their\nconstructions are less ambitious, their arguments more detailed, and\ntheir methods more empirical; in all these respects they show more\nkinship with the modern scientific outlook. On the other hand, Locke\nand Hume, if not Berkeley, approach philosophy too exclusively from\nthe side of psychology, and are concerned to study Man rather than the\nuniverse.\n\nLocke was a contemporary and friend of Newton; his great book, _An\nEssay concerning Human Understanding_, was published at almost the\nsame moment as Newton\u2019s _Principia_. His influence has been enormous,\ngreater, in fact, than his abilities would seem to warrant; and this\ninfluence was not only philosophical, but quite as much political and\nsocial. He was one of the creators of eighteenth century liberalism:\ndemocracy, religious toleration, freedom of economic enterprise,\neducational progress, all owe much to him. The English Revolution of\n1688 embodied his ideas; the American Revolution of 1776 and the\nFrench Revolution of 1789 expressed what had grown, in a century, out\nof his teaching. And in all these movements, philosophy and politics\nwent hand in hand. Thus the practical success of Locke\u2019s ideas has been\nextraordinary.\n\nWhen, knowing all this, one comes to read Locke himself, it is\ndifficult to resist a feeling of disappointment. He is sensible,\nenlightened, minute, but uninspired and (to moderns) uninspiring.\nOne has to remember that his contemporaries found common sense\nexhilarating after a century of wars of religion and a long struggle\nwith obscurantism. Locke combatted the doctrine of \u201cinnate ideas\u201d,\naccording to which we learned only certain things by experience,\nbut possessed our abstract knowledge in virtue of our congenital\nconstitution. He regarded the mind at birth as a wax tablet, upon which\nexperience proceeded to write. Undoubtedly he was, in this matter,\nmore in the right than his opponents, although the terms in which the\ncontroversy was waged are not such as a modern could employ. We should\nsay that the innate apparatus of man consists of \u201creflexes\u201d rather than\n\u201cideas\u201d; also that our sense-organs, our glands, and our muscles lead\nto responses of certain kinds, in which our own organisation plays a\npart of the same importance as that played by the external stimulus.\nThe element in our knowledge-responses that corresponds to our own\nbodily organisation might, perhaps, be regarded as representing what\nLocke\u2019s opponents meant by \u201cinnate\u201d. But it does not represent this\nat all accurately so far as our feelings towards it are concerned.\nThe \u201cinnate\u201d ideas were the ideas to be proud of; they embraced pure\nmathematics, natural theology, and ethics. But nobody is proud of\nsneezing or coughing. And when Locke tried to show, in detail, how our\nknowledge is generated by experience, he was liberating philosophy from\na great deal of useless lumber, even if his own doctrines were not\naltogether such as we can now accept.\n\nLocke used his own principles only in ways consistent with common\nsense; Berkeley and Hume both pushed them to paradoxical conclusions.\nThe philosophy of Berkeley, to my mind, has not received quite the\nattention and respect that it deserves--not that I agree with it,\nbut that I think it ingenious and harder to refute than is often\nsupposed. Berkeley, as everyone knows, denied the reality of matter,\nand maintained that everything is mental. In the former respect I\nagree with him, though not for his reasons; in the latter respect, I\nthink his argument unsound and his conclusion improbable, though not\ncertainly false. However, I will leave the development of my own views\nto a later chapter, and confine myself to Berkeley\u2019s argument.\n\nBerkeley contended that when, for example, you \u201csee a tree\u201d, all\nthat you really _know_ to be happening is in you, and is mental. The\ncolour that you see, as Locke had already argued, does not belong to\nthe physical world, but is an effect upon you, produced, according\nto Locke, by a physical stimulus. Locke held that the purely spatial\nproperties of perceived objects really belong to the objects, whereas\nsuch things as colour, softness, sound, etc., are effects in us.\nBerkeley went further, and argued that the spatial properties of\nperceived objects are no exception. Thus the object perceived is\ncomposed entirely of \u201cmental\u201d constituents, and there is no reason to\nbelieve in the existence of anything not mental. He did not wish to\nadmit that a tree ceases to exist when we do not look at it, so he\nmaintained that it acquires permanence through being an idea in the\nmind of God. It is still only an \u201cidea\u201d, but not one whose existence\ndepends upon the accidents of our perceptions.\n\nThe real objection to Berkeley\u2019s view is rather physical than\nmetaphysical. Light and sound take time to travel from their sources\nto the percipient, and one must suppose that something is happening\nalong the route by which they travel. What is happening along the route\nis presumably not \u201cmental\u201d, for, as we have seen, \u201cmental\u201d events are\nthose that have peculiar mnemic effects which are connected with living\ntissue. Therefore, although Berkeley is right in saying that the events\nwe know immediately are mental, it is highly probable that he is wrong\nas to the events which we infer in places where there are no living\nbodies. In saying this, however, we are anticipating the results of a\nfuller discussion in a later chapter.\n\nHume, proceeding from a starting-point essentially similar to that\nof Locke and Berkeley, arrived at conclusions so sceptical that all\nsubsequent philosophers have shied away from them. He denied the\nexistence of the Self, questioned the validity of induction, and\ndoubted whether causal laws could be applied to anything except our own\nmental processes. He is one of the very few philosophers not concerned\nto establish any positive conclusions. To a great extent, I think, we\nmust admit the validity of his reasons for refusing to feel the usual\ncertainties. As regards the Self, he was almost certainly right. As we\nhave already argued, a person is not a single entity, but a series of\nevents linked together by peculiar causal laws. As regards induction,\nthe question is very difficult, and I shall devote a subsequent chapter\nto it. As regards causal laws, the question, as we shall find later, is\nthe same as the question of induction. On both points Hume\u2019s doubts are\nnot to be lightly dismissed.\n\nThe usual modern criticism of Locke, Berkeley, and Hume is that they\nwere unduly \u201catomistic\u201d. They thought of the mind as a collection of\n\u201cideas\u201d, each as hard and separate as a billiard-ball. They had not\nthe conception of continuous change or of integral processes; their\ncausal units were too small. As we have already seen in connection with\n_Gestaltpsychologie_ and with sentences, the causal unit is often a\nconfiguration which cannot be broken up without losing its distinctive\ncausal properties. In this sense, it is true that the traditional\nBritish philosophy was too atomistic. But in another sense I do not\nthink it is true, and I think much modern philosophy is confused on\nthis point. Although a configuration may lose its _causal_ properties\nwhen broken up into its elements, it nevertheless does consist of\nthese elements related in certain ways; analysis into \u201catoms\u201d is\nperfectly valid, so long as it is not assumed that the causal efficacy\nof the whole is compounded out of the separate effects of the separate\natoms. It is because I hold this view that I call the philosophy which\nI advocate \u201clogical atomism\u201d. And to this extent I regard Locke,\nBerkeley, and Hume as in the right as against their modern critics. But\nthis also is a topic which will be resumed in a later chapter.\n\nHume\u2019s criticism of the notion of cause was what led Kant to his new\ndeparture. Kant\u2019s philosophy is difficult and obscure, and philosophers\nstill dispute as to what he meant. Those who disagree with him are held\nby his supporters to have misunderstood him; I must therefore warn the\nreader that what follows is my view of what he meant, and that there is\nno agreed view.\n\nKant maintained that, in virtue of our mental constitution, we deal\nwith the raw material of sense-impressions by means of certain\n\u201ccategories\u201d and by arranging it in space and time. Both the categories\nand the space-time arrangement are supplied by us, and do not belong to\nthe world except as known by us. But since our mental constitution is\na constant datum, all phenomena _as known_ will be spatio-temporal and\nwill conform to the categories. Among the latter \u201ccause\u201d is the most\nimportant. Thus although there may be no cause in the world as it is\nin itself (a point on which Kant was inconsistent in the interest of\nmorals), yet phenomena, _i.e._ things as they seem to us, will always\nhave other phenomena as their causes. And although there is no time in\nthe real world, things as they appear to us will be some earlier and\nsome later. Space, again is supplied by us, and therefore geometry can\nbe known _a priori_, without our having to study the outer world. Kant\nthought that Euclidean geometry was quite certainly true, although it\ncould not be proved by logic alone, since Euclid\u2019s axioms could be\ndenied without self-contradiction.\n\nIt was on this question of geometry that the weakness of Kant\u2019s\nsystem first became obvious. It was found that we have no grounds for\nregarding Euclidean geometry as quite true. Since Einstein, we have\npositive grounds for regarding it as not quite true. It appears that\ngeometry is just as empirical as geography. We depend upon observation\nif we want to know whether the sum of the angles of a triangle is two\nright angles just as much as if we want to know how much land there is\nin the western hemisphere.\n\nWith regard to the \u201ccategories\u201d there are equally great difficulties.\nLet us take \u201ccause\u201d as our illustration. We see lightning, and then we\nhear thunder; as phenomena, our seeing and hearing are connected as\ncause and effect. But we must not--if we are to take the subjectivity\nof \u201ccause\u201d seriously--suppose that our seeing or our hearing has an\noutside cause. In that case, we have no reason to suppose that there\nis anything outside ourselves. Nay, more: what _really_ happens when\nwe see is not, according to Kant, what we perceive by introspection;\nwhat really happens is something without a date, without a position\nin space, without causes and without effects. Thus we do not know\nourselves any better than we know the outside world. Space and time\nand the categories interpose a mirage of illusion which cannot be\npenetrated at any point. As an answer to Hume\u2019s scepticism, this\nseems a somewhat unsuccessful effort. And Kant himself, later, in the\n_Critique of Practical Reason_, demolished much of his own edifice,\nbecause he thought that ethics at least must have validity in the\n\u201creal\u201d world. This part of his philosophy, however, is usually ignored\nby his followers or apologetically minimised.\n\nKant gave a new turn to an old philosophical controversy, as to how\nfar our knowledge is _a priori_ and how far it is based on experience.\nKant admitted that without experience we could know nothing, and that\nwhat we know is only valid within the realm of experience. But he held\nthat the general framework of our knowledge is _a priori_ in the sense\nthat it is not proved by means of particular facts of experience, but\nrepresents the conditions to which phenomena have to conform in order\nto be capable of being experienced. Before his day, the tendency had\nbeen for continental philosophers to regard almost everything as\n_a priori_ while British philosophers regarded almost everything as\nempirical. But both sides thought that what is _a priori_ can be proved\nby logic, at least in theory, whereas Kant held that mathematics is\n_a priori_ and yet \u201csynthetic\u201d, _i.e._ not capable of being proved\nby logic. In this he was misled by geometry. Euclidean geometry,\nconsidered as true, is \u201csynthetic\u201d but not _a priori_; considered\nmerely as deducing consequences from premisses, it is _a priori_\nbut not \u201csynthetic\u201d. The geometry of the actual world, as required\nby engineers, is empirical; the geometry of pure mathematics, which\ndoes not inquire into the truth of the axioms but merely shows their\nimplications, is an exercise in pure logic.\n\nIt should be said, however, that, if the correct analysis of knowledge\nbears any resemblance at all to that which has been suggested in this\nbook, the whole controversy between empiricists and apriorists becomes\nmore or less unreal. All beliefs are _caused_ by external stimuli; when\nthey are as particular as the stimuli they are of the sort which an\nempiricist might regard as _proved_ by experience, but when they are\nmore general difficulties arise. A foreigner arrives in America and\nsees the immigration officials, who lead him to the generalisation that\nall Americans are rude; but a few minutes later the porter upsets this\ninduction in the hope of a tip. Thus sometimes a given belief will be\ncaused by one event and destroyed by another. If all the events in a\nman\u2019s life, so far as they affect the belief in question, are such as\nto cause it, he counts the belief true. The more general a belief is,\nthe more events are relevant to it, and therefore the more difficult\nit is for it to be such as a man will long consider true. Roughly\nspeaking, the beliefs which count as _a priori_ will be those which\nwell might have been upset by subsequent events, but in fact were\nconfirmed. Here as elsewhere we are driven to the view that theory of\nknowledge is not so fundamental as it has been considered since Kant.\n\nThere is one more traditional controversy which I wish to consider,\nnamely, that between monists and pluralists. Is the universe one,\nor is it many? If many, how intimately are they interconnected? The\nmonistic view is very old: it is already complete in Parmenides (fifth\ncentury B.C.). It is fully developed in Spinoza, Hegel, and Bradley.\nThe pluralistic view, on the other hand, is found in Heraclitus,\nthe atomists, Leibniz, and the British empiricists. For the sake of\ndefiniteness, let us take the monistic view as found in Bradley, who\nis in the main a follower of Hegel. He maintains that every judgment\nconsists in assigning a predicate to Reality as a whole: the whole is\nthe subject of every predicate. Suppose you start by saying \u201cTommy\nhas a cold in the head\u201d. This may not _seem_ to be a statement about\nthe universe as a whole, but according to Bradley it is. If I may\nbe allowed to set forth his argument in popular language which his\nfollowers might resent, I should put it something like this: First of\nall, who is Tommy? He is a person with a certain nature, distinguished\nfrom other persons by that nature; he may resemble others in many\nrespects, but not in all, so that you cannot really explain who Tommy\nis unless you set forth all his characteristics. But when you try to\ndo this, you are taken beyond Tommy: he is characterised by relations\nto his environment. He is affectionate or rebellious or thirsty, noisy\nor quiet, and so on; all of these qualities involve his relations\nto others. If you try to define Tommy without mentioning anything\noutside him, you will find this quite impossible; therefore he is not a\nself-subsistent being, but an unsubstantial fragment of the world. The\nsame thing applies even more obviously to his nose and his cold. How do\nyou know he has a cold? Because material substances of a certain kind\npass from his nose to his handkerchief, which would not be possible\nif he alone existed. But now, when you take in the environment with a\nview to defining Tommy and his nose and his cold, you find that you\ncannot define his immediate environment without taking account of its\nenvironment, and so on, until at last you have been forced to include\nthe whole world. Therefore Tommy\u2019s cold is in reality a property of\nthe world, since nothing short of the world is sufficiently substantial\nto have properties.\n\nWe may put the argument in a more abstract form. Everything which is\npart of the world is constituted, in part, by its relations to other\nthings; but relations cannot be real. Bradley\u2019s argument against\nrelations is as follows. First he argues that, if there are relations,\nthere must be qualities between which they hold. This part of the\nargument need not detain us. He then proceeds:\n\n\u201cBut how the relation can stand to the qualities is, on the other\nside, unintelligible. If it is nothing to the qualities, then they\nare not related at all; and, if so, as we saw, they have ceased to\nbe qualities, and their relation is a nonentity. But if it is to be\nsomething to them, then clearly we shall require a _new_-connecting\nrelation. For the relation hardly can be the mere adjective of one or\nboth of its terms; or, at least, as such it seems indefensible. And,\nbeing something itself, if it does not itself bear a relation to the\nterms, in what intelligible way will it succeed in being anything to\nthem? But here again we are hurried off into the eddy of a hopeless\nprocess, since we are forced to go on finding new relations without\nend. The links are united by a link, and this bond of union is a link\nwhich also has two ends; and these require each a fresh link to connect\nthem with the old. The problem is to find how the relation can stand to\nits qualities, and this problem is insoluble.