Philosophical Papers of Georg Henrik von Wright I Practical Reason Il Philosophical Logic Ill Truth, Knowledge, and ModalityTruth, Knowledge, and Modality GEORG HENRIK VON WRIGHT Philosophical Papers Volume IIT BASIL BLACKWELL© GH. von Wright 1984 First published 1984 Basil Blackwell Publisher Limited 108 Cowley Road, Oxford OX4 1JF, England Basil Blackwell Inc. 432 Park Avenue South, Suite 1505, New York, NY 10016, USA All rights reserved. Except for the quotation of short passages for the purposes of criticism and review, no part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the publisher. Except in the United States of America, this book is sold subject to the condition that it shall not, by way of trade or otherwise, be lent, re-sold, hired out, or otherwise circulated without the publisher’s prior consent in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed on the subsequent purchaser. British Library Cataloguing in Publication Data Wright, G.H. von Philosophical papers. Vol.3: Truth, knowledge and modality 1. Philosophy I. Title 192 B21 ISBN 0-631-13367-4 Typeset by MHL Typesetting Limited, Coventry Printed in Great Britain by The Pitman Press Ltd., BathContents Introduction and Acknowledgements Determinism and Future Truth Demystifying Propositions Truth and Logic Truth-Logic and ‘‘Dialectical Synthesis’’ Truth-Logic and Antinomies The Logic of Predication Determinism and Knowledge of the Future Knowledge and Necessity “‘Omne quod est quando est necesse est esse” Note on Preventing and Producing On Causal Knowledge Diachronic and Synchronic Modality Logical Modality Natural Modality Laws of Nature Subject Index Index of Persons Vii 14 26 36 39 42 52 68 72 82 86 104 117 134 150 154Introduction and Acknowledgements After the publication of Explanation and Understanding in 1971 my main concern in philosophy came to centre round the idea of ‘‘determinism’’. It had been a tenet of that work that determinism in the sphere of natural events and the sphere of human actions differ radically from one another. In two subsequent books, Causality and Determinism (1974) and Freedom and Determination (1980), | tried to clarify my thoughts on this difference. In the course of these efforts new questions constantly cropped up and many of them were oriented not so much towards the distinction between events and actions as towards the ideas of truth and knowledge and the modal notions of necessity and possibility. Some of the problems, moreover, had notable traditions in Ancient and Medieval philosophy which had been revived in recent thought, particularly by people working in the logico-analytical mainstream of modern philosophy. Examples are the Aristotelian problem of the ‘‘Sea Battle’, the Diodorean xuptevov, and the Schoolmen’s difficulties in reconciling God’s omniscience with human freedom and responsibility. The work of distinguished contemporaries—Anscombe, Hintikka, Prior, and Jules Vuillemin to mention a few—stirred me to efforts of my own in a similar direction. They were given tentative expression in my Nellie Wallace Lec- tures at Oxford in 1978 on ‘‘Some Ancient Problems of Time, Truth, and Necessity’ and in a lecture course at the Collége de France called ‘‘Truth, Knowledge, and Certainty”’ two years later. The studies here published under the titles ‘‘Determinism and Future Truth’’, ‘‘Determinism and Knowledge of the Future’, and ‘Knowledge and Necessity’ got their final form in the Belgum Memorial Lectures which I gave at Saint Olaf College, Northfield, Minnesota in the Autumn 1983. I am grateful for the many opportunities I have had to present my ideas to responsive audiences representing different backgrounds and traditions. For some time I planned to fuse the work I was doing on truth, knowledge, and modality into the unity of a book. The heterogeneity of the material forced me in the end to abandon the idea. What is presented here is essentially a collection of separate essays. But they have a greater thematic unity than the papers collected in the volumes Practical Reason and Philosophical Logic and there are frequent cross-references between the individual studies. In the late 1950s I published a paper on the concept of negation which came to have a seminal influence on my subsequent work in philosophicalIntroduction logic.' Its basic idea was a distinction between two forms of negation, a weak and a strong form. ‘‘Classical’’ negation is weak—it signifies, roughly speaking, ‘‘nothingness’’, mere absence of something. Strong negation is related to the notion of a contrary rather than to that of a con- tradictory. It stands for the opposite to something ‘‘positive’’—like priva- tion in relation to possession of a property or forbearance in relation to performance of an action. A similar distinction can be made between “not true’ and ‘‘false’’. Falsehood, however, may be defined as truth of the contradictory of a proposition. This opens a possibility of defining strong negation in the more ‘‘classical’’ terms of (weak) negation and truth. The ‘‘non-classical’’ feature of the logic which can then be built is, not that it employs an unorthodox kind of negation, but that it incor- porates the concept of truth in its object-language as an “‘operator’’ on Propositions, analogous to the notions of possibility and necessity in modal logic. How this is done I tried to show, for the logic of proposi- tions, in a paper published in a Festschrift for Professor Séren Halldén? and, for predicate logic, in a Festschrift for Professor Oiva Ketonen.* Both papers are from the year 1973. They may be regarded as early forms of the essays ‘‘Truth and Logic’? and ‘‘The Logic of Predication’’ published in this collection. I believe that truth-logic, or alethic logic as it may also be called, can be used for shedding light on the relation between what may be called logic in the “traditional” sense and another type of logic which, though it too has early roots in the history of thought, is chiefly associated with the name of Hegel and with so-called Dialectical Logic. Efforts are today made in dif- ferent quarters at a rapprochement between these two traditions. It is premature to evaluate their success—and it seems to me best to let the efforts be pursued independently of one another, at least for the time be- ing. This is why, in the two appendices to the paper “Truth and Logic” which deal with questions bordering on dialectical logic, I have not made any reference to related ventures in recent literature. For stimulating discussions on the topics of contradiction and “dialectical synthesis” [am indebted to two young scholars, Antti Hautamaki in Helsinki and Hristo Smolenov in Sofia. ' “On the Logic of Negation”, Commentationes physico-mathematicae Societatis Scien- tiarum Fennicae, XXII (4), 1959; available in reprint by University Microfilms Interna- tional. See also the Introduction to Philosophical Logic (Philosophical Papers II), Basil Blackwell, Oxford, 1983. 2 «Truth as Modality”. In Modality, Morality and Other Problems of Sense and Nonsense. Essays dedicated to Séren Halldén, Gleerups, Lund, 1973. 3 “Remarks on the Logic of Predication”, Ajatus, 35, 1973.Introduction ix The essay “Determinism and Future Truth” had two published predecessors which were partly also historical studies on Aristotle. One was called “Determinismus, Wahrheit und Zeitlichkeit, ein Beitrag zum Problem der zukiinftigen kontingenten Wahrheiten”;* the other, which appeared in a Festschrift in honour of Elizabeth Anscombe, had the title “Time, Truth, and Necessity” .* A preliminary version of “Determinism and Knowledge of the Future” with a postscript on “Knowledge and Necessity” was published in the series of publications of the Finnish Society for Future Research.® The essay “On Causal Knowledge” again first appeared in a Festschrift for Norman Malcolm and is here republished with the kind permission of the copyright holder.” The comments on the Aristotelian dictum “Omne quod est quando est necesse est esse” was in origin a paper for a symposium of the Nordic Plato Society in Copenhagen in 1979. A first version of the paper on diachronic and synchronic modalities was published that same year in the Spanish periodical Teorema and a second version some years later in Acta Philosophica Fennica.* The research embodied in these papers was in origin allied to my efforts to reconstruct the much-debated Master Argu- ment of Diodorus Cronus published in an essay, not included here, in a Festschrift for Jaakko Hintikka.” The three concluding studies, on logical and natural modality and on laws of nature, have no predecessors in my earlier literary output. Georg Henrik von Wright * Studia Leibnitiana, 6, 1974. $ in Intention and Intentionality. Essays in Honour of G.E.M. Anscombe, ed. by Cora Diamond and Jenny Teichman. The Harvester Press, Brighton, 1979. ° Tulevaisuuden Tutkimuksen Seuran Julkaisu A4, Turku, 1982. 7 Knowledge and Mind, ed. by Carl Ginet and Sydney Shoemaker. Copyright © 1983 by Oxford University Press, Inc. * Intensional Logic: Theory and Applications, edited by I. Niiniluoto and E. Saarinen. Acta Philosophica Fennica, vol. 35, 1982. ° “The ‘Master Argument’ of Diodorus”. In Essays in Honour of Jaakko Hintikka, ed. by E, Saarinen, R. Hilpinen, 1. Niiniluoto, and M.B. Provence Hintikka. D. Reidel Publishing Co., Dordrecht, Holland, 1979.Determinism and Future Truth 1 The ninth chapter of De Interpretatione has long puzzled philosophers and logicians. Did Aristotle, or did he not, hold that contingent proposi- tions about the future are (now) either true or false? He would evidently have agreed that it is (now) certain that there will be a sea battle or will not be a sea battle tomorrow, and he realized that from accepting this it does not follow that it is either certain (now) that there will be a sea battle tomorrow or certain (now) that there will not be a sea battle tomorrow. But would he have agreed that one of the two propositions, viz. the proposition that there will be a sea battle tomorrow and the proposition that there will not be a sea battle tomorrow, is (now) true? On this ques- tion interpreters of Aristotle disagree. More interesting than the question of what was Aristotle’s way out of the difficulties which vexed him, I find the question of how the problems themselves which we encounter here have to be solved. I have dealt with this in earlier papers.' It has seemed to me that my “way out” would also have had Aristotle’s approval. But on this last point I have not been able to rid myself entirely of doubts. In this new attack on the problem I shall altogether disregard the possible opinions of the Stagirite. Il Consider the following statement: (1) Itis true that there will be a sea battle tomorrow or there will not be a sea battle tomorrow. Is this not an indisputably true statement? But, if so, is it then not also indisputably true that (2) Itis true that there will be a sea battle tomorrow or it is true that there will not be a sea battle tomorrow? If we admit this distribution of (1) into (2) have we then not thereby implicitly admitted that now the truth of one of the two, mutually ex- " See above, Introduction, p. ix.2 Determinism and Future Truth clusive and jointly exhaustive, alternatives is settled for tomorrow—even though we need not know which one of the two will turn out to be true? It may thus look as if from accepting (1) we are led, by logical argu- ment, to accept that the truth of everything which will be is predeter- mined, already settled before it is—in fact at any time beforehand. Because the substance of the above argument would not be changed if for “tomorrow” we substitute a reference to any time, however remote, in the future. Accepting this consequence of (1) and (2) is tantamount to accepting (a form of) determinism. Il Let us next take a closer look at the statements (1) and (2). Statement (1) I am going to call an application of the Law of Excluded Middle to the proposition that there will be a sea battle tomorrow. By the Law of Excluded Middle I then understand the principle of logic which says that the disjunction of any given proposition and its negation is logically, and therewith also necessarily, true. Statement (2) is a disjunction. The second disjunct says that it is true that there will not be a sea battle tomorrow. But this is tantamount to say- ing that it is false that there will be a sea battle tomorrow. Generally speaking: I shall regard it as a definition of the predicate “false” that falsehood is the truth of the negation (contradictory) of a proposition. This is not as trivial as it may sound. We must distinguish between the truth of the negation (= falsehood) and the negation of truth, between the phrases “true that not” and “not true that”. The importance of this distinction will be discussed in a later chapter.” Accepting the equivalence of meaning between “true that not” and “false that” allows us to transform (2) into (3) Itis true that there will be a sea battle tomorrow or it is false that there will be a sea battle tomorrow. Allowing this move, statement (2) thus says, implicitly, that the propo- sition that there will be a sea battle tomorrow is either true or false. The general principle which says that any given proposition is either true or false, i.e. has one of the two truth-values “true” and “false”, is known as the Principle or Law of Bivalence. Our statement (2) is then an applica- tion of the Law of Bivalence to the particular proposition that there will be a sea battle tomorrow. The proposition that there will be a sea battle tomorrow exemplifies a ? See below, especially p. 27f.Determinism and Future Truth 3 future contingency. And similarly its negation, that there will not be a sea battle then. The impression that the application of the Laws of Excluded Middle and of Bivalence to contingent propositions about the future entails a commitment to determinism I shall label an “illusion”. The prob- lem with which I am here concerned is how to dispel the illusion. I shall not maintain that what has been said above gives an accurate and exhaustive characterization of the problem which worried Aristotle in the famous ninth chapter. But I think it is difficult to read this chapter without being worried oneself by the problem as here described.* IV Before presenting my own solution to the puzzle, I shall briefly mention two other proposed ways out of the difficulties. I think neither of them is acceptable; but both contain ideas which deserve to be further discussed. The first denies the legitimacy of the distributive step from (1) to (2) (or (3)). This solution was suggested by Lukasiewicz to whom, I think, belongs the chief credit for having revived the debate about future con- tingencies in contemporary philosophical logic.‘ Lukasiewicz stressed the importance of keeping distinct the two prin- ciples which I here, following him, call the Laws of Excluded Middle and of Bivalence respectively. In order to dispel our “deterministic illusion” it is not necessary to reject or restrict the validity of the former, Lukasiewicz notes. But the latter, according to him, is not unrestrictedly valid. It does not, for example, hold good for contingent propositions about the future. Such propositions lack truth-value, are neither true nor false. Lukasiewicz also makes the following interesting observation:*® The two disjuncts in the disjunction after the phrase “it is true that” in (1) stand in 3 Opinions differ on the question of what Aristotle’s main concern in De Int. 9 was. See the essay by Jaakko Hintikka: “The once and future sea fight: Aristotle’s discussion of future contingents in De Interpretatione 9” , in his collection of essays Time and Necessity, Studies in Aristotle’s Theory of Modality, The Clarendon Press, Oxford, 1973, especially pp. 147-53. * Jan Lukasiewicz, “O Determinizmie”, published posthumously in the collection of papers by Lukasiewicz, Z zagadnieh logiki i filosofii, ed. by J. Stupecki, Warszawa, 1961. The paper was in origin an address which Lukasiewicz delivered as Rector (Vice-Chancellor) of the University of Warsaw at the inauguration of the academic year 1922-3. There is an English translation by Z. Jordan in Polish Logic 1920-1939, ed. by Storrs McCall, Oxford, 1967 and another by Rose Rand in The Polish Review, 13, 1968. A German translation, by Gunther Patzig, appeared in Studia Leibnitiana, 5, 1973. * Op. cit., Section 11. © Op. cit., Section 9.4 Determinism and Future Truth what he calls a contradictory relationship to one another. This is why one may, with Aristotle, assume that (1) is necessarily true. But the two dis- juncts which make up (2) do not stand in a contradictory relationship; neither of the two is the negation of the other. Therefore (2) cannot necessarily be true. It may, in fact, be false. Lukasiewicz’s rejection of the Principle of Bivalence for future con- tingencies was a starting point for his grand conception of a many-valued logic, related to “classical” two-valued logic in a way which is analogous to the relation between non-Euclidean and Euclidean geometry. Lukasiewicz was not the first to entertain the idea of a polyvalent logic. But he gave a decisive impetus to its modern development. Vv The second suggested solution to our problem which I shall reject as un- satisfactory has to do with the semantic status of the phrase “it is true that”. This phrase, when prefixed to a sentence which expresses a propo- sition, is often said to be semantically otiose or redundant or vacuous. The idea is sometimes given the form of an identity: “it is true that p” = “p’ where the letter might represent, for example, the sentence “there will be a sea battle tomorrow”. It is not clear by itself how we shall understand the identity sign here—whether it designates identity of meaning of the two sentences or identity of truth-value of the propositions which they express. But whether we understand it in one way or the other, the idea seems to imply that the two members of the equation are intersubstitutable sa/va veritate at least in some contexts. Let us assume that our statements (1) and (2) qualify as such contexts. Then (1) reduces to the statement “py” (4) There will be a sea battle tomorrow or there will not be a sea battle tomorrow. But (2), after the substitution, also reduces to this. Thus (1) and (2) both say exactly the same, at least from the point of view of truth, viz. that there will be a sea battle tomorrow or there will not be a sea battle tomor- row. And if this is what (1) and (2) both say, the impression that the passage from (1) to (2) involves a commitment to determinism must be an illusion. Accepting the idea about the vacuous character of the phrase “it is true that”, one could also say that any attempt to drive a wedge between theDeterminism and Future Truth 5 Law of Excluded Middle and the Law of Bivalence is a vain manoeuvre. With this insight the “detérministic illusion” is dispelled. This “solution”, however, is too cheap to be good. Because, as we shall see later (below, Section XID), it hinges on the assumption that contingent propositions about the future have a truth-value. And this is precisely what some people, for example Lukasiewicz, have thought necessary to deny in order not to have to accept the consequence that the truth-value of propositions about the future is predetermined. VI Let tomorrow be day f. Assume that there is a sea battle on that day. Then it is true that there is a sea battle at ¢ (on that day). When is this true? Tomorrow? Or “already” today? Assume that somebody had said yesterday that there will be a sea battle the day after tomorrow. Or, since tomorrow (when writing this) is 14 January 1981, assume that somebody had said on 14 January 1881 that there will be a sea battle somewhere exactly 100 years later—or that somebody in ancient Greece had said that after so and so many sunrises in the place where he was living there would somewhere be a sea battle before sunset in his place. Jf tomorrow there was a sea battle these people were surely right. They spoke the truth. What they said was true. What did they say? They said that there will be a sea battle on 14 January 1981. So was it not true when they made the statement? But “it” is the proposi- tion that there will be a sea battle tomorrow, i.e. on 14 January 1981. We are thus led to the following idea: If there is a sea battle on 14 January 1981 the proposition that there is a sea battle then is true at any time before, on, or after the date in question. Or, putting “p” for “there is a sea battle” and “at ¢” for “on 14 January 1981”: If p at ¢, then the proposition that p at ¢ is always or sempiternally true. But rather than calling its truth “sempiternal”, I think we should call it “atemporal” and say that “true at ¢” here simply means “true”. I shall refer to this atemporal notion of truth as plain truth. When plain truth is in question such locutions as “true today” or “already true” mean nothing over and above “true”, and “not yet true” means the same as “not true”. VIL When plain truth is concerned, “is” in “it is true that” is not the tensed “is” which occurs in the combination “is now” or “is already”. It is a6 Determinism and Future Truth tenseless “is” like in “twice two is four”. The tenseless “is” has no past form “was” or future form “will be”. Nor can it be temporally specified as in the phrase “is at f”. In spite of this, however, tensed uses of the verb “is” often occur in phrases attributing plain truth to a proposition. I doubt whether we could label such uses “ungrammatical”. But they can be replaced by locutions using the tenseless “is” in “it is true that”. Some examples will be given to illustrate this: One might say “it is now true that there is a sea battle” meaning that it is true that there is a sea battle (going on) now or, simply, that a sea battle is now being fought. One might also say “it was true in 1750 that France was a monarchy then”. Or, “that France is a monarchy was true in 1750”. And this again can be restated in the form “It is true that France was a monarchy in 1750”. The future form “it will be true” can also be used in attributions of plain truth, as when we say “it will be true tomorrow that the sky is clearing up” meaning that it is true that the sky will clear up tomorrow. Generally speaking, when plain truth is in question the schematic phrase “it is (was, will be) true at ¢ that ...” can be transformed into the phrase “it is true that at ¢ ...”, where the eliminated tensed form of the verb “is” is reflected in the tense of the sentence expressing what is said to be true. VIL It follows from what has been said, that plain truth is something “unchangeable” or “imperishable”. It does not come into being, nor does it cease to be. One could say of it that “once true, always true”—but then one must distinguish the temporal “once” and “always” from an atem- poral meaning of these words in that combination. But do we not speak of things becoming true and ceasing to be true and of things sometimes being true and sometimes not true? Surely we do this—also when the truth under consideration is what I have here called the atemporal notion of plain truth. So how then shall we understand these locutions? The answer is obtained from a closer inspection of the “things” to which truth is attributed. These things I have already called propositions. Some logicians and philosophers would prefer to call them sentences. This may be done—but in order to explain what it means to attribute truth to a sentence we must, I