by C. D. Broad
Trinity College, Cambridge.

Published in Inquiry I (1958): 99-129. Page numbers from the published version are inserted in brackets. These are to be understood as occurring at the top of the page. Transcribed into hypertext by Andrew Chrucky, Aug. 1, 2001.

    Anyone who is a professional teacher of philosophy must be well aware that many of his colleagues regard his subject as at best extremely queer and at worst completely bogus. He knows too that this impression is not confined to stupid or ignorant or unfriendly persons, whose opinions on the matter he can lightheartedly set aside. We may usefully begin by asking ourselves what are the causes of this unfavourable estimate, and how far, if at all, they are valid reasons.

    We may remind ourselves at the outset that doubts about the validity and the fruitfulness of philosophy, as hitherto pursued, have not been confined to outsiders. At least three of the most eminent European professionals, Descartes, Locke, and Kant, were moved to initiate the philosophic developments for which they are famous by their profound dissatisfaction with all that had gone under the name of 'philosophy' before their time. It is true that each of them claimed to have found a remedy and to have set philosophy at long last on the right course. But their successors can derive little comfort from this. For, whilst we cannot but admit that each of these great men was largely justified in his criticisms of philosophy as practised up to his time, we must also own that the remedy proposed by each of them has to all appearance failed to cure the disease. The philosophies of Descartes, of Locke, and of Kant seem to be open to very much the same kind of objections as these writers launched against the philosophy in which they were brought up.

    These objections may be summarized as follows.

    (1) Philosophy claims to be a branch of knowledge or of well-founded belief. It is not content to be regarded as mythology or poetry or the expression of a personal emotive reaction. Now, if this claim were valid, we should [100] expect philosophy to be a cooperative work, in which generally agreed results are reached by common and accepted methods. There would, no doubt, be plenty of legitimate disagreement. There would also be a certain amount of heated and needless controversy, due to common human defects and to the personal failings of particular individuals. All this we find in other branches of human enquiry, such as mathematics, natural science, and history. But we should expect the controversies to have led to a very large amount of agreement at any moment among experts, and we should expect such agreed results to form a firm basis for further developments. Now, it will be said, we do not find this or anything like it in philosophy. At any moment we find philosophers divided into sects and schools, whose problems, methods, and very language are so alien to each other that mutual understanding is difficult and fruitful discussion and cooperation impossible. One has only to note, e.g., the coexistence at the present time of a school of logical positivists and a school of existentialists, in order to illustrate this contention. At most we may find one fashion prevailing for some years and over a considerable area, and ousting temporarily another fashion which sooner or later returns. In illustration of this it is amusing to compare and contrast the titles of articles in the famous philosophical journal MIND to-day and fifty years ago. At the beginning of this century many pages in each number were devoted to the discussion of intimate details in the life of the Absolute. Nowadays it is rare indeed for the very phrase 'the Absolute' to occur in a number of MIND, except perhaps in a book-review. Now no-one in his senses would allege that, e.g., Bradley and Bosanquet and McTaggart and Royce were any less able or learned or devoted to their subject than the leaders of English and American philosophy to-day. The inference which suggests itself is that there is something radically wrong with the subject.

    It is worth while to remark that Descartes and Bacon quite fairly brought a similar objection against the natural science which had prevailed up to their time. They thought that the defect could be remedied by the introduction of certain new methods, and the subsequent history of natural science has abundantly confirmed their anticipations. Descartes, who had himself invented a method which revolutionized the treatment of geometrical problems, had similar high hopes for philosophy. It must be admitted that these hopes have been completely disappointed. Yet we can hardly be surprised that the wonderful transformation wrought in geometry and in physics by the introduction into each of an appropriate new method should have encouraged in- other philosophers too, [101] such as Locke and Kant, the belief that their own subject might in a similar way be 'set in the sure path of science'. The repeated frustration of such hopes strongly suggests that the analogies which seemed to justify them were illusory. Either philosophy is not a branch of knowledge at all, or, if it is, it is radically unlike both pure mathematics and natural science.

    (2) I think that the considerations which I have outlined constitute the main cause of wide-spread dissatisfaction with philosophy. Critics who are familiar with the history of scientific thought in Europe may be inclined to reinforce these objections with the following reflexions. They will say that the various branches of human knowledge which are now recognized sciences, e.g., astronomy, mechanics, and formal logic, began by being parts of philosophy. So long as they remained in that state they exhibited all the characteristic defects which have been noted above. But sooner or later the appropriate concepts, methods, and principles were discovered in each case, and thereafter the subject began to progress and at the same time ceased to be a part of philosophy. It might be alleged with some plausibility that we can see this process taking place in our own day with psychology. Does not this strongly suggest that philosophy is not a genuine branch of human knowledge at all, but is simply the ever-diminishing field within which baseless speculation and profitless controversy still have free play because the appropriate conditions for scientific treatment have not yet been elicited?

    (3) If one considers the way in which philosophy is taught to students, and contrasts it with the methods of teaching in subjects which are admittedly genuine branches of human knowledge, one may well be confirmed in these suspicions. A considerable part of the training of a student of philosophy consists in studying the writings of eminent philosophers of the past, such as Plato, Aristotle, Descartes, and Kant. But it is no part of the training of a mathematician or physicist to read and criticize the works of Archimedes, of Newton, of Faraday, and of Maxwell. The methods and results of these great men, so far as they have proved to be valid and useful, have been embodied in the science and now appear in simplified and improved form in contemporary textbooks. If the original writings are now studied at all, this is only because of their historical interest as milestones on the road of mathematical or physical progress. Does not this peculiarity in the method of training students of philosophy reinforce the suspicion that the subject is not a branch of human knowledge but only a collection of obiter dicta dressed up in argumentative form?

    [102] Many professional philosophers nowadays would be ready to admit that there is a great deal of truth in these contentions, and would be content to make very modest claims for their subject. An influential contemporary school, with many very able adherents in England and the U.S.A., would reduce philosophy to the modest task of attempting to cure the occupational diseases of philosophers. In their writings the word 'philosopher' is commonly used to denote the holder of some opinion, or more accurately the utterer of some sentence in the indicative mood, which the writer regards as characteristically fatuous. If this is what one thinks about one's own occupation, it is certainly honest to announce the fact. It is not for me to judge whether it is altogether prudent for professional philosophers thus publicly to proclaim that their business is to take in and wash each other's dirty linen. Nor will I speculate on how long an impoverished community, such as contemporary England, will continue to pay salaries to individuals whose only function, on their own showing, is to treat a disease which they catch from each other and impart to their pupils.

    What in older idioms were called 'philosophical problems' are often referred to by such philosophers as 'worries', with the implication that they are comparable to the pimples and emotional upsets of adolescence. The 'treatment' proposed for these 'worries' ranges from the agonized soul-searchings of the late Professor Wittgenstein and his pupils in Cambridge to the avuncular man-of-the-world good sense of Professor Ryle in Oxford. We might perhaps respectfully summarize the latter prescription in the phrase: "I've been through it myself in my time, my boy, and I know that it hurts like Hell; but I came through, and so will you if you'll only look at it sensibly!"

    On the other hand, philosophers nowadays find themselves repeatedly admonished, not only by well-meaning non-professionals, but also by some of their professional colleagues, to be up and doing and to deliver a message which shall restore to humanity its lost spiritual values and save civilization. How gladly would I deliver such a message, if I had one which I believed to be well-founded, and if I saw any reason to think that it would not be either ignored or misunderstood or deliberately suppressed or distorted where it was most needed!

