Deepening our Understanding of Wittgenstein

DEEPENING OUR UNDERSTANDING OF WITTGENSTEIN

Grazer Philosophische Studien INTERNATIONALE ZEITSCHRIFT FÜR ANALYTISCHE PHILOSOPHIE

GEGRÜNDET VON Rudolf Haller HERAUSGEGEBEN VON Johannes L. Brandl Marian David Leopold Stubenberg

VOL 71 - 2006

Amsterdam - New York, NY 2006

DEEPENING OUR UNDERSTANDING OF WITTGENSTEIN

Edited by

MICHAEL KOBER

Die Herausgabe der GPS erfolgt mit Unterstützung des Instituts für Philosophie der Universität Graz, der Forschungsstelle für Österreichische Philosophie, Graz, und wird von folgenden Institutionen gefördert: Bundesministerium für Bildung, Wissenschaft und Kultur, Wien Abteilung für Wissenschaft und Forschung des Amtes der Steiermärkischen Landesregierung, Graz Kulturreferat der Stadt Graz

In memoriam Georg Henrik von Wright

The paper on which this book is printed meets the requirements of “ISO 9706:1994, Information and documentation - Paper for documents Requirements for permanence”. Lay out: Thomas Binder, Graz ISBN-10: 90-420-2010-5 ISBN-13: 978-90-420-2010-8 ISSN: 0165-9227 © Editions Rodopi B.V., Amsterdam - New York, NY 2006 Printed in The Netherlands

TABLE OF CONTENTS

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii

System of Reference to Wittgenstein’s Writings . . . . . . . . . . . . . . .

ix

Hans SLUGA: Family Resemblance . . . . . . . . . . . . . . . . . . .

1

Klaus PUHL: Only Connect … — Perspicuous Representation and the Logic of Nachträglichkeit . . . . . . . . . . . . . . . . . . . .

23

Joachim SCHULTE: The Pneumatic Conception of Thought . . . . . . .

39

Felix MÜHLHÖLZER: “A Mathematical Proof Must Be Surveyable”: What Wittgenstein Meant by This and What It Implies . . . . . . .

57

Michael KOBER: Wittgenstein and Religion . . . . . . . . . . . . .

87

Wolfgang KIENZLER: Wittgenstein and John Henry Newman on Certainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

117

Hans-Johann GLOCK: Thought, Language, and Animals . . . . . . . .

139

Severin SCHROEDER: Moore’s Paradox and First-Person Authority

161

Eike VON SAVIGNY: Use, Meaning, and Theoretical Commitment

175

Wolfgang HUEMER: The Transition from Causes to Norms: Wittgenstein on Training . . . . . . . . . . . . . . . . . . . .

205

Barbara SCHMITZ: Grammatical Propositions . . . . . . . . . . . .

227

Richard RAATZSCH: On Knowing What One Does . . . . . . . . . .

251

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), vii.

INTRODUCTION From the 1st to the 3rd of July 2004, a group of German-speaking Wittgenstein scholars met at Ulm, Southern Germany, in the Villa Eberhardt. Their papers, mostly distributed before the meeting, were succinct, the discussions intense, and criticism constructive. No doubt, what is published here had its origins in the papers presented at Ulm, though many contributions were substantially revised as a result of the ensuing discussions. They all go to show that Wittgenstein scholarship — both in terms of exegesis and systematic relevance — is flourishing. As the title of the present volume indicates, the following papers aim at improving our understanding of Wittgenstein’s thinking (Richard Raatzsch’ contribution which consists of philosophizing in a Wittgensteinian manner might be viewed as an exception). Andreas Kemmerling played an active role in the discussions but refrained from publishing anything here. Although Barbara Schmitz had prepared a paper, she was later unable to join the group in Ulm; her contribution is printed here nevertheless. The sequence of papers does not follow any particular order. The Wittgenstein Workshop, as it was called, was made possible by funding from the Ulmer Universitätsgesellschaft e.V., by the hospitality of the University of Ulm and Villa Eberhardt, and by the organisational support of the Humboldt-Studienzentrum für Philosophie und Geisteswissenschaft at the University of Ulm. The Philosophy Department of the University of Freiburg (Germany) helped support the editorial process, and collaboration with the Grazer Philosophische Studien was a pleasant and rewarding experience. All support is gratefully appreciated. Michael Kober

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), ix–xi.

SYSTEM OF REFERENCE TO WITTGENSTEIN’S WRITINGS Reference to works and other sources (conversations, lectures, letters, manuscripts …) of Wittgenstein are made by means of the following abbreviations (listed in alphabetical order). Unless otherwise clarified, all references are either to numbers of sections, indicated by “§” (e.g. “OC § 206”), or to pages of the edition cited (e.g. “BB 27”). Some authors refer by “191a” to the first paragraph on page 191, by “191b” to the second paragraph, etc.; similarly, “PI § 7c” denotes the third paragraph of PI § 7. AWL BB BT

CB CD

CE CV

FF GB KGE

Wittgenstein’s Lectures, Cambridge 1932–1935, from the notes of A. Ambrose and M. MacDonald, ed. by A. Ambrose, Oxford: Blackwell 1979. The Blue and Brown Books, Oxford: Blackwell 1958. The Big Typescript (= TS 213), ed. by M. Nedo, Vol. 11 of the Wiener Ausgabe / Vienna Edition, Wien, New York: Springer 2000. References are made not to the pagination of Nedo’s edition, but to the pagination of TS 213 (which is included in Nedo’s edition). O. K. Bouwsma, Wittgenstein, Conversations 1949–1951, ed. by J. L. Craft and R. E. Hustwit, Indianapolis: Hackett 1986. “Some Notes on Conversations with Wittgenstein” and “Conversations with Wittgenstein”, by M. O’C. Drury, in R. Rhees (ed.), Ludwig Wittgenstein, Personal Recollections, Totowa, NJ: Rowman & Littlefield 1983, 91–189. “Cause and Effect: Intuitive Awareness”, as printed in PO 368–426. Culture and Value / Vermischte Bemerkungen, A Selection from the Posthumous Remains, ed. by G. H. von Wright in collaboration with H. Nyman, revised second edition of the text by A. Pichler with English translation by P. Winch, Oxford: Blackwell 1998. Frühfassung or early version of the Philosophical Investigations, TSS 225, 220, 221, as printed in KGE. “Remarks on Frazer’s Golden Bough”, as printed in PO 115–155. Philosophische Untersuchungen, Kritisch-genetische Edition (= criticalgenetic edition), ed. by J. Schulte, in Zusammenarbeit mit (in collaboration with) H. Nyman, E. von Savigny, und G. H. von Wright, Frankfurt am Main: Suhrkamp 2001.

LC LE LPE LPP

LSD LW I LW II LWL ML

MS

NB

NL OC PG PI

PI II

PPO

Lectures and Conversations on Aesthetics, Psychology and Religious Belief, ed. by C. Barrett, Oxford: Blackwell 1966. “A Lecture on Ethics”, as printed in PO 36–44. “Notes for Lectures on ‘Private Experience’ and ‘Sense Data’”, as printed in PO 200–288. Wittgenstein’s Lectures on Philosophical Psychology 1946–47, notes by P. T. Geach, K. J. Shah and A. C. Jackson, ed. P. T. Geach, Hassocks: Harvester Press 1988. “The Language of Sense Data and Private Experience — Notes taken by R. Rhees of Wittgenstein’s Lectures, 1936”, as printed in PO 289–367. Last Writings on the Philosophy of Psychology, Vol. I, ed. by G. H. von Wright and H. Nyman, Oxford: Blackwell 1982. Last Writings on the Philosophy of Psychology, Vol. II, ed. by G. H. von Wright and H. Nyman, Oxford: Blackwell 1992. Wittgenstein’s Lectures, Cambridge 1930–32, from the notes of J. King and D. Lee, ed. by D. Lee, Oxford: Blackwell 1980. Letters to Moore, as printed in Briefe, ed. by B. F. McGuinness and G. H. von Wright, correspondence with B. Russell, G. E. Moore, J. M. Keynes, F. P. Ramsey, W. Eccles, P. Engelmann, and L. von Ficker, Frankfurt am Main: Suhrkamp 1980; references are to pages and/or to dates as specifically as possible. Manuscript from the Nachlass, available on CD-ROM as Wittgenstein’s Nachlass: The Bergen Electronic Edition, Oxford: Oxford University Press 1998ff. References are to manuscript-numbers of von Wright’s catalogue (in: G. H. von Wright, Wittgenstein, Oxford: Blackwell 1982, 35–62), followed by page number. Notebooks 1914–1916, ed. by G. E. M. Anscombe and G. H. von Wright, Oxford: Blackwell 1979; references are either to pages or to dates of the entries. Notes on Logic, as printed in NB 93–107. On Certainty, ed. by G. E. M. Anscombe and G. H. von Wright, Oxford: Blackwell 1969. Philosophical Grammar, ed. by R. Rhees, Oxford: Blackwell 1974. Philosophical Investigations, Part I, ed. by G. E. M. Anscombe and R. Rhees, Oxford: Blackwell 1958, 1st edition 1953; references are to sections, except for footnotes. Philosophical Investigations, Part II, as printed in PI; references are to pages. Note: Severin Schroeder uses, e.g., the notation “PI IIxi”, indicating that the passage referred to is in chapter xi of PI II. Public and Private Occasions, ed. by James C. Klagge and Alfred Nord-

x

mann, Lanham, Boulder etc.: Rowman & Littlefield 2003. Philosophical Occasions, ed. by J. Klagge and A. Nordmann, Indianapolis: Hackett 1993. PR Philosophical Remarks, ed. by R. Rhees, Oxford: Blackwell 1975. RFM Remarks on the Foundations of Mathematics, ed. by G. H. von Wright, R. Rhees and G. E. M. Anscombe, revised edition Oxford: Blackwell 1978; references are to parts – indicated by capital roman numerals – and sections. RPP I Remarks on the Philosophy of Psychology, Vol. I, ed. by G. E. M. Anscombe and G. H. von Wright, Oxford: Blackwell 1980. RPP II Remarks on the Philosophy of Psychology, Vol. II, ed. by G. H. von Wright and H. Nyman, Oxford: Blackwell 1980. RUL Letters to Russell, as printed in Briefe, ed. by B. F. McGuinness and G. H. von Wright, correspondence with B. Russell, G. E. Moore, J. M. Keynes, F. P. Ramsey, W. Eccles, P. Engelmann, and L. von Ficker, Frankfurt am Main: Suhrkamp 1980; references are to pages and/or to dates as specifically as possible. SF Spätfassung or final version of the Philosophical Investigations, TS 227, as printed in KGE. TLP Tractatus Logico-Philosophicus, references are to numbered sections. The authors use two different translations: (1) tr. C. K. Ogden and F. P. Ramsey, London: Routledge 1990, first published 1922, or (2) tr. D. F. Pears and B. F. McGuinness, London: Routledge & Kegan Paul 1961. TS Typescript from the Nachlass, available on CD-ROM as Wittgenstein’s Nachlass: The Bergen Electronic Edition, Oxford: Oxford University Press 1998ff. References are to typescript-numbers of von Wright’s catalogue (in: G. H. von Wright, Wittgenstein, Oxford: Blackwell 1982, 35–62), followed by page number. TS 213 is referred to by means of “BT”. UF Urfassung or earliest version of the Philosophical Investigations, MS 142, as printed in KGE. VWVC Ludwig Wittgenstein and Friedrich Waismann: The Voices of Wittgenstein. The Vienna Circle, transcribed and edited by Gordon Baker, original German Texts and English translations, London: Routledge 2003. WVC Ludwig Wittgenstein and the Vienna Circle, shorthand notes recorded by F. Waismann, ed. by B. F. McGuinness, Oxford: Blackwell 1979. Z Zettel, ed. by G. E. M. Anscombe and G. H. von Wright, Oxford: Blackwell 1967. ZF Zwischenfassung or intermediate version of the Philosophical Investigations, TS 227, as printed in KGE. PO

xi

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), 1–21.

FAMILY RESEMBLANCE Hans SLUGA University of California, Berkeley Summary

Wittgenstein’s remarks about family resemblance in the Philosophical Investigations should not be construed as implying a comprehensive theory of universals. They possess, rather, a defensive function in his exposition. The remarks allow one, nevertheless, to draw certain general conclusions about how Wittgenstein thought about concepts. Reflection on the notion of family resemblance reveals that kinship and similarity considerations intersect in it in a problematic fashion.

“It is folly for you to say that you can discern the habits of daughters from those of their parents …: for I wouldn’t know from where either fathers or you could learn the secrets of mothers; and even if you knew them, daughters are often quite dissimilar to their parents.” Boccaccio, The Decameron

1.

Instead of producing something common to all that we call language, I am saying that these phenomena have no one thing in common which makes us use the same word for all, - but that they are related to one another in many different ways. And it is because of this relationship, or these relationships, that we call them all ‘language’[…] And the result of this examination is: we see a complex network of similarities overlapping and criss-crossing: sometimes overall similarities, sometimes similarities of detail. I can think of no better expression to characterize these similarities than ‘family resemblances’; for the various resemblances between members of a family: build, features, colour of eyes, gait, temperament, etc. etc. overlap and criss-cross in the same way. (PI §§ 65–67)

Wittgenstein’s words are familiar. They are, indeed, too familiar to induce much thought any longer. We know what Wittgenstein means to say in

this passage. Isn’t he making a general point about the nature of concepts or (as we might also and better put it) the use of terms — an anti-Platonic point that some have hailed as a breakthrough and others have utterly rejected from their essentialist perch? Renford Bambrough once wrote in a well known essay that in these lines “Wittgenstein solved what is known as ‘the problem of universals’” (Bambrough 1961, 186). According to Bambrough, Wittgenstein’s words “can be paraphrased into a doctrine which can be set out in general terms and can be related to the traditional theories, and which can then be shown to deserve to supersede the traditional theories” (Bambrough 1961, 192). We may or may not agree with his final assessment but there is a broad consensus that Bambrough was, indeed, right about Wittgenstein’s ambition. While Bambrough did not assert straightforwardly that Wittgenstein produced a theory in his remarks on family resemblance, he assumed that one could extract some such thing from them, a “doctrine” that competes with the traditional theory of universals and that, in fact, supersedes it. But can this be right? It is certainly far from evident that Wittgenstein’s words were intended to present or even sketch a theory. Did he not repeatedly object to the possibility of philosophical theories? Did he not tell us in both the Tractatus and the Philosophical Investigations that there are no philosophical theories and no philosophical propositions? We could, of course, follow those interpreters who dismiss such claims as hyperbolic. But even then it is not obvious that the quoted words express a theory. Even if we take Wittgenstein to have been in fact (though, perhaps, unintentionally) engaged in philosophical theorizing, we cannot conclude that each and every one of his words is part of a theory and we certainly cannot conclude that he intended to advance a substitute for the traditional theory of universals. We should, in any case, not presume from the start that Wittgenstein’s remarks about family resemblance concern all general terms. He seems to be speaking, instead, of a distinctive class of terms which includes “language,” “proposition,” “game,” “family,” (in the human sense) and “number.” He appears to be saying only that these terms characterize family resemblances, not that all terms do. And he does so (at least in the first instance) not in order to justify a general theory but in order to bolster the argument of the initial sections of the Philosophical Investigations. Those sections had considered questions of language and meaning and for this purpose he had constructed a number of illustrative language-games. But had he not taken the easy way out in his argumentation, Wittgenstein’s

2

interlocutor objects at this moment: “You talk about all sorts of languagegames, but have nowhere said what the essence of a language-game, and hence of language is” (PI § 65). This apparently willful omission, the interlocutor notes, appears to be in stark contrast to Wittgenstein’s own earlier concern with “the general form of the proposition” and thus with the essence of language. The considerations of the first sixty-four sections of the Philosophical Investigations may therefore be incomplete and perhaps also inconclusive. Wittgenstein is forced to admit the accusation that he has not determined the essence of language and what the general form of the proposition looks like and that he has not told us how the concept of language is to be formally defined. But he now seeks to explain this apparent omission and defend himself against his objector. There is no such thing as the general form of the proposition, he insists, there is no fixed essence to language, and thus the concept of language-game cannot be formally defined, and all this is so because “proposition,” “language,” and “language-game” are, in fact, family resemblance terms. Wittgenstein’s remarks on those terms must thus be seen as defensive in character and as strategically directed against the attempt to question and undermine what he has said before in his Philosophical Investigations. This does not prove that one could not derive a theory from his words — even if he himself did not intend to do so. But a theory of what? Of universals in general or, better, of general terms in toto? Or a theory of distinctive terms like “language”, “game,” “family”, and “number”? If we are to speak of a theory at all, I believe it can only be one of this latter, more limited kind. The terms that Wittgenstein singles out in the discussion of family resemblance seem to be, indeed, of a very special sort and sharply distinct from various other kinds of terms. It will help to distinguish them from at least two other classes of terms. There are, first of all, those terms that we use to pick out directly observable characteristics of things and the things that have them. We call something “red,” or “round,” or “hot” because we are or can be directly aware of its color, its shape, and its temperature. We can call such terms “basic terms” for present purposes as long as we attach not too much philosophical weight to the word “basic.” We can say, in any case, with some confidence that basic terms have no formal definitions and are instead explained by ostension. “This,” we say, pointing to an appropriate sample, “is red (round, or hot).” A second class consists of terms that we apply to things with various distinguishing marks. Thus, we call a “bachelor” someone who has the

3

distinguishing marks of being both a man and unmarried. We call a gourmet “a person who likes and is an excellent judge of fine foods and drink,” as Webster’s Dictionary puts it, and a gourmand “a person with a hearty liking for good food and drink and a tendency to indulge in them to excess.” Such terms have all a composite structure and they are the darlings of traditional logic which holds that every definition of a term is achieved through a conjunction of distinguishing marks. Epistemically, we can speak of such terms as being of higher determination. In order to determine that someone is, indeed, a bachelor we must first determine that this someone is both a man and unmarried. In order to determine that a composite term T applies to an object we must first determine that certain distinguished marks A, B, C, … apply to it. It is evident from this that not every term can be composite. Otherwise we would have to think of the marks A, B, C, … as again being terms of higher determination, and if this were true in every case, an infinite regress would follow. There are, then at least two kinds of terms: basic and indefinable terms and composite and definable ones. Wittgenstein gets us to understand that this easy division is insufficient, that there are still other kinds of terms, and that family resemblance terms, in particular, are neither basic nor composite. They surely do not pick out a single observable characteristic nor do they refer us to a conjunction of distinguishing marks. The members of a particular family are, instead, linked to each other by an array of interconnected and criss-crossing marks. Among those are figure, eye color, temperament, gait or, more precisely, having a particular figure, a particular eye color, a particular temperament, a particular gait, and so on. These distinguishing marks are typically such that one can have them to a degree. Two people can be more or less similar in figure, eye color, temperament, or gait. Someone’s eyes, for instance, can be more or less blue. This in contrast, for instance, to the characteristic of being a citizen of some particular country. No one can be more or less of a citizen. Similarly, no number can be more or less even. 2. Such observations have intrinsic interest with regard to the way we use the concept of family and other such notions. But Wittgenstein, as I have said, has a very specific purpose in mind in bringing such concepts up. He is defending himself against the accusation that his considerations about language and meaning in the first sixty-four sections of the Philosophical Investigations lack proper weight since he has failed to define the

4

concept of language. In order to defuse this criticism he needs to show not only that (a) “language” is a family-resemblance term but also and more significantly that (b) family resemblance terms are not capable of formal definition. If there is any doctrine to be found in Wittgenstein’s remarks, it must consist in these two assertions. He would certainly have no particular reason to argue at this point in his exposition that (c) all terms are family resemblance terms. On the contrary, what he says conflicts directly with that possibility. Proposition (b) implies, in fact, the negation of (c) if we assume, at least, that some terms are capable of being defined, and I find no place where Wittgenstein denies that. Bambrough must therefore be wrong when he takes Wittgenstein to be gesturing at a general theory of universals. There is still another reason for denying that every term is of the family resemblance variety and hence for denying Bambrough’s interpretation. We can speak of family resemblance only when things are characterized by overlapping and criss-crossing marks. But these marks cannot in turn always be of the family resemblances variety. Otherwise we would land once more in the infinite regress to which the assumption that everything is a composite term gives rise. Family resemblance terms are, in other words, similar to composite ones in that both are of higher determination. It will be understood that not every term can be of that sort. There remains the question how to establish that family resemblance terms are not formally definable. And why should we think of “language” as a family resemblance term? I will concern myself first with the more general proposition because Wittgenstein’s proposed answer to his objector will certainly fail if it proves without justification. What then is to be said for the indefinability of family resemblance terms? The argument cannot be that they are not composite terms. We know surely since Frege that not all definition is by conjunction. Consider, for instance, the term “prime number.” We can easily define it but not through a conjunction of its distinguishing marks. Instead we say: n is a prime number if and only if there is no number x different from 1 and n such that n is divisible by x. We can see thus that our initial division of terms into three large classes: basic, composite, and family resemblance is unsatisfactory. We have evidence that Wittgenstein must have known this. He must have understood that not all definition has to be through conjunction of marks. Thus, he

5

considers the possibility that we might define family resemblance terms through a disjunction of their marks but dismisses it in the specific case family resemblance terms as “only playing with words” (PI § 67). But why is that so? There is surely nothing wrong with disjunctive definitions as such. Webster’s Dictionary defines the term “corner” (in one of its senses) as “the point or place where lines or surfaces join and form an angle,” and there is surely nothing wrong with that characterization. We may get clearer on this when we look at what Wittgenstein says about the term “game” rather than that of “family.” He writes that we can explain that term by describing various games and by adding that “this and similar things are called ‘games’” (PI § 69). And this he appears to argue is the only way to explain it because “the concept ‘game’ is a concept with blurred edges” (PI § 71). Now it is inviting to think that a concept with blurred edges cannot be sharply defined. But we must surely allow for the possibility of defining such a concept through other concepts with blurred edges. We can, for instance, define the term “game plan” as “a strategy planned before a game” and thus use the family resemblance term “game” — which according to Wittgenstein has blurred edges — to define what is presumably another family resemblance term with blurred edges. If the term “game” is not definable that must then be so for some other reason. That reason is, in fact, provided in Wittgenstein’s further observation that games display “similarities, relationships, and a whole series of them at that” and that these “similarities crop up and disappear” as we consider more and more games (PI § 66). It follows from this that the set of “overlapping and criss-crossing” marks by which we determine what a game is not fixed in advance, that new marks will crop up and will become relevant as we consider new games and lose sight of old ones. Let us call terms of this kind open-ended. Wittgenstein’s ultimate reasoning may then be that open-ended terms have no formal definition and that family resemblance terms are typically open-ended. 3. This does not, however, completely resolve the question whether Wittgenstein has a comprehensive view of general terms or concepts. There may be, after all, more to Bambrough’s view than we have so far allowed for. It may not be that Wittgenstein takes all terms of the family resemblance kind, but this does not exclude the possibility that behind his remarks on family resemblance lies a comprehensive doctrine of general terms. It may even be that his remarks on family resemblance lead us (and are perhaps

6

even meant to lead us) to such a doctrine. In order to see this more clearly we must turn from the Philosophical Investigations to Wittgenstein’s Blue Book which anticipates the later remarks on family resemblance, just as it anticipates so much else in Wittgenstein’s later thought. It is not for nothing, it seems, that the Blue and Brown Books are subtitled “preliminary studies for the ‘Philosophical Investigations.’” It may be useful to recall this characterization now when we are trying to understand the implications of Wittgenstein’s remarks about family resemblance terms in the Investigations.1 The Blue Book’s comments on family resemblance (or, rather, “family likeness”) are triggered by Wittgenstein’s provocative claim that thinking is really an operating with signs (BB 6). This leads him to write somewhat later: “If we say that thinking is essentially operating with signs, the first question you might ask is: “What are signs?’” (BB 16). His response to that question foreshadows his subsequent refusal in the Philosophical Investigations to provide a formal definition of language. In The Blue Book Wittgenstein proposes instead of “any kind of general answer” to look at “particular cases” in which we operate with signs (BB 16). We note that just as in the Philosophical Investigations, his motivations are defensive in character; but it also appears that the Blue Book remarks hint at more. They suggest that the defensive argument is based on a general view of the nature of concepts. It is certainly true that in a fashion similar to the later work, The Blue Book rejects the idea that there must be “something in common to all the entities which we commonly subsume under a general term” (BB 17). But Wittgenstein goes on to say that “the idea of a general concept being a common property of its particular instances connects up with other primitive, too simple, ideas of the structure of language” (BB 17). And this sentence certainly gives us grounds for pause. While Wittgenstein is rejecting one particular picture of “the structure of language,” he seems to be willing to grant that there is such a structure and this allows for the possibility of some kind of comprehensive view of the nature of language and of concepts. While that may be so, it is, of course, by no means immediately clear what that comprehensive conception would look like. We can be sure only that it will have both a negative and a positive component. On the negative side Wittgenstein asserts that we suffer from 1. The assumption of such a continuity must, however, be handled with caution. Rush Rhees’s characterization of The Blue and Brown Books as preliminary studies for the Philosophical Investigations obscures that these two works belong to a distinct middle period of Wittgenstein’s thought.

7

a pervasive craving for generality which is “the resultant of a number of tendencies” and that these are, in turn, linked to “particular philosophical confusions” (BB 17). The positive content of his view appears to be, on the other hand, that the meaning of general terms can (and can only) be clarified by looking at “simple examples,” “primitive forms of language,” i.e. simple “language-games” where activities and reactions” are “clear-cut and transparent” (BB 17). Wittgenstein brings these two components together at the end of this section of The Blue Book when he writes: “The idea that in order to get clear about the meaning of a general term one had to find the common element in all its applications has shackled philosophical investigation; for it has not only led to no result, but also made the philosopher dismiss as irrelevant the concrete cases, which alone could have helped him to understand the usage of the general term” (BB 19–20). With its all-inclusive talk of “the usage of the general term” this statement certainly sounds more constructive (if only in a programmatic fashion) than anything we find in the corresponding sections of the Philosophical Investigations. In order to see this constructive conception more clearly we must turn to what Wittgenstein writes about the tendencies that encourage or cause our craving for generality. He lists four classes (a) to (d) on pages pages 17 to 18 of The Blue Book. The latter two need not concern us here since they are linked to specific considerations about the mind-body distinction and about the methods of science. The first two, which are closely related, appear to be, however, of immediate relevance to the question of a comprehensive view of the nature of general terms. Wittgenstein lists under (a) the already cited tendency to look for something in common to entities subsumed under a general term and illustrates this tendency in two ways: (i) by our belief that there is something in common to all games and (ii) by the “primitive” idea that properties are ingredients of the things which have them. His discussion of games initiates, in turn, consideration of the matter of family resemblance (or, rather, as Wittgenstein puts it in his own English, “family likeness”). He writes that games form a family “the members of which have family likeness. Some of them have the same nose, others the same eyebrows and others again the same way of walking; and these likenesses overlap” (BB 17). Just as later on in the Philosophical Investigations, Wittgenstein refrains assiduously from asserting in this context that all terms are family resemblance terms. The suggestion is, indeed, to the contrary, that family resemblance terms constitute a specific subclass. His remarks can, once again, not be taken to lend support to

8

the idea that Wittgenstein seeks to construct a comprehensive theory of general terms. With respect to that issue one must turn to what he says under (b) and subject this passage to detailed scrutiny. Wittgenstein writes under (b): “There is a tendency rooted in our usual forms of expression, to think that the man who has learnt to understand a general term, say, the term ‘leaf ’, has thereby come to possess a kind of general picture of a leaf, as opposed to pictures of particular leaves” (BB 17–18). And this belief is, in turn, “connected to the idea that the meaning of a word is an image, or a thing correlated to the word” (BB 18). The man who learns to understand the term “leaf ” was shown different leaves and the belief is that “showing him the particular leaves was only a means to the end of producing ‘in him’ an idea which we imagined to be some kind of general image. We say that he sees what is common to all these leaves” (BB 18). It is to these remarks that we must attend, if we are to progress in the question whether Wittgenstein’s words imply a comprehensive account of general terms. This will become clearer only when we realize that Wittgenstein has borrowed his example from Nietzsche’s essay “On Truth and Lie in a NonMoral Sense.” Nietzsche writes in the relevant passage: “Every concept arises from the equation of unequal things. Just as it is certain that one leaf is never totally the same as another, so it is certain that the concept ‘leaf ’ is formed by arbitrarily discarding these individual differences and by forgetting the distinguishing aspects. This awakens the idea that, in addition to the leaves, there exists in nature the ‘leaf ’: the original model according to which all the leaves were perhaps wove, sketched, measured, colored, curled and painted” (Nietzsche 1873, 83). It should be evident that Nietzsche, at least, is proposing a comprehensive account of concepts. But does this show that the same must be true of Wittgenstein? It is certainly noteworthy that Wittgenstein uses the same example as Nietzsche and does so in order to make the same (anti-Platonic) point. But this by itself does not mean that there exists a broader agreement between the two. Nietzsche is evidently saying something about every concept. Surely, we cannot reason that Wittgenstein of The Blue Book must be doing the same since he borrowed one or two other things from Nietzsche. But this is not meant to be the structure of the proposed argument. It is rather that attention to the actually existing link between Nietzsche and Wittgenstein may reveal other affinities in their thought that would otherwise escape us. The case is like this: When I am friends with Loren and then come to know his brother Laurence, my knowledge of Loren’s

9

person and character may help me to see things in Laurence which would not be visible to me, if I did not know them as brothers. My knowledge of their kinship gives rise to my recognition of their resemblance just as, in other cases, the recognition of resemblance can generate knowledge of a kinship. Or, to put it more generally, the recognition of a causal link of some sort or other between an A and a B can serve as means for bringing out similarities between them just as, reversely, the recognition of similarities between an A and a B can serve as a clue to their causal relations. I will return to the substantive significance this point, which serves here only to justify a particular course of reasoning. That there might be a significant link between Wittgenstein and Nietzsche should not surprise us given Wittgenstein’s substantial debt to Schopenhauer. We know now, too, that Wittgenstein bought and took copious notes on Nietzsche’s Antichrist during the First World War (McGuinness 1988, 225f.; Monk 1990, 121–123). Of greater interest still in the present context is the likelihood that he took the term “family resemblance” from Beyond Good and Evil where Nietzsche speaks of “the peculiar family resemblance of all Indian, Greek, and German philosophizing” (Glock 1996, 120).2 Nietzsche declares this resemblance to be due to the kinship of the three languages and he diagnoses from this “a common philosophy of grammar” which “through the unconscious domination and guidance of the same grammatical operations has prepared everything from the start for the same kind of evolution and sequence of philosophical systems.” It is thus not only the term “family resemblance” that unites the passage to Wittgenstein’s thought in The Blue Book but also the idea that the syntax of our language suggests to us metaphysical pictures and that the task of philosophy is to deconstruct the resulting metaphysical systems through a critique of language. Metaphysical questions arise, as Wittgenstein in The Blue Book suggests, from “an unclarity about the grammar of words” (BB 35), and the task of philosophy is to relieve the puzzlement and the mental discomforts that result. It is probably then not too much to say that Nietzsche’s and Wittgenstein’s philosophical concerns mirror each other here in certain important respects — though, of course, not in every detail. 2. Glock’s entry “Family Resemblance” in his Wittgenstein Dictionary focuses on Wittgenstein’s uses of the term and does not elaborate on the significance of the possible link between Nietzsche and Wittgenstein. It should be noted that the word Familienähnlichkeit was not coined by Nietzsche but was already in literary use in the early nineteenth century. Grimm’s Wörterbuch cites its occurrence in Jean Paul and Tieck and suggests its derivation from the Latin “gentilis similitudo”.

10

There is surely, for instance, a difference in tone between Nietzsche who often writes in an excitedly hyperbolic fashion and Wittgenstein who cherishes a more sober and cautious language. Nietzsche, moreover, does not shy from ambitious and large-scale formulations whereas Wittgenstein adopts throughout a deflationary strategy towards all philosophical claims and doctrines. What the Wittgenstein of The Blue Book shares, however, with Nietzsche is the conviction that human thought proceeds throughout by analogies (as Wittgenstein says) or metaphors (as Nietzsche puts it) and that these are built into our language and are capable of generating the profound philosophical illusions. Thus, Nietzsche writes in his essay “On Truth and Lie …:” “We believe that we know something about the things themselves […] and yet we possess nothing but metaphors for things” (Nietzsche 1873, 82f.). Truth is for him, indeed, “a movable host of metaphors” (Nietzsche 1873, 84). Truths, he concludes, are in consequences illusions or even lies (in a non-moral sense) which may prove, however, to be of distinctive use to us. For Nietzsche “everything which distinguishes man from the animals depends upon this ability to refine (verflüchtigen) perceptual metaphors in a schema, and thus to dissolve an image into a concept” (ibid.). These schemata allow, in turn, “the construction of a pyramidal order […] — a new world, one which now confronts that other vivid world of first impressions as more solid, more universal, better known, and more human than the immediately perceived world” (ibid.). By recognizing the existence of an actual link between Nietzsche and Wittgenstein we can come to understand the crucial role that the concept of analogy plays in Wittgenstein’s Blue Book. (It is certain, in any case, that we didn’t see it until we recognized the actual connection between the two.) In this way we can come to see, in particular, that Wittgenstein conceives of concepts, similarly to Nietzsche, as built on analogies. This is made most apparent in what he writes about the concept of a person. We read in The Blue Book: For the ordinary use of the word ‘person’ is what one might call a composite use suitable under the ordinary circumstances. If I assume, as I do, that these circumstances are changed, the application of the term ‘person’ or ‘personality’ has thereby changed; and if I wish to preserve this term and give it a use analogous to its former use, I am at liberty to choose between many uses, that is, between many different kinds of analogy. (BB 62)

The remark gives two indications that Wittgenstein thinks of concepts as grounded in analogy. When he speaks of the ordinary use of the term “per-

11

son” as composite, he means presumably not that the concept is defined through a conjunction of distinguishing marks but that we apply it to a series of distinct yet similar phenomena. His statement rejects, in other words, the idea that there might be one single thing that all persons share. Wittgenstein notes, second, that in forming new concepts, such as a new concept of “person,” we extend and modify existing concepts analogically. Like Nietzsche, he believes, moreover, that proceeding in this way analogically is of distinctive value to us. He writes: “The use of expressions constructed on analogical patterns […] may be extremely useful. […] Every particular notation stresses some particular point of view” (BB 28). But he notes also that “where words in ordinary language have prima facie analogous grammars we are inclined to interpret them analogously” (BB 7) and that this can be the source of philosophical confusion. We must therefore always ask: “How far does the analogy between these uses go?” (BB 23; Wittgenstein italicizes this question). Philosophical problems are made difficult because of “the fascination which the analogy between two similar structures in our language can exert on us” (BB 26). Augustine’s problems with the reality of time, metaphysical quarrels between idealism and realism, and confusions about the relations between body and mind are all due to our fascination with analogical structures. The method of philosophy must therefore be “to counteract the misleading effects of certain analogies” (BB 28). This is not easy since, as already noted, analogy is essential to language and there is, moreover, “no sharp boundary […] round the cases in which we should say that a man was misled by an analogy” (BB 28). We must therefore study and assess the use of analogies case by case. In this enterprise we may sometimes find it helpful to construct new notations. In order to break the spell of old analogies we introduce “ideal languages,” i.e. symbolic notations — not for the purpose of replacing ordinary language, which on the whole is perfectly all right as it is, but to test the analogies built into our ordinary ways of speaking. “Thus we sometimes wish for a notation which stresses a difference more strongly, makes it more obvious, than ordinary language does, or one which in a particular case uses more closely similar forms of expression than our ordinary language. Our mental cramp is loosened when we are shown the notations which fulfill these needs. These needs can be of the greatest variety” (BB 59). Wittgenstein’s remarks must suffice here as indications of the significant affinities between Nietzsche and his thought in The Blue Book. These affinities will eventually fade away, and in the Philosophical Investigations

12

the notion of analogy will play no longer a major role. But its occurrence in The Blue Book entitles us to conclude that there Wittgenstein largely agrees with Nietzsche concerning the nature of concepts. There is no doubt that in the essay “On Truth and Lie …” Nietzsche advances a general theory of concepts: it holds that concepts are grounded on our perception of similarities, that every concept embodies a metaphor and that in our language these metaphors get schematized and regularized. Wittgenstein’s picture in The Blue Book seems similarly broad and appears to include a similarly general understanding of the nature of concepts. Like Nietzsche he sees concepts as grounded in the perception of similarities; he appears to hold that every concept embodies an analogy and that in our notations these analogies get once again schematized and regularized. That Nietzsche and Wittgenstein should be here in agreement may, in turn, be due to their shared debt to Schopenhauer. It was Schopenhauer, after all, who had argued before them that concepts are grounded in recognitions of similarity, that they express metaphors or analogies. In his Parerga and Paralipomena he had written: “Similes are of great value in so far as they refer an unknown relation to a known […]. Even the formation of concepts rests at bottom on similes in so far as it results from our taking up what is similar in things and discarding what is dissimilar” (Schopenhauer 1851, 550). We seem to have moved far from a consideration of the notion of family resemblance, but we can summarize the outcome of our excursus in the following propositions: On The Blue Book view, concepts are based on recognitions of similarity, and general terms have an analogical structure. Family resemblance terms are, it seems, a special variety of general term; they are like composite terms of higher determination; but the distinguishing marks to which they refer us are to be considered as a network of overlapping and open-ended similarities, not in conjunctive terms. It appears then that Bambrough is partially right in thinking that Wittgenstein’s remarks on family resemblance are linked to a comprehensive view of the nature of concepts or, as we might also say, of the meaning of general terms. But the view was not that all terms are of the family resemblance kind. In The Blue Book Wittgenstein seems to have held, however, that consideration of family resemblance terms might alert us to the wider fact that all general terms are analogical structures. That view seems to have been limited to Wittgenstein’s Blue Book period and is not to be found (as far as I see) in his Philosophical Investigations. This observation may help us to appreciate more clearly the difference between the two texts. It is,

13

in fact, misleading to think of The Blue Book as a preliminary study for the Philosophical Investigations. The Blue Book represents, rather, a distinct period in Wittgenstein’s thought, one that we have come to identify as a middle period. One of the characteristics of this period is Wittgenstein’s willingness to gesture at a comprehensive conception of concepts or general terms — a conception that owes something to Nietzsche. 4. The notion of family resemblance is more problematic than Wittgenstein allows for and that is one reason why it cannot serve as grounding for a comprehensive theory of concepts or general terms. It is problematic in that its characterization draws on two quite different sets of ideas, two different vocabularies but treats them as if they were one and the same. The first is the vocabulary of kinship, of descent, of some sort of real and causal connection and of the links established by them, the second is that of similarity, resemblance, affinity, and correspondence. It appears that Wittgenstein fails to appreciate the genuine difference between these two ways of speaking and that his characterization of family resemblance combines both in a single formula. Thus, he writes, on the one hand, that the phenomena we call language have no one thing in common, but “are related (verwandt) to one another” and that it is “because of this relationship, or these relationships (because of this Verwandtschaft, oder diese Verwandtschaften)” that we call them all “language” (PI § 65). On the other hand, family resemblances are said to show us “a complicated network of similarities overlapping and criss-crossing: sometimes overall similarities, sometimes similarities of detail” (PI § 66). That the two vocabularies are different is hidden to some extent by the English translation of “verwandt” as “related” and of “Verwandtschaft” as relationship. But even the German words prove ambivalent: in its older uses “Verwandtschaft” may mean as little as the English word “similarity,” but in modern usage it also means more specifically a relationship built on blood. It is not easy to say of every single occurrence whether Wittgenstein takes the word in one or the other of these senses. But his exposition and his examples indicate that he is playing on its ambiguity. When he insists, for instance, that in games “we will not see something that is common to all, but similarities, relationships (Verwandtschaften), and a whole series of them at that” (PI § 66), the words “similarity” and “relationship” may simply repeat the same idea or he may be tacitly appealing to our intuition that games are activities that have grown out of each other. Is a human family really only

14

united by relations of similarity or is it not defined also through bonds of kinship? Do two sentences belong to one and the same language only because they are linked by a network of resemblances or because they are part of one and the same linguistic history? We must, so it seems, distinguish between two kinds of concepts which Wittgenstein lumps together under the heading of family resemblance. The first we may want to call kinship concepts, meaning not necessarily only biological kinship, but any kind of appropriate causal connection. The second variety we may for obvious reasons call similarity concepts. The two kinds of concepts are very different and while both may turn out to resist formal definition the reasons why they do so are at variance. I will try to explain some of this by constructing examples. Example 1: My mother is pregnant. I am supposed to be getting a little brother but due to a bout of diphtheria my mother gives birth to a terribly disfigured child. It doesn’t look like anyone in the family; the baby looks, indeed, hardly human and more like a monster. But there is no doubt: the child is my brother. I am related to him. We have the same father and mother, are part of the same bloodline. This is decisive in establishing that we are of one and the same family; similarity and dissimilarity has nothing to do with this. The only important factor is, in this case, that this child is my kin. That does not mean that membership in a family is determinately fixed in terms of biological kinship. The concept of family is, in fact, highly complex, as anthropologists, sociologists, and lawyers will tell us. Biological, cultural, and legal assumptions enter its definition. A husband and a wife are in our culture said to form a family even though there may be no prior biological bond between them. An adopted child belongs to its adoptive family even though there may be no biological relations to others members of that family. In some cultures an abandoned child or an abandoned wife are no longer counted as part of the family. While none of this has to do with biological kinship it is also the case that none of it has to do with similarity or dissimilarity. The question of “build, features, colour of eye, gait, temperament, etc. etc.” does not even arise. It is, of course, true that those who are biologically related may display some such similarities, but this is a consequence of their kinship and may thus also be said to be a consequence of their being part of the same family. It is equally true that we can sometimes determine membership in a family by discovering similarities between family members (the long lost son is recognized by a distinguishing family feature), but these similarities do

15

not constitute that membership, they only provide evidence for it. The conclusion I draw is then that “family” in the human sense is a kinship and not a similarity term. Example 2: We are traveling to the Moon and we meet there the wellknown Moon people. Surprisingly, they look like Germans; they talk and act exactly like Germans. They speak of themselves as the nation of thinkers and poets while watching their soccer and drinking their beer. Among them is even a Sluga family and the Moon Slugas look amazingly like my relatives back home. Some of them might even be my brothers. But they are not. The Moon Slugas are not a newly discovered branch of my family. We are not bound to invite them to our next family feast. Their claims to legal rights in our inheritance disputes will be rejected by any Earth court. Similarity or dissimilarity has nothing to do with the matter. The Moon people lack the appropriate biological and legal relations to me and my own. The Moon people are not even Germans, however much they may look like them. The concept of nationality wavers between biological, cultural, and legal considerations just as the concept of family. There is a primitive view according to which those who belong to a nation must be of common biological descent, but the assumption is highly dubious. Who is to determine that Otto von Bismarck and I share a common biological descent (the idea strikes me right now as unlikely) and yet we are both indubitably German. I know, in any case, that my student in Frankfurt is German as much as I am even though her family came a generation ago from Turkey. But I can’t recognize the Moon people as German, however much they may look similar to me and other Germans. They have, for instance, no right to immigrate to the Federal Republic; they have no right to vote in German elections. Things might be different, if the story of Jack and the bean stalk proved true after all and Jack had been accompanied to the moon by some Germans and perhaps even by members of the Sluga family. My meeting with the Moon Slugas might then turn into a family reunion and our encounter with the Moon people the discovery of an exotic colony of exile Germans. We could even imagine a bizarre act of the German parliament that declares all people we will encounter in space to be honorary Germans. But in no case will similarities alone establish that Moon people are Germans and Moon Slugas part of my family. This re-enforces the earlier conclusion that “family” is a kinship term and adds to this the observation that terms referring to nationalities are likewise.

16

Example 3: This time we are flying to Mars. There we find creatures that look in no way like Germans; they don’t even look human. But they are living in a highly developed culture. They have, for instance, organs with which they can trace complex movements in the air. As they do so, the air congeals for a moment in the elaborate pattern they draw. Slowly it becomes clear to us that this is a system of communication. Eventually we learn to translate their airy formations into a human idiom. No doubt, we conclude, the Martians do have a language. That means, of course, not that their language stands in some actual relation of descent to human language, that there are causal and historical links between their form of communication and ours. What convinces us that the Martians possess a language is rather that their peculiar gestures and the temporary forms in the air serve similar purposes as what we call languages. The concept of language is for us, in other words, a similarity concept. Anything sufficiently similar to what we call language is for us also a language. Here similarity is decisive and what we observe is presumable a network of similarities, a “family resemblance,” or, to put it more clearly, a loose array of resemblances that may not form a family in the more specific sense of the word. In contrast to the terms “family” and “nationality” the term “language” proves thus not to be a kinship but a similarity term. Example 4: We are flying once again into space. On this journey Venus is our goal. Very strange beings live on that planet. They consist of loose clouds of smoke but the strangest thing is that we can talk to them without effort. Perfect German, emanates, in fact, from those clouds. But how is that possible? The inhabitants of Venus are certainly not Germans either biologically or by law. “German,” we have decided, is a kinship and not a resemblance term. Hence, it would seem that “the German language” should be likewise. But we have also come to understand that “language” is a similarity term and this might incline us quite reasonably to conclude that “the German language” must be the same. Is it possible that our Venusians are, after all speaking German? Our question may, however, have no direct answer. Perhaps we can use the term “German language” in one way or the other or, what is even more likely, the term may refer us to both kinship and similarity relations. That raises, admittedly, all kinds of questions. There may, for instance, be terms in which the kinship element dominates and others where the similarity element is most prominent. Nietzsche seems to have had some such complex formation in mind when he spoke of the family resemblance of Indian, Greek, and German philosophizing in Beyond Good and Evil. He certainly meant to

17

indicate that these different traditions exhibit overlapping and criss-crossing similarities. But he also wanted to say that their affinity was due to a common descent. 5. Wittgenstein did not concern himself with such niceties — neither in The Blue Book nor in the Philosophical Investigations. In both works he thought it sufficient to have shown that there is a category of term that lacks formal definitions and that “language,” “sign” and other such terms belong among them. This returns us, however, once more to the question why we should consider such terms indefinable. Having distinguished kinship and similarity terms we can come to see that there may be two very different explanations for this state of affairs. In the case of kinship terms, it seems, we have no formal definition because we cannot say, in advance, who will in the future be considered a kin. Kinship relations are open-ended but are so specifically with respect to the future. Consider the concept of art as such a term. It is in the nature of art that its course be unpredictable, that we cannot say in advance what new works of art will look like, or what we will count as art at some future moment. Historical concepts have this kind of openness (think of “culture,” “religion,” “politics,” etc.) and they can have for that reason no formal definition. We can, of course, arbitrarily define them by limiting ourselves to a particular moment, or to how these terms have been applied in the past. But the peculiar way in which we use such terms to project a future cannot be captured in this restrictive manner. Nietzsche was therefore right when he declared at one point that only what is unhistorical can be defined. But there is, in addition, another form of open-endedness and it manifests itself in similarity terms. It is the kind of open-endedness to which Wittgenstein draws our attention when he says that in our survey of games “many common features drop out, and others appear” (PI § 66). What makes such terms indefinable is not that they refer us to a multitude of similarities and common features or that these have potentially blurred edges. We have seen already that concepts with blurred edges may well have formal definitions. The indefinability of the Wittgensteinian family resemblance terms is due rather to the fact that the range of relevant similarities is not fully determined. Such terms are, in other words, also open-ended but with respect to relevance rather than to the future. Either way, Wittgenstein is right in pointing out that terms like “language,” “sign,” “family,” game,” “number,” and so on have no formal definitions.

18

That this may be so for different reasons need not concern him in the course of the argument in which he is engaged in both the Philosophical Investigations and The Blue Book. It is enough for him to have shown that for whatever reasons the terms “language,” “sign,” and “game” are among these indefinable terms. But for other purposes the finer distinctions may turn out to be important. Again an example may help to explain this, but this time an example of a more realistic sort. Example 5: Analytic philosophy is today considered the dominant form of philosophy — at least in the English speaking world. But what do we mean by “analytic philosophy?” It appears that we can understand the term in two quite different ways. We can treat it as either a similarity or a kinship term. When we take it one way we will consider everything analytic philosophy that stands in some appropriate resemblance relation to what we are doing today. We will conclude then that even a Plato and Aristotle, a Leibniz, Kant, and Hume was (at least, at times) doing analytic philosophy in so far as their philosophical work is similar to that of contemporary analytic philosophers. If we take “analytic philosophy,” on the other hand, to be a kinship term we will think of it as an enterprise connected by actual, causal relations. We may then set thinkers like Frege, Russell, Moore, Wittgenstein, Carnap, and Quine and their contemporary successors apart because of the ways these philosophers relate each other personally and through their interactions, through their reading each others writings and through the fact that they responded to each other’s ideas in certain ways. On this view Plato and Aristotle and other earlier thinkers will not count as analytic philosophers because they are not part of the intricate real and causal web that holds “analytic philosophy” together. Both kinship and similarity concepts have a place in our thinking. But it may also be true that family concepts are particularly indispensable in historical accounts. In history we are, after all, not only concerned with mapping similarities of one sort or another; we are also trying to establish direct and real connections, causal links, dependencies and “influences.” Similarity terms prove insufficient for this kind of undertaking. Thus kinship terms are essential when we are telling the history of philosophy but not when we are comparing types of philosophical ideas. Kinship terms are essential in the history of art but not in the examination of style. They play a role in literary history but not the same in comparative literature. The distinction between the two kinds of terms is, however, not an absolute one. Something that starts of as a kinship term may eventually

19

become in our use a similarity term, and the same thing will hold vice versa. Thinking in terms of relations of kinship, of descent, of developments, of family trees, etc. can satisfy certain intellectual needs. But once we have established connections and descriptions in this way we may want to shift attention to similarities and dissimilarities in the field of phenomena so described. This permits us to ignore historical contexts and causes, questions of descent, and family trees and may open up a new way of seeing things. It is plausible, for instance, to think of our concept of language as originally a kinship notion. The ancient Greeks certainly pretended that only they and their kin spoke a real language and that everything else was incoherent noise. But for us today “language” is, undoubtedly, a similarity term. Anything that looks like, sounds like, or functions like language is for us language. Hence, we can conceive even of the utterings of our Venusian friends as language. But similarity concepts may also over time become family concepts. When we speak today of a family tree of the Indo-European languages, when we postulate that these languages have developed out of each other, and that they refer us to a shared cultural (and perhaps even biological) heritage, then we do so because of the similarities that philologists have discovered between the various Indo-European tongues. The term “IndoEuropean language” has thus developed from being a similarity term to a kinship term and this development marks a substantial advance in our understanding. Our grasp of the phenomena becomes firmer when we think not only in terms of similarities but also in the richer and more suggestive vocabulary of a causal order. In the course of my exposition I have already referred to the illuminating possibilities of a shift from considerations of kinship to a consideration of similarities. The use I made of Nietzsche in order to illuminate Wittgenstein was of exactly this kind. I close with an example that illustrates that the distinction is relevant not only to terms of higher determination, that is, in particular family resemblance terms. Even what we have called basic terms can be understood either in the vocabulary of similarity or kinship. Consider that color terms are for us almost always similarity terms. Black is what looks like black. Blue looks similar to blue and red and to red. But the painter knows that some black is created from blue and some from red. And anyone who understands these possible derivations will begin to see black itself in a new way. He will recognize, for instance, relations and differences between a cold and a warm black. Once again (and just as in our discussion of Nietzsche and Wittgenstein) recognition of kinship can

20

lead to sharpened perception of similarities and dissimilarities. But the reverse is also true and holds also for basic terms. As the painter’s power of visual discrimination sharpens, he will learn to create and mix colors in a new and different ways.

REFERENCES Bambrough, Renford 1961: “Universals and Family Resemblance,” reprinted in George Pitcher (ed.), Wittgenstein, London: Macmillan 1968, 186–204. Glock, Hans-Johann 1996: A Wittgenstein Dictionary, Oxford: Blackwell. McGuinness, Brian 1988: Wittgenstein: A Life. Young Ludwig 1889–1921, London: Duckworth. Monk, Ray 1991: Ludwig Wittgenstein. The Duty of Genius, London: Vintage. Nietzsche, Friedrich 1873: “On Truth and Lie in a Nonmoral Sense,” in Philosophy and Truth. Selections from Nietzsche’s Notebooks of the Early 1870’s, translated and edited by David Breazeale, New Jersey: Humanities Press International 1979. Schopenhauer, Arthur 1851: Parerga and Paralipomena, vol. 2, translated by E. F. J. Payne, Oxford: Clarendon Press 1974.

21

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), 23–38.

ONLY CONNECT … 1 PERSPICUOUS REPRESENTATION AND THE LOGIC OF NACHTRÄGLICHKEIT

Klaus PUHL University of Innsbruck Summary In what follows, I will concentrate on the type of temporality which structures Wittgenstein’s method of a perspicuous representation, or of a synoptic overview (übersichtliche Darstellung). I will argue that the temporal order which applies to (giving) a perspicuous representation is best to be described as retroactivity, deferred action or afterwardness (Nachträglichkeit), a concept which calls into question the ordinary conception of time as a linear and irreversible process as well as of a clear break between present and past. First, I will turn to Sigmund Freud and the way he developed the concept of Nachträglichkeit or retroactivity in some of his case-studies. The basic idea here is that the meaning or content of a past experience is constituted by repeating it in a different or disguised way. I will then argue that the relation between a perspicuous representation and what is made perspicuous is best interpreted as a relation of Nachträglichkeit. This interpretation also accounts for the involvement of the subject in the object made perspicuous — as it is stressed by Wittgenstein and illustrated by his discursive method.

1. Nachträglichkeit The concept of retroactivity (Nachträglichkeit), as it is employed in this paper, was first articulated by Sigmund Freud in some of his case-studies, for instance in his analysis of the psychotic Emma and in his analysis of the Wolfman. Freud applies this concept to account for the non-linear and reversible (i.e. circular) time-structure involved in psychical causality. He implicitly questions the view widely held among his followers and critics that, during psychoanalytic treatment, the repressed memory 1. “Only connect …” is the epigraph of E. M. Forster’s novel Howard’s End (1910).

of a traumatic childhood-event is recovered from repression and then reunited with the affect that had become separated and transformed into a bodily symptom. In contrast to this rather simple account, Freud suggests that, by reaching puberty and triggered by new experiences, earlier memories, experiences etc. are reworked or “retranscribed” (Freud) and thus turned into traumatic events. Having been reworked means that the earlier experiences get their sexual meaning and hence their psychic efficacy retroactively. A new experience, such as a dream, leads the subject to unconsciousnessly repeating, but at the same time also reinterpreting an old experience in such a manner that the old experience gets a delayed psychical efficacy, one it did not have at the earlier time. Thus, the trauma in particular, the decisive event in neurosis, never occurred as such but is constructed afterwards. 1.1 To begin with, the focus will be put on Freud’s case of Emma Eckstein whose phobic condition forbids her to enter shops by herself. Freud discovers two events in Emma’s life. The first event is remembered by Emma herself: At the age of twelve, she enters a shop where she comes across two shop assistants, one of whom she remembers. Emma sees them laughing and flees the shop in a panic attack (Schreckaffekt). This event marks the beginning of her phobic neuroses. Freud traces this scene — which is, as above mentioned, consciously remembered by Emma — to an earlier traumatic event which is repressed by her: When Emma was eight years old, a shopkeeper touched Emma sexually. At that time the incident did not have a sexual content for Emma, as she was too young to make sense of what was going on; hence she forgets all about it. The later event — the attendants’ laughing — did not have any sexual content. What had happened? According to Freud, the laughing of the shop assistants triggered the unconscious memory of the earlier scene because the laughing reminded Emma (unconsciously) of the grimace of the shopkeeper, when he was touching her through her clothes, causing a sexual stimulus, and the fear that the shop-assistants will repeat the earlier assault which is now understood sexually. As it is unconscious, the memory of the first scene is not experienced as a memory but relived as new experience, and hence repeated, not remembered. The two scenes are linked by the repetition of two elements, the laughing and the clothes, combined with a sexual feeling which Freud calls “sexual release”. The later scene repeats unconsciously the earlier one, but differently and in a distorted way. For Emma now suddenly understands, or rather feels, the sexual meaning of the first scene, although she is transferring this meaning (and “the sexual

24

release” she feels) to the second one. By being repeated 4 years later, albeit disguised, the earlier scene gets its sexual meaning and traumatic effect belated, nachträglich. The sexual meaning is then wrongly transferred to the later quite harmless scene. This makes the repetition pathological. Freud writes: The memory aroused what it was certainly not able to at the time, a sexual release [Entbindung], which was transformed into anxiety. With this anxiety, she was afraid that the shop assistants might repeat the assault, and she ran away. […] Here we have the case of a memory arousing an effect, which it did not arouse as an experience, because in the meantime the change in puberty had made possible a different understanding of what was remembered. Now this case is typical of repression in hysteria. We invariably find that a memory is repressed which has only become a trauma by deferred action. (Freud 1895, 354, 356)2 The traumata of childhood operate in a deferred fashion [wirken nachträglich] as though they were fresh experiences; but they do so unconsciously. (Freud 1896, 166 – 7; emphasis by Freud) It is not the experiences themselves which act traumatically but their revival as a memory after the subject has entered sexual maturity. (Freud 1896, 164)

Thus the linear time of cause and effect is reversed, as it is the occurrence of the later event that turns the earlier event into a traumatic cause of the symptom. To put it paradoxically, the later event causes the earlier to have a traumatic effect, or rather, the earlier event can only cause the symptom (Emma’s being afraid of entering shops by herself ) as its traumatic significance only becomes effective once the later event has happened. According to our ordinary conception of causation, however, cause and effect are asymmetrically related. A cause precedes its effect; they cannot happen at the same time. Cause and effect can be identified independently of one another. The priority of the cause to its effect consists in the capacity of the former to produce the occurrence of the later, not the other way round. Also, the explanation of why the effect happened must invoke the cause, while the occurrence of the effect only serves to deduce that the cause has 2. The original reads: “[…] daß eine Erinnerung verdrängt wird, die nur nachträglich zum Trauma geworden ist”. A better translation of “nachträglich” would be “after the event” or “retrospectively”, since “deferred” only suggests a temporal gap between cause and effect, while the whole point of “nachträglich” is that the delay turns the original event into a psychic cause by re-transcribing it.

25

occurred, not why. Causal time is objective, linear and irreversible. There is a clean break between past and present, the past cannot return. In contrast, Nachträglichkeit means that the earlier event can only be identified as having the causally relevant properties when the later event has occurred. The earlier event exists as a causally efficient one only in its later effects. In this sense the past is present in the present. The (present) effects repeat and change or distort (disguise) the meaning of the past event and constitute its psychic efficacy. Nachträglichkeit does not just mean that an event has a delayed effect, in the sense that it was already traumatic when it occurred — like the headache I have today is caused by having had too many drinks the night before. This alone would still be compatible with our ordinary linear time-conception. Neither does it just mean, as Freud makes clear against C. G. Jung, that the earlier event is only a figment of somebody’s imagination. Nachträglichkeit presupposes at least two events, and it is the latent phase — puberty in Freud’s analysis, extending between the first and the second event — that retroactively constitutes the meaning and efficacy of the first scene by turning the second one into a (distorted and unconscious) repetition of the first.3 1.2 The second of Freud’s case studies where Nachträglichkeit plays an important role is the famous Wolfman case (Freud 1917/18). At the age of 4, the Wolfman dreams about white wolves sitting on a tree outside his bedroom window and staring at him. This dream initiates his neuroses and his phobic fear of wolves. Freud reconstructs a primal scene in which the Wolfman watches, at the age of one and a half, his parents engaged in a coitus a tergo — an event which at that time does not have any sexual meaning for the child. Yet, it gets its sexual character retroactively by being repeated in the wolf-dream. This dream, happening at the age of four, repeats in a distorted way the primal scene of the coitus-observation, and it is only by being repeated in this way that the primal event gets its traumatic meaning and efficacy. The link between the two events are the wolves, which are associated to a fairy tale, told to the Wolfman shortly before the dream. This tale is about a wolf that had his tail ripped out by a tailor and which later asks 3. Jacques Lacan, who was the first among Freudians to stress the importance of Nachträglichkeit in Freud’s work (he translates “Nachträglichkeit” as “après coup”), uses the “future perfect” to describe the logic of traumatic efficacy and, in general, of the reworking of past impressions in the light of present experiences. The event which was not traumatic when it occurred will have been traumatic when it is repeated in the future. Grünbaum 1984 does not mention Nachträglichkeit in Freud, neither does Bouveresse 1995.

26

other wolves to get on its back in order to climb a tree where the tailor is hiding. This element links the dream to the primal scene: Being mounted from behind with one’s tail ripped out repeats his father mounting his tailless mother from behind. The dream triggers an unconscious memory of the primal scene which is now charged with a sexual meaning, made possible by the sexual development of the 4 year old. The dream manifests the unconscious desire of the little Wolfman to be taken from behind by his father, just like his father took the mother from behind during the original scene. In order to be taken by his father, however, one has to be castrated like the mother, hence his fear of his father, which is repressed and turned into a phobia about wolves. Freud considers the possibility that the primal scene, watched by the Wolfman, had not taken place the way it was reconstructed by psycho-analysis. He concedes that the primal scene could also have been constructed by the Wolfman. Freud insists, however, that there must have been some kind of experience in the meantime, e.g. the observation of copulating dogs. For only a scene of a coitus a tergo, be it between humans or animals, can have been repeated in the dream and can have the nachträgliche traumatic effect of a phobic fear of wolves. 1.3 Neither in the case of Emma nor of the Wolfman have the different scenes, considered in isolation, a traumatic content. As they happened, neither the sexual attack of the shopkeeper on Emma nor the Wolfman’s parents having sex were understood as being of a sexual nature. Furthermore, the later scenes in the clothing shop and the Wolfman’s dream had no sexual content at all. It is rather the connecting of the two experiences or events that turns the shopkeeper’s attack and the sight of the parents having sex into a traumatic cause. The earlier events exist as causally efficient ones only in their later effects. The theory of Nachträglichkeit stresses that psychic time is not linear, but circular, and that, in order to be meaningful and efficient, an experience cannot happen just once (once is nothing) but has to be repeated and constantly reworked and rearranged. The analytic situation is also the site of such repetitions. The chronologically later events are the first to appear during analysis, for instance within transference, announcing the traumatic events, which they are already repeating, as the events to be recovered. Although the later events happen in the analytic setting before the earlier ones, they count only as events in so far as they repeat the earlier ones. And the earlier events only count as (traumatic) events in so far as they are repeated. Hence, in order for an event or an experience to be meaningful and psychologically efficient, it has to be

27

connected to other events (either by repeating them or being repeated by them). At the same time, this reworking or repeating of the past through and in the present is a form of synchronizing past and present, a form of revealing the past in the present, and this brings us to Wittgenstein’s method of developing a synoptic overview or a perspicuous representation (übersichtliche Darstellung), i.e. of making events meaningful by arranging them on a synchronic level. What I want to show subsequently is that the basic idea of Nachträglichkeit — namely that the meaning or content of a past experience is constituted by repeating it in a different or disguised way — is contained in Wittgenstein’s method of coming up with perspicuous representations. Perspicuous representations are responding to puzzling events and situations by repeating them in a way that turns them into something meaningful. 2. Perspicuous Representations 2.1 In Philosophical Investigations §§ 122– 123, Wittgenstein writes: A perspicuous representation produces […] that understanding which consists in ‘seeing connections’. Hence the importance of finding and inventing intermediate cases. The concept of a perspicuous representation is of fundamental significance for us. It earmarks the form of account we give, the way we look at things. […] A philosophical problem has the form: “I don’t know my way about”.4

As almost every commentator has remarked: The concept of a perspicuous 4. PI § 122 is taken from Remarks on Frazer’s ‘Golden Bough’. Against Frazer’s alleged historical explanations, Wittgenstein stresses the importance of connecting things: “It is just as possible to see the data in their relation to one another and to embrace them in a general picture without putting it in the form of an hypothesis about temporal development” (PO 131). What Wittgenstein claims here links him with Forster’s epigraph used in the title of my paper. According to Oliver Stallybrass, the editor of the Penguin edition of Howard’s End, “only connect …” alludes to a recurrent theme in Forster’s work, namely the “possibility of reconciliation between certain pairs of opposites: the prose and the passion, the seen and the unseen, the practical mind and the intellectual, the outer life and the inner” (Stallybrass 1992, 10). S. P. Rosenbaum quotes Goronwy Rhees writing that “only connect …” “had more influence in shaping the emotional attitudes of the English governing class between the two world wars than any other single phrase in the English language“ (Rosenbaum 1995, 80).

28

representation is not itself perspicuous. However, the following features can be discerned from remarks by Wittgenstein: P1 What motivates constructing a perspicuous overview is marked by a (philosophical) problem of the form of “I don’t know my way about” (PI § 123), or by sensing “a vague mental uneasiness” or “intellectual discomfort” (PO 114). Its subject-matter can be some puzzling aspect of the use of certain words, mathematical proofs, magical rites, religious ceremonies, and dreams (as in Wittgenstein’s reading of Frazer’s The Golden Bough and Freud’s Interpretation of Dreams). P2 A perspicuous representation proceeds by finding intermediate cases (Zwischenglieder) so far unnoticed, or by inventing them. P3 However, contrary to empirical accounts, a perspicuous representation does not come up with new pieces of information, or with discoveries or hypotheses, but has the character of reminding somebody of something which has been there all the time. “The work of the philosopher consists in assembling reminders for a particular purpose” (PI § 127)5. A perspicuous representation is, just as Freud’s theory of Nachträglichkeit, a form of understanding and knowing more by arranging and connecting things already there. P4 Being perspicuous is not an intrinsic property of a representation, but rather dependent on being acknowledged by the person addressed by it — as something which makes things clear and satisfies her. As regards this respect, Wittgenstein compares his method to psychoanalysis (PO 164) and to aesthetic explanations (PO 108). P5 As to justification, perspicuous representations are on the same level as the puzzle cases they make perspicuous. They show other contrasting, but coherent ways of seeing things and thereby make one realize that one has always been dealing with forms of representing things (like traumata) and not the things themselves. P6 A perspicuous representation responses to a local problem. It does not abstract from the particular (contextual, historical, or psychological) relations the subject has to its object, but tries to make this relation clear, in order to show how the way the object represented is constituted by this particular relation. Thus, the subject of knowledge is 5. Moore reports in Wittgenstein’s Lectures 1930–32: Wittgenstein “also said that he was not trying to teach us any new facts: that he would only tell us ‘trivial’ things — ‘things which we all know already’; but that the difficult thing was to get a ‘synopsis’ of these trivialities” (PO 114).

29

not a universal subject standing in a detached relation to an equally detached object. Because of P3 to P6, perspicuous representations are not scientific, if “scientific” means giving causal explanations, where cause and effect, subject and object, past and present are independent of one another, and acceptance cannot be essential for the correctness of explanations (PO 108). Perspicuous representations do not present something that exists selfidentically and objectively understood by the subject before it is made perspicuous.6 Rather, making a case perspicuous at the same time constitutes it as having a certain meaning and coherence. It is this constitutive relation between something that happened earlier, i.e. the puzzling object or experience, and something that happens later, that is achieved by the perspicuous representation. It is best understood as a relation of Nachträglichkeit, i.e. as making something perspicuous by repeating it in a certain way. As in Freud, we start with an affect, something happens to us, a response is prompted, something seems to have meaning, moves us before we understand it, such that it effects our lives without being fully present and transparent. By representing connections in a perspicuous way, we put a face to what puzzles us (“the point is to hit on the physiognomy of the thing exactly”, PO 165). Perspicuous representations allow us to experience something in relation to other things. In this way, every new situation is turned into, or at least connected to, something already known and understood. For Wittgenstein, as for Freud, getting to know oneself follows a logic of Nachträglichkeit: “That is what that was, what felt so intense”, “Now I see what bothered me”, or “So, that was me, after all”. Wittgenstein was well acquainted with Freud’s tenets and commented on the latter one’s The Interpretation of Dreams and Jokes and their Relation to the Unconsciousness. Wittgenstein also liked to compare his method with Freud’s therapy. However, he criticizes Freud — as well as, by the way, Frazer and Darwin — for confusing what in fact are perspicuous representations or “a good way of representing a fact” (PO 107) with scientific hypotheses about the fact itself, that is, for confusing the search for the reasons for 6. Hence, Anscombe’s translation of “Darstellung” by “representation” is misleading, in so far as it suggests that there is an identity or a presence waiting to be re-covered or re-presented by a perspicuous representation. Wittgenstein, of course, dismisses the “regime of representation” in any of its forms. In addition, “Darstellung” has an active connotation, stressing that something is done or carried out by an agent. Notice that in the present context repetition also is to behave in a certain manner.

30

dreams (and symptoms in general) or, in the case of Frazer, strange rituals with the search for empirical causes. G. E. Moore quotes Wittgenstein saying that what is the most striking about Freud is “the enormous field of psychological facts which he arranges” (PO 107; cf. LC 19). 2.2 In his discussion of Freud’s The Interpretation of Dreams, Wittgenstein offers an account of interpreting a dream that is structurally similar to the logic of Nachträglichkeit: To interpret a dream means to turn it into something coherent by repeating it in a perspicuous way. When a dream is interpreted we might say that it is fitted into a context in which it ceases to be puzzling. In a sense the dreamer re-dreams his dream in surroundings such that its aspect changes. It is as though we were presented with a bit of canvas on which were painted a hand and a part of a face and certain other shapes, arranged in a puzzling and incongruous manner. Suppose […] that we now paint forms — say an arm, a trunk, etc. — leading up to and fitting on to the shapes on the original bit; and that the result is that we say: “Ah, now I see why it is like that, how it all comes to be arranged in that way, and what these various bits are …” […] What is done in interpreting dreams is not all of one sort. There is a work of interpretation which, so to speak, still belongs to the dream itself. In considering what a dream is, it is important to consider what happens to it, the way its aspect changes when it is brought into relation with other things remembered, for instance. […] If one now remembers certain events in the previous day and connects what was dreamed with these, this already makes a difference, changes the aspect of the dream. If reflecting on the dream then leads one to remember certain things in early childhood, this will give it a different aspect still. And so on. (All this is connected with what was said about dreaming the dream over again. It still belongs to the dream, in a way.) (LC 45– 6; emphasis added by K.P.)