\u201d\n\nI cannot deal adequately with this argument without abstruse\ntechnicalities which would be out of place. I will, however, point out\nwhat seems to me the essential error. Bradley conceives a relation as\nsomething just as substantial as its terms, and not radically different\nin kind. The analogy of the chain with links should make us suspicious,\nsince it clearly proves, if it is valid, that chains are impossible,\nand yet, as a fact, they exist. There is not a word in his argument\nwhich would not apply to physical chains. The successive links are\nunited not by another link, but by a spatial relation. I think Bradley\nhas been misled, unconsciously, by a circumstance to which I alluded\nin an earlier chapter, namely, the fact that the _word_ for a relation\nis as substantial as the _words_ for its terms. Suppose A and B are\ntwo events, and A precedes B. In the proposition \u201cA precedes B\u201d, the\nword \u201cprecedes\u201d is just as substantial as the words \u201cA\u201d and \u201cB\u201d. The\nrelation of the _two_ events A and B is represented, in language, by\nthe time or space order of the _three_ words \u201cA\u201d, \u201cprecedes\u201d, and\n\u201cB\u201d. But this order is an actual relation, not a word for relation.\nThe first step in Bradley\u2019s regress does actually have to be taken in\ngiving verbal expression to a relation, and the word for a relation\ndoes have to be related to the words for its terms. But this is a\nlinguistic, not a metaphysical, fact, and the regress does not have to\ngo any further.\n\nIt should be added that, as Bradley himself recognises, his\ndifficulties break out afresh when he comes to consider the relation of\nsubject and predicate when a character is assigned to Reality, and that\nhe is therefore compelled to conclude that _no_ truth is quite true.\nA conclusion of this sort, based upon an extremely abstract argument,\nmakes it natural to suspect that there is some error in the argument.\n\nPluralism is the view of science and common sense, and is therefore\nto be accepted if the arguments against it are not conclusive. For\nmy part, I have no doubt whatever that it is the true view, and that\nmonism is derived from a faulty logic inspired by mysticism. This logic\ndominates the philosophy of Hegel and his followers; it is also the\nessential basis of Bergson\u2019s system, although it is seldom mentioned in\nhis writings. When it is rejected, ambitious metaphysical systems such\nas those of the past are seen to be impossible.\n\n\n\n\nCHAPTER XXIV\n\nTRUTH AND FALSEHOOD\n\n\nThe question of truth and falsehood has been wrapped in unnecessary\nmystery owing to a number of causes. In the first place, people wish to\nthink that their beliefs are more apt to be true than false, so that\nthey seek a theory that will show that truth is normal and falsehood\nmore or less accidental. In the second place, people are very vague as\nto what they mean by \u201cbelief\u201d or \u201cjudgment\u201d, though persuaded that they\nknow beliefs or judgments to be the objects to which the predicates\n\u201ctrue\u201d or \u201cfalse\u201d apply. In the third place, there is a tendency to\nuse \u201ctruth\u201d with a big T in the grand sense, as something noble and\nsplendid and worthy of adoration. This gets people into a frame of mind\nin which they become unable to think. But just as the grave-diggers in\n_Hamlet_ became familiar with skulls, so logicians become familiar with\ntruth. \u201cThe hand of little employment hath the daintier sense,\u201d says\nHamlet. Therefore it is not from the logician that awe before truth is\nto be expected.\n\nThere are two questions in our present subject: (1) What are the\nobjects to which the predicates \u201ctrue\u201d and \u201cfalse\u201d apply? (2) What is\nthe difference between such as are true and such as are false? We will\nbegin with the first of these questions.\n\n_Prima facie_, \u201ctrue\u201d and \u201cfalse\u201d apply to statements, whether in\nspeech or in writing. By extension, they are supposed to apply to the\nbeliefs expressed in those statements, and also to hypotheses which are\nentertained without being believed or disbelieved. But let us first\nconsider the truth and falsehood of statements, following our practice\nof going as far as we can with the behaviourists before falling back on\nintrospection. We considered the meaning of words earlier; now we have\nto consider sentences. Of course a sentence may consist of a single\nword, or of a wink; but generally it consists of several words. In\nthat case, it has a meaning which is a function of the meanings of the\nseparate words and their order. A sentence which has no meaning is not\ntrue or false; thus it is only sentences as vehicles of a certain sort\nof meaning that have truth or falsehood. We have therefore to examine\nthe meaning of a sentence.\n\nLet us take some very humble example. Suppose you look in a time-table\nand find it there stated that a passenger train leaves King\u2019s Cross for\nEdinburgh at 10 A.M. What is the meaning of this assertion? I shudder\nwhen I think of its complexity. If I were to try to develop the theme\nadequately, I should be occupied with nothing else till the end of\nthe present volume, and then I should have only touched the fringe of\nthe subject. Take first the social aspect: it is not essential that\nanybody but the engineer and fireman should travel by the train, though\nit is essential that others should be _able_ to travel by it if they\nfulfil certain conditions. It is not essential that the train should\nreach Edinburgh: the statement remains true if there is an accident or\nbreakdown on the way. But it is essential that the management of the\nrailway should intend it to reach Edinburgh. Take next the physical\naspect: it is not essential, or even possible, that the train should\nstart _exactly_ at ten; one might perhaps say that it must not start\nmore than ten seconds before its time or more than fifty seconds after,\nbut these limits cannot be laid down rigidly. In countries where\nunpunctuality is common they would be much wider. Then we must consider\nwhat we mean by \u201cstarting\u201d, which no one can define unless he has\nlearnt the infinitesimal calculus. Then we consider the definitions of\nKing\u2019s Cross and Edinburgh, both of which are more or less vague terms.\nThen we must consider what is meant by a \u201ctrain\u201d. Here there will be\nfirst of all complicated legal questions; what constitutes fulfilment\nof a railway company\u2019s obligations in the way of running \u201ctrains\u201d? Then\nthere are questions as to the constitution of matter, since evidently\na train is a piece of matter; also of course there are questions as\nto methods of estimating Greenwich time at King\u2019s Cross. Most of the\nabove points have to do with the meaning of single words, not with the\nmeaning of the whole sentence. It is obvious that the ordinary mortal\ndoes not trouble about such complications when he uses the words: to\nhim a word has a meaning very far from precise, and he does not try to\nexclude marginal cases. It is the search for precision that introduces\ncomplications. We think we attach a meaning to the word \u201cman\u201d, but we\ndon\u2019t know whether to include Pithecanthropus Erectus. To this extent,\nthe meaning of the word is vague.\n\nAs knowledge increases, words acquire meanings which are more precise\nand more complex; new words have to be introduced to express the less\ncomplex constituents which have been discovered. A word is intended to\ndescribe something in the world; at first it does so very badly, but\nafterwards it gradually improves. Thus single words embody knowledge,\nalthough they do not make assertions.\n\nIn an ideal logical language, there will be words of different kinds.\nFirst, proper names. Of these, however, there are no examples in\nactual language. The words which are _called_ proper names describe\ncollections, which are always defined by some characteristic;\nthus assertions about \u201cPeter\u201d are really about everything that is\n\u201cPeterish\u201d. To get a true proper name, we should have to get to a\nsingle particular or a set of particulars defined by enumeration,\nnot by a common quality. Since we cannot acquire knowledge of actual\nparticulars, the words we use denote, in the best language we can make,\neither adjectives or relations between two or more terms. In addition\nto these, there are words indicative of structure: _e.g._ in \u201cA is\ngreater than B\u201d, the words \u201cis\u201d and \u201cthan\u201d have no separate meaning,\nbut merely serve to show the \u201csense\u201d of the relation \u201cgreater\u201d, _i.e._\nthat it goes from A to B, not from B to A.\n\nStrictly speaking, we are still simplifying. True adjectives and\nrelations will require particulars for their terms; the sort of\nadjectives we can know, such as \u201cblue\u201d and \u201cround\u201d, will not be\napplicable to particulars. They are therefore analogous to the\nadjective \u201cpopulous\u201d applied to a town. To say \u201cthis town is populous\u201d\nmeans \u201cmany people live in this town\u201d. A similar transformation would\nbe demanded by logic in all the adjectives and relations we can know\nempirically. That is to say, no word that we can understand would occur\nin a grammatically correct account of the universe.\n\nLeaving on one side the vagueness and inaccuracy of words, let us ask\nourselves; in what circumstances do we feel convinced that we know a\nstatement to be true or false as the case may be? A present statement\nwill be regarded as true if, _e.g._ it agrees with recollection or\nperception; a past statement, if it raised expectations now confirmed.\nI do not mean to say that these are the only grounds upon which we\nregard statements as true; I mean that they are simple and typical,\nand worth examining. If you say \u201cit was raining this morning\u201d, I may\nrecollect that it was or that it wasn\u2019t. One may perhaps say that the\nwords \u201cthis morning\u201d are associated for me with the word \u201craining\u201d or\nwith the words \u201cnot raining\u201d. According to which occurs, I judge your\nstatement true or false. If I have neither association, I do not judge\nyour statement either true or false unless I have material for an\ninference; and I do not wish to consider inference yet. If you say \u201cthe\nlights have gone out\u201d, when I can see the lights shining, I judge that\nyou speak falsely, because my perception is associated with the words\n\u201clights shining\u201d. If you say \u201cthe lights will go out in a minute\u201d, you\nproduce a certain familiar kind of tension called \u201cexpectation\u201d, and\nafter a time you produce a judgment that you spoke falsely (if the\nlights do not go out). These are the ordinary _direct_ ways of deciding\non the truth or falsehood of statements about past, present, or future.\n\nIt is necessary to distinguish between direct and indirect grounds\nfor accepting or rejecting statements. Pragmatism considers only\nindirect grounds. Broadly speaking, it considers a statement false when\nthe consequences of accepting it are unfortunate. But this belongs\nto the region of inference. I ask you the way to the station, you\ntell me wrong, and I miss my train; I then _infer_ that you told me\nwrong. But if you say \u201cthe lights are out\u201d when I see them shining,\nI reject your statement without inference. In this case, something\nin my present circumstances is associated with words different from\nyours, and different in ways which I have learnt to regard as involving\nincompatibility. The ultimate test of falsehood is _never_, so I think,\nthe nature of the consequences of a belief, but the association between\nwords and sensible or remembered facts. A belief is \u201cverified\u201d when a\nsituation arises which gives a feeling of expectedness in connection\nwith it; it is falsified when the feeling is one of surprise. But\nthis only applies to beliefs which await some future contingency for\nverification or refutation. A belief which is an immediate reaction\nto a situation--_e.g._ when you are waiting for a race to begin and\npresently you say \u201cthey\u2019re off\u201d--has no need of verification, but\nverifies other beliefs. And even where the confirmation of a belief is\nin the future, it is the expectedness, not the pleasantness, of the\nconsequences that confirms the truth of the belief.\n\nI think it is a mistake to treat \u201cbelief\u201d as one kind of occurrence,\nas is done in traditional psychology. The sort of belief which is\nbased upon memory or perception alone differs from the sort which\ninvolves expectation. When you find in the time-table that a train\nleaves King\u2019s Cross at ten, your belief that this statement occurs in\nthe time-table does not await future confirmation, but your belief\nabout the train does: you may go to King\u2019s Cross and see the train\nstart. A belief which concerns an event may be a recollection, a\nperception, or an expectation. It may be none of these, in the case of\nan event which you have not seen and do not expect to see--_e.g._ C\u00e6sar\ncrossing the Rubicon, or the abolition of the House of Lords. But\nsuch beliefs always involve inference. I do not at this stage consider\nlogical and mathematical beliefs, some of which must be, _in a sense_,\nnon-inferential. But I think we shall find that this sense is different\nfrom that in which memories and perceptions are non-inferential.\n\nA belief, I should say, interpreted narrowly, is a form of words\nrelated to an emotion of one of several kinds. (I shall give a broader\nmeaning later.) The emotion is different according as the belief\nembodies a reminiscence, a perception, an expectation, or something\noutside the experience of the believer. Moreover, a form of words\nis not essential. Where the emotion is present, and leads to action\nrelevant to some feature of the environment, there may be said to be\na belief. The fundamental test of a belief, of no matter what sort,\nis that it causes some event which actually takes place to arouse\nthe emotion of expectedness or its opposite. I do not now attempt to\ndecide what an emotion is. Dr. Watson gives a behaviouristic account\nof emotions, which would, if adopted, make my definition of \u201cbelief\u201d\npurely behaviouristic. I have framed the definition so as not to\ninvolve a decision on the question of introspection.\n\nThe subject of truth and falsehood may be subdivided as follows:\n\n  A. _Formal Theory._--Given the meanings of the component words, what\n      decides whether a sentence is true or false?\n\n  B. _Causal Theory._--Can we distinguish between truth and falsehood\n      by (_a_) their causes, (_b_) their effects?\n\n  C. _Individual and Social Elements._--A statement is a social\n      occurrence, a belief is something individual.\n\n      How can we define a belief, and what is it when not composed of\n      words?\n\n  D. _Consistency and Truth._--Can we get outside the circle of\n      beliefs or statements to something else which shows them true,\n      not merely consistent? In other words, what possible relation is\n      there between propositions and facts?\n\nIt is very hard to disentangle these questions. The first question,\nas to formal theories, leads to the fourth, as to the relations of\npropositions to facts. _E.g._ \u201cBrutus killed C\u00e6sar\u201d is true because\nof a certain fact; what fact? The fact that Brutus killed C\u00e6sar. This\nkeeps us in the verbal realm, and does not get us outside it to some\nrealm of non-verbal fact by which verbal statements can be verified.\nHence our fourth problem arises. But this leads us to our second\nproblem, as to causes and effects of what is true or false, for it\nis here that we shall naturally look for the vital relation between\npropositions and facts. And here again we must distinguish between\n_thinking_ truly and _speaking_ truly. The former is an individual\naffair, the latter a social affair. Thus all our problems hang together.\n\nI will begin with C, the difference between a belief and a statement.\nBy a \u201cstatement\u201d I mean a form of words, uttered or written, with a\nview to being heard or read by some other person or persons, and not\na question, interjection, or command, but such as we should call an\nassertion. As to the question what forms of words are assertions, that\nis one for the grammarian and differs from language to language. But\nperhaps we can say rather more than that. The distinction, however,\nbetween an assertion and an imperative is not sharp. In England,\nnotices say \u201cVisitors are requested not to walk on the grass\u201d. In\nAmerica, they say \u201cKeep off! This means you.