    Those who make this demand may be invited to consider the following facts. In the first place, even if it be part of the business of philosophy to produce a Weltanschauung, it is no part of the business of any philosopher to produce one on demand in order to pop a leak in the ship of civilization. Secondly, there is no guarantee whatever that an honest [103] attempt to philosophize will lead to a view of the world and of human nature and society which liberal-minded contemporary Western Europeans and Americans would find to their liking. Plato, Spinoza, Hobbes, and Hegel were all great philosophers who had thought very deeply on these high matters, and their philosophies constitute fairly impressive instances to the contrary. Lastly, we may remind ourselves that the cruelty and violence and loss of liberty, which now prevail in a great part of the world, are the products of a fanatically held philosophical Weltanschauung. If we want to find an historical parallel to them, we have only to go back to the seventeenth century in Europe. Now at that time all those philosophical doctrines, to the decline of which our present troubles are often confidently ascribed, were in full bloom. Almost everyone that mattered believed quite seriously in the existence and providence of God, in the immortality of the human soul, in an objective system of moral law, and in rewards and punishments in a future life. The prevalence of these elevating and consolatory views of man and the universe did not prevent the sack of Magdeburg or the reduction of parts of central Europe to starvation mitigated by cannibalism. Indeed, if these doctrines had never been heard of, the combatants in the 30 Years' War would have needed to seek, and would doubtless have found, other pretexts for fighting and persecuting.

    I think that we may at once concede the following points to the critics of philosophy. The facts adduced by them are largely genuine. These facts do make it certain that philosophy, as hitherto pursued, is not a science or a collection of several sciences, either in the sense in which pure mathematics is so, or in the sense in which physics, chemistry, biology, etc., are so. Some of these facts, moreover, make it almost certain that philosophy never can become a science in either of these senses, and that hopes that it may do so, or regrets that it has not done so, are based on a complete misunderstanding of its nature.

    Whilst accepting all this, I would like before going further to make the following three remarks.

    (1) We must distinguish between being non-scientific and being un-scientific. What I have admitted is that philosophy is a subject which is almost certainly of its very nature non-scientific. We must not jump from this purely negative statement to the conclusion that it has the positive defect of being unscientific. The latter term can be properly used only when a subject, which is capable of scientific treatment, is treated in a way which ignores or conflicts with the principles of scientific method.

    (2) One of the adverse comments which we noted above was that [104] philosophy makes no progress and that opposite views on the same question simply alternate with each other again and again. I think that, at any rate in some important instances, this is a superficial appearance which is greatly modified by closer inspection and more intimate knowledge of the facts.

    We must first distinguish here two different cases, viz., certain very general and far-reaching alternative types of philosophic theory, and alternative views on certain fairly specific problems. As an example of the former we might take what I call 'Substantival Monism' and 'Substantival Pluralism', with the philosophy of Spinoza and that of Leibniz as typical and extreme instances of each. As an example of the latter we might take the cluster of interconnected problems about the nature of sense-perception and its objects.

    As regards alternative theories of the first kind, I think we must expect them to co-exist with each other at every period in the history of philosophy. Unity and plurality, in intimate correlation, are obvious features in the world as it presents itself to any reflective person. The aspect of unity appeals intellectually and emotionally to thinkers of one temperament, and that of plurality appeals equally strongly to those of another. Each type of theory is likely to be held, though in various specific forms, by equally intelligent philosophers at every period of history. Neither party is in the least likely to convert the other to its own view by argument. At best, argument may be used to show that a particular theory of one or other of these alternative types fails to do justice to a certain important aspect of unity or of plurality, as the case may be. Apart from this it is a waste of time to argue. One has to make one's choice of one or another of these types of theory on grounds which cannot be presented in the form of arguments appealing equally to all intelligent and instructed persons. Having done this, one may try to develop the chosen type of theory into as coherent and comprehensive a specific form as possible. In that undertaking there is plenty of room for argument, and there is no reason why there should not be progress.

    As an example of the second case we might take the controversies about sense-perception and its objects which have so much occupied English philosophers from the time of Locke to the present day. It might well seem to an unsympathetic external observer that idealistic and realistic answers have merely alternated with each other and that no progress has been made. Now this is certainly a mistake. The course of philosophical thought here, amongst really able and expert thinkers at any rate, has taken the form of a spiral and not of a mere circle. [105] Any competent philosopher who puts forward either a realistic or an idealistic answer to these questions nowadays must state it in a far more subtle and qualified form than his remote predecessors would have done, just because of the distinctions which have been drawn and the criticisms which have been made by intermediate philosophers in the course of their controversies with each other. In consequence of this, his theory, of whichever type it may be, should at least be immune to many of the objections which have been legitimately brought against earlier forms of the same type of theory.

    Another excellent illustration of the same fact is provided by the controversies in regard to phenomenalism and the so-called 'verification-principle' which have taken up so much of the energies of philosophers in England and the U.S.A. during the last twenty-five years. I think it might fairly be said that phenomenalism began as a new gospel, that its adherents strove with immense ingenuity and evangelical fervour to apply the verification-principle to cases which seemed prima facie to be most recalcitrant, and that in the end it has had to be admitted that a satisfactory phenomenalist analysis of, e.g., material-object propositions or memory propositions, cannot be given. To an unsympathetic observer it might well seem that we were simply back at where we started, and that a number of very able men, who might surely have been better occupied, had wasted a quarter of a century in desperately defending positions which any sensible person could have seen at the outset to be untenable.

    I am quite sure that such a judgment is superficial and mistaken, though I must confess that I have sometimes myself felt inclined to say such things in my haste. The fact is that, in the course of the defense and the attack, ambiguities in the at first vaguely formulated verification-principle have been noted and cleared up, and much light has been thrown on the conditions which must be fulfilled by any satisfactory analysis of material-object propositions or of memory propositions.

    I think that the fate of the verification-principle provides an instance in favor of a rather depressing generalization which the history of philosophy suggests. It is this. Time and again some new general principle is brought forward with a flourish of trumpets. As originally formulated it is vague and ambiguous, but this is not at first apparent. In the course of controversy the ambiguities are cleared up, and then it is found to cover two or more alternatives. On one interpretation it is certainly true, but is little more than a tautology and has no interesting consequences. On the other interpretation it is synthetic and would have [106] exciting consequences, but there is no reason to think it true. The time for evangelical fervour is when the ambiguity is still undetected and the principle seems to be at once evidently true and fruitful in consequences. This depressing generalization extends beyond the field of philosophy. It is admirably illustrated, e.g., by the so-called 'economic interpretation of history'. Such illusions are not to be despised. They may be compared with those which cause men to marry and beget children. They are carrots which nature dangles before the noses of philosophers, luring them on to do much valuable thinking which would otherwise never have been undertaken.

    (3) The last remark which I will make here is the following. There certainly are changes of fashion in philosophy. When a new kind is in vogue, many things in the philosophy of an earlier day, which are of permanent value and perhaps highly relevant to contemporary problems, tend to be altogether forgotten or carelessly and ignorantly dismissed, simply because they occur in an out-of-date setting and are clad in an unfashionable dress. It is now quite certain that much of permanent value in Scholastic philosophy was ignored or contemned from this cause by Locke, Berkeley, Hume, and their followers. I have little doubt that the same is true mutatis mutandis of the attitude of many present-day philosophers towards the systems of monistic idealism which were fashionable at the beginning of this century. It is consoling to a philosopher's vanity not to pry too closely into the history of his subject, for otherwise he is liable to find that his discoveries have been anticipated and his fallacies refuted in advance by predecessors whom he has ignored or despised.