For Wittgenstein the interpretation of a puzzling dream, at least from the first-person perspective, is not to be separated from the dream itself but “still belongs to the dream” insofar as it repeats it. (For Freud, “psychic reality”, contrary to material and psychological reality, also exists not independently of the ways it is approached in analysis: free association and transference.) But is this not a very strange claim? Is not a dream very different from its later interpretation? After all, when I dream I am asleep, whereas when I am interpreting I am awake and doing very different things. Why does the dreamer not just remember the dream when he interprets it? I think

31

Wittgenstein wants to stress, firstly, the constitutive relation between dream and interpretation. While remembering usually means representing or reproducing something which was meaningful independently of being remembered, a dream only becomes meaningful by interpreting it. Secondly, the very memory of the puzzling nature of the dream asks for an interpretation. Thirdly, Wittgenstein seems to have cases in mind where connecting, for instance, “certain events in the previous day” with the dream may affect the dreamer as if he were re-dreaming the dream in the present. Wittgenstein also appears to agree with Freud that we have to repeat what we cannot remember or do not understand. As Wittgenstein stresses, the dream is not just repeated as the same, i.e. by dreaming it again, but in a different way: The difference consists in connecting it “with other things remembered” or “certain events in the previous day” (what Freud calls “day’s residues”, Tagesreste). Like Freud, Wittgenstein does not identify a dream with its manifest content, that is, with what the dreamer tells about her dream before its interpretation. In other words, he agrees with Freud on a structural or formal level: A dream cannot be reduced to something self-identical which happened in the past and has a meaning or coherence of its own, independently of what the subject does later, of how she responds to it, etc. — in the sense that the dream-interpretation would only re-present or repeat an original experience. Rather, it is the content, meaning, coherence of a dream, in short, its identity, which is constituted retroactively (nachträglich) by repeating it, i.e. by connecting its manifest content with day-residues, memories etc. In this sense the dream as a meaningful or perspicuous whole only exists in the effects it has on the subject. Again, this does not mean that the dream exists only if it is later interpreted in the sense that the interpretation would bring the dream into existence — constitution is about meaning, not about existence. As with Emma and the Wolfman, there have to be at least two events: an earlier experience and its later repetition. Only by repeating it in a different way, i.e. retroactively repeating it, its relevant meaning gets constituted. 2.3 Of course, there are important differences between Freud and Wittgenstein. Although both agree that repeating something means repeating it differently, for Wittgenstein the difference introduced by the repetition does not consist in disguising or distorting what is repeated — as in the cases of Emma and the Wolfman. (Additionally, Emma and the Wolfman do not know that they repeat; for them, the repetitions that triggered their neuroses felt as though they were new and unprecedented experiences.)

32

In The Interpretation of Dreams Freud explicitly identifies the dream with the unconscious dream-work, and not with its latent content (Freud 1900, 506), and it is the dream-work that results in distorting — via condensation and displacement — the latent sexual thoughts and wishes into the manifest content. Wittgenstein, however, rejects the assumption of a dream-work, because he is sceptical of dreams “as camouflaged wish fulfilments” (LC 7) and of what he takes as Freud’s reductive project of looking for the hidden meaning of symptoms, which takes the form of “this is really only this” (LC 24). Also for Freud, but seemingly not for Wittgenstein, a dream itself, and not just its interpretation, is a repetition (of a repressed wish). 2.4 The retroactive connection between an object or experience and its perspicuous representation works not only for the interpretation of dreams, but also for the representation of strange and foreign cultural rites and rituals as illustrated by Frazer’s The Golden Bough.7 Representing a past ritual in a perspicuous way synchronizes past and present. The same feelings that in the past have called the ritual to life are repeated, and the effect the rituals may have on us is thus constituted. Frazer himself, of course, would not agree. In The Golden Bough, he offers genetic as well as instrumentalist explanations of ritual practices, superstitions, and myths. Frazer’s aim was to demonstrate by an evolutionary theory how mankind had elevated itself from superstition, magic, and religion to rationalism and science. Our enlightened mind, he argued, could hardly understand the irrational practices of the “savages” (Frazer’s term) and past civilizations. Hence, the picture Frazer wanted to draw was one of utter difference and discontinuity between the primitive and the contemporary Western mind. However, when one is reading Frazer’s violent and often erotic descriptions of apparently irrational practices, one cannot help but feeling puzzled, deeply impressed, fascinated, bewildered — though certainly not detached. At least these were Wittgenstein’s and his contemporaries“ reac7. I argue elsewhere (in Puhl 2002 and 2004) that Wittgenstein’s remarks on rulefollowing can also be read as following the logic of Nachträglichkeit. For the identity of rules, their normative power, is dependent on the manner we follow them (repeat them) in situations not yet considered. A rule is constituted retroactively according to the behavior that counts as following it. Here the two events linked are, first, an empirical regularity — if people are trained in certain ways, they in most cases get 4 as the result of multiplying 2 by 2 — and, secondly, repeating this behavior in a different, i.e. in a normative context where you have to come up with “4” if your behavior is to count as following the rule for multiplying “2” by “2”. Similar to Freud’s cyclic time of Nachträglichkeit, Wittgenstein also uses the metaphor of a circular movement for the retroactive constitution of the rule out of behavioral regularities (cf. RFM VI §§7–8, and my comment in Puhl 2004, 163).

33

tions.8 Why this response, this involvement, this not being able to keep a detached attitude? Wittgenstein reacts by making perspicuous connections and similarities between our practices and the foreign ones. He calls attention to practices we already understand and find unproblematic and which are related to the ones we are puzzled about. For example, when I am bewildered by the practice of burning a puppet during the Beltane fire festival (which are still held in Scotland) or by people who kiss small pieces of paper, I might be reminded that I also have sometimes thrown things against the wall, hit the floor with a stick or the table with my fist, or that I keep pictures of persons I love, that I knock on wood for good luck and much more. Seeing such connections might satisfy me, might dissolve the puzzle by allowing me to understand my reaction. In a surprising move, Wittgenstein extends the perspicuous representation to Frazer’s method itself. He thereby reinterprets Frazer’s alleged scientific and detached approach and makes clear that Frazer has misunderstood his own project, because he ignored his own subjective relation to the alien rituals and practices. Wittgenstein points out that Frazer, by the tone, structure etc. of his text, shows a personal response to the facts he describes: When Frazer begins by telling us the story of the King of the Wood of Nemi, he does this in a tone which shows that he feels, and wants us to feel, that something strange and dreadful is happening. But the question “why does this happen?” is properly answered by saying: Because it is dreadful. (PO 121)

Frazer’s relation to the phenomena he describes is one of being impressed and puzzled. He does not quite understand what is going on. Something he cannot put his finger on has touched him. However, according to Wittgenstein, by looking for causal explanations of the ritual practices, Frazer is looking for explanations in the wrong place. He gives an answer that is not appropriate to his puzzled response, because a causal explanation goes back to causes or evolutionary stages and ignores any subjective relation 8. Frazer, whose Golden Bough is hardly read anymore, had a huge impact on many modernists, including Freud and T. S. Eliot, and their fascination with mythical thought (see, for instance, the many references to Frazer in Freud’s Totem and Taboo). However, Frazer’s euro- and ethnocentrism and his “armchair” method have been severely discredited by recent ethnology. Frazer himself never left Europe, and he was getting his “data” from the famous questionnaires he handed out to travellers, colonial civil servants, and missionaries. Wittgenstein harshly criticizes Frazer’s eurocentrism: “What a narrow spiritual life on Frazer’s part! As a result: how impossible it was for him to conceive of a life different from that of the England of his time!” (PO 125)

34

between the subject and the objects of its inquiry. Causal explanations are just hypotheses and cannot account for the urgency of Frazer’s response. What is needed for understanding is not a correct belief about the rituals of foreign cultures, but a way of responding to or repeating these rituals that is revealing the meaning of our reactions to them. What is appropriate to Frazer’s and to our own response is to reinterpret Frazer’s explanatory story as part of a perspicuous representation, such that it allows us to understand ourselves by making perspicuous connections and similarities between our cultural practices and the alien ones. Seeing these connections makes us realize “that there is also something in us which speaks in favor of those savages’ behavior” (PO 131). “Indeed, if Frazer’s explanations did not in the final analysis appeal to a tendency in ourselves, they would not really be explanations” (PO 127). (Like many modernists, Wittgenstein believes that we share “a common spirit” with the ancient practices and their participants; PO 151.) This reinterpretation, however, is an inversion of Frazer’s “scientific” intentions to demonstrate our difference to the “primitives”. Notice that Wittgenstein does not give a better explanation of the rituals and practices Frazer describes. Neither does he come up with new information.9 He also does not ask for a neutral description, one which “would contain nothing superstitious or magical in itself ” (PO 133), although he gives an example of such a description. He wants to remind us of connections which are already there in Frazer’s text and in our responses to its descriptions.10 3. Wittgenstein’s Method: Perspicuous Representation and Nachträglichkeit The most important point here is the link between Wittgenstein’s analysis of rituals and repetition. As with dreams, the interpretation of a cultural practice belongs to this practice and repeats it: But the question “why does this happen?” is properly answered by saying: Because it is dreadful. That is, precisely that which makes this incident strike us as dreadful, magnificent, horrible, tragic, etc., as anything but trivial and 9. Wittgenstein is certainly more sceptical about Frazer’s project than, for instance, Freud, whose Totem and Taboo (1913) relies heavily on Frazer’s work, especially his Totemism and Exogamy (1910). 10. That the connections are already there is to be understood in a constitutional, not in a representational way, that is, they cannot be identified independently of our responses to them.

35

insignificant, is also that which has called this incident to life. (PO 121; first italics added)

In a sense Wittgenstein treats Frazer like a patient in analysis, as Frazer does not know that he repeats the feelings that in the past have “called … to life” the rituals and practices he writes about in his monumental work. As with dreams, the question is why the practice is repeated by its interpretation, and why the interpretation belongs to the practice. After all, the interpretation happens in a different historical and cultural context, in a book, in writing, as part of a theory. Participating in a Celtic Beltane fire festival is certainly different from writing about it. Wittgenstein’s point seems to be that Frazer in his work repeats the feelings that were motivating the past religious rituals, because he cannot help but expressing his own bewilderment, embarrassment, consternation etc. about the rituals, although he wants to give a neutral, scientific, causal approach. What is repeated, albeit in a different context, is the feeling “that something strange and dreadful is happening” (PO 121), which is the same feeling that motivated the ancient ritual in the first place. Frazer in fact gives the beginning of a perspicuous representation that would make explicit the connections between the ancient practice and our practices, habits, and beliefs. Like a puzzling dream, the strange practice is “fitted into a context in which is ceases to be puzzling” (LC 45). Thus, the subjective aspects of perspicuous representations, their dependence on our acceptance and whether they satisfy us, together with the temporal logic of Nachträglichkeit, i.e. of constituting something by repeating it differently — all these aspects do not allow for separating the phenomenon made perspicuous and its representation. A perspicuous representation not only comes later than what it represents but also constitutes its meaning and identity. Constantly working backwards to earlier experiences, practices or beliefs — i.e. a reworking and retranscribing a puzzling past through the present — links Wittgenstein’s method to psychoanalysis and to the logic of Nachträglichkeit. Wittgenstein in a sense repeats Freud, albeit with a twist, of course. The “orthodox normalization of Wittgenstein” (as Henry Staten put it) ignores or plays down this side of his method. For according to the orthodox interpretation, Wittgenstein has developed a new method for bringing philosophy back to earth, for returning home to the bedrock of our form of life, where all justifications end and the famous spade is turned (cf. PI § 217). And, of course, one can easily come up with statements by Wittgenstein which seem to underscore this view. On the other 36

hand, Wittgenstein did not stop with returning home, with “This is what we do”, contending himself with affirming our practice by making it perspicuous (cf. OC § 204). As Henry Staten has nicely put it, Wittgenstein shows a “renewed perplexity about what he himself has already resolved, a loosening-up of the bonds of syntax he has already written in order to write anew” (Staten 1984, 65). A philosophical movement that started with being lost and puzzled may come to an end, an end which is reached with the help of perspicuous representations. However, this end is only the end for a short time, only to mark the beginning of new perplexities, puzzles and problems, and their repetition and reworking in a new context in which they “cease to be puzzling”. What Wittgenstein really seems to be affirming is not any return to the ordinary, but coming up with novel configurations by repeating past problems in the present. Like Freud, Wittgenstein works backwards in order to rid himself of puzzles, fixations, and constraints. For him, doing “work on philosophy […] is really more a work on oneself ” (CV 24).11

REFERENCES Bouveresse, Jacques 1995: Wittgenstein Reads Freud, Princeton University Press. Freud, Sigmund 1895: “Project for a Scientific Psychology”, in The Standard Edition, Vol. I. –– 1900: The Interpretation of Dreams, in The Standard Edition, Vol. V. –– 1896: “Further Remarks on the Neuro-Psychoses of Defence”, in The Standard Edition, Vol. III. –– 1917/18: “From the History of Infantile Neurosis”, in The Standard Edition, Vol. XVII. –– 1953–74: The Standard Edition of the Complete Works of Sigmund Freud, ed. by James Strachey, The Hogarth Press, 24 Volumes. Grünbaum, Adolf 1984: The Foundations of Psychoanalysis, Berkeley University Press. 11. I am grateful for helpful comments to Monika Seidl, Michael Kober, and the participants of the Wittgenstein Workshop held at Ulm where an earlier version of the paper was presented. I also want to thank the Deutsche Forschungsgemeinschaft (DFG) for a research grant on the method of perspicuous representations and modernist conceptions of ordering, which made the present paper possible.

37

Puhl, Klaus 2002: “Die List der Regel“, in Ulrich Baltzer, Gerhard Schönrich (eds.): Institutionen und Regelfolgen, Paderborn: mentis, 81–99. –– 2004: “Rule-Following: Difference and Repetition”, in T. Demeter (ed.): Essays on Wittgenstein and Austrian Philosophy, New York: Rodopi, 155– 66. Rosenbaum, S. P. 1995: “Wittgenstein in Bloomsbury: 1911–1931”, in: Jaakko Hintikka, Klaus Puhl (eds.), The British Tradition in 20th Century Philosophy, Proceedings of the 17th International Wittgenstein Symposium, Kirchberg am Wechsel 1994, Vienna. Stallybrass, Oliver 1992: “Preface of the Editor”, in E. M. Forster: Howard’s End, Penguin, London. Staten, Henry 1984: Wittgenstein and Derrida, Lincoln & London, University of Nebraska Press.

38

Grazer Philosophische Studien 71 (2006), 39–55.

THE PNEUMATIC CONCEPTION OF THOUGHT

Joachim SCHULTE University of Bielefeld Summary This paper is an attempt at presenting a convincing reading of the first sentences of PI § 109, especially of its third sentence. There Wittgenstein mentions what he calls “the pneumatic conception of thought”, which by Miss Anscombe is translated as “the conception of thought as a gaseous medium”. By comparing the relevant sentences with their sources in Wittgenstein’s manuscripts and additional parallels it is found that Anscombe’s rendering is liable to be misleading. Wittgenstein’s notion of “pneuma” is likely to be inspired by a conception of the kind elucidated by Oswald Spengler, for instance, according to whose account pneuma is to be understood as a sort of body (Seelenkörper). If you take this as your central image, the pneumatic conception alluded to in PI § 109 is one which can be tracked down in Wittgenstein’s early as well as in his middle-period work. In this remark (PI § 109), Wittgenstein tries to mention both helpful and misleading features of his earlier ways of thinking, and what he now calls the “pneumatic conception” can be seen to be the foil for both sorts of features. The central idea is that the core of language contains a scaffolding of rules whose (“pneumatic”) substance is the same as that of our thought. In accordance with this view the logical structure of thought is held to be identical with that of language and reality — a notion which fits parts of the Tractatus as well as parts of the original “grammatical” conception developed in the early 1930ies.

1. The topic of this paper are the first three sentences of § 109 of Wittgenstein’s Philosophical Investigations: [1] It was true to say that our considerations could not be scientific ones. [2] It was not of any possible interest to us to find out empirically ‘that, contrary to our preconceived ideas, it is possible to think such-and-such’ — whatever that may mean. [3] (The conception of thought as a gaseous medium.) [[The pneumatic conception of thought.]] And we may not advance any kind of theory. There must not be anything hypothetical in our considerations.

We must do away with all explanation, and description alone must take its place. And this description gets its light, that is to say its purpose, from the philosophical problems. These are, of course, not empirical problems; they are solved, rather, by looking into the workings of our language, and that in such a way as to make us recognize those workings: in despite of an urge to misunderstand them. The problems are solved not by giving new information, but by arranging what we have always known. Philosophy is a battle against the bewitchment of our intelligence by means of language.

Some readers of this remark must have wondered which uses and interpretations of the concept “pneuma” (or “pneumatic”) Wittgenstein was familiar with. Certainly, an answer to that question would be of interest, even though it would be more important to arrive at an explanation of Wittgenstein’s own use of that notion. After all, his use of the word may have been different from certain uses he came across in reading the works of various authors. But still, it may be instructive and helpful as regards the understanding of the quoted remark to form at least a vague idea of the use, or uses, of the word “pneuma” Wittgenstein was acquainted with. Among those authors of whom we know with certainty that Wittgenstein had read some of their works there are several from whom we might learn something about the concept of pneuma. From the point of view of our question, however, no one seems more promising than Oswald Spengler, who was one of the two “muses” of Wittgenstein’s later philosophy (to quote a phrase used by Brian McGuinness).1 As Wittgenstein records in an often-quoted passage, the second muse was his Cambridge colleague Piero Sraffa to whom he is said to be indebted for his “anthropological” way of tackling certain philosophical problems (see Rhees 1982).2 In a journal kept in 1931 Wittgenstein wrote down a few points that had struck him in the course of his by no means uncritical reading of Spengler’s work. In this context, he places himself in the same group as 1. At this point as well as in the remainder of this paper, I heavily rely on an as yet unpublished article by McGuinness on “What Wittgenstein Owed to Sraffa”. Many of the following remarks were inspired by comments made by McGuinness. On the other hand, I am by no means sure that McGuinness would agree to every aspect of the occasionally very speculative interpretation I am going to suggest. 2. Rhees also refers to Wittgenstein’s use of the phrase “anthropological way of looking at things” (Rhees 1970, 55). Cf. Kienzler 1997, ch. 1, section 3 e.

40

Adolf Loos, Spengler, and Freud and adds that all the members of this group belong “to that same class which is so characteristic of the present time” (PPO 37 = MS 183, 29; cf. PPO 25, 27 = MS 183, 16, 19). There are several passages in his manuscripts where Wittgenstein points out that certain distinctive notions typical of his later thought owe something to Spengler’s influence. Examples are the ideas of family resemblance, surveyability (Übersichtlichkeit) and object of comparison. In the earliest two versions of the Philosophical Investigations, Spengler is mentioned by name (UF §§ 115 (117) and 122 (124); FF §§ 100 (102) and 107 (109)).3 While both passages appear in the final version of the text (SF §§ 123, 131), Spengler’s name was crossed out in the course of Wittgenstein’s revision of the early version. In Spengler’s chief work The Decline of the West the notion of pneuma plays a considerable role. Even a superficial reading will easily reveal more than the eleven passages referred to in the book’s index. At one place Spengler points out that Chrysippus was of the opinion that the divine pneuma is a kind of body (Spengler 1972, 229). In the context of contrasting the “Faustian” concept of the soul with the “Apollonian” concept he makes a distinction between “soul space” (Seelenraum) and “soul body” (Seelenkörper) and suggests that the strikingly rigid gaze of people represented by Byzantine works of art was meant to portray their participating in divine pneuma (Spengler 1972, 389–91). In magic culture, a macrocosmic dualism between god and devil corresponds, according to Spengler, to a microcosmic dualism between pneuma and psyche (Spengler 1972, 399). Even the jurisprudence and various political theories prevalent in certain cultures of the East are claimed to have been affected by the notion of pneuma (Spengler 1972, 635). There is a section in Spengler’s book entitled “The Magic Soul” where a number of aspects of this notion are discussed in great detail and developed against the background of a distinction between two soul substances (distinguished by labels like “spirit” or “soul”, “pneuma” or “psyche”, “ruach” or “nephesh” and so on) — a distinction that, of course, had been crucial for St Paul’s reading of Christian doctrine. Some of the ideas repeatedly mentioned by Spengler are summarized in the following passage: 3. The abbreviations and the terminology (UF = Urfassung, FF = Frühfassung, ZF = Zwischenfassung, and SF = Spätfassung) are used in the critical-genetic edition of the Philosophical Investigations (i.e. Schulte 2001).

41

Faustian man is an Ego, a power that is committed to relying on itself while ultimately in charge of the infinite. Apollonian man, as one soma among many, is responsible to no one but himself. Magical man, on the other hand, in his spiritual being is nothing but a constituent of a pneumatic We which, floating down from above, is one and the same in all partakers. As body and soul man belongs exclusively to himself. But there is something else, something alien and higher, present in him; and that is why he himself together with all his discoveries and beliefs is felt to be a member of a consensus which, as an emanation from the deity, excludes error as well as every possibility of an Ego capable of establishing a form of value (Spengler 1972, 843).

Spengler suggests that a variety of views that, from our present Western point of view, appear characteristic of Christian thought are inconceivable without the concept of pneuma as a divine substance in which man can at most partake. Pneuma is a divine spark and, as Spengler vividly puts it at one point, can be seen “not as an ego but as a kind of visitor” (Spengler 1972, 869). 2. The mere fact that Wittgenstein was familiar with this idea described by Spengler can be regarded as evidence supporting criticisms of Miss Anscombe’s translation with its implicit exegesis of the relevant passage. Where the German text has “Die pneumatische Auffassung des Denkens”, the English version speaks of “The conception of thought as a gaseous medium”. In his commentary, Eike von Savigny (1994, 163) says that, “unfortunately”, this rendering is “very restrictive”. Brian McGuinness, who in the article mentioned speaks of “the pneumatic theory of thought”, offers an even more pronounced criticism of Anscombe’s English translation and emphasizes that “pneuma is certainly not gas”. McGuinness writes that something’s being pneumatic does not amount to its being gaseous; the pneumatic is something structural, perhaps even something concrete, that is meant to support the sense of our thoughts and utterances. This figurative reading of Wittgenstein’s figurative formulation is not difficult to render compatible with the notion explained by Spengler and familiar to Wittgenstein, and it surely points in a different direction from that indicated by Miss Anscombe’s implicit interpretation. However, McGuinness doesn’t primarily refer to Philosophical Investigations § 109 but to a manuscript passage which, on the surface, is not easily recognized as an earlier version of the printed remark. This passage from MS 157b (which will be quoted below) poses a number of additional problems that 42

we cannot go into in the present context. It does, however, form part of an entire series of manuscript remarks that may prove helpful in the attempt to give a consistent reading of our text. The passage referred to by McGuinness can be found in the second of two small consecutive notebooks that Wittgenstein used while he was working on the Ur-version of the Philosophical Investigations. It was jotted down on one of the two last days of February 1937 or a little later. These entries (which were written in Norway) are among the most impressive in Wittgenstein’s entire Nachlass: they permit us to catch an extraordinarily direct glimpse of Wittgenstein’s thought and his struggle with earlier positions recently, or not yet entirely, abandoned. Here, however, this passage will be mentioned at a later point of the following chronological survey. Altogether there are roughly two dozen interesting occurrences of the word “pneumatic” (or its cognates) to be found in Wittgenstein’s manuscripts. But as regards these passages, two important points are to be taken into account. First, several of these occurrences are repetitions or slightly reformulated revisions of passages that were written earlier. Second, there are three distinct periods in Wittgenstein’s philosophical development where these passages occur, and these periods are of varying relevance to our story. These periods are the following: (1) February 1932 (that is, at a time when Spengler’s and Sraffa’s influence was particularly strong); (2) February (perhaps March) 1937, when Wittgenstein was in Norway working on the earliest version of the Philosophical Investigations; (3) in 1945 when he was busy producing the intermediate version and perhaps part of the final version of the Philosophical Investigations. On 29 February 1932 Wittgenstein notes down the following remark: (A)

The sense of a sentence is not pneumatic; it is what is answered in reply to a request for an explanation of the sense. And — or — one sense differs from another the same way an explanation of the former differs from an explanation of the latter. (MS 113, 82)

On the same day he writes three manuscript pages further on: (B)

And the sense of a sentence is after all not something we explore and which may in part remain impossible to find out about. In such a way that it might be possible for us to hit on it only later that this very sentence may be known in another way by beings who are different from ourselves. In such a way that it would remain this sentence with this sense, where this sense would have attributes we today don’t have the

43

slightest inkling of. The sentence, or its sense, is not a pneumatic being that has a life of its own and is now going through adventures we need not know anything about. That would be as if we had breathed a spirit into it of the same kind as our own — its sense — but now it had — as if it were our child — a life of its own, and we were (merely) able to explore it and to understand it more or less. (MS 113, 85)

Both quotations raise certain problems of interpretation that are connected with Wittgenstein’s use of the word “or” (A: “And — or — one sense differs from another […]”, B: “The sentence, or its sense, is not a pneumatic being […]”). Considerations of the context of these remarks show that they are both connected with a number of further central thoughts developed by Wittgenstein (e.g. the notion of a calculus, of internal relations, etc.). Neither these exegetical questions nor the wider connections can be taken into account at this point. What will be stressed are a few fairly clear ideas that can be gleaned from those two quotations. From (A) we may learn that the sense of a sentence is not something pneumatic in this respect: that it is completely determined by an explanation — a public explanation. This insight is supplemented by (B): Sense is not something that can be explored. At any rate, it cannot be explored the same way one goes about exploring an unknown country or the bottom of the sea. We cannot comb through it to find out whether we shall encounter something of interest. It neither is nor contains a great secret that might remain hidden from the explorers’ glance.4 The identity of sense5 does not rest on features that are (as a matter of principle) inaccessible to us. The respect in which the sense of a sentence is said to be non-pneumatic is this: it does not have an existence independent of us 4. The way McGuinness describes it one is led to think of Wittgenstein’s well-known remarks on dogmatism from his conversations with Schlick and Waismann (WVC, 182ff., 9 December 1931). Here Wittgenstein says: “The wrong conception which I want to object to in this connection is the following, that we can hit upon something that we today cannot yet see, that we can discover something wholly new. That is a mistake.” In notebook 157b, from which I shall quote below, Wittgenstein articulates his objections to dogmatism as follows: “The only way we can avoid the unfairness, or emptiness, of our claims is by presenting the ideal as what it really is, viz., by presenting it in the context of our considerations as an object of comparison — as a standard, as it were — and not as a kind of prejudice everything must conform to. For that would be what dogmatism lies in” (MS 157b, 15v f., cf. the corresponding passage in CV 30). Immediately after this passage Wittgenstein makes a comparison with Spengler. — The connection between the trend of these anti-dogmatic remarks and the thrust of § 109 will be obvious. 5. Of course, the idea of the “identity” of sense, or of a context-invariant form of sense, is a notion insistently brought into question in Wittgenstein’s later philosophy.

44

(the speakers of the relevant language). In other words, while something pneumatic would indeed be something whose spirit was of the same kind as ours — and here the parallel with the idea of partaking in a divine kind of spiritual substance becomes obvious — it would be beyond our control and might lie outside the realm of what we can know. 3. The two passages from February 1932 quoted above (A and B) travel, along with a great number of other remarks of that time, via TSS 211 and 212 into the so-called Big Typescript (TS 213). From there, they continue their careers and end up, after a few but interesting modifications6 in the “Umarbeitungen” (reworkings) contained in MSS 114 and 115. What is of particular interest for us are the immediate predecessors of the final version of §109 of the Philosophical Investigations. In their earliest coherent form the majority of the formulations of this remark can be found in notebook 152 (called “C8” — i.e. the last notebook of the C-series; cf. von Wright 1982, 45).7 From here, they are transferred to MS 142 (the Ur-version of the Philosophical Investigations) only to be revised after a short time and in the same manuscript book.8 The most striking change made by Wittgen6. MS 114, 125f.: “The sense of a sentence is not pneumatic (nor is a thought); it is what is answered in reply to a request for an explanation of the sense. Or rather, one sense differs from another the same way an explanation of the former differs from an explanation of the latter. And that means furthermore that the sense of one sentence differs from the sense of another sentence the way one sentence differs from the other. / The sense of a sentence is not a kind of soul.” MS 115, 77: “The sense of a sentence is, after all, not something we explore the same way we explore the structure of matter and which may in part remain impossible to find out about. (Unsolved problems in mathematics.) In such a way that it might be possible for us […].” 7. MS 152, 94f.: „It was true to say that our considerations were not to be scientific ones. [Marginal note with no exact location: Nothing hypothetical may attach to them.] The known fact ‘that something can be thought //that this or that can none the less be thought// (whatever that may mean) was of no possible interest to us. We had to do away with all explanation — and description alone must take its place //be in its place// //and description alone was to take its place//. And this description got its light, that is to say its purpose, from the philosophical problems. These are, of course, not empirical problems; they are solved, rather, by grasping [Einsicht] the nature //working// functioning// of our language! And that in such a way as to disclose this nature: in despite of a tendency //temptation// to misunderstand it. Not by new information //empirical facts// [Erfahrung //Erfahrungstatsachen//], but by arranging what we have always known. | Philosophy is a battle against the fascination of language [this I have put in a better way]. [[I have not been able to discover an earlier formulation of these words; the earliest occurrence of the well-known formulation “against the bewitchment of our intelligence by means of language” can be found in UF.]] 8. The first versions of the relevant remarks can be found in the so-called “island” — a number of remarks between pages 77 and 91 of MS 142 which were crossed out immediately after their

45

stein in the course of this revision is the addition of the following words inserted above the line: “(The pneumatic conception of thought.)” (UF 110 (112)) We may be certain that this addition is — in the same way as a number of further changes on these pages of the Ur-version — derived from a few pages written down a little earlier in notebook 157b, which was begun in Norway, on 27 February 1937, as an immediate continuation of notebook 157a. It is tempting to quote pages and pages from this notebook, but here I shall limit myself to a few particularly relevant lines: (C)

It really seems as if logic lost what was essential to it: its rigour. As if it had been lured away. [PI § 108] Now, however, it plays a different role. What used to be a prejudice concerning reality has turned into one form of representation. Where has its crystal clearness ended up? It has turned into a form of representation, nothing else.9 Understanding not a pneumatic process. The concept of family two axe strokes against the — — — 10 For it became manifest that I did not have a general concept of a sentence and of language. I had to recognize this and that as signs (Sraffa), even though I could not assign them a grammar. Understanding and knowledge of rules. The pneumatic element of understanding disappeared completely, and together with it disappeared the pneumatic element of sense. At first strict rules seemed to be something |still| hidden in the background, in the |nebulous| medium of understanding; and it was possible to say: they must be there — or: I can see them, through a thick medium,

composition (cf. my Kritisch-genetische Edition, 133, note 18, and 1090). In the meantime, I have come to the conclusion that one of the most serious defects of this edition consists in the fact that this “island” has been left out of consideration. It would have been important to give at least a rough idea of its content. 9. At this point, the manuscript contains a symbol indicating connection. The intended order of paragraphs is unclear; it is not even clear whether they are meant to be paragraphs or fragments of sentences. 10. McGuinness supplies the following reading: „Wittgenstein (says in the notebook) that the idea of the family [i.e. family resemblance, by inference and by other references to Spengler] and [the realization that] understanding was not a pneumatic process were two axe strokes against [his previous doctrine — of the crystal clarity of logic in itself ].“

46

as it were, but I do see them [cf. PI § 102]. And hence they were concrete. I had used an analogy (of a method of projection, etc.) but because of the grammatical illusion of a unified concept it did not seem to be an analogy. (MS 157b, 5r–6r)11

Obviously, this series of jottings is a kind of self-reflective report on Wittgenstein’s own intellectual development. From the perspective reached in 1937, it gives an account of those traits of his earlier philosophy that, around 1930 or a little later, were seen to be erroneous or misleading and had accordingly to be removed or modified. There is an allusion to the well-known story of Sraffa’s Neapolitan gesture performed while demanding an elucidation of the logical form of this gesture (cf. Malcolm 1984, 57f.). Here there was no space for a logical (or “grammatical”) analysis. At the same time, the gesture was perfectly intelligible. This way it became clear that understanding is not a kind of tracing or grasping of an, as it were, “concrete” scaffolding of rules hidden behind our linguistic signs. Understanding is not a pneumatic process: it is not dependent on partaking in the substance of a logical structure (MS 157b, 9r ) underpinning and regulating language — not even if the structure is assumed as a mere model. But if the idea of understanding as a process of grasping pneuma (the logical structure, or a meaning body) collapses, then what disappears at the same time is the notion of a sense functioning like a kind of pneuma. It is exactly this idea which is expressed by the quotations (A) and (B) mentioned earlier. The “nebulous” or “thick” medium mentioned in the last paragraph of the above-quoted passage (C) is not identical with pneuma. The pneumatic would be what is hidden in this sort of medium or what is visible through the medium. Accordingly, one might say that the pneumatic conception goes together with or is accompanied by the idea of understanding as a kind of (possibly gaseous) medium.12 What is pneumatic, however, would be 11. The transcription on which this translation is based can claim neither completeness nor ultimate correctness. (The same is true of some of the other quotations from manuscript sources.) 12. A special role is played by the picture of a gaseous medium whenever the difference between two substances — the physical and the psychological — is emphasized. In this context, Wittgenstein sometimes favours the expression “ethereal”, cf. BB 47: “The mental world in fact is liable to be imagined as gaseous, or rather, ethereal.” In the passages we have looked at previously, however, the word “pneumatic” is used to underline the idea of sameness of, or sharing in a, substance. Presumably, this is the notion which is at the basis of the following idea, alluded to in the Philosophical Investigations § 96: “These concepts: proposition, language, thought, world stand in line one behind the other, each equivalent to each.”

47

something structural, a kind of scaffolding consisting of rules, for instance, in whose substance thought partakes. This sort of partaking would then make it possible for thought to discover that a certain use of language is in accordance with the rules — it would re-cognize something of its own kind. Making this possible, however, would not be due to an additional means (a medium, a vehicle or something of that kind); it would be due to sameness of substance or structure, that is, sameness of “logical structure”. 4. To conclude this survey of Wittgenstein’s uses of the expression “pneumatic” (etc.) I shall now proceed to discuss the relevant passages from the last of the three periods distinguished above. In 1945 Wittgenstein wrote down an early version of what became § 308 of the Philosophical Investigations and remarked, among other things: (D)

The first step is the innocent pneumatic conception, where (however) the ‘kind’ of processes or states is left open. But the next step consists in seeing that whatever kind this something one wishes to talk about may belong to, it won’t explain anything and will be a useless fiction. If now, however, one drops this fiction, one seems to discard everything mental and to claim at the same time that only what is physical exists. (MS 116, 332)

Of course, this conception can really be (or anyway appear) “innocent” only if the word “pneumatic” is used in a completely modest and neutral sense involving no theoretical alignment, that is, only if it is used in an entirely unbiased and plain or, perhaps, pedestrian sense of “psychological” or “mental”. As a matter of fact, an even earlier version of this remark, which was noted down a few lines before the passage quoted, runs as follows: “The first step which is taken is the seemingly innocent one of talking of something mental whose exact classification is left open for the time being.” Accordingly, I shall assume that in this passage the word “pneumatic” is used in a sense of “mental” which involves absolutely no theoretical element.13 On the first pages of that part of MS 130 which was presumably writ13. In 1945 or thereabouts Wittgenstein was working on ZF, i.e. the intermediate version of the Philosophical Investigations, and shortly afterwards he proceeded to produce the final version (SF). Thus it is easy to imagine that in the relevant remarks originating from this time his use of the word “pneumatic” was inspired by § 109 or its predecessor. On the other hand, in all the relevant passages written around 1945, the meaning of the word “pneumatic” is clearly different from the meaning it bears in the earlier passages quoted above.

48

ten in 1945 the word “pneumatic” occurs several times. The first of these passages is the following one: (E)

What I object to is the pneumatic conception. That is, to the extent it is the same as the conception of the gaseousness of the soul. The opposite of the pneumatic conception is the behaviouristic one; and they are both bad. “You claim: He is merely saying it, and nothing is going on inside him.” My answer is: “The word ‘merely’ is misleading, and so is talking of an accompanying inner process.” (MS 130, 3)

In this passage, the word “pneumatic” is evidently used to describe that roughly speaking “mentalist” position which intends to give our ordinary use of psychological concepts a foundation by appealing to psychological processes and states, and which is the target of a great number of critical remarks in the Philosophical Investigations. Maybe it was this passage that inspired Miss Anscombe to translate PI § 109 the way she did. But apart from the fact that this use of the word “pneumatic” (in the sense of a mentalist position opposed to reductionist versions of behaviourism) does not really fit our sentence from PI § 109, it must be remembered that the “pneumatic conception” mentioned in (E) is not unconditionally identified with the “conception of the gaseousness of the soul”. No, the pneumatic conception is attacked only to the extent it really involves the idea of gaseousness. That means that in this passage Wittgenstein appears to allow for a less doctrinaire sort of pneumatic conception which talks of the mental in an unbiased fashion. Probably it is this same harmless way in which the word is used in the following passage, which is to be found on the page following the previous quotation: (F)

If you wish, you may call the use of certain words ‘pneumatic’. But that does not mean that in this case they refer to a pneuma. (MS 130, 4f.)

It is likely that the pneuma which is said to be not referred to by words used in the “pneumatic” way is supposed to be a kind of soul substance; and in the context of the passage quoted the notion of a soul substance meant is likely to be one held by a mentalist position opposed to reductionist versions of behaviourism. At any rate, what ought to be remembered is the fact that Wittgenstein appears to allow for an harmless, non-doctrinaire use of the word “pneumatic”, which counts as different from the censured mentalist point of view. The non-doctrinaire use would then

49

more or less amount to the unbiased everyday use of words like “mental” or “psychological”. The last occurrence of the word “pneumatic” is to be found in TS 235, which is an undated list of 163 remarks that are enumerated by means of brief summarizing key words. In his catalogue of Wittgenstein’s Nachlass writings Georg Henrik von Wright gives the following succinct description of this typescript: “Typescript of a Table of Contents to an unidentified work. Date unknown.” I have never come across a study discussing these nine pages in any detail. After a superficial glance at this list I am tempted to conjecture that it too was written around 1945. (McGuinness, who has had a look at the original, is inclined to think that it was written a little later, perhaps 1947 or 1948.) A more accurate description would require a comprehensive comparison between each entry of this list and its counterpart in Wittgenstein’s identifiable projects from his later years. At any rate, on page 3 the following entry can be found: (G)

50. The ethereal character of mental phenomena. The pneumatic conception.

The question of the identity of the intended remark “50.” cannot be answered with any degree of certainty. What is likely, however, is that here again the expression “pneumatic conception” is meant to refer to a mentalist kind of position. On the other hand, it must remain open whether the mentalism intended is of the vague and harmless everyday kind or to be identified with a philosophically ambitious theory entailing substantial ontological and epistemological consequences. 5. Now that we have assembled all these reminders we are in a position to embark on our real exegetical task, and this task is not an easy one. Particularly obvious problems are the following ones: (a) The tense in sentences [1] and [2] of § 109; (b) the scope of “It was true to say” in [1]; (c) the sense, or point, of [2]; (d) the link between [1], [2] and [3] — after all, there must be a way of connecting [3] with the preceding sentences if we are to make any sense of it. (a) The question of tense: the striking past tense (“was”) in [1] and [2] has often been read as a kind of reference to Wittgenstein’s preoccupation with certain ideas of the Tractatus (cf. Hallett 1977; Baker and Hacker 1980; re § 109). The correctness of this approach is confirmed by earlier versions of PI remarks mentioned or quoted above. It is well-known that 50

Eike von Savigny has suggested a competing interpretation which deliberately ignores questions of the origin of the text. According to this interpretation, the word “was” refers back to § 89b while the word “seemed” used in this latter paragraph refers back to § 81a. In my view, the readings resulting from these apparently opposed approaches can under certain conditions turn out to be quite compatible: a criticism which, from the author’s point of view, has a specific reference may, from a reader’s point of view, have a far wider field of application. What is clear and uncontroversial is the observation that in Philosophical Investigations §§ 89ff. Wittgenstein harks back to the Tractatus. Not only unmistakable allusions are contained in these remarks but even explicit references occur in the text of the Philosophical Investigations. Moreover, in his preface the author says that he feels that only by being contrasted with his former views and against the background of these views can his more recent ideas be seen in an appropriate light. This of course means neither that in each and every single case the intended reference to the Tractatus will stick out, nor that taking account of allusions to the early book will render looking for links with other parts of the Philosophical Investigations superfluous. In the present case, however, the connection between PI §§ 109[1, 2] and 89b does not seem sufficient to explain the choice of tense with its insistent allusion to once upon a time. Here and in the surrounding paragraphs the back-reference to Wittgenstein’s earlier thought is too obvious to permit any other association to take pride of place. However, in the case of Wittgenstein’s self-critical remarks of this type one should never forget that the Tractatus is not the only possible target: he may also be referring to views held at a later time. (b) As regards the scope of “It was true to say”: here we are dealing with the question whether this expression only applies to the remainder of the first sentence or to certain parts of the sequel too. If one takes the structure of the whole paragraph into account — its starting from “true” insights contained in the early philosophy and its gradual leading up to especially characteristic formulations of Wittgenstein’s later philosophy –, one is inclined to think that the field of application of the expression “It was true to say” extends beyond the first sentence. What also speaks in favour of this reading is the further fact that the past tense used in the earliest version of our paragraph (MS 152, see footnote 7, above) stretches as far as the seventh sentence of PI § 109.14 More reliable statements regarding 14. Of course, the text of the remark as expressed in MS 152 is shorter than the printed

51

the scope of “It was true to say” can only be made after the sense of [2] has become clear. (c) Now, let us look into the question of the sense of [2]. Surely the most glaring uncertainty will be the problem whether the author regards it as a welcome or an unwelcome fact that the experience mentioned in quotation marks was of “no possible interest to us”? If our lack of interest was justified (“true”), the experience mentioned must be regarded as something unwelcome. Maybe this will come as a surprise. After all, the notion that “contrary to our preconceived ideas, it is possible to think such-and-such” is often used by Wittgenstein to support his arguments, as Eike von Savigny rightly points out. For this reason one should according to von Savigny emphasize the word “Erfahrung” (“experience”). That could be understood as meaning that “experience” is — as in many other passages of Wittgenstein’s writings15 — to be read in the sense of “passively undergone feeling” or “sensation” or “qualitatively specific stimulation” (this reading is not the one suggested by von Savigny). But, of course, stressing the word “Erfahrung” is not incompatible with reading it in the sense of “what is given in experience” or “empirical finding”.16 A further problem of interpretation lies in the fact that there are two words that correspond to immediately preceding or immediately following portions of the text in a conspicuous way. One of them is the word “think” in [2], which could be seen as corresponding to the word “thought” in [3]; the other one is the expression “Vorurteil” [“preconceived ideas”], which could be taken to refer back to PI § 108.17 The question is whether version. The present tense used in “Nothing hypothetical may attach to them” is irrelevant to our present concern: this sentence was added at a later stage and was inserted at the bottom of the page. 15. There is plenty of evidence for this way of using the expression. A particularly striking passage is PI § 611: “One is inclined to say, ‘Willing too is merely an experience’ […]” 16. It seems to me that at this point Miss Anscombe’s translation is a little too free. Her rendering runs as follows: “It was not of any possible interest to us to find out empirically ‘that, contrary to our preconceived ideas, it is possible to think such-and-such’ — whatever that may mean.” If one accepts the “empirical” reading, one might perhaps wish to translate the passage this way: “The fact that one had found ‘that, contrary …’ … was of no possible interest to us.” 17. The fact that three paragraphs seem to lie between the italicized word “Vorurteil” [“preconceived idea”] in PI § 108 and the occurrence of the same word in § 109 is due to the editors’ possibly erroneous decision to incorporate these paragraphs from a separate sheet inserted into the typescript of the final version (SF) of the Philosophical Investigations into the body of § 108 (see my critical-genetic edition, 808f.). In Wittgenstein’s relevant manuscripts and typescripts the occurrences of this word are separated by very small portions of text — in MS 142 they can be found on one and the same page.

52

these correspondences are of real significance and ought to be taken into account in serious attempts at giving an interpretation of this passage. This question points in the direction of (d), for if the word “think” in [2] hangs together with “thought” in [3], the parenthetical remark “(The pneumatic conception of thought)” ought to be associated with that part of [2] which is enclosed in quotation marks. Now I want to try to remove at least a few of these perplexities by way of presenting my own reading, starting from question (d): What is the connection between [1], [2] and [3]? In trying to appreciate the following considerations it will be useful to bear in mind that when working on the relevant passage of the Ur-version (PI §§ 108, 10918) Wittgenstein very probably had MSS 152 and 157a and b in front of him. PI § 109 was transferred from MS 152, and in this process it was considerably modified. This happened after Wittgenstein had, on the same page of his manuscript, completed the Ur-version of PI § 108, which is based on MSS 157a and b (cf. the lengthy quotation in footnote 10 on pages 142f. of the critical-genetic edition).19 It is not unlikely that what stimulated Wittgenstein to insert his allusion to the pneumatic conception of thought was his re-reading of passage (C), which was quoted above. A further addition to the version given in MS 152 was evidently written down straight away or after a very short period of reflection, and this addition consists of the words “contrary to our prejudice” (or “our preconceived ideas”, as Miss Anscombe has it) in [2]. In my view it is likely that this addition is due to Wittgenstein’s noticing the word “Vorurteil” in his manuscript, where it occurs several times and appears underlined in the manuscript version of PI § 108. Thus, the prejudice which is meant is the prejudice of the crystal clearness of logic (“nothing amorphous”!). In MS 152 the idea had simply been this: that a correct part of the conception held at the time of the Tractatus had been the exclusion of “scientific” considerations. And equally correct had been the idea that feelings or experiences of the conceivability of this or that were — probably because 18. In UF they bear the numbers 109 (111) and 110 (112). To avoid taxing the reader’s memory unduly I shall here and in the sequel use the paragraph-numbering of the Philosophical Investigations to refer to the relevant paragraphs of other manuscripts or typescripts. 19. It is true that earlier versions of the two relevant paragraphs can be found in the area of the “island” mentioned above (footnote 8). But various features of the text indicate that when Wittgenstein was making his second attempt at formulating these paragraphs in MS 142 he did not content himself with consulting MS 142, 90f., but was working from the notebooks mentioned above.

53

of their “scientific” character — to play no role in the author’s reflections. The “scientific” character of such feelings or experiences would have been due to their role as evidence that could be used to support or refute this or that claim. What would run counter to our prejudice of crystal clearness (which is mentioned not in the early notebooks but only in the Ur-version and its successors) are all kinds of notions or conceivable ideas that cannot be derived with crystalline exactness and clarity from certain rules — the scaffolding of rules inherent in our language. (In MS 157a Wittgenstein speaks of “crystal systems”.20) The reason why feelings or sensations or experiences of the kind mentioned were of no possible interest to us was the following: Our conception was a pneumatic one, that is, it was a conception according to which thought obeys the directions enshrined in the crystalline structure of the scaffolding of rules inherent in our language. And the directions implicit in that structure can be obeyed by thought because it itself participates in the same substance as the logical structure of language. Thus, what was correct in the pneumatic conception was its excluding scientific, or empirical, features from our considerations. From Wittgenstein’s earlier point of view this exclusion (which in 1937 continued to count as a “correct” move) had been regarded as justified for a reason which had its roots in the idea that what is scientific or empirical is, not sufficiently exact, but, owing to its contingence and its lack of transparency, impure, possibly amorphous, and not compatible with the sublimity,21 or purity, of logic. At the time of the Tractatus, that exclusion could be justified by reference to the pneumatic conception held by the author of that book: the exclusion followed as a consequence from this conception.22 In conclusion and by way of paraphrasing Wittgenstein’s words, I want to summarize sentences [1] to [3] as follows: What was right about the conception of the Tractatus was the conviction expressed by it, according to which no part of it was allowed to be scientific in the sense of being empirical or based on empirical evidence. That was the reason why at that time nothing that would have run counter to the prejudice of crystalline 20. This indicates that the “crystal” metaphor signifies, not only translucency and transparency, but also straight linearity of shape and structure. 21. As regards Wittgenstein’s use of the word “sublime”, see PI §§ 89 and 94 as well as the vicinity of our relevant quotations from MSS 152 and 157a. Cf. also a number of passages from the journal kept in Norway while working on the Ur-version of the Philosophical Investigations (MS 183), 152, 158, 164. 22. Thus, according to this reading there is no direct connection between the pneumatically conceived kind of “thought” mentioned in [3] and the “thinking” in sentence [2].

54

purity could have been of any possible interest. For instance, an experience not supported by the logical scaffolding of rules inherent in language would have been of no interest. At the time, the reason for this view had been the pneumatic conception of thought, that is, the conception according to which the core of language contains a scaffolding of rules whose substance is the same as that of our thought: the logical structure of thought is identical with that of language and reality. Here it is important to remember the following two points: First, while the pneumatic conception is indeed criticized in this part of PI (§§ 89ff.) it remains unscathed in the relevant sentences of § 109. Second, it may very well be the case that the criticized pneumatic view was, according to Wittgenstein’s later lights, at the basis not only of his early views but also of some of the ideas characteristic of his philosophy of the early 1930s.

REFERENCES Baker, Gordon and Peter Hacker 1980: An Analytical Commentary on Wittgenstein’s Philosophical Investigations, Oxford: Blackwell, new edition revised by Peter Hacker, 2005, Volume I. Hallett, Garth 1977: A Companion to Wittgenstein’s “Philosophical Investigations”, Ithaca and London: Cornell University Press. Kienzler, Wolfgang 1997: Wittgensteins Wende zu seiner Spätphilosophie, Frankfurt am Main: Suhrkamp. Malcolm, Norman 1984: Ludwig Wittgenstein, A Memoir, Oxford: Oxford University Press, 2nd. edition. Rhees, Rush 1970: Discussions of Wittgenstein, London: Routledge. –– 1982: “Wittgenstein on Language and Ritual”, in B. McGuinness (ed.): Wittgenstein and his Times, Oxford: Blackwell, 69–107. Schulte, Joachim (ed.) 2001, in Zusammenarbeit mit (in collaboration with) Heikki Nyman, Eike von Savigny, and Georg Henrik von Wright: Ludwig Wittgenstein: Philosophische Untersuchungen, Kritisch-genetische Edition (=critical-genetic edition), Frankfurt am Main: Suhrkamp. von Savigny, Eike 1994: Wittgensteins “Philosophische Untersuchungen”: Ein Kommentar für Leser, 2nd edition, vol. 1, Frankfurt am Main: Klostermann. von Wright, Georg Henrik 1982: “The Wittgenstein Papers”, in G. H. von Wright, Wittgenstein, Oxford: Blackwell.

55

This page intentionally left blank

Grazer Philosophische Studien 71 (2005), 57–86.

“A MATHEMATICAL PROOF MUST BE SURVEYABLE” WHAT WITTGENSTEIN MEANT BY THIS AND WHAT IT IMPLIES Felix MÜHLHÖLZER University of Göttingen Summary In Part III of his Remarks on the Foundations of Mathematics Wittgenstein deals with what he calls the surveyability of proofs. By this he means that mathematical proofs can be reproduced with certainty and in the manner in which we reproduce pictures. There are remarkable similarities between Wittgenstein’s view of proofs and Hilbert’s, but Wittgenstein, unlike Hilbert, uses his view mainly in critical intent. He tries to undermine foundational systems in mathematics, like logicist or set theoretic ones, by stressing the unsurveyability of the proofpatterns occurring in them. Wittgenstein presents two main arguments against foundational endeavours of this sort. First, he shows that there are problems with the criteria of identity for the unsurveyable proof-patterns, and second, he points out that by making these patterns surveyable, we rely on concepts and procedures which go beyond the foundational frameworks. When we take these concepts and procedures seriously, mathematics does not appear as a uniform system, but as a mixture of different techniques.

Part III of Wittgenstein’s Remarks on the Foundations of Mathematics (hereafter Remarks), which was written from October 25, 1939 until March 19, 1940, begins with the sentence “‘Ein mathematischer Beweis muß übersichtlich sein’” — “‘A mathematical proof must be surveyable’”1 — which 1. Translations of quotations are my own; but, of course, I have consulted Elizabeth Anscombe’s translation in RFM. There, the first sentence reads: “‘A mathematical proof must be perspicuous’”. The word “perspicuous”, however, suggests understanding (a ‘perspicuous proof ’ as one that ‘can be easily understood’) and this is not what Wittgenstein intends when using the word “übersichtlich” in RFM III. The same is true of the expression “can be taken in”, which is Anscombe’s translation of Wittgenstein’s word “übersehbar” in RFM III § 2 and which, too, may suggest understanding (to take in = to understand). As we will see, Wittgenstein, when applying words like “übersehbar” and “übersichtlich” to mathematical proofs — and also the word “überblickbar” (see MS 122, 43r: “‘Ein Beweis muß überblickbar sein’”) — , uses them

is put in simple quotes. With this Wittgenstein does not signify the utterance of a philosophical opponent, or interlocutor, but he strikes the main subject of his subsequent investigations. At the same time this statement is a kind of self-citation which refers back to RFM I § 154, where Wittgenstein — without quotes! — has written: “Surveyability is part of proof ” (“Zum Beweis gehört Übersichtlichkeit”). However, in RFM I (written in 1937 and 1938, with the last note dated October 10, 1938) this statement stands totally isolated, and only one year later, in RFM III, Wittgenstein really takes it up and devotes an extensive and scrupulous investigation to it.2 1. Surveyability as Reproducibility Initially, the statement “A mathematical proof must be surveyable” sounds like an extremely strong, revisionist and, because of the “must”, breathtakingly dogmatic thesis, reminding one of the days of the Tractatus. Whoever has studied only a bit of mathematics knows that complicated, unsurveyable proofs are the order of the day. It cannot be that the later Wittgenstein, who permanently struggles against the dogmatism of his early philosophy, should deny all of them their character as genuine, respectable proofs. By in a purely formal sense that has nothing to do with understanding, at least not in the sense of ‘mathematical understanding’. Furthermore, it is quite obvious that he treats these words as interchangeable, and in my translations I will use the term “surveyable” for all of them. This is, I think, the best terminological choice in order to convey Wittgenstein’s intentions. 2. I was not able to find statements about the ’surveyability’ (“Übersichtlichkeit”, “Übersehbarkeit”, “Überblickbarkeit”) of mathematical proofs in Wittgensteinian texts before 1937. To be sure, there are earlier allusions to the phenomena of surveyability or nonsurveyability in mathematics, e.g. in BT 579f., but Wittgenstein there does not apply terms like “übersichtlich”, “übersehbar” or “überblickbar” to proofs. However, and quite interestingly, in BT 733f., he uses the word “übersehen” to mark the difference between the finite and the infinite: “Imagine an infinitely long row of trees, and, so that we can inspect them, a path beside them. All right, the path must be endless. But if it is endless, then that means precisely that you can’t walk to the end of it. That is, it does not put me in a position to survey [übersehen] the row. That is to say, the endless path does not have an end ‘infinitely far away’, it has no end” (this is Anthony Kenny’s translation in PG 455.) — Of course, Wittgenstein uses the expression “übersichtliche Darstellung” as early as 1931 in his remarks on Frazer’s Golden Bough (see PO 132), but with this he characterizes his own philosophical method and not mathematical proofs. Much has been written about Wittgenstein’s term “übersichtliche Darstellung” (see for a most recent essay on it Hacker 2004 [or Puhl’s contribution in this volume; insertion by the editor M. Kober]), whereas the Wittgensteinian subject of the ‘Übersichtlichkeit’ of mathematical proofs remained underexposed. The present paper is concerned only with the latter one.

58

the word “surveyable” (“übersichtlich”, “übersehbar”, “überblickbar”) he must mean something — something rather uncustomary — that excludes such absurd consequences, and in RFM III §1 he immediately explains what he means. It can be expressed in four meaning postulates: (S1) The surveyability of a proof consists in its possibility of reproduction. (S2) This reproduction must be an easy task. (S3) We must be able to decide with certainty whether the reproduction produces the same proof. (S4) The reproduction of a proof is of the sort of a reproduction of a picture. In what follows, the word “surveyable” should always be understood in this sense. It is the sense explicitly introduced at the beginning and then presupposed in all of Part III of the Remarks. Let us consider the four meaning postulates in detail, in particular with regard to the question why Wittgenstein wants to give the word “surveyable” precisely this meaning; i.e., why it is important to him to highlight just these four properties of mathematical proofs. So, why is it important to him, as expressed in S1, to lay stress on the reproducibility of proofs? — The reason is that, when one pays attention to the specific sort of reproducibility, the distinction between proofs and experiments becomes particularly clear. Wittgenstein says explicitly that this is the point of his notion of surveyability: “Proof must be surveyable” really means nothing but: a proof is not an experiment. We do not accept the result of a proof because it results once, or because it often results. But we see in the proof the reason for saying that this must be the result. (RFM III § 39)

The issue “proof versus experiment” is a characteristic Wittgensteinian invention, and it pervades the Remarks from beginning to end. It arises from the tension between the following two modes of appearance of mathematical proofs: On the one hand, a proof, when performed by a human agent, is a process in time which, like an experiment, causally produces a certain result;3 on the other hand, a proof can also be considered as something 3. See RFM I § 157: “Is it experimentally settled whether one proposition can be derived from another? — It looks as if it were! For I write down certain sequences of signs, am guided by doing so by certain paradigms […] and of what I get in this procedure I say: it follows.” See

59

non-temporal which shows us, as Wittgenstein says in RFM I §4, “the peculiar inexorability of mathematics”. To Wittgenstein, this inexorability does not lie in the mathematical entities and facts, to which the mathematical signs and sentences refer, but in the way we use the mathematical signs and sentences, and particularly in the way we use mathematical proofs. Wittgenstein’s problem is how the point of view, which puts the peculiar inexorability of mathematics to the fore, can be dissociated from the former, process-oriented view in such a way that one avoids drifting into dubious, metaphysical channels. His notion of surveyability, with its stress on the reproducibility of proofs, serves precisely this purpose. Wittgenstein’s most important observation concerning the difference between proof and experiment is that we regard the identity of an experiment already as fixed when the experimental setup — i.e. the conditions under which the experimental result should appear — is fixed and that we are prepared, then, to accept this or that result.4 In contrast to this, the identity of a proof essentially depends on the proof ’s result itself: the proof would be a different one if it had another result. This fundamental difference is most obvious when one considers the reproduction of an experiment and of a proof. In case of the experiment, we have to reproduce only the aforementioned condition; in case of the proof, we must also reproduce the result. With his notion of reproducibility, Wittgenstein wants to refer to precisely this contrast between the repetition of a proof and the repetition of an experiment. He explicitly says it in RFM III § 55: “To repeat a proof means, not to reproduce the conditions under which a particular result was once obtained, but to repeat every step and the result”. To further clarify this contrast (which certainly is necessary), Wittgenstein therefore starts a thorough investigation of the characteristic way mathematical proofs are reproduced and allocates a special concept to this investigation: the concept of surveyability as introduced in RFM III § 1. also the vivid description in RFM III §69: “We let ourselves unwind and get the result of the calculation.” 4. This is not true of ‘experiments’ done in classrooms or lecture halls, where, when the expected result does not occur, the ‘experimenter’ says: “Something must be wrong with the experimental setup”. These, however, are not genuine experiments but demonstrations which in a sense function like mathematical proofs, because their identity essentially depends on their results. — Furthermore, even in the case of a genuine experiment, if its result deviates from what we take as reasonable scope, we also will call its setup into question. But it seems to me that this doesn’t change the basic correctness of Wittgenstein’s observation that in normal cases the identity of the experiment is regarded as fixed by the experimental setup alone, without being dependent on the experimental result.

60

2. Reproducibility’s Formal Character Meaning postulate S2 says that the reproduction of a proof must be an easy task.5 Wittgenstein’s reflections in Part III of the Remarks show that with this ‘easiness’ he does not have in mind that “we could take all such proofs in at a glance”, as we read in Steiner (1975, 41), or that they should be memorable, as claimed in Frascolla (1994, 142; and repeated in Frascolla 2001, 284). Demands of this sort cannot be intended by Wittgenstein because they obviously contradict our mathematical practice, where long and complicated proofs, which overstrain our memory, not to mention our ability to take them in ‘at a glance’, are nothing out of the ordinary. What S2 excludes are not very long and complicated proofs per se, but proofs that are unsurveyable in the manner in which they, e.g., represent numbers. This would be the case, say, if we represented the natural number n by a string of n strokes, as envisaged in Hilbert’s stroke calculus,6 or as the sum of n ones, such that for very big n it is not an ‘easy task’ — and it may even be a practically impossible task: for human beings anyway, and even for machines! — , to reproduce the sign representing n. A fortiori it is, then, not an easy task (to say the least) to reproduce a proof containing such a sign. Wittgenstein wants to direct our attention to notations of just this sort. It is a trivial and indisputable fact of our mathematical practice that we do not use such notations: that we replace them by feasible ones. This point is deepened in meaning postulate S3, which demands certainty in reproducing a proof. Wittgenstein here, of course, does not 5. When Wittgenstein explains surveyability via reproducibility, he, of course, always thinks of reproductions by human beings, not by gods and also not by machines. He is concerned with the human practice of doing mathematics and not with the practice of super- or non-human beings. Consequently, the reproduction of proofs should be an easy task for us. However, human beings use machines which, then, they can also employ to reproduce proofs, or even to produce proofs, and Wittgenstein’s thoughts concerning the surveyability of proofs should take this possibility into account and investigate its relevance. This, however, does not really happen, and therefore, in what follows, I will ignore the possible aid of machines, despite the fact that the consideration of these aids very often suggests itself. There is a serious gap here in Wittgenstein’s philosophy of mathematics that should be closed by separate investigations. 6. As described in Hilbert 1922 and 1925 — where, to be precise, Hilbert does not use strings of strokes but strings of the sign “1”, or even of the sign “1” combined with the sign “+”, thereby producing packages of the form “1 + 1+ … + 1”. But Hilbert, of course, could also have used simple strokes instead. In this case, according to his calculus, the natural numbers are identified with the signs “~”, “~~”, “~~~”, ... . Our usual numerals in decimal notation then designate these signs: “1” designates “~”, “2” designates “~~”, and so on (such that, when using “1” instead of “~”, Hilbert’s calculus implies that the sign “1” designates itself !).

61

have in mind Cartesian but merely practical certainty; a certainty that is shown in the fact that there are practically never any disputes over the question whether someone really has produced the same proof or not. As remarked in RFM III § 1, this is due, among other things, to the fact that we consider it irrelevant in which handwriting the proof is written, or in which colour, and so on. In other words: the criteria of identity for proofs should be applicable without problems, and these criteria should only involve “what is essential to the proof ” (as Wittgenstein says in RFM III § 1). Everything involving the reproduction of handwritings, colours and similar things, certainly does not belong to that. But what does belong to it? For example, is it essential to Cantor’s diagonal proof exactly how the diagonal is altered such that the new diagonal differs from the former one at every place? Or is it essential to an H-G-proof in analysis exactly which G is chosen to a given H in order to conduct the proof? In both cases there is a great freedom of choice, and the choice which one makes is inessential to the idea of the proof. Do Wittgenstein’s ‘reproductions of a proof ’ allow variations of this sort, which not only concern the form of the proof, as in the case of handwritings etc., but also its idea? The examples discussed by him in RFM III7 clearly show that the answer is no. Wittgenstein’s criteria of identity for proofs, which determine what a correct reproduction of a proof is, are of a purely formal nature, and they do not take into account ‘proof ideas’. In fact, the Wittgensteinian criteria seem to be those presupposed in proof theory, and Wittgenstein’s conception of a proof in RFM III reminds one of the proof theory conceived by Hilbert, where mathematics is considered as a rule-governed ‘formula game’. The formalized proofs occurring in this game are characterized by Hilbert as follows: “[A] formalized proof, like a numeral, is a concrete and surveyable object. It can be communicated from beginning to end.”8 “Surveyability” (“Überblickbarkeit”) is a central notion in Hilbert’s foundational outlook, which is presented in Hilbert’s 7. And also in the much more voluminous manuscripts MS 122 and 117, from which the remarks published as RFM III have been taken. 8. Hilbert 1925, 383; whereby, in Hilbert’s German, the word for “surveyable” is “überblickbar”. This formulation is repeated word for word in Hilbert 1928, 471. The term “formula game” (“Formelspiel”) is Hilbert’s and it occurs several times on page 475 of Hilbert 1928. The first occurrence of the word “überblickbar” in Hilbert’s work may be in a lecture held at the beginning of 1920, where Hilbert expresses the desire that “proof procedures become completely surveyable” (“[…] das Verfahren der Beweisführung vollständig überblickbar wird”). This statement can be found in Sieg 1999, 23f.. Sieg’s paper is a treasure trove for information about the development of Hilbert’s proof-theoretic project.

62

programmatic paper “The new grounding of mathematics” of 1922 in an impressive way: [A]bstract operation with general concept-scopes and contents [as practised by Dedekind and Frege] has proved to be inadequate and uncertain. Instead, as a precondition for the application of logical inferences and for the activation of logical operations, something must already be given in representation [in der Vorstellung]: certain extra-logical discrete objects, which exist intuitively [anschaulich] as immediate experience before all thought. If logical inference is to be certain, then these objects must be capable of being completely surveyed in all their parts, and their presentation, their difference, their succession (like the objects themselves) must exist for us immediately, intuitively [unmittelbar anschaulich], as something that cannot be reduced to something else. Because I take this standpoint, the objects [Gegenstände] of number theory are for me — in direct contrast to Dedekind and Frege — the signs themselves, whose shape [Gestalt] can be generally and certainly recognized by us — independently of space and time, of the special conditions of the production of the sign, and of insignificant differences in the finished product. [Hilbert here adds in a footnote: “In this sense, I call the signs of the same shape ‘the same sign’ for short.”] The solid philosophical attitude that I think is required for the grounding of pure mathematics — as well as for all scientific thought, understanding, and communication — is this: In the beginning was the sign. (Hilbert 1922, paragraph 25)

Wittgenstein knew and discussed some of Hilbert’s foundationalist writings of the 1920s, and I am sure that, despite his harsh criticism of Hilbert’s metamathematics, he was strongly influenced by it.9 And it appears very plausible, then, that the quotation marks of the first sentence of RFM III not only indicate that this sentence expresses the subject of the subsequent investigations, or that it is a self-citation referring back to RFM I §154, but that they also hint at the corresponding view of Hilbert, which Wittgenstein adopts in important respects and then expounds and uses in his own way! 9. The most obvious traces of Wittgenstein’s dealing with Hilbert’s foundationalist program can be found in WVC in its entirety (and especially in Wittgenstein’s numerous remarks on consistency, e.g. where, on December 17, 1930, he even uses the Hilbertian term “formula game”, i.e., “Formelspiel”) and in the Big Typescript, sections 108-111 (BT 530–49; with section 109 entitled: “There is no metamathematics”, BT 539). A thorough investigation of Hilbert’s influence on Wittgenstein, and a systematic comparison of their views of mathematics, would be rewarding. In the following I can give only some hints at what such an investigation might disclose.

63

All the obvious differences notwithstanding, there are in fact important points of agreement between Wittgenstein and Hilbert, the most important one, as just mentioned, consisting in their formalistic outlook10 with its stress on the surveyability of the formulae and especially the proofs. Furthermore, for Hilbert as well as for Wittgenstein our mathematical thinking reveals itself in the rules of the mathematical formula game. In Hilbert’s own words, “this formula game is carried out according to certain definite rules, in which the technique of our thinking is expressed. […] The fundamental idea of my proof theory is none other than to describe the activity of our understanding, to make a protocol of the rules according to which our thinking actually proceeds” (Hilbert 1928, 475; emphasis Hilbert’s). In Wittgenstein’s words (dictated to his students in 1933/34): “We may say that thinking is essentially the activity of operating with symbols” (BB 6); and Wittgenstein says this shortly after having rejected Frege’s ridicule of the formalist conception of mathematics (see BB 4)!11 Years later, in PI §§ 316–362, he carefully discusses also other aspects of what we call “thinking”, but there can be no doubt that the idea of ‘thinking as operating with symbols’ remains of central importance to him, especially in the context of our mathematical thinking. According to Hilbert, concentrating on the mathematical formula game has two important philosophical consequences (besides the desired consequence of promising mathematically rigorous consistency proofs, which lies at the center of Hilbert’s genuinely mathematical program): first, to protect us from the subjectivism of the intuitionism of Brouwer and Weyl 10. Frascolla, therefore, has coined the term “quasi-formalism” for Wittgenstein’s position (see Frascolla 1994, vi and 163). Though I myself would refrain from attributing any form of ‚ism‘ to Wittgenstein, there is without doubt considerable legitimacy in Frascolla’s terminology. When talking about “Hilbert’s outlook”, I am, of course, referring only to Hilbert’s proof theoretical project. As a practicing mathematician, outside proof theory, Hilbert was no ‘formalist’ at all. 11. Hilbert, by the way, shows the philosophical naivety typical of scientists when giving his statement about the ‘technique of our thinking’ the following mentalistic backing: “Thinking, it so happens, parallels speaking and writing” (Hilbert 1928, 475), which is, of course, the antithesis to everything Wittgenstein wants to teach us about ‘mental activities’. In fact, it appears not too far-fetched to assume that Wittgenstein might have been stimulated to develop his own, alternative view about thinking (and understanding, intending, meaning etc.) by just this statement of Hilbert’s. The idea of understanding as being able to operate with symbols is already present in WVC, e.g. in Wittgenstein’s remarks of September 21, 1931, and there it is in fact intimately connected with Wittgenstein’s extensive discussion of Hilbert’s view. — Note that, when the later Wittgenstein considers mathematics as a rule-governed formula game, he should not be understood as having in mind only mechanical rules, i.e. rules expressible by recursive predicates. His notion of a rule is much broader.