\u201d Effectively, the two have\nthe same meaning: yet the English notice consists only of a statement,\nwhile the American notice consists of an imperative followed by a\nstatement which must be false if read by more than one person. In so\nfar as statements are intended to influence the conduct of others, they\npartake of the nature of imperatives or requests. Their characteristic,\nhowever, is that they endeavour to effect their aim by producing a\n_belief_ which may or may not exist in the mind of the speaker. Often,\nhowever, they _express_ a belief, without stopping to consider the\neffect upon others. Thus a statement may be defined as a form of words\nwhich either expresses a belief or is intended to create one. Our next\nstep, therefore, must be the definition of \u201cbelief\u201d.\n\n\u201cBelief\u201d is a word which will be quite differently defined if we take\nan analytic point of view from the way in which we shall define it if\nwe regard the matter causally. From the point of view of science, the\ncausal point of view is the more important. Beliefs influence action\nin certain ways; what influences action in these ways may be called a\nbelief, even if, analytically, it does not much resemble what would\nordinarily be so called. We may therefore widen our previous definition\nof belief. Consider a man who goes to the house where his friend used\nto live, and, finding he has moved, says, \u201cI _thought_ he still lived\nhere\u201d, whereas he acted merely from habit without thought. If we are\ngoing to use words causally, we ought to say that this man had a\n\u201cbelief\u201d and therefore a \u201cbelief\u201d will be merely a characteristic of\na string of actions. We shall have to say: A man \u201cbelieves\u201d a certain\nproposition _p_ if, whenever he is aiming at any result to which _p_\nis relevant, he acts in a manner calculated to achieve the result if\n_p_ is true, but not otherwise. Sometimes this gives definite results,\nsometimes not. When you call a telephone number, it is clear that you\nbelieve that to be the number of the subscriber you want. But whether\nyou believe in the conservation of energy or a future life may be\nharder to decide. You may hold a belief in some contexts and not in\nothers; for we do not think in accordance with the so-called \u201cLaws\nof Thought\u201d. \u201cBelief\u201d like all the other categories of traditional\npsychology, is a notion incapable of precision.\n\nThis brings me to the question whether the truth or falsehood of a\nbelief can be determined either by its causes or by its effects. There\nis, however, a preliminary difficulty. I said just now that A believes\n_p_ if he acts in a way which will achieve his ends if _p_ is _true_.\nI therefore assumed that we know what is meant by \u201ctruth\u201d. I assumed,\nto be definite, that we know what is meant by \u201ctruth\u201d as applied to\na form of words. The argument was as follows: From observation of\na person\u2019s acts, you infer his beliefs, by a process which may be\nelaborate as the discovery of Kepler\u2019s Laws from the observed motions\nof the planets. His \u201cbeliefs\u201d are not assumed to be \u201cstates of mind\u201d,\nbut merely characteristics of series of actions. These beliefs, when\nascertained by observation, can be expressed in words; you can say,\n_e.g._ \u201cThis person believes that there is a train from King\u2019s Cross at\n10 A.M.\u201d Having once expressed the belief in words of which the meaning\nis known, you have arrived at the stage where formal theories are\napplicable. Words of known meaning, put together according to a known\nsyntax, are true or false in virtue of some fact, and their relation to\nthis fact results logically from the meanings of the separate words and\nthe laws of syntax. This is where logic is strong.\n\nIt will be seen that, according to what we have said, truth is\napplicable primarily to a form of words, and only derivatively to\na belief. A form of words is a social phenomenon, therefore the\nfundamental form of truth must be social. A form of words is true when\nit has a certain relation to a certain fact. What relation to what\nfact? I think the fundamental relation is this: a form of words is true\nif a person who knows the language is led to that form of words when\nhe finds himself in an environment which contains features that are\nthe meanings of those words, and these features produce reactions in\nhim sufficiently strong for him to use words which mean them. Thus \u201ca\ntrain leaves King\u2019s Cross at 10 A.M.\u201d is true if a person can be led\nto say, \u201cIt is now 10 A.M., this is King\u2019s Cross, and I see a train\nstarting\u201d. The environment causes words, and words directly caused by\nthe environment (if they are statements) are \u201ctrue\u201d. What is called\n\u201cverification\u201d in science consists in putting oneself in a situation\nwhere words previously used for other reasons result directly from\nthe environment. Of course, given this basis, there are innumerable\nindirect ways of verifying statements, but all, I think, depend upon\nthis direct way.\n\nThe above theory may be thought very odd, but it is partly designed\nto meet the fourth of our previous questions, namely, \u201cHow can we get\noutside words to the facts which make them true or false?\u201d Obviously we\ncannot do this within logic, which is imprisoned in the realm of words;\nwe can only do it by considering the relations of words to our other\nexperiences, and these relations, in so far as they are relevant, can\nhardly be other than causal. I think the above theory, as it stands,\nis too crude to be quite true. We must also bring in such things as\nexpectedness, which we discussed earlier. But I believe that the\ndefinition of truth or falsehood will have to be sought along some such\nlines as I have indicated.\n\nI want in conclusion to indulge in two speculations. The first concerns\na possible reconciliation of behaviourism and logic. It is clear that,\nwhen we have a problem to solve, we do not always solve it as the rat\ndoes, by means of random movements; we often solve it by \u201cthinking\u201d,\n_i.e._ by a process in which we are not making any overt movements.\nThe same thing was sometimes true of K\u00f6hler\u2019s chimpanzees. Now what is\ninvolved in the possibility of solving a problem by verbal thinking?\nWe put words together in various ways which are not wholly random, but\nlimited by previous knowledge of the _sort_ of phrase that is likely to\ncontain a solution of our problem. At last we hit upon a phrase which\nseems to give what we want. We then proceed to an overt action of the\nkind indicated by the phrase; if it succeeds, our problem is solved.\nNow this process is only intelligible if there is some connection\nbetween the laws of syntax and the laws of physics--using \u201csyntax\u201d\nin a psychological rather than a grammatical sense. I think this\nconnection is _assumed_ in logic and ordinary philosophy, but it ought\nto be treated as a problem requiring investigation by behaviourist\nmethods. I lay no stress on this suggestion, except as giving a hint\nfor future investigations. But I cannot think that the behaviourist has\ngone far towards the solution of his problem until he has succeeded in\nestablishing a connection between syntax and physics. Without this,\nthe efficacy of \u201cthought\u201d cannot be explained on his principles.\n\nMy second speculation is as to the limitations which the structure\nof language imposes upon the extent of our possible knowledge of\nthe world. I am inclined to think that quite important metaphysical\nconclusions, of a more or less sceptical kind, can be drawn from simple\nconsiderations as to the relation between language and things. A spoken\nsentence consists of a temporal series of events; a written sentence\nis a spatial series of bits of matter. Thus it is not surprising that\nlanguage can represent the course of events in the physical world;\nit can, in fact, make a map of the physical world, preserving its\nstructure in a more manageable form, and it can do this because it\nconsists of physical events. But if there were such a world as the\nmystic postulates, it would have a structure different from that of\nlanguage, and would therefore be incapable of being verbally described.\nIt is fairly clear that nothing verbal can conform or confute this\nhypothesis.\n\nA great deal of the confusion about relations which has prevailed in\npractically all philosophies comes from the fact that relations are\nindicated, not by relations, but by words which are as substantial as\nother words. Consequently, in thinking about relations, we constantly\nhover between the unsubstantiality of the relation itself and the\nsubstantiality of the word. Take, say, the fact that lightning precedes\nthunder. We saw earlier that to express this by a language closely\nreproducing the structure of the fact, we should have to say simply:\n\u201clightning, thunder\u201d, where the fact that the first word precedes the\nsecond means that what the first word means precedes what the second\nword means. But even if we adopted this method for temporal order, we\nshould still need words for all other relations, because we could not\nwithout intolerable ambiguity symbolise them also by the order of our\nwords. When we say \u201clightning precedes thunder\u201d, the word \u201cprecedes\u201d\nhas a quite different relation to what it means from that which the\nwords \u201clightning\u201d and \u201cthunder\u201d have to what they respectively mean.\nWittgenstein[12] says that what really happens is that we establish a\nrelation between the word \u201clightning\u201d and the word \u201cthunder\u201d, namely\nthe relation of having the word \u201cprecedes\u201d between them. In this way he\ncauses relations to be symbolised by relations. But although this may\nbe quite correct, it is sufficiently odd to make it not surprising that\npeople have thought the word \u201cprecedes\u201d means a relation in the same\nsense in which \u201clightning\u201d means a kind of event. This view, however,\nmust be incorrect. I think it has usually been held unconsciously, and\nhas produced many confusions about relations which cease when it is\nexposed to the light of day--for example, those which lead Bradley to\ncondemn relations.\n\n     [12] _Tractatus Logico-Philosophicus_ (Kegan Paul).\n\nIn all this I have been considering the question of the relation\nbetween the structure of language and the structure of the world. It is\nclear that anything that can be said in an inflected language can be\nsaid in an uninflected language; therefore, everything that can be said\nin language can be said by means of a temporal series of uninflected\nwords. This places a limitation upon what can be expressed in words.\nIt may well be that there are facts which do not lend themselves to\nthis very simple schema; if so, they cannot be expressed in language.\nOur confidence in language is due to the fact that it consists of\nevents in the physical world, and, therefore, shares the structure of\nthe physical world, and therefore can express that structure. But if\nthere be a world which is not physical, or not in space-time, it may\nhave a structure which we can never hope to express or to know. These\nconsiderations might lead us to something like the Kantian _a priori_,\nnot as derived from the structure of the mind, but as derived from the\nstructure of language, which is the structure of the physical world.\nPerhaps that is why we know so much physics and so little of anything\nelse. However, I have lapsed into mystical speculation, and will leave\nthese possibilities, since, by the nature of the case, I cannot _say_\nanything true about them.\n\n\n\n\nCHAPTER XXV\n\nTHE VALIDITY OF INFERENCE\n\n\nIt is customary in science to regard certain facts as \u201cdata\u201d, from\nwhich laws and also other facts are \u201cinferred\u201d. We saw in Chapter VII\nthat the _practice_ of inference is much wider than the theories of any\nlogician would justify, and that it is nothing other than the law of\nassociation or of \u201clearned reactions\u201d. In the present chapter, I wish\nto consider what the logicians have evolved from this primitive form of\ninference, and what grounds we have, as rational beings, for continuing\nto infer. But let us first get as clear a notion as we can of what\nshould be meant by a \u201cdatum\u201d.\n\nThe conception of a \u201cdatum\u201d cannot be made absolute. Theoretically,\nit should mean something that we know without inference. But before\nthis has any definite meaning, we must define both \u201cknowledge\u201d and\n\u201cinference\u201d. Both these terms have been considered in earlier chapters.\nFor our present purpose it will simplify matters to take account only\nof such knowledge as is expressed in words. We considered in Chapter\nXXIV the conditions required in order that a form of words may be\n\u201ctrue\u201d; for present purposes, therefore, we may say that \u201cknowledge\u201d\nmeans \u201cthe assertion of a true form of words\u201d. This definition is not\nquite adequate, since a man may be right by chance; but we may ignore\nthis complication. We may then define a \u201cdatum\u201d as follows: A \u201cdatum\u201d\nis a form of words which a man utters as the result of a stimulus, with\nno intermediary of any learned reaction beyond what is involved in\nknowing how to speak. We must, however, permit such learned reactions\nas consist in adjustments of the sense-organs or in mere increase of\nsensitivity. These merely improve the receptivity to data, and do not\ninvolve anything that can be called inference.\n\nIf the above definition is accepted, all our data for knowledge of\nthe external world must be of the nature of percepts. The belief in\nexternal objects is a learned reaction acquired in the first months of\nlife, and it is the duty of the philosopher to treat it as an inference\nwhose validity must be tested. A very little consideration shows that,\nlogically, the inference cannot be demonstrative, but must be at best\nprobable. It is not logically _impossible_ that my life may be one\nlong dream, in which I merely imagine all the objects that I believe\nto be external to me. If we are to reject this view, we must do so on\nthe basis of an inductive or analogical argument, which cannot give\ncomplete certainty. We perceive other people behaving in a manner\nanalogous to that in which we behave, and we assume that they have had\nsimilar stimuli. We may hear a whole crowd say \u201coh\u201d at the moment when\nwe see a rocket burst, and it is natural to suppose that the crowd\nsaw it too. Nor are such arguments confined to living organisms. We\ncan talk to a dictaphone and have it afterwards repeat what we said;\nthis is most easily explained by the hypothesis that at the surface\nof the dictaphone events happened, while I was speaking, which were\nclosely analogous to those that were happening just outside my ears.\nIt remains _possible_ that there is no dictaphone and I have no ears\nand there is no crowd watching the rocket; my percepts _may_ be all\nthat is happening in such cases. But, if so, it is difficult to arrive\nat any causal laws, and arguments from analogy are more misleading\nthan we are inclined to think them. As a matter of fact, the whole\nstructure of science, as well as the world of common sense, demands\nthe use of induction and analogy if it is to be believed. These forms\nof inference, therefore, rather than deduction, are those that must be\nexamined if we are to accept the world of science or any world outside\nof our own dreams.\n\nLet us take a simple example of an induction which we have all\nperformed in practice. If we are hungry, we eat certain things we see\nand not others--it may be said that we infer edibility inductively\nfrom a certain visual and olfactory appearance. The history of this\nprocess is that children a few months old put everything into their\nmouths unless they are stopped; sometimes the result is pleasant,\nsometimes unpleasant; they repeat the former rather than the latter.\nThat is to say: given that an object having a certain visual and\nolfactory appearance has been found pleasant to eat, an object having\na very similar appearance will be eaten; but when a certain appearance\nhas been found connected with unpleasant consequences when eaten, a\nsimilar appearance does not lead to eating next time. The question is:\nwhat logical justification is there for our behaviour? Given all our\npast experience, are we more likely to be nourished by bread than by a\nstone? It is easy to see why we think so, but can we, as philosophers\njustify this way of thinking?\n\nIt is, of course, obvious that unless one thing can be a sign of\nanother both science and daily life would be impossible. More\nparticularly, reading involves this principle. One accepts printed\nwords as signs, but this is only justifiable by means of induction.\nI do not mean that induction is necessary to establish the existence\nof other people, though that also, as we have seen, is true. I mean\nsomething simpler. Suppose you want your hair cut, and as you walk\nalong the street you see a notice \u201chair-cutting, first floor\u201d. It\nis only by means of induction that you can _establish_ that this\n_notice_ makes it in some degree probable that there is a hair-cutter\u2019s\nestablishment on the first floor. I do not mean that you employ the\nprinciple of induction; I mean that you act in accordance with it,\nand that you would have to appeal to it if you were accompanied by a\nlong-haired sceptical philosopher who refused to go upstairs till he\nwas persuaded there was some point in doing so.