    I want now to consider in some detail a view of the nature of philosophy which has been accepted in recent years by a number of very able thinkers in England and the U.S.A. It may be formulated roughly as follows. The sole or the main business of philosophy is to analyze the various kinds of proposition which constitute the common-sense view of the world. For many philosophers this line of thought originates in the paper A Defence of Common-Sense, which was Professor G. E. Moore's contribution to the collection of essays published in 1925 under the title Contemporary British Philosophy. It will be worth while to devote some careful attention to this famous and very influential essay of Moore's.

    Moore does not attempt to define the word 'common-sense' or the phrase 'a common-sense proposition'. Instead he begins by enumerating a long list of propositions, each of which would, I think, readily be [107] admitted to be a common-sense proposition. In order to classify them briefly and intelligibly I will begin by introducing a few simple technical terms. Let us call a proposition 'epistemic' if and only if it asserts that a certain individual, or some one or other, or everyone knows a certain proposition or knows propositions of a certain class. Thus the following propositions would be epistemic: Jones knows that the angles at the base of an isosceles triangle are equal; Everyone knows that twice two is four; Some people know all the propositions in the First Book of Euclid. On the other hand, the proposition 2 X 2 = 4 would be non-epistemic. An epistemic proposition may be of the first or the second or of a still higher order. The examples which I have given are all of the first order. But the proposition: I know that Jones knows that 2 X 2 = 4, is an epistemic proposition of the second order.

    The first point to be noted is that Moore's propositions fall into two primary groups. The first cousins of a large number of non-epistemic propositions. The second consists of a single epistemic proposition of the first order concerning the members of the first group.

    The non-epistemic propositions which constitute the first primary group may themselves be subdivided first into physical and psychological propositions. The physical propositions may then be subdivided into autobiographical, heterobiographical and non-biographical. The auto-biographical physical propositions are assertions by Moore himself of the existence, and certain spatial and other relationships, of Moore's own body during a certain period. The heterobiographical ones are similar assertions made by Moore about other living human bodies and the same or earlier periods. The non-biographical ones are similar assertions rnade by Moore about certain ostensibly non-living bodies, e.g., the earth and the sun, and the same or earlier periods. Finally, the psychological non-epistemic propositions may be subdivided into a number of autobiographical ones and a single heterobiographical one. The autobiographical ones are assertions by Moore that he himself has from time to time had experiences of each of certain specific kinds. The single heterobiographical one is an assertion by Moore, with regard to the human bodies other than his own, whose existence he asserted in the heterobiographical physical propositions. It is that many of these have been the bodies of persons who have had during the lifetime of their bodies experiences of the various kinds which Moore himself has had during the lifetime of his present body.

    So much for the first primary group, i.e., the non-epistemic propositions. We come now to the second primary group, which, as I have said, [108] consists of a single epistemic proposition of the first order. This may be put as follows. It is true of many of the persons other than Moore himself, whose existence is implied by the heterobiographical psychological proposition just stated, that they have often during the life time of their present bodies known propositions corresponding mutatis mutandis to each of those enunciated in the first primary group. The mutanda here are of course such terms as the person denoted by the proper name 'Moore', the date at which Moore was born, the date at which he wrote the essay, and so on, in the autobiographical propositions.

    Now Moore claims to know many propositions in each subdivision of the first group. Since this is an assertion of knowledge, it is an epistemic proposition. Since all the propositions in the first group are non-epistemic, it is an epistemic proposition of the first order. But Moore also claims to know the one proposition in the second group. Since this is itself an epistemic proposition of the first order, the assertion that Moore knows it is an epistemic proposition of the second order.

    We may sum up this complicated business as follows:

    (1) Moore claims to know that he has a body of a certain kind, viz., a living human body, and that he has had experiences of certain kinds. He claims to know that there are other living human bodies beside his, and that there are bodies of other kinds, e.g., trees, chairs, etc. He claims to know, with regard to each living human body, that it is the body of a person who has had experiences of the same kinds as he himself has had. {All this is an epistemic proposition of the first order.}

    (2) Moore daims to know, with regard to each such person, that that person knows facts corresponding mutatis mutandis to each of the kinds of fact enumerated above which Moore himself claims to know. {This is an epistemic proposition of the second order.}

    The next point to be noted is this. The word 'know' is used in English in such a way that it would be nonsensical to say of a proposition that it is known by someone but may yet be false. Therefore anyone who admits Moore's claims must admit that there are true propositions in all the subdivisions of Group I. He must also admit that the proposition which is the only member of Group II is true. There are two ways in which he might seek to whittle down this admission. One is to say that in each of the final subdivisions enumerated above there are propositions which are at any rate partly true, though even one of them is also partly false. The other is to say that for the propositions in each of the final subdivisions it is possible to find an interpretation in which some of them are wholly true, though in their ordinary sense they are [109] all false. We must note that Moore explicitly rejects both these expedients. He is committed to holding that in each of the final subdivisions there are propositions which, when taken in their ordinary sense, are wholly true.

    This last point deserves further consideration. Moore asserts that each of the common-sense propositions which he has taken as examples, and which he claims to know, has one and only one meaning, and that everyone who knows English understands that meaning. Take, e.g., the proposition: There is a penny on the table before me now. Moore would say that this has one and the same meaning (allowing for the systematic ambiguity of the words 'I' and 'now') for everyone who understands English. But he says that there might be, and in fact are, great differences of opinion as to the right analysis of this and similar propositions, and he does not claim to know the right analysis of them. Unfortunately he does not further discuss these two vitally important notions of the meaning and the correct analysis of a proposition, nor does he explain in any detail how he supposes the two to be connected. All that he says on that point is this. It is impossible to raise the question: What is the correct analysis of such and such a proposition?, unless one understands the meaning of the proposition.

    On this I would make the following comments.

    (1) Surely it is not propositions, but sentences, which can significantly be said to have a meaning. A proposition is what a sentence means, or more accurately it is the common meaning of a whole multitude of equivalent sentences, spoken or written, in English and in German, and so on. If this be so, we must substitute for Moore's original assertion some such statement as follows. Any person who knows English will think of one the same proposition whenever he sees or hears or utters or images an instance of the type-sentence 'There is a penny on the table before me now'. Now this seems to me to be an empirical assertion about the behaviour of members of a certain class of men when placed in situations of a certain kind. I do not see how Moore or anyone else can be justified in affirming it with complete confidence. On general grounds it seems to me highly doubtful. The utmost that I could admit is that we could describe fairly accurately the kind of circumstances which would be necessary, though not sufficient, conditions for a sane waking adult Englishmen to utter or to write (as distinct from babbling or scribbling) an instance of this type-sentence. We could also describe fairly accurately the sort of circumstances which would be held to be relevant for testing the truth or falsity of such an utterance.

   1 (2) Since a proposition cannot significantly be said to have a meaning, we cannot significantly contrast the meaning of a proposition with its analysis. We must substitute for this the contrast between a proposition itself and that set of interconnected propositions which together constitute the correct analysis of it. For Moore's assertion about meaning and correct analysis we could then substitute the following statement. Although everyone who understands English thinks of one and the same proposition whenever he sees or hears or utters or images an instance of the type-sentence 'There is a penny on the table before me now', yet there is no agreement as to the set of interconnected propositions which together constitute the correct analysis of that proposition. I have already commented adversely on the first part of this assertion. Let us, however, waive these objections and consider the second part of it.