think, refer to the proposition which the sentence in question expresses, i.e. to what the sentence says. Truth, therefore, isDeterminism and Future Truth 7 primarily an attribute of propositions and not of sentences. (On the no- tion of “proposition” see the next essay.) With regard to propositions we have to notice here the following distinction. That it is raining is a proposition. But this proposition is not “by itself” true or false. In order to be either it must become associated with a time and a place. That it was raining in Paris on 1 January 1900 is true or false, as the case might have been. The association with time and space is often implicit in the linguistic context. If I use the sentence “it is raining” for making a statement, I nor- mally mean that it is raining “here and now”, i.e. in the place and at the time of my making the statement. I have elsewhere’ called the two types of proposition generic and individual, Generic propositions are a species of what is also called propo- sitional functions. Their truth-value is a function of a spatio-temporal individuation. We can here ignore the spatial component. I shall say that in the schematic phrase “p at ¢” the letter “p” stands for a sentence ex- pressing a generic proposition such that “p at ¢” expresses an individual proposition. For a given “p” the schema “p at f” may turn out to be a true proposi- tion for some value of “r” and a false proposition for some other value. One can then say that the generic proposition that p, for example that it is raining in Paris, is sometimes true and sometimes not. One can also speak of its coming true (at a certain time) and its ceasing to be true. If the schema turns out to be a true proposition for all values of “1” we say that the generic proposition that p is sempiternally true. For example, the propositions that the earth is shifting its position relative to the sun and that it is rotating round its axis may be thought of as (“practically”) sem- piternal truths. The sempiternal truth of the generic proposition that the earth is moving must not be confused with the atemporal (and in this sense “eternal” or “sempiternal”) truth of the individual proposition that the earth is moving at time ¢. Truth, when attributed to generic propositions, is temporal, or better: temporalized; (plain) truth attributed to individual propositions is atem- poral. But the meaning of an attribution of (temporalized) truth to a generic proposition is, as we have seen, explicated in terms of an attribu- tion of (atemporal) truth to some individual proposition(s). The former type of attribution is thus secondary to the latter—and the concept of 7 In Norm and Action, A Logical Inquiry, Routledge & Kegan Paul, London, 1963, Chapter I], Section 4.8 Determinism and Future Truth truth involved in both types of attribution is what I have here called “plain truth”. I have dwelt on these distinctions at some length because I think they are important when discussing the idea of truth. A failure to make them clear may have also contributed to confusions in the debates concerning future contingencies. Ix So far I have been talking about “plain” truth. But not all truth is “plain”. The phrase “it is true” and its tensed variations such as “it was true”, “will be true”, “is now true”, “today true”, “already true”, “not yet true”, when applied to individual propositions, a/so has a use which is genuinely temporal and cannot be reduced to uses of the atemporal notion of plain truth. For example: “That there will be a sea battle tomorrow is now true; the admirals have just decided that the fleet will fight.” Or: “It is already true that there will be an eclipse of the moon on such and such a night in the year 2000; this may be calculated from the laws of planetary motion. That this be true was fixed from the dawn of creation.” It cannot be disputed that these are genuinely temporal uses of the phrase “it is true” and its tensed variations. But one might dispute that they are genuine uses of the word “true”. What I mean is the following: In the examples mentioned one can for the word “true” substitute some other words which express the intended meaning of the sentences at least equally well or maybe with greater clarity. One such word, perhaps the most telling one, is “certain”. “It is now certain that there will be a sea battle tomorrow. The admirals have just decided that the fleet is going to fight. But one hour ago this was not yet certain.” Other words of a very colloquial character which will also do here are “fixed” or “settled”. A more “philosophical” one is the word “necessary”. It would suit the second example we gave particularly well: “It is already necessary, and was so from the dawn of creation, that there will be an eclipse of the moon on such and such a night in the year 2000.” It should be observed that the notion of certainty which we encounter here is nor the epistemic (“subjective”) idea of someone being certain of something. It is an ontic or objective certainty which is there indepen- dently of whether anybody knows about it.* Even if man had never been ® 1 must therefore disagree with Moore, when he writes in his paper “Certainty” (Philosophical Papers, Allen & Unwin, London, 1959, p. 240) as follows: “It is, indeed, ob- vious, I think, that a thing can’t be certain, unless it is known: this is one obvious point thatDeterminism and Future Truth 9 able to figure out the regularities of planetary motion, it would be certain now that there will be an eclipse on such and such a future day. But this certainty would, of course, not be there if the regularities concerned did not, as a matter of fact, hold good. When the phrase “it is true that” is used in a genuinely temporal sense, one cannot only substitute for “true” the words “certain” or “necessary”. One can also substitute for it a compound “certainly true” or “necessarily true”. In the compound we distinguish two components, viz. truth and certainty or necessity. The first is “plain” truth and, as such, atemporal. It is the second component which is accountable for the temporal character of the phrase. Certainty and necessity function here as what I propose to call diachronic modalities.’ An individual proposition which is certain now need not have been this always. The proposition was always true in that spurious sense of “always” which really designates atemporality. But it became certain—and therefore also true in the compound sense of cer- tainly true. Similar observations apply to the idea of necessary truth. A man falls from the top of the Eiffel tower and is killed. That his death occurs at time tis an atemporal, “eternal” truth. That he should die precisely then was necessary from the moment he fell down. After having fallen over the bar- rier he was “doomed”. But that he fell was, let us assume, accidental. Before he fell it was not necessary that his death was going to occur at time ¢. When the phrase “it is true that” is used temporally, I say that the use of “true” is not “plain” but spurious." But this spurious use, as we have distinguishes the use of the word ‘certain’ from that of the word ‘true’; a thing that nobody knows may quite well be true, but cannot possibly be certain. We can, then, say that it is a necessary condition for the truth of ‘It is certain that p’ that somebody should know that pis true.” —But this cannot be so. Assume that there was an earthquake in a district and that nobody had anticipated it. Afterwards geologists find that, in view of some hitherto unknown facts about the geology of the region, the earthquake was in fact, as the experts then would say, certain to happen. Must this not be considered an entirely normal and quite common use of “certain”? To suggest that one should say that the earthquake was, in view of those facts, (causally) necessary, but not say that it was certain, would seem to me to be misdirected pedantry. 9 See the essay “Diachronic and Synchronic Modalities” later in this volume. ° A nice example of the spurious use of “true” in philosophic argumentation is provided by the paper by Lukasiewicz mentioned above. If something is the case at time ¢ then it is at any later time true that this was the case then, but not (necessarily) at any time before , Lukasiewicz says (op. cit., Section 2). If it had been true “beforehand” that a certain thing was going to be the case at ¢ then this thing would of (causal) necessity have had to be (ibid., Section 8.) Somewhat surprisingly, Lukasiewicz also says in one place that “every truth is atemporal” (ibid., Section 2.)10 Determinism and Future Truth seen, can be debunked as signifying a compound idea of two components. One of the components is “plain” atemporal truth; the other is a diachronic modality of certainty or necessity. x We now return to the puzzle about future truth. The phrase “‘it is true that” in (1) and (2) above may be given either an atemporal or a temporalized reading, i.e. be understood as referring either to the notion of “plain” truth or to truth which is certain or necessary. Moreover, if the distributive step from (1) to (2) involves a commitment to determinism, it is necessary that the phrase as it occurs in (2) be given the temporalized reading. This is what actually happens when we say, with emphasis, that if the passage from (1) to (2) is logically allowed then either the proposition that there will be a sea battle tomor- row or its negation (contradictory) is true already now. If we do not notice the distinction between the two readings of “it is true that” and do not wish to accept this deterministic conclusion, then we must either, as did Lukasiewicz, deny the logical validity of the distributive step from the Law of Excluded Middle to the Law of Bivalence or, if we do not question the validity of this step, deny the validity of the Law of Excluded Middle itself for future contingencies. When the distinction between the two readings is duly noted, the dif- ficulties disappear. “It is certain that there will be a sea battle tomorrow or there will not be a sea battle tomorrow” sounds as acceptable as “it is true that there will be a sea battle tomorrow or there will not be a sea battle tomorrow”. But it is also quite obvious that from the former it does not follow that it is certain that there will be a sea battle tomorrow or it is cer- tain that there will not be a sea battle tomorrow. The same observations hold good if for “certain” we put “fixed”, “settled”, or “necessary”. None of these words can be distributed in the same way as the word “true” when plain truth is concerned. One could also express these insights as follows: The principle which we have called the Law of Excluded Middle has the same degree of plausi- bility for plain truth as for necessary truth and certainty. But with the principle we have called the Law of Bivalence things stand differently. When plain truth is concerned it seems just as acceptable as the Law of Excluded Middle, and the distributive step from the one to the other seems unproblematic. But when the notion of necessary truth is concern- ed it seems much less obvious than the Law of Excluded Middle and the distributive step is quite clearly logically invalid. The same observation holds for certainty.Determinism and Future Truth I Thus the wedge which Lukasiewicz and some other logicians have wanted to drive between the two laws certainly is there when the type of truth under consideration is certain or necessary truth. To remove the wedge would be to lapse into determinism. But its removal would also be contrary to sound logical reasoning. What I have called the “illusion of determinism” has one of its roots in a tacit oscillation or shift in the understanding from an atemporal reading of the phrase “it is true that” to