\n\nThe principle of induction, _prima facie_, is as follows: Let there\nbe two kinds of events, A and B (_e.g._ lightning and thunder), and\nlet many instances be known in which an event of the kind A has\nbeen quickly followed by one of the kind B, and no instances of the\ncontrary. Then either a sufficient number of instances of this\nsequence, or instances of suitable kinds will make it increasingly\nprobable that A is always followed by B, and in time the probability\ncan be made to approach certainty without limit provided the right\nkind and number of instances can be found. This is the principle we\nhave to examine. Scientific theories of induction generally try to\nsubstitute well-chosen instances for numerous instances, and represent\nnumber of instances as belonging to crude popular induction. But in\nfact popular induction depends upon the emotional interest of the\ninstances, not upon their number. A child who has burnt its hand\n_once_ in a candle-flame establishes an induction, but words take\nlonger, because at first they are not emotionally interesting. The\nprinciple used in primitive practice is: Whatever, on a given occasion,\nimmediately precedes something very painful or pleasant, is a sign\nof that interesting event. Number plays a secondary part as compared\nwith emotional interest. That is one reason why rational thought is so\ndifficult.\n\nThe logical problem of induction is to show that the proposition \u201cA\nis always accompanied (or followed) by B\u201d can be rendered probable by\nknowledge of instances in which this happens, provided the instances\nare suitably chosen or very numerous. Far the best examination of\ninduction is contained in Mr. J.\u00a0M. Keynes\u2019s _Treatise on Probability_.\nThere is a valuable doctor\u2019s thesis by the late Jean Nicod, _Le\nProbl\u00e8me logique de l\u2019induction_, which is very ably reviewed by R.\u00a0B.\nBraithwaite in _Mind_, October 1925. A man who reads these three will\nknow most of what is known about induction. The subject is technical\nand difficult, involving a good deal of mathematics, but I will attempt\nto give the gist of the results.\n\nWe will begin with the condition in which the problem had been left by\nJ.\u00a0S. Mill. He had four canons of induction, by means of which, given\nsuitable examples, it could be demonstrated that A and B were causally\nconnected, if the law of causation could be assumed. That is to say,\ngiven the law of causation, the scientific use of induction could\nbe reduced to deduction. Roughly the method is this: We know that B\nmust have a cause; the cause cannot be C or D or E, etc., because we\nfind by experiment or observation that these may be present without\nproducing B. On the other hand, we never succeed in finding A without\nits being accompanied (or followed) by B. If A and B are both capable\nof quantity, we may find further that the more there is of A the more\nthere is of B. By such methods we eliminate all possible causes except\nA; therefore, since B must have a cause, that cause must be A. All\nthis is not really induction at all; true induction only comes in in\nproving the law of causation. This law Mill regards as proved by mere\nenumeration of instances: we know vast numbers of events which have\ncauses, and no events which can be shown to be uncaused; therefore, it\nis highly probable that all events have causes. Leaving out of account\nthe fact that the law of causality cannot have quite the form that Mill\nsupposed, we are left with the problem: Does mere number of instances\nafford a basis for induction? If not, is there any other basis? This is\nthe problem to which Mr. Keynes addresses himself.\n\nMr. Keynes holds that an induction may be rendered more probable by\nnumber of instances, not because of their mere number, but because of\nthe probability, if the instances are very numerous, that they will\nhave nothing in common except the characteristics in question. We want,\nlet us suppose, to find out whether some quality A is always associated\nwith some quality B. We find instances in which this is the case; but\nit may happen that in all our instances some quality C is also present,\nand that it is C that is associated with B. If we can so choose our\ninstances that they have nothing in common except the qualities A and\nB, then we have better grounds for holding that A is always associated\nwith B. If our instances are very numerous, then, even if we do not\n_know_ that they have no other common quality, it may become quite\nlikely that this is the case. This, according to Mr. Keynes, is the\nsole value of many instances.\n\nA few technical terms are useful. Suppose we want to establish\ninductively that there is some probability in favour of the\ngeneralisation: \u201cEverything that has the property _F_ also has the\nproperty _f_\u201d. We will call this generalisation _g_. Suppose we have\nobserved a number of instances in which _F_ and _f_ go together, and\nno instances to the contrary. These instances may have other common\nproperties as well; the sum-total of their common properties is called\nthe _total positive analogy_, and the sum-total of their _known_ common\nqualities is called the _known positive analogy_. The properties\nbelonging to some but not to all of the instances in question are\ncalled the _negative analogy_: all of them constitute the _total\nnegative analogy_, all those that are _known_ constitute the _known\nnegative analogy_. To strengthen an induction, we want to diminish the\npositive analogy to the utmost possible extent; this, according to Mr.\nKeynes, is why numerous instances are useful.\n\nOn \u201cpure\u201d induction, where we rely solely upon number of instances,\nwithout _knowing_ how they affect the analogy, Mr. Keynes concludes\n(_Treatise in Probability_, p. 236):\n\n\u201cWe have shown that if each of the instances necessarily follows\nfrom the generalisation, then each additional instance increases the\nprobability of the generalisation, so long as the new instance could\nnot have been predicted with certainty from a knowledge of the former\ninstances.... The common notion, that each successive verification of\na doubtful principle strengthens it, is formally proved, therefore\nwithout any appeal to conceptions of law or of causality. _But we have\nnot proved_ that this probability approaches certainty as a limit, or\neven that our conclusion becomes more likely than not, as the number of\nverifications or instances is indefinitely increased.\u201d\n\nIt is obvious that induction is not much use unless, with suitable\ncare, its conclusions can be rendered more likely to be true than\nfalse. This problem therefore necessarily occupies Mr. Keynes.\n\nIt is found that an induction will approach certainty as a limit if two\nconditions are fulfilled:\n\n(1) If the generalisation is false, the probability of its being true\nin a new instance when it has been found to be true in a certain\nnumber of instances, however great that number may be, falls short of\ncertainty by a finite amount.\n\n(2) There is a finite _a priori_ probability in favour of our\ngeneralisation.\n\nMr. Keynes uses \u201cfinite\u201d here in a special sense. He holds that not all\nprobabilities are numerically measurable; a \u201cfinite\u201d probability is one\nwhich exceeds some numerically measurable probability however small.\n_E.g._ our generalisation has a finite _a priori_ probability if it is\nless unlikely than throwing heads a billion times running.\n\nThe difficulty is, however, that there is no easily discoverable way of\nestimating the _a priori_ probability of a generalisation. In examining\nthis question, Mr. Keynes is led to a very interesting postulate\nwhich, if true, will, he thinks, give the required finite _a priori_\nprobability. His postulate as he gives it is not quite correct, but I\nshall give his form first, and then the necessary modification.\n\nMr. Keynes supposes that the qualities of objects cohere in groups,\nso that the number of _independent_ qualities is much less than the\ntotal number of qualities. We may conceive this after the analogy\nof biological species: a cat has a number of distinctive qualities\nwhich are found in all cats, a dog has a number of other distinctive\nqualities which are found in all dogs. The method of induction can, he\nsays, be justified if we assume \u201cthat the objects in the field, over\nwhich our generalisations extend, do not have an infinite number of\nindependent qualities; that, in other words, their characteristics,\nhowever numerous, cohere together in groups of invariable connection,\nwhich are finite in number\u201d (p. 256). Again (p. 258): \u201cAs a biological\nfoundation for Analogy, therefore, we seem to need some such assumption\nas that the amount of variety in the universe is limited in such a way\nthat there is no one object so complex that its qualities fall into an\ninfinite number of independent groups ... or rather that none of the\nobjects about which we generalise are as complex as this; or at least\nthat, though some objects may be infinitely complex, we sometimes have\na finite probability that an object about which we seek to generalise\nis not infinitely complex.\u201d\n\nThis postulate is called the \u201cprinciple of limitation of variety\u201d. Mr.\nKeynes again finds that it is needed in attempts to establish laws\nby statistics; if he is right, it is needed for all our scientific\nknowledge outside pure mathematics. Jean Nicod pointed out that it is\nnot quite sufficiently stringent. We need, according to Mr. Keynes,\na finite probability that the object in question has only a finite\nnumber of independent qualities; but what we really need is a finite\nprobability that the number of its independent qualities is less than\nsome assigned finite number. This is a very different thing, as may be\nseen by the following illustration. Suppose there is some number of\nwhich we know only that it is finite; it is infinitely improbable that\nit will be less than a million, or a billion, or any other assigned\nfinite number, because, whatever such number we take, the number\nof smaller numbers is finite and the number of greater numbers is\ninfinite. Nicod requires us to assume that there is a finite number _n_\nsuch that there is a finite probability that the number of independent\nqualities of our object is less than _n_. This is a much stronger\nassumption than Mr. Keynes\u2019s, which is merely that the number of\nindependent qualities is finite. It is the stronger assumption which is\nneeded to justify induction.\n\nThis result is very interesting and very important. It is remarkable\nthat it is in line with the trend of modern science. Eddington has\npointed out that there is a certain finite number which is fundamental\nin the universe, namely the number of electrons. According to the\nquantum theory, it would seem that the number of possible arrangements\nof electrons may well also be finite, since they cannot move in all\npossible orbits, but only in such as make the action in one complete\nrevolution conform to the quantum principle. If all this is true, the\nprinciple of limitation of variety may well also be true. We cannot,\nhowever, arrive at a proof of our principle in this way, because\nphysics uses induction, and is therefore presumably invalid unless\nthe principle is true. What we can say, in a general way, is that\nthe principle does not refute itself, but, on the contrary, leads to\nresults which confirm it. To this extent, the trend of modern science\nmay be regarded as increasing the plausibility of the principle.\n\nIt is important to realise the fundamental position of probability in\nscience. At the very best, induction and analogy only give probability.\nEvery inference worthy of the name is inductive, therefore all inferred\nknowledge is at best probable. As to what is meant by probability,\nopinions differ. Mr. Keynes takes it as a fundamental logical category:\ncertain premisses may make a conclusion more or less probable, without\nmaking it certain. For him, probability is a relation between a premiss\nand a conclusion. A proposition does not have a definite probability\non its own account; in itself, it is merely true or false. But it has\nprobabilities of different amounts in regard to different premisses.\nWhen we speak, elliptically, of _the_ probability of a proposition,\nwe mean its probability in relation to all our relevant knowledge.\nA proposition in probability cannot be refuted by mere observation:\nimprobable things may happen and probable things may fail to happen.\nNor is an estimate of probability relevant to given evidence proved\nwrong when further evidence alters the probability.\n\nFor this reason the inductive principle cannot be proved or disproved\nby experience. We might prove validly that such and such a conclusion\nwas enormously probable, and yet it might not happen. We might\nprove invalidly that it was probable, and yet it might happen.\nWhat happens affects the probability of a proposition, since it is\nrelevant evidence; but it never alters the probability relative to\nthe previously available evidence. The whole subject of probability,\ntherefore, on Mr. Keynes\u2019s theory, is strictly _a priori_ and\nindependent of experience.\n\nThere is however another theory, called the \u201cfrequency theory\u201d, which\nwould make probability not indefinable, and would allow empirical\nevidence to affect our estimates of probability relative to given\npremisses. According to this theory in its crude form, the probability\nthat an object having the property _F_ will have the property _f_ is\nsimply the proportion of the objects having both properties to all\nthose having the property _F_. For example, in a monogamous country\nthe probability of a married person being male is exactly a half. Mr.\nKeynes advances strong arguments against all forms of this theory that\nexisted when his book was written. There is however an article by R.\u00a0H.\nNisbet on \u201cThe Foundations of Probability\u201d in _Mind_ for January 1926,\nwhich undertakes to rehabilitate the frequency theory. His arguments\nare interesting, and suffice to show that the controversy is still an\nopen one, but they do not, in my opinion, amount to decisive proof.\nIt is to be observed, however, that the frequency theory, if it could\nbe maintained, would be preferable to Mr. Keynes\u2019s, because it would\nget rid of the necessity for treating probability as indefinable, and\nwould bring probability into much closer touch with what actually\noccurs. Mr. Keynes leaves an uncomfortable gap between probability and\nfact, so that it is far from clear why a rational man will act upon a\nprobability. Nevertheless, the difficulties of the frequency theory\nare so considerable that I cannot venture to advocate it definitely.\nMeanwhile, the details of the discussion are unaffected by the view\nwe may take on this fundamental philosophical question. And on either\nview the principle of limitation of variety will be equally necessary\nto give validity to the inferences by induction and analogy upon which\nscience and daily life depend.\n\n\n\n\nCHAPTER XXVI\n\nEVENTS, MATTER, AND MIND\n\n\nEverything in the world is composed of \u201cevents\u201d; that, at least, is the\nthesis I wish to maintain. An \u201cevent\u201d, as I understand it, is something\nhaving a small finite duration and a small finite extension in space;\nor rather, in view of the theory of relativity, it is something\noccupying a small finite amount of space-time. If it has parts, these\nparts, I say, are again events, never something occupying a mere point\nof instant, whether in space, in time, or in space-time. The fact that\nan event occupies a finite amount of space-time does not prove that it\nhas parts. Events are not impenetrable, as matter is supposed to be; on\nthe contrary, every event in space-time is overlapped by other events.\nThere is no reason to suppose that any of the events with which we are\nfamiliar are infinitely complex; on the contrary, everything known\nabout the world is compatible with the view that every complex event\nhas a finite number of parts. We do not know that this is the case,\nbut it is an hypothesis which cannot be refuted and is simpler than\nany other possible hypothesis. I shall therefore adopt it as a working\nhypothesis in what follows.\n\nWhen I speak of an \u201cevent\u201d I do not mean anything out of the way.\nSeeing a flash of lightning is an event; so is hearing a tire burst,\nor smelling a rotten egg, or feeling the coldness of a frog. These\nare events that are \u201cdata\u201d in the sense of Chapter XXV; but, on the\nprinciples explained in that chapter, we infer that there are events\nwhich are not data and happen at a distance from our own body. Some of\nthese are data to other people, others are data to no one. In the case\nof the flash of lightning, there is an electro-magnetic disturbance\nconsisting of events travelling outward from the place where the flash\ntakes place, and then when this disturbance reaches the eye of a\nperson or animal that can see, there is a percept, which is causally\ncontinuous with the events between the place of the lightning and the\nbody of the percipient. Percepts afford the logical premisses for all\ninferences to events that are not precepts, wherever such inferences\nare logically justifiable. Particular colours and sounds and so on are\nevents; their causal antecedents in the inanimate world are also events.\n\nIf we assume, as I propose to do, that every event has only a finite\nnumber of parts, then every event is composed of a finite number of\nevents that have no parts. Such events I shall call \u201cminimal events.