    (3) I suppose it would be admitted that in order for a conjunction of propositions p1 & p2 & ... pn to count as the correct analysis of a proposition p the following conditions are necessary.

  1. That if p is true, then p1 & p2 & ... pn is true, and conversely.
  2. That this equivalence is not merely contingent, like the equivalence between being a ruminant and having cloven hoofs, but is in some sense necessary.
  3. That the equivalence, though necessary, is not merely linguistic. By this I mean that it really is a case of two different propositions, viz., p, on the one hand, and p1 & p2 & . . . pn, on the other, and not merely a case of two different sentences, e.g., 'This is a negro' and 'This is a black man', which in a certain language stand for one and the same proposition.

    Now, although these conditions are severally necessary for a set of propositions to count as the correct analysis of a given proposition, I think it is certain that they are not jointly sufficient. Some further condition seems to be needed, and I do not know what it is. I think that this can be made plain by the following example.

    Take the sentence 'n is a prime number'. Here I think it is probably true that practically everyone who understands English and has learned elementary mathematics does attach one and the same meaning to it, viz., that n is an integer which is not exactly divisible by any other integer. Now consider the following sentence: -- 'The immediate successor of the product of all the integers less than n is divisible by n'. I think it is plain that this really does mean a different proposition from that which is meant by the sentence 'n is a prime number'. So our third condition is fulfilled.' Now it can be proved that the proposition which is meant by this sentence entails and is entailed by the proposition which [111] is meant by the sentence 'n is a prime number'. (This mutual entailment is known as Wilson's Theorem.) So the first and second of our conditions are also fulfilled. But, although all three conditions are thus fulfilled, would anyone be prepared to say that this complicated proposition is the 'correct analysis' of the proposition that n is a prime number? I do not think that anyone would.

    If anyone did, he might be invited to consider the following questions. There are plenty of other complicated propositions which can be shown to entail and be entailed by the proposition that n is a prime number. Are all of them to be counted as correct analyses of the proposition that n is a prime number? If so, the phrase 'the correct analysis' has no application. If, on the other hand, one of them is to be singled out as the correct analysis, whilst the rest are to count only as logical equivalents and not as analyses, on what principle is the distinction to be drawn?

    The only suggestion that I can make here is the following. It might be said that the proposition that n is a prime number does not by itself entail the proposition that the immediate successor of the product of all the integers less than n is divisible by n. The latter proposition is entailed only by the conjunction of the former with the axioms of arithmetic. We are inclined to overlook this because these axioms are themselves necessary propositions which are the common premisses of all deductions within pure arithmetic. Perhaps, then, we ought to add the following as a fourth condition which is necessary if a conjunction of propositions p1 & p2 & .. . pn is to count as an analysis of a given proposition p. The proposition p must suffice by itself to entail the conjunctive proposition p1 & p2 & . . . pn, and similarly the latter must suffice by itself to entail the former. Whether these four conditions are jointly sufficient as well as severally necessary to mark out an analysis, I do not know. Nor do I feel any confidence that the fourth of them is compatible with the third, viz., the condition that analyzandum and analysis shall be two propositions and not just two different sentences which in a certain language mean the same proposition.

    (4) I think that I can sometimes say with confidence that a certain set of propositions is not an analysis at all (and therefore not the correct analysis) of what I understand by a certain sentence. But I cannot state any satisfactory criterion by which I decide this. Still less can I state any criterion by which I could recognize that a certain set of propositions is the correct analysis of what I understand by a certain sentence when that is in fact the case. In view of all this, I do not think that it is [112] particularly illuminating to say that what is meant by such a sentence as 'There is a penny on the table before me now', when understood in its ordinary sense, is and is known to be wholly true, and that the business of philosophy is to seek for the correct analysis of such propositions and not to question their truth.

    It should be noted that Moore himself never alleged that this was the only business of philosophy. Still less did he claim that the philosophical analyses of common-sense propositions must themselves be expressed in the language of common-sense. Some philosophers in recent years seem to me to have written as if they held the latter view. If anyone really does hold it, I can only say that it seems to me completely unjustifiable. The philosophical analysis of common-sense propositions, whatever it may really consist in, is obviously a reflective activity carried out by certain specialists upon materials provided by common-sense. It is no part of the business of common-sense itself. There is therefore not the least reason to expect that the language of common-sense would contain a suitable terminology for expressing the results of philosophical reflexion even on nothing but common-sense propositions. So it is no prima facie objection to a proposed philosophical analysis that it is expressed in technical terms, e.g., 'sense-datum', 'appearance', 'pure ego', etc., which either do not occur at all in common-sense language or occur there in a different and non-technical sense. One might as well object to a theoretical physicist because he uses the word 'entropy', to which no word in everyday speech corresponds, or because he uses the word 'energy' in a specialized technical sense which is very different from that which it has in daily life.

    Certainly it is most desirable to bring high-sounding technical terminology to the test of concrete situations, as Professor Moore has so often done with such devastating effect. It is also true and important to note that a technical term may and often does presuppose a certain theory, and that those who habitually use it are therefore in danger of unwittingly assuming that theory without question, when it is in fact open to doubts and objections. But, provided that a philosopher always bears these facts in mind, he need not reproach himself if he sometimes expresses the results of his reflexions in terms which would be unintelligible to his bedmaker or unfamiliar to his bookmaker. It is as well to remember that the word 'jargon', which is sometimes bandied about in this connexion, is a question-begging epithet. In practice it generally denotes simply the technical terminology favoured by one's opponents.

   [113] I will condude to-day's lecture by summarizing my own reaction to the view that philosophy is or should be the analysis of common-sense propositions.

    I think that one very important part of philosophy is something which it is convenient to call the 'analysis' of various important types of proposition, though I must confess that I cannot give a satisfactory definition of 'philosophical analysis' or a criterion for deciding whether a certain set of propositions is or is not the correct analysis of what is meant by a certain sentence. I also think that one very important class of propositions to be subjected to philosophical analysis are those which plain men and philosophers alike express by the sentences which they utter, and understand by the sentences which they hear, in the most ordinary business of their everyday life.

    I would point out however that, in my opinion, the importance of these latter propositions, and therefore the importance of a correct analysis of them, would be very little diminished if I rejected, as I am inclined to do, Moore's claims to knowledge here. Suppose I were content to say that in my unreflective moments I unhesitatingly believe such non-epistemic propositions as Moore claims that he and I know, including propositions that there are other living human bodies and that many of them are the bodies of persons who have experiences of similar kinds to those which I unhesitatingly believe myself to have had. Suppose, further, I were content to say that in my unreflective moments I unhesitatingly believe that these other persons, whom I unhesitatingly believe to exist and to have experiences like mine, unhesitatingly believe in their unreflecting moments propositions similar mutatis mutandis to those which I unhesitatingly believe in my unreflective moments. Then the importance to me of these propositions, and therefore of a correct analysis of them, would still be very great. And presumably the importance of similar propositions mutatis mutandis to another person would be no less great, if, as I unhesitatingly believe in my unreflective moments, there are other persons like myself.

    However that may be, I do not think that the analysis of propositions of various important kinds is the whole business of philosophy. Nor do I think that the propositions of common-sense, as exemplified by Moore in his list, are the only suitable materials for philosophical analysis. It seems to me that another important part of the business of philosophy is to consider the consistency of common-sense propositions of various kinds with each other and with propositions of other important kinds, e.g., those of mathematics, of natural science, of morality, of psychical [114] research, and of religion. It is also part of its business to consider the internal consistency of each of these systems of propositions.