a temporalized reading of it. A crucial step in the argument which seems legitimate and uncontroversial under the first reading is quite obviously not allowed under the second reading. Nothing which has been said here in order to dispel the illusion amounts to a denial or a proposed refutation of determinism. Maybe it is already today settled whether there will be a sea battle tomorrow or not—only we do not know “which way” this has been settled. Maybe it was settled even “from the dawn of creation”. Surely there are some propositions about future events, the truth of which is certain or necessary in advance—such as, for example, forecasts of eclipses of the moon or the sun. What is, at best, controversial is whether this holds good for all propositions and whether it ever, for any proposition, does this “from the dawn of creation”. XI In order to solve the puzzle about future contingent truth it is thus not only not necessary to deny the validity of the Law of Excluded Middle. It is also not necessary to doubt its equivalence with the Law of Bivalence when ‘‘plain truth’’ is concerned. I hope that I have been able to show this convincingly. Two things, however, I have not shown. One is that the two laws actually are equivalent for the notion of truth (I shall henceforth drop the adjective ‘‘plain’”). The other is that these principles are true (valid). I shall not here discuss these questions. In another essay (‘‘Truth and Logic”, below p. 26ff.), I shall try to show, on the one hand, that the two laws of logic are equivalent, but that, on the other hand, neither of them is unrestrictedly valid for all propositions whatsoever. Before concluding the present paper, I shall return to a point which was already touched upon in the previous discussion (Section V). It concerns the supposedly otious or vacuous character of the phrase ‘‘it is true that’. Some people, including myself in past writings, have thought that there is ashort cut to the solution of the puzzles relating to future contingent truth based upon an assumed equivalence of the two schemata “‘it is true that p” and ‘‘p”’ when ‘‘p’’ stands for a sentence which expresses a proposi-12 Determinism and Future Truth tion. The reason why I think this proposal no good for solving the puzzles is that it begs the question of truth-value. I shall now elaborate my objec- tion to it in some detail. XIL As already indicated in Section V, it is not entirely clear what the pur- ported equivalence or, as it is sometimes also called, ‘‘identity’’ of the two schemata amounts to. But if it is to serve as a short cut to a solution of the problem about future contingencies it must, at least, amount to inter- substitutability sa/va veritate (in the relevant contexts). If this condition is to be satisfied, the two schemata must, when instantiated, necessarily yield propositions with the same truth-value. This means the following: If it is true that it is true that p, then it is also true that p, and vice-versa. And if it is false that it is true that p, then it is false that p, and vice-versa. There are thus four relations of implication involved. Do they all hold good? We pick out for consideration the one which may seem least obviously true, and ask: Is it necessarily the case that if it is false that it is true that p, then it is false that p? Assume that the proposition that p lacked truth-value, i.e. were neither true nor false. Then, obviously, it would be false to say that it is true that p. But it would not follow that it is false that p—since this proposition is neither true nor false. Or, to put the point about ‘‘not follow” a little dif- ferently: It would then be true that it is false that it is true that p—but it would not be true that it is false that p. Therefore the second does not follow from the first. Because that something follows from something else must here mean that if the second is true, the first is also true. And this is not now the case. Thus at least one of the four implications is not logically valid. Therefore the two schemata “it is true that p” and “p” cannot be equivalent in the sense of being intersubstitutable salva veritate either. My counterargument hinges upon the assumption that the proposition that pis neither true nor false. But is this not a queer assumption to make? Are not all propositions either true or false—“by definition” somebody may even wish to say? This question must still be discussed. Be it observed, however, that if the schema “it is true that p” = “p’ is valid for all propositions whatsoever then it follows from (1) above that it is either true that there will be, or true that there will not be, a sea battle tomorrow. It can indeed be disputed that this involves a commitment toDeterminism and Future Truth 13 determinism. But in order to dispel successfully what I called “the deter- ministic illusion” here we must not (tacitly) take for granted the very thing which some of the disputants think necessary to reject. The trick then begs the question.Demystifying Propositions I In the essay on future truth I used the term “proposition” and said that sentences may be pronounced true or false only via the propositions which they express (above, p. 6f.). Truth and falsity are commonly and naturally predicated of beliefs and judgements. Also of statements and, maybe, assertions—although, I must confess, it slightly offends my “logical ear” to hear assertions pro- nounced true or false. And to say this of sentences seems to me definitely barbaric. I have this impression independently of whether sentences are conceived of as types or as tokens. It is noteworthy, however, that the “logic” built into different languages differs here. I suppose that the German “Satz” is as adequate a translation of the English “sentence” as one might wish for. But to say of a Satz in German that it is true or false does not offend the ear. To translate into German a philosophic text in English, the author of which has carefully observed the sentence—proposition distinction, can therefore be connected with difficulties. Similarly it is sometimes awkward to render Satz in a German text consistently as “sentence” in English—as, for example, translators of Wittgenstein must have experienced. Talk of true or false propositions, using that term, can hardly be said to be part and parcel of colloquial English. “Proposition” as a technical term of philosophical logic sounds like a philosopher’s invention. I do not know the history of the term. Its use is characteristic of a Cambridge tradition in philosophy, chiefly associated with the names of C. D. Broad, W.E. Johnson, G. E. Moore, and Bertrand Russell. The ordinary way of speaking about truth which most closely parallels or corresponds to the philosopher’s talk about propositions is when we say that it is true, or false, that so and so (is the case). This, somehow, seems the most basic or direct attribution of truth-value which there is. That certain things are (“objectively”) true, others not, is what “makes” beliefs, judgements, statements true or false. The truth-value of these latter is, in a certain way, derivative or secondary. Primarily, truth and falsity belongs to the content of beliefs, judgements, and statements, i.e. to that which is believed, judged, or stated, and therewith also to that which sentences say or mean or express.Demystifying Propositions 15 It is for speaking about those “that which”-things that philosophers may find it useful to employ the term “proposition”. But use of this term is apt to engender lots of philosophical mist. Once one has the noun one also gets the idea of some entity or substance. What are propositions? What form of being or existence do they partake in? How are they related, on the one hand, to the objective features of reality and, on the other hand, to the phenomena of belief, judgement, and state- ment which are tied to persons or subjects? What, in particular, is their relationship to language? Il The bewilderment caused by the notion of “proposition” is amply reflected in the writings of the Cambridge philosophers just mentioned from the first half of this century. In his early work The Principles of Mathematics, Russell ventured to define the notion as follows. A proposi- tion, he says there,' is anything which implies itself. “Hence to say ‘pis a proposition’ is equivalent to saying ‘p implies p’; and this equivalence may be used to define propositions.” Implication is indefinable.” But “the assertion that q is true or p false turns out to be strictly equivalent to ‘p implies q’”.? From this it follows that to say that p is a proposition is equivalent to saying that pis true or pis false. “By definition” therefore a proposition, on Russell’s view here, is either true or false, has a truth- value. The above quotations from Russell will give rise to some questions. “p” and “q” are letters. “p is a proposition” is not meant to say that the letter pis a proposition. The letter is a variable, a schematic representation for something which is asserted to be a proposition. What can we substitute for “p” so as to obtain an example? A sentence? A sentence would be, for example, “it is raining”. “it is raining is a proposition” does not sound even grammatically correct. If instead of the sentence we substitute a name of the sentence and use the familiar device of forming the name of a sentence by enclosing it in quotes, we get “‘it is raining’ is a proposition”. This is grammatically in order. What it says is that a certain sentence is a proposition. “Proposition” would then name a kind of sentence— perhaps any sentence in the indicative mood. In Principia Mathematica Russell distinguished between propositions ' The Principles of Mathematics, Allen & Unwin, London, 1903, p. 15. ? Ibid., p. 14. * [bid., p. 15.16 Demystifying Propositions “considered factually” and “considered as vehicles of truth and falsehood”.* Considered factually, propositions are “classes of similar occurrences”; considered as vehicles of truth and falsehood, they are “particular occurrences” .° Occurrences of what? the reader may ask. The answer is: of sentences.® Russell is here referring to the well-known distinction between type and token of a linguistic expression. Both in Principles and in Principia, Russell thus regarded propositions as a kind of “linguistic entity”, as a “form of words”. And in Introduc- tion to Mathematical Philosophy he said expressly: ‘We mean by a “pro- position” primarily a form of words which expresses what is either true or false.’? But in a paper® from the very same year as Introduction he somewhat modified his position. He now distinguished between “word- propositions” and “image-propositions”. A word-proposition refers to the objective fact which makes it true or false and means a corresponding image-proposition.’ This presumably echoes a distinction familiar from Frege which has later come to play an important role in philosophical semantics and the philosophy of language generally. We may note here, on the one hand, a tendency to view propositions as “linguistic entities” and, on the other hand, a reluctance to identify them with merely “a form of words”. The same tension is reflected already in the well-known textbook on formal logic by J.N. Keynes.'° Here it is said: “So far as we treat of propositions in logic, we treat of them not as gram- matical sentences, but as assertions, as verbal expressions of judgements”."' But is not a verbal expression a “mere” form of words then? The question of how propositions are related to language was also of great concern to W.E. Johnson." His opinions on the matter seem to have vacillated considerably." But at least in his only major work Logic he is * A. N. Whitehead and Bertrand Russell, Principia Mathematica, Cambridge University Press, 1910; 2nd edn, 1925, Volume I, p. 664. 