\u201d\nIt will simplify our discussion to assume them, but by a little\ncircumlocution this assumption could be eliminated. The reader must not\ntherefore regard it as an essential part of what follows.\n\nA minimal event occupies a finite region in space-time. Let us take\ntime alone for purposes of illustration. The event in question may\noverlap in time with each of two others, although the first of these\nothers wholly precedes the second; for example, you may hear a long\nnote on the violin while you hear two short notes on the piano. (It\nis not necessary to suppose that these are really minimal events; I\nmerely want to illustrate what is meant.) I assume that every event\nis contemporaneous with events that are not contemporaneous with each\nother; this is what is meant by saying that every event lasts for a\nfinite time, as the reader can easily convince himself if he remembers\nthat time is wholly relational. If we look away from the world of\nphysics for a moment, and confine ourselves to the world of one man\u2019s\nexperience, we can easily define an \u201cinstant\u201d in his life. It will be\na group of events, all belonging to his experience, and having the\nfollowing two properties: (1) any two of the events overlap; (2) no\nevent outside the group overlaps with every member of the group. By\na slightly more complicated but essentially similar method, we can\ndefine a point-instant in space-time as a group of events having two\nproperties analogous to those used just now in defining an \u201cinstant\u201d\nin one biography.[13] Thus the \u201cpoints\u201d (or point-instants) that\nthe mathematician needs are not simple, but are structures composed\nof events, made up for the convenience of the mathematician. There\nwill be many \u201cpoints\u201d of which a given minimal event is a member;\nall these together make up the region of space-time occupied by that\nevent. Space-time order, as well as space-time points, results from the\nrelations between events.\n\n     [13] See _The Analysis of Matter_, by the present author,\n          chap. xxviii.\n\nA piece of matter, like a space-time point, is to be constructed out of\nevents, but the construction is considerably more complicated, and in\nthe end is only an approximation to what the physicist supposes to be\nreally taking place. There are, at the moment, two somewhat different\nviews of matter, one appropriate to the study of atomic structure, the\nother to the general theory of relativity as affording an explanation\nof gravitation. The view appropriate to atomic structure has itself\ntwo forms, one derived from Heisenberg, the other from De Broglie\nand Schr\u00f6dinger. These two forms, it is true, are mathematically\nequivalent, but in words they are very different. Heisenberg regards\na piece of matter as a centre from which radiations travel outward;\nthe radiations are supposed really to occur, but the matter at their\ncentre is reduced to a mere mathematical fiction. The radiations are,\nfor example, such as constitute light; they are all avowedly systems\nof events, not changes in the conditions or relations of \u201csubstances.\u201d\nIn the De Broglie-Schr\u00f6dinger system, matter consists of wave motions.\nIt is not necessary to the theory to postulate anything about these\nwave-motions except their mathematical characteristics, but, obviously,\nsince they are to explain matters they cannot serve their purpose if\nthey consist of motions of matter. In this system also, therefore,\nwe are led to construct matter out of systems of events, which just\nhappen, and do not happen \u201cto\u201d matter or \u201cto\u201d anything else.\n\nGravitation, as explained by the general theory of relativity, is\nreduced to \u201ccrinkles\u201d in space-time. Space-time being, as we have\nalready seen, a system constructed out of events, the \u201ccrinkles\u201d in it\nare also derived from events. There is no reason to suppose that there\nis a \u201cthing\u201d at the place where the \u201ccrinkle\u201d is most crinkly. Thus in\nthis part of physics, also, matter has ceased to be a \u201cthing\u201d and has\nbecome merely a mathematical characteristic of the relations between\ncomplicated logical structures composed of events.\n\nIt was traditionally a property of substance to be permanent, and to a\nconsiderable extent matter has retained this property in spite of its\nloss of substantiality. But its permanence now is only approximate,\nnot absolute. It is thought that an electron and a proton can meet and\nannihilate each other; in the stars this is supposed to be happening\non a large scale.[14] And even while an electron or a proton lasts,\nit has a different kind of persistence from that formerly attributed\nto matter. A wave in the sea persists for a longer or shorter time:\nthe waves that I see dashing themselves to pieces on the Cornish coast\nmay have come all the way from Brazil, but that does not mean that a\n\u201cthing\u201d has travelled across the Atlantic; it means only that a certain\nprocess of change has travelled. And just as a wave in the sea comes\nto grief at last on the rocks, so an electron or a proton may come to\ngrief when it meets some unusual state of affairs.\n\n     [14] See _The Analysis of Matter_, by the present author,\n          chap. xxviii.\n\nThus \u201cmatter\u201d has very definitely come down in the world as a result of\nrecent physics. It used to be the cause of our sensations: Dr. Johnson\n\u201cdisproved\u201d Berkeley\u2019s denial of matter by kicking a stone. If he had\nknown that his foot never touched the stone, and that both were only\ncomplicated systems of wave-motions, he might have been less satisfied\nwith his refutation. We cannot say that \u201cmatter\u201d is the cause of our\nsensations. We can say that the events which cause our sensations\nusually belong to the sort of group that physicists regard as material;\nbut that is a very different thing. Impenetrability used to be a noble\nproperty of matter, a kind of Declaration of Independence; now it is a\nmerely tautological result of the way in which matter is defined. The\nevents which are the real stuff of the world are not impenetrable,\nsince they can overlap in space-time. In a word, \u201cmatter\u201d has become\nno more than a convenient shorthand for stating certain causal laws\nconcerning events.\n\nBut if matter has fared badly, mind has fared little better.\nThe adjective \u201cmental\u201d is one which is not capable of any exact\nsignificance. There is, it is true, an important group of events,\nnamely percepts, all of which may be called \u201cmental\u201d. But it would be\narbitrary to say that there are no \u201cmental\u201d events except percepts,\nand yet it is difficult to find any principle by which we can decide\nwhat other events should be included. Perhaps the most essential\ncharacteristics of mind are introspection and memory. But memory\nin some of its forms is, as we have seen, a consequence of the law\nof conditioned reflexes, which is at least as much physiological\nas psychological, and characterises living tissue rather than\nmind. Knowledge, as we have found, is not easy to distinguish from\nsensitivity, which is a property possessed by scientific instruments.\nIntrospection is a form of knowledge, but turns out, on examination, to\nbe little more than a cautious interpretation of ordinary \u201cknowledge\u201d.\nWhere the philosopher\u2019s child at the Zoo says \u201cThere is a hippopotamus\nover there\u201d, the philosopher should reply: \u201cThere is a coloured pattern\nof a certain shape, which may perhaps be connected with a system of\nexternal causes of the sort called a hippopotamus\u201d. (I do not live up\nto this precept myself.) In saying that there is a coloured pattern,\nthe philosopher is practising introspection in the only sense that I\ncan attach to that term, _i.e._ his knowledge-reaction is to an event\nsituated in his own brain from the standpoint of physical space, and\nis consciously avoiding physiological and other inference as far as\npossible. Events to which a knowledge-reaction of this sort occurs are\n\u201cmental\u201d; so are, presumably, other events resembling them in certain\nrespects. But I do not see any way of defining this wider group except\nby saying that mental events are events in a living brain, or, better,\nin a region combining sensitivity and the law of learned reactions to\na marked extent. This definition has at least the merit of showing that\nmentality is an affair of causal laws, not of the quality of single\nevents, and also that mentality is a matter of degree.\n\nPerhaps it is not unnecessary to repeat, at this point, that events in\nthe brain are not to be regarded as consisting of motions of bits of\nmatter. Matter and motion, as we have seen, are logical constructions\nusing events as their material, and events are therefore something\nquite different from matter in motion. I take it that, when we have\na percept, just what we perceive (if we avoid avoidable sources of\nerror) is an event occupying part of the region which, for physics, is\noccupied by the brain. In fact, perception gives us the most concrete\nknowledge we possess as to the stuff of the physical world, but what\nwe perceive is part of the stuff of our brains, not part of the stuff\nof tables and chairs, sun, moon, and stars. Suppose we are looking at\na leaf, and we see a green patch. This patch is not \u201cout there\u201d where\nthe leaf is, but is an event occupying a certain volume in our brains\nduring the time that we see the leaf. Seeing the leaf consists of\nthe existence, in the region occupied by our brain, of a green patch\ncausally connected with the leaf, or rather with a series of events\nemanating from the place in physical space where physics places the\nleaf. The percept is one of this series of events, differing from the\nothers in its effects owing to the peculiarities of the region in\nwhich it occurs--or perhaps it would be more correct to say that the\ndifferent effects _are_ the peculiarities of the region.\n\nThus \u201cmind\u201d and \u201cmental\u201d are merely approximate concepts, giving a\nconvenient shorthand for certain approximate laws. In a completed\nscience, the word \u201cmind\u201d and the word \u201cmatter\u201d would both disappear,\nand would be replaced by causal laws concerning \u201cevents\u201d, the only\nevents known to us otherwise than in their mathematical and causal\nproperties being percepts, which are events situated in the same\nregion as a brain and having effects of a peculiar sort called\n\u201cknowledge-reactions\u201d.\n\nIt will be seen that the view which I am advocating is neither\nmaterialism nor mentalism, but what we call \u201cneutral monism\u201d. It is\nmonism in the sense that it regards the world as composed of only one\n_kind_ of stuff, namely events; but it is pluralism in the sense that\nit admits the existence of a great multiplicity of events, each minimal\nevent being a logically self-subsistent entity.\n\nThere is, however, another question, not quite the same as this, namely\nthe question as to the relations of psychology and physics. If we knew\nmore, would psychology be absorbed in physics? or, conversely, would\nphysics be absorbed in psychology? A man may be a materialist and\nyet hold that psychology is an independent science; this is the view\ntaken by Dr. Broad in his important book on _The Mind and its Place\nin Nature_. He holds that a mind is a material structure, but that it\nhas properties which could not, even theoretically, be inferred from\nthose of its material constituents. He points out that structures very\noften have properties which, in the present state of our knowledge,\ncannot be inferred from the properties and relations of their parts.\nWater has many properties which we cannot infer from those of hydrogen\nand oxygen, even if we suppose ourselves to know the structure of the\nmolecule of water more completely than we do as yet. Properties of a\nwhole which cannot, even theoretically, be inferred from the properties\nand relations of its parts are called by Dr. Broad \u201cemergent\u201d\nproperties. Thus he holds that a mind (or brain) has properties which\nare \u201cemergent\u201d, and to this extent psychology will be independent of\nphysics and chemistry. The \u201cemergent\u201d properties of minds will only be\ndiscoverable by observation of minds, not by inference from the laws\nof physics and chemistry. This possibility is an important one, and it\nwill be worth while to consider it.\n\nOur decision to regard a unit of matter as itself not ultimate, but an\nassemblage of events, somewhat alters the form of our question as to\n\u201cemergent\u201d properties. We have to ask: Is matter emergent from events?\nIs mind emergent from events? If the former, is mind emergent from\nmatter, or deducible from the properties of matter, or neither? If the\nlatter, is matter emergent from mind or deducible from the properties\nof mind, or neither? Of course, if neither mind nor matter is emergent\nfrom events, these latter questions do not arise.\n\nLet us coin a word, \u201cchrono-geography\u201d, for the science which begins\nwith events having space-time relations and does not assume at the\noutset that certain strings of them can be treated as persistent\nmaterial units or as minds. Then we have to ask ourselves first: can\nthe science of matter, as it appears in physics and chemistry, be\nwholly reduced to chrono-geography? If no, matter is emergent from\nevents; if yes, it is not emergent.\n\n_Is matter emergent from events?_ In the present state of science it\nis difficult to give a decided answer to this question. The notion\nof matter, in modern physics, has become absorbed into the notion of\nenergy. Eddington, in his _Mathematical Theory of Relativity_ shows\nthat, in virtue of the laws assumed concerning events, there must\nbe something having the observed properties of matter and energy as\nregards conservation. This he calls the \u201cmaterial-energy-tensor\u201d, and\nsuggests that it is the reality which we sometimes call \u201cmatter\u201d and\nsometimes \u201cenergy\u201d. To this extent, matter has been shown to be not\nemergent. But the existence of electrons and protons (to the extent\nthat they do exist) has not yet been deduced from the general theory\nof relativity, though attempts are being made and may at any moment\nsucceed. If and when these attempts succeed, physics will cease to be\nin any degree independent of chrono-geography, but for the present\nit remains in part independent. As for chemistry, although we cannot\npractically reduce it all to physics, we can see how, theoretically,\nthis could be done, and I think it is safe to assume that it is not an\nultimately independent science.\n\nThe question we have been asking is: could we predict, theoretically,\nfrom the laws of events that there must be material units obeying\nthe laws which they do in fact obey, or is this a new, logically\nindependent, fact? In theory we might be able to prove that it is _not_\nindependent, but it would be very difficult to prove that it _is_. The\npresent position is, broadly speaking, that the continuous properties\nof the physical world can be deduced from chrono-geography, but not the\ndiscontinuous facts, viz. electrons and protons and Planck\u2019s quantum.\nThus for the present materiality is practically, though perhaps not\ntheoretically, an emergent characteristic of certain groups of events.\n\n_Is mind emergent from events?_ This question, as yet, can hardly be\neven discussed intelligently, because psychology is not a sufficiently\nadvanced science. There are, nevertheless, some points to be noted.\nChrono-geography is concerned only with the abstract mathematical\nproperties of events, and cannot conceivably, unless it is radically\ntransformed, prove that there are visual events, or auditory events, or\nevents of any of the kinds that we know by perception. In this sense,\npsychology is certainly emergent from chrono-geography and also from\nphysics, and it is hard to see how it can ever cease to be so. The\nreason for this is that our knowledge of data contains features of a\nqualitative sort, which cannot be deduced from the merely mathematical\nfeatures of the space-time events inferred from data, and yet these\nabstract mathematical features are all that we can legitimately infer.\n\nThe above argument decides also that mind must be emergent from matter,\nif it is a material structure. No amount of physics can ever tell us\nall that we do in fact know about our own percepts.\n\nWe have still to ask whether we are to regard a mind as a structure of\nmaterial units or not. If we do so regard it, we are, so far as mind is\nconcerned, emergent materialists in view of what we have just decided;\nthis is the view favoured by Dr. Broad. If we do not so regard it, we\nare in no sense materialists. In favour of the materialist view, there\nis the fact that, so far as our experience goes, minds only emerge in\nconnection with certain physical structures, namely living bodies, and\nthat mental development increases with a certain kind of complexity\nof physical structure. We cannot set against this the argument that\nminds have peculiar characteristics, for this is quite consistent with\n_emergent_ materialism. If we are to refute it, it must be by finding\nout what sort of group of events constitutes a mind. It is time to\naddress ourselves to this question.