64

(see again Hilbert 1928, 475), and second, to restore the certainty that mathematics seemed to have lost in the ‘Grundlagenstreit’. There has been some dispute in the philosophical literature about what Hilbert meant, or should have meant, by the term “certainty” — ”Sicherheit” — used by him in many prominent places (as, e.g., in the long passage from Hilbert 1922 cited above).12 However, one can attain some sort of minimalist position in this dispute by looking at the actual use Hilbert made of this term. It is indeed striking that he uses it only in the context of the foundational debate where the agreement among mathematicians — the pervasive intersubjective agreement that is so characteristic of mathematics as an exact science — broke down. That this is quite an essential aspect of his notion of certainty, becomes particularly obvious in his lecture at the International Congress of Mathematicians in Bologna (Hilbert 1929) where he explained his worries, and at the same time his own achievement, in the following words: [I]f we use contentual axioms as starting points and foundations for the proofs [as it is done in non-formalized mathematics], then mathematics thereby loses the character of absolute certainty [des absolut Sicheren]. With the acceptance of assumptions [Voraussetzungen] we enter the sphere of what is problematic. Indeed, the disagreements among people [Meinungsverschiedenheiten der Menschen] are mostly due to the fact that they proceed from different assumptions. (Hilbert 1929, 228; emphasis mine) My lecture today shows how many problems still await solution. However, in a general and basic sense, not even the faintest trace of obscurity is possible any longer: Every basic question can be answered in a mathematically precise and univocal manner on the basis of the proof theory I have sketched. Now the theorems at issue can in part be proved, in an absolutely certain [absolut sichere] and purely mathematical fashion, with the help of the present results, and they have therefore been removed from the dispute [dem Streite entzogen]. Whoever wants to confute me must show me, as has always been customary 12. Detlefsen (1986, 64–71) conveys something of the flair of this dispute. In the Englishspeaking world this dispute is aggravated by the fact that Hilbert’s word “Sicherheit” — and Hilbert always uses this one word! — is sometimes translated as “certainty” (e.g. in the translations of Hilbert 1922 and 1929 by Ewald and Mancosu) and sometimes as “reliability” (e.g. in the translations of Hilbert 1925 and 1928 by Bauer-Mengelberg and Føllesdal, and also in the, otherwise excellent, discussion of the development of Hilbert’s thoughts in Sieg 1999). “Reliability”, however, is not really a translation but already an interpretation! It seems advisable to use the relatively neutral word “certainty” for Hilbert’s “Sicherheit” and to look for an interpretation of it afterwards.

65

in mathematics and will continue to be so [wie es in der Mathematik bisher stets üblich war und bleiben wird], exactly where my supposed error lies. (Hilbert 1929, 233; emphases mine)

The Hilbertian ‘certainty’, as expressed in these passages, can be considered to be the certainty of scientists who simply agree about their fundamentals — and beyond that there is no question whatsoever about the ‘ground’ of this agreement. The agreement itself is enough for the science to function, and ‘deeper’ foundations are not needed. I think that this minimalist conception of ‘certainty’ captures an important aspect of what Hilbert meant by this term, an aspect that surely would be congenial to Wittgenstein. And the same is true of Hilbert’s motive for rescuing mathematics from the subjectivist moods of the intuitionists.13 In both respects we have considerable agreement between Hilbert’s and Wittgenstein’s outlook. There remain deep differences, of course, the most important one concerning Hilbert’s striving for certainty by means of consistency proofs. Wittgenstein notoriously delivered only polemics against this Hilbertian fixation upon consistency. And Wittgenstein from the outset rejected the Hilbertian project of metamathematics as a mathematical one. His own way of doing ‘metamathematics’, i.e. of talking about mathematics, is drafted not as an exact theoretical science, as Hilbert conceived it, but as a purely descriptive philosophical endeavour that aims at perspicuous representations of our actual use of signs. A particularly important aspect of this Wittgensteinian approach is that he in no way sees, as Hilbert apparently does,14 the mathematical formula game as a uniform one. On the contrary, Wittgenstein is notorious for sayings like “mathematics is a multicoloured mixture of techniques of proof ” (RFM III § 46) and “I want to explain the colourfulness of mathematics” (RFM III § 48).15 13. See LFM 237: “Intuitionism […] requires that we have an intuition at each step in calculation, at each application of a rule […]. Actually […] you simply do a certain thing. It is a question of a certain practice. Intutionism is all bosh — entirely.” 14. E.g. when he writes: “This formula game enables us to express the entire thought-content of the science of mathematics in a uniform manner” (Hilbert 1928, 475). 15. In Wittgenstein’s German: „Die Mathematik is ein buntes Gemisch von Beweistechniken“; „Ich will die Buntheit der Mathematik erklären“. I deviate from Anscombe’s (much quoted!) translation, which uses the word “motley” in these two sentences (“mathematics is a motley of techniques of proof ”; “I want to give an account of the motley of mathematics”), because this word has a derogatory tinge which strikes Wittgenstein’s true intentions in the face; and also because the use of this word in both sentences misrepresents the actual relation between them as formulated by Wittgenstein himself. (Furthermore, Anscombe’s translation of

66

Furthermore, Hilbert, despite his warnings of contentual thinking in mathematics (as in the passage quoted above), nevertheless develops his own brand of contentual mathematics, his ‘finitary’ mathematics, which should yield the desired consistency proofs. In this sort of mathematics, numbers are identified with signs and mathematical statements are about constellations of signs — again things that Wittgenstein strongly rejects. In view of this difference, one might suspect that with his first sentence of RFM III Wittgenstein does not express his own position, but the Hilbertian one, and that his reflections in RFM III do not refer to mathematical proofs as they occur in our ordinary mathematical practice, but only to the rigorously formalized proofs as presupposed in Hilbert’s proof theory. This suspicion, however, would be unfounded. It is true that in RFM III Wittgenstein mainly (but not exclusively!) considers formalized proofs, such as those found, e.g., in Whitehead/Russell’s Principia Mathematica, but it is the demand for surveyability with respect to all proofs which he uses as the standard against which the proofs of Principia Mathematica should be measured — with the result that, as we will see, these proofs turn out to be problematic. It is precisely the concentration on formalized proofs as they occur in foundationalist endeavours, like the logicist project of Whitehead and Russell, that prompts Wittgenstein to make the formal aspects of proofs the focus of his attention, and it is he himself who thinks that these formal aspects are essential. There can be no doubt, I think, that according to Wittgenstein the demand for surveyability, as some sort of minimal condition for proofs, should be upheld with respect to any proof that deserves this name. A further, and decisive, contrast to Hilbert consists in the way in which Wittgenstein understands and uses the term “surveyable”. It must be said that Hilbert did not take all too great pains to explain this term; whereas Wittgenstein gives the elaborate explanation which I discuss in this paper. Wittgenstein struggles with his notion of surveyability page after page, and he uses it, as I have just said, with a decidedly anti-foundationalist objective. For Hilbert, on the other hand, it was precisely his foundationalist motive that led him to lay stress on the surveyability — on what he took to be the surveyability — of proofs!16 So, in the end, despite the similarities in their ‘formalistic outlook’, the differences between Hilbert and Wittgenstein turn out to be profound. Wittgenstein’s word “erklären” by “to give an account” is no longer only a translation but really an interpretation — and an inadequate one, as we will see.) 16. This motive is very instructively displayed in Sieg 1999, 23–32.

67

3. Proofs as Pictures With his fourth meaning postulate, S4, which declares the reproduction of a proof to be of the sort of the reproduction of a picture, Wittgenstein alludes to a beautiful thought which he has already developed in Part I (and which he will develop further in Part VI) of the Remarks: that the real, temporal process of proving a mathematical theorem may very well be comparable to an experiment, but that the proof itself rather resembles the picture of such an experiment, in which the experiment is frozen, as it were, into something non-temporal. It is, above all, this aspect of visibility that justifies the choice of the term “surveyable” (that is, of the terms “über-sicht-lich”, “über-seh-bar”, “über-blick-bar”). To compare proofs with pictures is Wittgenstein’s way of doing justice to the non-temporality of proofs without falling prey to platonistic myths. The production and use of pictures is a familiar practice within our lifeworld, and its similarity with our production and use of proofs should shed light on our practice of proving theorems. Wittgenstein gives the following explanation of this idea: When I say “the proof is a picture” — it can be thought of as a cinematographic picture. We construct the proof once and for all. (RFM III § 22)

The cinematographic picture, which Wittgenstein has in mind here, records, and thereby fixes, a concrete process of proof-construction on a film, and in this way the construction is made ‘once and for all’, because this film now can be run again and again. Seen in this manner, the film is the proof; not in the sense of a temporal process, but in the sense of something non-temporal. This is one of the numerous down-to-earth methods of Wittgenstein, with the help of which he tries to grasp the peculiar non-temporality to be found in mathematics. The reproduction of such a picturelike proof, then, is a copying. It is a copying, however, that—now unlike the normal copying of a film—gets along without any technical aids like measurements and similar means, which are characteristic of experiments (see RFM III § 10). For, what we want to copy is that which is mathematical and non-temporal and which lies, so to speak, beyond anything empirical, process-like and causal. Thus, the proofreproduction itself should be conceived not in causal categories, but in quite different ones. Wittgenstein’s problem is, then, to find these ‘different categories’ and to present them in a way that is philosophically satisfying.

68

In precisely this opposition between the causal sphere of temporal processes, on the one hand, and the mathematical sphere of non-temporality, on the other, lies the key to Wittgenstein’s reflections concerning “surveyability”. In Part IV of the Remarks, which were written more than two years after Part III, he expresses the quintessence of these reflections as follows: When I wrote “the proof must be surveyable” that meant: causality plays no part in the proof. Or, also: the proof must be capable of being reproduced by mere copying. (RFM IV § 41)

With this ‘mere copying’ Wittgenstein has in mind an activity which we have completely under control; whereas causal processes are typically beyond our control and in many cases take their course, as Wittgenstein often says, ‘underground’ (“unterirdisch”).17 With this, however, surveyability is only negatively characterized: “causality plays no part”. The corresponding positive characterization, then, consists in Wittgenstein’s comparison of proofs with pictures which he, furthermore, underpins with the philosophically charged notion of intuition, i.e., “Anschauung”. In RFM III § 42 the proof-process is called “intuitive” (“anschaulich”): “The proof must be an intuitive process [ein anschaulicher Vorgang]”. Then, the picture of such a process must be intuitive as well, and this is expressed in unpublished passages in the manuscript MS 122, from which the first 58 sections of RFM III are drawn.18 There, Wittgenstein says things like the following: The [proof ] must be intuitive [anschaulich sein]. (Remark about the grammar of the word “proof ”.) (MS 122, 74r) 17. E.g. see PR 212, where Wittgenstein’s German expression “unterirdische Verbindung” is translated as “subterranean connection”; and RFM I Appendix II, § 3, where Wittgenstein’s “durch einen unterirdischen Gang” is translated as “through an underground passage”. — The idea of a control can also be found in Hilbert; see the quotation from a Hilbertian lecture in Sieg 1999, 24 (in Sieg’s translation): “For this purpose [namely: to secure the foundations of mathematics] it seems appropriate to connect the mathematical constructions to what can be concretely exhibited and to interpret the mathematical inference methods in such a way that one stays always within the domain of what can be checked [daß man immer im Bereiche des Kontrollierbaren bleibt]”. 18. One could write a separate essay about the peculiar editorial policy that led to the highly fragmentary character of RFM. I will refrain from commenting upon this problem here. It is quite clear, however, that a sufficient understanding of the remarks presented in RFM can only be reached by a careful study of all of the material in the manuscripts, from which RFM has been extracted.

69

“The proof must be surveyable” — is to mean (approximately): the identity of a transformation of a proof [i.e., the sign-transformation that makes up the proof (F. M.)] is not to be established by an experiment but (immediately) by intuition [(unmittelbar) durch die Anschauung]. (MS 122, 78v–79r) The proof must be intuitive [anschaulich sein]: if we are no longer convinced by what we see, then the proof has lost its force. No matter, whether it is construed according to the ‘logical’ schema of Russell or otherwise. (MS 122, 80v)

The terms “intuitive” and “intuition” (“anschaulich” and “Anschauung”) are used here in the non-ambitious sense of ‘plain to view’19 and in contrast to something like ‘demonstrable by a physical or other empirical investigation’ (see RFM III § 39). What Wittgenstein has in mind is in no way a philosophical conception of intuition. Obviously, he plays with philosophically ambitious associations, but his bracketed hint: “Remark about the grammar of the word ‘proof ’” (MS 122, 74r), calls us to order and expressly warns against placing his thought into a Kantian or otherwise theoretically charged project. In order to clarify the philosophical problems concerning mathematical proofs, Wittgenstein merely wants to bring before our eyes our ordinary use of the word “proof ”. The simple observation which he has in mind in the present context concerns the fact that, if someone refuses to accept a mathematical statement even if she has got to know its complete, detailed proof, she cannot justify her refusal by pointing out that the proof does not show everything that is relevant, that there might be, as it were, certain things going on underground that still have to be settled. For, in the complete proof everything that is relevant is also plain to view — and if someone still refuses to accept it, we treat her as a person who does not understand what mathematics and, in particular, what a mathematical proof is. All our reasons to accept the correctness and conclusiveness of a proof only refer to things which we can see in the proof.20 The proof, for Wittgenstein, is a road that we survey from beginning to end. This is, of course, again reminiscent of Hilbert; compare his statement 19. And such is, in fact, Anscombe’s translation of “anschaulich” in RFM III § 42. However, this is not a legitimate translation but already an interpretation of Wittgenstein’s term “anschaulich”. When using this philosophically charged term in the present context, Wittgenstein certainly had his ulterior motives, the existence of which should remain visible also in the translation. 20. “When we say in a proof: ‘This must come out’ — then this is not for reasons that we do not see” (RFM III §39).

70

cited above, that “a formalized proof, like a numeral, is a concrete and surveyable object. It can be communicated from beginning to end” (Hilbert 1925, 383). And Hilbert, like Wittgenstein, uses the term “intuitive” (“anschaulich”) when characterizing the surveyable proofs: “A proof is a figure, which as such must be given to us intuitively”.21 It is true that, unlike Wittgenstein, Hilbert has a tendency, when using terms like “intuitive” and “intuition”, to align himself with the Kantian tradition;22 but one can doubt whether he is really justified in doing so, because he in no way uses “intuitive” as a theoretical term, as it is certainly meant in Kant’s philosophy, but rather as a term belonging to our familiar lifeworld.23 Be that as it may, it is again not improbable that, as already mentioned with respect to the term “surveyable”, Wittgenstein was inspired by Hilbert to also lay stress on the intuitiveness of proofs, and again, in Wittgenstein’s work, this thought is then given its own, characteristic Wittgensteinian stamp. It is given this stamp by Wittgenstein’s comparison of proofs with pictures. And in a further step, Wittgenstein then concentrates on the specific use of pictures (“Bilder”) as paradigms (“Vorbilder”) — where we can, once more, orient ourselves towards our familiar practice, belonging to our normal lifeworld, of acting according to paradigms. As already cited, Wittgenstein writes: “We construct the proof once and for all” (RFM III § 22). However, this can in a sense also be said of experiments: We carry out our experiments once and for all, that is, we do not think it necessary 21. This is my own translation of Hilbert’s statement “Ein Beweis ist eine Figur, die uns als solche anschaulich vorliegen muß” in Hilbert 1922, which in the translation of Ewald reads as follows: “A proof is a figure, which must be able to view as such” (Hilbert 1922, paragraph 57). And this is Bauer-Mengelber’s translation of the same Hilbertian formulation in Hilbert 1925 (which merely differs in the missing emphasis and the replacement of “Beweis” by “mathematischer Beweis”): “A mathematical proof is an array that must be given as such to our perceptual intuition”. The spectrum of different translations of Hilbert’s German term “anschaulich” turns out to be even wider when one further takes a look at the following places: Hilbert 1922, paragraph 25 (as quoted above); Hilbert 1925: 376, 381, 386 and 392; Hilbert 1928: 464ff.; Hilbert 1929: 323. In all these places, Hilbert uses the one term “anschaulich”, and I think it is — as already mentioned in the corresponding, Wittgensteinian case — advisable to translate this term always in the most direct and verbatim way by the English term “intuitive”. The interpretation of this term should then be a different task, uncoupled from questions about the correct translation. 22. See Majer 1992, 267–270, for material concerning this issue. 23. An interpretation along these lines is hinted at in Weyl 1946, 604: „Hilbert bases his mathematics on the practical manipulation of concrete symbols rather than on some ‚pure consciousness‘ and its data“, and in Weyl 1949, 342.

71

to repeat them again and again. We trust Nature, so to speak, that further repetitions would only confirm what we once observed. So, in this sense, we certainly perform also experiments, and not only proofs, ‘once and for all’. But this is not the sense in which we perform — or construct — proofs once and for all. In the case of a proof, when we have constructed it and accept it, we take it as a paradigm, and thereby as a standard, of what is right or wrong in mathematics. This has nothing at all to do with our ‘trust in Nature’, but rather with the specific way we use the proof in the future: we use it like a picture taken as paradigm. In case of a proof consisting of a simple calculation, for example, which shows that a + b equals c, we construct a paradigm of the equality of the numbers a + b and c. In a passage in the Nachlass Wittgenstein expresses this as follows: “But why do you trust this calculation that it really yields the same number?” — I do not trust it. This is what I call now ‘the same number’. (MS 122, 23v)24

That is, this calculation is now a paradigm for ‘the same number’ in this case. The same is true of the characteristic sort of yielding (“Ergeben”) in mathematics, which is not causal and process-like, but non-temporal; for example, when we say “200 and 200 added together yield 400”. What the characteristic mathematical yielding actually is, can only be shown by the corresponding proof, which serves as a picture and as a paradigm of what can be called “yielding” in the situation at issue. There are no other paradigms for the characteristically mathematical sort of yielding. Therefore, Wittgenstein says in RFM III § 24: “The proof must [!] be our paradigm, our picture, of how these operations have a result.” I do not want to say more here about the function of proofs as paradigms. Wittgenstein struggles with this subject at many places (he is unsure, e.g., about what are precisely all the different things for what proofs might be paradigms), not only in the published text, but also in the unpublished material of the Nachlass which is relevant to RFM III, and his struggle is too complicated and too dissipated to be dealt with in the present paper. However, it is clear that with anything Wittgenstein says about mathematical proofs, and about their specific use and function, he does not want to say more than what belongs to our common, indisputable practice of dealing with them. And this is true in particular with respect to his statement “A mathematical proof must be surveyable”, 24. “»Aber warum vertraust Du dieser Rechnung, daß sie Dir wirklich die gleiche Anzahl liefert?« — Ich vertraue ihr nicht. Das ist, was ich jetzt ›gleiche Anzahl‹ nenne.”

72

meant by him in accordance with the explications S1 – S4. That proofs must be surveyable in this sense, should be seen, then, as one of those theses of which Wittgenstein says in the Philosophical Investigations §128, that “it would never be possible to debate them, because everyone would agree to them”. 4. The Non-Surveyability of Proofs in Foundational Systems Let us consider now the consequences that can be drawn from Wittgenstein’s observations about proofs and their surveyability. In RFM III he mainly draws consequences with regard to foundationalist programs in mathematics, whereby it is, above all, the logicist program of Russell, with Frege’s logicist ideas in the background, which he has in mind. But his reflections are equally relevant to the set-theoretic foundations of mathematics that are customary today. In what follows, I will call mathematicians, who think that mathematics is in need of foundations of this sort, “foundationalists”. The foundational systems they favour typically make use of only very few primitive non-logical symbols (e.g. ZermeloFraenkel set theory gets along with the element-sign “” alone), such that their expressions, and correspondingly the proofs, very soon become unsurveyable — and this, indeed, in Wittgenstein’s sense of “unsurveyable”: their reproduction becomes uncertain or downright impossible. In the foundational system of Nicholas Bourbaki, e.g., the expression for the number 1 already consists of 4 523 659 424 929 primitive symbols (not counting the further 1 179 618 517 981 so-called links between these symbols which Bourbaki needs to disambiguate the whole expression).25 So, proofs on the foundational level, where only the primitive symbols are used,26 normally are unsurveyable; hence, according to Wittgenstein, they are in effect not proofs, for: ‘A mathematical proof must be surveyable’. However, it is characteristic of foundational systems that we immediately move away from the foundational level by means of definitions, and 25. These beautiful numbers have been calculated by Mathias 2002. In view of such dimensions, Wittgenstein’s term “unsurveyable” turns out to be an euphemism. 26. In what follows, I always use the term “foundational level” to mean the level where the expressions of the foundational system only contain primitive symbols and have not yet been made surveyable by abbreviating definitions. Bourbaki calls this most basic and most elementary level “the horizon” (see Guedj 1985). It is what we see when we look, so to speak, ‘through’ the abbreviated expressions to the form they assume when all the abbreviations have been retracted.

73

the proofs thus can be made surveyable. In order to have something simpler and less eerie than the Bourbakian 1 before our eyes, let us consider numerals of big numbers (which also occur, of course, in proofs), say: of the number 100 000 (or of a bigger number, if 100 000 does not seem big enough), and let us consider Peano arithmetic, where this number is represented by the — unsurveyable — expression “1 + 1+ … +1” (with hundred-thousand 1’s), which can be expressed, then, in decimal notation and thereby made surveyable (as I just did by using the sign “100 000”). However, procedures of this sort do not seem to be of any help to the foundationalist, for Wittgenstein, quite unmoved, states: “If, by a change of notation, you make an unsurveyable proof-pattern surveyable, then you are creating a proof, where there was none before” (RFM III § 2).27 Thus, his result remains valid: The foundational level contains almost no proofs. How could it be, then, foundational for mathematics! A foundationalist, of course, can easily defend himself against this brisk attack, and Wittgenstein’s reflections in Part III of the Remarks — which often are of agonising slowness — do not seek quick argumentative triumphs at all. On the contrary, their aim is to scrutinise and evaluate all the resources which a foundationalist might have at his disposal when explicating and defending his position. Wittgenstein is anxious to avoid any dogmatism, and therefore he presents the statement just quoted — “If, by a change of notation, you make an unsurveyable proof-pattern surveyable, then you are creating a proof, where there was none before” — with the preceding clause: “I want to say”. Such clauses pervade his whole text (and even more so the unpublished material to be found in the Nachlass), and they document his effort to do as much justice to the foundationalist ideas as he can afford. It is precisely this scrupulous attitude which makes his reflections valuable. He has to cope with two rejoinders the foundationalist will put forward against his attack of RFM III § 2: the basicality rejoinder and the theoreticity rejoinder (as I will call them). The basicality rejoinder consists of the impatient reaction typical for mathematicians when they are confronted with the practical limitations of human beings: Although we humans obviously have difficulties (to say the least) in reproducing proofs at the foundational 27. “Wenn man eine nicht übersehbare Beweisfigur durch Veränderung der Notation übersehbar macht, dann schafft man erst einen Beweis, wo früher keiner war.” Anscombe’s translation of this sentence — “if you have a proof-pattern that cannot be taken in, and by a change in notation you turn it into one that can, then you are producing a proof, where there was none before” — appears to me rather off the mark.

74

level, this, according to the foundationalist, is only a trivial, contingent fact without any basic relevance to mathematics as such.28 And in accordance with this attitude, also the definitions — which create real (and not only fancied) surveyability, where there was none before — are seen as nothing but abbreviating devices that serve our convenience but do not have any serious function beyond that.29 Such is the basicality rejoinder, which is presented by Wittgenstein himself straight away in RFM III § 2 (“The assumption is that the definitions serve merely to abbreviate the expressions for the convenience of the calculator [or theorem-prover (F.M.)]”). The theoreticity rejoinder says that proofs on the foundational level should only be thought of as theoretically postulated entities.30 The existence of such a proof can be proved, then, within proof theory, and there this existence can in fact be proved in a surveyable way! But from the standpoint of proof theory, it does, strictly speaking, not make any sense at all to demand of those proofs, which are their theoretical objects, that they should be surveyable, because nobody really wants to see these proofs 28. Within their ‘finitist mathematics’, Hilbert and Bernays have recourse to the notion of a thought-experiment in order to transcend our factual limitations: “The characteristic trait of this methodological standpoint [namely: the standpoint of finitism] is that the reflections are pursued in the form of thought-experiments with objects which are presumed as concretely given” (Hilbert/Bernays 1968, 20; see also p. 32). 29. The foundationalist literature abounds with claims of this sort. See, e.g., Whitehead/ Russell 1925, 11: “Theoretically, it is unnecessary ever to give a definition: we might always use the definiens instead, and thus wholly dispense with the definiendum. […] Practically, of course, if we introduced no definitions, our formulae would very soon become so lengthy as to be unmanageable; but theoretically, all definitions are superfluous.” This point of view gets its perhaps funniest expression in the following words of Alonzo Church in his book Introduction to Mathematical Logic: “They [= the definitions] are concessions in practice to the shortness of human life and patience, such as in theory we disdain to make. The reader is asked, whenever we write an abbreviation of a wff [= well-formed formula], to pretend that the wff has been written in full and to understand us accordingly. […] Indeed we must actually write wffs in full whenever ambiguity or unclearness might result from abbreviating. And if any one finds it a defect that devices of abbreviation […] are resorted to at all, he is invited to rewrite this entire book without use of abbreviations, a lengthy but purely mechanical task” (Church 1956, 75ff.). As we shall see, this is close to being the complete antithesis to what Wittgenstein wants to teach us! 30. This rejoinder, too, is formulated by Wittgenstein himself, albeit only in passages of the unpublished Nachlass and, moreover, only in a rather indirect way. But the allusions to it are nevertheless sufficiently discernible; for instance, when Wittgenstein writes in MS 122, 77v: “The proof is that which convinces us — hence, not that, of which we think that it would convince us if we could survey it. Or: There is nothing which, theoretically, had to be the proof. Because nothing has — so to speak — the duty to be the proof. ‘Such would be a proof if I could survey it’ — what makes you so sure of this? — A proof?”

75

and to reproduce them. Our theoretical knowledge of their existence is enough.31 The philosophical problem, with which Wittgenstein struggles here, consists in the fact that, on the one hand, these rejoinders are, or might be, very characteristic of the thinking of many mathematicians and are supported by their authority, such that we might be disposed to simply put up with them (with the attitude: “That’s the way these people just think”). On the other hand, this nonchalant way of thinking is in sharp contrast to the obvious fact that in our real mathematical practice we only handle proofs that are (really) surveyable, and despite of all sympathy we might have with the metatheoretic statements of mathematicians, it sounds rather overdone to rate this fact as of no interest whatsoever for our understanding of mathematics. Considered soberly, it appears very plausible that the nonchalant thinking of the mathematicians must somehow be rectified. However, it is initially unclear in which way this should precisely be done. This is Wittgenstein’s problem! It has the form “I don’t know my way about”, which Wittgenstein in PI § 123 designates as characteristic for philosophical problems. According to his philosophical agenda, he wants to resolve it, then, by suitable reminders and descriptions of our familiar mathematical practice. 5. The Significance of Foundational Systems Let us start with the basicality rejoinder. Its presumption that the definitions, with which we make our proofs surveyable, merely serve our convenience, is certainly not to be taken seriously. Even if one does not 31. The proof theorists themselves, however, often shy away from adopting this point of view. Hilbert since 1922 repeated, almost in the manner of a prayer-wheel, statements with the following message (see notes 8 and 21 above): “a formalized proof, like a numeral, is a concrete and surveyable object. It can be communicated from beginning to end” (Hilbert 1925, 383); as if this were an adequate record of the practice of mathematicians who actually do proof theory. In reality, these mathematicians deal with the formalized proofs as theoretical objects which no one wants to survey and communicate from beginning to end. Hilbert has a tendency to simply forget this elementary truth. In a likewise forgetful mood, the proof theorist Wolfram Pohlers, in a text written in 1998, says: “A proof, of course, first of all has the function of convincing the research mathematician of the correctness of his theorem. In this form it serves as a means to find the truth. […] Furthermore, a proof should also convince other mathematicians of the correctness of the own trains of thought. Here we see the function of the communication of truth” (Pohlers 1998, 475). But the formalized proofs, as referred to in proof theory, in no way appear in these functions!

76

have sympathy for Wittgenstein’s inclination to deny unsurveyable symbol-patterns the name “proof ”, the Wittgensteinian distinction between surveyable and nonsurveyable patterns is weighty enough to consider the talk of ‘mere convenience’ totally inadequate. And this all the more so, when one takes into accout that in most cases it is not even possible to write down the unsurveyable proofs (think of a ‘proof ’ with Bourbaki’s one!). When, with the help of suitable definitions, we produce surveyable proofs, in most cases we do not replace something inconvenient by something convenient, but rather something fictitious — or only theoretically postulated — by something real. The talk of ‘convenience’ is, in the present context, certainly beside the point and should be simply understood as the expression of a certain arrogance typical for people working with formal methods. The foundationalist can present himself more convincing by dropping this convenience issue, which is only a side issue anyway, and by explicitly bringing his actual foundationalist motives into play. If he is very ambitious, it is natural for him to say that the foundational system reveals the essence of the mathematical domain: the essence of numbers, of addition, multiplication, and so on. In Wittgenstein’s Lectures on the Foundations of Mathematics, given in the spring of 1939, this point of view was expressed by Alan Turing, who actively took part in these lectures, as follows: “Russell’s definitions show us the point of having these ideas of addition and finite cardinals and so on” (LFM 262).32 In a somewhat less ambitious mood, the foundationalist may perhaps wish to avoid talking about the ‘essence’, or the ‘point’, of addition (and so on), but he will insist that Russell, or the set theorist, teaches us his way of adding — or of adding in his way —, which we very well understand and which we feel to be instructive. This foundationalist motive, be it the more or the less ambitious one, need not necessarily lead to a rejection of the Wittgensteinian view that in the transition from the non-surveyable, foundational level to a surveyable level we create proofs (i.e. proofs that really deserve this name), where there have been none before. But the foundationalist can still claim that the relation of the surveyable to the foundational level remains effective in such a way, that the foundational level is important for our understanding of what happens at the surveyable one. Wittgenstein thinks, as we shall see, that this point of view is in danger 32. Wittgenstein replied to Turing’s statement as follows: “Yes — and it is just what I want to deny” (LFM 262). Wittgenstein’s remarks in RFM III, however, written at the end of 1939, can be read as showing a more cautious and less dogmatic attitude.

77

of underrating those mathematical ideas and concepts that are operative when we move from the foundational to the surveyable levels. At first sight, however, the foundationalist can easily dispel this misgiving by referring to the fact that it is exactly one of the essential aims of a foundational system to simulate, so to speak, all of our familiar mathematics by means of suitable definitions. And these simulations, of course, include our common devices for making proofs surveyable, such as our decimal notation, and similar ones. This is, in a sense, just the point of a foundational system, and that the actual systems, which are on the market, are really able to simulate our familiar mathematics, is certainly a remarkable and important fact. Nevertheless, despite this indubitable fact,33 Wittgenstein wants to show that much of what happens on the surveyable levels is, as it were, external to the foundational system; i.e., that it relies on concepts and ideas which, in cases in which they have not yet been thought by us, are also not already been ‘anticipated’ in any way by the foundational system. Let us call this “Wittgenstein’s externality objection”. To develop this objection, i.e., to show in precisely what sense these new concepts and ideas, which are used in making our proofs surveyable, are ‘external’ to the foundational system, is perhaps the deepest problem with which Wittgenstein struggles in Part III of the Remarks. 6. The Value of Making Proofs Surveyable Wittgenstein deals with the simulational potential of foundational systems in two completely different ways. First, he considers the criteria of identity for proofs on the foundational level, and he argues that these criteria, in order to be adequate for mathematics, depend on conceptual tools which are at our disposal only at the higher, simulating levels. But this means that the foundation of the simulations is essentially dependent on the simulations already at hand — which obviously casts doubt on the idea of a ‘foundation’. Second, Wittgenstein develops his externality objection by 33. A fact in which Wittgenstein was, of course, well versed and which he presupposes in all of RFM III (as it is obvious especially in §§ 13 and 45-49; see also LFM 159f. and 259). Astonishingly, Mark Steiner, in his discussion of RFM III (in chapter 1 of Steiner 1975), presumes that Wittgenstein was ignorant of this fact and ‘refutes’ Wittgenstein by long-windedly showing how our familiar decimal notation can be simulated in a system like Russell’s. If this were an adequate reply to Wittgenstein, then Wittgenstein certainly would not deserve any attention as a serious philosopher.

78

discussing numerous pertinent mathematical examples. Let us consider these two argumentative strategies in detail. To ask for the criteria of identity for entities of different sorts is a characteristic Wittgensteinian move.34 But it is Frege who is famous for having raised this question with regard to numbers and who erected much of his foundational system on precisely such a criterion — with the effect, that the so-called ‘neo-logicism’ of Bob Hale and Crispin Wright, which is intensively discussed today, circles around just this aspect of Frege’s thinking.35 However, lofty thoughts of this sort surely cannot be expected from Wittgenstein! Instead of considering criteria of identity for numbers, he is interested in those of numerals, and instead of the thoughts expressed by proofs, he considers the proofs as such, as they present themselves in the form of patterns of symbols, and he asks for their criteria of identity. In this respect, Wittgenstein again is in the same boat with Hilbert! Notoriously, Hilbert was wary of Frege’s loftiness, because he diagnosed in it an all too loose, and in its consequence disastrous, dealing with the infinite. Hilbert’s cure, then, was to turn his eyes to the symbols themselves and their finite patterns.36 Wittgenstein, of course, did not share Hilbert’s obsession with the disastrous consequence of Frege’s thinking, namely, the Russell antinomy; he had his own view of what had gone wrong in Frege’s thinking (which need not interest us here). But he proposed the same cure as Hilbert: Let us look at the symbols and the way we use them.37 It is natural, then, to look at those patterns of symbols that make up 34. See the useful information in Glock 1996, 168f. and 312–14. In the present context, Glock’s reference to BB 61f., is particularly important, for we find there that Wittgenstein, after having discussed criteria for the identity of a person, adds in brackets: “This kind of consideration is of importance in the philosophy of mathematics. Consider the use of the words ‘proof ’, ‘formula’, and others.” Many passages in RFM III can be read as scrupulous elaborations of this short suggestion. 35. See McBride 2003, for a useful survey of Hale and Wright’s program. The Fregean criterion just mentioned is, of course, the so-called Hume’s principle, which states that the number which belongs to a concept F is the same as the number which belongs a concept G if and only if there exists a bijection from the extension of F onto the extension of G. 36. See Hilbert 1925, 392: “We gain a conviction that runs counter to the earlier endeavours of Frege and Dedekind, the conviction that, if scientific knowledge is to be possible, certain intuitive conceptions and insights are indispensable; logic alone does not suffice. The right to operate with the infinite can be secured only be means of the finite.” 37. As already mentioned in footnote 2, it is interesting — and may in fact be a further indication of Hilbert’s influence on Wittgenstein — that when Wittgenstein first used the term “übersehen” (“survey”) in the context of mathematics (in BT 733f.), he used it to mark the difference between the finite and the infinite! Furthermore, it is a Wittgensteinian leitmotif in RFM III that, as Hilbert says (see the foregoing note), “logic alone does not suffice”.

79

proofs and to clarify the criteria of identity which we accept, or presuppose, in their case. In RFM III, Wittgenstein primarily concentrates on unsurveyable proof-patterns at the ground level of a foundational system, and he makes the thought-experiment that two such patterns may be given to us at this level and that we ask ourselves whether they are the same proof. Such patterns will contain symbolic strings like “yuz …”, or “1 + 1 + 1 + … + 1”, with huge numbers of quantifiers, variables and 1’s; and because of the unsurveyability of these strings, we can establish the numerical identity of these sets of quantifiers, variables and 1s — which is necessary to verify the identity of two patterns of this sort — only by counting the quantifiers, variables and 1s with the help of our normal decimal notation, or by using similar procedures to make the pattern surveyable. But this means that, in order to establish the identity of proofs at the foundational level, the procedures of our normal counting, or similar procedures, are necessary. However, these procedures do have their place only at the much higher levels of the system which simulate our normal procedures. Consequently, in this respect the foundational level cannot be considered self-sufficient. In other words: the concept of proof does not seem to be sufficiently determined at the foundational level; higher levels are needed for this. In this sense, then, the higher levels are not based on the foundational level. Certainly, in a foundational system, the concepts, sentences and proofs on the higher levels should be constituted by the concepts, sentences and proofs on the foundational level. But if the identity of the foundational sentences and proofs are dependent on what happens on the higher levels, this idea of a ‘constitution’ is compromised. An argument of this sort is developed in RFM III §§ 4–21 (in much more detail than I can convey here). In these sections, there is also a discussion of eventual empirical criteria of identity that one may resort to in cases of symbolic unsurveyability; e.g. by drawing lines or stretching threads in order to establish a 1-1 correlation between signs (see RFM III § 10). At the back of our minds we might have the idea, here, that such empirical procedures and states of affairs should compensate for the unsurveyability of proofs which restricts our capability to actually managing them. But these empirical criteria are certainly not those operative, or presupposed, in our mathematical practice (or: in our mathematical practice), and they have therefore to be discarded if our aim is to better understand just this practice. This is an anti-foundationalist manner of argumentation which, I think, never occurred to someone like Russell, who otherwise was very reflected

80

and anything but naive with respect to foundational issues. So, Russell leaves no doubt that the foundational function of his system — the system of Principia Mathematica — should not be considered as an epistemological one: the reasons which in fact make us hold true the familiar mathematical propositions, and especially the propositions of arithmetic, are obviously quite different from the reasons which his own system provides, and they give our familiar propositions much more convincing power than a derivation in his own system could achieve. Russell, in fact, turns the tables in this respect and lets the acceptance of his system be dependent on its ability to derive our familiar mathematical propositions in it.38 Wittgenstein’s thoughts, however, take a path that was foreign to Russell. As far as I know, the Wittgensteinian question about the identity of the symbols and proofs occurring in Principia Mathematica was never raised by Russell himself.39 It throws new light on the character of the foundation offered by Russell. This foundation is not only, as Russell admits, not an epistemological one, but also none in which the identity of the symbols and proofs that are constructed by chains of definitions should be considered as based on the identity of the signs and proofs at the foundational level. On the contrary, just the opposite seems to be true. Wittgenstein draws our attention to the fact that in the case of proofs at the foundational level, it is not only, as Russell observed, our acceptance of them which is at stake but also they themselves, as proofs with their own, characteristic identity. Russell’s system cannot be regarded as a system in which the higher levels are constituted on the basis of the foundational one, because the sentences and proofs on the higher levels must already be equipped with their own criteria of identity, which they do not obtain from Russell’s system alone. A foundationalist can try to defend himself against this Wittgensteinian undermining with two arguments that have already been mentioned. First, he can insist that the foundational system, despite all its unsurveyabilities, reveals the essence of what is mathematical — the essence of numbers, of 38. See Whitehead/Russell 1925, p. v, and also the reflections in Russell 1907, 272–275. 39. This question can already be found, with respect to strings of strokes and sums of 1’s (and, as this suggests, perhaps prompted by Wittgenstein’s study of Hilbert), in the Big Typescript: “[W]hat is the criterion in the stroke-notation for our having the same numeral in front of us twice? […] How can I know that ~~~~~~~~ and ~~~~~~~~ are the same sign? After all it is not enough that they look alike. For having roughly the same gestalt can’t be what is to constitute the identity of the signs, it is rather just the numerical identity. (The problem of the distinction between 1 + 1 + 1 + 1 + 1 + 1 + 1 and 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 is much more fundamental than appears at first sight. […])” (BT 580; cf. PG 331).

81

addition, and so on —, and that precisely this makes up its foundational character. And second, he can retreat to the theoreticity rejoinder, which says that the proofs on the foundational level are important only as theoretically postulated ones, such that we need not care about the verification of the identity of proof-patterns; this identity, too, should be considered as a purely theoretical one. Wittgenstein, however, is not satisfied with these rejoinders. As I read him, the claim that a foundationalist system reveals the essence of mathematics is handled by him quite undogmatically. He in no way needs to dispute that a foundational system might present, as he says in RFM III § 18, ‘original’ (‘ursprüngliche’) mathematical ideas and procedures; for example ideas concerning the intimate connection between addition and disjunction (as discussed by Wittgenstein in MS 122, 9r–9v and 17r–21v). And in RFM III § 19 he admits that, in a sense, “the calculation is based on the unit steps” (as suggested by the Peano axioms); there he only warns of the danger “of looking at the shortened procedure as a pale shadow of the unshortened one”. At the same time, however, he insists that there are excellent reasons also to consider the shortened proof procedures, which are located at the higher levels, as ‘original’ in important respects. In RFM III § 18 he mentions one of these reasons: “A shortened procedure tells me what ought to come out with the unshortened one. (Instead of the other way round.)” That is, because of their unsurveyability, the unshortened procedures do not have the force to convince us that something must be the case, and our insight in this ‘mathematical must’ makes the unshortened procedures at the foundational level dependent on the shortened ones at higher levels. In this important respect the so-called foundational level can not be regarded as foundational at all. The same is true, when we consider the proofs at the foundational level as only being theoretically postulated, as the theoreticity rejoinder would have it. A mathematical proof convinces us by what we can see (RFM III § 39), whereas a merely theoretically postulated proof is not an object of our perception from the outset. Therefore, in the case of the latter it does not make sense at all to ascribe the convincing force characteristic of mathematical proofs to it. Wittgenstein expresses this as follows: The proof must be intuitive [anschaulich sein]: if we are no longer convinced by what we see, then the proof has lost its force. No matter, whether it is construed according to the ‘logical’ schema of Russell or otherwise. (MS 122, 80v)

82

In other words, the so-called foundational level fails in founding mathematical necessity!40 What I just said still belongs to the topic “criteria of identity for proofs”, and Wittgenstein’s argument was that these criteria depend on conceptual tools that are not available at the foundational level, but only at higher levels in which our familiar mathematical procedures are simulated. And his further argument was, as we have just seen, that the foundationalist is not allowed to deal with this fact in a nonchalant way, because it touches upon important aspects of mathematical proofs, such as their convincing force which gives the mathematical results their peculiar necessity. Let us turn now to Wittgenstein’s second way of dealing with the simulational potential of foundational systems, his externality objection (as I have called it). It says that in going away from the foundational level by means of definitions, we bring new concepts into play — or, as Wittgenstein also says, new ‘aspects’ (see RFM III §§ 46–50) — which are added to the foundational system from outside, such that, without these additions, it cannot be considered ‘foundational’ at all. This objection, as I already said, is certainly in need of careful explanations, since these new concepts, or aspects, can very well be simulated within the machinery of a foundational system. In which sense, and in what way, should they be regarded, then, as ‘external’ to the system? This question has to be answered by means of examples. In RFM III § 47 Wittgenstein uses the example of exponentiation, i.e., the replacement of expressions of the form “a u a u … u a”, with n factors a, by expressions of the form “an”.41 What is new — the new aspect, the new concept — that is created by this abbreviating notation? Wittgenstein observes, first, that when writing “an” instead of “a u a u ... u a” we have to take notice of the number of a’s occurring in the product, and our calculating procedure will then be different: when calculating 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16 u 16, we can, so to speak, blindly check off the factors until we are finished with them, without in the least caring about 40. The topic “mathematical necessity” is of central importance to Wittgenstein’s philosophy of mathematics (see Mühlhölzer, forthcoming, for more on this subject), and we should not be surprised that it also proves to be important to Wittgenstein’s discussion of the surveyability of proofs. It certainly would deserve a much more extensive treatment than I can give it in this paper. 41. It is interesting that Wittgenstein uses just this example, which also proves to be ticklish with regard to Hilbert’s finitary mathematics, as shown in Parsons 1998. However, Wittgenstein’s discussion is not directed at finitism, but at foundationalism in the manner of Russell, and therefore his reflections are of a quite different sort than Parsons’.

83

their number; whereas, when calculating 1615, it is precisely this number we have to heed. This leads to the establishment of new mathematical connections, for example connections between (as Wittgenstein writes in RFM III §47) “the technique of counting factors and the technique of multiplying”, which can be condensed, then, in new mathematical laws, like am+n = am u an, and similar ones. Furthermore, Wittgenstein asks for the circumstances in which, on the basis of connections and laws of this sort, only the abbreviating notation, and not also the original one, can be used in a manner that is mathematically important. It is not difficult to specify circumstances of this sort. Wittgenstein gives the example (which can only be found in the unpublished Nachlass42) of an inductive proof, in which the inductive step from n to n+1 refers to an exponent, i.e., to the variable n in an expression of the form “an”. Obviously, such an induction is not possible if we have not yet introduced the notation “an”. In this example we meet one of the many ‘aspect changes’ that are so characteristic of, and so familiar in, our mathematical practice and which have such important consequences for our techniques of proof. Wittgenstein wants to direct our attention to precisely such phenomena. In contrast to the uniform and pale foundational landscape, mathematics will then present itself as the rich field — as the multicoloured mixture of techniques of proof (RFM III § 46) — that it really is. In RFM III § 48 Wittgenstein says that he wants to explain the colourfulness of mathematics. The explanation, which he suggests in the present context, is that in our mathematical practice, when confronted with situations that threaten to become unsurveyable, we are permanently forced to introduce new techniques. These techniques should not be considered as already anticipated by what we possessed before, but as creative, and multifarious, advancements of mathematics.43

42. Even there, it is only hinted at, but the hints are clear enough: see MS 122, 103r– 104v. 43. I am grateful to the participants of the Wittgenstein Workshop, held in July 2004 in Ulm and organized by Michael Kober, where the original, German version of this paper was presented and discussed. Furthermore, I am indebted to the Deutsche Forschungsgemeinschaft for awarding me two research semesters (grant MU 687/4–1) that gave me the opportunity to think about Wittgenstein’s view of mathematical proofs in peace and quiet. As always I thank my wife, Marianne Mühlhölzer, for the many discussions I could have with her about Wittgenstein’s philosophy of mathematics. I also profited from discussions with Ulrich Majer and Christian Tapp.

84

REFERENCES Church, Alonzo 1956: Introduction to Mathematical Logic, Princeton University Press. Detlefsen, Michael 1986: Hilbert’s Program, D. Reidel. Ewald, William (ed.) 1996: From Kant to Hilbert: A Source Book in the Foundations of Mathematics, Vol. II, Clarendon Press. Frascolla, Pasquale 1994: Wittgenstein’s Philosophy of Mathematics, Routledge. Frascolla, Pasquale 2001: “Philosophy of Mathematics”, in Wittgenstein — A Critical Reader, ed. by Hans-Johann Glock, Blackwell, 268–288. Frege, Gottlob 1994: Die Grundlagen der Arithmetik, Felix Meiner, 1988; English translation by John L. Austin: The Foundations of Arithmetic, 2nd. rev. ed., Blackwell, 1953; citations are from this edition. Glock, Hans-Johann 1996: A Wittgenstein Dictionary, Blackwell. Guedj, Denis 1983: “Nicolas Bourbaki, Collective Mathematician: An Interview with Claude Chevalley”, The Mathematical Intelligencer 7, No. 2, p. 18–22. Hacker, Peter M. S. 2004: “Übersichtlichkeit und übersichtliche Darstellungen”, Deutsche Zeitschrift für Philosophie 52, 405–420. Heijenoort, Jean van (ed.) 1967: From Frege to Gödel: A Source Book in Mathematical Logic, 1897–1931, Harvard University Press. Hilbert, David 1922: “Neubegründung der Mathematik”, Abhandl. aus dem Math. Seminar d. Hamb. Univ., Bd. 1, 157–177 (also in David Hilbert, Gesammelte Abhandlungen. Band III, J. Springer, 1935, 157–177; and in Hilbertiana, Wissenschaftliche Buchgesellschaft, 1964, 12–32; English translation by William Ewald in Ewald 1996, 1117–1134; citations are from this edition and use the paragraph numbers added by Ewald). –– 1925: “Über das Unendliche”, Mathematische Annalen 95, pp. 161–190 (also in Hilbertiana, Wissenschaftliche Buchgesellschaft, 1964, 79–108; English translation by Stefan Bauer-Mengelberg in van Heijenoort 1967, 369–392; citations and page numbers are from this edition). –– 1928: „Die Grundlagen der Mathematik“, Abhandl. aus dem Math. Seminar d. Hamb. Univ., Bd. 6, 65–85 (also in David Hilbert, Die Grundlagen der Mathematik, Hamburger Mathematische Einzelschriften, Teubner, 1–21; English translation by Stefan Bauer-Mengelberg and Dagfinn Føllesdal in van Heijenoort 1967, 464–479; citations and page numbers are from this edition). –– 1929, “Probleme der Grundlegung der Mathematik”, Mathematische Annalen 102, 1–9 (English translation by Paolo Mancosu in Mancosu 1998, 227–233; citations and page numbers are from this edition).

85

Hilbert, David und Bernays, Paul 1968: Grundlagen der Mathematik I, Springer, Berlin. MacBride, Fraser 2003: „Speaking with Shadows: A Study of Neo-Logicism“, British Journal for the Philosophy of Science 54, 103–163. Majer, Ulrich 1995: “Geometry, Intuition and Experience: From Kant to Husserl”, Erkenntnis 42, 261–285. Mancosu, Paolo (ed.) 1998: From Brower to Hilbert: The Debate on the Foundations of Mathematics in the 1920s, Oxford University Press. Mathias, A. R. D. 2002: “A Term of Length 4 523 659 424 929”, Synthese 133, 75–86. Mühlhölzer, Felix (forthcoming): “Putnam, Wittgenstein and the Objectivity of Mathematics”, to appear in The Philosophy of Hilary Putnam (in the Library of Living Philosophers), ed. Randall E. Auxier, Open Court. Parsons, Charles 1998: “Finitism and Intuitive Knowledge”, in The Philosophy of Mathematics Today, ed. by Matthias Schirn, Clarendon Press, 249–270. Russell, Bertrand 1907: “The regressive method of discovering the premises of mathematics”, reprinted in B. Russell, Essays in Analysis, ed. by D. Lackey, Braziller, 1973, 272–83. Sieg, Wilfried 1999: “Hilbert’s Programs: 1917–1922”, The Bulletin of Symbolic Logic 5, 1–44. Steiner, Mark: 1975, Mathematical Knowledge, Cornell University Press. Weyl, Hermann 1946: “Review: The Philosophy of Bertrand Russell”, The American Mathematical Monthly 53, 208–214 (reprinted in Hermann Weyl, Gesammelte Abhandlungen, Vol. IV, Springer, 1968, 599–605; page numbers are from this edition). Weyl, Hermann 1949: “Wissenschaft als symbolische Konstruktion des Menschen”, Eranos-Jahrbuch 1948, 375–431 (reprinted in Hermann Weyl, Gesammelte Abhandlungen, Bd. IV, Springer, 1968, 289–345; page numbers are from this edition). Whitehead, Alfred North and Russell, Bertrand 1925: Principia Mathematica, Vol. 1, second edition, Cambridge University Press.

86

Grazer Philosophische Studien 71 (2006), 87–116.

WITTGENSTEIN AND RELIGION Michael KOBER University of Ulm Summary It will be shown that Wittgenstein’s philosophical approach to religion is substantially shaped by William James’ The Varieties of Religious Experience. For neither during the Tractatus period nor later does Wittgenstein thematise religious doctrines, but rather struggles to determine what it means for a sincere person to have a specific religious attitude (James called these attitudes “experiences”). Wittgenstein’s almost exclusive focus on attitudes explains, (i) why he is able to strictly discriminate between scientific and empirical claims on the one hand and religious utterances on the other, (ii) why religious and mythical narrations should not be understood as promoting (pre-scientific) theories, (iii) why Wittgenstein non-cognitively interprets religious utterances such as “This is God’s will” as avowals, and (iv) why he seems to promote fideism. Wittgenstein’s one-sided way of reflecting on religious matters, however, should not be understood as adumbrating or even promoting any more specific account of religion, especially bearing in mind that many of his remarks concerning religion are connected to or motivated by reflections on his own life. This thesis is meant to imply that Wittgenstein does not, as the usual understanding holds, offer a ›theology for atheists‹.

It is a striking fact that hardly any other philosopher has written so little, in a systematic fashion, on religion and yet has been adopted by so many (mainly Christian) theologians and/or believers. Many authors, in a host of articles and books, claim that Wittgenstein’s remarks do philosophically substantiate, or authoritatively provide security for, their own (mutually incompatible) views. Nevertheless, it remains an uncontroversial philological fact that Wittgenstein’s remarks concerning religion are a mixed bag of more or less vague hints. Although they are, as will be shown, systematically linked to his general philosophical approaches, they are also interwoven with reflections on the course of his own life and personal development. For that reason, his remarks concerning religion are often one-sided and

leave many questions open. Therefore, they need to be treated with care when read merely from a philosophical or non-personal perspective, and I will argue that they do not contain a systematic exposition of, or recommendation for, any specific approach in theology or philosophy of religion. Wittgenstein’s hesitation to promote any particular account of religion is certainly connected to his dislike of proselytising, and I will explain why this is just a consequence of his overall approach to that aspect of the phenomenon of religion in which he was interested. On the other hand, some scholars think that Wittgenstein is advocating a sort of fideism, i.e. the view that a religious conviction is but a matter of a specific non- or pre-propositional intuition, or an emotional stance, such that religious conviction need not be connected to nor even rest upon a particular religious doctrine. However, I have found no argument in support of Wittgenstein’s holding such a view. In what follows, I will provide an explanation for the fact that Wittgenstein’s texts, notes, and lectures are vague and appear to promote fideism; and in general, I will discuss critically and from a philosophical point of view his own attitude towards religion. 1. William James on Religious Attitudes 1.1 It is well-known that Wittgenstein read a great number of texts possessing more or less spiritual undertones, by authors generally concerned with religious matters, such as Dostoievski, Tolstoy, Angelus Silesius, Cardinal Newman, or Kierkegaard. But we also know that Wittgenstein appreciated another book on religion: He carefully read already in 1912 William James’ 1902 Gifford Lectures The Varieties of Religious Experience, and he told Russell in a letter: “This book does me a lot of good” (RUL 231, 22 June 1912). I will unfold in which way James’ discussion of ›religious experiences‹ substantially shaped Wittgenstein’s approach to the subject of religion, and then will go on to speculate as to precisely what it was that ‘did him a lot of good’. In his lectures, James covers numerous related subjects, such that my presentation must focus only on one chain of reasoning constructed out of the rich material he supplies.1 To begin with, as he wishes to restrict 1. The gist of the following discussion of James and its relevance for understanding Wittgenstein’s Tractatus appeared in a less elaborated and differently oriented version already in Kober 2005, 234–40 (reproduced with permission of Palgrave Macmillan).

88

himself to that domain with which he considers himself to be sufficiently acquainted, James is almost entirely concerned with aspects of practised Christian religion — without thereby downgrading other forms of faith, confession, or myth (J 2–3).2 He then distinguishes, due to problems of equivocation, between four different uses of the word “religion”, although he concentrates almost entirely on the last one. According to James, the term “religion” may mean (i)

›the church‹, meaning ecclesiastical organisations and institutions (J 3, 334–6), (ii) a set of conventional religious practices such as saying one’s prayers, going to confession, baptising one’s children, going on pilgrimages, and so forth (J 6), (iii) a set of specific religious beliefs, doctrines, or (theological) theories that attempt to systematise the religious belief-contents of particular religious communities (J 337), and (iv) religious experiences, as for example awakenings, revelations or specifically religious feelings, and — sometimes resulting from them — religious attitudes towards the world, towards human life in general, one’s friends, neighbours, and family, or towards one’s own life in particular (more on this below). Ecclesiastical organisations (i) and religious practices (ii) are usually investigated by religious studies of a sociological, ethnological, or historical kind, but while both James and Wittgenstein disparage the inherent will to power of ecclesiastical organisations, their accompanying trait of developing “a hierarchy, honours, and official positions” (CV 35), and tendency to corruption and cruelty (J 334–6), neither of them show a deeper interest in these. Though Wittgenstein shows in his Remarks on Frazer’s ›Golden Bough‹ that prima facie peculiar magical, mystical, or religious rituals can be rendered understandable — especially if they are not interpreted as naive or pre-scientific techniques — he never genuinely scrutinises religious practices within particular communities (such as, for example, which rituals believers of a particular community might be expected to go through during their lifetime). 2. References to page-numbers of James 1902 will henceforth be made by “J”. — Throughout his life, Wittgenstein’s concerns on religion will, like James’, be confined to sorts and aspects of Christian, often catholic belief (he is, for example, reflecting on Christian notions like redemption, resurrection, or confession, and not of concepts like nirvana, dharma, or being kosher).

89

Whereas philosophers often critically discuss specific religious doctrines (iii), questioning their claims to truth and reasonableness — most prominently with proofs of God’s existence —, James declares himself a psychologist3 who is simply incompetent in such matters (J 2). He hence refrains from further investigating this aspect of lived religious life (– and it will be argued that Wittgenstein, at least during his Tractatus-period, found his own way to avoid the necessity of discussing theological questions). James furthermore stresses that in scrutinising “religious experiences” (iv), he does not aim at investigating the religious life of a conventional religious believer, since “his religion has been made for him by others, communicated to him by tradition, determined to fixed forms by imitation, and retained by habit”; it is a “second-hand religious life” (J 6). Being himself a creative thinker who craves authenticity, James searches rather for “the original experiences which were the pattern-setters to all this mass of suggested feeling and imitated conduct”. He hastens to acknowledge that many reports of awakenings, revelations and the like may be ascribed to “eccentric” persons who show “symptoms of nervous instability” and have been “liable to obsessions and fixed ideas”; indeed some of these people have certainly been “subject to abnormal psychical visitations” (all quotes J 6–7). However, (keeping his own religious attitude in mind) James insists that there are many known reports which show such a high degree of authenticity and veracity, that it would be — perhaps due to a ›scientistic‹ stance — inexcusably narrow-minded to belittle and neglect all of them (J 11–8). To substantiate his reflections, James cites countless examples, many of which indeed made an impression on Wittgenstein (Drury 1976/81, 108; McGuinness 1988, 129, 156, 158), for reasons that shall become clear. What is of importance in the present context are less the reports of revelations or awakenings, than the attempts of many of James’ sources to describe how their experiences — or lack of specific experiences — that shaped or affected their attitude towards the world and human life. That is, what concerns us most are attitudes—by James often described as “experiences”—that do not necessarily result in religious truth claims or proselytising. 1.2 It will he helpful to construct out of James’ text a typology of religious attitudes. These types should be understood as instances of Max Weber’s 3. In 1902, James was indeed primarily known as the author of The Principles of Psychology, which appeared 1890.

90

ideal types, that is, examples which due to some abstraction, show typical characteristics and need not necessarily stand for a enormous set of empirically demonstrable objects (cf. Weber 1904). I presume we all have at least heard of some of the following five types of quite extreme cases: (1) The happy or “healthy-minded” agnostic: To provide an example, James quotes from a 1890s interview with a 67 years old successful businessman who presents himself as scientifically-minded, believing in progress, and enjoying nature, music, and literature. He insists not to have any need for traditional religion, since it is “pernicious”: it teaches us “to rely on some supernatural power when we ought to rely on ourselves. […] All of my thoughts and cogitations have been of a healthy and cheerful kind, for instead of doubts and fears I see things as they are, for I endeavour to adjust myself to my environment.” “The God-idea”, he adds in a Russellian or Freudian manner,4 was begotten in ignorance, fear, and a general lack of any knowledge of nature. […] It seems to me that [thinking by means of the notion of ] sin is a condition, a disease, incidental to man’s development not being yet advanced enough. […] We should think that a million of years hence equity, justice, and mental and physical good order will be so fixed and organised that no one will have any idea of evil or sin. (from J 92–3)

James comments that someone in search of help in case of a personal crisis, would certainly not contact such an overly-optimistic person, so “coarsemeated and incapable of wounded spirit” (J 92): “His contentment with the finite encases him like a lobster-shell and shields him from all morbid repining at his distance from the Infinite” (J 93). James additionally declares: “The writer’s state of mind may by courtesy be called a religion, for it is his reaction [or attitude] to the whole nature of things, it is systematic and reflective, and it loyally binds him to certain inner ideals” (J 92). (2) The unhappy agnostic: This type, as the label indicates, consists of the counter-type of the ›happy agnostic‹. Here is one example: Away down to the bottom of my heart, I believe I was always more or less sceptical about “God”; […] When I was sixteen I joined the church and was asked if I loved God. I replied “Yes”, as was customary and expected. But instantly with a flash something spoke in me, “No, you do not.” I was haunted for a long time with shame and remorse for my falsehood […]. At 4. Cf. B. Russell, Why I am not a Christian, 1957; S. Freud, The Future of an Illusion, 1927, as well as Civilization and Its Discontents, 1930.

91

nineteen, […] I heard told a story of a brute who had kicked his wife downstairs, and then continued the operation until she became insensible. I felt the horror of the thing keenly. Instantly this thought flashed through my mind: “I have no use for a God who permits such things.” This experience was followed by months of stoical indifference to the God of my previous life, mingled with feelings of positive dislike and a somewhat proud defiance of him. I still thought there might be a God. If so he would probably damn me, but I should have to stand it. I felt little fear and no desire to propitiate him. (from J 177n)

James describes ›unhappy agnostics‹ as persons who show, in comparison with ordinary people, an enhanced sensitivity to the cruelties of nature (earthquakes, tsunamis etc.) and human affairs, from both of which they suffer. They take these cruelties for granted — as obviously belonging to human life — and do not expect a transcendent being claiming responsibility for such disastrous events (i.e. they are not interested in problems of theodicy). ›Unhappy agnostics‹ are usually melancholic (cf. J 148). (3) The happy ›once-born‹ religious believer: For reasons that will subsequently become clear, James adopts the terms ›the once-borns‹ and ›the twice-borns‹ from Francis W. Newman (Newman 1852, 89, 91; J 80). The ›once-born‹ happy believers are usually remarkably optimistic, since they consider the condition of the world as a whole, the position of man in nature, the relationship between men, and between God and men, as thoroughly benevolent and are therefore often in an elated or even enthusiastic mood (cf. J 67–8). They “see God not a strict judge, […] but as the animating spirit of a beautiful harmonious world, beneficent and kind, merciful as well as pure” (Newman in J 80). One ›once-born‹ expresses his stance as follows: The great central fact of the universe is that spirit of infinite life and power that is back of all, that manifests in and through all. […] In just the degree in which you realise your oneness with the infinite spirit, you will exchange disease for ease, inharmony for harmony, suffering and pain for abounding health and strength. (from J 100–1)

James concedes that “one can but recognise in such writers as these the presence of a temperament organically weighted on the side of cheer and fatally forbidden to linger […] over the darker aspects of the universe”, and he continues: “In some individuals optimism may become quasi-pathological. The capacity for even a transient sadness or a momentary humility seems cut off from them as by kind of congenital anaesthesia” (J 83).

92

(4) The unhappy religious believer, or the sick soul: ›Sick souls‹ display the opposite temperament of happy believers, for they do not glorify the condition of the universe but are well conscious of injustice, suffering, and evil, which may affect in extreme cases a “melancholy in the sense of incapacity for joyous feeling” (J 147). James furthermore explains: Such sensitiveness and susceptibility to mental pain is a rare occurrence where the nervous constitution is entirely normal; one seldom finds it in a healthy subject even where he is the victim of the most atrocious cruelties of outward fortune. (J 145) In none of these cases was there any intellectual insanity or delusion about matters of fact; […] the whole universe [rather is] coagulating about the sufferer into a material of overwhelming horror, surrounding him without opening or end. Not the conception or intellectual perception of evil, but the grisly blood-freezing heart-palsying sensation of it close upon one, and no other conception or sensation able to live for a moment in its presence. (J 161–2)

The following report provides an impressive illustration: Everything I saw seemed to be a burden to me; the earth seemed accursed for my sake: all trees, plants, rocks, hills, and vales seemed to be dressed in mourning and groaning, under the weight of the curse, and everything around me seemed to be conspiring my ruin. My sins seemed to be laid open; so that I thought that every one I saw knew them […]: sometimes it seemed to me as if every one was pointing me out as the most guilty wretch upon earth. […] When I waked in the morning, the first thought would be, “Oh, my wretched soul, what shall I do, where shall I go?” And when I laid down, would say, “I shall be perhaps in hell before morning.” (from J 159)

James comments: “How irrelevantly remote seem our usual refined optimisms and intellectual and moral consolations in presence of a need of help like this! Here is the real core of the religious problem: Help! Help!” (J 162). (5) The twice-born religious believers: Some people are able to leave their stance of being a ›sick soul‹. It happened to them ›to escape the tunnel at the other end‹ and experience a so-to-speak second birth. It amounts to “the loss of all worry, the sense that all is ultimately well with one, the peace, the harmony, the willingness to be, even though the outer conditions should remain the same” (J 248). Sometimes, this change is effected by some kind of awakening or revelation, through altered feelings or because

93

of altered powers of action, perhaps sometimes even through insights into one’s position in life, but sometimes it just happens (see below). “It may come gradually, or it may occur abruptly […]. However it come, it brings a characteristic sort of relief; and never such extreme relief as when it is cast into the religious mould” (J 175). Quoted by James, Francis W. Newman characterises a ›twice-born‹ after his transformation as follows: The individual finds himself ›at one with all creation‹. He lives in the universal life; he and man, he and nature, he and God, are one. That state of confidence, trust, union with all the things, following upon the achievement of moral unity, is the Faith-state. Various dogmatic beliefs suddenly, on the advent of the faith-state, acquire a character of certainty, assume a new reality, become an object of faith. As the ground of assurance here is not rational, argumentation is irrelevant. (from J 247)

Perhaps the most prominent example of such a ›twice-born‹ was Tolstoy, whose My Confession of 1879 was quoted by James at length. Here are some illustrative excerpts: I felt that something had broken within me on which my life had always rested, that I had nothing left to hold on to, and that morally my life had stopped. […] It was an aspiration of my whole being to get out of life. […] All this took place at a time when so far as all my outer circumstances went, I ought to have been completely happy. I had a good wife who loved me and whom I loved; good children and a large property which was increasing with no pains taken on my part. I was more respected by my kinsfolk and acquaintances than I had ever been […]. And yet I could give no reasonable meaning to any actions of my life […] and I sought for an explanation in all the branches of knowledge acquired by men. […] I sought like a man who is lost and seeks to save himself, — and I found nothing. I became convinced, moreover, that all those who before me had sought for an answer in the sciences have also found nothing. […] Yet, whilst my intellect was working, something else in me was working too, and kept me from the deed — a consciousness of life, as I may call it, which was like a force that obliged my mind to fix itself in another direction and draw me out of my situation of despair […]. I can call this by no other name than that of a thirst for God. This craving for God had nothing to do with the movement of my ideas, […] but it came from my heart. It was like a feeling of dread that made me seem like an orphan and isolated in the midst of all these things that were so foreign. And this feeling of dread was mitigated by the hope of finding the assistance of some one. (from J 153–6)

94

Tolstoy’s religious experience, his abandonment of the sciences, along with James’ and Newman’s characterisation of the ›twice-born‹ enables us to understand Wittgenstein’s philosophical treatment of religion during his Tractatus period more easily. 2. Wittgenstein on Religion during the Tractatus Period 2.1 As is well-known, Wittgenstein in the Tractatus conceptualises language as consisting of propositions that are meaningful or significant, if and only if they depict possible states of affairs (TLP 4.1, 4.2). Hence, “the totality of true propositions is the whole of natural science” (TLP 4.11; cf. TLP 6.53). Therefore, propositions are understood to be concatenations of ›names‹ that ultimately, i.e. subsequent to a penetrating logical analysis, refer to existing and somehow empirically determinable ›objects‹ in the world (TLP 3.21, 3.22, 4.22). This account easily enables the author of the Tractatus to rate propositions such as “God is omniscient” — i.e. propositions that claim to express true religious theses or doctrines (“true” in the sense of: agreeing with something that actually exists as part of the world) — nonsensical (unsinnig; TLP 4.003, 4.4611). For such propositions contain ›names‹ like “God” (and “omniscient”, as well) that purportedly refer to ›objects‹ which do not belong to the scientifically determinable world — at least according to the Tractatus, where Wittgenstein maintains that “God does not reveal himself in the world” (TLP 6.432). Wittgenstein therefore establishes a strict distinction between scientific and religious theses and is thereby able to explain why the sciences could not be of any help to Tolstoy during his period of desperation — or why, as F. W. Newman put it, “argumentation is irrelevant” (in J 247, quoted above). On the other hand, this account also implies that there is no need to care, intellectually speaking, about religious doctrines or theological theories (cf. James’ class (iii) in section 1.1. above), since the assumption that they have propositional and therefore truth-value relevant contents merely rests on a confusion. 2.2 In addition to purportedly religious propositions that are nonsensical, one may encounter another kind of ostensibly religious propositions: These are deemed “inexpressible” (unaussprechlich, TLP 6.522), since they try to ›say‹ what can only be ›shown‹ (TLP 4.1212). No doubt, Wittgenstein’s Tractatus doctrine of ›saying‹ and ›showing‹ is mainly designed to apply to

95

logical matters, i.e. to the philosophical interpretation of predicate logic and to the (supposed) logical scaffolding of language and the world. It is designed to express Wittgenstein’s view, firstly (and in opposition to Russell),5 that logic is entirely different from the natural sciences, and secondly, that the Tractatus-Wittgenstein is unable to provide any ›deeper‹ explanation of why predicate logic is as it is. Hence, “logic must take care of itself ” (TLP 5.473), and this thesis implies that we, to our astonishment, are forced without any further reason to accept logic as such. This (supposed) matter of fact is the paradigm case of “the mystical” (TLP 6.522). Similarly, we are also compelled to accept that the world exists as it does, since there is — according to the Tractatus — no possible significant proposition which concerns or even explains the world as a whole — for example, in the case where one tries to account for or merely to think of its origin or creation. For the ›name‹ “the world” in propositions of the form “The world is F ” refers already to the existing and scientifically determinable totality ›the world‹, such that there cannot be a significant predicating expression “F ”. For in order to be significant, the above mentioned concatenating-model of the proposition requires the predicating expression “F” to refer to an ›object‹ outside the world. However, since such an ›object‹ would not be scientifically determinable, the predicating expression “F ” has no meaning; i.e. there is no empirically specifiable ›object‹ to which “F” refers. From this Wittgenstein is able to conclude: “Not how the world is, is the mystical [since that can be significantly expressed by true empirical propositions that refer to existing states of affairs being part of the world], but that it is” (TLP 6.44; cf. LE 41). Of course, such an insight is, properly speaking, ›inexpressible‹ by significant propositions and can therefore only be ›shown‹ (TLP 6.54). 2.3 Tolstoy in his (above quoted) Confession did not talk about the condition of the world, but attempted to describe the change of his own attitude, stance, or mood towards the world. The most striking thing is that he altered his attitude etc. “even though the outer conditions […] remain[ed] the same” (as James had it; J 248), i.e. despite the fact that all relevant existing states of affairs continued to be. Similarly, James pointed out regarding ›sick souls‹ that their drastic unhappiness does not depend on any “delusion about matters of fact” (J 162, as quoted above). Therefore, 5. “[L]ogic is concerned with the real world just as truly as zoology, though with its more abstract and general features” (Russell 1919, 169).