5 Jbid., p. 665. * Ibid. 7 Introduction to Mathematical Philosophy, Allen & Unwin, London, 1919, p. 155. “On Propositions: what they are and how they mean”, Proceedings of the Aristotelian Society, Supplementary Volume Il, 1919. Here quoted from the collection of essays by Russell, Logic and Knowledge, Essays 1901-1950, ed. by R. Ch. Marsh, Macmillan, New York, 1956. Logic and Knowledge, pp. 308-9. '© JN. Keynes, Formal Logic, 4th edn., London, 1906. " Op. cit., p. 66. "2 W.E. Johnson, Logic, Part I, Cambridge University Press, Cambridge, 1921. Cf. the statement by C.D. Broad in Examination of McTaggart’s Philosophy, Volume I, Cambridge University Press, Cambridge, 1933, p. 69f.Demystifying Propositions 17 anxious to separate propositions from their counterparts in language. “It has been very generally held”, he says,’ that the proposition is the verbal expression of the judgment; this, however, seems to be an error, because such characterizations as true or false cannot be predicated of a mere verbal expression, for which appropriate adjectives would be ‘obscure’, “‘ungrammatical’, ‘ambiguous’, etc.”. Johnson then makes a distinction between propositions “subjectively regarded” and “objectively regarded”. When viewed under the first aspect a proposition is, he says, an assertible; when viewed under the second aspect it is a possibile.'* G.E. Moore always strictly observed the distinction between proposi- tion and sentence. In the chapter “Propositions” in the early lectures published under the title Some Main Problems of Philosophy, Moore wrote: “Whenever I speak of a proposition, I shall always be speaking, not of a mere sentence—a mere collection of words, but of what these words mean.”'® The term “proposition”, he says there, “is a name for what is expressed by certain forms of words—those, namely, which, in grammar, are called ‘sentences’. It is a name for what is before your mind, when you do not only hear or read but understand a sentence. It is, in short, the meaning of a sentence—what is expressed or conveyed by a sentence: and is, therefore, utterly different from the sentence itself—from the mere words.”'” In Moore’s view, “To say that an expression is true is simply to say that it expresses a true proposition.”'® In partial agreement with the opinion Russell had expressed in Principia (cf. above p. 16) on how truth is related to the type—token distinction, Moore in lectures from the 1920s makes the following good observation: “It is primarily token-expressions which are ‘true’ in this sense: with regard to type-expressions to say that it was ‘true’ would be to say that every token-expression which was an instance of it was true, and this is a thing which it would be rash to affirm was ever the case.”"” In the lectures, Moore thus admitted that there also is a sense, albeit a secondary one, in which sentences can be true or false, viz. when they express true propositions. “We certainly do say that sentences—mere form of words—are true”, he says. But in a footnote from the year 1952 "4 Logic, I, p. 1. 'S Logic, 1, p. 14. 'S G.E, Moore, Some Main Problems of Philosophy, Allen & Unwin, London, 1953, p. 57. "” Tbid., p. 259. '® G. E. Moore, Lectures on Philosophy, ed. by C. Lewy, Allen & Unwin, London, 1966, p. 57. ” Tbid., p. 142f.18 Demystifying Propositions when the lectures were edited for publication, he added: “I see no reason now to think that we ever do call sentences or forms of words ‘true’ except in such archaic-sounding expression as ‘A true word is often spoken in jest’”.”° It was this taste which I myself acquired in the numerous discus- sions which I had with Moore during my Cambridge years in the late 1940s and early 1950s. Hl The Cambridge philosophers may be said to have agreed that truth and falsity “primarily” or “properly” belong to propositions.”! But on the nature of propositions they, as we have seen, held no very clear and still less unanimous opinion. A question which, as far as I know, none of these philosophers seriously considered was whether a proposition must have a truth-value or whether there can be propositions which are neither true nor false. Moore noted, in passing, that not all sentences express propositions. He said he “is not sure” whether “an imperative, such as ‘Go away’, expresses a proposition at all”. The reason for his hesitation was, presumably, that not all sentences express something which is either true or false. From the “definition” of the concept of a proposition which Russell gave in the Principles (see above p. 15) it seemed to follow that proposi- tions have to be either true or false. Because the Russellian relation of implication can hardly be thought to hold between entities void of truth- value. But in the Principia Russell divides propositions into significant and non-significant ones. “Significant”, he says,” means “either true or false”. Russell may thus here be said to have introduced a trichotomy “true—false—neither true nor false” in place of the traditional dichotomy “true—false” when speaking about truth and propositions. Russell only rarely uses the term “meaningless” which later became current with the logical positivists and was used by them in a sense corresponding, roughly, to that which Russell called “non-significant”. ? Some Main Problems of Philosophy, p. 262. 2! Moore, Some Main Problems, p. 63: “Propositions are, then, a sort of thing which may be properly said to be true or false.” F.P. Ramsey, “Facts and Propositions” (in The Foun- dations of Mathematics, ed. by R.B. Braithwaite, Kegan Paul, London, 1931), p. 142: “Truth and falsity are ascribed primarily to propositions.” C.D. Broad, Examination of McTaggart’s Philosophy, Volume I, p. 64: “Truth and falsity, in their primary sense, attach to propositions.” Some Main Problems, p. 61. % Principia, Volume I, p. 45.Demystifying Propositions 19 As we have seen above (pp. 15f.), Russell was, on the whole, of the opinion that propositions are linguistic entities or (even) “forms of words”. In his article on Russell’s logic in the Schilpp volume, Reichenbach hails the Russellian trichotomy as “one of the deepest and soundest discoveries of modern logic”. It was Russell’s great merit, Reichenbach says, to have realized “that the division of linguistic expressions into true and false is not sufficient; that a third category must be introduced with includes meaningless expressions.””* But if, as Moore was anxious to point out, truth and falsehood are not (directly) attributes of “forms of words” and if propositions, as he said, are the meanings of (certain) sentences, then the talk of meaningful and meaningless propositions would itself be nonsense.” The two pairs “true—false” and “meaningful—meaningless” are then simply not distinc- tions on the same logical level of discourse. Both distinctions are impor- tant; and that Russell’s work greatly contributed to making the conditions of meaningfulness a major problem in modern philosophical logic and logical semantics is undeniable. Thus we must conclude that the replacement of the dichotomy “true—false” with the trichotomy “true—false—meaningless” does not answer the question whether propositions, when conceived of as the con- tents of beliefs, judgements or statements, and not as linguistic expres- sions, must be either true or false or may be void of truth-value altogether. I shall now turn to this question. But before I answer it, I must venture an opinion of my own on the troublesome entities called “propositions”. Iv Even though I do not think it right to call propositions “linguistic entities” I think that in order to clarify (“demystify”) the notion we must start from considerations about language. I shall here regard as basic the linguistic notion of a well-formed or grammatically correct sentence (of a given language). This notion I shall not attempt to clarify or define. One may relegate the task to gram- marians and linguists—without thereby implying that the notion itself of a sentence is “philosophically unproblematic” . ** Hans Reichenbach, “Bertrand Russell’s Logic”, in The Philosophy of Bertrand Russell, ed. by P.A. Schilpp, Tudor Publishing Company, New York, 1944, p. 37. Ibid. ?6 It may be noted that Gédel in his paper “Russell’s Mathematical Logic” in the Schilpp volume does not hesitate to speak about “meaningful propositions” (p. 149).20 Demystifying Propositions Not all well-formed sentences express something which may be associated with a truth-value. For example, an imperative sentence like “Open the window!”, or an optative one like “May he come today” or an interrogative one like “What is your name?” do not do this. Aristotle mentions” prayers as examples of sentences which say neither the true nor the false. “What a handsome face!”. Is this a sentence? Not, if a sentence must contain a verb. Here the notion of sentence itself begins to waver. No philosopher would feel tempted to say that sentences such as those mentioned above constitute exceptions to the Law of Bivalence or that they express propositions which lack truth-value. The obvious thing to say is that they do not express propositions at all.”* In saying this, one would implicitly ask on what conditions a grammatically correct sentence expresses a proposition. One possible answer is that a sentence expresses a proposition only on condition that it says something which is either true or false.”? Truth-value would then “by definition” belong to propositions. But there is also another possibility to be considered. If to the sentences mentioned above we prefix the phrase “it is true that” we do not get a new well-formed sentence of the language. “It is true that open the window” is ungrammatical and therefore “meaningless”. So are, “it is true that may he come today” and “it is true that what is your name”. If prefixing the phrase “it is true that” to a well-formed sentence s yields another well-formed sentence of the language, then I shall say that the sentence s expresses a proposition and speak of the proposition expressed by s. Since the iteration of the phrase does not change the well-formed character of the sentence, it follows from our definition that also the sentence “it is true that s” expresses a proposition. (“It is true that it is true that s” is well-formed if “it is true that s” is well-formed.) A sentence s with the above properties is usually what grammarians would call an indicative sentence. But sentences other than those in the 27 De Interpretatione, Chapter IV, 1724-6. Cf. the quotation from Moore above, p. 18, fn. 22. ® Aristotle, loc. cit., 17al—4’Eou 8& Myos Gnas ev onpavtnxds ... amopavtrxds 8 ob nds, BAX év t SAndeder H yedSeobor Smdpyer. In the Loeb translation: “But while every sentence has meaning ... . not all can be called propositions. We call propositions those only that have truth or falsity in them.” Also the translation by E.M. Edghill in the Ross edition of Aristotle’s work renders &nopavtutde by “proposition”. This seems to me awkward also on independent philological grounds. A better translation of the term would be “statement”. This is used in the new translation by J.L. Ackrill Aristotle’s “Categories” and “De Interpretation”, Clarendon Press, Oxford, 1963.Demystifying Propositions 21 indicative mood can also satisfy our condition for expressing a proposi- tion. An example would be a conditional sentence such as “if he were with us, we might be saved”. Its grammatical mood is not indicative. As a common name for proposition-expressing sentences I shall use the term constative or declarative (sentence). Whether this is a “grammatical” category or not is arguable. But this question need not occupy us here. v According to the criterion which we gave, “you must not open the