   Both for this reason, and for its own sake, it is part of the business of philosophy to try to analyze, e.g., scientific propositions and moral propositions, as well as common-sense propositions. Here we shall have to extend the phrase 'analysis of propositions' to cover the attempt to show that certain kinds of sentences in the indicative, e.g., moral ones such as 'Lying is wrong' and 'People ought to pay their debts', do not in fact express or convey propositions at all. This extension can easily be made by substituting for the phrase 'analysis of propositions' the phrase 'analysis of what is expressed by sentences in the indicative'.

   In this lecture I have confined myself in the main to negative and critical statements about philosophy itself and about certain views as to its nature and functions. In the next lecture I shall try to develop and to illustrate the ideas which I have thrown out in the course of my criticism of the view that the sole function of philosophy is to analyze common-sense propositions.


    In the present lecture I shall be covering in part the same ground as in the address, entitled Some Methods of Speculative Philosophy, which I gave at the joint session of the Mind Association and the Aristotelian Society at Cambridge in July 1947. As I have not substantially altered my views since then, it is inevitable that I should to some extent repeat my own words. For this I must apologize to those who happen to have read and to remember that address.

    We may begin with the platitude that philosophy is what philosophers do. If, then, we want to decide what philosophy is, we shall naturally begin by considering what kind of activities have been pursued by men whom everyone would regard as great philosophers when engaged in what everyone would regard as their characteristically professional work. There-would be no great difficulty in giving instances which would satisfy most people. Spinoza and Leibniz were great philosophers; Shakespeare and Gibbon, though men of the highest intellectual calibre, were not philosophers at all. Spinoza was engaged in philosophical work when he wrote his Ethics and not when he wrote his Hebrew Grammar, and [115] Leibniz was so engaged when he wrote his Monadology and not when he wrote his history of the House of Brunswick. But there are undoubtedly plenty of marginal cases. One and the same man may be eminent both as a philosopher and as a scientist or a mathematician. Descartes, Leibniz and Whitehead are obvious examples. Then, again, it may be impossible to draw a sharp line between philosophical and non-philosophical activities in many cases. Einstein, e.g., was obviously a great mathematical physicist and not primarily a philosopher, but it might well be held that the reflexions which led him to formulate first the special and then the general theory of relativity were predominantly philosophical. Leibniz's writings on dynamics and on the differential calculus seem to be an intimate blend of what everyone would call 'science' and what most people would call 'philosophy.

    We may notice next that what would generally be admitted to be philosophical activities seem to cover at least two very different kinds of intellectual undertaking. Hume's attempt to analyze causal propositions and Spinoza's attempt to establish by deductive argument the fundamental nature of God or the Universe are extreme instances of the two. In the philosophical work of most great philosophers these two kinds of activity are blended in various proportions. But it is not at all obvious that there is any necessary connexion between them. It might be held that the former is a practicable and useful activity, which should continue to be pursued; whilst the latter is an impracticable but seductive activity, which should be dropped, and against which the unwary should be warned. (Those who take this view might claim to find analogies in the original amalgam of astronomy with astrology and of chemistry with alchemy. )

    Now I am inclined to think that there is one feature which is characteristic of all work that would generally be regarded as philosophical. This I will call 'Synopsis'. I am inclined to think that there are certain other features, of which the following things may be said. One or more of them must be present, in addition to synopsis, in any work which would be called 'philosophic'. But they need not all be present. In some work which everyone would call 'philosophic' one or more of these features may be absent or evanescent. I distinguish four such features, and I will give to them the following names,

'Analysis of Propositions and Concepts',
'Detection and Formulation of Presuppositions',
'Critical Appraisal of Presuppositions', and
The first three of these are very closely bound up with each other, and together they are characteristic of what I call 'critical philosophy'. The fourth, viz., synthesis, is [116] specially characteristic of what I call 'speculative philosophy'. The rest of this lecture will be devoted to explaining and illustrating the above statements and pointing out inter-connexions between these various features. I will begin with synopsis, which I regard as fundamental.

    There are different departments of fact, or different regions or levels within a single department, which are very seldom viewed together or seen in their mutual relationships by the plain man or even by the professional scientist or scholar. Yet they do co-exist and are relevant to each other and must presumably be inter-connected into some kind of coherent whole. All men at most times, and many men at all times, conduct various parts of their living and thinking in relatively water-tight compartments. They turn blind eyes to awkward, abnormal, or marginal facts, and they skate over the surface of phenomena. A strong and persistent desire to see how the various aspects of experience hang together is perhaps the one characteristic common and peculiar to philosophers. I understand by the word 'synopsis' here the deliberate attempt to view together aspects of human experience which are generally viewed apart, and the endeavour to see how they are inter-connected. To illustrate what I have in mind I will take two of the examples which l used in my address of 1947.

    As our first example we will take the problem of sense-perception. We commonly think of a material thing as something which has all the following characteristics. It is something which has at every moment a certain shape, size, and position in a common three-dimensional space; which has at every moment various qualities, such as colour and temperature, in certain determinate forms; which has various causal properties, such as inertia, impenetrability, elasticity, etc.; and which persists, moves or rests, retains or alters its determinate shape, size, qualities, and causal properties, and interacts with other material things through impact, gravitation, radiation, etc. We commonly think that our senses reveal to us the presence and certain of the qualities, mutual relations, changes, and causal properties of certain bodies. We assume that these bodies would have existed; would have had precisely the same qualities, mutual relations, and causal properties; and would have undergone precisely the same changes, even if they had never been perceived by anyone, human or non-human. We take for granted that at the time when a person sees a certain object, e.g., a star, that object exists and is in the state and at the place in which it then visually appears to him to be. We think that the same person can perceive the same part of the same body at the same time by different senses, e.g., sight and touch; and that different persons [117] can perceive the same part of the same body at the same time by the same sense, e.g., sight.

    Now there as a problem about sense-perception for the following reasons.

    (1) If we attend carefully, we note such facts as these.

  1. Of two persons, who would be said to be seeing the same part of the same thing at the same time, one may see it, e.g., as round and the other as elliptical.
  2. One and the same person, who would be said to be seeing and feeling the same part of the same thing at the same time, may, e.g., see it as elliptical and feel it as round.
  3. One and the same person, who would be said to have been seeing the same unchanged part of the same thing at different times from different positions, may, e.g., see it on one occasion as round and on the other as elliptical.
Common-sense is more or less aware of such minor systematic variations in normal sensible appearances, and it has certain modes of expression for describing them, but in the main it ignores them. Certain sciences and arts, e.g., geometrical optics and perspective drawing, deal explicitly with some of them.

    (2) There are visual perceptual experiences which are abnormal in various ways and degrees, but are similar to and continuous with normal visual perceptions. They range, e.g., from mirror-images and sticks which feel straight but look bent when half immersed in water, through 'seeing double' when one eyeball is pressed aside or the percipient is drunk, to dreams and full-blown waking hallucinations. Now those which come at the latter end of this scale cannot plausibly be interpreted in the naively realistic way which ordinary language inevitably suggests for normal sense-perceptions, and yet qualitatively the series is continuous from one end to the other.

