(I) Nature, Subdivisions, and Methods of Philosophy

     Under this heading come the whole of Professor Körner's essay and parts of the essays by Professors Nelson and Patterson.

(A) "CRITICAL" AND "SPECULATIVE PHILOSOPHY".

      I find nothing to dissent from and little to comment upon in Professor Nelson's remarks on my dealings with the branches of philosophy which in some of my later writings I distinguished under the names of "Analysis," "Synopsis," and "Synthesis." The history of English and American philosophy since 1923 has shown clearly that, when I wrote Scientific Thought, I greatly overestimated the certainty which could be hoped for in what I called "Critical Philosophy." It is no less true that I failed to notice the extent to which a philosopher's practice of Analysis is influenced (often unwittingly) by "metaphysical" presuppositions which, if made explicit, would fall within the province of Synopsis and of Synthesis.

      Since the days when I first used the expressions "critical" and "speculative" philosophy, and alleged that an essential part of the former is "analysis of concepts," much work has been done in analysing the concept of analysis, and distinctions have been drawn which I had not recognised Professor Körner may be said to be examining through a modern telescope a nebula which I had scanned with an old-fashioned operaglass.

      Professor Körner distinguishes two kinds of analysis, which he calls "exhibition" and "replacement." Both presuppose "rules which we or others have accepted for the use of signs as concepts." We are told that a sign is made "conceptual" by being used by a person in accordance with a certain rule which he accepts, and that a person may habitually conform to rules which he could not formulate. The business of exhibition-analysis is to elicit and formulate the de facto rules governing the use of certain words, phrases, etc., without attempting to criticise those rules, or, if they should be unsatisfactory, to substitute others for them.

      It seems to me that this does include an essential part of the business of critical philosophy, but that, unless certain restrictions are put on the terms "rule," "sign," "usage," etc., it covers much that would not commonly be included in philosophy. Surely e.g., as it stands it would include the work of a grammarian, in the strict sense of that term, and of a prosodist. What is a writer of a Greek grammar doing, except to formulate the rules which ancient Greeks unwittingly followed in using words, phrases, and sentences when speaking or writing? And what is a writer on Greek prosody doing, except to formulate the rules in accordance with which words are strung together in lines, and lines in verses, in various kinds of Greek poetry?

      Professor Körner says, quite rightly, that exhibition-analysis leads to empirical propositions about usage. In view of this it is important to note the following fact, and neither to underestimate nor to exaggerate its importance. With nearly all general names, e.g., "body," "animal," "person," etc., the situation is as follows. There are

  1. innumerable cases where hardly anyone familiar with the language would refuse to apply the name;
  2. innumerable cases where hardly any such person would consent to do so; and
  3. a great many marginal cases where such a person would hesitate whether to apply it or to withhold it.
In the group of marginal cases various possibilities exist. It may be that many such persons would unhesitatingly consent to apply the name, that many would unhesitatingly refuse to do so, and that many would hesitate. Again, it is often possible to present, not merely one, but several different series of marginal cases of the following kind. At one end of such a series most of such persons would feel little hesitation in consenting to apply the name; at the other end most of them would feel little hesitation in refusing to do so; and there would be a more or less continuous change in this respect as one presented intermediate instances in order.

      This is certainly an important fact, and neglect of it may lead to tiresome and futile controversy. But it does seem to me to have gone to the heads of some contemporary philosophers, and to have produced the impression that endless dithering about series of marginal cases is all that is required of them. What is needed is, not to stand moonstruck at a very simple fact which has been well recognised since the time of Locke (to go no further), but to proceed roughly as follows. In the first place, to compare and contrast the cases where people unhesitatingly consent to apply the name with those where they unhesitatingly refuse to do so, and to note the features common and peculiar to the first group. We thus get a set of rules for the application of the name "N" to what might be called "typical" or "indubitable" N's. Next, in the light of this, to look into the various series of marginal cases which diverge in different directions from the indubitable N's, and to note

  1. what features distinguish one such series from another, and
  2. the characteristic ways in which the distinguishing features of each such series vary as it diverges further and further from the indubitable N's.
Thus we may hope to end with a set of rules for the application of "N" to indubitable N's, qualified by a set of generalizations as to typical ways in which the applicability of "N" shades off in various directions from the indubitable N's to the indubitable non-N's.