96

by “the world of the happy man is different from that of the unhappy man” (TLP 6.43), Wittgenstein means that one’s being happy or unhappy depends on one’s (religious) attitude while one is experiencing the world, not necessarily on the facts. Yet, even this can only be ›shown‹, since it is, according to the Tractatus, not possible to describe significantly the mental status, attitudes, or moods of a person. Propositions such as “Leo is a ›twice-born‹ believer” or “Edna is a ›sick soul‹“ are declared to be just as inexpressible as “John is sad” or “Ralph believes that the man on the beach is a spy.” For the ›names‹ that occur in these propositions (“Leo” etc.) would need to have an ›object‹ in the world as their reference in order to be meaningful, such that the respective ›objects‹, i.e. persons, would be the bearers of respective mental properties. In this case, Wittgenstein in the Tractatus calls these ›objects‹ souls or ›subjects‹ (cf. TLP 5.5421). However, the Tractatus develops the view that souls or ›subjects‹ do not belong to the world, rather to the limits of the world (TLP 5.632). For this reason, Wittgenstein in the Tractatus treats propositions of the form “A believes that p” as follows. The believing or thinking ›subject‹ A manifests itself in the world through the existence of a thought, for example the thought “p”. The proposition “A believes that p” therefore may be transformed with constant meaning into “A thinks p” or “’p’ thinks p”, which is supposed to mean the same as “There is a thought that (has as its content:) p” (TLP 5.541–5.5421). It is hence the well-established view that the Tractatus accounts philosophically for how we are able to understand the world as well as (scientific) propositions concerning the world, but not for persons and their mental properties. According to the Tractatus, we are thus not able to say and to understand propositions concerning the religious attitudes of persons, and consequently do not understand religious attitudes themselves. If that is so, it follows that there is no significant argumentation concerning any possible religious attitude. Still in his Lecture on Ethics (1929), Wittgenstein insists that propositions such as “I wonder at the existence of the world” or “I am safe, nothing can injure me whatever happens” should be deemed nonsense (LE 41); that is, propositions which in non-Tractarian terminology could clearly be understood as religious avowals, perhaps uttered by ›once-born‹ or ›twice-born‹ believers (avowals are utterances by way of which a person more or less unintentionally expresses his or her mental state, e.g. by “Ouch!” or “I am in pain” in the case of being suddenly in pain; cf. PI § 244). From this general point of view, Wittgenstein

97

has philosophically justified F. W. Newman’s declaration: “As the ground of assurance here is not rational, argumentation is irrelevant” (J 247). 2.4 It was certainly with texts such as Tolstoy’s Confession in mind, that Wittgenstein wrote in his Notebooks 1914–16: “To believe in a God means to see that the facts of the world are not the end of the matter [dass es mit den Tatsachen der Welt noch nicht abgetan ist]. To believe in God means to see that life has a meaning” (NB 8.7.16, p. 74). One may have the impression that the allusion to God here can be understood as a gesture granting that any philosophical explanation or justification has come to an end, particularly any explanation of why there is ›meaning‹ anyway — which is to say, the meaning of life in general, of one’s own life in particular, of the world, or of that expressed by means of propositions; it appears to be up to God, or some divine ›pneuma‹, that perceptible and scientifically determinable, though somehow ›dead‹ ›signs‹ may turn into meaningful ›symbols‹; cf. TLP 3.11, 3.31, 3.326).6 However, consistent with his overall Tractatus approach, Wittgenstein sighs at the view just outlined: “I am conscious of the complete unclarity [or inexpressibility; cf. TLP 6.522] of all these sentences” (NB 2.8.16, p. 79). And he certainly alludes to Tolstoy again when he summarises: “We feel [that is: it is not significant knowledge] that even if all possible scientific questions be answered, the problems of life have still not been touched at all. Of course there is then no question left, and just this is the answer. — The solution of the problem of the problem of life is seen in the vanishing of the problem” (TLP 6.52f ). To my mind, Neil Young’s line “Though my problems are meaningless, that don’t make them go away”7 captures an appropriate objection to the Tractarian approach. As we will see, both the actual person and the philosopher Wittgenstein came to recognise similar pressures in the 1930s and later. 2.5 Before I comment on Wittgenstein’s later remarks on religion, I wish to further consider the above mentioned question of why he wrote to Russell in 1912, that James’ Varieties of Religious Experience had ‘done him a lot of good’. To be sure, at that time Wittgenstein had not yet fully developed 6. Cf. Joachim Schulte’s considerations as regards ›pneuma‹, in his contribution to the present volume. 7. This line appears in Neil Young’s song “On the Beach” on the album by the same title, 1974.

98

his Tractarian doctrine. Yet, I am certain that it was not James’ own ›pragmatic‹ evaluation of religion that interested Wittgenstein (as far as I know, Wittgenstein never commented on the interpretation James gives towards the end of the book). Rather, I imagine that what appealed to Wittgenstein was James’ (relatively) clear, factual and unbiased, though thoroughly non-scientific treatment of the phenomenon of religious attitude. James was, as indicated above, critical, but not unfair or narrow-minded (like Frazer in his Golden Bough), and did not presume an omniscient point of view. He presented in a sensitive manner an enormous variety of material, representing people who authentically, sincerely, and without unnecessary pathos struggle to express their respective religious experiences, moods, or attitudes. In other words, Wittgenstein was attracted by James’ examples of existential authenticity, and by the way in which they are presented, rather than by James’ own philosophical discussion. No doubt, these examples of religious attitudes, in combination with their unpretentious presentation would be attractive and reassuring to someone like Wittgenstein, authentically searching for his own healthy or appropriate state of mind. Wittgenstein wrote in his 1912 letter to Russell: “Whenever I have time I now read James’s ‘Varieties of Religious Experience’. This book does me a lot good. I don’t mean to say that I will be a saint soon, but I am not sure that it does not improve me a little in a way in which I would like to improve very much: namely it helps me to get rid of the Sorge [worry, anxiety, dread8]” (RUL 22.6.12, p. 231). To know that others are worried too, and that some of them, like Tolstoy, were able to overcome their worries, must have been a comfort to Wittgenstein. From a philosophical point of view, James’ above quoted characterisation of “a religion” as a person’s “reaction [or attitude] to the whole nature of things” which “is systematic and reflective, and […] loyally binds him to certain inner ideals” (J 92), proved to be fruitful for Wittgenstein: the correct or appropriate attitude definitely matters. James’ characterisation enables one to view the phenomenon of religion as if it were a matter of distinguishing not between reason and faith — both are a search for truths — but between science on the one hand and non-scientific or nontheological attitudes on the other; that is, between claims for truth, argumentation, justification and the like, and the sincere attempt to come to grips with one’s own position in the world and way of life (that is: one’s 8. Cf. the appearances of the word “dread” in Tolstoy’s Confession, quoted above.

99

acting). It will become obvious that this latter is a recurrent theme in Wittgenstein’s life and thinking. 3. The ›Later‹ Wittgenstein on Religious Attitudes 3.1 Although Wittgenstein’s account of language underwent substantial change after the Tractatus period, his overall perspective on religion remains remarkably constant — due to the fact that it focuses mainly on religious attitudes. Considerations which thematise aspects of religion can be found spread across his Diaries 1930–32, 1936–1937 (PPO 3–255), in his Lectures … on … Religious Belief (LC 53–72), in Culture and Value, as well as in his conversations with friends and acquaintances — although these hardly amount to discussions of ecclesiastical organisations, religious and mystical rituals, or particular theological doctrines or theories. Yet Wittgenstein definitely aims at a deeper understanding, firstly, of how religious attitudes are different from scientific or empirical ones, and secondly, of what is, or what it means to have, a religious attitude towards human life and human fate — in other words, what it means to live a religious life. All of this compelled Wittgenstein to rethink his own view of religious expressions such as “God”, or “belief ”, and he came to acknowledge that the use of such expressions need not be nonsensical or ›genuinely inexpressible‹, but may indeed be significant for all those who wish to understand the religious aspect of human life. This in turn required him to rethink the question of what these expressions might mean. Nevertheless, it must be repeated: Wittgenstein did not develop any comprehensive account of religion, such as he (arguably) did for ordinary language, mathematics, and philosophical psychology. Concerning religion, only scattered remarks are to be found, although they do in some respects adumbrate a particular view. 3.2 As is well-known, Wittgenstein after the Tractatus period abandoned the search for a closed, systematic theory of language, as well as for a general criterion of linguistic significance. Crudely speaking, the Wittgenstein of the Philosophical Investigations holds that what an expression or a sentence means can be explained by listing rules that govern the use of the respective expression or sentence in particular, i.e., contextually situated practices or language games (PI §§ 43, 560; Z § 560; CV 53, 97). In other words: It is human practice that accounts for the meaning of signs in use, and not God or a dubious ›pneuma‹ who creates meaning or significance (one may

100

be inclined in this respect to speak of a thoroughgoing secularisation of our philosophy of language). It is the task of philosophy, firstly, to make us aware of the actual usage of expressions and sentences, and secondly, to make us aware that prima facie similar, analogous or parallel applications of particular expressions in different language games may produce different usage and meaning (PI § 116). 3.3 To be sure, one may still say of so-called empirical or scientific propositions — i.e. empirical judgements, statements, or claims — that they are true or false (depending on how the respective matters of fact are understood). Accordingly, if a person denies an empirical or scientific proposition p, she usually accepts non-p as true. Furthermore, it belongs to propositional logic as well as to the ›logic‹ or ›grammar‹ of many scientific practices that, if another person believes that p or is convinced of (the truth) of p, and I am not, then I contradict her; in most cases, there will be a way or a method that is accepted by both of us which can help to determine who is right and who is wrong. However Wittgenstein stresses in his Lectures … on … Religious Belief, that things are different where religious matters are concerned: If someone believes, for example, in the Last Judgement, and I do not, then it is not necessarily the case that one of us is right and the other wrong (LC 53). There is no empirical or rational method acceptable to both of us which could settle the matter. Wittgenstein points out that the utterance of a proposition such as “I believe in the Last Judgement” does not constitute a true-or-false statement, but expresses one’s own attitude towards human life, by evaluating possible courses of action from a quite demanding moral point of view. In other words: such utterances are avowals. Religious attitudes of this kind are not convictions, they escape the true/false-bipolarity of empirical or scientific claims and are, so-to-speak, truth-value insignificant or truth-value agnostic. Wittgenstein rather suggests in his Remarks on Frazer’s ›Golden Bough‹ that religious beliefs or attitudes should be seen as analogous to ritualistic action, which must not be confused with bad or confused technique: “[T]he characteristic feature of ritualistic action is not at all a view, an opinion, whether true or false” (GB 129i; cf. GB 137b). Still in line with the Tractatus account, he considers it a matter of confusion or misunderstanding if one looks at religious beliefs or faith from the standpoint of being right or wrong. For were one to think so, one would have failed to realise the difference between making a (scientific) claim and expressing an attitude, that is, one would transform religion into supersti-

101

tion, which is “a sort of false science” (CV 82; LC 58–9). Wittgenstein stresses anew that a religious belief or faith “is not a scientific belief, has nothing to do with scientific convictions” (CV 72). In order to point out the distinction between science and religion, Wittgenstein sometimes applies a specific terminology — one that in fact can be traced back to James’ book (which assisted me in describing Wittgenstein’s view in the Tractatus). According to Wittgenstein, verbal expressions of religious beliefs or convictions such as, “I believe in the Last Judgement”, are “foreshadowed by the word ‘attitude’” (LC 71; emphasis added by M.K.), in the sense that they express a certain personal stance, or “guidance for this life” (LC 53). The sincerity and depth of such avowal-like confessions of faith is revealed in the respective person’s actions; for example, in the case of a happy ›twice-born‹ believer, whether or not she acts and reacts with great confidence and trust, come what may. If the believer in God looks around and asks “Where does everything I see come from?” […], what he hankers after is not a (causal) explanation [as, e.g., in scientific practices]; and the point of his question is that it is the expression of this hankering. He is expressing, then, an attitude [Einstellung; a variant runs: a stance, Stellungnahme] towards all explanations. — But how is this manifested in his life? / It is the attitude of taking a certain matter seriously […]. / Really what I should like to say is that here too what is important is not the words you use or what you think while saying them, so much as the difference that they make at different points in your life. (CV 96f; Wittgenstein’s emphasis) It appears to me as though a religious belief [i.e. a specific religious attitude] could only be (something like) passionately committing oneself to a system of coordinates. Hence although it’s belief, it is really a way of living, or a way of judging life. Passionately taking up this interpretation. (CV 73) We don’t talk about hypothesis, or about high probability [as in scientific practices]. Nor about knowing. (LC 57) [In a conversation with Drury:] If you and I are to live religious lives, it mustn’t be that we talk a lot about religion, but our manner of life is different. (Drury 1976/81, 109; emphasis M.K.)

In accordance with this view — namely, that what matters is a person’s attitude towards her own or her neighbour’s life, an attitude which need not be based on justifications or truth claims — Wittgenstein makes the following remark:

102

Christianity is not based on a historical truth, but presents us with a (historical) narrative and says: now believe! But not believe this report with the belief that is appropriate to a historical report, — but rather: believe, through thick and thin[,] and you can do this only as the outcome of a life. […] Queer as it sounds: the historical grounds of the Gospels might, in the historical sense, be demonstrably false, and yet belief [i.e. an attitude] would lose nothing through this […], because the historical proof (the historical proof-game) is irrelevant to belief [i.e. an attitude]. (CV 37f.)

Even if a religious believer were to assure us that he or she had proofs for his or her faith, these proofs would hardly be compelling to anyone else (i.e. one should not confuse them with logical or mathematical proofs): “He has what you might call an unshakeable belief [i.e. an attitude]. It will show, not by reasoning or by appeal to ordinary grounds for belief, but rather by regulating for all in his life” (LC 54). Obviously, James’ account of religious experience, as well as his above-quoted remark concerning the epistemic foundation of religious beliefs (“[a]s the ground of assurance here is not rational, argumentation is irrelevant”; J 247) lie in the background of these considerations. 3.4 Whereas the sciences and humanities doubtlessly also aim at explaining our natural and social world, such that we may rightfully hope to exercise an influence upon, or even completely control some specific upcoming course of events, a religious attitude might be motivated by a conscious awareness of the fact that our fate is sometimes decided beyond our sphere of hope and influence: “Fate is the antithesis of natural law. A natural law is something you try to fathom, and make use of, fate is not” (CV 70). “It simply isn’t a theory”, Wittgenstein stresses, rather it is “a sigh, or a cry” (CV 34f ). That is, utterances like “I believe in the Last Judgement” are again to be seen as avowals, i.e. as expressing a complicated attitude which is severely different from wanting to make any claims or to engage in engineering. Accordingly, Wittgenstein proposes to interpret utterances such as “We are not masters of our fate” or “It is God’s will” as expressing our acceptance that not everything in human life can be controlled, or that particular human beings cannot always be made responsible for what has occurred (CV 69). As a matter of fact, for some people, only the development of such an attitude will make it possible to go on living, after a cruel blow of fate has been suffered. To be sure, it is tempting to view this interpretation of “It is God’s will” as a paradigm for the development of a general technique to account for 103

the meanings of the word “God” without taking this word to be a name which refers to a particular transcendent being.9 Wittgenstein even says: “The way you use the word ‘God’ does not show whom you mean, but what you mean” (CV 58). Thus, “I believe in God” need not mean the same as “I believe in the existence of a transcendent being”, but “I am convinced (without being able to give any reasons) that in the end everything will turn out well.” Really what I should like to say is that here too what is important is not the words you use or what you think while saying them, so much as the difference that they make at different points in your life. (CV 97)

That is to say: Religious terms are used in a way that should not be taken too literally. For what they mean depends upon the context and manner in which they are used; for example, to express a highly personal attitude. Similarly, a practice such as prayer need not be understood as a (one-sided) discourse between God and a believer, but may be interpreted as a rulegoverned activity through which a person expresses his or her religious attitude. Furthermore, religious utterances in the indicative mode, being avowals of a certain religious stance, do not necessarily obey the logic of scientific or empirical claims and/or negation: Suppose that someone believed in the Last Judgement, and I don’t, does that mean that I believe the opposite to him, just that there won’t be such a thing? I would say: “not at all, or not always.” / Suppose I say that the body will rot, and another says “No. Particles will rejoin in a thousand years, and there will be a Resurrection of you.” / If someone said: “Wittgenstein, do you believe this?” I’d say: “No.” “Do you contradict the man?” I’d say: “No.” (LC 53)

That does not mean, first of all, that religious practices have no rules of logic or argumentation at all. Wittgenstein rather indicates that there are, as a matter of fact, religious utterances belonging to religious practices that are different from the ones that can be described by what we call propositional or scientific logic. And that is so because the religious utterances of the kind Wittgenstein mentioned are not truth-value relevant claims. 3.5 It is uncontroversial that the leitmotif of Wittgenstein’s Remarks on Frazer’s ›Golden Bough‹ is his insistence that religious or mystical narrations 9. I myself subscribe to such a generalisation in Kober 2005, 243. Thus, the following considerations are also to be understood as a piece of self-criticism.

104

not be regarded as true or false reports of events that actually happened, but rather as the quasi-explanatory efforts of members of a cultural community trying to cope with their fate, or somehow to integrate into their common world-picture favoured and less favoured events — such as terrible earthquakes, a bad harvest, or the unexpected death of a young loved one. Myths or religious narratives may function as frameworks for determining specific types of action in specific circumstances, or for motivating particular attitudes towards events in a natural or social environment. In this sense, myths or religious narratives identify the point at which everyday justifications or explanations come to an end (e.g. how the world came into existence, why there are human beings, or what gives our lives a meaning?), such that argumentation — if there is any — becomes irrelevant. Correspondingly, a particular religious attitude towards natural or social issues may manifest itself as an engagement with a cause, such as nature conservation (cf. ‘preserving creation’, e.g., by involvement with a group such as Greenpeace), or defending the dignity of other persons (perhaps by being an active member of Amnesty International). In other words: The religious attitude of a person is revealed in his or her actions. 3.6 Wittgenstein’s later approach to religion, as outlined so far, may yet be regarded as a descendant of his earlier Tractatus account. It is, so-tospeak, “a theology for atheists [or agnostics; M.K.], an understanding of religion from the outside (as an anthropological phenomenon) which does not accuse it of being either mistaken, unfounded or nonsensical”, as Hanjo Glock nicely put it (Glock 1996, 321). This interpretation can also be characterised as agreeing with slogans such as “the non-descriptive and non-cognitive nature of religion” and “religious terms like ‘God’ do not [need to] refer to entities” (both Glock 1996, 321). To be sure, such an approach to religion does not commit the scientistic fallacy of understanding religious beliefs as naive and false convictions (as James’ prime example of a ›happy agnostic‹ did), since it respects the fact that there exist highly intelligent and trustworthy religious believers. Nevertheless, if this were all that need be said of religion, such an approach would dismiss the whole phenomenon as a mere attitude. For the reasons given below, I do not think that Wittgenstein would endorse such a restricted view.

105

4. Wittgenstein’s Reflections on his own Religious Attitude 4.1 As regards scientifically- or technologically-oriented practices, one may develop one’s own competencies — one’s own epistemic attitudes, so-to-speak10 — by making particular observations, undergoing certain experiences, reading appropriate books, listening to other competent people, or receiving instruction in certain techniques. Having developed a certain basic competence, one may then decide on more or less good ground to accept a particular view or theory in spite of other competing views or theories, or one may decide to go on with further instruction or to rethink some problems anew. However, authentic religious attitudes — as can be seen from James’ examples — are usually not the result of intellectual decisions. One may rather say that these attitudes arise or set in, suddenly or gradually — sometimes in spite of other intentions (one hardly intends to live the life of a ›sick soul‹). A religious attitude may be consciously realised only after it has developed. It is hard to imagine that a member of a present-day, scientifically-oriented community would become acquainted with the narration of Jesus’ resurrection, or with the idea of the Last Judgement, and then decide: “I believe that!” (if it did happen, such a decision would rest upon a confusion). A change of one’s religious attitude — a new “passionately taking up this interpretation” (CV 73) — requires rather a complicated network of smaller changes in several aspects of a person’s life, changes which are usually beyond one’s own control. In other words, a plurality of factors in one’s life can educate you to ‘believing in God’. And experiences too are what do this […,] e.g. sufferings of various sorts. And they do not show us God as a sense experience does an object […] — life can force this concept on us. (CV 97) Amongst other things Christianity says, I believe, that sound [scientific etc., presumably even theological] doctrines are all useless. That you have to change your life [i.e. your attitude towards life]. (Or the direction of your life). (CV 61)

These considerations indicate that it is impossible for one to force oneself into a particular religious attitude by means of intellectually established decisions.

10. Wittgenstein’s account of (unjustified) epistemic attitudes is developed in Kober 2005.

106

4.2 The historical figure Wittgenstein, however, seems to have been in an awkward position as regards this systematically relevant point. Recollections of him by his friends, his own diaries, many of his autobiographically-shaped notes in Culture and Value, as well as his biographies indicate that he was not an altogether happy person. Fania Pascal, for example, describes him in conclusion as “a tragic character” (Pascal 1981, 61), though “he did not suffer from irrational fears and insecurity” (Pascal 1981, 59). Wittgenstein seems to have suffered from the shortcomings of his own intellectual capacity and his own morally- (or religiously-) relevant actions — measured, of course, against his own, perhaps inhumanlyhigh standards and expectations.11 He seems to have craved relief from his own insufficiencies and failures. In religious terms (cf. CV 51f,g), he was longing for redemption (CV 51d). Fania Pascal again characterises him to be “above all a person in search of a spiritual salvation” (Pascal 1981, 62 fn.4). In particular, Wittgenstein may have thought that if only he could adopt or develop his own appropriate religious attitude, the traces of vanity in himself of which he became aware would vanish (cf., e.g., the preface to the Philosophical Remarks, PR 7, or CV 53j, 54f ). Additionally, one may surmise that he believed a religiously-motivated and thus sincere attitude of modesty towards his own achievements (cf. CV 77ab), his friends, and human life in general, would bring him some peace of mind, existentially-speaking. But these desires obviously remained unsatisfied: I am reading: “And no man can say that Jesus is the Lord, but the Holy Ghost.” And it is true: I cannot call him Lord; because that says absolutely nothing to me. […] Because I do not believe that he will come to judge me; because that says nothing to me. And it could only say something to me if I were to live quite differently […].What fights doubt is, as it were, redemption. (CV 37–8; Wittgenstein’s emphasis) ‘Believing’ means, submitting to an authority. [… But:] I cannot kneel to pray, because it’s as though my knees were stiff. I am afraid of dissolution (of my own dissolution) should I become soft. (CV 52+63) 11. Fania Pascal (1981, 50) remembers him making a confession that made her angry: “[…] At one stage I cried out: ‘What is it? You want to be perfect?’ and he [Wittgenstein] pulled himself up proudly, saying: ‘Of course I want to be perfect’. This memory alone would make me burst with the wish to describe the scene.” The adverb “proudly” and the final comment are presumably meant to indicate that Wittgenstein, in Pascal’s eyes, was not at all ironic or selfcynical in this exclamation.

107

It appears as though Wittgenstein were trying to force himself into a (Christian) religious attitude which he — by his own admittance — is not capable of developing — at least insofar as an attitude of submission before the Lord is concerned. It is as if he were longing to undergo a transformation similar to that of the ›twice-borns‹, which, for example, Tolstoy had encountered. It seems as though Wittgenstein had not accepted his own fate, and therefore could not rid himself of Sorge (cf. RUL 231). This, of course, was his fate. For, as was already said above (section 4.1), one’s own religious attitude cannot be changed at will; that is, it is not manipulable by means of an intention or decision. One is rather compelled to accept the attitude one has — the attitude, so-to-speak, that ›one’s fate has been chosen on one’s behalf‹. If one still senses that one’s religious attitude is somehow inappropriate, seeking for religious doctrines or theological theories will be pointless, since they do not function, in the way that scientific theories do, as theoretical foundations for a recipe or prescription. Religious doctrines are not manuals for a better, more satisfactory, or more appropriate life. They are rather interpretations of the events that shape human lives, they are attempts of coming to terms with what we care about. One may say: religious doctrines are intellectually developed avowals which hint at, or manifest, specific attitudes.12 Religion says: Do this — Think like that! but it cannot justify this and it only need try to do so to become repugnant [think of missionaries]; since for every reason it gives, there is a cogent counter-reason. It is more convincing to say: “Think like this” — however strange it may seem. (CV 34)

In other words: In order to become a ›true‹ believer, i.e. someone with a ›true‹ or deeply sincere attitude, one does not need to make intellectually well-founded decisions. If there is any active part on the respective person’s behalf — that is: if the change did not set in by fate –, it is rather a nonintellectual engagement, such as a Kierkegaardian jump.13 If I am really redeemed, — I need certainty — not wisdom, dreams, speculation — and this certainty is faith. And faith is what my heart, my soul, needs, not my speculative intellect. […] One may say: it is love that believes in the Resurrection. (CV 38–9) 12. In saying this, I had in mind Barnes’ reading of Sextus Empiricus’ Outlines of Pyrrhonism as being something like a complicated avowal; cf. Barnes 1983. 13. Cf. Kierkegaard’s Concluding Unscientific Postscript to the Philosophical Fragments (1846).

108

Wittgenstein seemed unable to jump. Or to develop the required ›love‹.14 In fact, he did not even expect any kind of revelation: A miracle [in Humean terms: an event that strikingly deviates from what can be expected according to our natural laws; M.K.] is, as it were, a gesture which God makes. […] It would be an instance if, when a saint has spoken, the trees around him bowed, as if in reverence. — Now, do I believe that this happens? I don’t. (CV 51)

However, I am now in danger of engaging in ›biographical speculation‹; that is, when a biographer without evidence (such as a diary) claims to know what the subject of a biography actually thought. The systematic point of the present discussion is rather this: As if giving grounds did not come to an end sometime. But the end is not an ungrounded presupposition: it is an ungrounded way of acting (OC § 110; cf. OC § 204).

That is to say: If the religious attitude of a person does change, as in the case of Tolstoy, what really matters (in Wittgenstein’s eyes) is not what the arguments are, or what the new religious views or doctrines are, but how the respective person changes her own “way of living” or her own “way of judging life” (both quotes CV 73e) — for example, in which mood she is (happy or unhappy, elated or depressed), which commitments she accepts, how she treats others, or for which course of events she feels responsible. Drury recounts an anecdote which illustrates the main point: Wittgenstein told me that one of his pupils had written to him to say that he had become a Roman Catholic, and that he, Wittgenstein, was partly responsible for this conversion […]. Wittgenstein told me he had written back to say: “If someone tells me he has bought the outfit of a tight-rope walker I am not impressed until I see what is done with it.” (Drury 1976/81, 103)

As a matter of fact, we need not only think of conversions or of becoming a ›twice-born‹, since these are extreme or dramatic cases. There might be smaller changes in a person’s life, and even they need to be made — at least this seems to be Wittgenstein’s demand — sincerely: “A confession has to be part of one’s new life” (CV 16; emphasis M.K.). 14. It is difficult to understand what kind of love Wittgenstein had in mind in the above quote. Is it an acceptance of the world and of other people “without the guarantees”, as Cavell once put it (according to Putnam 1992, 178). Or is it a particular relationship to a transcendent realm?

109

5. Evaluation 5.1 On the grounds of the above discussions, it may be tempting to generate something like ›a Wittgensteinian account of (a philosophy of ) religion‹. But even if one is willing to grant — as I do — that the ›later‹ Wittgenstein indeed developed something like a philosophy of language, of mathematics, or of mind — perhaps even an epistemology15 –, it remains a fact that what has been passed on to us concerning religion are only scattered remarks. Wittgenstein did not elaborate a perspicuous phenomenology of religion, and did not wish to. His remarks often concern very specific issues, such that it is highly questionable whether or not they can be understood as paradigm cases of a particular kind; in other words, it is difficult to determine what the respective kind may be. To my mind, they should not be read as paradigm cases, since conclusions drawn from his small number of examples may result in a ›one-sided diet‹ of problems or issues. The only thesis with regard to religion that Wittgenstein seems to advocate throughout his philosophical life, is that of wishing to distinguish science from religion, and thus the ›logic‹ of science from the ›logic‹ of religion. As already mentioned, the examples he uses (“This is God’s will” or “I believe in the Last Judgement”) in order to illustrate that utterances containing religious terms can be understood as avowals, do not substantiate the conclusion that all religious talk can be transformed salva significatione into an expressive mode of speaking. This manoeuvre would anyhow effect an empirically false claim, since there are in fact religious believers who use such utterances literally; such as, for example, that there will actually be an event in the future that is the Last Judgement. And some speakers certainly do believe their utterances containing the word “God” to be true or false reports, such that “God” is used as a name whose reference is supposed to be a transcendent being. That is to say: Questions of theological convictions, whether or not naive, need sometimes to be distinguished from questions of meaning and use. It would be unfair and arrogant to say: “He should rather mean …”. A more promising hypothesis might be: Only against the highly restricted background of conceiving one aspect of the phenomenon of religion in terms of ›being in a certain attitude‹, can utterances of a religious kind be understood as avowals (although they need not be!). In other religious contexts, religious terms or utterances 15. Cf. Kober 1996 or Moyal-Sharrock 2004.

110

may have other meanings. Wittgenstein in his few remarks certainly did not develop an extended ›grammar‹ of religious terms. It is rather open to us to engage in such a project — if one wishes. One should keep in mind that Wittgenstein, like James in The Varieties of Religious Experience, does not claim to know more than the believers he is thinking of or talking about. Although James made critical comments (as indicated in section 1.2 above), he did not question the sincerity of the reports or their respective belief-contents. He did not aim at enlightening the authors or speakers as to the ›true‹ meaning of their utterances (such as their being avowals as opposed to statements). An attempt to dispose entirely of religious claims through the interpretation of all religious utterances as avowals, would merely reveal one’s own narrow-mindedness; that is, one would thereby uncover only one’s thoroughly secularised view of reality (another form of scientism, I suppose). Such an attitude, lacking humility so-to-speak, would indicate only one’s insensitivity to human tragedy — as in the case of James’ ›happy agnostic‹ above. 5.2 As a matter of fact, there are many aspects of religion which are not thematised by Wittgenstein (and that is why the title of the present contribution is not “Wittgenstein on Religion”). That he put his focus almost entirely on religious attitudes, is due to the fact of this having struck him as most important — first from a personal and then from a philosophical point of view; namely, what does it mean to be in such an attitude? It is authentic sincerity that mattered to Wittgenstein,16 and not necessarily specific propositional belief contents. Are we allowed to conclude that Wittgenstein dismisses any religious doctrine or theological theory? Here is what he writes: Really what I should like to say is that here too what is important is not the words you use or what you think while saying them, so much as the difference that they make at different points in your life. […] Theology that insists on certain words and phrases and prohibits others makes nothing clearer. (Karl Barth)” (CV 97, written in 1950)

Again, from this it does not follow that Wittgenstein devalues Barthian, Protestant or Christian theology as a whole. It only means, I presume, that 16. Drury reports that Wittgenstein told him in 1936 that “he had been reading Newman’s Apologia and that he admired Newman’s obvious sincerity. But when he came to read the last sermon Newman preached to his friends at Littlemore, he thought to himself, ‘I wouldn’t wish to speak to my friends like that’” (Drury 1976/81, 145).

111

Wittgenstein at some moment in 1950 thought that a particular kind of theology is not helpful if one is trying to come to terms with a specific personal attitude, and that somewhere in some specific writing of Karl Barth we can find evidence for this claim. We do not even know what Wittgenstein thought of Karl Barth in particular, if we take into account what Drury claims to remember: […] In another letter he [Wittgenstein] told me that he had been reading a Swiss theologian, Karl Barth. “This writing must have come from a remarkable religious experience.” In reply I reminded him that years ago in Cambridge I had tried to read him something of Barth’s to him, and he had dismissed it as ‘very arrogant’. He did not refer to this again.17

This shows that Wittgenstein’s remarks — even those in Culture and Value — must be read with great care. Drury’s report, however, makes us aware that we should not conclude from two or three remarks on particular issues, that Wittgenstein is making a general claim, e.g., dismissing theological theories altogether. Perhaps he did not like theological theories, or — better — did not like one of them at one particular moment. Like James, he is not interested in reflecting upon them, or in developing one of his own. From this it does not follow that we should not care about them in a philosophy of religion. And from the insight that Wittgenstein — being occupied in a personal struggle to determine his own appropriate attitude — is not interested in theology, it does not follow that he therefore promotes fideism. The non-cognitivistic ›argumentation is irrelevant‹-tone of his remarks might be exclusively effected by his focus on attitudes. Attitudes, indeed — like moods — need not be justified (cf. Kober 2005, 246–9). But from this it does not follow that specific attitudes are all that matters concerning the phenomenon of religion. For even attitudes call forth consequences which need to be reflected upon with regard to content, as attitudes may play a part in decision-making and acting. 5.3 It is perhaps our task to reflect on the use that is made of religious doctrines or theological theories: What is their role in our or other culture’s practices? Why do people engage in religious rituals? Why do they form ecclesiastical organisations? 17. Drury 1976/81, 160, dating the event “January 1940” and referring back to an incident of 1930 that he outlines on p. 134. — Drury’s reports were pointed out to me by Klaus von Stosch.

112

As I take it, Wittgenstein’s later approach to problems of x (with x as, e.g., language, mathematics, knowledge, mental states, …) consists in pointing out that the phenomenon of x, as a matter of fact, is divided up into several aspects and can be considered from several perspectives. We should therefore not expect to develop an overall systematic account of x that adequately covers all aspects of x in a particular order, since — although they are all aspects of x — it might be the case that the phenomenon of x exhibits a family resemblance (that is, all aspects of x need not necessarily have certain ›essential‹ features in common; cf. PI §§ 65–77) implying that additional practices may always be developed anew. Religion, I hold, is of such a kind. For, as James already pointed out (cf. section 1.1), religion consists of — at the least — ecclesiastical organisations, specific rituals, doctrines, and of course, experiences and attitudes. If Wittgenstein in the Tractatus and in his later notes concentrates mainly on attitudes, we should not infer from this that the other aspects are irrelevant. Though it may be true that the rituals, doctrines and organisations of a religious community are due to a somehow ›fossilised‹ tradition and therefore part of a “second-hand religious life” (J 6), they nevertheless belong to our religious practices. As a matter of fact, such diverse religious practices are present — like our life (cf. OC § 559). It is no accident that doctrines, rituals, and organisations belong to religious practices. Remember James’ qualification of “having a religion”, even if the respective person happens to be our prime example of a ›happy agnostic‹: “The writer’s state of mind may by courtesy be called a religion, for it is his reaction [or attitude] to the whole nature of things, it is systematic and reflective, and it loyally binds him to certain inner ideals” (J 92). James and Wittgenstein obviously agree that a religious attitude can be expressed in one’s way of life, which is connected to the problem of living one’s life in a right, good, less good, or even wrong way. Generally, i.e., less individualistically-speaking, if a religious life constitutes a practice, and if that practice is institutionalised within a community,18 whose members share a common sense of the right and wrong ways of partaking in their practices, then there will be rules, spelled out by way of doctrines or theories, and organisational constitutions, which influence or even regulate the religious attitude of that community. Such rituals and doctrines belong ›logically‹ to (our) established practices of religion, and this is also true for ecclesiastical organisations (of course, it does not matter in 18. Kober 2005a discusses what a practice or an institution is.

113

the context of the present discussion if some of these rituals are perplexing, some doctrines incoherent or naive, and some organisations corrupt and dangerous, for these aspects concern only the quality of a certain practice, not its ›logic‹). It is only because Wittgenstein does not consider these other elements, neither developing them nor discussing specific ›rule books‹ for them, that we find no endorsement of any particular religious view — not even fideism — in his writings. Hence it is no surprise that so many theologians and believers find their own view to be compatible with Wittgenstein’s notes, as these are compatible with virtually any respectable religious practice. On the other hand, if people are sincerely in search of their own religious attitude, they will sooner or later feel tempted to articulate what that attitude consists in; that is, they will develop a religious doctrine, including their own interpretation. For these are acts of agreeing with and evaluating one another (for instance through the development of a canon of accepted narrations, such as myths, the Bible, the Koran, or the Torah). The developed doctrine will concern, among other things, the issues of what really matters in life, and what ought to be done or heeded at all costs. Religious utterances may then be used as norms for regulating the life of a community and its members.19 However, even though there may be no ultimate justification for such a doctrine, this does not mean that ›anything goes‹. Religious doctrines — and this is merely a ›grammatical‹ remark — somehow respond to people’s needs and questions (e.g., regarding death); they indicate what should be considered good in respect to human life, and must exhibit a certain coherence. This shows that the phenomenon of religion is not an autonomous practice which can simply be disregarded; on the contrary, it penetrates every aspect of our lives (which is not to say that a religious doctrine or practice may be justified through its being useful or purposeful). Traditional doctrines of already established religious practices, e.g., the practice of Protestant or Catholic confession, do thematise the issues just mentioned — even though their interpretation is difficult and/or controversial. I surmise that it is also these traditional (religious) practices of which Wittgenstein says that “[w]hat has to be accepted, the given, is — so one could say — forms of life” (PI II, 226). Traditional religious practices “are not based on grounds” and are “not reasonable (or unreasonable)” (OC 19. The idea of unjustified moral certainty regulating a community’s practices is spelled out in Wittgensteinian terms in Kober 1997.

114

§ 559). The striking thing is that Wittgenstein himself did not engage in these traditionally established religious practices (though he sometimes attended services). Being a genius, he was looking for his own view — e. g., by critically reading Dostoievski, Tolstoy, Angelus Silesius, Cardinal Newman, or Kierkegaard –, since it is the duty of a genius that he (or she) develops his (or her) own approach.20 This may appear like fideism, or subjectivism, but Wittgenstein does not claim to be making any general systematic point here. It is his personal affair.21

REFERENCES Barnes, Jonathan 1983: “The Beliefs of a Pyrrhonist”, in Elenchos IV, 1983, 5–43. Drury, Maurice O’C. 1976/81: “Some Notes on Conversations with Wittgenstein” and “Conversations with Wittgenstein”, in R. Rhees (ed.): Ludwig Wittgenstein, Personal Recollections, Totowa/NJ: Rowman & Littlefield 1981. Glock, Hans-Johann 1996: A Wittgenstein Dictionary, Oxford: Blackwell. James, William 1902: The Varieties of Religious Experience, ed. by M. E. Marty, Harmondsworth: Penguin 1982. Kober, Michael 1996: “Certainties of a World-Picture. The Epistemological Investigations of On Certainty”, in H. Sluga, D. Stern (eds.), The Cambridge Companion to Wittgenstein, Cambridge: Cambridge University Press 1996, 411–441. –– 1997: “On Epistemic and Moral Certainty: A Wittgensteinian Approach”, in International Journal of Philosophical Studies 5, 1997, 365–381. –– 2005: “›In the Beginning was the Deed‹: Wittgenstein on Knowledge and Religion”, in D. Moyal-Sharrock, W. H. Brenner (eds.): Readings of Wittgenstein’s ›On Certainty‹, Houndmills etc.: Palgrave Macmillan 2005, 225–250. 20. Wittgenstein feels the pressure of being a genius when he reflects in religious undertones that a genius requires talent, courage, and modesty: “There is no more light in a genius than in any other honest human being — but the genius concentrates this light into a burning point by means of a particular kind of lens. [CV 41 …] Genius is not ‘talent and character’, but character manifesting itself in the form of a special talent. [CV 40 …] Courage is always original [CV 42 …] Genius is what makes us unable to see the master’s talent. Only where genius wears thin can you see the talent”. [CV 50] 21. I am grateful to Klaus von Stosch who made very critical, though highly appreciated comments on an earlier version of the present contribution. I suspect that he still disagrees with much of what is said here; cf. e.g., von Stosch 2001 and 2002. — I have also profited from discussion of an intermediate version of this paper at the Philosophy Department of the Federal University of Santa Maria, Brasil.

115

–– 2005a: “Soziales Handeln und die Frage der Verantwortung”, in M. Kober (ed.), Soziales Handeln, Beiträge zu einer Philosophie der 1. Person Plural, Ulm: Humboldt-Studienzentrum 2005, 63–84. McGuinness, Brain F. 1988: Wittgenstein, A Life: Young Ludwig 1889–1921, London: Duckworth. Moyal-Sharrock, D. 2004: Understanding Wittgenstein’s ›On Certainty‹, Houndmills etc.: Palgrave Macmillan. Newman, Francis W.: The Soul; Its Sorrows and Its Aspirations, 3rd. ed. 1852. Pascal, Fania 1981: “Wittgenstein: A Personal Memoir”, in R. Rhees (ed.): Ludwig Wittgenstein, Personal Recollections, Totowa/NJ: Rowman & Littlefield 1981. Putnam, Hilary 1992, Renewing Philosophy, Cambridge/Ma.: Harvard University Press. Russell, Bertrand 1919: Introduction to Mathematical Philosophy, London: George Allen & Unwin. von Stosch, Klaus 2001: Glaubensverantwortung in doppelter Kontingenz, Untersuchungen zur Verortung fundamentaler Theologie nach Wittgenstein, Regenburg: Friedrich Pustet. –– 2002: “Grundloser Glaube? Zur Glaubensverantwortung nach Wittgenstein”, in Freiburger Zeitschrift für Philosophie und Theologie 49, 2002, 328–346. Weber, Max 1904: “‘Objectivity’ in the Social Sciences and Social Policy”, in Max Weber, The Methodology of the Social Sciences, New York: Free Press 1949, 50–112.

116

Grazer Philosophische Studien 71 (2006), 117–138.

WITTGENSTEIN AND JOHN HENRY NEWMAN ON CERTAINTY Wolfgang KIENZLER University of Jena Summary Wittgenstein read and admired the work of John Henry Newman. Evidence suggests that from 1946 until 1951 Newman’s Grammar of Assent was probably the single most important external stimulus for Wittgenstein’s thought. In important respects Wittgenstein’s reactions to G. E. Moore follow hints already given by Newman. If you know that here is a1 hand, we’ll grant you all the rest. (When one says that such and such a proposition cannot be proved, of course that does not mean that it cannot be derived from other propositions; any proposition can be derived from other ones. But these may be no more certain than it is itself.) (On this a curious remark by H. Newman.) (OC § 1)

This is the first remark in a notebook published as On Certainty in 1969. Since then there have been a number of attempts to compare Wittgenstein and John Henry Newman on a more or less general level, but no one has yet identified which (or what kind of ) remark Wittgenstein had in mind here.2 References to Newman are conspicuously absent from the great bulk of literature devoted to Wittgenstein in general, and he does not feature in any discussion of On Certainty in particular.3 This paper explores, firstly, 1. The printed translation has “one hand” but this gives the wrong suggestion that Wittgenstein is thinking about numbers here. 2. FitzPatrick (1978) is a notable early exception, but in spite of his title “Eine komische Bemerkung H. Newmans” he makes no attempt to determine which remark Wittgenstein has in mind. 3. Curiously, the index to the Blackwell edition that was prepared by G. Hallett contains Moore, New York, and even Goethe (who is quoted but not mentioned), Wilhelm Meister and Saturn, but not Newman. Monographs devoted to On Certainty do not even mention Newman (see e.g. Kober 1993, Stroll 1994, and Rhees 2003 [this is also true for Moyal-Sharrock 2004; note by the editor M.K.]). Hallett (1979, 770) lists Newman among the “authors

Wittgenstein’s reading of Newman, secondly, it discusses some features of Newman’s thought, and finally, it traces the first remark of On Certainty back to Newman’s writings. It concludes by elucidating the specific kind of interest that Newman’s work can have for Wittgenstein studies. 1. Wittgenstein and John Henry Newman There is no quick and straightforward proof that Wittgenstein alludes to John Henry Newman in the first remark of On Certainty. He refers to “H. Newman” and not to “J. H. Newman” and fails to supply any further information. The name “Newman” occurs only once more in Wittgenstein’s written work. In MS 117, 208 he notes: “A description, not an explanation (Newman) leads to clarity here.”4 This remark, however, refers almost certainly to M. H. A. (Max) Newman (1897–1984), who was a colleague of Wittgenstein and a Fellow of Christ’s College in Cambridge.5 Wittgenstein wrote down this latter remark about “Newman” between February 2 and March 13, 1940. On February 19 of the same year Wittgenstein gave a lecture at the Trinity Mathematical Society. The minutes of the discussion record: “After many unsuccessful attempts, the President finally took advantage of a momentary lull in the Newman-Wittgenstein controversy to thank the professor for his lecture and to declare the meeting social.” This rules out the possibility that the remark in question should refer to J. H. Newman.6 The precise nature of Wittgenstein’s brief reference to J. H. Newman in On Certainty will be discussed below. We know from various sources that Wittgenstein was familiar with Wittgenstein knew or read”, commenting that in view of his affinity to Kierkegaard it would seem natural for Wittgenstein to dislike the theological writings of Newman. He also remarks: “Miss Anscombe has said that nobody understood Wittgenstein’s views on religion” (Hallett 1979, 426, commenting on PI § 373). Newman is also absent from books like Wittgenstein on Religion (Phillips 2001). 4. “Eine Beschreibung, nicht eine Erklärung (Newman) leitet hier zur Klarheit.” A different phrasing of the same remark is printed in RFM III § 78: “What we want is to describe, not to explain.” 5. Alan Turing was a student of Wittgenstein’s as well as of Newman’s, and for a while both were on Alice Ambrose’s Ph.D. examination committee. 6. For the documents and some more on the Wittgenstein / Max Newman relation, see PPO 374–5. The opposite guess, that the remark in On Certainty alludes to Max Newman, is excluded by the use of the initial. — Incidentally, the remark on explaining and describing would fit some remarks that occur in J. H. Newman’s work (see below).

118

several of John Henry Newman’s writings. In 1936 he told Drury that he “had been reading Newman‘s Apologia and that he admired Newman’s obvious sincerity” (CD 130). Norman Malcolm relates in his Memoir of Wittgenstein: “He disliked the theological writings of Cardinal Newman, which he read with great care during his last year at Cambridge” (Malcolm 1984, 59). This remark is puzzling: Why would Wittgenstein read Newman “with great care” if he disliked what Newman wrote? While it seems natural that he would sharply disagree with most of Newman’s views on religion, the remark suggests that Wittgenstein must have found something of great interest in Newman’s work if he read his books in spite of this. Cyril Barrett supplies some further indirect evidence of Wittgenstein’s interest in Newman: Yorick Smythies, a former student of Wittgenstein’s, told me that Wittgenstein had said of J. H. Newman’s Grammar of Assent that Newman thought the grammar was supporting the Christian faith whereas, in fact, the faith was supporting the grammar, as if it were suspended from a balloon. Thus assent to religious beliefs is an ascent or an elevation rather than the result of an upward climb. The driving force that impels this upward thrust is love of Christ and trust in his redemptive power. (Barrett 1991, 181)7

The most detailed account of such conversations, however, is given in O. K. Bouwsma’s notes of 22 August 1949. They deserve to be quoted at length: On the way as we passed the Jewish synagogue, he remarked that he did not understand modern Judaism. He did not see what could be left of it since sacrifice was no longer practiced. And now? What was left was too abstract. […] In any case a religion is bound up with a culture, with certain externals in a way of life, and when these change, well, what remains? Then he went on to cite the Oxford Movement as a symptom of the same hollowness, lifelessness, in the Anglican church. I didn’t understand all these things.[8] […] Later he asked me, had I read Newman? He was much impressed by 7. The statement is somewhat puzzling because Newman nowhere explicitly discusses the relation of grammar and faith. It seems to argue that Newman tries to give something like a rational construction of propositions to make the truth of Catholic belief evident while in Wittgenstein’s view nothing we say can be more than an expression of faith. William James quotes Newman’s definition of theology as “the Science of God, or the truths we know about God, put into a system, just as we have a science of the stars and call it astronomy, or of the crust of the earth and call it geology” (James 1985, 435, quoting from The Idea of a University II, § 7). 8. Bouwsma apparently did not know about Newman and the Oxford Movement (see below).

119

Newman. Kingsley accused him of insincerity. [9] But Newman was sincere. He, Wittgenstein, had read Grammar of Assent, too. That was puzzling. How a man of such learning and culture could believe such things! Newman had a queer mind. Later I pressed him for an explanation. Did he mean by ‘queer’ that a man like Newman should have become a Roman Catholic? Oh, no. My best friends and the best students I had are converts. What is queer about Newman is the kind of reasons he gives for becoming a Roman Catholic. On miracles, Newman cites the case of Christians, who taken by savages had their tongues cut out, and yet they could speak. [10] He gives a natural explanation for this — if the tongue is only half cut off a man cannot speak, but if wholly cut off a man still can — but Newman then goes on to say that it may nevertheless have been a miracle. Again: The pope excommunicated Napoleon. Napoleon said he didn’t care so long as his soldiers’ weapons did not fall out of their hands. Some years later in Moscow, in Russia, this is literally what happened. [11] What was Newman doing? He argued that miracles occur still? How? What God has done once he contrives to do — usually. This is the sort of thing that is so queer in Newman. Later […] he remarked that twenty years ago he would have regarded Newman’s action as incomprehensible, as insincere perhaps. But no more. […] He came gradually to see that life is not what it seems. […] It is like this: In the city, streets are nicely laid out. And you drive on the right and you have traffic lights, etc. There are rules. When you leave the city, there are still roads, but no traffic lights. And when you get far off where there are no roads, no lights, no rules, nothing to guide you. It’s all woods. And when you return to the city you may feel that the rules are wrong, that there should be no rules, 9. Newman wrote his Apologia in response to Charles Kingsley who had accused him of insincerity. 10. Newman quotes from a Persian source: “I am not an anatomist, and I cannot therefore give a reason, why a man, who could not articulate with half a tongue, should speak when he had none at all; but the facts are as stated” (Apologia 414). He then goes on to discuss the matter in further detail. Similar reports are collected in Newman’s essay The Miracles of Early Ecclesiastical History, Section 9: The Power of Speech continued to the African Confessors deprived of their Tongues (1842/43). 11. Newman writes: “‘What does the Pope mean’ said Napoleon to Eugene, in July 1807, ‘by the threat of excommunicating me? Does he think the world has gone back a thousand years? Does he suppose the arms will fall from the hands of my soldiers?’ Within two years after these remarkable words were written, the Pope did excommunicate him, in return for the confiscation of his whole dominions, and in less than four years more, the arms did fall from the hands of his soldiers; and the hosts, apparently invincible, which he had collected were dispersed and ruined by the blasts of winter” (Newman 1900, 215–16).

120

etc. [12] […] Of course, a man need not argue his religious beliefs. Newman did. Once he does this he must argue clearly — soundly. But one may believe without argument. (CB 33–37)13

Finally, in notes taken from Wittgenstein’s lectures in 1946 we find him discussing the question: “What is thinking?”: To say thinking is indefinable will not do. Why? No concept is indefinable. We must ask: what are its connections? We might have a definition, but it may in no way be useful. Take the primitive propositions of Cardinal Newman.[14] In some cases, inference is important, in others not. Where primitive propositions include everything, inference is not interesting. What is interesting? It is the use of a word. In it are included the inferences we make from it. However, thinking is not learnt by definition — except when learning a foreign language. If so, it should be the easiest enquiry. Everything is known about the use. Yet it is an excruciatingly difficult enquiry. Therefore, we must be hopelessly wrong somewhere as to what question we are asking or in regard to the answers. [15] (LPP 120)16 12. This fascinating passage approaches a topic that could be dubbed On Uncertainty. 13. Bouwsma’s own essays on questions of religion and philosophy of religion, many of them discussing Kierkegaard, are collected in the posthumous volume Without Proof or Evidence (Bouwsma 1984). Wittgenstein’s first disciples apparently had no clear idea of Wittgenstein’s interest in Newman: Elizabeth Anscombe said in 1977 that Wittgenstein had read the Apologia but not Grammar of Assent, but could report nothing more (FitzPatrick 1978, 42). Bouwsma, Smythies and Wittgenstein for a while entertained plans to read a book of Dostoevsky together (Bouwsma 1984, xi). 14. Newman does not speak of “primitive propositions” (as does Russell), but this is probably a slip in hearing. It seems that Wittgenstein is referring to Newman’s “first principles”: “These socalled first principles, I say, are really conclusions or abstractions from particular experiences; and an assent to their existence is not an assent to things or their images, but to notions, real assent being confined to the propositions directly embodying those experiences. Such notions indeed are an evidence of the reality of the special sentiments in particular instances, without which they would not have been formed; but in themselves they are abstractions from facts, not elementary truths prior to reasoning” (Grammar of Assent 65). Newman also discusses one specific example: “Thus the constitutional formula, ‘The king can do no wrong’, is not a fact, or a cause of the Constitution, but a happy mode of bringing out its genius, of determining the correlations of its elements, and of grouping or regulating political rules and proceedings in a particular direction and in a particular form” (Grammar of Assent 67). This brings out nicely the insight shared by Wittgenstein that apparently fundamental or primitive propositions have a role in the language game much different from what one in inclined to suppose. Furthermore, Newman and Wittgenstein agree in viewing such “first principles” as modes of representations. The connection between Newman’s discussion of first principles and On Certainty is stressed in Ferreira 1986, 160–3. 15. This last sentence echoes a motto from Hertz which Wittgenstein liked to quote (see BB 169). 16. Wittgenstein’s lecture, October 11, 1946. Newman’s name occurs only in one of the

121

These remarks from the first lecture of the series explore some conceptual difficulties that are quite similar to the first remark in On Certainty. One should note that Wittgenstein twice commenced a series of philosophical investigations using an insight of Newman’s as a point of departure. This is so in his 1946 lectures on the philosophy of psychology and also in his 1950 notes on certainty. We can see so far that in conversation Wittgenstein discussed above all Newman’s religious views which he regarded to be quite interesting albeit basically mistaken. In his lecture, however, he used an example from Newman relating to his own logical or grammatical investigations. Although several of Wittgenstein’s students and a number of early Wittgenstein scholars were either Catholic or even converts like Newman himself, they have failed to explore his interest in Newman. For example, Anthony Kenny gave one of the Centenary Lectures on Newman in 1990, supplying ample quotes to illustrate, and eventually criticize, Newman’s attempts to understand the justification and the grammar of religion. However, he did not make the connection to Wittgenstein (Kenny 1992). On the other hand, P. J. FitzPatrick seems to have been the first to note some striking similarities between On Certainty and Newman’s Grammar of Assent. In his short piece he notes with great insight that Newman’s position is “superior to, say, the response offered by G. E. Moore” (FitzPatrick 1978, 44). 2. Newman’s Question and Paradox17 Newman’s life was one of continuous religious struggle from his beginnings in Oxford around 1820 until his conversion to Catholicism in 1845. From 1833 to 1841 he played a leading role in the Oxford Movement three sets of student’s notes. 17. There exists hardly any literature on Wittgenstein and Newman written from a philosophical point of view, while Wittgenstein is mentioned rather often in writings on Newman. Some items are listed in Bottone 2003, 5. That paper tries to find views common to Newman and Wittgenstein, but Wittgenstein is not so much interested in Newman’s general views but rather in some special methods of investigation. For a partial but balanced account of Newman’s work see Ker 1990, including sections on The educator, The philosopher, The preacher, The theologian, and The writer. Ker remarks: “Newman’s treatment of doubt anticipates Wittgenstein’s fundamental insight into the absurdity of universal skepticism, since to doubt everything is to nullify the language of doubt itself ” (Ker 1990, 71; however, this is not quite the point of convergence, because Wittgenstein is less interested in such a general thesis but rather in Newman’s

122

that investigated the historical and religious foundations of the Anglican Church in opposition to modernist and liberal tendencies. In his autobiographical Apologia pro vita sua: Being a History of his religious Opinions (1864)18 Newman defends himself against charges of insincerity, and he relates and interprets those investigations and struggles that eventually led him to conversion. In this book he does not primarily address the question of the truth, concerning his Catholic faith, rather he focuses on the issue of his sincerity in passing from the Anglican to the Catholic Church.19 Newman reflects on his earlier conceptions regarding the nature of religious belief and religious truth in terms of probabilities in the following way: Speaking historically of what I held in 1843–4, I say, that I believed in God on a ground of probability, that I believed in Christianity on a probability, and that I believed in Catholicism on a probability, and that all three were about the same kind of probability, a cumulative, a transcendent probability, but still probability; inasmuch as He who made us, has so willed that in mathematics indeed we arrive at certitude by rigid demonstration, but in religious inquiry we arrive at certitude by accumulated probabilities — inasmuch as He who has willed that we should do act, cooperates with us in our acting, and thereby bestows on us a certitude which rises higher than the logical force of our conclusions. (Apologia 199)

Newman also extensively studied early ecclesiastical history which was of tremendous importance for his intellectual and religious development.20 In passing he also sketched his view of scientific growth: “I am far from denying that scientific knowledge is really growing, but it is by fits and starts; specific examples, without the generalizations). Commenting on The Philosophical Notebook of John Henry Newman, Ker remarks: “Although the notebook is fragmentary and was never intended for publication, it contains some penetrating and suggestive observations that are sometimes uncannily similar to the gnomic utterances of Wittgenstein, who appreciated Newman’s originality” (Ker 1998, 823). 18. The book was published in installments in 1864 and again as a book, including some revisions, in 1865. Most reprints reproduce the 1865 edition. In 1913 a combined edition was published. 19. Newman tries to refute the charge that he could not possibly be sincere as a preaching Anglican as well as a preaching Catholic. 20. Alfred North Whitehead had met Newman in 1889 and was much impressed by his personality. He lists Newman’s Essay on the Development of Christian Doctrine next to Gibbon’s Decline and Fall of the Roman Empire as one main source of influence in the Preface to his Adventures of Ideas. He also quotes Newman’s motto from the Grammar of Assent: “This saying, quoted by Cardinal Newman, should be the motto of every metaphysician” (Whitehead 1948, 7 and 339).

123

hypotheses rise and fall; it is difficult to anticipate which will keep their ground” (Apologia 262).21 After his conversion the nature of Newman’s struggle with religion changed: instead of pursuing the question of how we can find true religion, he now addressed the even more fundamental issue of by what standards we can evaluate our justifications for religious faith. These investigations culminated in his main work on the philosophy of religion: An Essay in Aid of a Grammar of Assent (1870).22 Newman’s reasoning can be seen as a type of nominalist empiricism that takes individual facts as “real” and discounts everything general as only “notional”.23 This empiricism goes back to Newman’s time at Oxford where he studied logic under Richard Whately.24 Whately too had tried to give an empiricist account of the truth of Christian religion. He was the author of Elements of Logic (1826),25 which became the standard logic textbook of the time. Newman took an active part in writing this book, as Whately mentions in the Preface (Whately 1848, ix). One of Whately’s main claims was that the truth of Christian religion can be seen as evident in the miracles from the New Testament, which he took to be a collection of factual reports beyond any reasonable doubt. In order to underline this view Whately wrote a satirical pamphlet, Historic Doubts Relative to Napoleon Bonaparte, anonymously published in 1819.26 Employing the figure of a reductio ad absurdum, Whately put forward the 21. This attempt to interpret the history of science on the model of ecclesiastical history may well deserve further attention. 22. Newman uses the term “grammar” in his title to indicate that he is trying to get clear about some elementary steps, not to indicate a “grammatical investigation” in Wittgenstein’s sense. However, as a matter of fact Newman conducts quite extensive “grammatical investigations”. The words „An Essay in Aid of“ are also intended to indicate that Newman is not offering a “theory” or a scholarly work. 23. Chapter Four of his Grammar of Assent is about Notional and Real Assent. Newman’s notorious “argument from conscience” takes the fact of our conscience as evidence for the existence of God, but this line of thought cannot be discussed here. 24. Whately’s work has been almost entirely forgotten except for his place in the history of early 19th century logic (see Evra 1984 for details on this aspect). Whately served as archbishop of Dublin from 1831 to his death in 1863. 25. While W. Hamilton declared it ridiculous that Whately took logic to be about language, Elements of Logic was one of the first works to practice analytic philosophy of language. The book also contains extensive exercises in logical analysis of language based on ordinary language as a standard for what makes sense. See especially Appendix I “On certain terms which are peculiarly liable to be used ambiguously” and Appendix III “Praxis of Logical Analysis”. A comparison of Whately and Wittgenstein will have to wait for another occasion. 26. See the editor’s introduction to Whately 1985 for more background information on this remarkable piece of writing.

124

view that it would be every bit as silly to doubt the miracle reports of the Scriptures as it would be to doubt or even deny Napoleon’s existence. Even though one might argue over the precise details of such an historical event as represented by different reports, this in itself is not sufficient to deny its occurrence. In a footnote in his Elements of Logic Whately summarizes his views as follows: “In [the pamphlet] it was shown that the existence of that extraordinary person could not, on Hume’s principles, be received as a well authenticated fact; since it rests on evidence less strong than that which supports the Scripture-histories” (Whately 1848, Book I, § 3).27 Whately also gave an example of the surprising reversal of doubt and rational inquiry where the exercise of doubt appears as an instance of credulity: And there may even be cases in which doubt itself may amount to the most extravagant credulity. For instance, if anyone should “doubt whether there is any such country as Egypt,” he would be in fact believing this most incredible proposition; that “it is possible for many thousands of persons, unconnected with each other, to have agreed, for successive ages, in bearing witness to the existence of a fictitious country, without being detected, contradicted, or suspected.” (Whately 1848, Book II, § 3)

Unfortunately, Whately did not elaborate any more on this paradoxical situation but just stated it as a fact before continuing his exposition. While he pointed out this paradox, he left it for Newman to analyze it in more detail. When he discusses Hume’s notion of “experience” Whately also points to cases where “experience and common-sense” are in opposition to “theory”, whereas Hume had claimed that all common-sense and theory were founded on experience: In former times, men knew from experience that the earth stands still, and the sun rises and sets. Common-sense taught them that there could be no antipodes, since men could not stand with their heads downwards, like flies on a ceiling. Experience taught the King of Bantam that water could not become solid. (Whately 1826, App. I, “Experience”)28 27. Whately fiercely attacks the views of his contemporaries “representing the ‘Faith’ that is required of a Christian as wholly independent of evidence” and repeatedly emphasizes that “all the narrative, and all the teaching, of the New Testament, presupposes evidence as the original ground on which belief had been all along demanded” (Appendix III). This evidence is always empirical evidence, especially as to the reality of miracles. He rejects any attempt to give demonstrative proofs for the existence of God, because in his view, no fact can be proven by reason. Newman follows Whately in this regard, too. 28. In a note to his Essay on Miracles Hume remarked: “The inhabitants of Sumatra have always seen water fluid in their own climate, and the freezing of their rivers ought to be deemed

125

For Whately the King of Bantam was quite rational (on Hume’s account) when he refused to listen to anyone who tried to tell him that water could turn solid: “Not only was the King of Bantam justified (as Hume admits) in listening to no evidence for the existence of ice, but no one would be authorized on this principle to expect his own death.”29 In Whately’s view, Hume, who rejected reports about miracles, and the king of Bantam, who rejected reports about snow and ice, made the same mistake in equating “experience” with their own personal experience. This, however, is a deviation from the manner in which this word is actually used in language, and Whately goes on to distinguish “personal experience” from more general conceptions of experience that would include reports of incidents other people have experienced. On the basis of such a wider and more natural concept of experience it is rational to listen critically to reports and to develop a broader view of world events. While Newman adhered to Whately’s general empiricism he did not apply it to the Scriptures as records of historical facts, but rather to the development of the church through the centuries.30 His studies of Arians of the Fourth Century (published in 1833) and of other aspects of early church history convinced him in the end that there could be only one true church in history. For Newman, the Anglican Church could not convincingly hold this claim but was rather to be compared to the Arian heresy. The main problem Newman investigates in his Grammar of Assent is the question of evidence that can be produced for religious belief.31 He takes his cue from Locke’s conception of the relationship between probability and certainty: “These probabilities rise so near to certainty, that they govern our thoughts as absolutely, and influence all our actions as fully, as the most evident demonstration; and in what concerns us, we make little or no difference between them and certain knowledge. Our belief thus grounded, rises to assurance” (Locke, Essay, IV, XVI, 6; quoted in Newman’s Grammar a prodigy.” It may be noted that Bantam is on the island of Java, not Sumatra. (Thanks to Temilo van Zantwijk for pointing this out.) Maybe Hume is to be credited with opening up this line of thought, but Whately seems to have been the first who became aware of this “paradox of doubt and credulity”. 29. In offering this example Whately subverts the alleged rationality of the King of Bantam, who is following the principle to believe only what he had encountered in his own personal experience, because it is quite obviously rational to expect one’s own death. 30. Newman also was very much interested in miracles, but he mainly discussed those reported in the early centuries A.D. 31. He is also very much interested in the phenomenon of changing religious beliefs. This topic, too, will surface in On Certainty.

126

of Assent 161). According to Locke we can reach a level of probability that can serve on all practical accounts as well as certainty.32 Newman declares this seemingly self-evident conception to be plainly wrong because even in everyday life there are many examples where we are absolutely certain about something without this certainty being the result of the gradual growth that Locke claims to be necessary:33 We are sure beyond all hazard of a mistake that our own self is not the only being existing; that there is an external world; that it is a system with parts and a whole, a universe carried on by laws; and that the future is affected by the past. We accept and hold with an unqualified assent, that the earth, considered as a phenomenon, is a globe; that all its regions see the sun by turns; that there are vast tracts on it of land and water; that there are really existing cities on definite sites, which go by the names of London, Paris, Florence, and Madrid. We are sure that Paris or London, unless suddenly swallowed by an earthquake or burned to the ground, is today just what it was yesterday, when we left it. We laugh to scorn the idea that we had no parents though we have no memory of our birth; that we shall never depart this life, though we can have no experience of the future.”34 (Grammar of Assent 177)

The most important example, however, is religious belief. In his earlier writings Newman had criticized Locke along the line that in some cases it is possible to pass through probabilities in order to finally reach absolute certainty (as in the passage quoted above). After 1866 he reverses his perspective and starts with the description of a certain type of certitude that he calls “assent”. His Grammar of Assent, first published in 1870, tries to spell out and explain this basic insight. He notes in 1866: “You are wrong in beginning with certitude — certitude is only a kind of assent — you should begin with contrasting assent and inference” (Newman, quoted in Ward 1912, vol. 2, 278). Newman contrasts “certitude”, being the state of mind associated with propositions having “certainty”, and “assent”. Still following Locke on this, Newman believes that certitude can be attained only on the basis of collecting probabilities. Certitude therefore is a fundamentally intellectual achievement. This also means that it is not attainable by everybody. By 32. Locke does not distinguish religious from secular cases. 33. Newman’s conception of reason-based certainty, however, still owes very much to Locke. 34. In this passage Newman passes from more general considerations common in much philosophical writing (e.g. from the philosophers of common sense) to his own peculiar way of giving very detailed examples.

127

contrast, assent is the original and unconditional affirmation of something. Newman’s discovery, which he at length tried to express in his work, was that simple assent and not reason-based certainty is the most basic as well as essentially sufficient element in religious belief. This point of view is connected to an important question, which Newman put this way: “What is the difference between faith and prejudice?”35 As it turned out there is no logical mark that could help us to distinguish with certainty between the two.36 Newman’s examples of assent are religious as well as everyday beliefs. Before we acquire the capacity to doubt, we already have a set of very firm beliefs that we did not gain by way of reflection but through our upbringing or just through everyday life. For Newman, the “discovery” of this basic phenomenon of assent and the elucidation of its logical structure is his most important philosophical insight.37 The notion of assent is fundamental to Newman’s philosophy of religion; because it enables him to answer the question that had occupied him for decades: Is natural, traditional religious belief as fully justified as a more sophisticated religious belief that is backed by sufficient knowledge of the foundations of one’s own religion? On Locke’s conception this could not be the case, because no belief can be stronger than the reasons and probabilities actually given for it. In the case of simple, straightforward believers this would imply that such persons are not really in full possession of their religion unless they can give adequate reasons for their belief. For Newman, however, simple assent is really the best kind of religious belief because it goes straight to the facts whereas informed certainty only rests on abstract reasoning. As Newman explains: “To give a real assent to [a dogma] is an act of religion; to give a notional, is a theological act. It is discerned, rested in, and appropriated as a reality, by the religious imagination; it is held as a truth, by the theological intellect” (Grammar of Assent 98). This attitude is also expressed in the motto Newman chose 35. Quoted in Ward 1912, vol. 2, 245. 36. According to Newman, it is only the entire body of evidence, but never single isolated items, that will help us to finally find the truth. 37. In an important sense Newman considered his Grammar of Assent as much a work on logic as did Wittgenstein his remarks On Certainty, see e.g. OC §§ 51, 56, 82, 501, and 628, where he insists that he is doing work in logic. It may be worth noting that Newman’s original plan for his book contained three parts, just like traditional treatises on logic: Assent — Certitude — Proof (Newman 1973, 104, letter, dated July 27, 1868). Curiously, both works have been widely misunderstood as to their true nature. Comparing both works may contribute to a better appreciation of the logical as opposed to epistemological character of each.