\n\n_What is a mind?_ It is obvious, to begin with, that a mind must be a\ngroup of mental events, since we have rejected the view that it is a\nsingle simple entity such as the ego was formerly supposed to be. Our\nfirst step, therefore, is to be clear as to what we mean by a \u201cmental\u201d\nevent.\n\nWe said a few pages ago that: _Mental events are events in a region\ncombining sensitivity and the law of learned reactions to a marked\nextent_. For practical purposes, this means (subject to a proviso to be\nexplained shortly) that a mental event is any event in a living brain.\nWe explained that this does not mean that a mental event consists of\nmatter in motion, which is what an old-fashioned physicist would regard\nas the sort of event that happens in a brain. Matter in motion, we have\nseen, is not an event in our sense, but a shorthand description of a\nvery complicated causal process among events of a different sort. But\nwe must say a few words in justification of our definition.\n\nLet us consider some alternative definitions. A mental event, we might\nsay, is one which is \u201cexperienced.\u201d When is an event \u201cexperienced\u201d?\nWe might say: when it has \u201cmnemic\u201d effects, _i.e._ effects governed\nby the law of association. But we saw that this law applies to purely\nbodily events such as the contraction of the pupil, with which nothing\n\u201cmental\u201d seems to be connected. Thus if our definition is to serve,\nwe shall have to define \u201cexperience\u201d differently; we shall have to\nsay that the mnemic effects must include something that can be called\n\u201cknowledge.\u201d This would suggest the definition: A mental event is\nanything that is remembered. But this is too narrow: we only remember\na small proportion of our mental events. We might have regarded\n\u201cconsciousness\u201d as the essence of mental events, but this view was\nexamined and found inadequate in Chapter XX. Moreover, we do not want\nour definition to exclude the \u201cunconscious\u201d.\n\nIt is clear that the primary mental events, those about which there can\nbe no question, are percepts. But percepts have certain peculiar causal\nproperties, notably that they give rise to knowledge-reactions, and\nthat they are capable of having mnemic effects which are cognitions.\nThese causal properties, however, belong to some events which are not\napparently percepts. It seems that any event in the brain may have\nthese properties. And perhaps we were too hasty in saying that the\ncontraction of the pupil on hearing a loud noise involves nothing\n\u201cmental\u201d. There may be other \u201cmental\u201d events connected with a human\nbody besides those belonging to the central personality. I shall come\nback to this possibility presently. Meanwhile, I shall adhere to the\nabove definition of a \u201cmental\u201d event, which, as we saw, makes mentality\na matter of degree.\n\nWe can now return to the question: What is a mind? There may be mental\nevents not forming part of the sort of group that we should call a\n\u201cmind\u201d, but there certainly are groups having that kind of unity that\nmake us call them one mind. There are two marked characteristics of\na mind: First, it is connected with a certain body; secondly, it has\nthe unity of one \u201cexperience\u201d. The two _prima facie_ diverge in cases\nof dual or multiple personality, but I think this is more apparent\nthan real. These two characteristics are, one physical, the other\npsychological. Let us consider each in turn as a possible definition of\nwhat we mean by one \u201cmind\u201d.\n\nIn the physical way, we begin by observing that every mental event\nknown to us is also part of the history of a living body, and we define\na \u201cmind\u201d as the group of mental events which form part of the history\nof a certain living body. The definition of a living body is chemical,\nand the reduction of chemistry to physics is clear in theory, though\nin practice the mathematics is too difficult. It is so far a merely\nempirical fact that mnemic causation is almost exclusively associated\nwith matter having a certain chemical structure. But the same may be\nsaid of magnetism. As yet, we cannot deduce the magnetic properties of\niron from what we know of the structure of the atom of iron, but no one\ndoubts that they could be deduced by a person with sufficient knowledge\nand sufficient mathematical skill. In like manner it may be assumed\nthat mnemic causation is theoretically deducible from the structure of\nliving matter. If we knew enough, we might be able to infer that some\nother possible structure would exhibit mnemic phenomena, perhaps in an\neven more marked degree; if so, we might be able to construct Robots\nwho would be more intelligent than we are.\n\nIn the psychological way of defining a \u201cmind\u201d, it consists of all the\nmental events connected with a given mental event by \u201cexperience\u201d,\n_i.e._ by mnemic causation, but this definition needs a little\nelaboration before it can be regarded as precise. We do not want the\ncontraction of the pupil to count as a \u201cmental\u201d event; therefore a\nmental event will have to be one which has mnemic effects, not merely\nmnemic causes. In that case, however, there cannot be a last mental\nevent in a man\u2019s life, unless we assume that it may have mnemic effects\non his body after death. Perhaps we may avoid this inconvenience by\ndiscovering the _kind_ of event that usually has mnemic effects, though\nthey may be prevented from occurring by special circumstances. Or we\nmight maintain that death is gradual, even when it is what is called\ninstantaneous; in that case the last events in a man\u2019s life grow\nprogressively less mental as life ebbs. Neglecting this point, which is\nnot very important, we shall define the \u201cexperience\u201d to which a given\nmental event belongs as all those mental events which can be reached\nfrom the given event by a mnemic causal chain, which may go backwards\nor forwards, or alternately first one and then the other. This may be\nconceived on the analogy of an engine shunting at a junction or where\nthere are many points: any line that can be reached, by however many\nshuntings, will count as part of the same experience.\n\nWe cannot be sure that all the mental events connected with one body\nare connected by links of mnemic causation with each other, and\ntherefore we cannot be sure that our two definitions of one \u201cmind\u201d\ngive the same result. In cases of multiple personality, some at least\nof the usual mnemic effects, notably recollection, are absent in the\nlife of one personality when they have occurred in the life of the\nother. But probably both personalities are connected by mnemic chains\nwith events which occurred before the dissociation took place, so that\nthere would be only one mind according to our definition. But there\nare other possibilities which must be considered. It may be that each\ncell in the body has its own mental life, and that only selections from\nthese mental lives go to make up the life which we regard as ours.\nThe \u201cunconscious\u201d might be the mental lives of subordinate parts of\nthe body, having occasional mnemic effects which we can notice, but\nin the main separate from the life of which we are \u201cconscious\u201d. If\nso, the mental events connected with one body will be more numerous\nthan the events making up its central \u201cmind\u201d. These, however, are only\nspeculative possibilities.\n\nI spoke a moment ago of the life of which we are \u201cconscious\u201d, and\nperhaps the reader has been wondering why I have not made more use of\nthe notion of \u201cconsciousness\u201d. The reason is that I regard it as only\none kind of mnemic effect, and not one entitled to a special place.\nTo say that I am \u201cconscious\u201d of an event is to say that I recollect\nit, at any rate for a short time after it has happened. To say that I\nrecollect an event is to say that a certain event is occurring in me\nnow which is connected by mnemic causation with the event recollected,\nand is of the sort that we call a \u201ccognition\u201d of that event. But\nevents which I do not recollect may have mnemic effects upon me. This\nis the case, not only where we have Freudian suppression, but in all\nhabits which were learnt long ago and have now become automatic, such\nas writing and speaking. The emphasis upon consciousness has made a\nmystery of the \u201cunconscious\u201d, which ought to be in no way surprising.\n\nIt does not much matter which of our two definitions of a \u201cmind\u201d we\nadopt. Let us, provisionally, adopt the first definition, so that a\nmind is all the mental events which form part of the history of a\ncertain living body, or perhaps we should rather say a living brain.\n\nWe can now tackle the question which is to decide whether we are\nemergent materialists or not, namely:\n\n_Is a mind a structure of material units?_\n\nI think it is clear that the answer to this question is in the\nnegative. Even if a mind consists of all the events in a brain, it\ndoes not consist of bundles of these events grouped as physics groups\nthem, _i.e._ it does not lump together all the events that make up\none piece of matter in the brain, and then all the events that make\nup another, and so on. Mnemic causation is what concerns us most in\nstudying mind, but this seems to demand a recourse to physics, if we\nassume, as seems plausible, that mental mnemic causation is due to\neffects upon the brain. This question, however, is still an open one.\nIf mnemic causation is ultimate, mind is emergent. If not, the question\nis more difficult. As we saw earlier, there certainly is knowledge in\npsychology which cannot ever form part of physics. But as this point is\nimportant, I shall repeat the argument in different terms.\n\nThe difference between physics and psychology is analogous to that\nbetween a postman\u2019s knowledge of letters and the knowledge of a\nrecipient of letters. The postman knows the movements of many letters,\nthe recipient knows the contents of a few. We may regard the light and\nsound waves that go about the world as letters of which the physicist\nmay know the destination; some few of them are addressed to human\nbeings, and when read give psychological knowledge. Of course the\nanalogy is not perfect, because the letters with which the physicist\ndeals are continually changing during their journeys, as if they were\nwritten in fading ink, which, also, was not quite dry all the time,\nbut occasionally got smudged with rain. However, the analogy may pass\nif not pressed.\n\nIt would be possible without altering the detail of previous\ndiscussions, except that of Chapter XXV, to give a different turn\nto the argument, and make matter a structure composed of mental\nunits. I am not quite sure that this is the wrong view. It arises not\nunnaturally from the argument as to data contained in Chapter XXV. We\nsaw that all data are mental events in the narrowest and strictest\nsense, since they are percepts. Consequently all verification of causal\nlaws consists in the occurrence of expected percepts. Consequently any\ninference beyond percepts (actual or possible) is incapable of being\nempirically tested. We shall therefore be prudent if we regard the\nnon-mental events of physics as mere auxiliary concepts, not assumed\nto have any reality, but only introduced to simplify the laws of\npercepts. Thus matter will be a construction built out of percepts, and\nour metaphysic will be essentially that of Berkeley. If there are no\nnon-mental events, causal laws will be very odd; for example, a hidden\ndictaphone may record a conversation although it did not exist at the\ntime, since no one was perceiving it. But although this seems odd, it\nis not logically impossible. And it must be conceded that it enables\nus to interpret physics with a smaller amount of dubious inductive and\nanalogical inference than is required if we admit non-mental events.\n\nIn spite of the logical merits of this view, I cannot bring myself\nto accept it, though I am not sure that my reasons for disliking it\nare any better than Dr. Johnson\u2019s. I find myself constitutionally\nincapable of believing that the sun would not exist on a day when he\nwas everywhere hidden by clouds, or that the meat in a pie springs into\nexistence at the moment when the pie is opened. I know the logical\nanswer to such objections, and _qua logician_ I think the answer a good\none. The logical argument, however, does not even tend to show that\nthere are _not_ non-mental events; it only tends to show that we have\nno right to feel sure of their existence. For my part, I find myself\nin fact believing in them in spite of all that can be said to persuade\nme that I ought to feel doubtful.\n\nThere is an argument, of a sort, against the view we are considering.\nI have been assuming that we admit the existence of other people and\ntheir perceptions, but question only the inference from perceptions to\nevents of a different kind. Now there is no good reason why we should\nnot carry our logical caution a step further. I cannot verify a theory\nby means of another man\u2019s perceptions, but only by means of my own.\nTherefore the laws of physics can only be verified by me in so far as\nthey lead to predictions of _my_ percepts. If then, I refuse to admit\nnon-mental events because they are not verifiable, I ought to refuse\nto admit mental events in every one except myself, on the same ground.\nThus I am reduced to what is called \u201csolipsism\u201d, _i.e._ the theory that\nI alone exist. This is a view which is hard to refute, but still harder\nto believe. I once received a letter from a philosopher who professed\nto be a solipsist, but was surprised that there were no others! Yet\nthis philosopher was by way of believing that no one else existed. This\nshows that solipsism is not really believed even by those who think\nthey are convinced of its truth.\n\nWe may go a step further. The past can only be verified indirectly,\nby means of its effects in the future; therefore the type of logical\ncaution we have been considering should lead us to abstain from\nasserting that the past really occurred: we ought to regard it as\nconsisting of auxiliary concepts convenient in stating the laws\napplicable to the future. And since the future, though verifiable if\nand when it occurs, is as yet unverified, we ought to suspend judgment\nabout the future also. If we are not willing to go so far as this,\nthere seems no reason to draw the line at the precise point where it\nwas drawn by Berkeley. On these grounds I feel no shame in admitting\nthe existence of non-mental events such as the laws of physics lead us\nto infer. Nevertheless, it is important to realise that other views are\ntenable.\n\n\n\n\nCHAPTER XXVII\n\nMAN\u2019S PLACE IN THE UNIVERSE\n\n\nIn this final chapter, I propose to recapitulate the main conclusions\nat which we have arrived, and then to say a few words on the subject of\nMan\u2019s relation to the universe in so far as philosophy has anything to\nteach on this subject without extraneous help.\n\nPopular metaphysics divides the known world into mind and matter, and\na human being into soul and body. Some--the materialists--have said\nthat matter alone is real and mind is an illusion. Many--the idealists\nin the technical sense, or mentalists, as Dr. Broad more appropriately\ncalls them--have taken the opposite view, that mind alone is real and\nmatter is an illusion. The view which I have suggested is that both\nmind and matter are structures composed of a more primitive stuff which\nis neither mental nor material. This view, called \u201cneutral monism\u201d,\nis suggested in Mach\u2019s _Analysis of Sensations_, developed in William\nJames\u2019s _Essays in Radical Empiricism_, and advocated by John Dewey,\nas well as by Professor R.\u00a0B. Parry and other American realists. The\nuse of the word \u201cneutral\u201d in this way is due to Dr. H.\u00a0M. Sheffer,[15]\nformerly of Harvard, who is one of the ablest logicians of our time.\n\n     [15] See Holt\u2019s _Concept of Consciousness_, preface.\n\nSince man is the instrument of his own knowledge, it is necessary to\nstudy him as an instrument before we can appraise the value of what our\nsenses seem to tell us concerning the world. In Part I we studied man,\nwithin the framework of common-sense beliefs, just as we might study\nclocks or thermometers, as an instrument sensitive to certain features\nof the environment, since sensitiveness to the environment is obviously\nan indispensable condition for knowledge about it.\n\nIn Part II we advanced to the study of the physical world. We\nfound that matter, in modern science, has lost its solidity and\nsubstantiality; it has become a mere ghost haunting the scenes of\nits former splendours. In pursuit of something that could be treated\nas substantial, physicists analysed ordinary matter into molecules,\nmolecules into atoms, atoms into electrons and protons. There, for\na few years, analysis found a resting-place. But now electrons and\nprotons themselves are dissolved into systems of radiations by\nHeisenberg, and into systems of waves by Schr\u00f6dinger--the two theories\namount mathematically to much the same thing. And these are not wild\nmetaphysical speculations; they are sober mathematical calculations,\naccepted by the great majority of experts.