      The above reflexions are not meant as a criticism on Professor Körner's remarks on exhibition-analysis. Passing now to what he calls "replacement-analysis," I find myself in general agreement with him. Certainly a person would have little motive for attempting to replace rules of usage, formulated as a result of exhibition-analysis, unless he held them to be defective in one way or another. That being granted, it is plainly desirable to make explicit the standards or requirements by which one is judging them, and the precise respects in which one thinks that they fall short. Lastly, it is important to make explicit what Professor Körner calls the "analysing relation," i.e., the kind of logical relation in which the replacing rules are supposed to stand to the rules which they are intended to replace. I have no doubt that I have often failed to fulfil these desiderata (if for no other reason, because they were not explicitly before my mind), and that the clear formulation of them by Professor Körner should help future analysts to do better in these respects.

      Professor Körner says that exhibition-analysis leads to contingent propositions, and replacement-analysis to necessary ones. What he has in mind is true; but it is important to understand precisely what that is, and not to confuse it with something else.

      It is true that any proposition to the effect that correct users of a language L have no hesitation in applying the name "N" when and only when the conditions c1c2 . . . . .cn are fulfilled, is contingent. And it is true that any proposition to the effect that conditions c'1c'2 . . . . . .c'm are a replacement of conditions c1c2 . . . . . .cn, given that the analysing relation is R, is necessarily true or necessarily false. But a proposition to the effect that the conditions c1c2 . . . . . cn are fulfilled in a given case may be either contingent or necessarily true or necessarily false. That will depend on the nature of the subject-matter. If, e.g., a particular animal fulfils (or fails to fulfil) the conditions under which an animal would unhesitatingly be called a "bird," that is a contingent fact about it. But, if the sum of a certain infinite series in pure mathematics fulfils (or fails to fulfil) the conditions under which a number would be unhesitatingly called "transcendental," that is a necessary fact about it. If the fulfilment (or the non-fulfilment), in a particular case, of the conditions formulated in an exhibition-analysis is contingent, then the fulfilment (or non-fulfilment) of the conditions substituted for them is also contingent. And the same would hold with "necessary" written for "contingent" in both places in the previous sentence.

      I pass now to Professor Körner's discussion about the nature of what I have called "speculative philosophy." In order to state clearly what I take to be the points at issue, I will begin with the notion of a sentence in the indicative mood. When a person utters or writes such a sentence he prima facie intends to assert or to deny or to offer for consideration or to put on record something which can significantly be said to be true or false. And when a person hears or reads such a sentence in a language which he understands, he expects prima facie to have presented for his consideration something which can significantly be said to be true or false. Every such sentence, then, serves prima facie to state, record, or convey factual information (correct or incorrect). We can sum this up by saying that every sentence in the indicative is "ostensibly informative."

      Now it is plausibly alleged that certain kinds of sentences in the indicative are in this respect misleading. Those who utter them are not in fact offering or recording information (correct or incorrect), though they may think that they are doing so. They are really only expressing an emotion, issuing a command, proferring advice, or so on. And those who hear or read such sentences understandingly are not thereby receiving any information (correct or incorrect), but are being emotionally stimulated, commanded, admonished, or so on. (This does not, of course, exclude the possibility that the hearer or reader may be led, either through explicit inference or through association, from hearing or reading such a sentence to forming a more or less confident opinion about the intentions, emotions, etc., of the speaker or writer.) I will describe such indicative sentences as "non-informative." Non-informative indicatives can then be classified in accordance with the positive functions which they perform, e.g., as evocative, admonitory, and so on.