128

for his book from St. Ambrose: “Non in dialectica complacuit Deo salvum facere populum suum.”38 Newman aims to clarify religious belief but he also finds abundant examples of assent in secular everyday life. He wants to convince his readers that there are obvious cases of simple assent both in everyday and in religious life. Especially these everyday examples are of particular interest from a Wittgensteinian perspective. In one example, Newman relates the story of a judge who was appointed to an important but unfamiliar post which required him to make far-reaching decisions. He consulted a friend on this who advised him as follows: “He bade him always lay down the law boldly, but never give his reasons, for his decision was likely to be right, but his reasons sure to be unsatisfactory” (Grammar of Assent 303). This striking imbalance between the quality of decisions or beliefs on the one hand and the justifications for these judgments on the other is “a phenomenon which seems to me as undeniable as to many it may be perplexing” (Grammar of Assent 293). It is important to note, however, that Newman does not talk about cases where decisions or propositions are actually unjustified.39 The peculiar phenomenon he is pointing out concerns the nature of what we might call sound judgments where we may not be able to provide adequate reasons for our judgments. Notwithstanding this we “have the feeling” that an adequate justification could eventually be given if we had the time and energy to think and investigate the matter all the way through. In the end, Newman provides a kind of explanation for this phenomenon by introducing the “illative sense” (Grammar of Assent, ch. 9). What he offers here is a description rather than an explanation of the conceptual situation: “My object in the foregoing pages has been, not to form a theory which may account for those phenomena of the intellect which they treat, viz. those which characterize inference and assent, but to ascertain what is the matter of fact as regard them“ (Grammar of Assent 343). Newman provides two striking examples of what he tries to capture with the “illative sense”.40 The first one concerns “Newton’s perception of 38. “It has pleased God to lay the salvation of his people not into dialectics.” In other words: “Salvation needs no theory.” Newman quotes the slogan repeatedly; historically he links it to the question of whether the church fathers needed Aristotelian philosophy (“dialectics”) in order to properly explain Christian religion. 39. Wittgenstein expresses a very similar idea in the following remark: “To use a word without justification does not mean to use it illegitimately” (PI § 289). The Anscombe translation has “… to use it without right”, but this obscures the point of the passage. 40. Newman also discusses in turn “formal inference”, “informal inference”, and “natural

129

truths mathematical and physical, though proof was absent” (Grammar of Assent 333), and the second one “Napoleon’s genius in reasoning, by which he was enabled to look at things in his own province, and to interpret truly, apparently without any ratiocinative media” (Grammar of Assent 333). These examples demonstrate the occurrence of excellent results in the absence of sufficient formal justification.41 Another example Newman provides is the belief that Great Britain is an island: We are all absolutely certain, beyond the possibility of doubt, that Great Britain is an island. We give to that proposition our deliberate and unconditional adhesion. There is no security on which we would be better content to stake our interests, our property, our welfare, than on the fact that we are living on an island. We have no fear of any geographical discovery, which may reverse our belief. We should be amused or angry at the assertion, as a bad jest, did any one say that we were at this time joined to the main-land in Norway or in France, though a canal was cut across the isthmus. We are as little exposed to the misgiving, “Perhaps we are not on an island after all”, as to the question, “Is it quite certain that the angle in a semi-circle is a right-angle?” It is a simple and primary truth with us, if any truth is such; to believe it is as legitimate an exercise of assent, as there are legitimate exercises of doubt or of opinion. This is the position of our minds towards our insularity; yet are the arguments producible for it (to use the common expression) in black and white commensurate with this overpowering certitude about it? Our reasons for believing that we are circumnavigable are such as these: — first, we have been so taught in our childhood, and it is so in all the maps; next we have never heard it contradicted or questioned; on the contrary, every one whom we have heard speak on the subject of Great Britain, every book we have read, invariably took it for granted, our whole natural history, the routine transactions and current events of the country, our social and commercial system, our political relations with foreigners, imply it in one way or another. Numberless facts, or what we consider facts, rest on the truth of it; no received fact rests on its being otherwise. […] The question is, Why do I believe it myself? A living statesman is said to have fancied Demerara an island; his belief was an impression; have we perinference” (Grammar of Assent, ch. 8). 41. It is important to note that this works only in areas of specific familiarity. Newton and Napoleon are examples of highly qualified and trained experts. The illative sense has nothing to do with guessing. There are some interesting similarities to thought experiments as described by Ernst Mach that cannot be explored here.

130

sonally more than an impression about Great Britain, like the belief, so long and so widely entertained, that the earth was immovable, and the sun careered around it? I am not insinuating that we are not rational in our certitude; I only mean that we cannot analyze a proof satisfactorily, the result of which good sense actually guarantees to us. (Grammar of Assent 294)

Newman’s description of the situation is concrete, detailed and convincing:42 There are many examples of extremely firm beliefs concerning everyday knowledge and situations where doubt is not answered by some good argument or some convincing evidence but by a joke or some expression of incredulity. We may even doubt the sanity of the person asking. It is a characteristic feature of such situations that the convictions in question are connected to many other everyday convictions: they are woven deep into the fabric of our world-picture (as Wittgenstein says). These are not isolated, “atomic”, facts that are believed to be true or false but belief without formal justification underpins all of our acting and judging.43 This becomes obvious in the example of Demerara: It is quite possible to believe that this (former) British colony is an island and not part of the main land44 because a false belief of this kind is (for most English persons of Newman’s time) of little consequence. An English person who is wrong about a small and unimportant colony is simply ill informed (and only if she is a politician a bit ridiculous). By contrast, some English persons who did not believe that Great Britain is an island would, so to speak, cease to be English, separating their beliefs deeply from their surroundings. Newman also stresses that beliefs of this kind are not just a matter of theory but mainly one of practical importance: The British act as inhabitants of an island. Newman draws a further comparison with mathematics: The certainty regarding the insularity of Great Britain is of exactly the same sort as our most basic certainties in geometry.45 He also maintains that such a belief is in a certain sense quite rational although it was not acquired in a “rational way” but simply by living in certain surroundings and on the example of older people without any explicit justification. Maybe without 42. Actually, Newman’s examples are much more detailed than those that Wittgenstein provides. 43. They are of the kind that would force us “to give up making judgments altogether” (as Frege expresses in the Preface to his Basic Laws of Arithmetic) if we would try to deny them. 44. Demerara is a part of British Guayana in South America. 45. Wittgenstein, too, gives and compares examples from empirical everyday life, from mathematics, and (to a much lesser degree than Newman) from religious life.

131

being explicitly stated, the proposition has been repeated (and alluded to) innumerable times in different ways and nobody ever felt any need to justify. On the contrary: The insularity of Great Britain has been used in turn to justify many other beliefs. Newman stresses that the statement seems to be supported by the fact that many other seemingly obvious facts about Great Britain would be incompatible with the falsity of the original statement.46 Newman brings forth two more examples.47 The first one concerns a rather bizarre example of speculation about far-reaching forgeries in literature: Father Hardouin maintained that Terence’s Plays, Virgil’s Aeneid, Horace’s Odes, and the Histories of Livy and Tacitus, were the forgeries of the monks of the thirteenth century. That he should be able to argue in behalf of such a position, shows of course that the proof in behalf of the received opinion is not overwhelming. (Grammar of Assent 296)

While we know and accept that there have been numerous forgeries in literature, we are in no way prepared to admit the possibility that an overwhelming share of classical literature might be spurious. People who make such sweeping claims are not refuted but rather ignored or laughed at. Newman takes his final example from Whately: “What are my grounds for thinking that I, in my own particular case, shall die?” (Grammar of Assent 298)48 In sum, Newman uses everyday examples to show that we give our “assent” in cases that are on the one hand quite simply empirical — there is nothing supernatural or mystical about them — and on the other hand of an “absolute” nature: These everyday convictions are in no way open to doubt, doubting them would not be critical or rational but quite ridiculous. They are exempted from doubt. Fundamentally, Newman is not interested in these everyday beliefs for their own sake but he uses them to build up his two-level model for religious belief. The first and original level is that of simple, unconditional assent. For most religious believers this is the state of unshaken and undoubted adherence to their native religion, 46. In another example from his own day, Newman states (in 1870): “I am quite certain that Victoria is our Sovereign, and not her father, the late Duke of Kent, without laying any claim to the gift of infallibility” (Grammar of Assent 225). 47. Newman views these examples as belonging to the past and future while the first one belonged to the present. 48. This last example is more abstract, lacking the vividness and color of the other two.

132

i.e. the religion they take over from their parents and surroundings. Newman holds that if this original assent is shaken and eventually lost49 it can be replaced by a new body of religious belief, based on a secure, rational foundation, from which every believed proposition can be justified. This point of view might be aptly called Newman’s “rationalism”. This rational belief Newman calls “certitude”. 3. Wittgenstein on Certainty There are some terminological50 differences between Wittgenstein and Newman. Wittgenstein’s “certainty” corresponds to Newman’s “assent”, and he is hardly interested in Newman’s rationalist conception. His own views on the nature of religious language differ widely from those of Newman. However, he must have been deeply fascinated by Newman’s analysis and description of the above-cited examples. Although he does not refer directly to any particular cases,51 even from the sparse first remark we can infer that he viewed Newman as somebody who had something illuminating to say on the questions he was investigating. If Moore was important for Wittgenstein because he had found some interesting sentences, even though he was quite mistaken about their role,52 then Newman was of 49. It is important to note that the original assent feels unassailable, but this does not mean that it cannot eventually erode and get lost. Newman cites his own early view that the Pope is the Antichrist. 50. Strictly speaking there is no worked-out terminology in Wittgenstein’s as well as in Newman’s writings. Both do, however, observe a certain consistency in expressing their ideas and in the particular cases under discussion they made some diverging choices. Wittgenstein was opposed to any kind of terminology even when he wrote the Tractatus. Closer scrutiny of his text shows that even the well-known terminological division of propositions into those that have sense, those that are senseless and those that are nonsensical (sinnvoll, sinnlos and unsinnig in German) is to a very large degree the invention of his well-meaning commentators. Some hints for this can be found in his correspondence with Ogden on translating the book where he insists on a non-literal translation and where he carefully corrects the punctuation style but does not react at all when he encounters some quite inconsistent translations of sinnlos and unsinnig. 51. It is true that Wittgenstein does not mention Newman a second time, not even in OC § 239 where he discusses some peculiarities of Catholic belief. In the light of the evidence of Wittgenstein’s interest in Newman this must seem surprising. 52. Wittgenstein took Moore’s sentences as a start for his own investigation, but he hardly bothered about what Moore was trying to do with them (see Kober 1993, 21–22 on this). The remark OC § 397 seems to contradict this: “Haven’t I gone wrong and isn’t Moore perfectly right?” It is difficult to correctly assess this remark. However, it does not tell us anything about Moore, but only about a passing doubt on Wittgenstein’s part. Thus the frequent occurrence of

133

even greater interest to Wittgenstein. He offered not only some interesting and much more detailed examples, but furthermore, his description of how these sentences work within the web of our beliefs is in important respects very close to Wittgenstein’s own analysis and description.53 In his first remark Wittgenstein refers to a “komische Bemerkung” by Newman.54 This is translated as “curious” but it could also mean “funny” or “comical”.55 In the latter sense, it would refer to a funny remark by Newman about propositions that can be inferred from one another even if there is really no point in making such inferences. There are very many funny remarks in Newman’s writings which often sparkle with humor, and among them there is at least one about such inferences. Maybe Wittgenstein was thinking of the following remark:56 I will take a question of the present moment. “We shall have a European war, for Greece is audaciously defying Turkey.” How are we to test the validity of the reason, implied, not expressed, in the word “for”? Only the judgment of diplomatists, statesmen, capitalists, and the like, founded on experience, strengthened by practical and historical knowledge, controlled by self-interest, can decide the worth of that “for” in relation to accepting or not accepting the conclusion which depends on it. The argument is from concrete fact to concrete fact. How will mere logical inferences, which cannot proceed without general and abstract propositions, help us on to the determination of this particular case? It is not the case of Switzerland attacking Austria, or of Portugal attacking Spain, of Belgium attacking Prussia, but a case without parallels. To draw a scientific conclusion, the argument must run somewhat Moore’s name is quite misleading in many respects. 53. The main reason that Wittgenstein used Moore’s examples may have been their conciseness. It would have taken Wittgenstein much more effort to explain where Newman’s investigations eventually went wrong. 54. The structure of this remark is quite interesting. It consists of a short sentence answering Moore, followed by a longer remark in parentheses that seems to be an aside comment to the first sentence. Then, finally, we get the sentence mentioning Newman included in parentheses, commenting on the first parenthetical remark. 55. There are, of course, also many remarks in Newman’s writings that Wittgenstein found “curious”, like the ones about the saints that he called ‘queer’. None of these, however, really relate to the topics discussed near the beginning of On Certainty. 56. This example explicitly discusses questions of inference. Other candidates might be the Great Britain remark already quoted or a longer passage discussing a controversy in Shakespeare scholarship about a line from Henry V: “His nose was sharp as a pen, and ’a babbled of green fields”, where emendations into “… and a table of green fields” or “… on a table of green frieze” are discussed along with the reasons given in trying to establish some conjecture, each of which sounds more or less nonsensical (Grammar of Assent 271–277).

134

this way: — “All audacious defiances of Turkey on the part of Greece must end in a European war; these present acts of Greece are such: ergo;” — where the major premiss is more difficult to accept than the conclusion, and the proof becomes an “obscurum per obscurius.” […] I take this instance at random in illustration; now let me follow up by more serious cases. (Grammar of Assent 304)

Newman himself thought the remark a bit “less serious”, and it certainly fits in well with Wittgenstein’s line of thought. It could also be compared to an earlier remark of Wittgenstein on the foundations of mathematics: “‘The stove is smoking, so the chimney is out of order again.’ (And that is how this inference is made!) Not like this: ‘The stove is smoking, and whenever the stove is smoking the chimney is out of order; therefore …’” (RFM I § 8).57 Newman and Wittgenstein approached these Moore-typepropositions58 from opposite directions. Newman started from empirical propositions, making no allowances for others, and he ended up finding propositions which could not be doubted but which were still empirical. Wittgenstein on the other hand had been investigating propositions expressing rules without any empirical foundation; in the end he was confronted with examples that acted like rules but all the while seemed to be purely empirical. Newman’s writings are interesting for Wittgenstein for several reasons: To begin with, Newman is not an academic “school man” but a writer of a precise and natural style with hardly any technical language. In addition, he also usually works not with general principles but with particular examples, and he presents these examples in a richness of detail that is quite uncommon in philosophical writing. As a convert who tried to be fair to his former convictions,59 Newman devoted much attention to all sides of any controversy. His accurate and detailed descriptions of controversial situations are much more important than the solutions he offered. His Apologia is a masterpiece in this regard. Furthermore, in his own personal life, Newman was living proof that our deepest convictions are able to change.60 Finally, both Newman and Wittgenstein came from 57. Felix Mühlhölzer drew my attention to this remark. 58. In On Certainty Wittgenstein discusses some examples from Moore but Newman is the only author referred to affirmatively. The only other exception is Goethe (see OC §402 for a quote from Faust). 59. Newman had his writings from the Anglican period of his life reprinted after his conversion. 60. Newman tried to build a model on which his change of convictions was, despite appear-

135

a strong logical background and later developed “a great dislike of paper logic” (Apologia 264). Newman writes: The concrete matter of propositions is a constant source of trouble to syllogistic reasoning, as marring the simplicity and perfection of its process. Words, which denote things, have innumerable implications; but in inferential exercises it is the very triumph of that clearness and hardness of head, which is the characteristic talent of the art, to have stripped them of all these connatural senses, to have drained them of that depth and breadth of associations which constitute their poetry, their rhetoric, and their historical life, to have starved each term down till it has become the ghost of itself, and everywhere one and the same ghost. (Grammar of Assent 267)61

As it turns out, Wittgenstein fundamentally disagreed with Newman’s views on religion but he was attracted by the method of Newman’s logical investigations which he developed to make his conception of religious belief convincing. Newman’s Grammar of Assent is the only known work on logical questions that Wittgenstein read “with great care” at that time,62 and much work will have to be done to work through this connection in more detail.63

REFERENCES Barrett, C. 1991: Wittgenstein on Ethics and Religious Belief, Oxford: Oxford University Press. Bottone, A. 2003: Newman and Wittgenstein after Foundationalism, www.swif. it/biblioteca/cxc. Bouwsma, O. K. (1984) Without Proof or Evidence, Lincoln and London: University of Nebraska Press. ances to the contrary, entirely gradual and rational, thus assimilating the history of his religious convictions to something like progress in science. 61. As we know, Wittgenstein did not think that the invention of modern symbolic logic had changed the situation much in this respect. 62. We have much less evidence of Wittgenstein actually reading Moore’s essays. 63. Thanks to Gottfried Gabriel for raising the question “Which Newman?”, to Brian McGuinness for his interest and suggestions on the translation of OC 1 and PI 289, and to Astrid Schleinitz for reading and discussing with me several earlier versions of this paper. Thanks also to Michael Kober for giving me the opportunity to discuss the matter at the Ulm Wittgenstein Workshop. Thanks finally to Anne Fuchs (University College Dublin) for checking my English.

136

Evra, J. V. 1984: ‘Richard Whately and the Rise of Modern Logic’, in History and Philosophy of Logic 5, 1–18. Ferreira, M. J. 1986: Scepticism and Reasonable Doubt. The British Naturalist Tradition in Wilkins, Hume, Reid and Newman, Oxford: Clarendon. FitzPatrick, P. J. 1978: ‘Eine komische Bemerkung H. Newmans’, in: W. Leinfellner et al. (eds.), Wittgenstein and his Impact on Contemporary Thought, Wien: Hölder-Pichler-Tempsky, 42–45. Hallett, G. 1979: A Companion to Wittgenstein’s ‘Philosophical Investigations’, Ithaca: Cornell University Press. James, W. 1985: The Varieties of Religious Experience [1902], Harmondsworth: Penguin. Kenny, A. 1992: ‘John Henry Newman on the Justification of Faith’, in: A. Kenny, What is Faith? Essays in the Philosophy of Religion, Oxford: Oxford University Press. Ker, I. 1990: The Achievement of John Henry Newman, London: HarperCollins. –– 1998: ‘Newman, John Henry’, in The Routledge Encyclopedia of Philosophy, London: Routledge. Kober, M. 1993: Gewißheit als Norm. Wittgensteins erkenntnistheoretische Untersuchungen in ‘Über Gewißheit’, Berlin: de Gruyter. Locke, J. 1975: An Essay concerning Human Understanding [1690], ed. by P. H. Nidditch, Oxford: Clarendon. Malcolm, N. 1984: Ludwig Wittgenstein. A Memoir, Oxford: Oxford University Press. Moyal-Sharrock, D. (ed.) 2004: Understanding Wittgenstein’s ‘On Certainty’, Houndmills etc.: Palgrave Macmillan. Newman, J. H. 1890: Essay in Aid of a Grammar of Assent, London: Longman Green. –– 1890a: The Miracles of Early Ecclesiastical History [1842/43], London: Longman Green. –– 1890b: Standard Edition, London: Longman Green (this edition is also available at www.newmanreader.org). –– 1900: Certain Difficulties felt by Anglicans in Catholic Teaching, Vol. 2 [1875], London: Longman Green. –– 1913: Apologia pro vita sua, The two versions of 1864 & 1865, ed. by W. Ward, Oxford: Oxford University Press. –– 1973: The Letters and Diaries of John Henry Newman, vol. xxiv, A Grammar of Assent January 1868 to December 1869, Oxford: Clarendon. Phillips, D. Z. 1993: Wittgenstein on Religion, Houndmills and London: Macmillan. Rhees, R. 2003: Wittgenstein on Certainty, ed. by D. Z. Phillips, Oxford: Blackwell.

137

Stroll, A. 1994: Moore and Wittgenstein on Certainty, New York and Oxford: Oxford University Press. Ward, W. W. 1912: The Life of John Henry Cardinal Newman, 2 vols. London. Whately, R. 1848: Elements of Logic [1826], 9th edition, London: John W. Parker. –– 1985: Historic Doubts Relative to Napoleon Bonaparte [1819], ed. by R. S. Pomeroy, Berkeley and London: Scolar Press. Whitehead, A. N. 1948: Adventures of Ideas, Harmondsworth: Penguin.

138

Grazer Philosophische Studien 71 (2006), 139–160.

THOUGHT, LANGUAGE, AND ANIMALS Hans-Johann GLOCK University of Reading Summary This paper discusses Wittgenstein’s ideas about the relation between thought, neurophysiology and language, and about the mental capacities of non-linguistic animals. It deals with his initial espousal and later rejection of a ‘language of thought’, his arguments against the idea that thought requires a medium of images or words, his reasons for resisting the encephalocentric conception of the mind which dominates contemporary philosophy of mind, his mature views about the connection between thought and language, and his remarks about animals. The aim is not just to get a clear picture of Wittgenstein’s position, but also to contrast it with contemporary approaches such as those of Fodor and Searle. While rejecting some of Wittgenstein’s claims about the role of the brain, I defend his basic idea, namely that the capacity for entertaining a thought is conceptually tied to the capacity for displaying that thought in behaviour, rather than to the possession of language as such or to the occurrence of specific neurophysiological phenomena.

In the history of philosophy, the notion of a thought or idea has been subjected to a creeping privatisation. In the Platonist tradition, revived for example by Frege and Moore, a thought or idea is a self-subsistent abstract entity, one which inhabits a separate ontological realm beyond space and time. In Augustine and Malebranche an idea or thought turns into an archetype in the mind of God, something which for its existence depends on the divine intellect. Finally, in the wake of Descartes an idea turns into a ‘modification of the mind’, a psychic entity or occurrence which inhabits the mind of an individual, whether divine or human. The differences between Platonism and Cartesianism can perhaps be resolved amicably by distinguishing between a thought as what is thought or believed — a so-called ‘propositional content’ — and a thought as the having of a thought, a thinking or believing. As Frege pointed out, what I think — e.g. that water is potable — is true or false, independently of my

thinking it. Furthermore, the very same thought can be entertained by different individuals (Frege 1891, 29–32; 1918). By contrast, the state of affairs of my thinking that p differs from the state of affairs of your thinking that p. My thinking that water is potable is a modification of me as an individual. Another conflict remains, however, one which concerns the thinking or having of thoughts. According to mentalism, the having of thoughts is a non- or pre-linguistic affair. It does not require possession of a language, and it is a phenomenon that doesn’t involve language notably the occurrence of some kind of mental image. By contrast, what I shall call lingualism holds that having a thought presupposes the possession of language, and that it always involves a linguistic process of some kind, even though the latter need not be public. This paper explores Wittgenstein’s contributions to these problems. The sections that follow argue respectively for the following contentions: 1. While the early Wittgenstein believed in a language of thought, the later Wittgenstein provided powerful arguments against this idea. 2. Contrary to received wisdom, the later Wittgenstein’s conception of thought avoids both mentalism and lingualism, because it undermines the assumption that thoughts are words or images that accompany our behaviour. 3. Wittgenstein casts doubt not just on neural representationalism, the idea that having thoughts consists in the occurrence of neurophysiological symbols in the brain, but also on the general encephalocentrism that predominates in contemporary analytic philosophy. 4. While Wittgenstein retained the idea of conceptual connections between thought and language, he tied the capacity for thought not to the possession of language, but to the capacity for complex behaviour, including behaviour of an expressive kind. 5. In line with this general approach, Wittgenstein did not deny that animals can think, but insisted that their thoughts are restricted to those which can be manifested in non-linguistic behaviour. 1. Wittgenstein and the Language of Thought In Wittgenstein’s picture theory, we find three apparently disparate explanations of the notion of a thought (Gedanke):

140

A thought is ‘a logical picture of facts’: its only pictorial form (Form der Abbildung) is logical form (TLP 3, 2.181–2.19). This means that it is an optimally abstract picture which does not rely on any specific medium of representation. (II) A thought is an ‘applied, thought (gedachtes) propositional sign’, a ‘proposition with a sense’ (TLP 3.5, 4). Accordingly, a thought is a sentence-in-use, a propositional sign which has been projected onto reality. This suggests that Wittgenstein tried to eschew both the Cartesian and the Platonist position. There is a difference between a mere sign and a proposition, a propositional sign expressing a thought. But the difference does not lie in the fact that the latter is associated with a mental process or an abstract entity. The relationship between a propositional sign and a proposition is analogous to that between a euro coin and a euro. The coin does not name a euro, but to present the coin is to present a euro. (III) Wittgenstein also writes, however: ‘In a proposition a thought finds an expression that can be perceived by the senses’ (TLP 3.1). And in a letter to Russell he insisted that a thought has ‘psychical constituents that have the same sort of relation to reality as words’, i.e. the constituents of the propositional sign (RUL 19.8.19). This suggests that a thought is a psychic phenomenon which is not identical but isomorphic with the propositional sign expressing it. (I)

I can see only one way of making these three specifications even remotely compatible. The early Wittgenstein subscribed to a particular version of lingualism, namely the idea of a language of thought. As he wrote in the Notebooks, ‘thinking is a kind of language’ (NB 12.9.16). By the same token, a thought is itself a proposition in a language of thought. This accounts for (I) in that this language is an optimally abstract mental language; unlike public languages, it does not rely on any specific medium of representation like sounds or inscriptions. Furthermore, unlike their various natural languages, it is shared by all linguistically competent human beings. The logical analysis of propositions in any sign-language will reveal the structure of the underlying proposition in the language of thought. It accounts for (III) on the assumption that the proposition (Satz) at issue in 3.1 is one belonging to a public language rather than the language of thought. Finally, it sort of accounts for (II): while the propositional sign of a thought public language stands in need of accompaniment, a thought is indeed a gedachtes propositional sign, one belonging to the language of thought.

141

The Tractatus was not the first to espouse the idea of a language of thought. It is prominent in William of Occam, for instance in Summa Logica (I.1–2). But it is among the closest ancestors of ideas which became popular through the work of Jerry Fodor (1975). According to Fodor’s language of thought hypothesis, when I have a thought (belief or desire) there occurs within me a neurophysiological phenomenon which functions as a token-sentence of an inner language (‘Mentalese’), just as typographic or acoustic phenomena function as tokens of the sentences of a public language. Although the Tractatus acknowledges an intimate connection between thought and language, it remains wedded to the traditional doctrine that it is the mind which gives meaning to our public language, that mental accompaniments breathe life into sounds and inscriptions that would otherwise be ‘dead’ (BB 3). While the precise nature of thinking is relegated to empirical psychology, the production of thoughts is conceived as a process which must accompany speaking, and which distinguishes it from the squawking of a parrot. In his later work, Wittgenstein intimates that the idea of a language of thought faces a dilemma. Either the alleged language of thought is genuinely linguistic, i.e. its symbols indeed ‘have the same sort of relation to reality as words’. But in that case they must themselves have symbolic content, which leads to a vicious regress. A proposition in Mentalese ‘would for us just be another sign’ (BB 5), which itself stands in need of being projected onto reality. This is obvious if one replaces the mental accompaniment by a physical one: a sentence plus a painting is no less capable of different interpretations than the sign by itself. The other option is that the propositions in the language of thought are somehow intrinsically representational. This idea captures one important feature of thought. While my words can be interpreted by reference to what I think, it makes no sense for me to interpret my own thoughts, save perhaps in the sense of asking myself why I have a particular thought. Unlike speech, ‘thought is the last interpretation’ (BB 34–5; PG 144–5). But this means that the psychic elements do precisely not stand in the same relation to reality as words. Bona fide signs are not intrinsically representational, they acquire their symbolic content through being used and understood in a certain way. Yet that possibility is precisely excluded in the case of thoughts. One might try to avoid the dilemma by reasoning as follows. On the one hand, thoughts are genuinely linguistic sentences, because they

142

have a propositional content (an idea prominent in Fodor); on the other, unlike public sentences they are self-interpreting, because they occur in the medium of the mind. This gambit, however, simply replaces a question about the indisputable capacities of sign-languages with a mystery about the capacities of a postulated language of thought. To suppose that the mind ‘could do much more in these matters’ because of its occult qualities is a ‘mythology of psychology’ (PG 99; Z § 211). Furthermore, the starting point of the defence is question-begging. It presupposes what would have to be established, namely that thoughts can have a propositional content only by being somehow linguistic. This argument rules out the language of thought position Wittgenstein grabbled with, most notably at the beginning of the Blue Book. It does not dispose of contemporary symbolic representationalism. Fodor postulates an innate language of thought which we use in learning public languages and which imbues the latter with meaning or the power to represent. Obviously, this raises the question of where Mentalese derives its meaning or representational content from. But in his later work Fodor (1987) addresses this difficulty by arguing that the tokens of the language of thought acquire meaning through standing in certain causal connections with objects, properties or situations in the external environment. At the same time, in other parts of his oeuvre Wittgenstein refutes causal theories of meaning, albeit of a simpler kind (see Glock 2005). He also provides reasons for denying that neurophysiological events could constitute tokens of a linguistic signs, reasons which I shall elaborate in section 3. 2. The Failure of both Mentalism and Lingualism For a while Wittgenstein continued to identify thought with language, albeit sign-language rather than the language of thought: philosophy is a ‘descriptive science […] of thought’; but thoughts and their logical relations ‘must be examined through the expressions conveying them’; thought is a ‘symbolic process’ and thinking ‘the activity of operating with signs’, which is performed by the hand — in writing — or mouth and larynx — in speaking (LWL 4, 25; BB 6; BT 408). This is obviously wrong. While we write with our hands, we only think with them in the sense of accompanying our speech by exuberant gestures. Residual traces of this misguided lingualism remain in the later work (PI §§ 329–30). In the main, however, Wittgenstein came to realize that think-

143

ing and speaking are conceptually related yet ‘categorially distinct’ (RPP II §§ 6–8, 183–93, 238, 248, 266–7). His mature discussion undermines the assumption behind both the mentalism he encountered in James (1890, 50ff.) and his own earlier lingualism. His first step was to abandon the catholic use of ‘thought’ and ‘think’, which, like the mentalist use of ‘idea’ and ‘representation’, skates over important differences between mental concepts. He treats thinking as a ‘widely ramified concept’, and discusses four major employments (Z §§ 110–2, 122; RPP II § 216): i) ii) iii) iv)

thinking about or meaning something; reflecting on a problem; believing or opining that p; occurrent thoughts which cross one’s mind at a particular moment.

None of them, he maintains, consists in either words or images crossing one’s mind. His strategy for undermining both alternatives is familiar from his account of linguistic understanding. Wittgenstein argues that inner goings-on are neither necessary nor sufficient for the phenomena at issue. In what follows, I shall leave aside (i) (see Glock 1996, 179–84; Glock/Preston 1995). Clearly, the longstanding ‘convictions’ of (iii), at present known as dispositional beliefs, could not consist in either images or words constantly crossing one’s mind. The point holds also for (ii). It would be foolish to deny, as some behaviourists have done, that when one is thinking mental images may cross one’s mind. However, such inner goings-on are neither sufficient nor necessary for me to think. In a delirium I may have mental images but do not think, and I may think about a problem without any images crossing my mind. Not all of our thinking can be characterized as having mental images, a point Berkeley and Kant made vis-à-vis ‘general ideas’ or concepts. The lingualist alternative fares no better. Saying that p and thinking that p are obviously not the same. Fortunately, we do not express all of our thoughts in words; and we sometimes say that p when we think that ~q. The obvious response is that in such cases we talk to ourselves in foro interno, and that thinking is a kind of internal monologue, as Plato had suggested (Theatet 189e). But speaking to oneself in the imagination is no more sufficient or necessary for thinking than having mental images. When

144

I count sheep in order to induce sleep I talk inwardly but I do not think; and one can perform even the most complex intellectual tasks without talking to oneself in the imagination. Both mentalism and lingualism fail even for (iv), what Wittgenstein calls ‘lightning-like thoughts’ (PI §§ 318–20; Z § 122). It is implausible to insist that when it suddenly occurs to a motorist: ‘You fool; there’s a radar control behind the bridge, you had better slow down to 50!’, her mind runs through either that string of words or a series of mental images within a split second. Thus the debate between mentalist and lingualist conceptions of thought seems to end in a stale-mate. We might try to overcome this dead-lock by striking a compromise. In this vein, Dummett (1993, ch. 12) has suggested that while genuine thought is essentially linguistic, animal thoughts or mechanical activities like driving involve ‘proto-thoughts’ which cannot be adequately explained in language. But a compromise between two misconceptions is not a promising idea, outside politics, that is. I suggest that instead we should take a critical look at two premises both sides of the debate accept. For one thing, both mentalists and lingualists subscribe to an accompaniment view of thought. They presuppose that there must be certain occurrent events which determine (and not just in a causal sense) what we think and whether we think. But our previous discussion suggests that internal goings-on, whether mental imagery or words, are irrelevant. When I studied mathematics as an undergraduate, we asked one of our professors to have a look at a problem which had withstood our sustained efforts. He glanced at the problem, briefly stared out of the window, walked to the blackboard, and sketched the solution. In reply to our question ‘How did you do that?’, he said: ‘Well, it’s simple. I glanced at the question, stared out of the window, walked to the blackboard and wrote down the solution.’ Of course, in some circumstances we would deny that the professor had been thinking. If he just reproduced an answer that he had arrived at previously, we would certainly do so. We might also hesitate if the solution occurred to him in a flash, for example on the grounds that this is less a case of thinking than of intuition. But what if at successive stages of this short tale, he would have given different answers to questions like ‘What are you thinking just now?’ or ‘Where have you gotten to with this problem?’. In that case there is no denying that he was thinking through the problem, and yet we would not be entitled to conclude that he must have been lying when he denied that specific mental images or symbols crossed his mind.

145

What we think is not determined by anything going on ‘in the head’. What people think is evident from what they avow to be thinking — if they are sincere —, from how they explain their thoughts when challenged, from their behaviour in a wide variety of contexts, and from what they are capable of doing. Mental images or inner speech are accompaniments of thinking, and may be ‘logical germs’ of thoughts (LW I § 843). As psychological studies in the wake of Vygotsky have shown, they give rise to thoughts and serve as heuristic or mnemonic devices. However, this dependency is contingent. Inner goings-on do not determine what I think, and they are not conceptually necessary for me to think. What we think is evident from what we would sincerely say and do, not by what images or words may flit across our minds. A motorist can be credited with the aforementioned thought if she sincerely avows it — either then or later (see PI § 343). Equally, whether I thought about a problem on a given occasion is not determined by internal accompaniments, but by what I am capable of doing, by the way I speak and act, and it may well depend on what went on before or after. To put it in Aristotelian terms, even those types of thinking which most suggest an actual going-on, namely reflecting on a problem and having an occurrent thought like that of our motorist, are more akin to a potentiality than an actuality. This may sound paradoxical. All it amounts to, however, is this: when we ascribe thoughts to others, we draw a cheque not on what crosses their minds, but on what thoughts they would sincerely avow, and on how they would explain and justify their actions and utterances. The second assumption of the debate between mentalism and lingualism which I want to question is that there must be a medium of thought. Both sides presuppose that thought requires a ‘vehicle’, and that we must always think ‘in’ something, either words or images. But after all, thoughts are not passengers and language is not a means of transportation. What seems to give substance to these metaphors is the fact that one can talk inwardly in a particular language. But this is not the same as to think in a particular language. For, as we have seen, interior monologue is neither necessary nor sufficient for thinking. Arguably, the question of what language I think in arises only with respect to a foreign language. And there it boils down to questions such as these: Do I speak that language hesitantly? Do I have to decide first what I want to say and then try to remember the equivalent in the foreign tongue, or can I simply say it? But there is no need to suppose that in general I must first think in inner representations of some kind, linguistic or mental, and then transpose my

146

thoughts into utterances of a different symbolism. This picture has the absurd consequence that I might always be mistaken about even the most simple of my own thoughts (BB 41; LPP 247–8). For I might first read them off incorrectly from my internal display of words or images, and then mistranslate them into the public language. As they stand, such objections undermine only unsophisticated versions of their targets, e.g. the mentalism of the British empiricists or the lingualism of Plato. Pointing out that the occurrence of mental images is neither necessary nor sufficient for thought has no bite against mentalist positions like that of Husserl, which try to disassociate the idea of a mental representation from its pictorial connotations. But these more refined positions face the task of explaining what having a representation amounts to. In Husserl’s case, for example, we seem to be left with the idea that it is ‘just like’ mental picturing, only without mental images. But that simply boils down to saying that having a representation of X or of the fact that p is to think about X or to think that p, which means that the explanation of thought has moved in a circle. An intentional relation, thinking about X or believing that p, has first been construed as a pictorial one, and has then been robbed of the pictorial aspects which alone can give it any substance (Tugendhat 1982, 62–3, 276–7). Mutatis mutandis for symbolic representation. If it is disconnected not just from the idea of inner speech, but also from the idea of a symbol that has meaning by being used and interpreted, then we are left with a flatus vocis. Ultimately, these problems confronting more abstract positions pose a radical challenge to the very idea of a mental representation. A representation requires not just a content or message, but also a medium in which to represent the former. But in the case of thoughts, no medium is required; thoughts are all message and no medium. The Tractatus dimly realized this point by calling a thought a logical picture; and the later Wittgenstein groped for it by insisting that thought is the final interpretation. Furthermore, his reflections indirectly support scepticism about representationalism, namely by showing that occurrences in either a pictorial or a symbolic medium are irrelevant to thought. Without the connection to a medium, however, the notion of a representation boils down to that of a thought, and thereby loses its explanatory value.

147

3. Against Encephalocentrism Even if one grants that thinking does not consist in having representations in the mind, whether pictorial or linguistic, is it not obvious that it consists in the occurrence of certain neurophysiological processes, just as many neuroscientists and neurophilosophers suppose? I want to enter three objections to his proposal. First, the idea that having thoughts is a matter of having neural representations constituted by patterns of neural firings fares even worse than the idea that it is a matter of having mental representations (see Glock 2003). One minimal condition on a thing R being a representation of another thing X is that R should convey information about X. But, following Peirce’s theory of signs (1933), one can distinguish between different types of representation: • Icons resemble what they represent. • Symbols are related to what they represent by convention. • Indices are connected to what they ‘represent’ by causal dependencies or by other natural relations such as spatial or temporal proximity. Thus realist paintings are icons of what they represent. Linguistic expressions, with the possible exception of onomatopoetic ones, are symbols of what they represent. The word ‘dog’, for example, is connected to the animals not through any kind of resemblance, but through an arbitrary convention. Smoke, finally, is an index of fire, because it is a causal result of fire. Given this distinction, one can deliver the following brief verdict on neural representationalism: Patterns of neural firings are certainly indices of external phenomena, they might be icons (but for the most part are not), they cannot be symbols. That neural firings are causal results of external events and causal preconditions of perception is agreed on all sides. The extent to which there is, for example, a spatial resemblance between the objects of perception and the neural activities that underlie perception, is a matter for empirical investigation. For the most part, no such iconic relation has been observed. Although experiments like those of Hubel and Wiesel show that particular neurons are involved in seeing lines of a particular orientation, there is no iconic similarity between the lines and the pattern of firing neurons. Finally, neural firings cannot be symbols because there is no one who uses them to represent anything in a conventional way. Peter Hacker (1987) has powerfully expressed the same point in Wittgensteinian terminology: no one could use neural firings in a rule-guided manner, and

148

hence they cannot have linguistic meaning. Accordingly, the claim that perception involves neural representations is • false, as far as icons are concerned; • nonsensical, as far as symbols are concerned; • vacuous, as far as indices are concerned. For that last claim boils down to the indisputable but unexciting observation that neural firings in the brain are causally related in law-like ways to certain events outside the brain. Secondly, Wittgenstein does not deny that a brain of a certain size and complexity is a precondition for the possession of mental capacities, and that mental phenomena, e.g. perceiving a flash of light, are correlated with specific neurophysiological processes (PI §§ 376, 412). What he denies is that these goings-on constitute our thoughts. Neural processes are necessary only as a matter of empirical fact, not as a matter of our mental concepts (see BB 7, 118–20; PI §§ 149–58, 339, 376; RPP I § 1063). They are causal preconditions for my having thoughts, but in a logical or conceptual sense they are neither necessary nor sufficient for having a thought. Their presence does not entail that of mental phenomena (whatever the reading of the EEG, I do not think that p if I cannot sincerely ascribe this thought to myself ). And it is logically possible that mental phenomena are present not just without neurophysiological accompaniments of a specific kind (I think that p without any significant reading on the EEG). Wittgenstein further suggests that there might be thought without any neurophysiological accompaniments whatsoever: ‘it is imaginable that my skull should turn out empty when it was operated on’ (OC § 4; see PI §§ 149–58; BB 118–20; RPP I § 1063). But this does specifically not mean that we could doubt that normal human beings have brains, since this is one of the hinge propositions which could be relinquished only at the price of a disintegration of our belief-system. It means that there is no conceptual connection between neurophysiological mechanisms and mental terms. The former play no role in our explanation and application of the latter: third-person uses of mental terms are based on behavioural criteria, first-person uses are not based on any criteria, let alone neurophysiological ones, although a belief in a general connection between neurophysiological and mental phenomena is part of our world-view (BB 47). Wittgenstein has been accused of ignoring that science informs us that neural processes are necessary for mental phenomena like thinking

149

and understanding in a metaphysical rather than physical or conceptual sense (McGinn 1984, 114–7). Wittgenstein, however, explicitly rejected metaphysical necessities of this kind. He would not have accepted the most plausible account of such necessities, namely the realist semantics of Saul Kripke and Hilary Putnam (see Glock 1996, 46–7; Hanfling 2000, ch. 12). Furthermore, even if this account works for natural kinds terms like ‘water’ and ‘gold’, it will not work for our mental terminology, since the latter are not used as rigid designators that refer to all things with the same microstructure as a given sample. That much is clear not just from Wittgenstein’s own discussions about the connection between mental terms and behavioural criteria, but also from perfectly mainstream philosophy of mind. As Putnam and Fodor have pointed out, mental phenomena are multiply realisable through psychochemical phenomena, not just in principle (Martians, computers) but in fact, and not just across species. When different test persons solve one and the same problem, slightly different parts of the brain are activated. The claim that reference to neurophysiology forms no part of the meaning of mental terms shocks not just Kripkean metaphysicians but also contemporary naturalists. But it shouldn’t. We human beings are social primates by nature. Our languages include mental terms because of our fundamental need to describe, explain, predict and otherwise understand the behaviour and behavioural dispositions of others, and because of the equally fundamental need to provide such information to others. No room here for the inner glow sought by Cartesians, or the neural mechanisms that captivate many contemporary philosophers. Wittgenstein rejected not just the idea of psychophysical parallelism but even the weaker idea of supervenience. He acknowledged that, as a matter of empirical fact, a brain of a certain size and complexity is a precondition for the possession of mental capacities, and that some mental phenomena, e.g. perceiving a flash of light, are correlated with specific neurophysiological processes (PI §§ 376, 412). But he denied that there must be any neurophysiological difference, for example, between someone who speaks a sentence with understanding and someone who does not, or between someone who remembers a certain event or fact and someone who does not. One justification for this claim is an attack on received ideas of causation. In the case of two apparently identical plant-seeds which produce different kinds of plants there need not be any difference in the seeds underlying these different dispositions. We could and should treat the

150

origin of these seeds not just as the basis for a prediction — ‘Seeds from a type A-plant will produce type A-plants’ — but also as a genuine explanation — i.e. add ‘… because they are from type A-plants’ (Z §§ 608–10; CE 410–11, 433–4). This line of reasoning is uncompelling. Even if there is nothing unintelligible in supposing that there is no structural difference between the two seeds, accepting phenomenal properties (concerning the origin of the seeds) as ultimate causal explanations would be to abandon a highly successful principle of the physical sciences: Causal explanations must ultimately be structural rather than phenomenal. It would be on a par with accepting astrological explanations, provided the latter can be backed by statistical evidence. But even if it should turn out, for example, that people born between between 21 January and 19 February have an above average IQ, this fact cannot be explained by the fact that these people were born under the sign of Aquarius. Wittgenstein also has a more promising argument against supervenience. Although whether a creature is capable of having certain thoughts does not depend on its origin, it may depend on the context in which it operates. Like contemporary externalists (Burge, Putnam), Wittgenstein holds that two physically identical individuals might have different thoughts, albeit for different reasons (Glock/Preston 1995). One can be in agonizing pain for a split second, irrespective of the setting, since pain can be displayed in characteristic pain-behaviour. But one cannot, for example, expect something for a split second, irrespective of the setting, since an expectation ‘is embedded in a situation from which it arises’ (PI § 581). Furthermore, intentions are embedded in human customs and institutions. One can intend to play chess only as a member of a culture in which the game is played (PI §§ 205, 337). For, a person A can intend to ) only if A can display or execute this intention. In the case of complex intentions, this in turn presupposes a social and historical context, since otherwise the relevant actions and utterances do not count as expressions of the intention. By the same token, Nofretete could not intend to cash a cheque, even if she had been in the same neurophysiological state and had comported herself in the very same manner as contemporary folks when they go to the bank. Even if Wittgenstein is right on these points, as I think he is, it overcomes supervenience only at a local level. At a global level, supervenience is more compelling. How could two individuals entertain different thoughts if everything were physically equal, including their history and social context? Admittedly, such global supervenience lacks bite, because everything

151

is never equal. We could never infer that an individual must be entertaining a particular thought by observing that her overall physical situation is precisely the same as that of another individual having that thought. Still, global supervenience may mark a point at which the mental is rooted in the physical, not because of metaphysical necessities, and not just because of contingent facts about the causal preconditions of mentality, but because of conceptual connections involving our mental vocabulary (Jackson 2005 features different arguments for a similar conclusion). 4. The True Connection between Thought and Language Wittgenstein rejects lingualism as much as mentalism and materialism. Symbolic representations are no more essential to thinking than mental images or neural firings. At the same time, he does think that there is an essential link between thought and language, yet it does not require any actual inner vocalization. For one thing, we identify thoughts/beliefs by identifying their linguistic expressions (see BB 4–5, 161; PI § 502; MS 108, 237). Although thoughts are not identical with their linguistic expression, they are not entities beyond language either. The answer to the question ‘What do you think?’ is not a description of an inner process, but an expression of my thoughts in words, e.g. ‘I think that it will rain’. If I am challenged by a Platonist or mentalist to express the thought behind that utterance, I do not re-examine some inner process to see whether I can describe it better. Instead, I paraphrase my utterance into other symbols. I am not prevented from doing more by an epistemological obstacle, Rather, my opponent’s insistence on something else is blocked by the very concept of thought. Similarly, two people think the same thought not by sharing a qualitatively identical mental entity, or relating to a numerically identical abstract entity, but if the expressions of their thoughts say the same thing. Equally, it is the expression of thoughts which allows one to speak of their having constituents, as Fregeans do. Consequently, language is not just the only, if distorting expression of thought, as Frege had it (Frege 1979, 225, 269–70), it is the ultimate expression. The second essential link between thought and speech is one which will occupy us in the sequel. According to Wittgenstein, the capacity for thought requires the capacity to manipulate symbols, not because unexpressed thoughts must be in a language, but because the expression of thoughts

152

must be. The reason is that ascribing thoughts makes sense only in cases where we have criteria for identifying thoughts. In Quine’s memorable phrase, ‘No entity without identity’ (Quine 1969, 23). Something must count as thinking that p rather than that q. This means that thoughts, although they need not actually be expressed, must be capable of being expressed. And only a restricted range of thoughts can be expressed in non-linguistic behaviour. What matters to Wittgenstein, therefore, is the capacity for behaviour, which need not include linguistic behaviour. This invites the charge of behaviourism, which has been vehemently rejected by many of Wittgenstein’s followers. In fact, however, Wittgenstein’s relation to behaviourism is complex (see Hacker 1990, 224–53). In his mature later work Wittgenstein diverges even from the logical behaviourism of Carnap and Ryle in crucial respects. For one thing, he denies that first-person psychological propositions can be analysed into propositions about one’s own behaviour, to be verified by self-observation. In fact, he later claimed that such propositions are not descriptions at all, let alone descriptions of behaviour, but avowals, expressions of the mental. Such avowals have a role similar to that of expressive behaviour, but they are not about behaviour. To moan is not to say ‘I moan’, to cry out ‘I am in pain’ not to say ‘I am manifesting pain-behaviour’ (PI §§ 179, 244; LSD 11; LPE 296; RPP I § 287; Z § 539). For another, Wittgenstein came to reject the idea that psychological statements can be analysed into statements about behavioural dispositions. The relationship between the mental and behaviour is much more complicated than logical behaviourists suppose. Sensation-terms like pain are ascribed to others on the basis of straightforward behavioural criteria. By contrast, intentional verbs (‘believes’, ‘desires’, ‘intends’, etc.) and expressions for moods cannot be applied simply on the basis of an individual’s momentary behaviour. The occasions for their use forms a highly complex syndrome. For example, what counts as a manifestation of sadness on one occasion, may not on another. And what someone is disposed to do as a result of holding a particular belief depends not just on the context, but also on his other beliefs, desires, intentions, etc. (see PI II 174, 229; LW I §§ 862–9, 942, 966; LW II 42–3, 55–6, 84; Z § 567–9; RPP I §§ 129, 314; Z § 492; cp. Ryle 1980, 128–30). At the same time, Wittgenstein’s later philosophy of psychology retains points of contact with logical behaviourism. It rejects the dualist account of the mental as inalienable and epistemically private. Instead, it claims

153

that the ascription of psychological predicates to other people is logically connected with behaviour. However, that connection is not one of logical equivalence between propositions (namely psychological and behavioural ones). Rather, Wittgenstein maintains that our mental terms would not mean what they do if they were not bound up with some behavioural criteria, however diverse and defeasible. Finally, as we shall see in section 5, the result is that Wittgenstein does not deny mentality to non-linguistic creatures per se; rather he denies that they are capable of thoughts too complex to be manifested in non-linguistic behaviour. The ensuing position eschews both mentalism and lingualism, both dualism and behaviourism. Mental events and states are neither reducible to, nor totally separable from their bodily and behavioural expressions. The relationship between mental phenomena and their behavioural manifestations is not a causal one to be discovered empirically, through theory and induction, but a criterial one: it is part of the concepts of particular mental phenomena that they have a characteristic manifestation in behaviour, and that they are manifested in recognizable circumstances (LPE 286; LSD 10). ‘The concept of pain is characterized by its particular function in our life. Pain has this position in our life; has these connections’ (Z §§ 532–3). It is a moot question whether this explains the concept by reference to a combination of input (circumstances) and output (behaviour), a suggestion close to contemporary functionalism. What is clear is that for Wittgenstein it is part of mental concepts in general that they have some behavioural manifestation. We would have no use for these expressions if they were not bound up with behavioural criteria. If we came across human beings who used a word which lacked any connection with painbehaviour and the circumstances in which we display it, we could not translate it as ‘pain’. Putnam’s assertion to the contrary notwithstanding (Putnam 1975, ch. 16), the idea that there might be ‘super-super-spartans’ who are in constant agony without ever showing it under any conceivable circumstances, is as incoherent as describing as soulless human beings who behave exactly like us (LPP 281). 5. Wittgenstein — a Brute to the Brutes? Rumour has it that Wittgenstein was a brute to the brutes (e.g. Rollin 1990, 137–41; Singer 1990, 14; Frey 1980, 101–10). He supposedly denied that non-linguistic animals can have thoughts on account of his

154

private language argument. Some animal lovers then consider a caricature of the private language argument — often derived from A. J. Ayer — which they demolish to their own unbridled satisfaction. Whatever its merits, however, the private language argument does not imply that animals do not think (see DeGrazia 1994). Its conclusion is that there is no such thing as a language which cannot be communicated or taught to others even in principle. This in turn implies that we cannot ascribe such a private language to animals that do not speak a public one. But to reach the further conclusion that such animals cannot think would require the lingualist assumption that thinking requires possession of a language. As I hope to have shown, far from endorsing that assumption Wittgenstein actually casts doubt on it. Furthermore, Wittgenstein explicitly countenanced the ascription of sensations and consciousness to animals: ‘Only of a human being and what resembles (behaves like) a human being can one say: it has sensations; it sees; is blind; hears; is deaf; is conscious or unconscious’ (PI § 281, see §§ 282–7, 359–61). It is clear, moreover, that according to Wittgenstein some animals do behave like a human being in the relevant respects. While he denies that dogs can simulate pain (infra), his treatment of the issue implies that they can be in pain. Indeed, he even countenances the ascription of pain to insects: Look at a stone and imagine it having sensations. — One says to oneself: How could one so much as get the idea of ascribing a sensation to a thing? One might as well ascribe it to a number! — And now look at a wriggling fly and at once these difficulties vanish and pain seems able to get a foothold here, where before everything was, so to speak, too smooth for it (PI § 284).

Next, Wittgenstein grants that animals have intentions because, like sensations, they can be displayed in non-linguistic behaviour. What is the natural expression of an intention? — Look at a cat when it stalks a bird; or a beast when it wants to escape ((Connexion with propositions about sensations.)) (PI § 647)

As concerns emotions and beliefs, Wittgenstein’s verdict is both ambivalent and tentative: One can imagine an animal angry, frightened, unhappy, happy, startled. But hopeful? And why not? The dog believes his master is at the door. But can he also believe that his master will come the day after to-morrow? And what can he not do

155

here? — How do I do it? — How am I supposed to answer this? Can only those hope who can talk? Only those who have mastered the use of a language. That is to say, the phenomena of hope are modes of this complicated form of life. (If a concept refers to a character of human handwriting, it has no application to beings that do not write). (PI II 174)

And in a similar vein: We say the dog is afraid his master will beat him; but not, he is afraid his master will beat him tomorrow. Why not? (PI § 650).

Obviously, we must distinguish two parameters here: the kind of propositional attitude on the one hand, the kind of ‘propositional content’ on the other. Expressing the point in the formal mode which is less presumptuous and more congenial to Wittgenstein: it is one question what intentional verbs can be applied to animals; it is another question what noun-clauses, and in particular what that-clauses can follow these intentional verbs. Wittgenstein reasons that a dog can satisfy some intentional verbs but not others, and that these can be complemented by some noun-clauses but not others. Let’s attend to the noun-clauses first. Some thoughts cannot be manifested in non-linguistic behaviour because they are about things remote in space and time. For example, the dog can manifest its fear of being beaten by cowering with its tail between its legs. But nothing in its behavioural repertoire can be a manifestation of its fear now that it will be beaten tomorrow. To be sure, Eike von Savigny has ingeniously envisaged canines capable of nuanced and complex behaviour which might give license to such an ascription (von Savigny 1995, 43–4). However, such ascriptions are at best highly speculative and at worst hopelessly underdetermined. Furthermore, a modified conceptual limitation remains: the sort of behaviour which dogs are actually capable of cannot license the ascription of complex intentional verbs. With respect to the intentional verb, there are two possible reasons for Wittgenstein’s restrictive claims. One is his form of life contextualism (the line pursued by von Savigny). A dog can look forward to being taken for a walk by fetching the lead, wagging its tail excitedly, etc. But this expressive behaviour is not embedded in the appropriate way in a social environment. The other is the strictly behavioural argument. Thus nothing distinguishes a dog hoping to be taken for a walk from its being joyously confident of being taken for a walk. In my view the behavioural interpretation is superior. For Wittgenstein does not allude to the kind of social life an animal is capable of. The difference between solitary bears and pack hunters like

156

wolves never registers in his reflections. What does register is the kind of expressive behaviour, including the kind of facial expressions. The same two options also apply to another famous passage: Why can’t a dog simulate pain? Is he too honest? Could one teach a dog to simulate pain? Perhaps it is possible to teach him to howl on particular occasions as if he were in pain, even when he is not. But the surroundings which are necessary for his behaviour to be real simulation are missing. (PI § 250)

Prima facie this supports the contextualist interpretation. But another passage points in the opposite direction: ‘A child has much to learn before it can pretend. (A dog cannot be a hypocrite, but neither can he be sincere.)’ (PI II 229) Here the emphasis is on what the dog is or rather is not capable of doing. Similarly in the following passage: ‘A clever dog might perhaps be taught to give a kind of whine of pain but it would never get as far as conscious imitation’ (RPP II § 631). Here it is not the social context but the absence of an intention to deceive which bars dogs from simulation and hypocrisy. The argument is very similar to the debate in current cognitive ethology about so-called Machiavellian intelligence (see Byrne 1995). Although several primates are capable of deceiving other creatures, the crucial question is whether they are performed with the intention to deceive. Such an intention can be displayed in non-linguistic behaviour, e.g. in the hide and peek behaviour of chimpanzees. But this requires more complex behaviour than dogs are actually capable of. Philosophical opinion on animal thought forms a spectrum of cases: At one end we find lingualists like Davidson, who deny that non-linguistic animals have any thoughts (cf. Glock 2003a, ch. 9). The other, mentalist end is occupied by empiricists like Hume, who think that the thoughts of animals differ from those of humans only in degree, due to their different perceptual inputs. Oysters do not have thoughts about bicycles, simply because they cannot perceive bicycles. In one sense it is also occupied by those cognitive psychologists who explain even simple animal behavior by reference to a rich variety of complex thoughts and calculations, except that these thoughts are held to be in a language of thought, not in a public language. Like Davidson, Wittgenstein approaches thought from a third-person perspective. They do not appeal to phenomena — whether mental or neurophysiological — that cannot be manifested in behaviour even in principle. Both hold in effect that one cannot attribute beliefs to creatures which are totally incapable of manifesting these beliefs, Davidson because he

157

cannot make sense of the notion of a belief as a private attitude completely detached from behaviour and its explanation (Davidson 1984, 170; 1985, 476), Wittgenstein because he insists that we can ascribe a thought that p to a creature a only if something counts as a thinking that p rather than that q. The difference is that, unlike Wittgenstein, Davidson insists that even for simple beliefs, the required behaviour must include linguistic behaviour. In addition to these two extremes, however, there is an intermediate position, adopted by a coalition (rare, some might say) of common sense and Wittgenstein. It holds that animals are capable of having thoughts of a simple kind, namely those that can be expressed in non-linguistic behaviour. I want to end this paper by contrasting Wittgenstein’s position with one that is in many respects congenial, but which repudiates the third-person perspective. John Searle shares Wittgenstein’s scepticism about scepticism in general and scepticism about other minds, including animal minds, in particular. He also agrees with the claim that animals can have some but not all of our thoughts. At the same time, Searle’s arguments to this effect are diametrically opposed to Wittgenstein’s. He rejects the third-person perspective on thought that Wittgenstein shares with some opponents of animal thought, notably Davidson. According to Searle, ‘behaviour is simply irrelevant’ to the attribution of thoughts, because ‘my car radio exhibits much more intelligent verbal behaviour, not only than any animal but even than any human that I know’ (Searle 1994, 216). If one were to trust this passage, one would not envy Searle his company. The production of noise by a radio hardly even qualifies as behaviour. But if it does, it is exceedingly stupid. The radio fails the Turing test miserably. Even to its non-linguistic environment, moreover, it cannot react in an intelligent, i.e. responsive and flexible manner. This is why during a traffic jam, in the midst of a chorus of honking, it is capable of uttering things like ‘Right now everything is serene and quiet here’. It is not the radio that behaves intelligently, but at best the person whose utterances it transmits; and even that very much depends on the station it is tuned to. According to Searle, the essential prerequisite of thought is the presence of neurophysiological phenomena rather than the capacity for intelligent behaviour. Suppose we had a science of the brain which enabled us to establish conclusively the causal bases of consciousness in humans. Suppose we discovered that certain electrochemical sequences [XYZ] were causally necessary and sufficient for consciousness in humans. […] Now if we found XYZ present in snails

158

but absent in termites, that would seem very strong empirical evidence that snails had consciousness and termites did not. (Searle 1994, 215–6)

But one can only establish that XYZ is the causal base of consciousness if the phenomenon of consciousness has been identified on independent grounds. Searle dismisses as irrelevant the criteria for consciousness and thought employed not just by lay-people but also by cognitive ethologists. According to him it is part of the meaning of mental terms that they apply only to creatures with a certain neural outfit. This has the unpalatable consequence that scepticism about animal minds must be alright after all, since even the most knowledgeable among us are ignorant about the precise causal base of consciousness and the extent to which it is common to humans and animals. In light of our previous discussion, Searle cannot appeal to supervenience to sideline behaviour. If two creatures were identical in every respect, including not just their overall physical properties but also their connection to their physical and social environment, then supervenience would require that they have the same mental properties. But of course human beings and animals are not identical in this way. There is absolutely no reason to suppose that the occurrence of some electrochemical phenomenon in a context which is different in other respects has any bearing on mental properties. In particular, no occurrence of XYZ can show that a completely inert thing is nevertheless conscious. Searle’s emphasis on the brain at the expense of behaviour is a manifestation of an encephalocentrism that is rife within contemporary analytic philosophy. In this respect, at least, Wittgenstein’s perspective is both more naturalistic and more realistic. Mental and biological phenomena reveal themselves only when we go beyond the brain and consider not just the whole organism, but the organism in the context of its environment, in the context of its form of life, as both Wittgenstein and cognitive ethologists could put it.

REFERENCES Byrne, Richard 1995: The Thinking Ape, Oxford. Dummett, Michael 1993: Origins of Analytical Philosophy, London. Fodor, Jerry: “Special Sciences”, Synthese 28, 77–115. –– 1975: The Language of Thought, New York.

159

–– 1987: Psychosemantics, Cambridge/Mass. Frege, Gottlob 1891: “Sense and Meaning”, in Frege 1984, cited after the original pagination. –– 1918: “The Thought”, in Frege 1984, cited after the original pagination. –– 1979: Posthumous Writings, Oxford. –– 1984: Collected Papers, ed. by B. McGuinness, Oxford. Frey, Ray 1980: Interests and Rights: The Case against Animals, Oxford. Glock, Hans-Johann 1996: A Wittgenstein Dictionary, Oxford. –– 2003: “Neural Representationalism”, Facta Philosophica Vol. 5, 147–71. –– 2003a: Quine and Davidson on Language, Thought and Reality, Cambridge. –– 2005: The Normativity of Meaning made Simple”, in Nimtz/Beckermann, 219–41. –– and John Preston 1995: “Externalism and First Person Authority”, The Monist 78, 515–534. DeGrazia, D. 1994: “Wittgenstein and the Mental Life of Animals”, History of Philosophy Quarterly 11, 121–37. Hacker, Peter 1987: “Languages, Minds and Brains”, in C. Blakemore, S. Greenfield (eds.): Mindwaves, Oxford, 485–505. –– 1990: Wittgenstein: Meaning and Mind, Oxford. Hanfling, Oswald 2000: Philosophy and Ordinary Language, London. Jackson, Frank 2005: “The Case for A priori Physicalism”, in Nimtz/Beckermann, 251–66. James, William 1890: The Principles of Psychology Vol. II, New York. McGinn, Colin 1984: Wittgenstein on Meaning, Oxford. Nimtz, Christian, Ansgar Beckermann (eds.) 2005: Philosophy and Science, Proceedings of GAP 5, Paderborn. Peirce, Charles Sanders (1933) The Collected Papers of Charles Sanders Peirce Vol. II, Cambridge/Mass.. Putnam, Hilary 1975: Mind, Language and Reality, Cambridge. Quine, Willard van Orman 1969: Ontological Relativity, New York. Rollin, Bernard 1990: The Unheeded Cry: Animal Consciousness, Animal Pain and Science, Oxford. Ryle, Gilbert 1949: The Concept of Mind, Harmondsworth 1980. von Savigny, Eike 1995: “Keine Hoffnung für Hunde”, in E. von Savigny, O. Scholz (eds.): Wittgenstein über die Seele, Frankfurt. Searle, John 1994: “Animal Minds”, Midwest Studies in Philosophy XIX, 206– 19. Singer, Peter 1990: Animal Liberation, Second Edition New York. Tugendhat, Ernst 1982: Traditional and Analytical Philosophy, Cambridge.

160

Grazer Philosophische Studien 71 (2006), 161–174.

MOORE’S PARADOX AND FIRST-PERSON AUTHORITY Severin SCHROEDER University of Reading Summary This paper explores Wittgenstein’s attempts to explain the peculiarities of the firstperson use of ‘believe’ that manifest themselves in Moore’s paradox, discussed in Philosophical Investigations, Part II, section x. An utterance of the form ‘p and I do not believe that p’ is a kind of contradiction, for the second conjunct is not, as it might appear, just a description of my mental state, but an expression of my belief that not-p, contradicting the preceding expression of my belief that p. Thus, ‘I believe that p’ is just a stylistic variant of ‘p’; the word ‘believe’ doesn’t seem to have a substantial role to play in such an utterance. Following Wittgenstein, I discuss why there could not be a first-person present-tense use of the word that was more akin to its use in the third person: why it is impossible to describe one’s own current beliefs in a detached manner without thereby expressing them. In the final section, I try to develop Wittgenstein’s suggestion that the non-epistemic authority we have regarding the contents of our beliefs can be clarified by considering its link with intention and action.

1. Moore’s Paradox is the observation that it makes no sense to say: (1) It is raining and I don’t believe it. What is the point of Wittgenstein’s discussion of this paradox in chapter x of MS 144, published as Part II of the Philosophical Investigations? A first answer to this question might be given as follows: One of the leitmotifs of Wittgenstein’s later philosophy is his critique of referentialism, the view that words have meaning in virtue of standing for objects (their reference). And Wittgenstein was particularly concerned with the application of referentialism to psychological language and the consequent view that words like ‘understand’, ‘think’, ‘pain’, ‘hope’, or ‘believe’ denote states or processes in the privacy of our minds (cf. PI § 308). According to this — highly natural — view, one would be inclined to say that the statement:

(2) I believe that it’s not raining. was a description of the speaker’s state of mind (PI IIx, 190ab), whereas the statement: (3) It is raining. was a description of something entirely different, namely the weather (cf. RPP I § 819). So one should expect these two statements to be perfectly compatible, as descriptions of two logically independent subjects matters. Suppose ‘I believe it’s going to rain’ meant: ‘When I say to myself the sentence “It’s going to rain” it gives me a feeling of assurance’ […] If that were the case, Moore’s paradox would not come about: I could then utter the assertion: ‘I believe it’s going to rain; and it’s not going to rain’. (MS 132, 98f.)

However, this conjunction is a sort of contradiction. So expressions of belief, such as (2), cannot be understood as merely descriptions of one’s own state of mind. As a more plausible reading Wittgenstein suggests that (2) is in fact used very much like the assertion: (4) It is not raining. In this case the introductory phrase ‘I believe that …’ is fairly unimportant (PI IIx, 190c). It may be seen as a rhetorical device to attract attention to oneself as opposed to others, comparable to the preface: ‘If you want to know my opinion …’ (cf. PI IIx, 191b). It is an introductory signal and not itself part of the statement thus announced. Another possibility is that the words ‘I believe that…’ are used — like ‘It is probable that…’ — to express a degree of uncertainty (cf. RPP I § 821). Again, they wouldn’t be part of the statement itself. They indicate the way a statement is presented, not its content: ‘Don’t regard a hesitant assertion as an assertion of hesitancy’ (PI IIx, 192m). Both readings of ‘I believe that …’ dispel the semblance of paradox. For once it is clear that by uttering (2) ‘I believe that it’s not raining’ one asserts that it’s not raining, the statement’s incompatibility with (3) ‘It is raining’ is obvious. Is that Wittgenstein’s solution of Moore’s paradox? So it seems, for the just quoted remark about hesitancy concludes the discussion in chapter x.

162

2. The account given above, however, cannot be the whole story, or it would be difficult to understand why Wittgenstein thought Moore’s paradox so extraordinarily explosive and illuminating: ‘Moore poked into a philosophical wasp nest with his paradox; and if the wasps did not duly fly out, that’s only because they were too listless’ (CV 87 = MS 137, 120). What I presented as a first approximation to a solution of Moore’s paradox is, from a different perspective, exactly what is so disquieting about the matter: The paradox disappears when we convince ourselves that the word ‘believe’ has a peculiar function in assertions of the form ‘I believe that…’. But exactly this asymmetry between different uses of the same verb is likely to appear paradoxical: Moore’s paradox can be put like this: the expression “I believe that this is the case” is used like the assertion “This is the case”; and yet the hypothesis that I believe that this is the case is not used like the hypothesis that this is the case. So it looks as if the assertion “I believe” were not the assertion of what is supposed in the hypothesis “I believe”! (PI IIx, 190cd)

Equally paradoxical is the asymmetry between the use of the word ‘believe’ in the first person singular present tense and its use in the past tense (PI IIx, 190ef ): (5) I believe that the butler did it. says roughly the same as: (6) The butler did it. On the other hand: (7) I believed that the butler had done it. cannot be shortened in such a way. For this really is a statement about my past belief. Obviously in statements such as (5) the word ‘believe’ does not have the same function as in (7), or in a number of other forms of statements, such as: (8) Suppose I believe that the butler did it, and in fact he didn’t. (9) Inspector Witherspoon believes that the butler did it.

163

So it ought to be possible for there to be a language, a variant of English, that takes this difference into account by not using the same word in both kinds of cases (PI IIx, 191ef ). Let us suppose, in that language instead of ‘I believe that p’ one simply says ‘p’; or, if one wants to express uncertainty: ‘Probably, p’; or, in order to attract attention to oneself one uses the words: ‘I say, p’ (cf. PI IIx, 192i). In this language Moore’s paradox would not exist; ‘instead of it, however, there would be a verb lacking one inflexion’ (PI IIx, 191f ). But of course this language is merely a clarified version of our language. A difference in use is clearly marked in difference of wording. This then is the disquieting phenomenon: When we go by use and function, rather than appearance on the page, the word ‘believe’ in statements of the form ‘I believe that p’ (for short, first-person use) is different from the word ‘believe’ in other statements (for short, third-person use). Thus the word ‘believe’ in its third-person use is grammatically incomplete. In the sense in which I can say of others that they believe something; and of myself that in the past I believed something or that I might have believed something (which as it happens I know to be false) — in that sense of the word I cannot say of myself that I believe something now. That sounds odd, because we are so strongly inclined — prejudiced — to take it for granted that language always functions in the same way. ‘Don’t look at it as a matter of course, but as a most remarkable thing, that the verbs “believe”, “wish”, “will” display all the inflexions possessed by “cut”, “chew”, “run”’ (PI IIx, 190k). Amongst other things, Wittgenstein is here also attacking ‘Frege’s idea that every assertion contains a hypothesis,1 which is the thing that is asserted’ (PI § 22; cf. RPP I § 500). For, as noted above, one aspect of the asymmetry in question is that the assertion: (5) I believe that the butler did it. does not assert what is supposed in the corresponding hypothesis: (10) Suppose I believe that the butler did it. The hypothesis (10) is about my state of belief which is not part of the content of the assertion (5). Moreover, it is wrong to regard a hypothesis 1. Frege rather uses the term ‘Gedanke’: ‘thought’.

164

as the basic element to which something else needs to be added in order to produce an assertion (as suggested by Frege’s notation). In fact, the assertorical use of sentences is far more elementary than the comparatively sophisticated language-game of hypothesizing (cf. RPP I § 478). Wittgenstein once called it ‘the main mistake made by philosophers of the present generation, including Moore’ when considering language only to look at a form of words and not the use made of the form of words (LC 2). Undoubtedly, a reason for his interest in Moore’s paradox and the importance he ascribed to it was that it affords a particularly clear example of that mistake, or the tendency to make it. Going by superficial appearances one is inclined to deny that (1) ‘It is raining and I don’t believe it’ involves a contradiction; and yet if anyone were to utter that sentence, one would say to him that: he contradicted himself! That is an indication of considerable gaps in logic. It is an indication — as are so many other things — that what ordinarily we call ‘logic’ is applicable only to a minute part of the game with language. That, indeed, is why logic is so uninteresting, while, on the face of it, it ought to be so interesting. (MS 132, 119f., follows after RPP I § 488; cf. LW I § 525)

In a similar vein Wittgenstein wrote in a letter to Moore, shortly after Moore had presented a paper on his paradox: This assertion [‘p is the case and I don’t believe that p is the case’] has to be ruled out and is ruled out by “common sense”, just as a contradiction is. And this just shows that logic isn’t as simple as logicians think it is. In particular: that contradiction isn’t the unique thing people think it is. It isn’t the only logically inadmissible form and it is, under certain circumstances, admissible. And to show this seems to me the chief merit of your paper. (ML 285, October 1944)

Wittgenstein saw Moore’s paradox as a striking example of the shortcomings of formal logic. That explains why he thought it so important and likened it to a poke into a wasps’ nest. But it does not fully explain his own continuing occupation with it (over some five years). Why should Moore’s paradox, apart from being an embarrassment to the aficionados of formal logic, also pose a serious philosophical problem to Wittgenstein himself? 3. We saw that the use of the word ‘believe’ is not uniform: in utterances of the form ‘I believe that p’ (what I called its first-person use) it doesn’t

165

play the same role as in other forms of statements (what I called its thirdperson use). Indeed, in its first-person use it may not have any real function at all, as often one can express one’s belief just as well by the mere statement ‘p’. Thus, if for the moment we look only at simple present-tense statements, we could say that the word ‘believe’ has no substantial use in the first person: [I believe that p.] You believe that p. He/she believes that p.