    (3) There are certain highly relevant facts which are still quite unknown except to a minority of grown-up and educated persons. These must have been completely hidden from everyone at the time when the language in which we express our perceptual experiences was first formed and for thousands of years afterwards. One of these is the physical fact that light takes time to travel, and that a visual experience referring to an event in a remote object does not begin until light which left that object simultaneously with that event reaches the percipient's eye. A consequence is that an experience which would naturally be described as 'seeing a certain remote object at a certain moment' does not guarantee the existence of any such object at that moment, nor does it reveal the contemporary shape, size, position, colour, etc., of such an object if it does still exist. At the best such an experience guarantees [118] only the existence of such an object in the more or less remote past, and reveals only the shape, size, position, colour, etc., which it then had. To this physical fact we must add the physiological fact that visual appearances vary with certain differences in the percipient's eye, optic nerve, and brain, even when the retinal stimulus is the same. We must also add the psychological fact that visual perceptions are determined to some extent by the percipient's past experiences and present mental attitude and expectations.

    Now there is a philosophical problem of sense-perception for those and only for those who try to envisage all these facts together and to give an account of sense-perception and its objects which fits them all into a coherent pattern. The language in which we express our sense-perceptions and talk about material things was formed unwittingly in pre-historic times to deal in a practical way with a kind of normalized extract from our total perceptual experience. It was formed in utter ignorance of a whole department of highly relevant physical, physiological, and psychological facts. It would surely be nothing short of a miracle if it were theoretically adequate and if it were not positively misleading in some of its implications. If a synoptic view is to be taken, it must be taken by persons with philosophic interests and training, who are adequately informed of these and other relevant facts beside the beliefs and linguistic usages of common-sense.

    As my second example I will take what may roughly be called the 'mind-body' problem. Here the main facts to be viewed synoptically are the following.

    (1) Everyone knows that many of his sensations and feelings follow closely upon, and vary concomitantly with, certain events in his eyes, ears, skin, joints, etc. On the other hand, many experiences, e.g., those of day-dreaming, reasoning, deliberating, etc., do not seem prima facie to be co-variant with events in one's body. Then, again, everyone knows that certain of his overt bodily movements follow closely upon, e.g., his intentions to express certain thoughts or to make certain changes in his own or foreign bodies. There is no doubt at all that the common-sense notions of cause and effect, and the associated ideas of agent, instrument, and patient, and so on, are mainly derived from such facts as these.

    (2) The sciences of anatomy and physiology make it almost certain that the immediate bodily antecedents and correlates of our sensations are not events in our eyes, ears, skin, etc., but are slightly later chemical or electrical changes in our brainy These can be made perceptible to the senses only indirectly through elaborate technical devices, such as the [119] electro-encephalograph. These sciences also make it almost certain that the immediate bodily consequents and correlates of setting oneself to fulfil an intention are not the overt bodily movements which one is setting oneself to make, but are slightly earlier chemical or electrical changes in one's brain. Of these one is completely ignorant. It is further alleged, on the authority of these sciences, that there are immediate cerebral antecedents and correlates, of the same general nature, even for those mental processes, such as day-dreaming, reasoning, deliberating, etc., which do not seem prima facie to be co-variant with events in the body.

    (3) The physical sciences have developed a concept of causation in terms of regular sequence and concomitant variation. In this the notions of agent and patient, activity and passivity, etc., play little if any explicit part. So far as the notion of acting and being acted upon survives in physics it comes to this. The physicist thinks of one system as acting on another when energy is transferred from the former to the latter, or when the former, without doing work on the latter, modifies the direction of motion of some of its parts by fixed constraints. Now there is fairly good empirical evidence that a living organism never gains or loses energy except by transference from or to some other part of the material world. And it is difficult to imagine a volition exercising guidance without work on moving atoms or electrons in the brain, as a material constraint, such as the rod of a pendulum, does on a moving macroscopic body, such as a pendulum-bob.

    Now these various mutually relevant facts and concepts and principles are hardly ever viewed synoptically, except by philosophers. Common-sense is quite ignorant of many of them, and common language has grown up and crystallized before they were known or suspected. On the other hand, scientists, though familiar with all of them, tend to concentrate on one at a time and temporarily to ignore the rest. When they confine their attention to the chemical, physical, and physiological facts they are inclined to take the view that men are 'conscious automata', i.e., that all our mental states, induding processes of deliberating, imagining, reasoning, etc., are mere by-products of states of the brain, which are themselves completely determined by physical and physiological antecedents. But their daily lives and all their professional activities in designing, carrying out, and interpreting experiments presuppose a view which is shared by plain men and which seems prima facie to be incompatible with the 'conscious-automaton' theory.

    A scientist always assumes in practice that, when he designs an experiment and carries it out, he is initiating certain changes in the material 1 world which would not have taken place then and there unless he had thought them out beforehand, willed them, and deliberately prepared the conditions for them. He assumes that his assent to or dissent from the various alternative explanations which might be put on the results of an experiment is determined by processes of reasoning, demonstrative or probable, in which assent is given or witheld in accordance with evidence, which may be favourable or unfavourable, weak or strong or coercive. Now all this certainly involves concepts, and seems prima facie to involve modes of causation, completely different from those in terms of which the 'conscious-automaton' theory is formulated.

    To sum this up briefly. A scientist who investigates and theorizes about man and his powers and activities is himself a man exercising certain characteristically human powers and activities. But the account which he is led to give of man, when he treats him as an object of scientific investigation, seems difficult to reconcile with the occurrence and with the validity of his own most characteristic activities as investigator, experimenter, and reasoner. It seems no less difficult to reconcile with what non-scientists, and the scientist himself in his daily life, unhesitatingly take for granted about themselves and their fellow-men in ordinary social intercourse. There is obviously need for synopsis by someone who is aware of all the main facts and can hold them steadily together in one view. It is for that reason that there is a philosophical problem of body-and-mind, even if the solution of it should consist in showing that it has been wrongly stated.

    I think that these two examples should suffice to show what I understand by 'synopsis' and why I think that it is a mental operation which is peculiarly characteristic of philosophy. Synopsis is, however, no more than an essential first stage. From it philosophical thinking may develop in various ways. I will now try to describe and illustrate these.

    Suppose that the results of taking a synoptic view of a number of different mutually relevant departments of knowledge or belief were to show that they all obviously fit together without difficulty into a single coherent whole. Then there would be little or no occasion there for philosophy. But actually, as illustrated by my examples, this is often not the case. It often happens that each of the several mutually relevant regions, which we habitually contemplate and react to separately, gives rise to its own system of concepts and principles; that each such set seems fairly satisfactory and coherent in isolation; but that, when we contemplate the various departments together, we find that the several sets of concepts and principles seem prima facie to conflict with each [121] other. It is synopsis, revealing prima facie incoherence, which is the main motive to philosophical activity.

    When we are faced with apparent conflict between different sets of propositions which, taken separately, seem to be satisfactory and evidently true, the most obvious first step is to try to analyze the terms in them and to formulate the propositions themselves more carefully. Such a process is an indispensable step towards deciding whether the inconsistency is real or only apparent, and towards formulating it precisely if it is real. And this is a precondition of any reasonable attempt to deal with the difficulty.

    Now the general principles, in accordance with which people do in fact think and behave in their everyday life, in their professional activities as scientists, in their religious life, and so on, are seldom if ever explicitly before their minds. They have never explicitly formulated these presuppositions for themselves, and they might not recognize them if another person were to do so for them. It is an important part of philosophic activity to try to elicit and formulate the presuppositions of each of the various important departments of knowledge and belief which are to be brought into one synoptic view. This is obviously a difficult and delicate task. It is particularly so in connexion with the presuppositions of common-sense. For here the philosopher is trying to do for the plain man what the latter could not attempt to do for himself without ceasing, by definition, to be plain. It is evident that there is a serious risk of putting into the mouths of babes and of sucklings what could have occurred only to the wise and prudent. Nevertheless this danger must, I think, be faced if we are to get anywhere.