      Now any treatise on speculative philosophy certainly consists of sentences in the indicative, and there is no doubt that most speculative philosophers have thought that most of the indicatives which they wrote or read in such treatises conveyed factual information. So one set of questions to be raised is this. Is this a complete mistake? Are all such indicatives really non-informative? If so, what functions do they perform

  1. for those who speak or write them, and
  2. for those who hear or read them understandingly?
Again, if so, why are those functions habitually performed by the inappropriate and misleading means of sentences in the indicative?

      Suppose that this extreme position were rejected. Suppose it were alleged that some at least of the indicatives in treatises on speculative philosophy really do state propositions (which the writers accept or reject or are uncertain about), and really do present those propositions to the consideration of those who read such treatises understandingly. Then two kinds of question could be raised, one logical or ontological, and the other epistemological.

      The logical or ontological question is this. Is every proposition, stated by those indicatives in treatises on speculative philosophy which are really informative, either necessarily true or necessarily false? Or are they all contingent? Or are some of one kind, and some of the other kind?

      The epistemological question is closely connected with this. But it is a different question, and it is important to distinguish the two. We may introduce it as follows. If a proposition is contingent, the only legitimate ground on which it can be accepted or rejected is empirical. If, on the other hand, a proposition is necessarily true, it is theoretically possible to have purely a priori grounds for accepting it; and, if it is necessarily false, it is theoretically possible to have purely a priori grounds for rejecting it. But a proposition might in fact be necessarily true (if true) and necessarily false (if false), and a person might even know that this is so and yet he might have nothing better than empirical grounds for accepting or rejecting it. (That is the case, e.g., in regard to some propositions in the Theory of Numbers, such as Fermat's "last theorem.") So the epistemological question is this. Supposing

  1. that some at least of the indicatives of speculative philosophy are informative, and
  2. that some at least of them state propositions which are necessarily true or necessarily false, has any speculative philosopher produced cogent a priori reasons for accepting any of the former or for rejecting any of the latter?

      The above questions were certainly not all of them distinctly before my mind when I wrote about the nature of speculative philosophy and its relations to critical philosophy. Therefore no unambiguous answers to them will be found in my writings or can be elicited from them. What little I can now say on this topic is perhaps best discussed in connexion with one section of Professor Patterson's essay. For, in the summary at the end of it, he raises the general objection that I have failed to make a satisfactory case against the a priori deductive method in metaphysics.

(B) NATURE AND METHODS OF SPECULATIVE PHILOSOPHY.

      I should evidently be in a much stronger position, if I held that metaphysical sentences in the indicative must from the nature of the case be non-informative. But I have never been in the least impressed by the general argument, from the alleged nature of significant assertion, which has been put forward to prove this. It seems to me to depend on taking a very narrow and highly arbitrary definition of "significant assertion," and then ruling out by definition those indicatives as non-informative. I suppose that one may fairly say by this time that this kind of argument is characteristic of a philosophical school which (in the words of Oscar Wilde) "has a great future behind it." It seems to me plain that those who use this argument have at the back of their minds a number of ontological and epistemological premisses which constitute a part of an unformulated system of speculative philosophy. It seems to presuppose inter alia that there can be no a priori concepts, that all necessarily true propositions are analytic and all necessarily false ones are self-contradictory, and that all significant synthetic propositions are such that they can be validated or invalidated by reference to sense-perception or introspection. I am not convinced of any of these presuppositions.