window” is a sentence expressing a proposition. “It is true that you must not open the window” is grammatically well formed. But now: cannot “you must not open the window” be used, on occasions, to mean exactly the same as “don’t open the window”? Obviously it can. Must we then say that the second expresses a proposition if we say that the first does? Here we must not let the notion of a proposition mystify us. When speaking about that which declarative sentences “express”, “mean” or “say” we use a linguistic device which consists in prefixing the word “that” to the sentence. We thereby transform the sentence into a that- clause, into something which is not itself a sentence. Thus we say that it is, not only true or not true, but also possible or obvious or well-known, etc. that so and so—the place of “so and so” in the schema being taken by the declarative sentence. Calling that which we are then talking about a “proposition”, or saying that the sentence in question “expresses a pro- Position”, is tantamount to saying that to turn the sentence into a that- clause is a grammatically admissible move in our language. If we say that the sentence “don’t open the window” does not express a proposition what this means is therefore simply that the move from it to “that don’t open the window” is grammatically inadmissible. But if the two sentences can be used for giving the same order (prohibi- tion), do they not, when thus used, “mean” the same? If they mean the same do they not then both express the same proposition? One could say this—but then one must be aware that “mean the same” and “express the same proposition” here means that the two sentences, although they belong to different grammatical categories of sentence, may both be used for giving the same order. I hope that what has been said will help to demystify the notion of a proposition. It should also demystify the idea of a proposition without truth-value. A prohibition stated with the words “the window must not be opened” or a permission “you may park your car here” is neither true nor false. But the form of words “it is true that the window must not be22 Demystifying Propositions opened” and “it is true that you may park your car here” are gram- matically correct. These two facts jointly may be “contracted” into saying that the propositions expressed by the two sentences, when used for giving an order or permission respectively, lack truth-value, are neither true nor false.” One could dispense with the term “proposition” entirely and say everything which is said with its aid talking only about grammatically well-formed sentences, that-clauses, and truth-values. Since the term is apt to create confusion and pseudo-problems, it may even be advisable to dispense with it. As a terminological (linguistic) device it is nevertheless useful and I shall continue to employ it now that, at least in principle, we need not let ourselves be mystified by it any more. But I shall avoid such locutions as calling propositions the reference of that-clauses or the mean- ings of sentences because these locutions are unnecessary and induce us to talk of propositions as of some entities with a shadow of existence. VI To say that some propositions lack truth-value, are neither true nor false, thus means, on the ruling which we have adopted, that some (tokens of some) grammatically correct sentences of the form “it is true that s” (where “s” is itself a grammatically correct sentence) say something which is neither true nor false. Weare already familiar with examples. Sentences such as “you ought to open the window” or “one must not smoke in this room”, when used for giving prescriptions (orders, norms, rules), do not say anything which is true or false. According to many philosophers, evaluative sentences, such as e.g. “it is better to suffer evil than to do evil” belong in the same 2° “The two sentences” here means instances or tokens of the two sentence-types or of what Russell (see above p. 16) called “classes of similar occurrences”. Tokens of the types are sometimes used prescriptively, sometimes descriptively. When used prescriptively they order or permit something; when used descriptively they state that something has been ordered or permitted. In the second case only do the sentences mean or say something which is either true or false. It would be a mistake to think that the use of the phrase “it is true that” would, “by itself”, serve to distinguish the two cases. Consider the following dialogue: “You may park your car here.” “True?” “Yes, yes, it is true that you may park your car here.” Adding the phrase “it is true that” does not make the sentence-token express a true or false proposition. Whether it does this or not depends upon whether I use it—with or without the addition of “it is true that”—for giving permission (myself) or for informing my interlocutor that parking has been permitted (by the authorities) in this area.Demystifying Propositions 23 category. Philosophers of the logical positivist orientation of the mid- century sometimes called such sentences “meaningless”. “Meaningless” in a more substantial sense are sentences which, although correctly formed of familiar words and declarative in form have no use in our language because of lack of criteria for pronouncing them (what they say) true or false. Consider, for example, the sentences “prime numbers are green” or “courage is round”. Do we understand these sentences? Yes and no. We understand that the first attributes a certain colour to certain numbers, but we cannot “make sense” of the attribu- tion. One is here almost tempted to say that the meaninglessness of the sentence is shown by its meaning!*! The meaninglessness is not a matter of grammatical correctness—like that of the sentence “Socrates is iden- tical”. It is not exactly a matter of logical correctness either. The lack of “meaning”, as already indicated, is due to the lack of truth-criteria. We could invent such criteria, for example give some rule for associating colours with numbers—and then it might be of interest (“make sense”) to find out whether prime numbers have this colour or that one. The sentence that such numbers are green would no longer be meaningless; and similarly for traits of character and geometrical shapes. A different kind of example is provided by sentences which, seemingly, have a clear meaning but the truth-value of which, for reasons of a logical nature, cannot be decided or established. Consider, for example, the statement that the relative frequency with which a certain characteristic occurs in the members of a potentially infinite extensional sequence such as, say, throws with a coin or a dice, approaches as a limit a given value. As is well known from the philosophy of probability, such a limiting fre- quency statement cannot be conclusively verified or falsified. But is it not nevertheless true or false? Some would perhaps say that it is, although one cannot decide which truth-value it has. Since, however, the impossibility of coming to know its truth-value is not due to any shortcoming of our epistemic faculties, but to the fact that there is no state of affairs cor- responding (“objectively”) to its truth or falsehood, it is a feasible thing to say here that the proposition asserted with the statement actually is one which has no truth-value. Some logical positivists of the 1920s and 1930s would have labelled also such statements “meaningless”. The logical positivists were also worried about unrestricted universal and existential propositions. The first cannot be conclusively verified in extension, i.e. on the basis of facts of experience; the second again cannot be conclusively falsified. Hence one may say that there is no state of *" This was how G. E. Moore once expressed himself in a conversation with me.24 Demystifying Propositions affairs (fact) answering to the truth of the first or to the falsehood of the second. It may be difficult to defend the view that unrestricted universal or existential generalizations are not “genuine” propositions if by “pro- position” one means something which is either true or false. Yet it is not unproblematic to say that they are such. (I shall return to this problem later; see below, pp.107ff.) VII We have already (p. 7) encountered the distinction between generic and individual propositions. This distinction has a bearing on the question of associating truth-values with propositions. Consider the proposition that it is raining. Is it true or false? One could answer: the proposition is either true or false of any given place and time—but unless a place and a time is specified its truth-value is indeter- minate. It is not, “by itself”, true or false. It is true of some places at some times and false of some places at some other times. Its truth-value can thus also be said to vary with variations in the spatio-temporal determination. The needed determination is often supplied by the context of a state- ment. If I make the statement that it is raining without saying where and when I should normally mean that it is raining then in the place where I happen to be. If, however, I am speaking about remote places or times, I ought usually to supply the spatio-temporal coordinates in language. When this is done, the sentence-token used in making the statement expresses a univocally true or false individual proposition. Sentences which are well-formed and admit the prefix “it is true that” but which need a linguistic supplementation in order univocally to express a proposition with a truth-value, I shall in conformity with received terminology call open sentences. When the appropriate linguistic supplementation is made, they become closed. Thus the sentences “it is raining” and also “it is true that it is raining” are open; the sentences “it is raining in Paris on 12 November 1980” and “it is true that it is raining in Paris on 12 November 1980” are closed. Consider the sentence “it is green”. It is grammatically well-formed. So also is “it is true that it is green”. But unless it is made clear what the “it” refers to, we cannot tell whether the sentence expresses a true or a false proposition or, maybe, a proposition without truth-value. The reference of the word “it” may be clear from the context or it may be indicated in language by a name or a definite description which can take its place. “The dome of the Pantheon in Paris is green” expresses a true proposi- tion; “the number 7 is green” presumably expresses a proposition which is neither true nor false.Demystifying Propositions 25 The open sentence “it is green” may be said to express a generic propo- sition with a variable truth-value. Since the linguistic supplementation here needed to close the sentence consists in replacing the word “it” by a name or descriptive phrase, this word is also said to be a variable. In books on logic one usually employs letters, say x, to indicate such variables or open places in a sentence. If the variable is replaced by the name of some thing and the sentence thus obtained expresses a true individual proposition, we may say that the generic proposition is true of that thing. One could of course say that the generic propositions which open sentences express have no truth-value, are neither true nor false. But the way in which they lack truth-value is quite different from the way in which some other propositions may be said to do this. Generic propositions are not by themselves true or false. They are, in a characteristic sense, “in- complete” or “unsaturated”. They have a “gap” in themselves which, when filled, may turn them into true or false (individual) propositions. This is why I prefer to say that they have an indeterminate or variable truth-value rather than saying that they /ack truth-value. VU Assume