=p

Is there an explanation for this irregularity? This is the question that a considerable part of Wittgenstein’s remarks on the topic are concerned with: Why do we not have a more substantial use for the word ‘believe’ in the first person present tense? Wittgenstein writes that he is ‘tempted to look for a different development of the verb [“to believe”] in the first person present indicative’ (PI IIx, 191j); that is, a development that would allow me to say with the word ‘believe’ of myself exactly what I can say of others; a development in which the words ‘I believe that’ could not be dropped, but would be as semantically indispensable as ‘You believe that’ or ‘He believes that’. Believing, Wittgenstein tentatively suggests, ‘is a kind of disposition of the believing person. This is shewn me in the case of someone else by his behaviour; and by his words. And under this head, by the expression “I believe …” as well as by the simple assertion (PI IIx, 191k–192a). So if the first-person use is to be assimilated to the third-person use, it too would have to be based on observation. In order to ascribe a belief to myself I would have ‘to take notice of myself as others do, to listen to myself talking, to be able to draw conclusions from what I say’ (PI IIx, 192a). It would have to be possible for me to say: ‘Judging from what I say, this is what I believe’ (PI IIx, 192d); or again: ‘I seem to believe’ (PI IIx, 192c). And then, if I am able to take such a distanced attitude towards my own words and beliefs, I could also say: (1) ‘It is raining and I don’t believe it’ (PI IIx, 192f ). This, however, seems impossible. Apparently, I cannot adopt such a distanced attitude towards my own behaviour (– and so Moore’s paradox arises). Why not?2 2. To this question Wittgenstein reverts with remarkable frequency; see RPP I §§ 703-5, 711, 712, 737, 739, 744f., 752, 814.

166

A natural and tempting answer is that there is just no need. I do not need to observe and interpret my own behaviour because I experience myself from the inside and so know immediately what I believe (RPP I §§ 704, 711, 738f., 744f.): ‘“One feels conviction within oneself, one doesn’t infer it from one’s own words or their tone.” — What is true here is: one does not infer one’s own conviction from one’s own words; nor yet the actions which arise from that conviction’ (PI IIx, 191i). Of course, there is no specific feeling by which one tells what one believes. The word ‘feel’ is just a convenient expression to fill a gap: when one regards something as true without being able to give reasons or evidence, one may say: ‘I feel that …’. — But is it not at any rate true that I know immediately what I believe? No, one can’t really speak of knowledge here, at least not in the ordinary sense of the word, which can be characterized by three features (PI § 246; PI IIxi, 221): (i)

There are reasons, evidence or a way of acquiring knowledge (e.g. sense perception). I.e., there is an answer to the question ‘How do you know?’. (ii) There is a possibility of doubt or error. (iii) There is a possibility of ignorance or surprise. But these are exactly the points that Wittgenstein is concerned to clarify in his discussions of Moore’s paradox. None of these features pertains to expressions of one’s belief: One’s claim that one has a certain belief is not based on reasons, evidence or self-observation; there is no room for doubt or error about the contents of one’s own belief; nor can one be entirely ignorant of, or surprised to find out, what one believes.3 These are the features of the concept of belief on which Wittgenstein is trying to shed more light. He is trying, in other words, to find an explanation for the non-epistemic authority we have regarding the contents of our beliefs. But is it really non-epistemic? Do I not know what I believe? True, one may say, the lack of those three features, which are present in more ordinary cases of knowing, shows that this is a somewhat special kind of knowledge. But that doesn’t mean that it cannot be a case of knowledge all the same. If it’s true that p and someone is able reliably to report that p, it would seem appropriate to say that he knows that p. Hence, as I 3. Even in cases of self-deception one cannot be entirely ignorant of what one tries to overlook, nor can one be truly surprised when eventually it resurfaces from the back of one’s mind. After all, calling it self-deception implies at least some vague awareness of the matter.

167

am reliably able to report what it is I believe, why not say that I know it? — There is, indeed, no reason to be dogmatic about the use of the word ‘know’. ‘Say what you choose, so long as it does not prevent you from seeing the facts’ (PI § 79). But then, if we use the word ‘to know’ in this diluted sense, all it means is: ‘to be able to state reliably’ (cf. PI IIviii, 185c). Hence, saying that I ‘know’ what I believe would not afford any explanation why I am able to state my beliefs, even without any observation of my behaviour. 4. Wittgenstein suggested tentatively that believing is ‘a kind of disposition’ (PI IIx, 191k) that can manifest itself both in utterances (e.g. ‘It’s raining’) and in non-verbal behaviour (e.g. taking an umbrella).4 So it should be possible to identify such a disposition through self-observation; as in the case of other dispositions. For example, I know from self-observation that after the second glass of wine I become melancholy; or that today I am uncommonly irritable. I can also observe my dispositions to believe or not to believe things: that for certain periods or with respect to certain things I am particularly credulous or suspicious (cf. RPP I § 502; RPP II § 417). But can I also observe that I believe something? Can I observe, for example, ‘that I reply “Yes” to the question “Will he come?” and also exhibit such-and-such other reactions’ (RPP I § 503)? We can distinguish between the manifestation (or expression) of a disposition and the self-ascription of a disposition (RPP I § 832). For example, my utterance ‘Tennis ist langweilig’ may be a manifestation of a disposition to speak German, but not its description or self-ascription; whereas ‘I have a disposition to speak German’ is a self-ascription, but not a manifestation of that disposition. As long as this distinction is applicable, the kind of incoherence Moore described is not a problem. Of course the self-ascription of a disposition may be accompanied by a manifestation of the opposite disposition; or the claim that one does not have a certain disposition may be at odds with a simultaneous manifestation of that 4. This is not meant to be more than an approximate location of ‘believe’ on the map of psychological concepts. Beliefs are very unlike states of consciousness with genuine duration, and more like dispositions, in that they are not interrupted by a break in one’s attention or consciousness, but manifest themselves in a person’s verbal and non-verbal behaviour under certain circumstances (RPP II § 45). On the other hand, the fact that there are crucial logical differences between ordinary dispositions and believing is exactly what the discussion of Moore’s paradox, and in particular Wittgenstein’s invitation to construe believing as ‘a kind of disposition’, is meant to bring out.

168

disposition; but that would never amount to nonsense. It would simply be an error, like the following: (11) I don’t use metaphors: I never beat about the bush. Or (RPP I § 503): (12) I’m incapable of pronouncing any word with four syllables. Such judgements are based on observations of one’s own linguistic habits or capacities; observations that may well be inaccurate. An utterance of the form ‘I believe that p’ we take as a manifestation (or expression) of one’s belief that p. Now the question is whether it could not, instead, be construed as a fallible self-ascription of a kind of noetic disposition; rather as ‘the report “I am in the state of belief …”’ (RPP I § 832; cf. RPP II § 281). There are indeed cases where observing one’s own responses one appears to learn something new about one’s beliefs or desires, as in this example from Tolstoy’s Anna Karenina: At first Anna sincerely believed that she was displeased with him [Vronsky] for daring to pursue her; but soon after her return from Moscow, having gone to a party where she expected to meet him but to which he did not come she distinctly realised by the disappointment that overcame her, that she had been deceiving herself and that his pursuit was not only not distasteful to her, but was the whole interest of her life.5

Clarifying one’s thoughts and feelings, deciding whether something is on balance a good thing or a bad thing, may take some time and may involve taking stock of one’s emotional responses in real or possible situations. An insufficiently considered judgement may not express one’s real views, yet sometimes one has reasons to prevent oneself from attending to all that is relevant, preferring to stick to a convenient prejudice, deceiving oneself; until one’s attention is forced back to what one tried to overlook. However, even when occasioned by an observation of one’s emotional response or behaviour, the expression of one’s thoughts and feelings is not for that matter just a hypothesis. What Anna’s disappointment brought home to her is that she had been deceiving herself. That means, even before the event 5. L. N. Tolstoy, Anna Karenina, II.iv; tr. R. Pevear and L. Volokhonsky, Harmondsworth: Penguin, 1954.

169

she should have been aware that she thought more highly of Vronsky than she cared to admit. And if now she were to say: ‘I believe that Vronsky’s attention is not at all disagreeable’, she would, again, express or manifest her belief (just as earlier she might have expressed the belief that Vronsky was a nuisance), and not just make a self-ascription on the basis of some fallible evidence. Could perhaps psychoanalysis provide us with a case of a statement of belief that was based entirely on disinterested self-observation, allowing for a direct conflict with a simultaneous expression of belief? Consider an example, cited by Freud, of a member of parliament, named Lattmann, who called for a ‘rückgratlos’ (spineless) declaration of loyalty to the Kaiser, when he intended to say ‘rückhaltlos’ (unreserved).6 There is no doubt that such slips of the tongue often betray what a person truly believes contrary to, or beyond, what he intended to say. The thought is in one’s mind while one is expressing something else, and inadvertently a part of it finds its way into the speech, especially when there is a phonetic similarity between a word one wanted to say and a word expressive of one’s parallel secret thought. So it is conceivable that it had just occurred to Lattmann, or had been lurking at the back of his mind, that such a public kow-tow to the Emperor would show some lack of character. But it is also possible that he could in all honesty deny any such thought, insisting that he really believed the declaration in question to be perfectly appropriate, reasonable and dignified. His slip of the tongue, he might protest, was just that, an unfortunate phonetic accident. Freud, however, does not accept this possibility. For him it is an axiom that in this sphere there are no coincidences: everything we do or say is a manifestation of our thoughts and desires. Hence, if indeed we are not aware of any such thoughts or desires, they must be unconscious.7 It would appear then that Lattmann, if psychoanalytically trained, could have been in a position to say: 6. S. Freud: Die Psychopathologie des Alltagslebens, Frankfurt/Main: Fischer, 1954, 80. 7. Admittedly, Freud also suggests that psychoanalysis can prove the existence of unconscious thoughts and desires by bringing them to consciousness: ultimately eliciting from the patient an acknowledgement that he did indeed have them. However, apart from the practical worry that such a confession may be entirely due to the analyst’s persuasive suggestions, this does not seem to be taken as a negative criterion. Which is to say that a patient’s failure to own up to the suggested thoughts and desires is not regarded as conclusive proof that they didn’t exist. And Freud has no qualms saying that analysis may uncover thoughts the patient will never recognise as his own (S. Freud, J. Breuer: Studien über Hysterie, Frankfurt/Main: Fischer, 1970, 218). Cf. M. Eagle, ‘Validation of Motivational Explanation in Psychoanalysis’, in: Psychoanalysis and Contemporary Science 2 (1973), 265-75.

170

(13) It is honourable to make a declaration of loyalty to the Kaiser, but (as my slip of the tongue shows) unconsciously I don’t believe it. However, the psychoanalytical inference is as dubious as the underlying idea that all our mistakes must be determined by, and expressive of, our believes and desires, a premise for which Freud offers no support. But setting such doubts aside, the idea of unconscious beliefs — which I am unable just to express, but must discover through behavioural evidence — implies a split personality (cf. RPP I § 820). In order to be able to say: ‘“Judging from what I say, this is what I believe.” […] One would have to fill out the picture with behaviour indicating that two people were speaking through my mouth’ (PI IIx, 192ef ). For as an observer of oneself one must step back from oneself, adopting the stance of another person. But then we lose hold of the very thing we were trying to construe: an observer’s attitude towards one’s own utterances. If there are two people speaking through my mouth, then it is no longer my own utterances I am contradicting. The case is similar to the (not unlikely) scenario of a railway official making the announcement: ‘Train No. … will arrive at … o’clock. Personally I don’t believe it’ (RPP I § 486). The first sentence we don’t regard as the speaker’s own statement, but as an announcement that he is obliged to make even if he does not believe it to be true. 5. Is there anything more that could be said to explain why it is impossible (pathological cases apart) to distance oneself from one’s own expressions of belief in the way required by Moore’s paradoxical sentence (1)? Chapter x of Part II of the Investigations does not provide much of an answer. The point is only touched upon in two laconic remarks: Do I myself not see and hear myself, then? — That can be said. (PI IIx, 191h) My own relation to my words is wholly different from other people’s. (PI IIx, 192b)

Yet from the underlying manuscript sources it appears that this must be the Archimedean point of the whole discussion. When talking about Moore’s paradox in a lecture (on 14th February 1947), Wittgenstein said (according to Geach’s notes): ‘It is colossally important that one doesn’t normally infer one’s own beliefs from one’s own behaviour’ (LPP 67). Moreover he

171

remarked: ‘Moore’s paradox is bound up with the problems of voluntary motion and intention’ (LPP 67).8 This connection I shall now attempt to reconstruct. It seems to me that Wittgenstein’s thoughts on this matter remained unfinished, but at the time he had hoped here to find an explanation for the infallibility of expressions of belief, and indeed did find such an explanation at least in outline. Thus, Wittgenstein’s discussions of Moore’s paradox are (amongst other things) a contribution to his discussion of first-person authority, complementing his earlier remarks on this topic (PI §§ 246ff.). In Philosophical Investigations II, chapter x, there is only one passage mentioning a link between belief and voluntary action. Wittgenstein presents the idea of a language in which the verb ‘to believe’ is replaced by the expression ‘to be inclined to say’, except in the first person singular present tense, where instead of ‘I believe it is so’ one simply says: ‘It is so’ (PI IIx, 191e). He continues: Moore’s paradox would not exist in this language; instead of it, however, there would be a verb lacking one inflexion. But this ought not to surprise us. Think of the fact that one can predict one’s own future action by an expression of intention. (PI IIx, 191fg)

What ought not to surprise us here is that the imagined verb would lack one inflexion (cf. PI IIx, 190k). For the same thing occurs elsewhere, namely with the future auxiliary ‘will’ in its use to predict future events on the basis of past or present experience (e.g., ‘It will rain tonight’; ‘He will be here presently’). This concept, too, is incomplete, as ‘will’ in the first person singular has a rather different meaning: ‘I will be there’ is (normally) an expression of intention, and not a prediction based on experience. Of course the analogy is not perfect, for in this case there is a first person singular after all, although expressed by a different verb form: ‘shall’. ‘I shall be there’ is not an expression of intention, but roughly the same kind of prediction as ‘My wife will be there’, treating a person’s presence at the place in question as an occurrence to be predicted on the basis of one’s knowledge, rather than as something within the speaker’s immediate power. Even one’s own voluntary actions one can often predict on the basis of experience: ‘Compliant as I am, I expect I shall eventually acquiesce in 8. Or, according to Shah’s notes: ‘This paradox hangs together with voluntary and involuntary activity, and intention, and (the fact) that intention is a prediction — not a prediction based on hypothesis’ (LPP 195).

172

his wishes’ (cf. LPP 66). There may even be an outright conflict between intention and prediction: ‘I intend to say no, but I expect I shall in the end be persuaded to say yes.’ This, however, is only possible where the action in question is to take place some way in the future, so that there is yet time for me to change my intentions. My future intentions may not be known to me, so that my predictions are, in principle, just like others’ predictions based on experience and quite fallible. By contrast, it is hardly possible for me to make an inductive prediction about what I am intentionally going to do right now. If, for example, I am really resolved to lock the door now, I may of course have misgivings about my success (perhaps the key won’t work), but I can neither doubt nor expect that I am at least going to try. To that extent intention ousts inductive expectation. The word ‘intentional’ implies that one is immediately aware of what one is doing. It is obvious that ‘my relation to my actions is not one of observation’ (RPP I § 712); rather more obvious than that I cannot be an observer vis-à-vis my own beliefs. Therefore Wittgenstein tries to draw attention to the close conceptual link between intentions and beliefs, to show that the latter are just as much part of our selves and not to be observed from the outside as are the former. Every intention involves beliefs or knowledge, and every belief can manifest itself in intentions or actions.9 When somebody else says ‘This train is for Hereford’ I can doubt whether the utterance is really the expression of a belief; not when I say it. For my beliefs are interwoven with intentions (e.g. to travel to Hereford). When I distrust another person’s statement I suspect that he might not be prepared to act in accordance with his words (e.g., not board this train even when he wants to go to Hereford by train). But it is logically impossible for me to be in doubt about my own intentions (except of course in the sense that I may be undecided), and to the extent that my beliefs are involved in my intentions, they too must be logically exempt from any doubt. For example, my intention to take the train on Platform 2 to Hereford involves the belief that the train on Platform 2 is the Hereford train. Thus I can just as little be in doubt that I have this belief as I can be in doubt that I have that intention.

9. Are there not also beliefs about entirely theoretical or historical matters that can hardly impinge on my actions? For any belief there will at least be the possible verbal action of saying truthfully what one holds to be true in the matter, for example, trying to give the correct answer when examined.

173

Why don’t I make inferences from my own words to a condition from which words and actions take their rise? In the first place, I do not make inferences from my words to my probable actions. (RPP I § 814; cf. RPP I § 738)

If I were to infer what I believe from my utterance ‘This is the Hereford train’, I would also have to infer what, given my goals, I am likely to do. But that I cannot infer, because I intend it (cf. LW II, 10d). My intentions do not leave any room for an uninvolved observer’s attitude towards the intended action (setting aside the qualification entered above: that longterm intentions may be expected to change); and so there is also no room for an uninvolved observer’s attitude towards the beliefs involved in one’s intentions, which could potentially be any of one’s beliefs.10

10. I am indebted to David Dolby for comments on an earlier version of this paper.

174

Grazer Philosophische Studien 71 (2006), 175–204.

USE, MEANING, AND THEORETICAL COMMITMENT Eike VON SAVIGNY Universität Bielefeld

Summary This paper is on theoretical commitments involved in connecting use and meaning. Wittgenstein maintained, in his Philosophical Investigations, that meaning more or less ‘is’ use; and he more or less proclaimed that in philosophy, we must ‘not advance any kind of theory’ (PI § 109). He presented a connection between use and meaning by describing a sequence of language-games where richness of vocabularies and complexity of embedding behaviour grow simultaneously. This presentation is very impressive in the sequence of PI §§ 2, 8, 15, and 21, even if it needs sympathetic touching up. If supplemented, the presentation makes a convincing case for claiming that there is a connection between use and meaning in the following sense: Within everyday, innocent talk about meanings of expressions, all questions and controversies about meanings are ultimately to be answered and to be decided by appeal to correct descriptions of the expressions’ use. This may be a very modest statement of the meaning-is-use connection. However, establishing even this modest statement requires that one implicitly relies on controversial, explanatory theories from philosophy of language, as sober analysis of the sequence presented by Wittgenstein will reveal. This is not to say that the modest statement is in any way fishy. Rather, I want to remind readers of how desirable it is to restrict the interpretation of Wittgenstein’s famous hostile remarks on theories to that kind of metaphysical misunderstandings of our everyday language which the context of PI § 109 is about. In (1) I characterize, by way of listing examples from the Philosophical Investigations, the area of what I think Wittgenstein regarded as innocent, everyday meaning talk, talk that is not yet infected by bad philosophy. In (2), I argue that what Wittgenstein wanted to show was that such talk is in some sense replaceable by use descriptions, i.e. by descriptions of language-games. In (3), I argue that not all kinds of language-games are relevant; in particular, those of teaching and explaining words have to be excluded. As I restrict myself to the four remaining ‘primitive’ language-games in PI §§ 2, 8, 15, and 21, I have to defend my approach, in (4), against Joachim Schulte’s case for reading Witt-

genstein’s comparison of these language-games with real languages as ironical. How the invitation to regard such a language-game as a complete, primitive language should in fact be construed is a question I discuss in (5), defending my interpretation against Richard Raatzsch in particular. How increases of expressive power are brought about by increases of the use repertoires is shown by an analysis of modified versions of the language-games in question, and of alternatives thereof, in (6), (7), (8), and (9) respectively, pointing out the places where theoretical commitments enter. Section (10) sums up commitments that have emerged from a sympathetic defence of a modest reading of the meaningand-use connection.

1. Innocent Meaning Talk In the Philosophical Investigations, Wittgenstein often refers to or even discusses occurrences, situations, developments, activities etc. that involve or invite meaning talk of a kind that he regards as manifestly innocent and ordinary (even if liable to mislead philosophers). I shall here give a rough overview; the references are preferably to sections early in the book, supplemented by later sections if these are particularly significant or particularly popular. In brackets I add specimens of related utterances to point out how varied meaning talk can look; sometimes these are PIallusions, but I shall not indicate that (thus the quotation marks do not indicate quotations): Teaching a child his first language: PI §§ 6, 8, 143, 145, 244 and passim (‘Doll’, ‘This is the king’). — Explaining a word by ostension or examples: PI §§ 69, 71, 75, 560 (‘Football and ring-a-ring-a-roses are games, but war isn’t’, ‘This is sepia’). — Asking something’s name: PI §§ 28, 31 (‘What do you call this?’). — Defining a word by ostension or examples or verbally, christening, making a word more precise: PI §§ 68, 132 (‘My doll is called Lizzy’, ‘I define the concept of number as the logical sum of cardinal numbers, rational numbers, real numbers, and complex numbers’). — Grading expressions according to exactness: PI §§ 70, 76 (‘”Good” is less exact than “apt for its purpose”’). — Removing a misunderstanding: PI § 10 (‘I did not mean gaming with dice when I asked you to show the children a game’). — Declaring one’s meaning: PI §§ 125, 513, 633 ff., 677, 692 (‘The comprehension axiom is not meant to apply to the property of being a non-self-element’). — (Mis)Understanding an expression in the sense of mastering it, having a way of grasping it, or failing to do so: PI §§ 139, 146, 201 (‘He understands “cube” as “prism”’, ‘He continues “1, 4” with “9” because he understands it as the beginning of

176

the series of square numbers’). — (Mis)Understanding an expression in the sense of beginning to master it or of grasping it in a flash: PI §§ 139, 151, 181, 321 (‘Now the meaning dawned on him’, ‘Light is only seeming to dawn on him’). — Interpreting an expression: PI §§ 84–87, 185–187, 201 (‘He understands the rule “No alcohol on duty” as containing the clause “except on birthday parties”’). — Being in doubt about the meaning of something, not knowing how to react: PI §§ 60, 85, 288 (‘Why do you put it so oddly?’). — Paraphrasing a sentence: PI § 502 (‘”It is raining” means that water is pouring down from the clouds’). — Accepting meaning identity in a restricted context but not in general (‘In a use manual, “We cannot recommend to use the cell phone under water” means the same as “The cell phone will be damaged if submerged in water”’). —Translating a sentence: PI §§ 20, 61–62 (‘”Come!” means “Komm her!”’). — Committing a Germanism: PI § 597 (‘He speaks English as if he were translating’). —Trying to find the fitting expression: PI § 335 (‘Well, this is not exactly what I mean to say’). — Speaking thoughtfully: PI § 330 (‘Switch on your brain before talking’). — Acting on understanding: PI § 505 (‘Make sure you’ve understood the order before obeying it’). — Declaring an expression senseless: PI §§ 498–500, 511 (‘“Abrakadabra” does not mean anything’).1

I hope the list suffices for characterizing the kind of meaning talk Wittgenstein takes to be innocent for the reason that in such cases the ‘everyday use’ of key words like ‘meaning’, ‘sense’, ‘to be called’ and the like does not stray beyond the borders of its ‘original home’ (PI § 116). I assume it is such ways of speaking about meaning that he has in mind when suggesting that descriptions of use can serve to underpin meaning talk in one way or another. I can imagine (ancient) societies where whenever a messenger delivers the sender’s message, the question whether or not his report is correct with respect to content would never arise. There would be no meaning talk like ‘My lord tells you “Comply on my man’s arrival”, therefore my lord wants his order to be obeyed today’. In such a society, if a bilingual speaker A acts as spokesman for B who wants to address C (B and C speaking different native tongues), it would never occur to anyone that A could have made a mistake (other than a slip of the tongue). E.g., C would not ask A, ‘Did 1. In spite of the space Wittgenstein dedicates to them, I have not included two items he may be understood to have regarded as problematic: The case of sentences that mean the same in the sense of saying the same but differ in meaning on account of their peculiar forms (PI §§531ff.), and the question how to distinguish, in terms of use, homonyms from one word with several meanings (PI §§551-568).

177

B mean “Someone ought to come” or “Everyone ought to come”?’ When children learn their native tongue, teaching may be absent altogether. No parent would say, ‘This is called a cart, kiddie’. They would just react to the child’s ‘Jack has got a bike’ by saying ‘Jack has got a cart, kiddie’. When a will is drafted, its wording would be rendered more exact by the solicitor’s simply suggesting ‘all my mobile and immobile property’ instead of the testator’s ‘all my possessions’ without any reasoning about relative exactness. Thus the meanings that are talked about in much of our everyday meaning talk might be there without being talked about at all. (To be sure, a language of the kind described would be poorer than ours.) This observation is not exciting, but it is of some relevance both to the question whether or not Wittgenstein’s primitive language-games can contain meaningful expressions although they do not provide for meaning talk, and to the question whether language-games of teaching and explaining are as relevant for meaning as is, in explicit contrast, ‘the practice of the use of language’ (PI § 7). 2. Replaceability Wittgenstein may have believed that there is an interesting relation between an expression’s meaning, as it is captured by innocent everyday meaning talk, and the use of this expression; also, he may have believed that there is an interesting relation between innocent everyday meaning talk and descriptions of use. In the first case, the most interesting relation is that of identity: the meaning of a word is its use in the language.2 Given Wittgenstein’s philosophical outlook in general, it is difficult to decide whether he might have hoped to make sense of such an identity statement; however, people sympathetic to his views may have reasons to find that he is committed to the identity view. If this is true, then the second belief seems to follow (not vice versa): If the meaning of X is the use of X, then talk about the meaning of X should be talk about the use of X (even if unbeknownst to the uninitiated who do not know that in talking about Lewis Carroll they are talking about Charles Ludwig Dodgson because they do not know that the mathematician was the writer and poet). Now I want to pin down Wittgenstein to being 2. Attention! Danger! This is not a verbatim quotation. As preceded by the first sentence of PI § 43a, the identical second sentence of PI § 43a expresses an altogether different proposition, which is about the explainability of an expression.

178

committed to a theory or theories because he held the weaker alternative; so I welcome evidence for the conjecture that he held the stronger one. To be sure, there are some clues pointing in that direction. The first one is a telltale rhetorical question: ‘But doesn’t the fact that sentences have the same sense consist in their having the same use?’ (PI §20b). This question plays the role of a supporting step in a longer argument, and the German original is still more suggestive; translated literally, it says: ‘But doesn’t their identical sense consist in their identical use?’ Similar cases where an identity assumption seems to be used are PI §§ 60f., PI §§ 138f., and in PI §§ 191–197. Examples where differences in meaning are established by pointing out differences in use are too numerous in the Philosophical Investigations to require references. The reverse is presupposed, even if not very explicitly, in PI § 244: where saying ‘I am in pain’ has the same use as crying from pain, saying ‘I am in pain’ is a cry of lament (rather than a description), and in PI § 270: if ‘E’ is uttered consistently, under the same circumstances and with the same consequences, then if it denotes a sensation it denotes the same sensation in every case. There is also the argument of PI §§ 403–411: If ‘Someone is in pain — I don’t know who’ had the same use as ‘I am in pain’, then since the first utterance does not refer to the speaker, the second one would not refer to him, either. There are more passages pointing in the same direction; however, the evidence is not conclusive. The case is more straightforward for the weaker belief which concerns replaceability — indeed, the reader sometimes cannot help feeling that Wittgenstein uses the question whether talk about meaning is replaceable by talk about use as a test for the meaning talk in question being everyday and innocent. From the very beginning of the Philosophical Investigations, he treats descriptions of use as rendering questions about meaning out of place or even odd. Here is the first example: ‘But what is the meaning of the word “five”? — No such thing was in question here, only how the word “five” is used.’ (PI § 1d) Since the shopkeeper example is meant to serve as a contrast to the philosophical theory of meaning sketched in the bulk of PI § 1, Wittgenstein’s studied way of missing the point in answering the question seems to imply that talking about the use of a word is a better way of achieving what you hope to achieve than talking about its meaning. Again: ‘Now what do the words of this language signify? — What is supposed to shew what they signify, if not the kind of use they have? And we have already described that’ (PI §10a). Wittgenstein implies that he has rendered the opening question superfluous by describing the use. (The second question hints at a reason

179

justifying his manoeuvre, a reason that would follow if meaning were use.) The implication amounts to the claim that the description of the use of ‘Slab’ etc. is more informative, or more intelligible, than the statement ‘“Slab” signifies building-stones of that form’. In PI § 43, in a context where he points out possible sources of confusing the meaning of a name with its bearer, the replaceability idea is stated in the first paragraph: ‘For a large class of cases — though not for all — in which we employ the word “meaning” it can be defined thus: the meaning of a word is its use in the language’ (PI § 43a). Wittgenstein here treats the idea as obvious at least as far as the meanings of proper names are concerned; ‘For a large class of cases — though not for all’ simply means that possibly controversial cases are uninteresting in the present context, so don’t bother about them. He does so in order to show that if we replace ‘meaning’ by ‘use in the language’ in PI § 43b, then instead of ‘the meaning of a name is sometimes explained by pointing to its bearer’ (PI § 43b), we get: the use of a name in the language is sometimes explained by pointing to its bearer. This is salutary because nobody will mistake the bearer of a name for its use. Another explicit statement, from the early sections, of the replaceability thesis is contained in the first sentences of PI § 51 in connection with PI § 53a. Famous further examples can be found in PI §§ 117b, 180, 246a, 370, and 486b. While Wittgenstein’s adherence to the replaceability idea may be uncontroversial, at least two questions, important for my purpose, are still open: How strong is ‘replacing’ to be construed, and what kind of use talk is included? Replacing might mean that the enterprise has successfully been carried out, for a given expression, as soon as there has been supplied a way of translating any innocent, everyday context dealing with the expression’s meaning into a context referring to its use. I do not think Wittgenstein would have gone so far. But he must have aimed at something interesting, and I take this interesting something to be that the truth or falsity of any innocent, everyday claim about an expression’s meaning is correct or incorrect exclusively in virtue of facts about the expression’s use; and for this something to be interesting Wittgenstein has to tell us how to single out the relevant facts. Thus rather than taking him to claim that meaning talk can be replaced by use talk, I take him to claim that questions about meaning can be replaced by questions about use. As regards the second open question, something specific must be said about the kinds of facts about use that are to count as relevant. Expres-

180

sions may differ in their ranges of application, in the consequences of uttering them, in being pejorative or euphemistic, in language level and acceptability, in obsoleteness, in syntactical versatility and morphological flexibility and so on and so forth. It may be possible to describe uses in such a comprehensive way that all these kinds of differences are mirrored in differences of use; but as I hope to show below, determining the relevant kinds of use is achieved by the theoretical means Wittgenstein is committed to making use of. However, we shall have to exclude one kind of use from the start, viz. the use of an expression as what the language-game deals with — e.g. that use of a word which consists in the word’s being explained in a language-game. 3. Meta-linguistic language-games excluded As I said above, the facts referred to in meaning talk, like the fact that a child’s doll is called Lizzy, or at least some of these facts, could obtain in a language without there being meaning talk dealing with them. At least to the extent that such meaning talk is of the kind that is correct or incorrect, it does not itself belong to the facts about use that are relevant to the meanings that are talked about. This holds for teaching, for explaining (in contrast to defining and christening), for paraphrasing and indirect speech report, and for translating. Thus such uses of language are not among those that are relevant to the meanings of the words, sentences, and utterances concerned. If you teach or explain a word wrongly, you do not teach or explain the meaning of the word. Therefore, describing such activities is not an appropriate candidate for elucidating a statement about the meaning of an expression. This observation requires some comments on Wittgenstein’s use of the term ‘language-game’. In the sections I shall consider (roughly, PI §§1–24) Wittgenstein demonstrates his way of describing facts about the (interesting) use of expressions by describing language-games. Teaching, learning by being taught, explaining and the like he calls ‘language-games’, too; I think this is not because he took them to contribute to meaning but because they share the characteristic of connecting utterances with activities. To take the relevant passages in their order: A child uses such primitive forms of language when it learns to talk. (PI § 5b)

181

The only examples described so far (shopkeeper and builders) serve as standards for the degree of primitiveness, but they cannot be literally included: a very small child can neither write (as does the customer in PI §1d) nor act as a shopkeeper, and the builders’ language-game is not a learning activity, either. We could imagine that the language of § 2 was the whole language of A and B; even the whole language of a tribe. The children are brought up to perform these actions, to use these words as they do so, and to react in this way to the words of others. (PI § 6a)

The wording implies that language learning is not part of the PI § 2 language. The same holds for the concrete example in the next paragraph: An important part of the training will consist in the teacher’s pointing to the objects, directing the child’s attention to them, and at the same time uttering a word; for instance, the word “slab” as he points to that shape. (PI § 6b)

It is one of the points of the PI § 8 language that only at this stage pointing is introduced into the language! And later in the same section Wittgenstein states the condition for teaching to be successful: But if the ostensive teaching has this effect, — am I to say that it effects an understanding of the word? Don’t you understand the call “Slab!” if you act upon it in such-and-such a way? (PI § 6c)

This is just what B is supposed to have mastered to be able to participate in the PI § 2 language-game. In PI § 7a, Wittgenstein explicitly distinguishes between actually using the PI § 2 language and teaching it. This means that where he considers, in PI § 7b, using a language of that kind in order to learn one’s native tongue, the native tongue in question is either more comprehensive than the PI § 2 language, or else its use by competent speakers like A and B is contrasted with the use made by a child that is not yet competent. This child would not be using the language either in the role of A or of B, that is, as a competent speaker or hearer. In PI § 7a and c, Wittgenstein points to further concrete educational measures and calls them ‘processes resembling language’ and ‘language-games’. So far, the picture is the following: On the one hand, there is a language-game that is played by competent speakers. On the other hand, there are ways for children to become competent in such language-games. The latter are called language-games, too, because they share the characteristic of being wholes, consisting of language, i.e. expressions, and actions into

182

which language is woven (PI § 7). Therefore, whatever meanings expressions have in virtue of being used in language-games, they have on account of their use by competent speakers. The picture seems to change in PI § 27b, where ostensive explanation (in contrast to teaching, see PI §§ 7a, 26) and christening are introduced as language-games in their own right. Explanations of words (ostensive or other) can play different roles, two of which are of interest for present purposes. The first is that of describing the meaning of a word in one or another of many possible ways. Such descriptions can be correct or incorrect. If correct, they describe the meaning the word has independently of the description. The descriptive activity is, then, a language-game, but does not belong to the facts which conjectures about the meaning of the word in question are to be checked against. The second role of explaining a word is that of christening an object, or of authoritatively defining an expression, as when parents give a name to their newborn child or when a scientific convention formally agrees on a unit of measurement. Given undisputed authority, the act sets a standard for correctly using the word; this is why the act fixes the meaning. Therefore, turning to the act is the appropriate way to find out about the meaning. In real life, however, things are not so clear. The child may come to be known under a different name, and the unit may come to be measured in ways that are more easily accessible but diverge from the original definition. The birth certificate and the convention minutes may lose their standing as ultimate courts of appeal. So the question about the meaning of the word will turn into the question whether the original act is (still) effective, which has to be investigated by turning to the actual use. Rather than fixing the meaning, the original act is the cause of a use which, as long as the act is effective, can be described as obeying the rule: ‘Appeal to the original act.’ So we have a very special kind of languagegame, one where a rule cannot be obeyed unless there are antecedently meaningful expressions. Thus when Wittgenstein says: ‘“The meaning of a word is what is explained by the explanation of the meaning.” I.e.: if you want to understand the use of the word “meaning”, look for what are called “explanations of meaning”’ (PI § 560), he is not saying that the explanation of a word is a prime example of a language-game that contributes to the word’s meaning; he is, of course, referring to correct explanations of meaning, i.e. to descriptive explanations. Rather than telling us how to discover the meaning of a given word, he advises us to look for correct descriptions of meanings of words if we want to find out something interesting about

183

what ‘meaning’ means. For it must be something that fits into our innocent, everyday meaning talk which explanations of meaning are a prime example of. 4. Wittgenstein’s Strategy in PI §§ 2, 8, 15, and 21 I believe that in the first 24 sections of the Philosophical Investigations Wittgenstein wants to show that meanings of expressions as well as differences between meanings emerge in languages (regarded as stocks of expressions and utterances thereof ) at the same pace as the repertoire of activities into which they ‘are woven’ (PI § 7) is growing. In PI § 2, there are four activities of the A-calls-‘Slab’-B-fetches-slab type corresponding to four calls; no sentences, therefore no words. In PI § 8, there are quite a number of structured expressions: 4 (as before) times 26 (letters of the alphabet) times x (number of directions in which A can reasonably point) times y (number of colour samples) for sentences of the type ‘17 red pillars there’ plus x times x for ‘this-there’ sentences, each of the sentences corresponding to a different activity. There are words because the sentences can be constructed out of them; the words belong to different kinds of words because the expressions match an activity only if they have been constructed in ways that make a difference between the words. There are general names for types of building-stones because they function distinctly as count nouns. (The names — samples — of colours would be modifying names if Wittgenstein had introduced ‘d-slabs-there’ but not ‘d-red-there’.) Paragraph PI § 15 introduces proper names by introducing an activity where individual objects matter, and PI § 21 introduces new types of utterances (questions and reports) by embedding ‘old’ sentences in new activities. For this construal of Wittgenstein’s purpose to be correct, the languagegames must be taken as illustrations of languages, even if extremely poor ones which, in spite of their poverty, contain linguistically meaningful utterances or expressions. In this respect, it must be possible to liken them to full-grown natural languages, even if they are much more unlike than like. It must be possible to regard them as the beginning of a series that, through a long succession of steps, leads up to our natural languages. This reading has been challenged by Joachim Schulte. He argues that language-games which are constructed for purposes of comparison — and our examples surely belong to that class — are actually used, by Wittgen-

184

stein, to teach us the lesson that in the sense in which two constructed language-games can be compared, a constructed language-game and a natural language cannot. Studying the former could help us understand what language really is only by showing what it isn’t (Schulte 2004, 35–6). Viewing the matter in this light, Schulte argues, helps us to understand the unmistakeable irony permeating the request, in PI §§ 2b and 6a, to regard the builders’ language-game as a complete (even if very simple) language (Schulte 2004, 25f.). For such language-games form one end of a chain of objects of comparison, full-grown natural languages forming the other end — and the chain is simply too long (Schulte 2004, 33). I shall discuss exegetical questions regarding PI § 2 that would remain even if Schulte were wrong in section (5); Schulte’s interpretation endangers my whole approach, of course. What are his reasons? Besides embedding his interpretation in a larger context (which, if convincing, would give his interpretation additional weight but which I shall not discuss), he adduces three pieces of textual evidence for his conjecture of irony3 (on which everything turns): 1. In PI § 2, ‘primitive’ is used three times, and the word was inserted relatively late in the text’s history. The meanings of the three occurrences are (roughly speaking) ‘silly’, ‘simple’, and ‘immature’ (Schulte 2004, 24–6). Thus it is a deliberate, invidious modification. — 2. The request to regard the builders’ language as a complete language (PI § 2), even as the whole language of a tribe (PI § 6), is intended to provoke protest (Schulte 2004, 30), and does so (Schulte 2004, 26f.). — 3. What is described in PI §§ 2 and 8 are obvious candidates for being called ‘language-games’; however, even after introducing this term in PI §7, Wittgenstein calls them ‘language’ (Schulte 2004, 28). (I have found such passages, up to PI § 64, in PI §§ 7–10, 16–18, 26, and 41.) Now the first observation, on the deliberate use of ‘primitive’, seems to be as compatible with a serious as with an ironical reading of the request to regard the PI § 2 language as ‘complete’ — everything depends on what ‘regarding as complete’ means. The second observation is significant for an ironical reading only if what Wittgenstein was concerned with was the question whether or not the builders’ language is in fact a real language; I shall discuss this point presently. The third observation would be more significant if Wittgenstein distinguished more consistently between ‘lan3. I am aware that sensing irony may be a matter of familiarity with the author’s way of expressing himself — perception rather than interpretation. There is sure to be no living person more perceptive of Wittgenstein’s style and language than Joachim Schulte. Here, however, I have to restrict myself to ‘ponderable evidence’.

185

guage’ and ‘language-game’; just note that up to PI § 64, he also calls the PI §§ 2 and 8 activities ‘language-games’ in PI §§ 7, 16, 21, 37f., 41f., and 48, even using both words in the same sections in PI §§ 7 and 41. My own guess is that there is a tendency in Wittgenstein, where he is paying heed to his usage, to refer to the stock of expressions as ‘language’ und to utterances of the expressions as embedded in relevant activities as ‘language-game’. So the fate of the ironical reading depends on our answer to the question whether what Wittgenstein was concerned with was whether or not the PI § 2 activities might constitute a real language. If he was, then carrying on after PI § 1 with PI § 2 would be motivated by something like the following thought: ‘If Augustine’s picture were correct, then even A and B would be using a real language; indeed, their calls might possibly constitute the complete native tongue of a human tribe — what rubbish!’ (This is my own phrasing of the ironical reading.) However, viewed in this light, the dynamics of Wittgenstein’s remarks up to PI § 24 seems weakly motivated, at any rate more weakly than on a reading that has him seriously, patiently, and vividly pointing out a connection between richness of meaning and complexity of use. To establish this, I turn to the text: The PI § 8 language is ‘an expansion of language (2)’ (PI § 8), the PI § 15 language an expansion of the PI § 8 language (it could be an expansion of the PI § 2 language, but see PI § 41). Whereas in the PI § 2 language there are only unstructured ‘calls’ (PI §§ 2, 6, 19; see PI § 19b in particular), in the PI § 8 language there are structured sentences, and therefore words, as well as gestures and colour samples. The PI § 8 language, as opposed to that of PI § 2, contains general names — signifying not in their own right but in virtue of their standing in contrast with words of other kinds. This is paralleled by the fact that the PI § 8 language-game comprises many more activities the differences between which are regularly connected with the differences between the sentences. In the PI § 15 language, there are additional proper-name labels; this is paralleled by the fact that it is only in this language that there are cases where it matters to carry a certain object rather than an object of a certain kind. In the text, the PI § 21 language is not called an expansion of the PI § 8 language (I shall later treat it as such); however, the way the roles are distributed, in the first sentence, between A and B speaks in favour of their being our old acquaintances on the familiar building site. This is confirmed by the fact that the number, colour and shape of building-stones ‘in such and such a place’ can be reported. Also, the sentences: ‘Now what is the

186

difference between the report or statement “Five slabs” and the order “Five slabs!”? — Well, it is the part which uttering these words plays in the language-game’ (PI § 21), are naturally — even if not conclusively — construed as referring to reports and orders in one and the same language-game, suggesting that we are dealing with the orders from PI § 8. (This reading is not inescapable; ‘the respective language-game’ may be intended. Both readings suffer from the disadvantage that up to now, no language containing the expression ‘five slabs’ has been described. It has to be supplied by the reader in any case, and he may either invent a new language-game or graft the supplement upon the one of PI § 8.) Regardless of whether PI § 21 continues the series of languages expanding in parallel with growing sets of activities, the really decisive thing happens anyway: The introduction of two new kinds of utterances — question and report — is tied to a difference ‘only in their use’4 (PI § 21). Given his way of describing the sequence PI §§ 2, 8, 15, and 21, it would have been more than misleading, on Wittgenstein’s part, not to seriously pursue the aim of convincing his reader that meanings of expressions and relations between meanings become more varied in parallel with the growing repertoire of activities embedding their use. This is not an obvious idea from the start; it is in need of methodical, vivid elucidation. Wittgenstein describes his language-games vividly, and he enriches them step by step. He states his intention explicitly in PI § 5: It disperses the fog to study the phenomena of language in primitive kinds of application in which one can command a clear view of the aim and functioning of the words. (PI § 5a)

This is definitely a special case of the following more general claim: The language-games are rather set up as objects of comparison which are meant to throw light on the facts of our language by way not only of similarities, but also of dissimilarities. (PI § 130)

There are, of course, enormous dissimilarities between the language-games under consideration and our native tongues. However, there is a similarity, and the similarity is what counts, and it appears most clearly in the ‘clear and simple language-games’ (PI § 130): Greater expressive power requires a correspondingly larger set of embedding activities. 4. Anscombe translates ‘Verwendung’ as ‘application’. The use of a sentence is not completely, and not even most significantly, described by stating the conditions under which it can be uttered; what matters much more are the ‘practical consequences’ of uttering it (PI § 268).

187

5. Carelessness in PI § 2 In order to show that Wittgenstein’s project is successful, I shall have to complement his description in PI § 2: Let us imagine a language for which the description given by Augustine is right. The language is meant to serve for communication between a builder A and an assistant B. A is building with building-stones: there are blocks, pillars, slabs and beams. B has to pass the stones, and that in the order in which A needs them. For this purpose they use a language consisting of the words “block”, “pillar”, “slab”, “beam”. A calls them out; — B brings the stone which he has learnt to bring at such-and-such a call. — Conceive this as a complete primitive language. (PI § 2b)

Even setting aside Schulte’s general point, the purpose of this is very controversial, and I cannot do more than locate my own worries somewhere in the exegetical controversies. The first problem is presented by the sentence introducing the description: ‘Let us imagine a language for which the description given by Augustine is right’ (PI § 2b). The expression ‘the description’ can refer to the text of the Confessions quotation only if the reference is to its last sentence: ‘[…] and after I had trained my mouth to form these signs, I used them to express my own desires’ (PI §1 ). (The mason A represents little Augustine.) The quotation text preceding this sentence reports a learning story of which there is no mention in PI § 2. The mason A would then express his wishes by uttering one of the four calls. On the other hand, ‘the description’ may refer to Wittgenstein’s PI § 1b interpretation (explicitly so characterized) of the quotation: ‘the individual words in language name objects — sentences are combinations of such names’ (PI § 1b). In this case, the expressions which A utters are names; at first sight, this appears to be true and may be understood implicitly in the last sentence of the quotation because part of its first sentence states, as one result of Augustine’s learning process: ‘I grasped that the thing was called by the sound they uttered when they meant to point it out’ (PI § 1). Sentences as combinations of names are still missing in PI § 2, but they follow in PI § 8. Since Wittgenstein seems not to care about the precise reference of ‘the description’, by and large he seems to invite the reader to construe PI § 2b in such a way that the mason A expresses his wishes in using names. How is the reader supposed to react? The most natural reaction is to take the invitation literally. This has been argued forcefully by Richard Raatzsch (2003, 118–125, 208–223, 416–418). According to him, Wittgenstein’s

188

‘method of § 2’ (PI § 48) consists in constructing examples that actually exhibit the important features of the idea under consideration and do so in a manifest and unmistakeable way. Rush Rhees (1959/60) and Norman Malcolm (1989) construe the invitation as serious, too. However, in contrast to Raatzsch who thinks that Wittgenstein has succeeded, they criticize Wittgenstein for having failed completely, since, according to them, PI § 2 does not describe a language (that could be used for expressing wishes in uttering names). From Rhees’ and Malcolm’s impression, one may instead draw the conclusion — actually drawn by Fred Mosedale (1978) and later by Gordon Baker and Peter Hacker (1980) — that Wittgenstein invites the reader to argue a reductio ad absurdum: ‘This activity fits Augustine’s description; try to construe it as a language; you will see that it is impossible.’ Whichever interpretation is correct — one point seems undeniable (even if Raatzsch denies it): Wittgenstein cannot be supposed to have regarded ‘slab’ etc. as a name that designates slabs. First of all: Names are words, and there are no words where there are no structured sentences. According to PI § 19b, the builders’ language contains ‘calls’ in explicit contrast to words. Furthermore, describing anything in this language as names or signifying expressions is scrupulously avoided up to PI § 7. Here, at the latest possible place before the introduction of general names in PI § 8, ‘to name’ actually occurs, but the occurrence is not to the point at all, for Wittgenstein gives it a very special meaning: ‘In instruction in the language the following process will occur: the learner names the objects; that is, he utters the word when the teacher points to the stone’ (PI § 7a). The German for ‘that is’ is ‘d.h.’ (‘das heißt’), so the ‘that is’ is to be understood in the sense of ‘i.e.’. If the fact that a pointing gesture elicits the pronouncing of a word characterized this word as a name of what is being pointed to, then ‘Stop’ and ‘Go’ would be names of (pictures of ) red and green traffic lights, because the learner utters the words when the driving instructor points to (pictures of ) red and green traffic lights. Wittgenstein introduces the situation described in order to point to a source of the mistake of taking ‘slab’ in the PI § 2 language game for a name. That is, names and designating do not occur before PI § 8, the purpose of which is to show under what kinds of conditions ‘slab’ etc. turn into (general) names. This is confirmed by the fact that PI §§ 10–14, explicitly and to a certain length, warn us not to call expressions names ‘unless we have explained exactly what distinction we wish to make’ (PI § 13) — a distinction that, according to PI §§ 9 and 10, has not been introduced before PI § 8.

189

For my own part, I agree that introducing PI § 2 has a critical point; it is something like this: There is no foundational activity of a solitary person that succeeds in guaranteeing that this person means a certain object by a certain name (see PI § 32). However, the invitation to regard the language-game as something that fits Augustine’s description, I regard still differently: ‘This example satisfies, as well as possible, the conditions stated in Augustine’s description. Conceive it as a language. Looking closely, you will find that this appearance is due to the fact that I smuggled into the example additional elements, viz. common purpose, division of labour, authority relation, and that I introduced expressions that sound like names but in fact aren’t.’ Thus I read the last sentence, ‘Conceive this as a complete primitive language’ (PI § 2b), in this way: ‘Whatever you want to say about Augustine’s view in the light of this example, do not unwittingly use any information that is not contained in my description.’ I have to assume, however, that Wittgenstein did not do enough smuggling; he ought to have added some further elements that are necessary, according to his own views, even for a very primitive language. What further characteristics I take to be necessary I shall state in section (6); they are minimal as compared to the anthropological chasms that have scared Rhees, Malcolm, and Baker and Hacker. Forgetting about the missing items may be due to the fact that when Wittgenstein first designed the language-game of PI § 2, he had not yet reached his later insights into the connections between meaning, understanding as a way of acting, and rule-following. I concede that interpretations ought to avoid the explanation that a writer has been careless; but in this case, the explanation is corroborated by the fact that Wittgenstein did not comment on PI § 2 (mentioned for the last time in PI § 86) in the context of his cock-and-hens example: We say: “The cock calls the hens by crowing” — but doesn’t a comparison with our language lie at the bottom of this? — Isn’t the aspect quite altered if we imagine the crowing to set the hens in motion by some kind of physical causation? (PI § 493a)

The context (PI §§ 491–498) concerns the question whether it is possible to regard linguistic competence as the disposition of an apparatus to influence others, and to be influenced by others, in certain ways. Now the cock’s crowing influences the hens; but that he speaks (‘calls’) in crowing is a feeble analogy. Wittgenstein wants to say that even if speaking meaningfully is influencing acoustically, influencing acoustically does not suffice

190

for speaking meaningfully. More complex behaviour and quite different surroundings are needed for noises to qualify as meaningful utterances. Although cock and hens lack a common purpose and division of labour (there may be an authority relation), Wittgenstein’s mason A and his helper B, as far as described, are not yet sufficiently different from cock and hens as regards the possibility of interpreting their behaviour as mere acoustical influencing. Mason A may react to slab lacunas and to beam lacunas by different calls, just as a weaving spider reacts differently to supportingthread lacunas and to catch-thread lacunas; and that helper B may react differently to the calls is still easier to understand. (As explicitly stated, B, as opposed to the spider, has learnt how to react; but that does not matter to the present question.) So if Wittgenstein had juxtaposed PI §§ 2 and 493, he ought to have commented on the difference. He may simply have overlooked the incoherence. I am quite aware, of course, that my reasons for adding features to the description in PI § 2 in order to turn it into a convincing description of a very primitive language count against the assumption that Wittgenstein intended it as such. However, I see no better way to come to terms with his overall strategy concerning the series; and this strategy seems all to well evidenced by the text. 6. Language 2.1 and Alternatives Language 2.1 is the PI § 2-language enriched in the following way: (1) There are about a dozen workers involved, and normally everyone of them hears and sees what the others call and do. This is to ensure that, whenever a rule of use is disobeyed, unanimous reactions will decide the matter on the spot; presenting the case to others would require a richer language. (2) It is clear to everyone what sort of building they are constructing, and they expect each other to construct it as well and as quickly as possible. (3) All the workers have the same degree of professional experience; they work in pairs consisting of mason and helper to be formed each day according to the order of arrival at the building site. (4) After having used a building-stone, the mason can in most cases proceed with at least two different building-stones. (5) It is rule-following behaviour5 among 5. I do not specify rule-following behaviour over and above the condition that it can exist among people who cannot speak about rules; they expect the behaviour in question, from others as well as from themselves. I hope that the proponents of any specification can get along with

191

the workers (without their being able to say that they follow rules or to justify their behaviour by invoking rules) that (a) in situations where the mason has finished using some building-stone and could make use of a slab (block/pillar/beam) and where he calls ‘Slab’ (…), his helper fetches him a slab (…); (b) if the mason has called ‘Slab’ (…) and has been passed a slab (…) by his helper, he does not refuse it. Given this description, we can say that in situations where the mason has finished working with some building-stone and could make use of a slab (…), his utterance ‘Slab’ (…) has the meaning of a request that his helper pass him a slab (…). It is not simply the expression of a wish, because the helper is expected to comply. It is not an order because orders presuppose authority relations that are not implied by our description. It is not information about which kind of building-stone the mason now needs; for all others either know that anyway or know that it is wrong because the mason could proceed with a different kind. Because of this latter fact, the calls do not simply mean something like ‘Go ahead’; the use of different calls accounts for different utterance meanings. Meaning talk has been explained only regarding the four utterancetypes; utterance-types are pairs consisting of a situation type and an uttered expression. No meaning talk has been explained for the uttered expressions. In particular, the language-game description does not warrant the assumption of either sentences or words, and not of general names, of course. The utterance meanings have been specified according to the principles of not adding details which make no difference in use and of avoiding details which are inconsistent with the use. The specification ‘request that the helper pass the mason a slab’ can reoccur as specifying the meaning of an utterance ‘blob’ in an alternative Language 2.2. Its use differs from that of the original in that the workers are paving the beach with patterns that are going to be used in a religious ritual. Beauty and novelty matter most, and there is a general laziness about their work, suitable for dozing off for some moments and getting new inspiration. They use ‘blob’ and three other calls, corresponding to the building-stone calls from Language 2.1. The helper is, however, only weakly expected to pass a building-stone after a call. If he does, there is nothing like a thank you from the mason; if he does not, the mason will either repeat his call or fetch the building-stone himself or doze off and the use I do make of the concept. In so far as invoking rules for the justification of actions is taken to be essential for rule-following, the concept has to be replaced by that of something like behaviour which is manifested as a matter of course (PI § 238).

192

come up with a new call. All the same, the calls are requests rather than proposals because the helper cannot initiate the use of a stone which the mason has not called for. Despite the weakness of the obligation on the part of the addressee, the calls are requests rather than simple expressions of wishes because there is neither thanking nor an expectation of return favours. So the calls are inter-translatable with those from language 2.1. This possibility gives us a chance to make the concept of use a bit more precise. Uttering ‘slab’ or ‘blob’ is interwoven, in both language-games, with different activities: purposeful, single-minded cooperative behaviour and a strong expectation on the one hand, and calmly responsive behaviour and patient indulgence, on the other. Since the utterances are interwoven with different activities, describing these activities might be expected to warrant the assumption of different meanings; but it does not. The difficulty can be solved by stipulating that different meaning assumptions imply different modifications of the language-users’ activities, and that identical meaning assumptions imply identical modifications. That is, where B is less strongly expected to do something from the start, as in helping to produce ornaments, than he is in a different case, as in building, he is weakly expected to fulfil a corresponding request in the first case, but strongly in the second. The features about use that matter to meaning talk are modifications of activities; whereas the activities may be as manifold as you please, modifications of activities may be identical, and this will warrant talk of identical meanings. Thus principles are necessary that connect meaning assumptions with descriptions that specify modifications of activities; in this way, facts are specified about use that are relevant for meaning talk. For Languages 2.1 and 2.2, such a principle may roughly be stated as follows: ‘If it would be useful for common purposes if the addressee did x, and if this is known to addressee and speaker, then if the speaker requests the addressee to do x, the addressee is more obliged to do x than he was before.’ (This is not complete; e.g. it does not distinguish a request from an order.) The relevant fact about use is the increase in the addressee’s obligation — depending on the starting point, the increasing obligation will reach different levels. To distinguish requests from orders (and their kin) such principles have to be more detailed. Orders6 of the kind given by an officer to a private are characterized by formally certified authority on the one hand (without 6. I am here using ‘to request’ for German ‘auffordern’ and ‘to order’ for German ‘befehlen’. There is no one-to-one mapping between English and German ‘exercitives’, to use Austin’s term.

193

such authority, no utterance will succeed in becoming an order) and by strong expectations of compliance and severe threats of sanction, on the other. Of course, if a teacher orders a pupil to do something and the pupil disobeys, he will not be marched to the wall. So there are different kinds of order, too. (You would not say in German nowadays that the teacher’s order is a ‘Befehl’, even if Wittgenstein consistently talks this way.) Thus the principles we need to connect meaning assumptions with use descriptions achieve two things: they specify that whatever activities there are be modified in certain ways; second, they specify the same or different modifications for the same or different meaning assumptions. I shall now describe an expansion of Language 2.1, Language 2.3, in order to prevent a frequent sort of misunderstanding. In the language-games under consideration, Wittgenstein notes this sort of misunderstanding only in PI § 21. The misunderstanding is to think that two say two different things you need two different expressions (or the same expression accompanied by two different pointing gestures or other indexical means). In Language 2.1, just four different things can be said, and there are four corresponding expressions. Language 2.3 contains the following additional regularity: (6) It is rule-following behaviour among the workers that in situations where the mason is still occupied with a building stone and where the helper has nothing to do and where the helper calls ‘slab’ (…), the mason either calls ‘slab’ (…) and the helper fetches a slab (…), or the mason remains silent until he has finished dealing with the building-stone and the helper remains inactive. Given this description, helper B’s utterance can either be interpreted as a question whether the mason next wants a slab or as a suggestion to fetch mason A a slab; A answers the question or suggestion in the affirmative by repeating B’s call and in the negative by ignoring it. In direct speech, the utterance might be translated as ‘Wouldn’t you like to have a slab next?’ It is a blending of a question and a suggestion. The first interesting fact about the expansion is that the use of one and the same expression, under different circumstances (including, in this example, a change of speaker) can make for utterances with different meanings. The second interesting fact is: Use by itself does not give us a reason for deciding between question and proposal. Meaning assumptions are warranted by use descriptions only in so far as the use descriptions could not with equal right be taken to warrant an alternative meaning assumption. In this example, we have to define an illocutionary force for simple languages that do not differentiate between questions and proposals in the ways possible in a complex language.

194

7. Language 8.1 and an Alternative In PI § 8, Wittgenstein introduces the use of demonstratives with pointing gestures, of numerals, and of colour-samples. It is unclear from the text whether questions of well-formedness can arise and whether word order plays a role. Language 8.1 is simpler; it includes, in addition to four building-stone expressions, only numerals and demonstratives with an equivalent to pointing gestures. Word order will be relevant. The demonstratives and gesture equivalents are artificial in order to avoid ‘obvious’ interpretations. The following details of use will be modified: It is rule-following behaviour among the workers that (a) in situations where it could be useful for a couple’s work if p were the case and where the speaker utters a sentence which means that p, the addressee sees to it that p; (b) if the speaker has uttered a sentence which means that p and if the addressee sees to it that p, the speaker accepts this. (Note that the helper can speak, too; it can of course be helpful, from time to time, if either one engages in transporting building material from one place on the building site to another.) The description now refers to utterances of sentences with given meanings rather than to utterances of given expressions; this is how use descriptions will be seen to be connected with sentence-meaning talk (in addition to the connection with utterance-meaning talk). Sentence meaning, in turn, will be specified by way of assigning meanings to sub-sentential expressions; this is how use descriptions will be seen to be connected with word-meaning talk. The connection is supposed to be the same as what it amounted to in the case of utterance meaning: The use descriptions must be such that they are required by the meaning specifications; no different use description must be suggested by the meaning specification with equal right. If all rival meaning specifications have been tried and there is one that passes the test best, then it is the one that is warranted by the use description. Phonetically, we find expressions of three structures: a-slab-id, id-slabid, and id-id; a can be replaced by any of b,c,…,j, and slab by pillar, block, and beam. Id is regularly accompanied by staring. Whenever a speaker utters expressions that are not covered by these specifications, the workers within earshot tend to look in a bewildered way, and the speaker will then normally utter a normal expression. Furthermore, we get utterance meanings which are consistent with the use as described above, if we assume that

195

• a(…j)-slab(…)-id means that 1(…10) slab(…) is (are) where the speaker stares (applying only when this place is empty). (I assume that ‘slab’ from Language 2.1 is replaced by uttering a-slab-id, with the speaker staring at the place where he is; ‘slab’ by itself might survive as a colloquial abbreviation.) • id-slab(…)-id means that all slabs(…) which are where the speaker stares at first are where the speaker stares later (applying only when there is at least one slab at the first place and the second place is empty). • id-id means that all building-stones which are where the speaker stares first are where the speaker stares later (applying only when there is at least one building-stone at the first place and the second place is empty). These specifications, in turn, follow from some obvious parsings as well as meaning specifications for sub-sentential expressions: • id-slab(…) denotes all slabs(…) which are where the speaker stares; • id- (without following slab(…)) denotes all building-stones which are where the speaker stares; • -id denotes the place where the speaker stares; • a(…j) means 1(…10). Constructing the sentences with their meanings from these dictionary entries is a matter of elementary meaning talk. This is why we need not bother about differences between ‘means’ and ‘denotes’, let alone about whether meanings are functions or sets of possible worlds or what have you. We are concerned with interpreting meaning talk in terms of use description, and that’s it. Specifying meanings for parts of sentences is avoidable (there are no new sentences); but people will do it, and in fact the description is much shortened because our three sentence-meaning schemata contain 45 meaning specifications; the word semantics does with 16. An utterance of idslab-id, if the sentence is meaningful (i.e. if the sentence semantics applies) and if it is being uttered under the circumstances specified for requests in 8.1 and 2.1, means that the speaker asks the addressee to see to it that the indicated slabs are at the indicated place — i.e. he asks him to take them there. Utterances of meaningful sentences are meaningless if made under

196

inappropriate circumstances. An utterance of the meaningful sentence islab-id, if made under circumstances appropriate for requests, is a request to take nine slabs there even if there aren’t enough slabs left, so that the request cannot be fulfilled. My impression is that slab (the word, not the utterance) in Language 8.1 is a general name for slabs. Can we express this fact in terms of facts about use? That is, is Wittgenstein right to say: ‘Now what do the words of this language signify? — What is supposed to shew what they signify, if not the kind of use they have? And we have already described that’ (PI §10a)? To be a general name in 8.1 must be a role, shared by slab with block, pillar, and beam, but with none of the other words, that can be specified by that description of 8.1 which did not yet contain meaning specifications. This can be done because 8.1 is so simple: x is a general name in Language 8.1 for objects of kind k if and only if it is rule-following behaviour among speakers of 8.1 (a) that if the speaker utters a sentence containing x as an element (under the circumstances described above) then the addressee sees to it that something is the case for an object of kind k, and (b) that the speaker accepts this. (Trivial interpretations are ruled out by the condition of rule-following behaviour which implies that the regularities in behaviour are law-like.) This is a use-talk specification of being a general name in 8.1. It is not a use-talk specification of a general name in general. Such a specification will be vastly more difficult to give when more uses of general names come into play; it may be greatly simplified if the language contains sentences with the meaning of ‘This is a …’, or procedures like explanations by examples. In the alternative Language 8.2, let id- be replaced by hoc, -id by huc (to be pronounced ‘hook’), both accompanied by our usual pointing gesture, hoc being used in pointing to a collection of building-stones (or one) and huc in pointing to an empty place. Instead of slab (…) we now have one word tale (to be pronounced ‘tah-le’), accompanied by our usual pointing gesture which must be directed toward one building stone or a collection of identical ones. a….j are replaced by a, a0, aa, a00, a0a, aa0, aaa, a000, a00a, a0a0, the series not breaking off, however. There is no word order; hoc and tale do not occur together in a sentence. There are only the sentence structures exemplified in the following paradigmatic meaning specifications:

197

• a0a, tale, huc (in any order) means that 5 building stones like the one pointed to in uttering tale are in the place pointed to in uttering huc (applying only if there is just one kind at the first place and if the second place is empty). • hoc, huc (in any order) means that the bulding-stone(s) pointed to in uttering hoc are in the place pointed to in uttering huc (applying only if there is at least one building-stone at the first place and if the second place is empty). Parsing and word semantics are obvious. Id-id from the original Language 8.1 is translatable into hoc-huc and huc-hoc; e-slab-id is not translatable into a0a-tale-huc even when the speaker of the second sentence is pointing to a slab in uttering tale. Thus, although the numerals up to 10 translate into each other, no sentence containing a numeral in either language is translatable into a sentence of the other language. However, an utterance of (meaningful) e-slab-id is the same request (given appropriate circumstances) as an utterance of a0a-tale-huc if in uttering tale, the speaker points to a slab (and if the indicated point is the same in both cases), for it is a request to take 5 slabs there; thus these utterances are intertranslatable. Clearly, -id translates into huc whereas id- translates into hoc only if not followed by a building-stone word. 8. Language 15.1 and an Alternative PI § 15 introduces what Wittgenstein calls proper names of tools (PI § 41). Our Language 15.1 differs from 8.1 by n additional different expressions of the form Ni -id (i ranging from 1 to n). Uttering Ni -id in appropriate circumstances means requesting that the addressee takes tool no. i (according to our numbering — we need not care about how the builders recognize their tools) to the indicated place (i.e. the utterance has the effect of such a request). The obvious way to describe the sentence semantics of Language 15.1 is the following supplementation to that of Language 8.1: • Ni-id means that tool no. i is in the indicated place. This we get from supplementing the word semantics of Language 8.1 with

198

• Ni denotes tool no. i. That this is our easiest way to come to terms with the impact of the corresponding requests is the reason why the use of Ni warrants the assumption that it is a proper name of tool no. i. Being a proper name in Language 15.1 can then be explained in terms of use in a way parallel to the explanation of being a general name in 8.1. For this purpose, it is completely insignificant for tool no. i to have the label Ni attached to it. Note that in PI § 15, Wittgenstein says: ‘The word ‘to signify’ is perhaps used in the most straightforward way when the object signified is marked with the sign.’ (My emphasis; the German ‘den es bezeichnet’ is clearly a qualifying relative clause, meaning ‘provided it is signified by the sign’.) Just imagine the names consisting in acoustic modifications of -id. You cannot possibly label a tool with a modulation of ‘there’ (as distinguished from a modulated ‘there’). Let us use this strange kind of names for the alternative Language 15.2 and call them Mi ; instead of containing additional strings, the alternative will distinguish a standard pronunciation of -id from m different ways of singing it, the sentences consisting of nothing but the musical demonstrative. They can have the same meanings as some Ni -id insofar as both linguistic communities use the same set of tools. The proper names in 15.2 will be intonation contours rather than expressions; but they are names for exactly the same reason as are those in the original. 9. Language 21.1 As an expansion of Language 8.1, I sketch Language 21.1 (very incompletely described in PI § 21) for the purpose of underlining the point of Language 2.3, a point that Wittgenstein here makes explicitly without really illustrating it, viz. that the expression may be one and the same where utterances of it have different meanings. Here is Wittgenstein’s suggestion: Imagine a language-game in which A asks and B reports the number of slabs or blocks in a pile, or the colours and shapes of the building-stones that are stacked in such-and-such a place. — Such a report might run: “Five slabs”. (PI § 21)

To achieve this without introducing additional expressions (sentences or words) into Language 8.1, we describe new uses for the sentences id-

199

slab(…)-id and id-id: If uttered under circumstances new to 8.1, they have consequences in rule-following behaviour that are new to 8.1. I give a translation manual for the new elements first; what counts are the subsequent explanations: • Questioner, staring at the same place both times: ‘id-slab-id’: ‘How many slabs are left there?’ • Answerer, staring at the point where the questioner stared: ‘a-slab-id’: ‘There is one slab left there.’ ‘id-id’: ‘There are no slabs left there.’ • Questioner, staring at the same place both times: ‘id-id’: ‘How many building-stones are left there?’ • Answerer, staring at the point where the questioner stared: ‘a-slab-id, e-pillar-id’: ‘There is one slab left there, there are five pillars left there.’ (Sentences for ‘straight’ answers like ‘a-id’ or ‘id-a-id’ are not antecedently available in 8.1.) ‘id-id’: ‘No building-stone is left there.’ That a question is being asked (rather then a request uttered) is determined by the following set of conditions: The addressee is either, as a mason, busy dealing with a building-stone or, as a helper, busy transporting buildingstones; the speaker is less well-placed than the addressee to see or find out how many and what kind of building-stones are on a given heap. That a report is being given (rather than a request or a question being uttered) is determined by the condition that the speaker has been asked a question. The remainder of the use is as follows: If the answerer has been asked a question, he has to look or go and find out and then give a report. (This consequence marks the question as a question.) If the answerer has given a report, the questioner is entitled to rely on the report wherever that may be relevant. (This consequence marks the report as a report.) Note that Wittgenstein is right: ‘Now what is the difference between the report or statement “Five slabs” and the order “Five slabs!”? — Well, it is the part which uttering these words plays in the language-game’ (PI § 21). In addition to the requests (which Wittgenstein calls orders) ‘a-slab-id’, ‘id-slab-id’, and ‘id-id’ in Language 8.1, we have utterances of identical sentences id-slab-id and id-id as questions, and utterances of identical sentences a-slab-id and id-id as reports answering the questions, where utterances of id-id are different reports, depending on the preceding questions. There are no new expressions (as would be there if ‘the tone of voice’, PI § 21,

200

were different); rather, there are new utterance conditions (role on the building site, working condition, information accessibility, preceding question) and new obligations and entitlements (obligation to inquire and to answer, entitlement to rely on some fact) — new uses, for short, that characterize the utterances as questions or reports. 10. Theoretical Commitments In my attempt to spell out Wittgenstein’s replaceability idea in a convincing manner, I had, explicitly or implicitly, to rely on the following ideas: A request needs an intelligible motive. The addressee of a request is generally expected to comply to a certain degree whereas the addressee of an expression of a wish isn’t. A request increases an antecedent level of obligation to do what one is requested to; it does not oblige you to do it tout court. In contrast to a request, an order requires formal authority and has to be enforced by sanctions. You can only ask for information you are expected not to possess, and you can ask only someone who is expected to have it or to be able to get it. If you ask for information, others are entitled to expect you not to possess it, so that they can report it to you by answering. You cannot report a fact to anybody who is expected to know it anyway; but if you report it to him, he is entitled to trust you. Your suggestion that you do something in both parties’ interest may have the same results (as regards obligations and entitlements) as the question whether you should do it. That is, I have made use of principles that restrict the very possibility of requesting, ordering, reporting, asking for information, and making a proposal to conditions characteristic of each of these illocutionary forces, and that connect their successful execution to specific changes in the expectations of competent speakers that will occur on account of utterances with such illocutionary forces. Spelled out in detail, such a principle might look like this: (Precondition) The speaker S is expected to know whether or not p is the case. The addressee A is expected to be interested in learning whether or not p is the case. Then if (speech act) the speaker S reports to the addressee A that p is the case, (result) the addressee A may rely, at the expense of the speaker S, on its being the case that p.