\n\nAnother department of theoretical physics, the theory of relativity,\nhas philosophical consequences which are, if possible, even more\nimportant. The substitution of space-time for space and time has\nmade the category of substance less applicable than formerly, since\nthe essence of substance was persistence through time, and there is\nnow no one cosmic time. The result of this is to turn the physical\nworld into a four-dimensional continuum of events, instead of a\nseries of three-dimensional states of a world composed of persistent\nbits of matter. A second important feature of relativity-theory is\nthe abolition of force, particularly gravitational force, and the\nsubstitution of differential causal laws having to do only with the\nneighbourhood of an event, not with an influence exerted from a\ndistance, such as gravitation formerly seemed to be.\n\nThe modern study of the atom has had two consequences which have\nconsiderably changed the philosophical hearing of physics. On the\none hand, it appears that there are discontinuous changes in nature,\noccasions when there is a sudden jump from one state to another without\npassing through the intermediate states. (Schr\u00f6dinger, it is true,\nquestions the need for assuming discontinuity; but so far his opinion\nhas not prevailed.) On the other hand, the course of nature is not so\ndefinitely determined by the physical laws at present known as it was\nformerly thought to be. We cannot predict when a discontinuous change\nwill take place in a given atom, though we can predict statistical\naverages. It can no longer be said that, given the laws of physics and\nthe relevant facts about the environment, the future history of an atom\ncan theoretically be calculated from its present condition. It may be\nthat this is merely due to the insufficiency of our knowledge, but we\ncannot be sure that this is the case. As things stand at present, the\nphysical world is not so rigidly deterministic as it has been believed\nto be during the last 250 years. And in various directions what\nformerly appeared as laws governing each separate atom are now found to\nbe only averages attributable in part to the laws of chance.\n\nFrom these questions concerning the physical world in itself, we were\nled to others concerning the causation of our perceptions, which are\nthe data upon which our scientific knowledge of physics is based. We\nsaw that a long causal chain always intervenes between an external\nevent and the event in us which we regard as perception of the\nexternal event. We cannot therefore suppose that the external event\nis exactly what we see or hear; it can, at best, resemble the percept\nonly in certain structural respects. This fact has caused considerable\nconfusion in philosophy, partly because philosophers tried to think\nbetter of perception than it deserves, partly because they failed to\nhave clear ideas on the subject of space. It is customary to treat\nspace as a characteristic of matter as opposed to mind, but this is\nonly true of _physical_ space. There is also _perceptual_ space,\nwhich is that in which what we know immediately through the senses is\nsituated. This space cannot be identified with that of physics. From\nthe standpoint of physical space, all our percepts are in our heads;\nbut in perceptual space our percept of our hand is outside our percept\nof our head. The failure to keep physical and perceptual space distinct\nhas been a source of great confusion in philosophy.\n\nIn Part III we resumed the study of man, but now as he appears to\nhimself, not only as he is known to an external observer. We decided,\ncontrary to the view of the behaviourists, that there are important\nfacts which cannot be known except when the observer and observed are\nthe same person. The datum in perception, we decided, is a private fact\nwhich can only be known directly to the percipient; it is a datum for\nphysics and psychology equally, and must be regarded as both physical\nand mental. We decided later that there are inductive grounds, giving\nprobability but not certainty, in favour of the view that perceptions\nare causally connected with events which the percipient does not\nexperience, which may belong only to the physical world.\n\nThe behaviour of human beings is distinguished from that of inanimate\nmatter by what are called \u201cmnemic\u201d phenomena, _i.e._ by a certain kind\nof effect of past occurrences. This kind of effects is exemplified in\nmemory, in learning, in the intelligent use of words, and in every\nkind of knowledge. But we cannot, on this ground, erect an absolute\nbarrier between mind and matter. In the first place, inanimate matter,\nto some slight extent, shows analogous behaviour--_e.g._ if you unroll\na roll of paper, it will roll itself up again. In the second place, we\nfind that living bodies display mnemic phenomena to exactly the same\nextent to which minds display them. In the third place, if we are to\navoid what I have called \u201cmnemic\u201d causation, which involves action at\na distance in time, we must say that mnemic phenomena in mental events\nare due to the modification of the body by past events. That is to\nsay, the set of events which constitutes one man\u2019s experience is not\ncausally self-sufficient, but is dependent upon causal laws involving\nevents which he cannot experience.\n\nOn the other hand, our knowledge of the physical world is purely\nabstract: we know certain logical characteristics of its structure,\nbut nothing of its intrinsic character. There is nothing in physics\nto prove that the intrinsic character of the physical world differs,\nin this or that respect, from that of the mental world. Thus from\nboth ends, both by the analysis of physics and by the analysis of\npsychology, we find that mental and physical events form one causal\nwhole, which is not known to consist of two different sorts. At\npresent, we know the laws of the physical world better than those of\nthe mental world, but that may change. We know the intrinsic character\nof the mental world to some extent, but we know absolutely nothing\nof the intrinsic character of the physical world. And in view of the\nnature of the inferences upon which our knowledge of physics rests, it\nseems scarcely possible that we should ever know more than abstract\nlaws about matter.\n\nIn Part IV we considered what philosophy has to say about the universe.\nThe function of philosophy, according to the view advocated in this\nvolume, is somewhat different from that which has been assigned to it\nby a large and influential school. Take, _e.g._ Kant\u2019s antinomies.\nHe argues (1) that space must be infinite, (2) that space cannot be\ninfinite; and he deduces that space is subjective. The non-Euclideans\nrefuted the argument that it must be infinite, and Georg Cantor refuted\nthe argument that it cannot be. Formerly, _a priori_ logic was used to\nprove that various hypotheses which looked possible were impossible,\nleaving only one possibility, which philosophy therefore pronounced\ntrue. Now _a priori_ logic is used to prove the exact contrary, namely,\nthat hypotheses which looked impossible are possible. Whereas logic\nwas formerly counsel for the prosecution, it is now counsel for the\ndefence. The result is that many more hypotheses are at large than\nwas formerly the case. Formerly, to revert to the instance of space,\nit appeared that experience left only one kind of space to logic, and\nlogic showed this one kind to be impossible. Now, logic presents many\nkinds of space as possible apart from experience, and experience only\npartially decides between them. Thus, while our knowledge of what is\nhas become less than it was formerly supposed to be, our knowledge of\nwhat may be is enormously increased. Instead of being shut in within\nnarrow walls, of which every nook and cranny could be explored, we find\nourselves in an open world of free possibilities, where much remains\nunknown because there is so much to know. The attempt to prescribe\nto the universe by means of _a priori_ principles has broken down;\nlogic, instead of being, as formerly, a bar to possibilities, has\nbecome the great liberator of the imagination, presenting innumerable\nalternatives which are closed to unreflective common sense, and leaving\nto experience the task of deciding, where decision is possible, between\nthe many worlds which logic offers for our choice.\n\nPhilosophical knowledge, if what we have been saying is correct, does\nnot differ essentially from scientific knowledge; there is no special\nsource of wisdom which is open to philosophy but not to science, and\nthe results obtained by philosophy are not radically different from\nthose reached in science. Philosophy is distinguished from science only\nby being more critical and more general. But when I say that philosophy\nis critical, I do not mean that it attempts to criticise knowledge from\noutside, for that would be impossible: I mean only that it examines\nthe various parts of our supposed knowledge to see whether they are\nmutually consistent and whether the inferences employed are such as\nseem valid to a careful scrutiny. The criticism aimed at is not that\nwhich, without reason, determines to reject, but that which considers\neach piece of apparent knowledge on its merits and retains whatever\nstill appears to be knowledge when this consideration is completed.\nThat some risk of error remains must be admitted, since human beings\nare fallible. Philosophy may claim justly that it diminishes the risk\nof error, and that in some cases it renders the risk so small as to be\npractically negligible. To do more than this is not possible in a world\nwhere mistakes must occur; and more than this no prudent advocate of\nphilosophy would claim to have performed.\n\nI want to end with a few words about man\u2019s place in the universe. It\nhas been customary to demand of a philosopher that he should show that\nthe world is good in certain respects. I cannot admit any duty of this\nsort. One might as well demand of an accountant that he should show\na satisfactory balance sheet. It is just as bad to be fraudulently\noptimistic in philosophy as in money matters. If the world is good,\nby all means let us know it; but if not, let us know that. In any\ncase, the question of the goodness or badness of the world is one for\nscience rather than for philosophy. We shall call the world good if\nit has certain characteristics that we desire. In the past philosophy\nprofessed to be able to prove that the world had such characteristics,\nbut it is now fairly evident that the proofs were invalid. It does not,\nof course, follow that the world does not have the characteristics in\nquestion; it follows only that philosophy cannot decide the problem.\nTake for example the problem of personal immortality. You may believe\nthis on the ground of revealed religion, but that is a ground which\nlies outside philosophy. You may believe it on the ground of the\nphenomena investigated by psychical research, but that is science, not\nphilosophy. In former days, you could believe it on a philosophical\nground, namely, that the soul is a substance and all substances are\nindestructible. You will find this argument, sometimes more or less\ndisguised, in many philosophers. But the notion of substance, in\nthe sense of a permanent entity with changing states, is no longer\napplicable to the world. It may happen, as with the electron, that a\nstring of events are so interconnected causally that it is practically\nconvenient to regard them as forming one entity, but where this happens\nit is a scientific fact, not a metaphysical necessity. The whole\nquestion of personal immortality, therefore lies outside philosophy,\nand it is to be decided, if at all, either by science or by revealed\nreligion.\n\nI will take up another matter in regard to which what I have said may\nhave been disappointing to some readers. It is sometimes thought that\nphilosophy ought to aim at encouraging a good life. Now, of course,\nI admit that it should have this effect, but I do not admit that it\nshould have this as a conscious purpose. To begin with, when we embark\nupon the study of philosophy we ought not to assume that we already\nknow for certain what the good life is; philosophy may conceivably\nmodify our views as to what is good, in which case it will seem to\nthe non-philosophical to have had a bad moral effect. That, however,\nis a secondary point. The essential thing is that philosophy is part\nof the pursuit of knowledge, and that we cannot limit this pursuit by\ninsisting that the knowledge obtained shall be such as we should have\nthought edifying before we obtained it. I think it could be maintained\nwith truth that _all_ knowledge is edifying, provided we have a right\nconception of edification. When this appears to be not the case it is\nbecause we have moral standards based upon ignorance. It may happen by\ngood fortune that a moral standard based upon ignorance is right, but\nif so knowledge will not destroy it; if knowledge can destroy it, it\nmust be wrong. The conscious purpose of philosophy, therefore, ought\nto be solely to _understand_ the world as well as possible, not to\nestablish this or that proposition which is thought morally desirable.\nThose who embark upon philosophy must be prepared to question all\ntheir preconceptions, ethical as well as scientific; if they have a\ndetermination never to surrender certain philosophic beliefs, they are\nnot in the frame of mind in which philosophy can be profitably pursued.\n\nBut although philosophy ought not to have a moral purpose, it ought\nto have certain good moral effects. Any disinterested pursuit of\nknowledge teaches us the limits of our power, which is salutary; at\nthe same time, in proportion as we succeed in achieving knowledge, it\nteaches the limits of our impotence, which is equally desirable. And\nphilosophical knowledge, or rather philosophical thought, has certain\nspecial merits not belonging in an equal degree to other intellectual\npursuits. By its generality it enables us to see human passions in\ntheir just proportions, and to realise the absurdity of many quarrels\nbetween individuals, classes, and nations. Philosophy comes as near\nas possible for human beings to that large, impartial contemplation of\nthe universe as a whole which raises us for the moment above our purely\npersonal destiny. There is a certain asceticism of the intellect which\nis good as a part of life, though it cannot be the whole so long as\nwe have to remain animals engaged in the struggle for existence. The\nasceticism of the intellect requires that, while we are engaged in the\npursuit of knowledge, we shall repress all other desires for the sake\nof the desire to know. While we are philosophising, the wish to prove\nthat the world is good, or that the dogmas of this or that sect are\ntrue, must count as weaknesses of the flesh--they are temptations to\nbe thrust on one side. But we obtain in return something of the joy\nwhich the mystic experiences in harmony with the will of God. This joy\nphilosophy can give, but only to those who are willing to follow it to\nthe end, through all its arduous uncertainties.\n\nThe world presented for our belief by a philosophy based upon modern\nscience is in many ways less alien to ourselves than the world of\nmatter as conceived in former centuries. The events that happen in\nour minds are part of the course of nature, and we do not know that\nthe events which happen elsewhere are of a totally different kind.\nThe physical world, so far as science can show at present, is perhaps\nless rigidly determined by causal laws than it was thought to be;\none might, more or less fancifully, attribute even to the atom a\nkind of limited free will. There is no need to think of ourselves as\npowerless and small in the grip of vast cosmic forces. All measurement\nis conventional, and it would be possible to devise a perfectly\nserviceable system of measurement according to which a man would be\nlarger than the sun. No doubt there are limits to our power, and it\nis good that we should recognise the fact. But we cannot say what\nthe limits are, except in a quite abstract way, such as that we\ncannot create energy. From the point of view of human life, it is\nnot important to be able to _create_ energy; what is important is to\nbe able to direct energy into this or that channel, and this can do\nmore and more as our knowledge of science increases. Since men first\nbegan to think, the forces of nature have oppressed them; earthquakes,\nfloods, pestilences, and famines have filled them with terror. Now at\nlast, thanks to science, mankind is discovering how to avoid much of\nthe suffering that such events have hitherto entailed. The mood in\nwhich, as it seems to me the modern man should face the universe is\none of quiet self-respect. The universe as known to science is not in\nitself either friendly or hostile to man, but it can be made to act as\na friend if approached with patient knowledge. Where the universe is\nconcerned, knowledge is the one thing needful. Man, alone of living\nthings, has shown himself capable of the knowledge required to give\nhim a certain mastery over his environment. The dangers to man in the\nfuture, or at least in any measurable future, come, not from nature,\nbut from man himself. Will he use his power wisely? Or will he turn\nthe energy liberated from the struggle with nature into struggles with\nhis fellow-men? History, science, and philosophy all make us aware\nof the great collective achievements of mankind. It would be well if\nevery civilised human being had a sense of these achievements and a\nrealisation of the possibility of greater things to come, with the\nindifference which must result as regards the petty squabbles upon\nwhich the passions of individuals and nations are wastefully squandered.