     That would of course leave it open to me to hold that some (or even all) of the indicatives which occur in works on speculative philosophy are in fact non-informative. It may well be that some of them are, but it seems to me that many of them convey to the reader propositions for consideration, for reasoned acceptance or for reasoned rejection or for suspension of judgment. Take, e.g., Leibniz's doctrine that what appears to us as an inorganic material thing, e.g., a stone, is in fact a collection of a vast number of animated organisms of a very low order; that what appears to us as an animated body is in fact a collection of minds of a lower order of intelligence related in a certain specific way to a single mind of a higher order; and that what we take to be the laws of inorganic matter are statistical regularities concerning such groups of very numerous minds of very low intelligence. I can understand these statements in outline by analogy with what I know, e.g., of a swarm of gnats appearing as a cloud, of habitual and instinctive action in men and animals, of crowd-psychology, and so on. They do not seem to me to be radically different in nature from the extremely difficult statements which theoretical physicists make about the ultra-microscopic constituents of macroscopic phenomena. They may happen to uplift, depress, or admonish the reader; but, if they should do so, that seems to me to be incidental, as contrasted with the informative function which they perform.

      Certainly they cannot be tested, as scientific theories can, by deducing from them consequences as to what should be perceptible by the senses under assigned experimentally producible conditions. A metaphysical theory has to be appraised by reference to such criteria as

  1. its internal consistency or inconsistency,
  2. its coherence or incoherence with certain very general principles (positive or negative) which seems self-evident to the reader, and
  3. its ability (given that it fulfils the first two conditions to the reader's satisfaction) to unify in an illuminating way a number of very general and pervasive features of the inorganic, the organic, and the psychological aspects of the world.

      It is evident that universal or even very general agreement can hardly be expected, in view of the fact that general principles which seem self-evident to some persons will not seem so to others even of the same period and culture, and that principles which seem self-evident to most persons of a given culture at a given period may not seem so to those of other cultures or at other periods.

      Suppose we take what Professor Patterson calls the "a priori deductive method in philosophy" to be the attempt to infer a set of far-reaching and surprising speculative conclusions from a comparatively few premisses, each of which is either found to be self-evident on reflexion or states a very general and obvious empirical fact which no-one would be likely to question. Then I am certainly not in a position to assert a priori that "the a priori method in philosophy" must be futile.

      For here too I am in certain respects much less fortunately situated than many contemporary "anti-metaphysicians." I am not convinced that every proposition which is necessarily true must be analytic. And, if there be propositions which are synthetic and necessary, I see no reason why some of them should not be self-evident on careful inspection to most intelligent persons of appropriate training and interests. Again, it is obvious that there are very general empirical facts, e.g., that there appears to be change in general and motion in particular, which no sane person is likely to question. I am therefore not prepared to deny in principle that there might be premisses available for a satisfactory system of deductive metaphysics.

      Granted this, it would be idle to make the general objection that no important and surprising conclusions are likely to be deducible from a few very abstract premisses. For in geometry the most beautiful and surprising consequences have been deduced from such premisses. And in theoretical physics such extremely abstract and negative principles as the Entropy Principle, the Relativity Principle, and the Uncertainty Principle have led to highly interesting and unexpected results.

      I am therefore reduced, as Professor Patterson says, to appealing to the alleged lack of success of deductive metaphysicians in the past. That is certainly not a very strong argument, and Professor Patterson questions even the empirical basis of it. Have Descartes, Spinoza, Leibniz, and Hegel, e.g., he asks, accomplished nothing?

      To this I would answer as follows. They have done much to illuminate the problems which they have discussed, but I do not know of anything of importance which they have established deductively. Spinoza is in fact the only one of them who even claimed to do this, and I think that most of his readers would feel that what is valuable in his work is independent of, and tends to be obscured and distorted by, his deductive method of expounding it.

      It does seem to me that nothing comparable to the results obtained in geometry and theoretical physics by the deductive method has in fact been achieved by that method in speculative philosophy. There seems to me to be good empirical ground for thinking it very unlikely that others will succeed where so many men of such outstanding ability through so many centuries have failed. And it may well be that a careful study of the peculiarities of the subject-matter of geometry and of theoretical physics would provide a more positive and detailed ground for scepticism as to the possibility of a system of mainly deductive speculative philosophy.