that the proposition expressed by “s” is neither true nor false. What then about the proposition expressed by the sentence “it is true that s”? This proposition has a truth-value. It is false, i.e. its negation is true. Because if it is neither true nor false that s, then it is obviously false to say that it is true that s—or, which is merely a verbal transformation, true to say that it is not true that s. This idea is fundamental to the “Logic of Truth” which I am going to construct next. It is therefore important that it be clearly grasped. The idea is that the attribution of truth-value to a proposition which lacks truth-value yields a proposition which is false—and not a proposition which is itself void of truth-value.Truth and Logic I I shall construct a calculus which I propose to call the Logic of Truth or Truth-Logic or, for short, TL. An alternative name would be Alethic Logic, from the Greek word for truth, &A1@e1. Its basic vocabulary con- sists of the following ingredients: (1) Variables p, g, ..., standing for declarative sentences, i.e. sentences which allow their transformation in language into a that-clause (see above, p. 21.) (2) Two sentential connectives— ~ for negation, corresponding to the word “not” of natural language, and & for conjunction, corresponding to the word “and”. (3) An operator T which reads “it is true that”. (4) Brackets. By definition, three more sentential connectives are introduced, viz. v , ~—,and +. They correspond, roughly, to the words “or”, “if ... then”, and “if, and only if, ... then” of ordinary language. But their precise meanings are given by their definition in the terms of negation and con- junction: s v s’ = g~(~ s& ~ s'), ss’ = g~(s& ~ s’) and sos’ = g~(s& ~ 5s’) & ~ (~ 5&5"), where s and s’ are meta- variables representing arbitrary sentences of the calculus. Il The well-formed expressions or formulas of the calculus will be called T- sentences (-expressions, -formulas). They are either atomic or molecular. An atomic 7-sentence consists of the operator T followed by a variable or by a molecular compound of variables or by an atomic 7-sentence or by a molecular compound of atomic 7-sentences or, finally, by a variable or molecular compound of variables and atomic 7-sentences. Thus, for example, Tp is an atomic T-sentence, and soisalso T(p v g)and TT ~ p and T(~ Tp & T(q > r)) and T(p > Tp). A molecular T-sentence is a molecular compound of atomic 7- sentences. For example: Tp > T ~ Tp. The conventions for bracketing expressions and for omitting brackets ITruth and Logic 27 shall not explain here. I shall regard them as being either known or self- explanatory. Il The “bases” of the calculus are the following five axioms: Al. T~p-7 ~ Tp. A2. Tp-+T~ ~p. A3. T(—p&q) ~ Te& Tq. A4. T~(&Q)oT~pvT~q. AS. T~ Tp + ~ Tp. Falsehood may be defined as the truth of the negation (contradictory) of a proposition (cf. above, p. 2). What the five axioms say can then be stated in words as follows: AI says that a false proposition is not true. A2 says that a proposition is true if, and only if, its negation is false. A3 says that the conjunction of two propositions is true if, and only if, both conjuncts are true. Ad says that a conjunction of two propositions is false if, and only if, at least one of the conjuncts is false. AS, finally, says that it is false that a proposition is true if, and only if, the proposition in question is not true. If in A4 we substitute “~ p” and “~ q” for “p” and “q” and apply to the left membrum of the equivalence the definition of disjunction and simplify the right membrum by virtue of A2, we obtain the formula T(p v q) + Tp v Tq. This may be regarded as an alternative form of A4.! The most interesting, and perhaps controversial, axiom is AS. In order to see what is interesting about it, consider a proposition to which we do not wish to accord truth-value, for example, that prime numbers are green. Then it is not true that prime numbers are green. Nor is it false. | There is also another, more restrictive, way of laying down the truth-condition for dis- junction. One would then say that a disjunction of two propositions is true if, and only if, both disjuncts are true or one is true and the other false. This is equivalent with saying that a conjunction is false if, and only if, both conjuncts are false or one is false and the other true. On this more restrictive conception, A3 and A4 together specify the truth-conditions for a conjunction (disjunction) in such a way that, if one conjunct (disjunct) happens to be neither true nor false, then the conjunction (disjunction) is void of truth-value, too. The choice be- tween the two conceptions of the truth-conditions seems to me very much a matter or arbit- rary convention. I have previously myself favoured the more restrictive view. Ad is then given the form T ~ (p&q) ~ T~ p&T~ qv Tp&T ~ qv T ~ p& Tq. Cf. my paper “Truth as Modality” in Modality, Morality and Other Problems of Sense and Nonsense, Essays dedicated to Séren Halldén, Gleerup, Lund, 1973.28 Truth and Logic This being so, it is true that it is not true that prime numbers are green and, since “true that not” means “false”, it is false that it is true that prime numbers are green. What AS says, in effect, is that although there may exist propositions which are neither true nor false, any (declarative) sentence beginning “it is true that” expresses a proposition which is either true or false (cf. above p. 12 and p. 25). Iv The class of theorems of our Logic of Truth or T-Logic is defined as follows: (1) A T-formula which is a substitution instance of a tautology of “classical” Propositional Logic (PL) is a theorem of TL. A substitution instance is that which we get when we replace all variables in a tautology of PL by 7-formulas. For example: Pv ~ pisa tautology of PL, hence Tp v ~ Tpisa theorem of TL. (2) The axioms Al—AS of TL are theorems. (3) Theorems, finally, are all T7-formulas which can be obtained from theorems of TL with the aid of one or several of the following rules of transformation: RI. Substitution, i.e. the replacement of variables p, g, ... by other variables or by molecular compounds of variables or by T-formulas. For example: since Tp v ~ Tpisatheorem, T ~ Tp v ~ T ~ Tpisalsoa theorem. R2. Ifsands— s’ are theorems, then s’ is a theorem too. The Rule of Detachment or of modus ponens. R3. If s is a theorem, then 7s is a theorem too. For example: Tp v ~ Tp isa theorem. Hence T(Tp v ~ Tp) is also a theorem. The Rule of Truth. Vv We need for our purposes three meta-theorems of our Logic of Truth. Ml. Ifs +s’ isatheorem of TL, then sand s‘ are intersubstitutable salva veritate in T-formulas. This I shall call the Rule of Exten- sionality or “Leibniz’s Law” for truth-logic. Sketch of a proof. Let s ~ s’ be a theorem of TL. From the correspond- ing tautology of PL we obtain the theorem of TL (s + s’) > ~ (S& ~ s')& ~ (~ s&s’). By the Rule of Detachment we get ~ (s & ~ s’)Truth and Logic 29 & ~ (~ s&s’), by the Rule of Truth T(~ (s& ~ s’)& ~ (~ s&s')), and from this by A3 T ~ (s & ~ s’)& T ~ (~ s & s’). From a corresponding tautology of PL we obtain T ~ (s& ~ s'‘)}& T~ (~ s& 5’) > T ~ (s& ~ s’) whereupon we can detach T ~ (s & ~ s’). To this last we apply A4 and A2 and get T~ s v Ts’. By Al wehave T~ s> ~ Ts. From the last two formulas, in combination with PL-tautologies and R2, we obtain ~ Ts v Ts’ and from this Ts > Ts’. By symmetrical reasoning, applied to T ~ (~ s&s’) we get Ts’ — Ts. The two implica- tion formulas, finally, yield Ts + Ts’. We have then established that, if s +s’ isatheorem of TL, then Ts + Ts’ is also a theorem of TL. This constitutes the essentials of a proof that provably equivalent formulas of TL are intersubstitutable salva veritate. M2. Every 7-formula is equivalent with a 7-formula of the first order, i.e. with one in which no symbol T occurs within the scope of another 7. We sketch a proof of this for strings of symbols 7 with or without a negation-sign occurring between them. By R1, substituting ~ Tp for pin AS, we obtain T ~ T ~ Tp + ~ T ~ Tp. By virtue of M1 and AS, wecan in this formula replace the parts T ~ Tp by ~ Tp. This givesus T~ ~ Tp ~ ~ ~ Tp. By virtue of A2, we simplify the left member of the equivalence to TTp and by virtue of PL, we simplify ~ ~ Tp to Tp. Herewith we have proved the following theorem: Tl. TTIp + Tp. The theorem says that the phrase “it is true that it is true that” equals “it is true that”. It is now possible, by successive applications of M1, A5, A2, and T1 to contract any string of 7s with or without negation-sign between them to one single occurrence of the symbol T with or without negation-sign in front. The extension of this result to overlaps generally of Ts, whether in strings of immediate succession or not, follows easily from the next meta- theorem. 2 If we had accepted TI as one of the axioms of TL, AS would have been provable. This is seen as follows: As already noted (p. 28), T(7p v ~ Tp) is a theorem and provable in- dependently of AS. Hence, by A4 and the definition of disjunction “p v q” = ar ~ (~ p& ~ q)” and A2, we have TTp v T ~ Tp. By virtue of T1 and principles of PL, we obtain from this ~ Tp + T ~ Tp. The converse is proved starting from AJ and substituting in it ~ Tp for p. We get T ~ Tp + ~ T ~ ~ Tp. Cancelling double negation by virtue of A2 gives us T ~ Tp + ~ TTp, and from this and T1 and principles of PL we obtain T ~ Tp + ~ Tp. Herewith the equivalence T ~ Tp ~ ~ Tp or AS has been proved from T1 and the other axioms of TL. I am indebted to Mr Antti Hautamaki for these observations.30 Truth and Logic M3. Every T-formula is provably equivalent with a molecular compound of atomic 7-formulas of the simple form con- sisting of the letter T followed either by a single variable or by the negation of a single variable. These atomic formulas will also be called sruth-constituents or T-constituents of the original T-formula. Sketch of a proof. Consider first an atomic T-formula 7(...). If the expression in front of which the first T stands contains signs for disjunc- tion, implication, or equivalence, we replace the parts containing these signs with parts containing only signs for negation and conjunction according to the given definitions. Any occurrence of the letter T in the expression thus transformed which stands in front of a conjunction or a negation of a conjunction can now be distributed by virtue of (M1 and) the axioms A3 and A4. Strings of negations may be reduced by virtue of A2, strings of successive Ts may be reduced by virtue of T1, and occurrences of T ~ Tmay be reduced to ~ T by virtue of AS. By repeated application of these distributive and reductive operations we reach, in a finite number of steps, a stage when no further distribution or reduction is possible. In the formula with which we have then ended after the transformation all symbols T occur immediately before a variable with or without a negation sign in front. Trivially, by applying the same procedures to all the atomic 7-formulas of which a given T-formula is a compound, we transform the given T- formula into a compound of T-formulas of the simple kind we have called truth-constituents. VI Consider an arbitrary 7-formula. In it occur in all n variables p, q, etc. The 2n atomic T-sentences Tp, T ~ p, Tq, T ~ q, etc. are the truth consti- tuents or T-constituents of this 7-sentence. According to M3, the given T-sentence is equivalent, in TL, with a molecular compound of its T- constituents. (Not all 7-constituents need actually appear in the compound; missing constituents can, if needed, be vacuously introduced through operations of PL.) Consider the two T-sentences Tp and T ~ p. According to Al, they cannot both be true. If the one is, the other is not. But nothing which we have laid