201

Andreas Kemmerling has suggested to call such principles ‘bridging principles’, and I am happy to accept the word because it highlights their function: They actually build a bridge from meaning talk to use talk. For the expectations, obligations, entitlements and so on are what is relevant about speakers’, addressees’, and third parties’ behaviour in connection with utterances with given meanings. E.g., what does it mean to say that the speaker is expected to know whether p is the case? He is in a position to know; he is expected to act on the knowledge; people would be surprised if he did not; he himself would be puzzled if he had forgotten. What does it mean to say that the addressee may rely, at the speaker’s expense, on its being the case that p? If the addressee continues building in such a way that he needs the number of slabs reported as being left and if in fact there are not enough left, he may become furious with the helper, and the latter has to put up with it (accepting anger is an important informal way of paying social debts). These principles are, of course, highly controversial; and since they can be defended only as belonging to an entire system that satisfactorily differentiates between speech acts, they form a controversial theory. As such, there are more popular alternatives, for instance David Lewis’ modus definitions in Convention (1969) or the Gricean approach of elucidating the concept of meaning on the basis of the concept of intention. Whatever bridging principles you prefer — you have to use some. The bridging principles I relied on define which part of use talk is suitable for my attempt to substantiate Wittgenstein’s replaceability thesis. Which use talk is relevant to warranting meaning talk is determined by which facts about use enter into the bridging principles. They connect meaning with use, and they partly determine what it is that they are expected to connect — very much like a theory that is invented in order to explain some kind facts and which is then used to determine what counts as being of that kind. The bridging principles I relied upon require a distinction between utterance meaning and expression meaning, a distinction that, far from being generally accepted, plays a negligible part in contemporary philosophy of language. Meaningful utterances are nothing but Austin’s speech acts, of course, although I regard only one of his infelicity conditions as decisive (the violation of which he called ‘breach’). That speech-act theory is a theory follows from facts like this: It explains why the explicit performative ‘I state that the dog bites’ is a statement, but not of the fact that the speaker is stating something; it explains why there can be no ‘token-

202

reflexive’ statement to the effect that it itself is true. That the meanings of utterances of sentences because of indexicality may differ in interesting ways from meanings of the sentences uttered is common knowledge; it is far from being common knowledge that an utterance of the sentence ‘This dog bites’ may, depending on circumstances, be a warning that the addressee refrain from approaching the dog, or a recommendation that the addressee who is looking for a watchdog buy the dog, or as a witness’s testimony in court that the dog has a habit of biting. Bridging principles lead us from utterance meanings to use; so if we accept the distinction, we have only gone halfway and are in need of a bridge from semantic meaning to utterance meaning. It is no use claiming that this result just means that speech-act theory is a bad means for defending Wittgenstein: for if we renounce it, we need some other means to solve the same problems, even if in different clothes, and such means will clearly be as theoretical and explanatory as speech-act theory — just think of theories postulating literal meanings or implicatures. We need the connection between use and semantic meaning because much everyday, innocent meaning talk concerns the meanings of expressions rather than of utterances; and since there is no such thing as the typical use of a sentence like ‘This dog bites’, we cannot expect the missing link to be identical for all languages (as we could in the case of bridging principles). Let me give an example of a rule that might possibly, though not very probably, connect the meaning of a sentence with one kind of meaning of its utterances in English: If the speaker is expected to know whether p is the case, and if the addressee is expected to be interested in learning whether p is the case, then if the speaker utters a sentence which (semantically!) means that p, he thereby reports to the addressee that p. — Having been taken, by this rule, from semantic meaning to utterance meaning, you proceed from utterance meaning to use by the bridging principle stated above for reporting. It seems to follow from these observations that even the weak replaceabilty assumption — questions and controversies about meaning are to be decided by appeals to use — can be upheld only at the cost of non-trivial theoretical commitments. Whoever forswears the means I have proposed will have to use others that will not be trivial, either. The decisive point is this: You cannot state that it is possible to replace meaning talk by use talk unless you are prepared to say how to do the replacing; and if you attempt to say it, you will realize that you need theoretical means. They are hypothetical, pace PI § 109, and they are sure to be controversial, pace

203

PI § 128, but heaven be praised: They involve neither metaphysical talk nor hidden entities.7

REFERENCES Baker, Gordon P., Peter M. S. Hacker 1980: An Analytical Commentary on the ‘Philosophical Investigations’, Vol. I: Wittgenstein, Understanding and Meaning, Oxford: Blackwell. Lewis, David 1969: Convention, A Philosophical Study, Oxford: Blackwell. Malcolm, Norman 1989: ‘Language Game (2)’, in D. Z. Phillips, P. Winch (eds.): Wittgenstein: Attention to Particulars, New York: St. Martin’s Press, 35 – 44. Mosedale, Fred 1978: ‘Wittgenstein’s Builders Revisited’, in Second International Wittgenstein Symposium, Wien: Hölder-Pichler-Tempsky, 340–343. Raatzsch, Richard 2003: Eigentlich Seltsames. Wittgensteins ‘Philosophische Untersuchungen’, Band I: Einleitung und Kommentar PU 1–64, Paderborn: Schöningh. Rhees, Rush 1959/60: ‘Wittgenstein’s Builders’, Proceedings of the Aristotelian Society 60, 171–186. Schulte, Joachim 2004: ‘The Builders’ Language — The Opening Sections’, in: E. Ammereller, E. Fischer (eds.): Wittgenstein at Work, Routledge: London, New York, 22–41.

7. Thanks for an unusually helpful discussion are due to all who participated in the Ulm Wittgenstein Workshop. In particular, I thank Andreas Kemmerling for opening the discussion with prepared comments. Only the text of sections 6 to 8 has survived; the remainder of the paper is completely new. On top of his share of such thanks, I am immensely indebted to Joachim Schulte for once more making my philosophical ‘insights’ accessible to readers of English.

204

Grazer Philosophische Studien 71 (2006), 205–225.

THE TRANSITION FROM CAUSES TO NORMS: WITTGENSTEIN ON TRAINING Wolfgang HUEMER Universität Erfurt Summary Anti-reductionist philosophers have often argued that mental and linguistic phenomena contain an intrinsically normative element that cannot be captured by the natural sciences which focus on causal rather than rational relations. This line of reasoning raises the questions of how reasons could evolve in a world of causes and how children can be acculturated to participate in rulegoverned social practices. In this paper I will sketch a Wittgensteinian answer to these questions. I will first point out that throughout his later philosophy Wittgenstein draws a sharp distinction between “teaching” and “training”: newly-born children are trained (conditioned) to react to specific stimuli in specific ways, which then allows them to acquire concepts and follow rules. I will then show that this picture presupposes a strong analogy between concepts and capacities, which is also present in Wittgenstein’s later philosophy. In the last section I will point out that Wittgenstein only discusses the ontogenetic question of how individual children can acquire speech, but not the phylogenetic question of how rule-governed behavior could evolve in the first place. I will argue that this strategy should not be seen as a shortcoming, but rather as an expression of Wittgenstein’s approach that can be characterized as naturalistic in a wide sense.

When philosophers reflect about perception, they typically focus on the question of how our empirical beliefs and knowledge can be justified by facts in the world, mediated by perceptual experience. Unlike neurophysiologists, who analyze the causal relations that take place in our nervous system and eventually bring about perceptual experience, philosophers ask for justification. Reductionist attempts to reduce philosophical theories of perception to their scientific counterparts notoriously face the difficulty of coping with the fact that justificatory relations contain an intrinsically normative element that cannot be accounted for in the language of sci-

ence.1 Anti-reductionist approaches, on the other hand, face the difficulty of explaining where the normative aspect of the mental comes from. After all, persons who perceive and hold empirical beliefs are part of the physical world, their mental episodes can be conceived as a function of their brain, the working of which can be explained by neurophysiology. According to an old ideal it is the ultimate goal of science to formulate one universal theory of everything. If this is more than a shallow claim, we should expect that scientific theories are able — at least in principle — to account for our having mental episodes as much as they can account for there being black holes, molecules, or electric fields. In consequence, antireductionists (given that they do not want to buy into ontological pluralism) stand in need to provide a perspective of how normativity can emerge from the level of causality without, of course, unveiling systematic relations between the two levels, for that would betray their anti-reductionist program. One might not want to go so far as to push them to elaborate a detailed account, but one can expect them to provide an outlook on which place normativity holds in a purely physical world, in which relation, in other words, the realm of norms stands to the realm of causes. In this paper I will analyze how Wittgenstein conceives the relation between norms and causes. I will first focus on Wittgenstein’s remarks concerning the acculturation or, to use Wittgenstein’s term, the training [Abrichtung] of children, a process by which they become part of a community that is constituted by (rule governed) social practices. Next I will discuss Wittgenstein’s remarks on the relation between the level of causes and that of norms, which provide interesting insights into the topic, but also have clear ramifications. Wittgenstein’s primary goal, I will argue, is not to provide a satisfying answer to this question, but rather to shift our perspective, challenging, as he does, our urge to formulate the question in the first place. We have to content ourselves that there is no further explanation for the level of norms — “explanations come to an end somewhere” (PI § 1). Our philosophical worries are relieved not by developing theories that explain how the realm of norms could evolve from the realm of causes, but by understanding the senselessness of this project and acknowledging that reasons, like causes, are part of our everyday world. The realm of the 1. Sellars brings it to the point when he states that “the idea that epistemic facts can be analyzed without reminder — even ‘in principle’ — into non-epistemic facts, whether phenomenal or behavioral, public or private, with no matter how lavish a sprinkling of subjunctives and hypotheticals is, I believe, a radical mistake — a mistake of a piece with the so-called ‘naturalistic fallacy’ in ethics.” (Sellars 1997, 19, § 5)

206

normative, thus, does not stand in need of explanation; “It is there — like our life” (OC § 559). 1. Wittgenstein on Training Throughout his later philosophy, Wittgenstein uses examples of learning, especially of language acquisition, to illustrate his position. Since he does not develop a systematic theory of learning — a fact, however, that, given Wittgenstein’s notorious aversion to theories, should not come as a big surprise — and since we can reasonably expect that the first language games a young child comes to play are simple and clear, one might be tempted to assume that Wittgenstein uses these examples merely for didactic reasons. In fact — as Meredith Williams (1999, 188f.) already indicated — most of the secondary literature regards Wittgenstein’s appeal to learning as an expository or a heuristic device, and thus tends to overlook the fact that it plays an important systematic role in Wittgenstein’s later philosophy. Furthermore, Williams rightly pointed out that the process of training should be seen as “pivotal in creating the logical space for the very distinction between the grammatical and the empirical. […] understanding the role learning plays sheds light on the nature of normativity itself ” (Williams 1999, 189). I will now turn to show how Wittgenstein’s remarks on language acquisition can sketch a picture of how we should imagine the transition from pattern-governed to rule-conforming behavior. These remarks are characterized by a sharp underlying distinction between teaching and training that is operative in all of Wittgenstein’s later philosophy; a distinction not always fully appreciated in the secondary literature. In the English-language literature, this might (at least in part) be based on a problem of translation. Already in § 5 of the Philosophical Investigations Wittgenstein states: “Das Lehren der Sprache ist hier kein Erklären, sondern ein Abrichten.” Anscombe translates this sentence as “Here the teaching of language is not explanation, but training,” thus translating the German verb “abrichten” with the English “to train.” Though the translation is definitely literal, it is crucial to note that there is an important difference between “abrichten” and “to train”: while the English word “to train” can be used for persons or animals — we speak of a trained piano player and can train children to ski or to ride a bike — the German “abrichten” is exclusively used for animals, for training dogs to sit down on the command “sit,” or horses to gallop when the rider performs a certain bodily

207

movement (typically she increases the pressure of the left leg and keeps the reins loose). The German word for training a child to play chess, to ski, or to speak a language is “lehren” or “beibringen” rather than “abrichten.” What Wittgenstein has in mind, thus, comes closer to conditioning than to training. He talks about a process that sets up stimulus-response patterns that do not involve any kind of intellectual activity on the side of the trainee. Unlike teaching or explaining, it does not involve any kind of linguistic instruction, but is based merely on the reinforcement of patterns of behavior. In his Brown Book, a text he dictated to his students in English, even Wittgenstein uses the word “to train” in places where in his German version of the text he uses “abrichten.”2 At the first occurrence of the word, he does add a clarifying remark, though: “I am using the word ‘trained’ in a way strictly analogous to that in which we talk of an animal being trained to do certain things. It is done by means of example, reward, punishment, and suchlike” (BB 77). Thus, when Wittgenstein emphasizes that teaching language is training rather than explanation, he insists that the first steps of language acquisition can be fully explained by a setting up of stimulus-response patterns; they take place at the level of patterngoverned rather than rule-conforming behavior. We condition children to occupy their first positions in the language game which, from the adult speakers’, but not the children’s point of view, contain an intrinsically normative element. Wittgenstein draws our attention to this difference when he remarks: I have been trained to react to this sign in a particular way, and now I do so react to it. But that is only to give a causal connexion; to tell how it has come about that we now go by the sign-post; not what this going-by-the-sign really consists in. On the contrary; I have further indicated that a person goes by a sign-post only in so far as there exists a regular use of sign-posts, a custom. (PI § 198)

While the adult members of the community are following a rule when they react to the signpost in a particular way, children are merely behaving in the way they were conditioned to behave. In consequence, the causal connection, the process of training, can explain only the children’s behavior, but not that of the adults who have the freedom to react in a different way. It is noteworthy, however, that children are trained by adults 2. One year after dictating the Brown Book to his students, Wittgenstein started to work on a (slightly altered and expanded) German version of the text, probably to prepare it for publication (cf. Rhees 1989, 10).

208

who already do engage in social practices that do contain an intrinsically normative element.3 The prominence of language acquisition in Wittgenstein’s later philosophy becomes apparent already in §1 of the Philosophical Investigations, where he introduces the Augustinian picture of language. According to (Wittgenstein’s) Augustine, we acquire language by hearing adults uttering words when drawing their attention to specific objects in their environment. By observing their bodily movements, the expression of the face, or the play of the eyes4 as well as the objects towards which the latter are directed, children draw a connection between words and objects; they so learn to understand the meaning of the words uttered. Wittgenstein’s critique of this Augustinian picture has two strands. First, he criticizes atomistic theories of meaning, according to which meaning can be explained on the basis of the relation of reference between single words and single objects. Atomistic theories hold that we can learn the meaning of words like “red,” “dark,” or “sweet”5 in isolation, independently of the rest of language. Wittgenstein, as is well known, replaces this view with a holistic theory of meaning, according to which the meaning of a word is determined by how it is used by the speakers of a language.6 There is, however, a second strand in Wittgenstein’s argument: he criticizes not only atomistic theories of meaning, but also turns against Augustine’s picture of language acquisition. Augustine cannot explain, 3. This explains why they are often eager to interpret the behavior of the child as “following a rule,” even though it does not yet. In his remarks on reading (PI §§ 156–73) Wittgenstein discusses the difficulties to decide when the child’s behavior is no longer a conditioned reaction, but it is rather following a rule. How can the teacher tell that the pupil knows to read? Is there a first word it can read? Wittgenstein denies this when he states: “Nor can the teacher say of the trained [vom Abgerichteten, which Anscombe translates misleadingly as “pupil”]: ‘Perhaps he was already reading when he said that word’. For there is no doubt about what he did. — The change when the pupil began to read was a change in his behaviour; and it makes no sense here to speak of a ‘first word in his new state’” (PI § 157). At some point we are justified in saying that the child is now following a rule, but there is no single moment when we can say that now, “for the first time,” the child is following a rule. Along similar lines we can argue that at some point a child does have mental episodes, but it does not make sense to speak of a “first mental episode.” I have discussed this point in Huemer 2005, 61ff. 4. I am quoting here Augustine’s formulations as cited in PI § 1. 5. I am using Wittgenstein’s examples (cf. PI § 87). Empirical concepts are normally thought to be the first that one can acquire, for (according to empiricist theories) they require only to establish a connection between a word and a certain kind of sense-datum. 6. A concise expression of this view can be found in the Philosophical Investigations: “For a large class of cases — though not for all — in which we employ the word ‘meaning’ it can be defined thus: the meaning of a word is its use in the language” (PI § 43).

209

Wittgenstein argues, how children acquire their first, but rather, how they learn a second language. The child, in other words, would need to already have a language (she would have to be able to think about objects, which presupposes her having concepts) in order to acquire language. Wittgenstein summarizes his critique in the following way: And now, I think, we can say: Augustine describes the learning of human language as if the child came into a foreign country and did not understand the language of the country; that is, as if it already had a language, only not this one. Or again: as if the child could already think, only not yet speak. And “think” would here mean something like “talk to itself ”. (PI § 32)7

Thus, when Augustine claims that children learn to establish a connection between words uttered by adults and objects in their direct environment, he presupposes too much: he must hold first that children already understand what an ostensive definition is and what role it plays in our language game; second, that they are already informed about the referential function of language before they become players of a language game; and third, that they are already in a position to perceive objects as objects, to perceive, for example, chairs as chairs. Wittgenstein addresses this last point when he notes that Augustine treats the child as if it were able to already think — where “think,” as he specifies, “would here mean something like ‘talk to itself ’” (PI § 32). In order to see objects, one must be able to categorize them as falling under a certain concept. In order to see this thing in front of oneself as a green apple, for example, one must be able to categorize it under the concept apple. Hence, in Augustine’s picture, the child must already possess concepts to learn a language. Since children have not yet mastered any of these aspects before learning a language, we cannot teach them the meaning of words by explanation — neither by nonverbal, ostensive definition, nor in the way our French teacher explained us that “pain” in French means “bread” in English — for that would presuppose that they have already acquired a language, have concepts, or have at least mastered a practice of non-verbal explanation. We rather train them to speak — in the drastic sense of abrichten explained above.8 7. I have slightly altered Anscombe’s translation, changing “strange country” to “foreign country” in order to avoid the connotation of “odd”, which is not present in Wittgenstein’s German formulation “fremdes Land.” 8. To the ears of German-speaking readers, Wittgenstein’s affirmation that we have to abrichten our children is quite drastic, since that seems to put human children at the level of animals, which would undermine all principles of enlightened education to which we have — fortunately enough — grown acquainted in the last few centuries. We can put Wittgenstein’s point less drasti-

210

Before acquiring language the child’s behavior is guided by natural impulses; it does not yet contain a normative aspect. Only in the process of training (in the sense of Abrichtung) do children become players of the language game, which is constituted by rules. They make a transition into the realm of the normative and thus become fully accepted members of our community. Their behavior is not any longer guided only by their instincts, but also by social norms; it can now be corrected by other members of the community. The second aspect of Wittgenstein’s critique of the Augustinian picture of language, thus, consists in his diagnosis that Augustine does not distinguish between the children’s pattern-governed behavior and the adults’ rule-conforming behavior. A newly-born child might cry and thereby signal hunger to her parents. The adults can react to this signal only by trying to satisfy the child’s need; they cannot question or discuss her justification to cry at this moment. The behavior of the newly-born is guided by purely biological mechanisms that play an important role: they ensure the child’s survival, but do not contain an intrinsically normative aspect. Things are different when a ten year old child communicates her being hungry. The parents might try to explain her, for example, that it is not a good idea to eat now, since dinner will be ready in an hour. Other than the newly-born, the ten year old is more likely to be open to reasoning of this kind since, in the process of being acculturated, the child makes a transition from craving to wishing to eat,9 which shows that only the latter cally by stating that in the Augustinian picture newly-born children are treated like small adults with a faculty to reason. All parents, even those who try to follow the most enlightened principles of education, however, cannot expect their newly-born children to understand arguments; they rather have to train them into certain patterns of behavior. Once children have acquired minimal capacities of reasoning, parents can try to explain them their educational measures at the respective level of understanding. “The basis of each explanation,” as Wittgenstein puts it in a manuscript dated January 20, 1948, “is training. (Educators should keep this in mind)” (MS 136, 135b) [my translation: “Die Grundlage jeder Erklärung ist Abrichtung. (Das sollten Erzieher bedenken.)”]. 9. Pirmin Stekeler Weithofer (forthcoming) distinguishes between craving [begehren] and wishing, where the former is an undifferentiated longing for something that is satisfied when it comes to an end. My craving for food, for example, can be satisfied by a nice meal, a punch on my stomach, or a sudden event that completely absorbs my attention. Wishing, on the other hand, must involve a propositional content. Consequently, there are strict criteria that determine when my wish to eat is satisfied: the consumption of a nice meal does count as a satisfaction of my wish, but not the other scenarios considered. Wishing, but not craving, is accompanied with a clear intention that determines these criteria. Animals and newly-born children can crave, but not wish in this sense.

211

can be considered a part of rule-conforming behavior that constitutes the social practices of our form of life. This very example might be thought to show that the distinction between pattern-governed and rule-conforming behavior is less strict than I have suggested. After all, one might argue, also the behavior of the newly-born can be described as following rules. The child can ensure her survival only by signaling hunger with crying. The newly-born, thus, follows a rule that we might express roughly along the following lines: “If you feel the need to be fed soon, you should cry loud.” This rule even allows for error: we can imagine children who fail to signal their hunger in this way, with the fatal consequence of drastically diminishing their survival value. This suggests that also the instinct-guided behavior of the newly-born contains a normative element, it is following norms with which we are well accustomed from evolutionary biology. It is important to note, however, that the norms we know from the sciences differ in an important respect from the norms that constitute our social practices. In both cases, containing a normative aspect means that there are independent standards, according to which a theory, an event, or a certain kind of behavior can be evaluated. It is crucial to note, however, that these standards play a very different role in scientific theories than in our social practices. We can see this difference clearly by a “direction-offit” consideration: theories of physics are committed to the ideal of truth; if they fail to satisfy this ideal — if the world, in other words, does not behave in the way predicted by the theory — we have to change the theory, not the world. Our social practices, on the other hand, are committed to conformity with the rest of the social group. If an individual fails to satisfy these norms, she has to change her behavior, not the rule she is following.10 Moreover, changing one’s behavior presupposes that one is in a position to reflect on one’s behavior and consider alternatives. The standards of evolutionary biology play a mongrel role: insofar as the 10. For this reason we might say that scientific theories are normative at a descriptive level — they follow epistemic norms like truth — while our social norms contain an intrinsically normative element. I discuss this difference in more detail in my “Intrinsic Normativity” (forthcoming). I should add that this short discussion works with an idealized picture of science which, however, should suffice for my concerns. When the world does not behave in the way predicted by the theory, we have not only the option to revise the theory, but also to reject the observation statement, as Neurath (1983) and Quine (1980) have pointed out. But even in this case, we alter or reject the linguistic entity, not the state of affairs it describes. Similarly, when we point out to someone that she violates a norm, she will typically correct her behavior. She can also react, however, by challenging the norm.

212

theory speaks about species that are not able to reflect their behavior, they are normative at the descriptive level. If the members of a species behave in a way that, according to our theory, drastically diminishes their survival value but survive nonetheless without there being an obvious explanation for the exception of the rule, we need to seriously consider reformulating the theory. If, on the other hand, the survival value of our own species is concerned, we can consider changing our behavior — we can consider to stop building atomic power plants or avoid burning carbonic fuel in order to increase the survival value of our species, for example. When doing so, we turn biological standards (survival value) into moral standards (ecologically correct behavior). This transformation requires rational insight and can be justified, which undermines the reductionist project to reduce the norms that govern our social practices to the norms that govern evolutionary biology.11 This short discussion shows that while the instinct-guided behavior of newly-born children might increase their survival value, it nonetheless cannot be described as intrinsically normative. 2. From Pattern-Governed to Rule-Conforming Behavior: Concepts and Capacities Wittgenstein’s notion of training can shed a light on how children are conditioned to behave according to the rules imposed by the adult members of the community. This does not yet explain, however, how children can enter the realm of the normative, how, in other words, they come to apply concepts and to follow norms. In order to understand the transition from pattern-governed to rule-conforming behavior along Wittgensteinian lines, we have to pay attention to the close relation between concepts and capacities that Wittgenstein elaborates in his late philosophy. In his last text, On Certainty, he remarks that learning a language does not require a child to have a certain amount of knowledge, but rather to have a certain range of capacities: 11. Arguments along these lines have been proposed by advocates of evolutionary epistemology who aim to reduce epistemic norms to the norms of evolutionary biology. It is by no means obvious, however, that holding true beliefs can increase one’s survival value, as Quine has suggested when he famously stated that “creatures inveterately wrong in their inductions have a pathetic but praiseworthy tendency to die before reproducing their kind” (Quine 1969, 126). Ruth Millikan (1993, 90) quotes examples of species — she writes about beavers who often wrongly believe to be in danger — that increase their survival value by holding predominantly false beliefs.

213

But is it wrong to say: “A child that has mastered a language-game must know certain things”? If instead of that one said “must be able to do certain things”, that would be a pleonasm, yet this is just what I want to counter the first sentence with. (OC § 534)

What kind of capacities does a child have to have in order to master a language game? It seems obvious that there are different kinds of capacities involved. First, mastering a language requires capacities children have by their very nature, innate capacities, as it were. If children were not able to react to stimuli that are similar (in some relevant sense) in similar ways or to imitate the behavior of the adults we could not train them in the first place.12 Wittgenstein hints at this point when he notes: Think about the gestures and moves one makes to bring a dog to retrieve. But not every animal will react to these gestures like the dog. A cat will not or misunderstand these gestures; that simply means in this case: it will not retrieve. And if the child does not react to our encouragements, like a cat one wants to teach to retrieve, it will not come to an understanding of an explanation; or rather, the understanding begins here with reacting in a certain way.13 12. It is a question of empirical research to determine the exact nature of these abilities. In a recent study, the cultural biologist Michael Tomasello argues that language acquisition requires two sets of skills: (a) various “intention-reading” skills such as the abilities to “share attention with other persons to objects and events of mutual interest […] to follow the attention and gesturing of other persons to distal objects and events outside the immediate interaction […] to actively direct the attention of others to distal objects […] to culturally (imitatively) learn the intentional actions of others,” and (b) “pattern-finding” skills that include the abilities “to form perceptual and conceptual categories of ‘similar’ objects and events […] to form sensorymotor schemas from recurrent patterns of perception and action […] to perform statistically based distributional analysis on various kinds of perceptual and behavioral consequences […] to create analogies across two or more complex wholes” (Tomasello 2003, 4f ). For a discussion of Tomasello’s views in connection with Wittgenstein’s account of aspect-seeing, cf. Eldridge (forthcoming). 13. Ludwig Wittgenstein, Eine philosophische Betrachtung (Das Braune Buch). Werkausgabe Bd. 5. Frankfurt: Suhrkamp, 1989, 131; my translation: “Denke an die verschiedenen Gebärden und Bewegungen, die man macht, um einen Hund zum Apportieren zu bringen. Aber nicht jedes Tier wird auf diese Gebärden reagieren, wie der Hund. Eine Katze wird diese Gebärden nicht, oder mißverstehen; das heißt in diesem Fall einfach: sie wird nicht apportieren. Und wenn das Kind auf unsere Ermunterungen nicht reagiert, wie eine Katze, die man das Apportieren lehren möchte, so gelangt es nicht zum Verständnis einer Erklärung; oder vielmehr, das Verstehen beginnt hier mit dem Reagieren in bestimmter Weise.” Wittgenstein’s German text is an elaboration of the text he dictated to his students in English. We find a parallel, but shorter version of this remark in the English version of the text (BB 90): “Imagine the gestures, sounds, etc., of encouragement you use when you teach a dog to retrieve. Imagine on the other hand, that you tried to teach a cat to retrieve. As the cat will not respond to your encouragement,

214

This clearly shows that training presupposes capacities the trainee has by her very nature; they are, as we could say, innate, i.e., the trainee has them due to her biological constitution. In addition to these innate capacities (without which she could not be trained in the first place), there are abilities a child learns from the members of the social community in which she grows up by imitating the behavior the adults perform in the world — and with the world: a great part of these practices are actions that are performed with and on objects in the environment of the child. By imitating this behavior the child learns to act, and, as Wittgenstein points out, “it is our acting, which lies at the bottom of the language-game” (OC § 204). In his Remarks on the Foundations of Mathematics Wittgenstein draws an analogy between “concept” and “grasping” when he notes: “And concepts help us to grasp things. They correspond to a particular way of dealing with situations” (RFM VII § 67).14 Children acquire a language by learning to move in the world and to manipulate some of the objects in this world: they so learn to recognize certain structures in the world. Wittgenstein gives an illuminating example of this process when he notes: “Children do not learn that books exist, that armchairs exist, etc. etc., — they learn to fetch books, sit in armchairs, etc. etc.” (OC § 476). Thus, before acquiring words like “chair” or “glass,” the child learns to sit on chairs and drink from glasses. Based on these capacities the child acquires concepts like chair and glass, etc., which, in turn, is a prerequisite for acquiring the respective words. With this close connection between concepts and capacities Wittgenstein sketches a picture according to which there is a direct link between world and language. We can acquire concepts only by manipulating objects in the world around us.15 Moreover, if the world were different, or if it would radically change, we would not — or even: could not — use the same concepts (any longer) to describe it. Certain events would put me into a position in which I could not go on with the old language-game any further. In which I was torn away from the most of the acts of encouragement which you performed when you trained the dog are here out of question.” 14. In the German formulation of this remark the analogy is even stronger, for Wittgenstein draws on the common etymological roots of the expressions Begriff (concept) and begreifen (grasping): “Und Begriffe dienen zum Begreifen. Sie entsprechen einer bestimmten Behandlung der Sachlagen” (RFM VII § 67). I have altered Anscombe’s translation, who translates “begreifen” with “comprehend.” 15. As a consequence, the sceptic who doubts that physical objects exist, cannot possibly be right: if he was, he could not even come so far as to formulate his doubts, for he could not have acquired language in the first place.

215

sureness of the game. Indeed, doesn’t it seem obvious that the possibility of a language-game is conditioned by certain facts? (OC § 617)

Wittgenstein, thus, resists the idea that there is a gap between language on the one hand and the world on the other, a gap that is to be bridged in some mysterious way by meaning or intentionality. He rather insists that language is part of a form of life that is constituted by practices that involve objects and facts in the world. In a way we can say that the world (or at least relevant facts in the world) is part of our language. About the fact that water boils and does not freeze under such-and-such circumstances, for example, he explicitly states “This fact is fused into the foundations of our language-game” (OC § 558). Wittgenstein does not, however, go so far as to equate concepts with capacities. This would be the kind of simplification that Wittgenstein warned about over and over again. His later philosophy is characterized by a method of grammatical investigations that allows him to appreciate the variety of phenomena and resist the reductionist aspect of theories.16 With respect to concepts he points in this direction when he states that “‘Concept’ is a vague concept” (RFM VII § 70). When children are first trained to engage in certain practices, they do not yet have concepts. In a remark from 1944 Wittgenstein says, referring to the language game introduced in Philosophical Investigations § 2: It is not in every language-game that there occurs something that one would call a concept. Concept is something like a picture with which one compares objects. Are there concepts in language game (2)? Still it would be easy to add to it in such a way that “slab”, “block” etc. became concepts. […] There is of course no sharp dividing line between language-games which work with concepts and others. What is important is that the word “concept” refers to one kind of expedient in the mechanism of language-games. (RFM VII § 71)

Language game (2) involves only two players and four words: “block,” “pillar,” “slab,” and “beam.” When player A calls out one of these words her assistant B brings the object she has learnt to bring when the respective word is uttered. We can easily imagine that children are trained to this 16. I do not want to go so far as to say that Wittgenstein was opposed to any kind of theory. He clearly warns of the dangers inherent to theories, however, and sets apart his own approach when he states, for example, “we may not advance any kind of theory. There must not be anything hypothetical in our considerations. We must do away with all explanation, and description alone must take its place” (PI § 109).

216

sort of behavior.17 When this is the case, they do not yet speak a language or have concepts. Since only these four words can be used and only commands can be given (Wittgenstein does not contemplate the possibility that B utters “slab?” to question the command) the players have no means to reflect the working of their language game. Only if the language game has been extended and thus reached a level of complexity that allows us to reflect the usage of a certain utterance and, thus, point out errors, it can contain concepts — as it is the case when we, i.e. adult speakers of English, speak about language game (2). Wittgenstein does not explain how and by whom the language game can be extended. Are the players themselves in a position to carry out this extension, or is this done by more sophisticated speakers who already possess a richer language? To answer this question, we should consider that Wittgenstein does not discuss the phylogenetic question of the origins of language. He always focuses on the training of a particular child, performed by adults who already do engage in social practices and speak a language. In addition, he says that the “simple language-games” introduced in the first paragraphs of the Philosophical Investigations are “set up as objects of comparison which are meant to throw light on the facts of our language by way not only of similarities, but also of dissimilarities” (PI §130). This clearly hints that for Wittgenstein this extension is not performed by players A and B, nor by the trained children, but rather by their trainers, adult speakers, who are already initiated in the realm of the normative and already speak a language. Moreover, in his remarks against the sceptic idealist, Wittgenstein argues that also intrinsically normative epistemic notions like truth, knowledge, or doubt do not evolve from the realm of the causal, but are already anchored in a social practice which children are trained to play. When they first learn to occupy certain positions in the game, they cannot yet make moves that go beyond the positions they have been trained to occupy. Children cannot, for example, doubt the existence of objects that are relevant to this language game. “The child learns by believing the adult. Doubt comes after belief ” (OC § 160). To doubt means to occupy a position in the language game that the child has not yet acquired. In language game (2), 17. Wittgenstein explicitly draws this connection in his Brown Book, where he already introduces the language game (2) and then notes: “The child learns this language from grownups by being trained to its use” (BB 77). This passage is followed by Wittgenstein’s clarification that training is here understood as an activity normally performed on animals, as quoted above.

217

for example, player A can only give commands, but cannot formulate doubts, questions, make ironic comments, etc. We teach a child “that is your hand”, not “that is perhaps (or “probably”) your hand”. That is how a child learns the innumerable language-games that are concerned with his hand. An investigation or question, ‘whether this is really a hand’ never occurs to him. Nor, on the other hand, does he learn that he knows that this is a hand. (OC § 374)

And a few paragraphs later he states: “In the language-game (2), can he say that he knows that those are building stones? — ‘No, but he does know it’” (OC § 396). Only when children acquire language games that allow them to speak about the truth value of propositions or to say that they know that this-and-this is so-and-so do they acquire the capacity to make more complex moves in the logical space of reasons. Adult speakers, on the other hand, can attribute knowledge to the child when they think it appropriate. Words like “knowing,” “truth,” or “agreement with reality” have their meaning only within and relative to language games of which they are part. “Here we see that the idea of ‘agreement with reality’ does not have any clear application” (OC § 215). Children acquire concepts like knowledge, truth, or doubt in the same way as they acquire the rest of language: by being trained to partake in complex social practices. Only when this level of complexity is reached we are in a position to reflect and, if we think it necessary, to modify our language game. It is an important aspect of Wittgenstein’s anti-sceptical argument that we cannot do this for the whole of language at once, replacing, as it were, one language with another, for if we were to try this we would lose the tools necessary to make these revisions. “If I want the door to turn, the hinges must stay put” (OC § 343). In sum, we can state that capacities play a crucial role in Wittgenstein’s account of language acquisition: children need to have certain (innate) capacities to qualify as a trainee. They then start to imitate certain forms of behavior from the adults, a process by which they come to structure the world and apply concepts. With continuous training the adults enrich the children’s capacities to react to an increasing number of situations with an increasing vocabulary and to make moves in the language game, until they have reached the minimal level of complexity needed to call them a speaker of the language — “Light dawns gradually over the whole” (OC § 141).

218

3. Causes and Norms: Wittgenstein’s “Naturalism” Wittgenstein’s remarks on training show how an individual child can be acculturated in a society of adults who already engage in complex social practices guided by rules. Wittgenstein hardly discusses, however, how such a society is possible; how, that is, the first human beings came to form a social group and to engage in social practices. He does, in other words, not discuss how the realm of norms could evolve from the realm of causes, but only, how we can initiate our children into the realm of norms. The close connection he conceives between concepts and capacities, however, suggests that these two realms are closely related. I will now turn to the question of how Wittgenstein conceives this relation; how he would, in other words, answer the question concerning the role of normativity in a world of causes that I have raised in the opening paragraphs of this paper. In his later philosophy, Wittgenstein develops a picture of language that has been characterized with the thesis of the autonomy of language. According to this thesis, language is autonomous; it is independent of (1) psychological episodes and mental representations, (2) the referential relation between words and the objects referred to, (3) the acquisition of language, and (4) pragmatic goals (the aim of language and epistemological arguments) (cf. Burri 1995, 208). With this picture, Wittgenstein resists the reductionist impulse to explain the rules that govern our language games on the basis of the causal regularities that hold in our physical world. A language is for Wittgenstein […] independent both of the psychological inner world and the material outer world. Neither its ontogenesis nor its integration into a goal oriented human praxis have an influence on the shape of its constituting rules. (Burri 1995, 208)18

Wittgenstein’s main goal is to argue against a picture of language that regards truth and reference as the basic notions and “against the idea that they [the rules of grammar] have to mirror a putative essence of reality” (Glock 1996, 50). The discussion of the close relation between concepts and capacities shows, however, that language is at least to some degree determined by our form of life, i.e. by the social practices we engage in, and by the world in which we perform these practices. In On Certainty, 18. My translation: “Eine Sprache ist für Wittgenstein […] sowohl von der psychischen Innenwelt als auch von der materiellen Außenwelt unabhängig. Weder ihre Ontogenese noch ihre Eingliederung in eine zweckorientierte menschliche Praxis haben einen Einfluß auf die Beschaffenheit der sie konstituierenden Regeln.”

219

Wittgenstein draws an analogy between the rules that govern our practice and the primitive, instinct-guided behavior an animal can come to play: I want to regard man here as an animal; as a primitive being to which one grants instinct but not ratiocination. As a creature in a primitive state. Any logic good enough for a primitive means of communication needs no apology from us. Language did not emerge from some kind of ratiocination. (OC § 475)

The primitive being (or, for that reason, the newly-born baby) only acts by instinct; it does not act for reasons. In general we can say that the first social practices human beings come to engage in are not the result of reasoning, they are merely caused: “The game proves its worth. That may be the cause of its being played, but it is not the reason” (OC § 474).19 These remarks clearly show that human beings have developed language through practices that were not justified; they did not (yet) have a normative aspect. The world imposes constraints that determine which practices we can and cannot engage in. The child can be trained to take part in a practice only if this practice proves successful in this world. It is possible that people of other cultures develop practices quite different from ours. All of these practices, however, have to fit this world, as it were. The world does not force us into a specific kind of practice, it excludes certain practices from the realm of the possible, though. In On Certainty, Wittgenstein states: “By this I naturally do not want to say that men should behave like this, but only that they do behave like this” (OC § 284). Later in the text, he states again that there are no reasons for the concrete form of our language game: “You must bear in mind that the language-game is so to say something unpredictable. I mean: it is not justified. […] It is there — like our life” (OC § 559).20 The thesis of the autonomy of language must not be misunderstood as a complete detachment of language from the physical world and from our form of life. Our social practices, including our language games, are, as I have tried to show in the preceding section, deeply rooted in the physical world. Moreover, our rule-governed behavior is based on instinct-guided behavior we share with animals. In this way, the world shapes our social 19. I have slightly altered Paul’s and Anscombe’s translation, replacing the last word of this sentence, “ground” (in German: “Grund”) with “reason.” Wittgenstein does not seem to be interested in questions concerning foundationalism, but, more general, in questions concerning justification. 20. Again, I have slightly altered the translation, replacing “it is not based on grounds” with “it is not justified.” The original German text has: “Es ist nicht begründet” (cf. OC § 559).

220

practices (including our language) by putting constraints on what practices are possible; it does not justify them, though. If we would try to explain why our ancestors adopted one set of rules rather than another when they first developed language, we can quote only causes, for there are no reasons (yet) that regulate this process. If, on the other hand, we try to account for the sophisticated practice of language we now find in our community, we realize that we cannot derive its rules from the causal regularities that govern the physical world — the rules of (our) language are autonomous, they belong to the realm of norms, not to that of causes. Wittgenstein’s picture is based on the assumption that we can clearly distinguish causes and norms. He does resist, however, the urge to draw an insuperable demarcation line between the two, distinguishing, as it were, two separate realms of reality. Rather, he insists that both, or better, a combination of both are part of our nature. “Commanding, questioning, recounting, chatting, are as much part of our natural history as walking, eating, drinking, playing” (PI § 25). In a way we could therefore call Wittgenstein’s position a form of naturalism. It is crucial to see, however, that this form of naturalism is radically different from reductionist naturalism that is widespread in contemporary philosophy of mind.21 Wittgenstein does not try to reduce the normative to the causal, nor does he think that a scientific theory could justify the rules that govern our practice and our language game. If the formation of concepts can be explained by facts of nature, should we not be interested, not in grammar, but rather in that in nature which is the basis of grammar? — Our interest certainly includes the correspondence between concepts and very general facts of nature. (Such facts as mostly do not strike us because of their generality.) But our interest does not fall back upon these possible causes of the formation of concepts; we are not doing natural science; nor yet natural history — since we can also invent fictitious natural history for our purposes. (PI II 230)

This remark shows that although Wittgenstein conceives causes and norms as two aspects of one and the same reality, he is not interested in the question of whether or how theories that describe this reality at different levels, say physics and psychology, can be unified into a universal theory of everything, a theory, that is, that develops one complete and coherent picture of the world. 21. For a discussion of the differences between Quine’s naturalism and Wittgenstein’s position with regard to concepts, cf. e.g. Raatzsch 1995. For a comparison of Davidson’s and Wittgenstein’s conception of causes and norms, cf. Schroeder 2001.

221

Moreover, in his late philosophy he sheds an interesting light on the question of what place norms have in a world of causes. The urge to answer this question arises when we draw a distinction between a scientific description of the world and one that can account for the normative aspects that govern our mental and social life — a distinction, to put it in Sellarsian terms, between the scientific and the manifest image. Wittgenstein warns against the abstractions of the scientific image, insisting, as he does, that we find ourselves in a world and in a community that are regulated by reasons and causes. The scientific image abstracts from the level of reasons in order to develop a comprehensive picture of the causal regularities in this world, without which the sciences could not have made such impressive progress over the last centuries. Notwithstanding these achievements, however, we must not fail to see that the scientific image could be developed only on the basis of the manifest image, from which it abstracts. According to Sellars, the manifest image is the “framework in terms of which man came to be aware of himself as man-in-the-world” (Sellars 1963, 6). It is the result of a “refinement or sophistication of what might be called the ‘original’ image” (Sellars 1963, 7). The scientific image, on the other hand, is derived from scientific theories that postulate “imperceptible objects and events for the purpose of explaining correlations among perceptibles” (Sellars 1963, 19). Sellars clearly suggests that though the manifest image comes first in time, preference should be given to the scientific image. He does not think that we should completely eliminate the manifest image — he explicitly argues for a synoptic vision that should incorporate elements of both. Nonetheless, he develops an outlook according to which irreducible elements of the manifest image, “the conceptual framework of persons,” should be joined to the scientific image, thus granting dominance to the latter.22 The predominant positions in contemporary philosophy of mind take the Sellarsian distinction between manifest and scientific image as their starting point (even though most philosophers prefer not to use Sellars’ terminology) and share his view that we should give priority to the latter. While some reductionists would go along with Sellars and opt for a synoptic view, mostly in the hope to explain the level of norms with the 22. Interestingly enough, Sellars admits that we are not (yet) in a position to understand how such a unification can be achieved: “We can, of course, as matters now stand, realize this direct incorporation of the scientific image into our way of life only in imagination. But to do so is, if only in imagination, to transcend the dualism of the manifest and the scientific images of man-in-the-world” (Sellars 1963, 40).

222

means provided by the language of science, eliminativists, convinced by the impossibility of this strategy, opt for eliminating the manifest image altogether. The main motivation for this strategy lies in the undeniable advantages of the abstraction from the framework of persons and the level of norms in the development of the sciences. The problem arises, in my view, only when what has started as a working assumption for scientific theories turns into a full fledged picture of the world. What works for science might not work for gaining an understanding of persons and their position in this world. The scientific image cuts ourselves out of the picture, thus overlooking that it is us who develop this picture in the first place. When it comes to understanding persons, we have to take a step back and take all relevant phenomena into account. Wittgenstein’s method to examine what lies there open before our eyes in detailed grammatical investigations can offer an interesting alternative to this scientistic view. Rather than focusing on the question of whether and how the framework of persons can be reconciled with the scientific image, it requires us to acknowledge that we find ourselves in a world of things and persons; a world that contains causal and rational relations. “Our mistake is to look for an explanation where we ought to look at what happens as a ‘proto-phenomenon’. That is, where we ought to say: this language-game is played” (PI § 654).23 The urge to do philosophy (and, in general, science) stems from our finding ourselves in this world and trying to gain orientation and to understand who we are. “A philosophical problem has the form: ‘I do not know my way about’” (PI § 123). The abstractions of science prove fruitful when we focus on specific, clearly defined phenomena. When it comes to develop a more comprehensive picture, though, this abstraction can prove too restrictive. Our (and Wittgenstein’s) restlessness stems from the temptation to universalize scientific theories and approaches that have proven locally successful into a universal theory of everything — a temptation that makes us overlook the limitations of the respective theories. The fact that Wittgenstein does not provide an answer to the question of how we can reconcile the scientific with the manifest image should not make us believe that his remarks are irrelevant to this topic. His aim is to make us shift our perspective. He reminds us that in order to achieve a deeper understanding of ourselves we must first acknowledge that we live 23. I have slightly changed Anscombe’s translation, who has the last sentence in the past tense (“where we ought to have said”); the original German sentence is formulated in the present tense.

223

in a world and in a community that are regulated by causes and norms. This world lies open before our eyes, and the task is not to grasp it with one abstract theory, but rather to describe it in detailed analyses and, while doing so, continually to reflect this very process. In this way we can hope to get a fuller grasp of what it means to speak a language, to be part of a community and engage in rule guided practices, or to have conceptual consciousnesses, in short, what it means to be human.24 REFERENCES Burri, Alex 1995: “Kritische Bemerkungen zur These der Sprachautonomie in Wittgensteins Grammatik”, in: Logos N.F. 2, 207–35. Eldridge, Richard (forthcoming): “Wittgenstein on Aspect-seeing, the Nature of Discursive Consciousness, and the Experience of Agency,” in: William Day, Victor Krebs (eds.): Seeing Wittgenstein Anew: New Essays on Aspect-Seeing, Cambridge: Cambridge University Press. Glock, Hans-Johann 1996: A Wittgenstein Dictionary, Oxford: Blackwell. Huemer, Wolfgang 2005: The Constitution of Consciousness. A Study in Analytic Phenomenology, New York: Routledge. –– (forthcoming): “Intrinsic Normativity”. Millikan, Ruth 1993: White Queen Psychology and Other Essays for Alice, Cambridge: M.I.T. Press. Neurath, Otto 1983: “Protocol Statements,” in: Robert Cohen, Marie Neurath (eds. and trans.), Philosophical Papers 1913–1946, Dordrecht: Reidel, 91–9. Quine, Willard Van Ornam 1969: “Natural Kinds,” in: Ontological Relativity and Other Essays, New York: Columbia University Press, 114–38. –– 1980: “Two Dogmas of Empiricism,” in: From a Logical Point of View, 2nd, rev. ed. Cambridge: Harvard University Press, 20–46. Raatzsch, Richard 1995: “Begriffsbildung und Naturtatsachen,” in: Eike von Savigny, Oliver Scholz (eds.): Wittgenstein über die Seele, Frankfurt: Suhrkamp, 268–80. Rhees, Rush 1989: “Vorbemerkungen des Herausgebers,” in: Ludwig Wittgenstein: Das Blaue Buch. Werkausgabe Bd. 5. Frankfurt: Suhrkamp, 9–14. Schroeder, Severin 2001: “Are Reasons Causes? A Wittgensteinian Response to Davidson,” in: Severin Schroeder (ed.): Wittgenstein and Contemporary Philosophy of Mind, Basingstoke: Palgrave, 150–70. 24. I want to thank Alex Burri, Richard Eldridge, Wolfgang Kienzler, Richard Raatzsch, and Joachim Schulte for helpful comments on earlier drafts of this paper.

224

Sellars, Wilfrid 1963: “Philosophy and the Scientific Image of Man,” in: Science, Perception and Reality, Atascadero: Ridgeview, 1–40. Sellars, Wilfrid 1997: Empiricism and the Philosophy of Mind, Cambridge: Harvard University Press. Stekeler-Weithofer, Pirmin (forthcoming): “Kategoriale Analyse von Erkenntnis und Selbsterkenntnis”. Tomasello, Michael 2003: Constructing a Language. A Usage-Based Theory of Language Acquisition, Cambridge: Harvard University Press. Williams, Meredith 1999: “The Philosophical Significance of Learning in the Later Wittgenstein”, in: Wittgenstein, Mind, and Meaning. Towards a Social Conception of Mind, London: Routledge, 188–215.

225

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), 227–249.

GRAMMATICAL PROPOSITIONS Barbara SCHMITZ University of Basel Summary First of all, this paper aims at a clarification of Wittgenstein’s conception of grammatical propositions. Their essential characteristics will be developed and some of the central questions concerning their status will be discussed: Should grammatical propositions be seen as arbitrary conventions? How do they work in practices? And how do they relate to natural facts? Later on, the two propositions “Every rod has a length” and “Sensations are private” will be discussed in more detail, for both fulfil three important features of grammatical propositions: They are not empirical descriptions, they are rules of a practice, and they are not knowledge-claims but express insights. Focussing on the grammar of pain language, it will be shown what kind of interdependency exists between grammar and natural facts. It will also turn out that some grammatical propositions are closely connected to our understanding of living a human form of life.

Wittgenstein himself calls the following propositions (or sentences) either “grammatical propositions” or “grammatical considerations”, or he cites them in a context that makes that label plausible: “Every rod has a length” (PI § 251), “My images are private” (PI § 251), “An order orders its own execution” (PI § 458), “This body has extension” (PI § 252), “Nothing can be red and green all over” (PR 51, 52), “Every event must have a cause” (AWL 16), “There can be no certain knowledge of the future” (LW I § 188), “Only you can know if you had that intention” (PI § 247); “Green is a colour” (PR 118); “One plays patience by oneself ” (PI § 248), “Sensations are private” (PI § 248), and “4 metres is a length” (PG 129). Grammatical propositions play a central role in Wittgenstein’s later philosophy, because showing that certain propositions have to be understood as grammatical is one of his aims. Grammatical propositions have a certain status, they play a certain role not played by non-grammatical propositions. According to Wittgenstein’s later philosophy, failing to recognise a proposition as grammatical may lead one to misunderstand the propositions as empiri-

cal, and that is dangerous: Many philosophical errors and wrongheaded metaphysical theories result from such misunderstandings. Wittgenstein’s thesis that philosophy must be a grammatical investigation that “sheds light on our problem by clearing misunderstandings away” (PI § 90) is directly linked to the project of showing that certain propositions have to be understood as grammatical. My aim is to clarify Wittgenstein’s conception of grammatical propositions. I will describe their essential characteristics and discuss some of the central questions connected to their status: Should they be seen as arbitrary conventions? How do they work in practices? And how do they relate to natural facts? Because the answers to these questions cannot always be the same for all of the propositions in the list above, I will mainly work with two singular propositions: “Every rod has a length” and “Sensations are private”. In the second part of my paper I will focus on the latter in order to show that there are grammatical propositions that are closely connected to our understanding of a human form of life: thus we see that the grammar of our concepts and facts of nature are interdependent. My paper consists of three parts. To begin with, I will depict a number of general characteristics of grammatical propositions, employing the proposition “Every rod has a length” (part 1). I will focus on three specific issues: The difference between grammatical and empirical propositions (1.1), grammatical propositions as a description of a practice (1.2), and grammatical propositions as insights (1.3). Then, I will consider the proposition “Sensations are private” (part 2). I will first show that that proposition fulfils the three mentioned characteristics (2.1), and then I will go on narrowing the focus on the example of pain by showing how the grammar of our pain language is connected with instinctive preverbal behaviour (2.2). Finally, I will consider the interrelation between grammar and natural facts (part 3) by discussing, firstly, what kind of dependency exists between grammar and natural facts, secondly, how a totally different pain language can be imagined, and thirdly, how propositions that describe natural facts should be seen. 1. General characteristics of grammatical propositions In the course of his philosophical development, Wittgenstein gradually arrives at his later conception of grammatical propositions through a number of steps. The three above-mentioned aspects of grammatical

228

propositions (that will be discussed more extensively below) form three chronological stages of that development. The first step consists in differentiating between grammatical and empirical propositions. Wittgenstein takes that first step as early as in the beginning of the 1920s, building on the conception of the Tractatus (1.1). During the 1930s, Wittgenstein’s investigations into the concept of a rule lead to a new understanding of what rules are and thus to an understanding of grammatical propositions as descriptions of practices. The results of this development can be seen in the Philosophical Investigations (1.2). The last step of treating grammatical propositions as insights is taken in the 1940s in the Philosophical Investigations and, most importantly, in On Certainty (1.3). 1.1 Grammatical vs. Empirical Propositions Wittgenstein uses the expression “grammatical proposition“ from the beginning of the 1930s. That expression is coined in the course of his development and critique of his early philosophy. From that time on, Wittgenstein takes the objective of philosophy to be mainly a description of “grammar”. In many cases, the notion of “grammar” replaces the notion of “logic” used in his early philosophy. It might appear plausible, therefore, that Wittgenstein’s use of “grammatical propositions” is intimately linked to the “logical propositions” of his earlier philosophy. However, his manuscripts from the 1930s show that this is not straightforwardly so. At first, Wittgenstein uses the expression “grammatical proposition” to characterise a certain type of senseless propositions. In the Philosophical Grammar, Wittgenstein comments on the proposition “Every rod has a length” — a proposition that is used later, in the Philosophical Investigations, as a prime example of a grammatical proposition. In the Philosophical Grammar, however, he writes: But why does one say “I can’t imagine how it could be otherwise” and not “I can’t imagine the thing itself ”? One regards the senseless sentence (e.g. “this rod has a length”) as a tautology as opposed to a contradiction. One says as it were: “Yes, it has a length: but how could it be otherwise; and why say so?” To the proposition “This rod has a length” we respond not “Nonsense!” but “Of course!” We might also put it thus: when we hear the two propositions, “This rod has a length” and its negation “This rod has no length”, we take sides and favour the first sentence, instead of declaring them both nonsense. But this partiality is based on a confusion: we regard the first proposition as verified (and the second as falsified) by the fact “that the rod has a length

229

of 4 metres”. “After all, 4 metres is a length” — but one forgets that this is a grammatical proposition. (PG 129)1

According to this remark, the sentence “Every rod has a length” is senseless. The criterion of senselessness consists in the fact that the sentence cannot be negated. This criterion is connected to one used in his early philosophy. In the Tractatus, Wittgenstein holds that a sentence has to be bipolar if it is to have sense, i.e., for a proposition to have sense it needs to be such that it can be both true or false. Only a proposition that has these two poles and is not either always true or always false can have sense (cf. TLP 6.111–6.126; NL 104). Now that condition is violated by the proposition “This rod has a length”, because it is impossible for this proposition to be false. Another symptom of this fact, according to Wittgenstein, is the fact that we cannot make sense of the negated sentence. The sentence “This rod has a length” can only have sense if the sentence “This rod does not have a length” also has sense, but that is not so. The criterion of bipolarity for meaningful propositions gives the first hint of the distinction between grammatical and empirical propositions. Empirical propositions can be both true and false (though not at the same time). One test for this is whether it is possible to imagine the sentence’s contrary. In the case of “This rod does not have a length”, we cannot imagine anything. We do not understand what such a proposition might mean. It is fundamentally without sense and has no place in our language. Another aspect of this insight is that a grammatical proposition cannot be confirmed or disconfirmed by experience. There can be no experience showing us that every rod has a length; neither can there be an experience of a rod without a length — we would not even know what such an experience would be like. Thus, the proposition “This rod has a length” cannot be checked against experience, and Wittgenstein argues in the Philosophical Grammar that the proposition can neither be verified nor falsified.2 It is, however, questionable whether we should follow Wittgenstein in declaring the proposition “This rod has a length” as the same kind of nonsense as its contrary, “This rod does not have a length”, as he argues in the Philosophical Grammar. It is true that neither of the propositions 1. The remark is originally from MS 110, written in 1931. 2. In the cited passage, Wittgenstein alludes to the criterion of verification as a test for a sentence’s sense. In the early 1930s, Wittgenstein for a time does subscribe to the idea that a sentence’s sense is the method of its verification (cf. MS 107, 143; MS 106, 16). In the present context, that idea can be seen as a development of the criterion of bipolarity: A sentence must be verifiable or falsifiable in order to have sense.

230

is bipolar, and neither can be verified nor falsified by experience. Thus both propositions fail the criterion of bipolarity for having sense. However, it appears reasonable to assume that the sentence “This rod has a length” is not nonsense, because it expresses something that concerns our conceptual scheme, or our grammar. It is no empirical proposition, but it expresses an aspect of the grammar of our words “rod” and “length”. Thus, the following remark from the Philosophical Investigations can be read as a development of Wittgenstein’s ideas from the considerations in the Philosophical Grammar: “‘This body has extension.’ To this we might reply: ‘Nonsense!’ — but are inclined to reply ‘Of course!’ — Why is this?” (PI § 252). To my mind, Wittgenstein’s question “Why is this?” indicates that he here considers the proposition “This body has extension“ to be a grammatical proposition, and in contrast to what is said in the Philosophical Grammar, that proposition and its negation are not of the same type of nonsense. Historically, the difference between grammatical propositions and empirical propositions is thus connected to the criterion of bipolarity. That criterion expresses the insight that empirical propositions describe reality and are verified or falsified by experience, whereas this is not so for grammatical propositions. This distinction between empirical and grammatical propositions is of high importance for Wittgenstein, because a failure to distinguish these types of propositions gives rise to wrongheaded metaphysical theories and philosopical errors. This is all the more important as there are propositions that can sometimes be used as empirical and sometimes as grammatical propositions. In The Voices of Wittgenstein, Wittgenstein points this out for the sentence “One cannot see into another’s heart”: It is said: “One can’t see into another’s heart”. As in many similar declarations, the sense of this statement hovers between an empirical proposition and a grammatical rule. In the first case the prosposition is used to describe the opposite of the case in which we can see into the heart of another. We say then: “I know that this man is not lying”, but of another I can’t yet say this: “I can see into this man’s heart, but not into that man’s”. (VWVC 309)

The proposition “One can’t see into another’s heart” is used empirically if it is meant to say that for this particular man we are not sure whether he is lying at the moment. It is empirical insofar as its contrary would be to say “I can look into his heart. I know he is not lying.” When the proposition is used grammatically, that second possibility is absent, because it is excluded on logical, i.e. grammatical grounds. The difference between the

231

two cases is grounded in the meaning of “You cannot”: This could express an empirical, but also a logical impossibility. Wittgenstein continues the quoted passage as follows: This case is similar to saying: “One can’t be certain that …”. One uses this expression in these ways: “You can’t be certain that he’s a honest man, you don’t know him well enough”, or “One can’t be certain that that’s a house, it’s already too dark”. In these cases the expression “one can” corresponds to a striving, hence not to what we call logical possibility. For the logical possibility that p is the case is expressed by its being the case that p has sense, i.e. that the logical grammar permits the construction of this sentence. Here the words “I can’t peer into his heart” have sense only where there is also the possibility that one can peer into his heart. (VWVC 309)

Grammatical propositions describe logical or grammatical possibilities and exclude other possibilities as nonsensical. The mistake of taking a grammatical use for an empirical use may lead one to look for empirical possibilities that have already been excluded on logical grounds. Thus, e.g., one might think that it might be possible to look into another person’s heart, that perhaps scientific and technical progress might enable us to gain access to the heart or the soul of another person. If one sees that the impossibility expressed in the proposition is grammatical, one is able to see that this attempt is bound to fail. To sum up: The distinction between grammatical and empirical propositions can be seen as a development of Wittgenstein’s criterion of bipolarity, which he used as a criterion for sentences to have sense. Grammatical propositions are distinguished from empirical propositions in that: Firstly, they cannot be both true and false, and secondly, they cannot be verified or falsified by experience. That latter aspect makes is plausible to understand grammatical propositions as rules. In the following section, I will deal with that understanding of grammatical propositions. 1.2 Grammatical Propositions as Descriptions of a Practice Grammatical propositions are rules, i.e. they are normative. They say what is taken to be true or false; they set standards. They form part of what Wittgenstein calls “grammar”: the whole of rules and uses of language. Grammar does not consist of grammatical propositions alone, but it also includes ostensive definitions, mathematical rules, and colour tables (cf. BB 4, 12, 90; PG 319, 347).

232

Many interpreters have characterised Wittgenstein’s use of grammatical propositions as giving rules for the uses of words. These rules declare how a word should be used, and according to these interpreters, the main function of grammatical propositions is to point out the part played by the word in language acquisition or in misunderstandings. The grammatical proposition “Every rod has a length” can in this sense be taken to explain the use of the word “rod” or “length”.3 It is true that grammatical propositions formulate rules for the use of words, and that they can be taken as explanations of the uses of these words. This conception of grammatical propositions is in accord with many ideas found in Wittgenstein’s later philosophy. I think, however, that this understanding of grammatical propositions is too simplistic, for it neglects the important point that grammatical propositions have their status only within a practice. Instead of characterising grammatical propositions as rules for the uses of words, I would suggest seeing them rather as descriptions of or commentaries on a practice. I will explain what I mean by this in the following, using a sketch of the development of Wittgenstein’s conception of rules as my background. Wittgenstein’s picture of grammatical rules at the beginning of the 1930s, most importantly in the Big Typescript (1932/33), was the following: Language is controlled by explicit grammatical rules. Language is a system, and the rules of that system are a calculus. The rules express the possible “moves” in the system. Speaking and understanding a language means being fluent in the use of these rules. Language is a system that can be described and in which derivations are made. Rules are “presuppositions of the sense of sentences ” (VWVC 268). In the course of the 1930s, this conception of language as a calculus is more and more replaced by the conception of language as a game, in which not all rules are clearly determined, in which many uses are not completely determined by the rules, and where a competent speaker does not have to have explicit knowledge of the rules (cf. PG 153, PI §§ 82, 83). Neither do the rules determine all of our language use, nor is it always clear what the rules would be in a particular case (cf. PI § 82). Yet, the notion of a rule is still fundamental for Wittgenstein’s understanding of grammar at that time, because rules express the normativity of language (cf. RFM VI 29). Still, the notions of “language game” and that of a “practice” become more and more central 3. This understanding of grammatical propositions is often linked to the idea that grammatical propositions are conventions — an idea which I will criticise in part 3 below.

233

in Wittgenstein’s explanations of language use. Only in a practice can rules be understood and employed. They do not form an independent calculus, but they work inside the practice. In many cases, it would therefore be wrong to say that the rules are constitutive of the practice: it is not the case that the practice is built upon the rules, but the rules are formulated later as descriptions or commentaries of the existing practice. This understanding of grammatical propositions as descriptions of a part of a practice can be illustrated for the sentence “Every rod has a length”. This sentence is a commentary on our practice of measuring length. It is no isolated rule that gives rise to a practice, but it expresses an important aspect of our existing practice of measuring length. In order for this sentence to play the role of a grammatical sentence and be understandable as such, the whole practice of determining length must already be presupposed.4 For the sentence can only be of some help to clarify the meaning of the word “length”, if the practice of measuring length is already given. The sentence cannot stand at the beginning of a practice. It is, therefore, not the case that one learns a language through learning grammatical propositions. One does not learn what a length is by learning grammatical propositions that use the word “length”, but by learning to measure — as Wittgenstein stressed: “What ‘determining the length’ means is not learned by learning what length and determining are; the meaning of the word ‘length’ is learnt by learning, among other things, what it is to determine length” (PI II 225). It is important that the sentence “Every rod has a length” expresses something every competent participant of the practice of determining length takes for granted. The proposition is accepted without further ado. This will be important for the distinction between knowledge and insight, which will be discussed in the next section. It also becomes clear that grammatical propositions often describe fundamentals of a practice, i.e. aspects of a practice that lie at its very heart and that could not be changed without questioning the whole practice as such. If it were not longer true that every rod has a length, our whole practice of determining length would be shattered. In summary, I want to stress that grammatical propositions are not just definitions or elucidations of linguistic expressions, but they must be understood to be descriptions of a part or an aspect of a practice. Under4. The expression “grammatical sentence” and “grammatical proposition” will be used synonymously.

234

standing them as rules for the uses of words misses a crucial point. Only through the practice in which they are embedded is their meaning fully understood. 1.2 Grammatical Propositions as Insights The third and last aspect of grammatical propositions that I want to stress is intimately connected to the preceding one. Grammatical propositions express what goes without saying in a practice. Therefore, they do not express knowledge, but formulate insights. The difference between knowledge and insight presupposed here is elaborated by Wittgenstein in the Philosophical Investigations and in On Certainty. Wittgenstein first criticises the quite plausible idea that we, being competent in a certain practice, know what can be expressed by grammatical propositions concerning that practice, so that uttering such a grammatical proposition amounts to making a knowledge-claim. At first it seems uncontroversial to say that I know that every rod has a length, that my sensations are private, that every effect has a cause. However, Wittgenstein’s arguments in the Philosophical Investigations and in On Certainty show that knowledge can only be claimed for empirical, but not for grammatical propositions. According to Wittgenstein, the sentence “I know that this rod has a length” is fundamentally different from the sentence “I know that this rod has a length of four metres”. The difference consists in the fact that the latter proposition is bipolar, that there are techniques through which I can substantiate my claim to knowledge. These techniques are grounded in the fact that the sentence can be verified. My knowledge is grounded, for example, in the fact that I have measured the rod. The claim that is made by means of the sentence can be verified publicly and intersubjectively, since there are commonly accepted methods in our practice of determining length by which we can check whether the sentence expresses indeed knowledge or whether I am wrong. These matters are completely different in the case of a grammatical proposition. The claim made here is not bipolar, and there are no techniques of checking the claim similar to those of the second sentence. However, according to Wittgenstein, justifying and the possibility of checking belong to our language games of making statements and thereby claiming knowledge (cf. Z § 549). Speaking of knowledge where these possibilities are absent makes no sense, according to Wittgenstein. Therefore, there are two important insights in our concept of knowledge: Firstly, it presupposes that there is the possibility

235

of error, and secondly, it is connected with methods of justifying the claims of knowledge. It is this understanding of knowledge that forms the background for the idea that grammatical propositions do not express knowledge but insights. Wittgenstein explicitly contrasts knowledge and insight in On Certainty: If “I know etc.” is conceived as a grammatical proposition, of course the “I” cannot be important. And it properly means “There is no such thing as doubt in this case” or “The expression ‘I do not know’ makes no sense in this case”. And of course it follows from this that “I know” makes no sense either. “I know” is here a logical insight. (OC §§ 58, 59)

In the quoted passage Wittgenstein makes it clear that it only makes sense to speak of knowledge if there is a possibility of ignorance or of being wrong. If we speak of knowledge in a different sense, e.g. in the case of grammatical propositions, then we are wrong on two counts: firstly, what is at issue is not that a certain person knows something, secondly, ignorance is ruled out in that case. Instead of saying “I know that every rod has a length”, one should say “There is no doubt that every rod has a length”. That in turn, however, means that the sentence “I do not know that every rod has a length” makes no sense — it makes no sense for anybody who is competently partaking in the practice of determining length, though the grammatical proposition goes without saying for everybody who is participating in the practice. In contrast to a knowledge claim, in the case of insights it is not presupposed that one might not know. We can also have insights in that which goes without saying. Thus, a grammatical description of a practice should not be expressed as knowledge but as a grammatical or logical insight. It is a way to reflect on a practice. The distinction between knowledge and insight can therefore be summarised as follows: Knowledge might be formulated, for instance, in empirical propositions. Grammatical propositions, however, express grammatical insights. A person can gain knowledge of an empirical fact, and this knowledge claim can be confirmed or disproved by other persons by using certain criteria to test it. An insight is not gained in the same way as a knowledge claim. We can have insights into certain features of our practices when we reflect on the practice. An insight cannot be confirmed or disproved using the methods by which an empirical proposition can be proved. It concerns what is self-evident of a practice, and very often it deals with the foundations of a practice. However, even though one cannot speak of making a knowledge claim with a grammatical proposition, one

236

can argue for the insights expressed by them. This is part of the task of Wittgenstein’s grammatical investigations and the work of philosophy. 2. “Sensations are private” as a Grammatical Proposition The sentence “Every rod has a length“ can be quite easily recognised as a grammatical proposition. However, there are other grammatical proposition for which it is much less clear that they have this status, e.g. the sentence “Sensations are private”. Prima facie this proposition seems not to have the same kind of self-evidence as “Every rod has a length”. It seems to be an empirical proposition that can be proved by our experience of having sensations. However, Wittgenstein shows in the Philosophical Investigations that this sentence does not express empirical knowledge, but a grammatical insight. In this part of my paper I want to argue that the sentence must indeed be understood as grammatical. I will do this by showing that the three characteristics which I discussed in part 1 belong to it (2.1). Then I will go on discussing what it means to understand the sentence as grammatical and whether there can be a justification for the particular status the sentence has (2.2). 2.1 The Three Characteristics of “Sensations are private” The first question is whether “Sensations are private” should be understood as an empirical or as a grammatical proposition. To begin with, let me note that the sentence could be understood like the sentence “One can’t see into another’s heart”, that I discussed above. In this case the proposition could have either an empirical or a grammatical meaning. Considered as being empirical, the proposition could just be a description of the fact that we sometimes do understand the sensations of another person, and sometimes we do not. However, this is not the meaning that is at issue when the proposition “Sensations are private” is discussed in the Philosophical Investigations, where Wittgenstein rather considers the question whether or not the general sentence “Sensations are private” expresses an empirical claim. What would it mean if the sentence “Sensations are private” were an empirical claim? This would mean that I had the experience that I have certain sensations that have the feature of being private. This feature is not necessarily connected with my sensations, but it is something which

237

I derive from my experiences. The privacy of sensations could mean that my sensations cannot be had by another human being, or that nobody else except me can know which sensations I have. In the first case the sensations would be ontologically private, in the second case they would be epistemologically private. Of course, the sentence can also capture both meanings of privacy. In all these three cases, the sentence “Sensations are private” would express an empirical proposition. The question whether or not the sentence “Sensations are private” expresses an empirical claim is investigated by Wittgenstein through the private language argument. In Philosophical Investigations §§ 258ff., he discusses the following situation: Imagine a person having a sensation that no other human being can have and that nobody else can know about. This ontological and epistomological privacy leads to the idea of a private language. This would be a language that contains expressions that name private sensations and that could not be understood by any other human being. As Wittgenstein shows in the private language argument, this thought experiment fails because there are no criteria of how to identify the respective private sensation as a sensation at all. For if sensations are disconnected from their public expression, which is normally a bodily or verbal expression, the possibility of having access to the sensation at all disappears. Private criteria which would make it possible to identify the sensation privately cannot be established because they would never enable us to distinguish between a correct and an incorrect identification. Thus, these criteria could not have the function of criteria at all (cf. PI §§ 253, 269). The private language argument is supposed to show that if sensations are understood as ontologically or epistemically private, the whole concept of a sensation becomes unclear. Thus, the proposition “Sensations are private” can not be an empirical claim. This conclusion is also reached if we consider the difference between empirical and grammatical propositions as it was stated in part 1 above. The proposition “Sensations are private” is not bipolar. It cannot be either true or false and its negation does not mean anything. What would a non-private sensation be? In which sense could a sensation be public? If this is taken literally, it does not make sense at all, because publicity is something we cannot ascribe to the sensation. We can only describe the bodily or verbal expressions that belong to it as public.5 5. It would be totally unclear what can be meant by saying that sensations are public events if this should mean something more than that there are public expressions of sensations which are directly connected to the sensation itself. However, even if there is a direct connection between public expression and “private” sensation, this does not make the sensation itself public.