    I will now give some examples of attempts to elicit and formulate presuppositions. I have already outlined one instance in the account of the common-sense view of material things and our perception of them with which I introduced my first example of synopsis. Another admirable example is Sidgwick's attempt in his Methods of Ethics to formulate the main principles of the morality of common-sense. He tries there to elicit and to state clearly the presuppositions at the back of everyday moral judgments about truth-telling, promise-keeping, justice, etc., and their opposites Iying, promise-breaking, injustice, etc. This seems to me a typically philosophical undertaking, performed with exemplary acuteness, fairness, and thoroughness.

    It is often far from easy to discover what is presupposed by a certain system of beliefs or behaviour. What one tries to do may be described very roughly as follows. One tries to formulate a minimal set of fairly [122] general propositions which would make it logically necessary or logically reasonable for a person who accepted them, and who thought and acted consistently, to believe and to act in particular relevant situations in the ways in question. Here it is important to remember that we cannot safely assume that people do think and act consistently. We know that each of us is occasionally inconsistent in detail, and it would not be surprising if all of us were always more or less inconsistent in principle in important departments of our belief and behaviour. It may therefore be impossible, from the nature of the case, to formulate a mutually consistent set of presuppositions even for a particular department of human thought and action, e.g., for common-sense morality or for the common-sense beliefs about sense-perception and material things. Sidgwick, at any rate, found it impossible to do so for common-sense morality. If the presuppositions of any important department of belief or action seem to be mutually inconsistent, or if those of one department seem to be inconsistent with those of another which is obviously closely connected with the former, then the need for a careful analysis of the propositions concerned becomes pressing.

    We can illustrate this also from the problem of sense-perception. In order to deal with prima facie conflicts and difficulties here we need to analyze carefully such notions as the following: the notion of perceiving the same part of the same thing by two different senses, e.g. sight and touch; the notion of two persons simultaneously seeing the same part of the same thing; the notion of a thing remaining objectively unchanged during a period in which one and the same part of it presents varying visual appearances to an observer; the notion of a perceived thing having parts, e.g., a back and an inside, which are not at the moment being seen or felt; and the notion of a perceptible thing existing and having a history during periods when no one is perceiving it.

    Another good example is provided by the problem of free-will and determinism. The situation here may be described roughly as follows. (1) There is a sense of 'could' in which each of us in his daily life is quite convinced that on many occasions he and others could have acted otherwise than they in fact did. (2) Unless this conviction is significant and true, all judgments of the form: You ought to have done X (which you did not do.), or: You ought not to have done Y (which you did do.), are in principle false. Moreover, if that be so, such morally directed emotions as remorse, moral indignation, etc., are and have always been altogether without appropriate objects. (3) On the other hand, it seems evident to many people on reflexion that, given the circumstances in [123] which any event happened and the laws of nature, including those of psychology, it is inevitable that precisely such an event should have happened then and there. In science and a large part of our daily life we do seem to presuppose this.

    Now, in view of this prima facie conflict of presuppositions, the first business of the philosopher is one of analysis. He must try to analyze the sense of 'could' in which each of us is convinced that he and others could on many occasions have done otherwise than they did. He must try to analyze the sense of 'could' in which the conviction that persons could have acted otherwise than they did seems to be presupposed by a whole class of ordinary moral judgments and morally directed emotions. He must ask whether these are the same or different. Then he must try to analyze the sense of 'could' in which it seems evident to many that, given the circumstances in which an event happened and the laws of nature, nothing else could have happened. He may find that the senses are different, and that 'could' in the former sense or senses does not conflict with 'could not' in the latter. If so, the problem vanishes. If not, we have at least taken the essential first steps towards any possible solution of it.

    I have now explained and illustrated what I mean by 'synopsis', 'detection and formulation of presuppositions', and 'analysis of concepts and propositions'. We can next consider what I have called 'critical appraisal of presuppositions'.

    It is one problem, and often a very difficult one, to elicit and formulate the de facto presuppositions of a certain department of human belief or conduct. It is another problem to analyze these presuppositions and the notions involved in them, though these two processes generally go on together and interact with each other. But, when all this is done to the best of our ability, the question remains: Is there any good reason for us as critical philosophers to accept these analyzed and formulated presuppositions, even if as plain men or as scientists or as religious men we cannot help continuing to believe and to act in accordance with them? This is the question of critical appraisal.

    Suppose that the presuppositions of a certain department conflict with each other, as Sidgwick thought that those of common-sense morality do. Or suppose that those of one sphere of thought and action conflict with those of another intersecting sphere, as, e.g., it might well be held that the presuppositions of common-sense morality and everyday social intercourse do with those of biology, physiology, and experimental psychology. Then it is plain that a rational reflective person cannot continue in theory [124] to accept either of the conflicting propositions without reservation. But, even when there is no apparent conflict after careful analysis has been made, a situation may arise which I will illustrate by comparing and contrasting the cases of deductive and inductive logic.

    Consider the rules of the syllogism as formulated by Aristotle and his successors. There is a considerable class of deductive arguments which can without great violence be reduced to syllogistic form. Those of them which would universally be admitted to be cogent break none of the traditional rules of the syllogism, whilst those which would universally be admitted to be invalid break one or more of those rules. We may say therefore that Aristotle and his successors formulated the presuppositions of valid reasoning for an important class of deductive arguments.

    Now these rules do not appear to conflict with each other or with the presuppositions of any other department of human activity. So far, so good. But it must be admitted that few if any of them have any trace of self-evidence. Is it in the least obvious, e.g., that a syllogism with two negative premisses cannot be valid; or that, if the conclusion of a valid syllogism is negative, then one of the premisses must be negative? Surely not. But in this case we can get behind the traditional rules to more ultimate principles which entail them and which are self-evident. This can be done in various ways which we need not consider in detail. So here the situation may be regarded as satisfactory.

    Now contrast this with the case of inductive reasoning and the endeavours which have been made by Bacon, Mill, Venn, Jevons, Johnson, Keynes, and others to formulate and justify its principles. We need not concern ourselves here with the eliminative methods elaborated by Bacon, Mill, and Johnson. For, although these raise many interesting and subtle questions of detail, as Professor von Wright has well shown, they are essentially deductive arguments of a perfectly familiar kind. They state the conditions under which we can reject as false certain proposed generalizations. They give us no direct information as to the conditions under which we may accept a proposed generalization. Nor do they justify us even indirectly in accepting any generalization unless the following conditions are fulfilled. We must already know somehow that one or another of a certain set of alternative generalizations is true, and we must have empirical evidence which enables us to reject all but one of them in accordance with these principles of elimination.

    It is plain, then, that the essential problem is that of inductive generalization. In its simplest form the question may be stated as follows. Suppose that a number of instances of a class α, e.g., swans, which is in [125] principle unlimited in extent, have been observed, and that they have all been found to have a certain property P, e.g., whiteness. What justification is there for concluding, either with certainty or with high probability, that every past instance of α had P, that every present instance of it has P, and that every future instance of it will have P? We all assume that many such arguments are valid, and the whole of natural science is built on that assumption. What is at the back of this conviction?

    Now one thing is evident on a moment's reflexion. The subject of the instantial premiss is some, but not all, members of α, viz. those which have been observed up to date. But the subject of the inductive generalization is all members of α. The predicate is the same in both cases, viz., the property P which all the observed instances were found to have. It is therefore clear that any attempt to regard the argument as demonstrative and leading to a condusion which is certain must be mistaken. It does not matter what one supposes the suppressed premiss to be. No premiss, however strong, added to a particular premiss, will enable one validly to deduce a universal conclusion with the same subject and predicate. If such arguments can be defended at all, then, the line taken must be that the instantial premiss, either by itself or in conjunction with some other concealed premiss, makes the inductive condusion highly probable.