down about TL excludes the possibility that neither of them is true. The sentences Tp and T ~ pcan thus say or not say the true in 3 different combinations. The same holds for the pair Tg and T ~ q. The four sentences can thus say or not say the true in 3 x 3 or 3? different combina-Truth and Logic 31 tions. The 2” T-constituents of a given 7-sentence can be true or not in 3” combinations. Since the given 7-sentence is equivalent with a molecular compound of its T-constituents, one can use a truth-table to investigate which truth- function of the constituents the sentence itself expresses. If it expresses their tautology, we shall call the given sentence (or the proposition which it expresses) a truth-tautology or T-tautology. Whether the given 7-sentence is, or is not, a 7-tautology can thus always, after the appropriate transformation of the sentence into a compound of T-constituents, be investigated and decided in a truth-table. The notion of a T-tautology provides the obvious criterion of logical truth in TL. M4. All theorems of TL are 7-tautologies, and all 7-tautologies are theorems of TL. The proof of this metatheorem is an adaptation of well-known analogous results in modal logic. Vu Instead of conceiving of the formulas of the T-calculus as two-valued truth-functions of pairs of truth-constituents of the forms Tp and T ~ p, Tq and T ~ q, etc. we can also interpret them as three-valued “truth”- functions of the variables p, q, etc. themselves. This connects our truth- logic with a polyvalent, in this particular case three-valued logic. Its three values are “true”, “false” and “neither true nor false”. We can designate them by 1, 0, and } respectively. The distributions of the three values over the variables in a table would then be subject to no restriction. The values of the pairs of T-constituents are calculated in accordance with the table below. The calculation of values of compounds of 7-constituents then proceeds in the “classical”, i.e. two-valued, way. Tp T~p One |D oon “oo As shown by the table, the value of a truth-constituent is always either “true” or “false”, although the proposition itself, of which the constituent affirms truth or falsehood, may be void of truth-value, be neither true nor false (cf. above p. 25 and p. 28).32 Truth and Logic VU We can now deal with the question of the putative vacuousness of the phrase “it is true that” and of the possible equivalence or non-equivalence of the two Laws of Excluded Middle and of Bivalence respectively. T(Tp = p) which would say that the equivalence (“identity”) of Tp and pisatruth, is not a theorem of TL. This is easy to grasp intuitively and the reasons we have already given (see above p. 5 and pp. 12f.). But let us look at the matter in more detail by transforming the formula into a compound of T-constituents Replacing the equivalence sign by its definition, we first obtain 1(~ (Tp) & ~ p) & ~ ( ~ Tp) & p)). By virtue of A3, we distribute the Tin front of the whole expression and get T ~((Tp) & ~ p) & T ~((~ Tp) & p). By virtue of A4, we distribute the negations in front of conjunctions and get (T ~ Tp v T ~ ~ p)&(T ~ ~ Tp v T ~ p). By virtue of A2, A5, and TL, we can simplify the last expression to (~ Tp v Tp)&(Tp v T ~ p)whichisthe same as Tp v T ~ psimpliciter. Herewith has been shown that the two formulas T(Tp + p) and Tp v T ~ pare equivalent. Thus the first would be a theorem of TL if, and only if, the second were a theorem. But the second is not a theorem, because, as a truth-table would show, Tp v T ~ p is false when the proposition that p is itself one which is neither true nor false, i.e. lacks truth-value. But now it is also clear under which cirumstances the equivalence between “it is true that p” and “p” holds good, i.e. T(Tp ~ p) would bea theorem. It holds good if, and only if, it is either true or false that p. So, for propositions which can safely be assumed to have (or are already presupposed to have) a truth-value, the addition of the phrase “it is true that” to the sentence expressing the propositions is otiose or vacuous. But it should also once again be clear why an attempt to solve Aristotle’s puzzle about future contingencies by a resort to this equivalence “begs the question” (cf. above p. 12f.). We thus have a theorem of TL which says T2. (Ip®T ~ p)~> T(Tp ~ p). “If itis either true or false that p, then it is true that it is true that p if, and only if, p.” Another theorem is T3. Tip v ~ p)~ Tp v T ~ p. “It is true that p or not pif, and only if, it is either true or is false that p.” The proof involves simple applications of A4 and A2. This theorem is theTruth and Logic 33 syntactic equivalent of our statement (above p. 11) that the Law of Excluded Middle and the Law of Bivalence actually are but two different expressions of the same logical principle. But it should also be noted that neither T(p v ~ p) nor Tp v T ~ pare, by themselves, theorems, i.e. logical truths of truth-logic. They are not valid for all propositions, i.e. for all admissible values of the variable “p” , whatsoever. They hold good only on the prior assumption that the proposition in question has a truth- value. This condition reduces the statement of the Law of Bivalence to a sheer triviality of PL, viz.to Tp v T ~ p— Tp v T ~ p. The statement of the conditional validity of the Law of Excluded Middle, viz. Tp v T ~ p-—T(p v ~ p), can be proved only in TL. Ix No proposition to the effect that a tautology of “classical” Propositional Logic is true is a theorem of Truth Logic. No tautology of PL, one could say with an air of paradox, is tautologically true. Any substitution instance, however, in TL of a tautology in PL is a tautology of TL. As already noted in passing (above p. 28), we have a theorem T4. Tp v ~ Tp. It says that an arbitrary proposition is either true or is not true. By the Law of Excluded Middle we previously (p. 2) understood the assertion that the disjunction of a proposition and its negation (necessarily) is ttue—and by the Law of Bivalence the assertion that any proposition (necessarily) is either true or false. We have shown that these two “laws” amount to the same (in TL). We could, if wanted, introduce a slightly revised terminology and decide to understand by the Law of Excluded Middle the principle which says that any given proposition is either true or is not true. This would distinguish the Law of Excluded Middle from the Law of Bivalence and make it possible for us to say that the first is unrestrictedly valid for all propositions whatsoever, whereas the second is not. As also already (p. 28) noted, since Tp v ~ Tp is a theorem, then by the Rule of Truth 7(7p v ~ Tp) isatheorem, too. From it is easily derived TS. TTp v T ~ Tp. This theorem says that any proposition of the form of what we have called a truth-constituent is either true or false. Generally: any proposition to the effect that it is true that something or other (is the case) is either true or false. With this too we are already familiar (above p. 28). Alis, by PL and A2, equivalent with ~ (Tp & T ~ p). This again is, by34 Truth and Logic A3, equivalent with ~ T(p & ~ p). It is thus a logical truth of truth logic that a contradiction in the sense of PL is not true. But it is not a logical truth of TL that a contradiction is false. T ~ (p & ~ p) is not a theorem. The formula is equivalent with Tp v T ~ p which, as we know, is not a theorem. Substituting in Al “~ p” for “p” and applying A2 we obtain Tp > ~ T ~ p from which, by AS, we obtain Tp > T ~ T ~ p. “Truth entails the falsehood of falsehood”. But the converse does not hold. Falsehood of falsehood does not necessarily amount to truth. Some of the above observations will remind one of certain peculiarities of intuitionist logic. (Some further similarities will become manifest when we proceed to quantification.) It seems to me that some of the “intuitions” to which Brouwer gave expression in his philosophy of mathematics are captured and given a natural interpretation in our Logic of Truth and its extension to a Logic of Predication (see below, pp. 42ff.). x Someone may find it baffling, or “suspect”, that TL should use freely the “inferential machinery” of PL and recognize any substitution instance in TL of a PL-tautology as a T-tautology—and yet refuse to admit as tauto- logically true any statement to the effect that a PL-tautology is true. But this need not baffle us at all. It is the “point” about TL that it is designed to do justice to the idea that beside declarative sentences which express true or false propositions there are also such which neither say the true nor the false. Accepting this, we must of course also think that molecular compounds of sentences void of truth-value are themselves void of truth- value—even if they happen to have the form of tautologies. A compound of the form of a tautology is tautologically true, however, if its atomic constituents are themselves true or false. P ~P T F F T It is this very assumption, viz. that the constituent propositions have truth-value, that underlies the traditional interpretation or modelling of the expressions of PL in truth-tables. Consider, for example, the truth- table above, for negation. How do we argue to this correlation of Ts and Fs? As follows: if it is true that p, we think, then it is false that not-p and if it is false that p then it is true that not-p. And these relations do indeedTruth and Logic 35 hold—and can be expressed in the form of theorems of TL. (Tp > T ~ ~ p and T ~ p > T ~ pare T-tautologies.) But what justifies us in thinking that T and F exhaust all possibilities of assigning a value to a given propo- sition? We just assume that propositions take one, and one only, of these two values—not both at once, or neither of them. But this assumption can itself be questioned. What if instead of “true” and “false” we had said “true” and “not true”? Necessarily, a proposition is either true or is not true, even though it need not be true or false. At least this is so according to our truth-logic. (Ip v ~ Tpisa T-tautology but Tp v T ~ pis not, we have said repeat- edly.) We therefore rewrite the truth-table as shown below. Now we can be sure that the two possibilities are at least exhaustive. We then argue: if it is true that p, then it is not true that not-p; and if it is not true that p, then it is true that not-p. But here the difficulty is with the second implica- tion. Tp > ~ T ~ pisa tautology of TL, but ~ Tp > T ~ pis not. The second implication holds only if it is assumed, not only that the proposi- tion under investigation is either true or not-true, which is trivial, but also that it is either true or false, which is not trivial. P ~P T ~T ~T T So also when the truth-table is constructed in this way, the presupposition is that propositions are true or false. The notion of a tautology, it should be remembered, is a notion of logical semantics, not of syntax. A tautology is a propositional compound of a well-defined type which is true for all distributions of truth-values over its components. But in distributing truth-values in truth-tables it is assumed or presupposed that these components are either true or false and that, if they are true, they are not false, and if false, not true. For this reason one can say that the Laws of Bivalence and of Contradiction are basic to the semantics of “classical” PL.’ With TL the situation is different. Its semantics does not presuppose the Law of Bivalence. Not any proposition will be either true or false. But any proposition to the effect that a proposition is true will be this, i.e. will 3 On this, the reader is also referred to my early papers “On the Idea of Logical Truth I” (1948) and “Form and Content in Logic” (1949), reprinted in Logical Studies, Routledge & Kegan Paul, London, 1957.