\n\nPhilosophy should make us know the ends of life, and the elements in\nlife that have value on their own account. However our freedom may\nbe limited in the causal sphere, we need admit no limitations to our\nfreedom in the sphere of values: what we judge good on its own account\nwe may continue to judge good, without regard to anything but our own\nfeeling. Philosophy cannot itself determine the ends of life, but it\ncan free us from the tyranny of prejudice and from distortions due\nto a narrow view. Love, beauty, knowledge, and joy of life: these\nthings retain their lustre however wide our purview. And if philosophy\ncan help us to feel the value of these things, it will have played\nits parts in man\u2019s collective work of bringing light into a world of\ndarkness.\n\n\n\n\nINDEX\n\n\n  \u00c6ther or empty space, 107\n\n  Analogy, positive and negative, 271\n\n  _Analysis of Sensations_ (Mach), 292\n\n  _Animal Intelligence_ (Thorndike), 30\n\n  Animal learning, study of, 29 ff.\n    Thorndike\u2019s laws of, 31 f.\n    learned reactions, 35 f.\n\n  _A priori_, causation not regarded as, 150\n    knowledge, 249 f., 265\n    probability, on Keynes\u2019s theory, is, 274\n    logic, 296\n\n  Aristotle, 226\n\n  Association, principle of, 33 f., 48, 64, 180\n\n  Aston, Dr. F.\u00a0W., 99\n\n  Atom, theory of the, 98 ff.\n    centre from which radiations travel, 157\n    philosophical consequences of modern study of the, 293\n\n  Attention, 205\n\n\n  Bacon, 80\n\n  Behaviourism, its view of man, 70 ff.\n    where it breaks down as a final philosophy, 129\n    dilemma put to, 133\n    its propositions as to thought examined, 169 ff.\n    and logic, 263\n\n  _Behaviourism_ (Watson), 22, 31, 33\n\n  \u201cBelief\u201d, 254, 258 ff.\n    definition of, 261\n\n  Beliefs, defects in common, 3 ff.\n\n  Bergson, 71, 73, 198\n\n  Berkeley, 246 f.\n\n  _Bodily Changes in Pain, Hunger, Fear and Rage_ (Cannon), 218\n\n  Body, human, 25, 139\n\n  Bohr, Niels, his addition to the theory of atoms, 101 ff.\n\n  Bradley, monistic view of, 251\n    criticism of his argument against relations, 252\n\n  Braithwaite, R.\u00a0B., 269\n\n  Brentano, 202\n\n  Broad, Dr., 188, 195, 282, 292\n\n  Buddha, 227\n\n  Butler, Samuel, 71\n\n\n  Cannon, 218\n\n  Cantor, Georg, 296\n\n  Casuistry, 225\n\n  Causation, as an _a priori_ belief, 5, 150\n    notion of \u201cnecessary\u201d sequence, 115\n    conception of, in science, 144 ff.\n\n  \u201cCause\u201d, Kant\u2019s category of, 248 f.\n\n  \u201cChrono-geography\u201d, 283\n\n  Cognition, 61, 202 f., 217\n\n  Conation, 202\n\n  Conception, 203\n\n  \u201cConditioned reflexes\u201d, 35\n\n  Confucius, 227\n\n  Conscience, 228\n\n  \u201cConsciousness\u201d, 60\n    William James\u2019s views on, 210\n    two different meanings of the word, 210\n    criticism of common sense view of, 211 ff.\n    self, 214\n    William James\u2019s views approved, 217\n    one kind of mnemic effect, 288\n\n  Continuity in nature, 108\n\n  Correlation, laws of, 117\n\n  _Critique of Practical Reason_ (Kant), 249\n\n  Curiosity, 220\n\n\n  Dalton, 98\n\n  \u201cData\u201d, 266 f., 276\n\n  De Broglie, 278\n\n  Decalogue, the, 227\n\n  Descartes, 9, 162 ff., 237 ff.\n\n  Desire, behaviourist view of, 90 f.\n    introspective view of, 221 ff.\n\n  Dewey, John, 292\n\n  Discontinuity in nature, 101, 106, 108\n\n  Dreams, 62, 127, 175, 176, 185, 189, 193\n\n  Dualism of mind and matter, 141, 239\n\n  Ductless glands, the, 218\n\n\n  Eddington, Professor, 273, 279\n\n  Education, 233\n\n  Einstein, 96, 116, 239, 242, 249\n\n  Electron, 99 ff., 118, 145\n\n  \u201cEmergent\u201d properties, 282\n\n  Emotions, essential physiological conditions of the, 118\n    subject to \u201cConditioning\u201d, 119\n    generate irrational opinions, 120\n\n  Energy, radiation of, from matter into empty space, 145\n    propagation in empty space, 145\n    impact on matter in empty space, 146\n\n  _Essays in Radical Empiricism_ (William James), 210, 292\n\n  Ethics, views of the ancients on, 227\n    theory that virtue consists in obedience to authority, 227 ff.\n    utilitarian theory of, 229 f.\n    the concept of \u201cgood\u201d, 230\n    mainly social, 233\n    the supreme moral rule, 234 f.\n\n  Events, in physics, 110 f.\n    string of, 118 f.\n    \u201cmental\u201d, 141, 280 ff.\n    structure and mathematical laws of, 157\n    minimal, 277\n    matter constructed out of, 278\n\n  Experience, effects of, in a reaction to stimulus, 180 ff.\n\n\n  Familiarity, a stage in memory, 195 f.\n\n  Fear and Rage, 219\n\n  Feeling, as mental occurrence, 202\n\n  Forces, 111, 114, 117, 120 f.\n\n  Form, reaction to, 85 f.\n\n  Freudian \u201cunconscious\u201d the, 221\n\n\n  Galileo, 80\n\n  Generalisations, 271 f.\n\n  Geodesic, 112, 117\n\n  Geometry, as empirical as geography, 249 f.\n\n  _Gestaltpsychologie_, 37, 41, 43, 68, 247\n\n  Gravitation, 116 f., 145, 279\n\n  Griffith, Mr. Percy, 118\n\n\n  Habit-formation, 36\n\n  Habit-memory, 188, 196\n\n  Hegel, 227, 229, 251\n\n  Heisenberg, 96, 105, 278, 293\n\n  Heisenberg-Schr\u00f6dinger theories of atomic structure, 243\n\n  Heraclitus, 251\n\n  Huc, Monsieur, 232\n\n  Hume, 180, 191, 247 f.\n\n\n  Images, visual, auditory and tactual, 176\n    behaviourist explanation of, 177 f.\n    difference between sensations and, 179 ff.\n    definition of, 184 f.\n    first stage in memory, 195\n\n  Imagination, analysis of, 190 ff.\n    essence of, 191\n    exceptional gifts of, 193\n    and belief, 193 f.\n    difference between memory and, 194\n\n  Induction problem of validity of, 14\n    as a practice, 80 f.\n    principle of, 268 f.\n    logical problem of, 269 ff.\n    Mr. Keynes\u2019s examination of, 270 ff.\n\n  Inference, \u201cphysiological\u201d, 13, 80 ff., 135\n    syllogistic, 79\n    inductive and mathematical, 83 ff.\n\n  \u201cInnate ideas\u201d, doctrine of, 245\n\n  Interval, space-like and time-like, 110 f.\n\n  Introspection, 10, 11, 12, 172 f., 201 ff.\n\n\n  James, William, 210, 223.\n\n\n  Kant, 80, 201, 239, 248, 296\n\n  Keynes Mr., on problems of induction, 269 ff.\n\n  K\u00f6hler, 37 ff.\n\n  Knowing, as mental occurrence, 202\n\n  Knowledge, as displayed in reactions to environment, 17 ff.\n    perceptual, 58 ff.\n    behaviourist view of, 88 ff.\n    difference between introspective and other, 215\n    _a priori_, 249 f.\n    limitations on, imposed by structure of language, 264 f.\n\n  Knowledge-reaction, 216, 282\n\n\n  Language, as a bodily habit, 43 ff.\n    psychological side of, 48\n    words in an ideal logical, 256 f.\n    and things, relation between, 264\n\n  Laws, causal, 144 ff.\n    evidence for, 147\n    universal characteristics of, 149\n\n  Learning, laws of, 23, 29 ff.\n    two ways of, 39\n    in infants, 41, 48\n    by increase of sensitivity, 95 f.\n\n  Leibniz, 239, 241 f.\n\n  _Le Probl\u00e8me logique de l\u2019induction_ (Jean Nicod), 269, 273\n\n  Locke, 244 ff.\n\n  Logic, 263, 296\n\n  \u201cLogical atomism\u201d, 248\n\n\n  Mach, 214, 292\n\n  Man, his relation to the Universe, 292, 295, 298 ff.\n\n  Materialism, as a philosophy, 159\n\n  _Mathematical Theory of Relativity_ (Eddington), 283\n\n  Matter, the structure of the atom, 98 ff.\n    essence of, 146 f.\n    as conceived in modern physics, 157, 293\n    old view of, now untenable, 158 ff.\n    constructed out of events, 278\n    permanence of, only approximate, 279\n    possibly a structure of mental units, 290\n\n  Maxwell\u2019s equations, 107, 145\n\n  Meaning, 52, 71, 82\n\n  Meinong, 202\n\n  Memory, behaviourist theory of, 71 ff.\n    its reference to the past, 188 ff.\n    feeling of pastness complex, 190\n    more fundamental than imagination, 190\n    vital difference between imagination and, 194\n    Dr. Broad\u2019s view on reference to the past, 195\n    stages of, 195 ff.\n    immediate, 196 f.\n    true recollection, 197 ff.\n    trustworthiness of, 199\n\n  Memory and testimony, 5 ff.\n\n  Mendeleev, 99\n\n  \u201cMental\u201d events, 114, 141 f., 280 f.\n\n  \u201cMental\u201d occurrences, 201, 212\n\n  _Mentality of Apes_ (K\u00f6hler), 37 ff., 62\n\n  Mill, J.\u00a0S., his canons of induction, 269 f.\n\n  Mind and matter, conventional notions of, 141\n    distinction between, illusory, 142, 201\n    gap between, how filled in, 148\n    interaction between, 150\n    theory of \u201cneutral monism\u201d, 206 ff.\n    Cartesian dualism, 239\n    Leibniz\u2019s theory of, 241\n\n  Mind, a cross-section in a stream of physical causation, 150\n    modern conception of, 280 ff.\n    emergent from events, 284\n    definitions of a, 285 ff.\n\n  Minkowski, 239\n\n  _Mneme_ (Semon), 49\n\n  \u201cMnemic\u201d effects, 49, 209, 295\n\n  \u201cMnemic\u201d occurrences, 49, 180 f.\n\n  Monads, 241\n\n  Monists and pluralists, controversy between, 251 ff.\n    pluralism the view of science and common sense, 253\n\n  Moore, Dr. G.\u00a0E., on notion of \u201cgood\u201d, 230\n\n  \u201cMoral issues\u201d, 227\n\n  Motion, 119, 163\n\n  Mystics, 229, 264, 300\n\n\n  Names, 53\n\n  Necessity, anthropomorphic notion of, 115, 117\n\n  \u201cNeutral monism\u201d, theory of, 206 ff., 210, 282, 292\n\n  Newton, 242\n\n  Nisbet, R.\u00a0H., on probability, 275\n\n\n  Object, what happens when we see an, 146 f.\n\n  Objective and subjective study, 30\n\n  Objectivity, 154 f., 169\n\n  Ogden and Richards, Messrs., 52\n\n\n  Parmenides, monistic view complete in, 251\n\n  Parry, Professor R.\u00a0B., 292\n\n  Perception, difference between introspection and, 10 f.\n    a species of sensitivity, 59, 123\n    and inference, 65 f.\n    from objective standpoint, 66 ff.\n    of external event, analysis of, 123 ff.\n    element of subjectivity in, 130 ff.\n    and causal laws of physics, 145 ff.\n    its relation to the object causal and mathematical, 149\n    from introspective standpoint, 201 ff.\n\n  Perceptive knowledge, stages in act of, 18 ff.\n\n  Percepts, 133, 135, 137 ff.\n\n  Perspective, 152\n\n  Philosophy, the business of, 2, 236\n    Behaviourism as a, 129 ff.\n    Utilitarian, 229 f.\n    systems of Descartes, Spinoza, and Leibniz, 237 ff.\n    Locke, Berkeley, and Hume, 244 ff.\n    conscious purpose of, 299\n\n  Physics, modern, 97\n    causal laws in, 114 ff., 145 ff.\n    and perception, 123 ff.\n    spatial relations in, 137 ff.\n    our knowledge of, 151 ff.\n    only mathematical properties of, discoverable, 157\n    less deterministic than formerly, 239\n    and psychology, 282, 289\n\n  Pictures, as representations, 183\n\n  \u201cPlanck\u2019s Constant\u201d, 101 f.\n\n  Plato, 226\n\n  Poetry, 220\n\n  Probability, fundamental in science, 274\n    _a priori_ on Mr. Keynes\u2019s theory, 274\n    \u201cfrequency\u201d, theory of, 274 f.\n\n  Psychology, 16, 172, 184\n    and physics, 282, 289\n\n  \u201cPsychophysical parallelism\u201d, 238\n\n  \u201cPublic good\u201d, the, 230\n\n  Publicity, in the case of physical phenomena, 170\n\n\n  Quantum changes, 106\n\n\n  Radio-activity, 99, 103\n\n  Reactions, learned, 21, 33, 35, 36, 49, 81\n\n  Realism, naive, 175\n\n  Recognition, two forms of, 196\n\n  Recollection, true, 197 ff.\n\n  Relations, Bradley\u2019s argument against, 252\n    cause of confusion about, 264\n\n  Relativity, theory of, \u201cspace-time\u201d instead of one cosmic time and\n        space, 108\n    some results of the, 108 ff.\n    \u201cevents\u201d instead of bodies moving, 110\n    relations between \u201cevents\u201d, 110 f.\n    no \u201cforces\u201d in the, 111\n    philosophical consequences of the, 293\n\n  \u201cRight conduct\u201d, 230\n\n  Rutherford, Sir E., 99, 101\n\n\n  Santayana, Mr., 230\n\n  Schiller, Dr. F.\u00a0C.\u00a0S., 79\n\n  Schr\u00f6dinger, 98, 105, 278, 293, 294\n\n  Self-observation, 126, 161 ff.\n    basis of Descartes\u2019s system, 162 ff.\n    Dr. Watson\u2019s views, 167 ff.\n    gives knowledge not part of physics, 175\n\n  Semon, 49, 180\n\n  Sensation, difference between images and, 179\n    acoleuthic, 197\n    as opposed to perception, 204\n\n  Sensitivity, 59 f., 88, 123, 177\n\n  Sentences, 51, 54, 75, 255, 264\n\n  Sequence, laws of, 116\n\n  Shakespeare, 192\n\n  Sheffer, Dr. H.\u00a0M., 282, 292\n\n  Sight, compared with touch, 156 f.\n\n  Size, sense of, 153\n\n  Socrates, 226\n\n  \u201cSolipsism\u201d, 291\n\n  Sommerfeld, 103\n\n  Space, one persistent, abolished in relativity theory, 108\n    physical and perceptual, 137 ff., 241 f., 294\n\n  Space-time, in theory of relativity, 108 ff.\n    structure of, 145\n    point-instant in, 278\n\n  \u201cSpecious present\u201d, 195, 197\n\n  Spinoza, 238, 251\n\n  _Stars and Atoms_ (Eddington), 279\n\n  \u201cStatement\u201d, definition of a, 260\n\n  Subjectivity, 129, 133, 135, 154 f.\n\n  Substance, 5, 242 ff., 293\n\n  Syllogism, the, 80\n\n  Syntax, influence of, on philosophy, 243\n    connection between laws of physics and laws of, 263\n\n\n  Talking without thinking, 190\n\n  Tendency, quantitative laws of, 144\n\n  Testimony, 11 f., 170\n\n  _The Analysis of Matter_ (Bertrand Russell), 278\n\n  _The Meaning of Meaning_ (Ogden and Richards), 52\n\n  _The Mind and Its Place in Nature_ (Dr. Broad), 76, 188, 282\n\n  Thorndike\u2019s \u201cprovisional laws\u201d, 31 ff.\n\n  Thought, 163 ff., 174, 240, 263\n\n  Time, not cosmic, 108 ff., 158\n\n  Touch, compared with sight, 156\n\n  _Treatise on Probability_ (Keynes), 269 ff.\n\n  Truth, 94, 261 f.\n\n  Truth and Falsehood, causes of mystery about, 254\n    two questions in, 254 ff.\n    meaning of a sentence examined, 255 f.\n    grounds on which statements are regarded as true or false, 257\n    ultimate test of falsehood, 258\n    \u201cbelief\u201d, 258 ff.\n    problems of, 259 ff.\n\n\n  Universals, 53, 203\n\n  Universe, the, philosophy concerned with, 236\n    man\u2019s relation to, 298 ff.\n\n  \u201cUnlearned Equipment\u201d, 22\n\n  Utilitarian philosophy, 229 f.\n\n\n  Vitalists, 25\n\n  Volition, 61\n\n\n  Watson, Dr. J.\u00a0B., 10, 21, 22, 31, 33, 35, 36, 37, 70 ff., 126 ff.,\n        162, 167 ff., 177, 188, 219, 223, 259\n\n  Waves in empty space, 107 f.\n\n  Whitehead, Dr., 159\n\n  \u201cWill\u201d, 223 f.\n\n  Willing, as mental occurrence, 202\n\n  _Winds of Doctrine_ (Santayana), 230\n\n  Wish-fulfilment and dread-fulfilment, 194\n\n  Wittgenstein, 264\n\n  Words, purpose of, 11 f.\n    as physical occurrences, 44 ff.\n    spoken and written, 46 f.\n    how acquired by infants, 48 ff.\n    meaning of, 52, 256\n    relations of, 56\n    in an ideal logical language, 256 f.\n\n  World, the physical, nature of our\n    knowledge of, 151 ff.\n    a four-dimensional continuum of events, 293\n    our knowledge of, purely abstract, 295\n\n\nTHE END\n\n\n\n\n[Illustration]\n\nNORTON BOOKS\n\n  Paul Bekker            THE STORY OF MUSIC        $3.50\n\n  Everett Dean Martin    PSYCHOLOGY                 3.00\n                         THE MEANING OF A\n                           LIBERAL EDUCATION        3.00\n\n  Daniel Gregory Mason   ARTISTIC IDEALS            2.50\n\n  H.\u00a0A. Overstreet       INFLUENCING HUMAN\n                           BEHAVIOR                 3.00\n                         A BOOK ABOUT OURSELVES     3.00\n\n  Bertrand Russell       PHILOSOPHY                 3.00\n\n  Thomas E. Tallmadge    THE STORY OF ARCHITECTURE\n                           IN AMERICA               3.50\n\n  John B. Watson         BEHAVIORISM                3.00\n\n\n                        _THE NEW SCIENCE SERIES_\n                         EDITED BY C.\u00a0K. OGDEN\n\n  C. Judson Herrick      FATALISM OR FREEDOM        1.00\n\n  I.\u00a0A. Richards         SCIENCE AND POETRY         1.00\n\n  Vilhjalmur Stefansson  THE STANDARDIZATION OF\n                           ERROR                    1.00\n\n         _Announcements of other volumes in this series, and a\n          complete catalogue of Norton Books sent on request_\n\n\n                     W \u00b7 W \u00b7 NORTON & COMPANY, INC.\n                     70 Fifth Avenue      New York\n\n\n\n\nTranscriber\u2019s Notes\n\n\nHyphenation, and spelling were made consistent when a predominant\npreference was found in the original book; otherwise they were not\nchanged.\n\nOriginal text mostly placed commas and periods after closing quotation\nmarks, but occasionally placed them before closing quotation marks.\nThis inconsistency has not been changed here.\n\nSimple typographical errors were corrected; unbalanced quotation\nmarks were remedied when the change was obvious, and otherwise left\nunbalanced.\n\nFootnotes have been sequentially renumbered and placed just below\nthe paragraphs that reference them.\n\nThe index was not checked for proper alphabetization or correct page\nreferences.\n\nPage 86: The variables in the equation were printed in italics. To\nimprove readability, the underscore characters used elsewhere to\nindicate italics have been omitted in this case.\n\nPage 207: \u201cthis is a causal word\u201d originally was printed as \u201cthis a a\ncausal word\u201d; changed here by Transcriber.","title":"Philosophy","tradition":"philosophy-ethics","translated_text":null,"word_count":108309}