    It follows at once that, if inductive generalization can be justified at all, some at least of its presuppositions must be certain formal principles of the logic of probability. These may perhaps be regarded as self-evident, like the ultimate principles on which the traditional rules of the syllogism rest. We must remember, however, that the correct analysis of probability-propositions is a matter of acute controversy. It might well be held that the only senses of 'probability' in which these formal principles of probability are self-evident are quite different from the sense in which the conclusion of a well-established inductive generalization is highly probable.

    Let us, however, waive this difficulty. Even so, it is generally admitted by experts that the utmost that could be proved is the following. In accordance with the formal principles of probability all that favourable evidence can do is to multiply the initial probability of a generalization. If and only if this initial probability exceeds a pre-assigned fraction, which may itself be as small as you please, then an indefinitely long repetition of exclusively favourable instances will raise the final probability as near as you please to unity.

    Now, if this analysis be accepted, an essential presupposition of inductive [126] generalization must be a proposition to the following effect. For any conceivable inductive generalization there is some fraction ε, such that the probability of this generalization, prior to the observation of any relevant instances, favourable or unfavourable, is greater than ε. Now, if this is an intelligible statement, I do not think that it conflicts with anything that we positively know or with the presuppositions of any other department of thought or action. But one might well ask whether it is even intelligible. What exactly are we to understand, e.g., by the probability, prior to any observations on the colour of swans, of the generalization that all swans are white?

    But, even if the sentence be intelligible, is there the faintest reason to accept the proposition which it states? Obviously it has no trace of self-evidence, and equally obviously it would be circular to offer inductive grounds for it. In its lack of self-evidence it is indeed no worse off than some of the traditional rules of the syllogism. But in the present case no one has been able to suggest any proposition or set of propositions which fulfil the two essential conditions of being self-evident and entailing this proposition. All suggestions that I know of, e.g., the Uniformity of Nature of J. S. Mill or the Prineiple of Limited Variety of Lord Keynes, break down on both tests. They are not in the least self-evident, and they are far too abstract to entail anything so determinate as the proposition in question.

    The fate of this attempt to elicit, formulate, and critically appraise the presuppositions of a very important department of human thought is typical of what may happen. When one tries to appraise critically the presuppositions which one has elicited and formulated one may find that, although they do not conflict with each other or with those of any other department of thought or action, yet there appears to be no reason, direct or indirect, for accepting them. They are not self-evident, and one cannot discover any set of self-evident propositions from which they follow.

    The last topic which I shall treat is the activity which I have called 'synthesis' and which I said was specially characteristic of speculative philosophy.

    The purpose of synthesis is to supply a set of concepts and principles which shall cover satisfactorily all the various regions which are being viewed synoptically. The concepts and principles characteristic of each separate region, in so far as they are valid, must be shown to follow from, or at least to cohere closely with, this more general set under the special conditions and limitations peculiar to that region. The apparent [127] conflict between the concepts and principles characteristic of different but overlapping spheres of thought or action must be shown to arise from the valid application of these common concepts and principles in different contexts and under different special limitations. Even when there is no conflict to be solved it is likely that the synoptic contemplation of several regions, which are usually contemplated and reacted to separately, will reveal certain analogies between their contents or their structure and certain inter-relations between them as collective wholes.

    As an example of philosophic synthesis I will return to Sidgwick's Methods of Ethics. I have already given his account of the presuppositions of common-sense morality as a good example of the processes of eliciting, formulating, and critically appraising the presuppositions of an important department of human thought and action. I will now take his Utilitarianism as an example of an attempt at synthesis.

    It appeared to Sidgwick that the presuppositions of common-sense morality are neither severally self-evident nor collectively consistent. On the other hand the following propositions did seem self-evident to him.

  1. When several mutually exclusive alternative courses of action are open to a person in a given situation he will do rightly if and only if he enacts one of them which will have at least as good consequences as any of the others would have if he were to enact it.
  2. The only things that can be intrimirally good or bad are actual experiences, and the only feature of an experience which makes it intrinsically good or bad is its pleasantness or unpleasantness, respectively.
  3. Two alternative states of affairs, in which the same net balance of pleasant over unpleasant experience is distributed among wholly or partly different individuals or is differently distributed among the same individuals, have precisely the same intrinsic value.
These three principles may be called respectively the Utilitarian Principle, Ethical Hedonists, and the lndifference Principle.

    Now none of these principles seems in the least evident to most people at first sight. So Sidgwick tries to point out and remove a number of confusions and irrelevancies which tend to make people think that they are contemplating these propositions when they are in fact contemplating, and quite rightly doubting or rejecting, certain others which they confuse with these. His hope is that, when they really contemplate the propositions which he has in mind, they will find them as self-evident as he does. This is evidently the only possible method of procedure in such cases.

   Suppose now that we take into account the actual psychological and sociological facts about men living in communities. One important fact of this kind is the limitation of men's natural sympathies, of their knowledge of others, and of their power of affecting others for good or ill. Another such fact is the importance for human welfare of having general rules of conduct in certain frequently recurring situations, which men can be relied upon to follow, without pausing to reflect on the utilitarian reasons for doing so, even when following them goes much against the grain Sidgwick then tries to show in detail the following two things. He tries to show that, from his three self-evident principles and these psychological and sociological facts about human nature, the rules of common-sense morality follow as generally reliable recipes for maximizing happiness and minimizing unhappiness in a number of important types of situation in which men are frequently placed. He also tries to show that his three principles, together with these facts about human nature, provide a reasonable basis for deciding what a person ought to do in those marginal cases where the rules of common-sense morality conflict with each other. He thinks that common-sense is itself inclined on the whole to appeal to the utilitarian principle when faced with such situations. I am not called upon here to express any opinion on the success or failure of this attempt of Sidgwick's. What I do say is that it provides a most excellent example of an attempt at philosophic synthesis, based upon synopsis and subsequent formulation, analysis, and critical appraisal of presuppositions.

    This example relates, however, to synthesis within a single region, and not to synthesis of a number of overlapping and mutually relevant regions, such as I had in view when I gave as examples of synopsis the mind-body problem and the problem of freedom and determinism. It is attempts to make a synthesis which shall cover a number of such regions, or which shall even be all-embracing, that are characteristic of the great speculative philosophers, such as Plotinus, Leibniz, Spinoza, Hegel, Whitehead, McTaggart, and Alexander. I have left the mention of such forms of philosophizing to the extreme end of my lecture because I have recently said all that I have to say about them in the address on Some Methods of Speculative Philosophy which I referred to at the beginning.

    All that I will say here in condusion is this. To many people these are the most typical and the most exciting products of philosophical activity. But they are also the ones for which it is most difficult at the present day to put up a convincing defence, if they are to be regarded, as their authors would wish them to be, not as poetry or mythology addressed to our emotions, but as speculations about the nature of [129] things, to be accepted or rejected after critical examination by our intellects. For this reason I have preferred to dwell upon those philosophical activities which, as I have tried to show by examples, probably can be pursued with some hope of genuine, if neither continuous nor spectacular, progress. I suspect that synopsis, analysis, the formulation and appraisal of presuppositions, and limited attempts at synthesis, will suffice to keep philosophers innocently, and perhaps even usefully, occupied for as long as the social conditions are likely to last which make philosophizing economically and politically possible.