238

Is it possible to compare the proposition with reality? No, as the private language argument shows, this is as impossible as it is for the proposition “Every rod has a length”. The next question concerns the status of “Sensations are private” as a rule or as a description of a practice. In the Philosophical Investigations, Wittgenstein compares this sentence explicitly with the rule of a game: “The proposition ‘Sensations are private’ is comparable to “One plays patience by oneself ” (PI § 248). This comparison between “Sensations are private “ and “One plays patience by oneself“ may indicate that Wittgenstein sees the first sentence being in the same way embedded in and constitutive of a language game of talking about sensations as the second sentence is embedded in and constitutive of the card game patience. Thus, in the same way as the second sentence comments on the practice of playing patience, the first sentence is supposed comment on the language game of sensations. However, I do not think that this a correct interpretation of the matter, and to some extent I am going to criticise Wittgenstein’s comparison of these two sentences because it is in many points misleading. It is true that both propositions are insights into our grammar, but their actual function, i.e. how they work in the practices they are a part of, is very different. The proposition “Patience is played by oneself ” is a rule that is constitutive of the game of patience. It describes a part of the practice of playing patience and it says that someone playing cards who does not play alone must be playing another game. This rule is established conventionally. How can this rule be changed? To begin with, it is of course possible that in the future we will call the game that is now called “bridge” “patience”, and vice versa. This kind of change is just a change in words. Secondly, we could also change the practice of playing patience, e.g. by introducing new rules. This change of rules could concern rules of minor importance like introducing a rule that patience should only be played with the left hand, or rules of major importance like changing the rule that patience is played by two players who each plays 15 minutes. We may still call this game “patience” and the practice would not be changed fundamentally, but the rule “One plays patience by oneself ” would no longer hold. The case is different, however, with a change of the grammatical proposition “Sensations are private”. That sentence is a description of a practice that we cannot just change like the rules of patience. It is connected to our human form of life. The rule at issue here is not determined conventionally, and it is not possible to change this rule without giving up certain important other claims. I will discuss this aspect of non-conventionality

239

of the proposition in the next section. Here it should just be noticed that the proposition “Sensations are private” does not work in the same way as a rule in a game often works. “Sensations are private” is a description of a practice that is not established conventionally. The third feature of grammatical propositions points to the difference between knowledge and insight. Is the proposition “Sensations are private” a knowledge claim or does it express a grammatical insight gained by reflection on a practice? We have to distinguish between two ways of understanding this question: Firstly, it may be the question whether it is meaningful to say “I know that sensations are private”, and this question prompts us to investigate the third feature of grammatical propositions. Secondly, however, the question could be understood in a completely different sense, as asking whether I have a privileged knowledge of my own sensations, i.e. if it is true that I know my own sensations. This second question would be the question whether sensations are epistemically private. To my mind, the first question is the more important one. It can be answered easily by pointing to the results of part one of my paper. It is not reasonable to speak of knowledge in a case where there is no possibility of going wrong. The sentence does not have the features that typify empirical claims such as being testable by comparison with reality. The sentence is an insight one gains when one is thinking about our way of talking about sensations. It is a reflection on a practice. The second question is discussed explicitly by Wittgenstein in the Philosophical Investigations. He points out that it is in one way nonsensical to say about myself that I know that I am in pain: “It can’t be said of me at all (except perhaps as a joke) that I know I am in pain. What it is supposed to mean — except perhaps that I am in pain?” (PI § 246). It is nonsensical to say that I know that I am in pain because the meaning of knowing which includes the possibility of error is lost when doing this. When one says that one is in pain, one does not utter a special priviliged knowledge about one’s sensations, one just expresses pain. This expression involves no knowledge claim because it is not possible for it to go wrong. The expression of pain is immune from error. Yet, this immunity from error is not founded epistemologically on a particular kind of knowledge, but it is part of our language game of talking about sensations. The idea that a specific kind of knowledge is involved in such utterances in the first person singular is a mistake that builds on a misunderstanding of the proposition “Sensations are private” as an empirical claim. To sum up: The investigation shows that the proposition “Sensations

240

are private” displays all three characteristics of a grammatical proposition. This leads to the question what this grammatical descriptions means and how it works in our practice of talking about sensations. 2.2 The First Person Authority of Pain Expressions and its “Foundation” One problem with the proposition “Sensations are private” is its generality. It is open to interpretation what falls under a sensation and what is meant by privacy. To gain a better understanding of the sentence, it is therefore necessary to look at a specific sensation, and I will concentrate on the sensation of pain, which is discussed by Wittgenstein at length, subsequently. As I have shown in the previous sections of this paper, the kind of privacy that is at issue in Wittgenstein’s investigations is neither ontological nor epistemic privacy. It is grammatical privacy that is at issue. What does this mean? The rule “Sensations are private” concerns our use of the concepts “sensation” and “private”, and it describes a practice of talking about sensations. The sentence expresses the insight of first person authority concerning matters of fact that belong to expressions of pain. The grammatical proposition “Sensations are private” says that a person expressing pain honestly and sincerely cannot be corrected by other speakers, since the proposition in the first person singular is immune from error. Thus, the rule “Sensations are private” is fundamental for our practice of talking about sensations. This understanding of the proposition “Sensations are private” leads to the question if there is any foundation or justification for it as a rule. Is it possible to explain why first person authority is valid for pain expressions? Is it just a conventional rule that can be given up if one wants to do that? Or is there a way of showing that the rule and the connected practice is non-conventional, and that the rule of first person authority is not arbitrary? As pointed out in the previous section, Wittgenstein’s remarks show clearly that there cannot be a foundation for this rule that is built on ontological or epistomological considerations. No ontological or epistemic privacy supports or even justifies grammatical privacy. However, this rule is not totally conventional or arbitrary either. The practice of expressing pain and reacting to it is part of our being human, hence also part of our language game of being human beings, i.e. it is built into our human form of life and has its specific place there. Pointing to this fact does not mean giving the grammatical proposition a causal explanation or foundation.

241

Rather it is a way of showing how a practice has its specific place in our form of life, and how certain facts about this human form of life — that can be seen as facts about our nature — make this rule plausible. To do this, Wittgenstein sketches a picture of the development of our language of pain and our reactions to pain behaviour. According to Wittgenstein, humans learn the language of pain as follows: A small child, a baby, cries instinctively when she feels pain. This natural expression of pain is a direct expression of pain, without any intermediate step of thinking. That we have this immediate expression of pain belongs to the foundations of our language of pain and of the possibility of first person authority. Later on, the immediate expression of pain, the cry, is step by step removed and supplanted by verbal expressions like “Ouch” and “I am in pain”, that is to say, the verbal expressions replace and supplement the nonverbal expression (cf. PI § 244). The verbal expressions never replace the cry completely, because in cases of very strong pain, human beings still cry and do not say “I am in pain”. What is important here is that the verbal expressions build on the non-verbal instinctive behaviour.6 According to Wittgenstein, the instinctive reaction to pain provides an important part of the foundation on which the rule of first person authority rests. Instinctive pain behaviour is, however, not the sole foundation of first person authority. In order for first person authority to work as a rule, it is necessary that there be an instinctive reaction to the pain behaviour of other people, too. Wittgenstein points out that in many cases we understand another person being in pain immediately, or directly, i.e. without developing an interpretation of their utterances in the first place. The possibility of pretence cannot stand at the beginning of our language game with sensations; pretence is learnt later (cf. PI § 246; RPP II § 612; LW I § 253). Our practice of reacting to other people’s sensation-utterances is rather built on an instinctive reaction to the other persons behaviour. Wittgen6. In MS 124, Wittgenstein develops his ideas that are later contained in the remarks about sensations in the Philosophical Investigations. He suggests that the words “I am in pain” are part of the relevant pain behaviour, and he continues by making a general claim: “Und so sind alle sprachlichen Äußerungen der Empfindungen mit den ursprünglichen Empfindungsäußerungen verknüpft”, i.e. “And in this way, all verbal expressions of sensations are tied up with the original expressions of sensations” (MS 124, 223). Yet, this claim does not seem to be true when it is understood as a general remark about all sensation words. For there are many sensations where it is unclear what their natural expression would be. It is, however, true for such central sensations as pain or joy.

242

stein remarks that “it is a primitive reaction to tend, to treat, the part that hurts when someone else in in pain, — and not merely when oneself is” (Z § 540). He goes on to explain the word “primitive” in this context: “But what does the word ‘primitive’ meant to say here? Presumably that this sort of behaviour is pre-linguistic: that a language game is based on it, that it is a prototype of a way of thinking and not the result of thought” (Z § 541). That is to say: Our practice of talking about sensations and understanding other people’s expressions of their sensations rests on preverbal behaviour in the sense that, firstly, there are instinctive expressions of pain and, secondly, that there are instinctive ways of reacting to the pain behaviour of someone else being in pain. The practice rests on a form of behaviour that is shared among all human beings (and some animals). This form of behaviour is a natural fact, a fact about us as human beings. The grammar of sensations, i.e. the grammar of the the word “pain” in particular, and the grammar of the proposition “Sensations are private” rests upon these natural facts. Of course, it is an open question whether the grammar of pain can be explained by these natural facts. But referring to these natural facts at least makes the grammar of pain and the validity of the grammatical proposition understandable. It creates a background for our understanding of the grammar of sensations and gives them a foundation in our preverbal behaviour. In the next part of my paper I will examine this connection between grammar and natural facts more closely by looking again at the connection between the immediate expression of pain and the grammar of the sentence “I am in pain”. 3. Grammar and Natural Facts in the Language of Sensations The described relationship between the grammatical proposition “Sensations are private” and our instinctive behaviour can be read as a contribution to a question that concerns Wittgenstein in the 1940s: What kind of relation is there between grammar and natural facts? It is unclear what Wittgenstein’s answer to this general question would be, because there are only some scattered remarks on this theme in his writings.7 Another 7. Wittgenstein discusses this question in its general form mainly in On Certainty and in some remarks in Philosophical Investigations II, e.g. PI II 230; RPP I § 48; Z § 350; OC §§ 211, 617. The main results of these investigations are that (a) the rules of grammar cannot be justified by facts of nature and (b) that facts of nature make a grammar practical or impractical.

243

reason could be that this question cannot be answered easily in its generality. Therefore, I want to investigate this question for a particular case: the relation between the grammar of pain (or of pain-expressions) and the natural facts of our instinctive behaviour. Asking the following three questions will help us to understand this relation better: (i)

What kind of dependency is there between the fact that the natural reaction to pain is to cry and the grammar of the sentence “I am in pain”? (ii) Is it possible to imagine a concept-word “pain” that has a radically different grammar from our normal grammar of “pain”? What would follow from a radical change of our grammar? (iii) Are propositions that account for natural facts like “Babies cannot pretend to smile” empirical or grammatical propositions?

The first question builds on what was said above about the relationship between crying as a natural reaction and the grammar of “pain”. It was shown that our concept of “pain” develops out of our normal, or natural, reaction to pain. However, it remains unclear so far whether or not this grammar of “pain” can be explained causally by pointing to the natural reaction. Wittgenstein, however, stresses in several remarks that there can be no causal link between grammar and natural facts, i.e. the formation of concepts cannot be explained exclusively in causal terms. For stating a causal connection between a natural fact and a concept would be a kind of hypothesis that cannot be proved. These kinds of hypotheses are not what is at issue when explaining grammar. There is no causal connection between our natural reaction to cry when we feel pain and the grammar of the word “pain”, such that the natural reaction alone gives a causal explanation for our uses of “pain”. Any explanation has to start with or presuppose a grammar that is already in use. The causal model simply is the wrong model as far as grammar is concerned. However, pointing to the fact that the grammar of “pain” rests partly on natural reactions to pain helps us to make that grammar understandable. It shows a connection that is important and that helps us to see the implications of the grammatical rule “Sensations are private” more clearly. The second question is directly connected to the first one and helps us to understand better how the connection between natural facts and grammar must be conceived. If there would be no causal relation between the grammar and the natural facts, would it then be possible to develop radically

244

different concepts of pain, given that our natural reaction of expressing pain remains the same? Wittgenstein discusses the case of a tribe with a radically different type of pain language in Zettel: Imagine that the people of a tribe were brought up from early youth to give no expression of feeling of any kind. They find it childish, something to be got rid of. Let the training be severe. ‘Pain’ is not spoken of; especially not in the form of a conjecture “Perhaps he has got …”. If anyone complains, he is ridiculed or punished. There is no such thing as the suspicion of shamming. Complaining is so to speak already shamming. (Z § 383)

This remark occurs in a context that discusses pretending. The difference between the people of this tribe and us consists in the way we and they relate to pretending. In their practice of not uttering any expressions of sensation (e.g. avowals) as well as of reacting to other people’s expressions of sensations, there is no place at all for pretence and no place for the normal reactions to pain. They do not share our ways of reacting to the pain of other people like we do, but they try to abandon all kinds of bodily or verbal expressions of pain. In other words, the proposition “Sensations are private” would not be valid here. An additional conclusion is also drawn by Wittgenstein in Zettel § 387: “I want to say: an education quite different from ours might also be the foundation for quite different concepts.” Wittgenstein here directs our attention to the fact that education and the way a society relates to sensations makes it possible to have concepts very different from ours. This, of course, implies that their life would be very different from our life. However, it is questionable whether this thought experiment really leads to such a conclusion. These doubts are supported by another remark: “These men would have nothing human about them.” Why? — We could not possibly make ourselves understood to them. Not even as we can to a dog. We could not find our feet with them. / And yet there surely could be such beings, who in other respects were human. (Z § 390)

In this remark, two positions are contrasted with one another: On the one hand, we are forced to say that the people of the tribe without a sensation language cannot be seen as human beings. They do not share our ways of acting, and we could not have any kind of understanding of their pratice. We could not recognize ourselves in what they do and how they react. On the other hand, Wittgenstein suggests in the last sentence of Z § 390 that people like this would still be human beings. How do these two positions work together?

245

I think it is right to say that we reach the limits of our understanding when we consider human beings so described. We cannot see them as ourselves, i.e. as human beings who practice a human form of life like we do. That these people could still be imagined as human beings does not mean more than that they would belong to humanity as a biological species. They might have the same kind of physiology as we do. However, this belonging to the human species in this sense is not what is relevant here. What matters is the concept of being human that is connected with living a specific human form of life. In this respect, the people of the tribe imagined are not human beings as we are. The thought experiment therefore leads to the following conclusion: if we consider the grammar of pain in a radically different way, we reach the limits of our understanding of a human being and the human form of life. A radically different grammar of sensations conflicts with our understanding of what is charateristic of such a form of life. This shows that we have an understanding of a human form of life which is connected to a certain grammar and to ways of dealing with some natural facts related to that form of life. A radically different grammar of pain therefore shows that our language and grammar of sensations has its foundation in an understanding of the human form of life. The third question concerns propositions that account for natural facts. Are these propositions empirical or grammatical? I will investigate this question by considering the proposition “Babies cannot pretend to smile”. This proposition conveys the fact that babies naturally react to other people by smiling. This reaction is considered instinctive, just as immediate crying is considered to be a natural reaction to pain. There is no possibility of pretence here. In the Philosophical Investigations, after having made his comparison between the privacy of sensations and patience (see above), Wittgenstein adds the following remark: “Are we perhaps over-hasty in our assumption that the smile of an unweaned infant is not a pretence? — And on what experience is our assumption based? / (Lying is a language game that needs to be learned like any other one)” (PI § 249). To be sure, this remark is in several respects mysterious. Its style suggests that Wittgenstein is being ironic here and wants to claim exactly the opposite of what he writes: that babies cannot pretend to smile is not an assumption based on an experience, but something else. The question is what kind of remark is made by this sentence. There is a debate among the interpreters about whether Wittgenstein’s remark here should be understood as grammatical or as empirical. Eike von Savigny claims that it must be a grammatical reminder about the concept

246

of “pretence” (von Savigny 1996, 141–55). It shows that we cannot ascribe the pretence of smiling to babies because our concept of pretence implies that we ascribe it to children only once they have certain abilities and are therefore responsible for their utterances. Of course, this interpretation faces the problem of having to deal with the remark in parentheses which seems to suggest that pretence is learnt. According to von Savigny, the fact that we start to talk about pretence involves a change in our attitude towards a human being — not something based on a new experience. Von Savigny even goes one step further: He describes grammatical propositions as conventions and claims that we could change our attitude towards the baby. Then we would say that she can pretend to smile. The “cannot” in the sentence “Babies cannot pretend to smile” indicates not a lack of ability, but a logical impossibility created by convention. Severin Schroeder criticises this view by pointing out that von Savigny’s interpretation involves a concept of “pretence” that is not our normal concept, but one that involves moralistic claims because von Savigny connects it with the claim that we do not see babies as responsible for their utterances (Schroeder 1997, 161–69). Schroeder compares the sentence “Babies cannot pretend to smile” with the sentence “Bachelors are not married” and shows that the former cannot be a grammatical proposition like the latter. “Babies can pretend to smile” is not contrary to the rules of our language in the same way as “There could be married bachelors”. Schroeder claims that there are reasons why the behaviour of the baby cannot be pretence. These reasons are partly empirical, partly conceptual. Experience has taught us that babies are unable to do certain things. There are criteria for intelligent behaviour, and babies are not capable of such behaviour. The proposition is therefore not grammatical, but empirical, even though there are some grammatical considerations involved. This sketch of the discussion shows that what is at issue is not only the question whether the proposition “Babies cannot pretend” is grammatical or empirical, but how we understand grammatical propositions per se. Eike von Savigny sees them as propositions that account for “conventional facts” that can quite easily be changed. Severin Schroeder, however, compares them to analytical sentences. To my mind, neither of them is right: Grammatical propositions certainly range more widely than analytical sentences, because they do not only function at the conceptual level, but also at a much deeper level, with the understanding of a practice. On the other hand, they are not just conventional either, since they work in

247

a framework of natural facts that are part of the human form of life and that cannot easily be changed. This understanding of grammatical propositions suggests the following solution to the debate whether the proposition we are examining is grammatical or empirical: The proposition must be seen as grammatical but it is so in a specific way. It is not only an explanation of the use of the words “pretence” and “baby”, but it is also a description of the practice of talking about sensations. Not all descriptions of this practice stand on the same level: the proposition “Sensations are private” accounts for an important feature of this practice. The proposition “Babies cannot pretend to smile” can be seen as of the same kind as the sentence “Babies cannot pretend to have pain”. Both sentences are connected to the proposition “Sensations are private” as a grammatical proposition. They describe a part of our human form of life that cannot be changed if we do not want to give up the way we talk and think about human beings. The sentence “Babies cannot pretend to smile” can therefore be seen as a grammatical proposition that describes something that could be called a “fact of nature”. In order to understand this, it is necessary to make a distinction between empirical facts that are described in empirical propositions and grammatical propositions that express facts of nature on another level. The latter are part of the framework in which we can understand the former. These answers to the three questions above lead to the following conclusion about the relationship between grammar and natural facts: There are propositions where grammar and natural facts work so closely together that it is impossible to disentangle them sharply. Grammar rests and builds on natural facts, though not in a causal way, but as a way of making connections visible. There are facts about the human form of life that are connected to our non-scientific understanding of a human being. Considerations about changes in grammar in this field lead to the insight that there are certain facts that are part of our understanding of a human form of life. Grammatical propositions are intimately connected with anthropological facts: They rest on the human form of life.

248

REFERENCES von Savigny, Eike 1996: Der Mensch als Mitmensch. Wittgensteins ›Philosophische Untersuchungen‹, München: Deutscher Taschenbuch Verlag. Schröder, Severin 1997, Das Privatsprachenargument. Wittgenstein über Empfindung und Ausdruck, Paderborn, Schönigh.

249

This page intentionally left blank

Grazer Philosophische Studien 71 (2006), 251–283.

ON KNOWING WHAT ONE DOES Richard RAATZSCH University of Potsdam Summary You can see me doing this or that. And your seeing me doing this or that is the source, or even the form, of your knowing what I am doing. As well as the source, or the form, of my knowing what you are doing might be my seeing you doing this or that. However, it would be strange to say that one is looking for what one is doing in order to know it. Nevertheless, it would also be strange to say that one does not know what one is doing when one is doing this or that. So, what is the source, or even the form, of one’s knowledge of what one is doing, given that one knows what one is doing? Or is there something strange about this being an assumption? When, then, do we say that one knows, and then also: that one does not know, what one is doing?

1. Suppose a friend of mine gives me a phone call. Let his first words be the following ones: Hello, what are you doing? My answer is: I’m preparing a cup of tea. He replies: And what is your wife doing? Whereupon I answer: Let me see … Well, she’s reading a book. Nothing is more natural than such a conversation. — Let this naturalness be our first fact. That something is natural usually stands out only against the background of something which is unnatural. A small change in the dialogue generates something of that kind. — Suppose the conversation had taken the following course:

Hello, what is your wife doing? She’s reading a book. And what are you doing? Let me see … well, I’m preparing a cup of tea. Here at least one thing is unnatural, odd, or strange: that I claim to see what I am doing in order to be able to know and, then, to tell it. — Let this oddity be our second fact. It already gives us the key to answer the question why it was only said that there is at least one thing strange in the second dialogue. That was said because, if this oddity attracted one’s attention, one easily becomes suspicious of the phrase “She’s reading a book”. Indeed, how does one know that? If the answer was “Just so!”, we would have to deal with a further oddity in the second dialogue. — We will come back to this point, but will put it aside for the time being. 2. Let us continue with looking at the strange fact in the second scene. What does the oddity of “Let me see … well, I’m preparing a cup of tea” consist in? 2.1 Of course, the oddity does not depend on the seeing as such. There may be actions which cannot be seen but only be heard, or felt, or whatever. But this does not make a difference for our problem. The oddity of our case goes along with all kinds of perceptions of one’s own doings. Claiming that I have to listen in order to know that I am talking, is no less strange than claiming that I have to see what I am doing in order to know that I am preparing a cup of tea. — Therefore, in what follows, seeing is just a substitute for all kinds of percipiences by which actions can be observed. 2.2 The problem also has nothing to do with me in particular. In that regard, the oddity is a general one. To say about someone else that she has to see what she is doing only means to create a further strange linguistic fact. In a word, it is strange that one looks for what one does.

252

2.3 Furthermore, the problem does not consist in the fact, if it is one, that what one claims to have to look for is right in front of one all the time, so that it will be hard to oversee it. If the oddity was of this kind, it should not be strange if, instead of the question what one is doing, the question would be: What do you see? But who would think that by this question one is asked what one is doing? One might object that one need not observe what is in front of one’s eyes all the time. For instance, one is looking out of the window, but does not realize that it is raining because one was observing something else. But one nevertheless arouses suspicion if one, while looking out of the window, is asked how the weather is and answers: “Well, let me look, ah, it is raining.” For one already does what one usually does in order to find out how the weather is. That is, the concepts of looking and observing are either not so different that, totally independent of what anyone might observe, one might say that she saw this or that, or they are indeed as different as just indicated, but this is, then, due to there being indicated more than one meaning of the concept of observing. In the first case, the objection does not work for lack of necessary conceptual independence; in the second case, the objection is not a case in point because it is based on a meaning of “to observe” different from the one on which our consideration is based. 2.4 Does the oddity consist in pretending to have to look for something which cannot be looked at at all? Is the case like the one in which one tries to figure out what color the number 3 has? Saying that there is nothing one could look for, as is true in the case of 3’s color, boils in our case down to saying that there is no doing at all. For one cannot see what color the number 3 has because it has no color at all. — But I am indeed preparing a cup of tea, so I am doing something, so there is an action. And if there was no action, and therefore nothing one could observe, how could someone else find out, by observing what I am doing, or I what someone else is doing? Now what we initially thought to be not strange becomes strange.

253

3. The last point directs our attention to the first dialogue and the first part of the second dialogue. That there is something which might be known meets halfway with the naturalness of sentences like “She knows what he is doing”. For if there is something to be observed if someone is doing something, then there is also something to be known, even if not for the one who is doing what might be known as being done. Observing is one kind of acquiring knowledge, or one form in which knowledge exists. One might know something because one observes it, or by observing it. However, that there is something to be known is not simply a consequence of there being something to be observed. For the sentence “She knows what he is doing” is as natural as the sentence “She observes what he is doing” is. Suppose now that the oddity of our second linguistic fact had to be explained by there not being anything to be known. Then the question would be: to which of these two kinds of naturalness should we attach a greater weight? But how should one answer that question? So far, there was no talk of grades of naturalness, and it is hard to see how these could be introduced in such a way that everyone who accepts the existence of this feature, if it is one, would also have to accept its scaling. Naturalness, it appears, either exists in full or not at all, at least with regard to the cases we are interested in here. Following up the last but one section, one could object that, compared to knowing, observing is primary in the sense that someone who observes that p also knows that p, but not conversely. But this would be like saying that the concept of a dog is primary to that of a mammal because, whenever there is a dog in the room, there is, conceptually (or logically) seen, also a mammal in the room, but not the other way round. But if there is a mammal in the room, there is also a particular kind of mammal in the room. If one knows something, there is at least also one of the forms in which knowledge consists or from which it results, if knowledge has such forms. However, the primacy which, according to the dog-mammal schema, one might allow observing to have vis-à-vis knowing has nothing to do with the fact of its existence, but only with its place in a hierarchy of concepts. But the primacy in question does not concern the order of things. It concerns their existence. Instead of saying that the possibility of knowledge was a consequence of the possibility of perception, one better restricts oneself to the thesis that the naturalness of each of them fits seamlessly the naturalness of the other one.

254

4. But however vaguely the thesis concerning the connection between the possibility of knowledge und that of looking after might be formulated, it confronts us at any rate with the following, somehow “bumpily” expressed question: How, in effect, does one know what one is doing? That seems to be totally unclear, compared to how one knows what someone else is doing. And we are led to this question by the oddity of our second dialogue! Here is the second question to be asked in this connection, as far as it is independent from the question just mentioned (otherwise it is the same as the question before, only put in a different, less bumpy?, way): What, in effect, is knowing what one does all about? This question has the advantage that, following section 2.1, it is formulated without reference to perceptions. But even if the second question is indeed different from the first one, it is obvious that both questions hang together closely. For, as said above, looking after is one kind of acquiring knowledge or one form in which knowledge exists (with the proviso that this kind is not available to the agent herself.) Now, let us assume that perception is indeed out of play. Then, to the extent to which one cannot have knowledge without acquiring it, the question arises: How about an alternative? And as far as knowledge can consist in perception, we do not even need this assumption in order to come across the following question: Of which kind is knowing what one does? And this question is, in evidence, akin to the first one above. As far as the first question does not coincide with the second one, they are combined by the third question. As it were, all three questions make up a unity. 5. That there is something to be known whenever someone is doing this or that, but at the same time nothing to be looked after for the one doing this or that, could be due to the fact that there is nothing to be looked after for the agent because there is nothing at all to be known for her. For, as far

255

as looking after is one of the forms in which knowledge consists, this form drops out if there is no knowledge at all; whereas, as far as looking after is a source of knowledge, this source is, as it were, blocked if there is nothing into which it could gush. Before dealing with the content of this point, a word about the two as-far-as-clauses. They serve the purpose of showing the argumentation round a problem which has already been touched upon several times. Claiming that perception is a source of knowledge makes it look as if one was dealing with two kinds of things: there is, first, the perceiving, and hereafter we come across knowledge. But this would suggest that one might perceive something without knowing it. That leads, in turn, to the sceptical question how we can know that nothing goes wrong on the way from perception to knowledge. That is, how, in the end, do we know anything at all (about the external world, as we might have to add now)? If one only says that perception is a form which knowledge may take, it becomes miraculous how one can continue to know something which one no longer perceives. Simply supposing that there is a kind of mechanism doing the transformation leads again to the sceptical question, whereas the assumption that there is no such mechanism makes it look as if one could only know what one is perceiving at the moment. Now one might think that posing two theses at the same time, each of which leads into trouble, makes things even worse. But the difficulty indicated above just results from holding only one of these theses. That life is easier for the blind and the deaf as long as they act together does not, of course, mean that the agglomeration of the two as-far-as-clauses prevents each and every problem. But it should at least save us from those problems which too easily direct our attention away from our main topic. Well, if there is, then, with regard to what one is doing, nothing to be known for the one doing it, then at least one appropriate answer to the question, directed at that very person, what she is doing seems to be this one: Don’t ask me; I do not know this! And also this one: How should I know? These answers would be as appropriate as my interjection when someone else is asking what a third party is doing:

256

Don’t ask him, he does not know this! How should she know? But all this is no less strange than the answer that I first have to look for what I am doing in order for me to know it. If the fact that this answer is strange leads us to the thought that, with regard to my own doing, there is nothing I could look for, shouldn’t, then, these answers direct us towards the insight that I can have knowledge of my own actions, and not only of other people’s actions? — and, furthermore, others also of their actions? — that one can indeed know what one is doing? 6. It even seems that one cannot only know what one is doing, but also that one usually knows it. For if one did not know it, shouldn’t one be surprised when someone else who is supposed to be able to know it discloses it to one? But with regard to one’s own doing, exactly the absence of any surprise seems to be characteristic. And one is not surprised by an event when one somehow expected it, when one knows about that which is already in progress.1 That one knows what one does explains why one is not surprised by it. Understood in this way, that one is not surprised suggests that one usually knows what one is doing. Yes, if we bear in mind what, as it were, suggests itself as an appropriate response to the answer “How should I know that?”, namely: But you surely must know what you are doing! it seems that one cannot only know what one is doing, and usually does know it, but that one knows it by necessity. No wonder, then, that one can know, und usually does know it! At this place one could be tempted to directly reject this consideration by pointing towards a case like that reported by Clementine Mukabalisa from Rwanda: “On April, 9th, they came to our house. They immediately slammed. We gave them all the money we had, and they left us alone. Then we closed the window, father took out the Bible and said that we had to pray to God now that He should forgive these people because they 1. Gilbert Ryle (1949, 174f.) writes: “Usually we are not surprised to catch ourselves having whistled, planned or imagined something and we say, if asked, that we are not surprised, because we knew we were doing these things, while we were doing them.” (More on this in note 15.)

257

did not know what they were doing.”2 — No doubt, we understand this report, and that alone, it seems, suffices to show that one does not need to know what one does. But is that really so? Does the fact that we make and understand this kind of reports really show more than that there is a sense of the expression “knowing what one does” according to which one might be uninformed with regard to what one does? Indeed, it even appears as if taking a closer look at the case in question now supports the consideration which ends in the thesis that, when doing this or that, one must know that one is doing this or that. For the fact, if it is one, that we ought to forgive the offenders because they did not know what they were doing suggests the idea that, had they known what they did, they would not have done it. But had they not done what they did, they would not need our forgiveness. Now, with regard to much, if not the most, of what people do, the question of forgiveness does not arise at all. If there are paradigmatic cases for this set of actions, preparing a cup of tea surely is an element of this set. That does not mean that there cannot be single cases of preparing a cup of tea with regard to which that question arises. For instance, it may be that someone prepares a cup of tea instead of hurrying to help his neighbours who are in desperate need of help. But if he does not know about this need, and the situation is such that there is no obligation for him to be informed about the state of his neighbours, then it will be difficult not to forgive him. Now, this ignorance is of a totally different kind from the one mentioned above. The latter ignorance would consist exactly in his preparing a cup of tea while, at the same time, being aware that his neighbours are in desperate need of his help. What he is not aware of, what he does not know, does not concern his activity under these circumstances, but, as it might be called, the moral quality, or meaning, of his activity under these circumstances. And one could almost say that, in order for him not to know about this meaning, he should actually know what he was doing. Therefore, there is a sense in which one obviously does not need to know what one does. But whatever a more exact determination of this sense will look like, it is plain that, to a large part, it coincides with what might be called “(Christian) ethics”. As far as it coincides with it, it is 2. From Chrismon, Das evangelische Magazin 11/2004, 16; cf. Hanfling 2003, 33ff. Elizabeth Wolgast (1993, 303) reminds us of the counter-part to the father’s words: “Think of the exclamation, ‘How could I do that?’ How is it that when acting deliberately we can surprise ourselves? Yet this is a common fact of moral experience. There is self-accusation and regret in such a remark, while behind it lies some misunderstanding about oneself.”

258

not necessarily beyond the limits of the present investigation (we will, for instance, come back to the problem of responsibility in sections 21.1 and 23), but it points beyond these limits. To the extent to which it looks as if, in a certain sense, one needs to know what one does in order to be able not to know it in the ethical sense, it lies inside the scope of the present investigation. Here we are after this particular sense. Thus, if there are objections against the idea that knowing one’s own actions belongs, with necessity, to performing these actions, these objections will have to come from another direction. 7. In section 2.3, it was indicated that the oddity of our second linguistic fact does not consist in something being said which goes without saying. But this does not mean that the oddity might not consist in something similar to it. That the reason for its being superfluous to say that one had to look after what one is doing is not that this goes without saying, does not mean that the reason is neither that knowing what one does belongs with necessity to what one does. This could in turn explain why it appears as if that part of the answer to the question what one does in which looking after is mentioned goes against a conversational implicature: one simply confused what must be there with what goes without saying. Much of what goes without saying must be there in some sense; and looking after often leads to, or is itself, knowledge. 8. As far as knowing what one does is necessary, whatever “necessary” might exactly mean, no satisfactory solution of the problems connected with this knowledge can simply consist in, or build upon, the idea that knowledge and action are different. The relevant point here is that one knows with necessity what one does, whereas someone else might know what another person is doing as little as the latter need to know what the former is doing. In other words, it is not just a gradual difference,3 but a qualitative one.

3. As it is for Ryle who takes this idea to be an essential part of his answer to the question what knowing what one does and self-knowledge consist in (cf. Ryle 1949, ch. VI, esp. p. 179; cf. also section 14 below).

259

9. If the important difference between knowing what someone else does and knowing what oneself does consists in the second knowledge’s being necessary, the oddity of “Let me see … well, I am preparing a cup of tea” must consist in some kind of “offense against necessity” which characterizes the relation between this knowledge and the related action. — Which kind of necessity is involved here? In order to get closer to an answer to that question, let us look at a further strange utterance: I prepare a cup of tea, but I will not prepare a cup of tea.4 Of course, this sentence must not be confused with the following one: Right now, at this moment, I am preparing a cup of tea, but later I won’t prepare a cup of tea. If we understand the first sentence in the sense of the second one, or, to be more exact, if we understand the second sentence in the sense in which we immediately understand the first one when we hear or see the latter, then there is nothing strange about the second sentence, except, perhaps, its style, and then, there is also nothing strange about the first sentence. But understood in a different way, the first sentence is strange. However, its oddity does not consist in its straightforwardly being a contradiction, like, for instance, the sentence “I prepare a cup of tea, and I do not prepare a cup of tea” is. Sure enough, it is also not a tautology, like “If I prepare tea, then I prepare tea”. Finally, it is not an ordinary assertion either, viz. something that might be true or false, as is “He prepares a cup of tea”. “It is not a straightforward contradiction” means as much as “It is not a contradiction, but how could it ever be the way the speaker says it is?” That is, the sentence has as least a modal pull. That it is not straightforwardly a contradiction is, of course, due to the fact that both of its clauses are not straightforward negations of one another. That the sentence nevertheless 4. This is evocative of Moore’s Paradox. The analogy is only limited, but as far as it holds, the problem does not only concern Moore’s sentence but already the sentence “It is raining, and I believe it”, although it is not as obvious in this case as in the other one. Regarding the oddity of a sentence of the form “I do D, but won’t do D”, compare the last paragraph of Anscombe 1963, 91-4. We allude to this section at several places; thus, for instance, to the remark according to which such a sentence is “a contradiction of a sort […]. And yet we feel that this is not, so to speak, a head-on contradiction, like that of pairs of contradictory orders, contradictory hypotheses, or opposed intentions” (Anscombe 1963, 92).

260

sounds like a contradiction, though it is not one, points at both the two clauses’ being independent from one another and their being so constituted that they still “cannot” be straightforwardly connected in the given way. Here the negation of the second clause does not seem to play an essential role. For the sentence “I do not prepare a cup of tea, but I will prepare a cup of tea” sounds odd as well. If anything, both clauses are so constituted that none of them can, by conjunction, be connected with the negated other one without there arising a strange whole. Therefore, the real problem does not reside in the strange whole as such. Exactly this is shown by the fact that it would not straightforwardly be a tautology if the sentence was “It is not the case that I prepare tea but will not prepare tea”. “Not straightforwardly a tautology” means as much as “It is true that it is not a tautology, by how could it be other than the speaker says it is?” Result: Both the sentence “I prepare a cup of tea” and the sentence “I will prepare no cup of tea” seem, on the one hand, to be independent of one another, but, on the other hand, make up, when joined together in the way described, an “almost-contradiction”, which in turn speaks in favour of their not being fully independent from one another. 10. What does their (limited) independence consist in? One could say that it consists in the fact that one may express an intention with the sentence I prepare a cup of tea, but can also make a prediction. Of course, one may do the last thing also by saying: I will prepare a cup of tea. But this does not turn this sentence into a prediction, instead of, or even contrary to, an expression of an intention. For one might use it in this and also in that way. Given the way language is designed, one cannot just take any sentence in order to express an intention or make a prediction. More exactly, although one can indeed take any sentence, namely if it is understood in the appropriate way, it is nevertheless not usual to take any sentence. Instead there are particular (forms of ) sentences in use for that purpose. However, the standardisation is not so strong that there is one

261

and only one generally used linguistic expression to make a prediction or express an intention. Distinguishing between expressing an intention and predicting can, of course, hardly be the whole story! For the combination of an expression of an intention with a prediction might be fully natural, that is, without any problem. Here are two examples: I prepare a cup of tea, but Peter won’t do so. Peter will prepare a cup of tea, though I won’t do so. As far as the second example is an example for a combination of predicting an event and expressing an intention, this is due to the fact that we not only use “I prepare a cup of tea” in order to express an intention to prepare a cup of tea, but also “I will prepare a cup of tea”. And similarly, we can use “I prepare a cup of tea” not only to express an intention, but also to predict something. (Do we then express ourselves somehow incorrectly? — No doubt, but this is incorrect only according to certain textbooks of grammar. The problem is that the latter receive their prestige, if they have some, from our actual way of speaking!) As far as the sentence “I prepare a cup of tea, but will not prepare a cup of tea” is an “almost-contradiction”, its clauses cannot both pose merely as expressions of an attitude or as predictions. For in that case, if we leave the temporal aspect aside, the word “almost” in the phrase “almost-expression” would be superfluous. But it is not. Therefore, one of the two clauses must function as a prediction, while the other must work as an expression of an intention — but they cannot both function in only one of these ways. Rather, the expression of intention must somehow also be a prediction. That there must be more than just a side by side, although it cannot be an identity either, is made visible by the following observation. If one considers what one did in this or that past situation in order to find out what one will do, one behaves in a way which might be illuminated by considering which weather so far followed which cluster of clouds. This is a paradigm for carrying out deliberations in order to make forecasts. Now the forecast one made might turn out to be wrong. And if others relied on one’s prediction in their actions, one should not be surprised if they hold one to account. For example, one might have to accept the reproach not to have made clear enough the uncertainty of the forecast. But there is one thing one does not have to accept here: the reproach to have changed one’s mind. But this is exactly what one has to take into account when one

262

makes a decision, declares it, but then does not put it into practice. In this respect, prediction and expression of intention are different. — However, that should not make one ignore that what one risks when changing an expressed intention might still be a reproach. In fact, one easily becomes subject to justified reproaches because, when one expresses one’s intention, others usually have a right to rely on one’s realizing the intention. Therefore, one takes the easy way out when one sees the connection between expression of intention and forecast only in the possibility of the listener’s using the intention’s expression as a basis for her forecast. If one catches a justified reproach due to one’s change of intention, then the connection is closer than it appears to be in this picture. — As said above, the problem is that the expression of an intention neither conceptually falls together with a forecast nor seems to be completely separable from the latter. Now, if a prediction is not an expression of an intention, and if we have only one (part of a) sentence in our example above, what is the nature of that part? It is the nature of a hermaphrodite and thus, as was said, of something strange! 11. A side remark: expressions of intentions are, in some sense, as well as forecasts, as it were, temporally directed. If this gets lost, then each and every act of doing something is another story. But then a sequence of such acts is nothing but a more or less accidental succession of events. There is, then, no longer any genuinely homogenous procedure which might be called “doing of something” and which extends beyond that what happens at the moment. But in this case one could not even say, in any exact manner at least, that this doing was the more or less accidental succession of events. Which doing, after all? No doubt, one may, for instance, present a long jump in a freeze picture such that this single picture shows the jumper at different moments, or stages, of his jump. Then one has one picture which pictures the jumper several times, at different places in the picture. One may even assign a special act to any single part of the picture by saying: “Here he jumps off; there he stretches his arms forward; now he’s landing.” (Hence the talk of “stages”, not simply “moments”; the start, or the beginning, of a doing is often a stage and thus a process, not a moment.) One can of course say that the jump as a whole consisted of these single acts. One might even, and often does, practise these acts for themselves. But the fact that one can speak and behave in these ways does not mean that the jump was the sum of these acts in a sense according to which each of

263

these acts could, simply by itself, be the one that it is, i.e., independently of the whole to which the sum of these acts supposedly amounts. In order for the majority of such pictures to be pictures of processes of long jumps, the long jump would, as it were, already have to be there. For, without it, those acts would not be determined as parts of a long jump. But then the long jump is not simply the result of these single acts. How could something which consists of parts exist before those parts existed of which it supposedly consists? The representation of a process as a sum of single events rests, at least in the sense indicated here, on a concept of this process, or comes along with it. In this respect, the process is prior, or at least not secondary, vis-à-vis the single events, although it is still true that a process can hardly be anything but a sequence of events. 12. But let us return to our actual problem. Before we digressed, we were concerned with an almost-contradiction and a hybrid being. Both hang together. In order for two sentences to be contradictory, they must, in some sense, contain the same concepts, although not necessarily the same words. “He sold his bank, but he kept his bank“ is a contradiction only if the word “bank“ in both parts of that complex sentence is used in the same way, that is, if the same concept is used. That something is a hybrid means, on the one hand, that it is the same as that of which it is a hybrid, while, on the other hand, it is at the same time different from it, which it must be, since those beings to whom it complies are different. Hence the almost-contradiction! — In the very same line of conceptual hotchpotch we find our linguistic oddities. As far as they belong there, the necessity with which one is supposed to know what one is doing comes as close to a conceptual necessity as a necessity may ever come without there arising a contradiction when a sentence which expresses it is conjunctively connected with a negation of it. Due to their conceptual nature, our oddities are connected with the facts only to that extent to which our concepts are. To put it paradoxically, our concepts are yet not defined by their definition. Prosaically speaking, we have the concepts we have because we follow certain definitions — and because it depends here on both the definitions and our following them, that is, the use of the words according to their respective definitions in the practice of our language.

264

13. Now it also seems to belong to the concept of knowledge that one might ask where a supposed knowledge stems from, or what it consists in, such as it seems to belong to the concept of action that the agent knows what she does. — That brings us back to the question how one can know what one does if, as a source or a form of that knowledge, looking for what one does drops out. And here a certain form of an answer forces itself upon us. An example should outline this form. With regard to the question of how one knows what one does, one could hit on the following idea: one knows this because one has thought of how one will behave in a certain situation. However, one has not reasoned in such a way that one asked oneself what one did so far in such situations. One does not reason such that one reaches a result of the following form: If I will really do what I did so far in such cases, I will do A. That would just be a forecast. The intention would again have dropped out. Rather, one reasons in such a way that the result has a different form, namely this one: I will do A. One simply considers what one should do. In other words, one considers reasons and, if that is something additional, one decides to perform an action.5 If the question at issue is whether it is going to rain or not, no such reasons come into play. When asking oneself what, in the light of what one did in the past, one is likely to do, one’s attitude resembles that towards the possible future rain. There is no doubt that knowledge which results from such an attitude also belongs to the so-called self-knowledge. To the extent to which such knowledge constitutes self-knowledge, this is not, or 5. According to Hampshire and Hart (1958), someone who has to decide on two or more alternatives need not be certain, but might become certain about what he will do in two different ways: (1) by considering what he had done so far in such situations — in this case he gains empirical certainty, and a respective explanation on his side would count as a prediction, not as a decision — or (2), by considering reasons and making up his mind — then what he says will count as an announcement of his decision, not as a prediction. Hampshire and Hart take it for granted that an expression of an intention is not a prediction, because the way of criticising them is quite different. On the question whether one might blame someone for changing his mind, compare section 11. — Wittgenstein writes: “When asked ‘Will you do such-and-such?’, I consider grounds for and against [the action]” (RPP I § 815).

265

not to the same degree, necessary as is knowing what one does. It seems as if one could be the one who one is and at the same time not know who one is. As far as this kind of self-knowledge is concerned, others may know better who one is. But even if it was a matter of course that we know ourselves best, this is only relevant to us insofar as it forces us to separate our problem from others, which might easily be confused with it, and so to get clear about our problem’s nature. Regarding this nature, we seem to have found what we were looking for in our considering reasons: a way of acquiring knowledge what one does, compared to knowledge who or what kind of person one is. And we have found this answer without leaving the realm of the conceptual. We simply bore in mind what it means to consider reasons. At the same time, this kind of knowledge acquisition explains why one is usually not surprised by one’s own actions. It explains this exactly in the way which was indicated above: because one thought of what to do (and reached a decision), one also counted on one’s doing, or already knew of it when it started. But it catches one’s eye that this solution is unsatisfactory in the sense of immediately prompting new problems. The main problem is not that one cannot infer action from decision, so that, even if one knew one’s decision, one would still not know what one does. This objection against the “solution” considered above might perhaps be rebutted by conceptually including the decision into the action such that the decision itself already counts as the first moment of the action. This reflection would resemble the one made above, which we tried to counteract by the clause “as far as this is something additional”. This clause expressed the suspicion that, even if weakness of the will is possible, this does not prove that, normally, an act of making up one’s mind has to be added to the considering of reasons in order for the agent really to act such that, if the action will not be executed, although the reasons have been considered, the agent has to suffer from weakness of the will. (These doubts may be justified in a way which resembles our remarks concerning the relation between acting and trying to act given below.) The real problem resides rather in the endless sequence of steps to which the “solution” we are investigating here leads us. For as long as one talks the way one ordinarily talks, considering reasons is itself an activity. One can consider reasons more or less carefully, it takes some time, may be interrupted and later continued, and so on. If it is an activity, then, according to the supposition on which this “solution” is based, one needs

266

to know about it whenever one is engaged in it. In a word, one must consider reasons before one can consider reasons; thus one never gets down to considering reasons, therefore one will never acquire knowledge and never act. One might, of course, allude to the idea that the considering of reasons with which we are concerned here determines the form of the whole; in other words, that we rather talk about an as-if-representation, where this concept of an as-if-representation contrasts with a representation which refers only to that which may be proved to exist. One can describe the movement of a billiard ball not only by saying how the billiard cue actually hit the ball, but also as if the ball’s movements were the result of an operation of several strikes of the ball, that is, by using the conception of a parallelogram of forces. — But in our case this simile is unconvincing, exactly because the objection was itself a conceptual one and did not consist of an appeal to facts. Put differently, even if the considering of reasons was only a consideration in a sense which allows that it might have been unconscious (cf. Dennett 1976), the usual, ordinary considering of reasons would still serve as the paradigm. And one made a clear statement when talking about a considering of reasons, be it conscious or unconscious, only to the extent to which one relies on this paradigm. One may also deny that the consideration of reasons falls itself under the concept of action, in which case it is not itself an action. But then one runs into further conceptual trouble: of which kind, then, is the considering of reasons? And, what is perhaps simply the reverse, why does it look as if it was a kind of action? And if it is no action, how can it then solve our problem at all? Defining the concept of considering reasons such that it is no action is, of course, out of the question. For if one changes a concept, and not only a façon de parler, one simply speaks of something other than before. 14. For us, the solution just outlined is not valuable in itself. It only serves as an exemplification of a defect of many approaches to a solution: a certain lack of being radical. One accepts, in other words, too much as given. This lack of being radical corresponds to a lack of insight into the nature of the strange. Remember that not only did we say that our first short dialogue is completely natural. We also said that we usually become aware of a naturalness only against the background of something strange. That sounds

267

like a psychological thesis, but it actually expresses a conceptual relation between the natural — or, as one might also put it, the normal, the usual, the ordinary –, on the one hand and the strange on the other. As far as we are interested in the strange, we only need to read, as it were, the relation between it and the ordinary the other way round — in order to see that the fact that something is strange means in the first place that it is so in relation to something else. In other words, one explains what it means that something is strange by saying that it is not like …, and here the empty space might be filled in in just as many different ways as there are ways in which something can be natural, normal, usual, or ordinary. It is true that one has to accept something in order for there to arise a problem. — We arrived at our problem by realizing something strange, entailing that something else was natural for us. That is, what one accepts as normal plays a role already with regard to the genesis of a philosophical difficulty, and not only with regard to its removal! Now, since in many cases of knowledge one may ask for the source of the knowledge, or what it consists in, it seems as if this should also be true of a case in which (we tend to say that) someone knows what she herself is doing. One may ask how she knows what he is doing; and a possible answer is: by looking at him, by seeing. However, it does sound odd to give this answer when the question refers to someone’s own doing. Is it thus true that one does not know what one is doing? If so, then it should not be strange if someone was looking for what he himself is doing. And so on. Regarding the facts, there is at least no obvious difference between his situation and mine which would allow to deny me the right to say: I know that I am preparing a cup of tea. Or also: I know what I am doing. Thus in general: One knows what one does. But since this does not remove the oddity of one’s looking for what one does, one appears to be forced to view the knowledge of one’s own doing

268

as a case of direct knowledge, a knowledge which is free of, or unmediated by, perception. However, it is also clear that knowing what one does is not free of, or unmediated by, perception in the sense in which one might perhaps say that mathematical knowledge is independent from perception. Knowing what one does cannot be independent from perception in this way because mathematical knowledge, if there is such knowledge, is independent from perception as a matter of principle. That is, it is not the case that perception does not play a role for my mathematical knowledge, although it does so for your knowledge. But when it comes to knowledge of human actions, perception does play a role, namely with regard to knowing what someone else is doing. (As far as perception plays a role in mathematics, it does so for everyone in every case.) To pick up the example of a solution mentioned in the preceding section, does mathematical knowledge resemble knowledge of one’s own actions insofar as considering reasons resembles calculating? — If so, the resemblance is too strong rather than too weak. For what corresponds in the case of mathematical knowledge to her knowing what he is doing? She does not simply know what he is doing by considering what he should do, even if her consideration is absolutely correct. If we adjust knowledge of one’s doing to mathematical knowledge, as it is understood here, we lose the unity within the variety of the concept “knowing what one does”. The fact that we have a concept of knowledge which is not connected with the possibility of perceiving that which is known lends a certain plausibility to our talk about knowledge without observation. But the sense this concept has in its original place does not fit, as it were, the case which interests us here. Hence, our problem is still what it might mean that one knows what one does, if the source, or the form of the existence, of this knowledge is completely dubious. 15. On the one hand, the question where the supposed knowledge comes from is appropriate in countless cases in which the claim that someone knows that p makes sense. Here this claim and that question are connected in such a way that any answer to the question stating in effect that the knowledge originates from nothing makes the claim itself questionable. “How do I know that she is reading a book? Well, just so!” — “This is not called ‘knowing’, this is called ‘guessing’.” On the other hand, the word “to know” is not ambiguous (as Davidson 1984 and 1991, 205 also empha-

269

sizes). Therefore, it seems that, if it is essential to the concept of knowledge that one can say how one acquired a given instance of knowledge, or in what it consists, then the question how one knows should be allowed also in the case in which one knows what one does. Furthermore, if we do sometimes come across a way of speaking which does not fit the schema, this need not compel us to sacrifice the schema. Seeing a cat with three legs does not compel us to eradicate four-leggedness from our concept of a cat, given that it belongs to it. A three-legged cat may simply be an exception, a pathological case of catness. Thus, it would appear that our problem consists in the need to explain how a special case of knowledge might be incorporated into our general concept of knowledge. The special case’s oddity consists exactly in its not neatly fitting into the general frame. 16. But if we formulate the problem in this way, we make, explicitly or not, the assumption that there is a certain kind of uniformity in the use of an expression of the form “A knows that p” such that whenever an utterance of an expression of this kind is in order, the question how A knows that p is also appropriate. And if it is not appropriate, it might be shown that we are confronted with an anomaly which might in turn be explained by specifying “interfering influences”, as it is the case with regard to a three-legged cat. — But need there really be such a uniformity? Isn’t the only thing one is indeed entitled to say the following: if, in certain cases of using the expression “A knows that p”, things are so and so, and if we are not dealing with an ambiguity when using “A knows that p” at some other place, then there must be an understandable connection between these different uses of “A knows that p” such that we do not need to talk about an ambiguity of that schema? After all, the opposite of an ambiguity is not just logico-conceptual uniformity. Regarding the pathological cases, a three-legged cat is indeed a pathological individual. But what if such a cat had three-legged offspring? Is it, then, still a pathological exemplar or rather a normal exemplar of a new kind of cat? That is, there are pathological cases of single kinds, but this does not mean that everything which does not seamlessly fit in a certain schema must be a pathological case of that schema. It may instead belong to another one or, as far as the schema will be kept, confers a new use on it, one which is akin to the old one. The kinship might be so close that one does not simply want to speak of two kinds, i.e. two meanings, but

270

rather finds another way of linguistically putting the matter. To speak, in the case of knowing what one does, of knowledge without observation, or of direct knowledge, a knowledge which one has just so, is the attempt to keep the unity of the concept of knowledge, while, at the same time, taking its variety into account. Here one has a standard case in mind, meaning that all cases which do not obviously fit to it appear as abnormalities in need of an explanation: one knows what one does, but one does not know it in this or that way. The problem is buried in the manner of speaking. There appears a predicate in it with the same number of variables as the predicate one uses when saying that Paul knows what Mary is doing. The number of variables of this predicate is, among other things, determined by the fact that one might add how Paul acquired his knowledge. “Knowledge without observation” looks, as far as the form of that expression, i.e., its logical multiplicity, is concerned, exactly like “knowledge by observation”. But its real form is a different one. For “… without observation” is not one substitution among others, that is, one cannot complete it by “…, but by X”. As far as the addendum “… without observation” is appropriate, it rather corresponds to the rejection of the question which addendum is appropriate in this case. This complies with the fact that, with regard to direct, unmediated knowledge, one cannot say that one acquired it, but only that it happened to one, for “direct”, or “unmediated”, is the rejection of the question “Acquired in which way?”. 17. To speak of an anomaly of the first person with respect to knowledge of one’s own actions may, therefore, show less about the facts and more about the way one views the world. This way of speaking bears witness to what one takes to be the normal case, the standard. Regarding the matter itself, we see nothing but diversity. It is as if someone was to say that the graph of an exponential function was, compared to that of a linear function, an anomaly, which, of course, says nothing about the graph, but only about what the speaker has in mind as the normal case of the graph of a function. As long as the question of priority is an open one, one cannot exclude that the answer is that none of the different phenomena is basic with regard to others. This insight allows one to view the fact that one does not have to look for what one is doing in order to know it as a primary fact, that is, as one which is just not subject to an explanation. To the extent to which we want to say that one knows what one does, but, at the same time, do not want to speak of a source of that knowledge, we do

271

not speak of knowledge in the sense in which the question regarding that source is appropriate. But this need not mean that the word “to know” is ambigious. Perhaps the uses of “to know” in the relevant cases resemble each other with regard to essential aspects, although with regard to other essential aspects they do not. That may be the case if the essence, or the concept, need not be something general. This possibility is easily ignored when one tries to solve the problem in one of the usual ways. In fact, one ignores this possibility not only when one is looking for a solution, but already when one is formulating the problem. 18. With regard to the meaning of “to know”, our problem rests on an oversimplified conception of understandable connections between its uses. Removing this conception comes along with the possibility of accepting individual (kinds of ) uses as autonomous. If single uses are autonomous in the sense of “not being reducible to others”, then the fact that “How should I know?” is in some cases a strange answer to the question how one knows does not license one to infer that this answer is strange in all cases. Thereby the idea that one needs to know what one does loses quite a lot of its plausibility, given that it is understood as a general, conceptually necessary feature of the (or a part of the) concept of knowing what one does. The oddity with which we started, and those with which we continued, do not at all point towards such general necessity. — What, then, do they refer to? 19. “Let me see what I am doing” need not be strange for the reason that here a self-evident necessary condition of my doing is made explicit. Instead, the oddity might consist in the fact that “How should I know that?” is, in this context, in contrast to other contexts, not a meaningful answer at all, while one does not see that at first glance by looking at this context. The senselessness of this answer corresponds to the senselessness of the other answer. In this case, “But surely you must know it” does not simply remind one of a necessary connection. It is rather a rejection of that answer in this context. But the fact that one says that one must know what one does because one wants to negate the idea that one does not know it — which makes one look as if one agreed to being at least able to know it — does not license the inference to the claim that one simply wanted to say that one does know it in the sense in which one knows what someone else is doing.

272

In many cases in which one knows something, one might not have known it. Maybe I know what she is doing, maybe I don’t. Here “I do not know this” is one out of several different appropriate utterances which are destined for this kind of situation. It would only be strange to put it forward in a situation which usually makes sure that one does know the thing in question, as it was strange to reply in this way in the situation, mentioned in section 2.3, in which one is already looking out of the window when being asked what the weather is like. In the sense of “to know” which comes along with the possibility of being ignorant, one just does not know what one does in a case like that displayed by the second of our dialogues at the beginning. As far as one knows it, one at best knows it in a different sense. But both senses, the one in which one knows it and the one in which one does not know it, need not be so different from one another that we have to speak of two meanings of “to know”. “He knows what he is doing” does not provoke the question “In which sense of ‘to know’?” in the same way, if at all, in which “He bought himself a bank” may provoke the question “In which sense of ‘bank’?”. In the latter case, the well-known ambiguity forestalls confusions which, due to the resemblances, are promoted in the former case. 20. As to generality, the fact that “I do not know this” may serve both as a rejection and an answer to a question means that we are concerned with two different, although related, conceptual patterns. Now, what kind of case is it in which “I know what I am doing, namely D” and “I do not know what I am doing” have to be considered as answers? A general characterization could read as follows. One says that one knows what one does, or that one does this or that, when the circumstances are such that one should not be surprised if others say that one does not know what one is doing. One could add that these are circumstances in which one would like to deny that anything is not the way it usually is, or should be, in this kind of context, in contrast to how it might appear to be. However, that should not lead one to neglect the possibility that the different cases are autonomous. To the extent to which this possibility comes to the fore, the attraction of generally characterizing all of the things which fall under a concept loses its power. Instead, lists of different cases become more and more illuminating. — The list to be presented in the next subsections (it has already been opened by mentioning the case in section 6 above) will cul-

273

minate in cases where one might even, by looking, find out what one is doing, contrary to what one might think in the first place. 20.1 A first way not to know what one does consists in not knowing the consequences of one’s doing. One might object that this is, strictly speaking, irrelevant. For what one does and which consequences one’s doing has are not only terminologically two matters. Otherwise, the discussion about the question to which extent one is not only responsible for what one does but also for the consequences of one’s actions would have no clear subject. Therefore, the fact that someone does not know what the consequences of his doing are can be no instance for someone not knowing what he does. Such an argumentation would be more than precise, it would be captious. For what one is responsible for is what one engages in by one’s action; and to this the consequences, at least partially, often belong. More exactly, the concept of what one engages in just serves, among other things, the purpose of bridging the gap between the concept of action and the concept of the consequences of an action in order to bind both concepts closer together as they are already bound together by the fact that one can only talk of consequences where one can also, at least in principle, answer the question: the consequences of what? Otherwise, the discussion about the question to which extent one is not only responsible for what one does but also for the consequences of one’s actions would not have the sense it actually has. For that sense just consists in separating, for different cases, the realm where the action and its consequences fall together from the realm where they fall apart. Where we talk about knowledge of what one does, there one might not know what one does to that extent to which it is hard to establish a border between those two realms.6 But here the knowledge has nothing to do with any kind of special access to one’s doing, an access which results, for instance, from a privileged access to one’s own intentions.

6. See Austin 1956/57 on the range of events certain expressions for actions may comprise. The example John Perry (1979) focuses on is relevant here: By driving around with a shopping cart containing a ripped bag of sugar, he leaves a trace of sugar on the ground and thus makes a mess. That is what he is getting clear about step by step, not that he is driving around with a shopping cart. (It is difficult to say whether Perry’s two further examples might be described in that way.) The difference in what he is doing (driving around, making a mess) finds an echo in the difference in the oddity of his respective not knowing what he does.

274

20.2 Although the following does not coincide with ignorance regarding the consequences of one’s own actions, it hangs closely together with it. One is doing something which has in this context another meaning than it has in a different context, and one behaves in the way one had to, in the eyes of the public, if one was in a context different from the context one, in their eyes, actually is in. For example, one understands an utterance as an expression of sympathy … but it is only a sign of politeness. If one responds, then, to the utterance according to the way one understood it, others will sometimes not understand one’s response “as it was meant”. As far as the others’ understanding fixes the content of what one does, and as far as truth belongs to knowledge, one does, once again, not know what one does. 20.3 A further kind of not knowing what one does is connected with the two mentioned before in such a way that this third kind might even be viewed as a combination of them. The point is that some actions are, compared to others, extremely complex. Here is an example. The Prime Minister, of whom we know that he likes to smoke cigars, aims at a change in the social policy of his cabinet. Such a change is, compared to smoking a cigar, extremely complex. With regard to this change, we also say that the Prime Minister follows a plan. But does one follow a plan when smoking a cigar?7 Nevertheless, to the extent to which a change in the government’s social policy is a kind of doing, it falls under the concept we are concerned with here, that is, the concept of knowing what one does. That the Prime Minister does not know what he does might mean that he got mixed up with his plan, lost the survey, tends to give up his plan without being clear about this yet, and other things. However, such cases are surely not strange in the sense that the Prime Minister makes a conceptual, a logical mistake when losing the survey etc. It is even open to debate whether this would be strange in the sense of “usually not happening”. To say that one knows what one does would, in such a case, simply mean to disclose, to affirm, or to confirm one’s plan. Since one usually does not follow a plan 7. Peter Hacker (2000, 239-59) writes that a plan is not an intention, and therefore intentional action is not necessarily acting according to a plan. One might, as he says with reference to RPP I § 598, be engaged in planning without intending to do something (Hacker 2000, 251f.). But does it always make a difference whether one says that the cabinet plans something or that it intends to do something? To the extent to which it makes no difference, one might indeed learn something about intending by looking at it as a planning, like Michael Bratman does (Bratman 1999).

275

when smoking a cigar, one neither knows what one does nor does one not know it in the sense of “knowing what one does” just outlined. As far as one might, in this regard, make a conceptual mistake, it could at best consist in one’s intending to do something which may, in advance, be known to one as being too complex in order not to lead to a loss of survey, or to raise more opposition than one will be able to resist, or to go too much against the morality of people standing close to oneself so that one will not be able to keep the relationship with them intact; in short, the mistake consists in intending to do something for which a clear, realistic plan cannot even be developed. 20.4 The case in which an action vacillates, as it were, between different, not necessarily complex actions resembles the two previous cases in different respects. One wants to encourage someone, but one also wants to admonish him. No wonder that the respective intentional doing is anything but definite. (It has even been said that this is the normal case, whereas each clear and definite action is an exception.) It would be no misuse of language if, in such a case, another person said that one did not know what one was doing. For in many cases one expects a single, definite action as the answer to the question what one does, and not a mixtum compositum. One expects the specification of an action, not of a see-saw etc. But such an answer is simply not available in the present case. In this way there is also nothing to be known. This impossibility of knowing what one does is here of a general kind. It is not restricted to the agent. Thereby the background against which it is strange to say “I do not know this” ceases to exist. 20.5 Akin to the preceding case is the one in which someone tends to a systematic confusion of actions: he is multiplying, but says that he is dividing; he is adding, while announcing to subtract etc. Taken in itself, what he says is wrong. As far as we use the word “to know” in the way in which we use it when successfully trying to give an example of the so-called standard-analysis of the concept of knowledge, he does not know what he does. But can one say that, instead of knowing what he does, he only believes to know it?8 If he indeed systematically confuses the expressions “to 8. Anscombe (1963, 14) writes: “[T]here is point in speaking of knowledge only where a contrast exists between ‘he knows’ and ‘he (merely) believes he knows’.” For Anscombe, intentional actions are a subclass of the “class of things known without observation”, because “I did not know I was doing that” is a rejection of the question “Why did you do it?”, which is the key

276

multiply” and “to divide” etc., one might just as much say that he knows what he does as one might say that he does not know it, depending on the rule of language use one chooses: his rule or the usual one. Even if one wanted to say that he does not know what he does, but only believes to know it, one still would have to add that his expression of this belief would be different from ours. In many cases in which someone believes something, but does not know it since things are not the way he believes them to be, his belief finds just no other expression as his knowledge would find, that is, were his belief to be true. 20.6 A further kind of not knowing what one does consists in doing something while dreaming, being hypnotized or being in a similar state. This kind is more immediate than the one mentioned first insofar as it has nothing to do with the consequences of one’s doing, however close these might be connected with the doing itself. There is, therefore, also no borderline area such that one might say, and at the same time not say, that someone does not know what he does or does not know the consequences of his actions, respectively. This case closely resembles the one in which, acting under shock and then, as it were, awaking from it, one wonders whether one is conscious or not. (Analogously to asking oneself, after an accident, whether one still has two legs.) While being under shock, one need not act in such a way that others would be forced to say that one acted intentionally. (And if we act in such a way, all problems concerning the concept of knowing what one does are transferable to the concept of intentional action.) The person may, for instance, act not unintentionally in the sense of being able to correct mistakes, comment the course of her action, respond, while acting, to questions and advices concerning her action etc. Think of those movie scenes in which someone hears about the loss of a loved one. A standard form of portraying such a situation consists in the person’s continuing the to determining the concept of intention. Therefore, “I knew I was doing that, but only because I observed it” is, for Anscombe, also a rejection of the why-question. One might criticize Anscombe for not giving a positive characterization of “knowledge without observation”. And if one thinks that knowledge always has a source or consists in something, then this ciriticism is a serious one. (For instance, in the chapter on “Self-Identification”, Gareth Evans 1982 always speaks as if, whenever one can say that someone knows something, one must also be able to say how one knows.) But if the argument presented here is at least partially correct, then Anscombe’s failure speaks in favour of her. For one might understand it as evidence that she takes the concept “knowledge without observation” not to be a complex notion, standing in a hierarchical order of our concepts besides the concept “knowledge by observation”.

277

action just performed, while behaving like being remote-controlled, but nevertheless displaying the abilities just mentioned. (The character might spill the coffee and then wipe it up.) It would not be strange for someone being in such a situation to respond to the question what she is doing by saying: “Well, let me look …” The only strange thing here would be the investigator’s expectation to get a reliable answer from the addressee, in his state. Next in this row come mental diseases expressing themselves in the agent’s being permanently surprised by what he does. As the name already indicates, these are perhaps the clearest examples of pathological cases. As such, they are conceptually secondary to the cases of which they are pathological instances. (However, if they were the rule in a community, one would perhaps not speak of a disease.) Seen that way, an utterance of the form “I know what I am doing” may simply mean that the speaker is (again) sane. (Or one of us.) 20.7 A closer family member of these cases is that kind of not knowing what one does which consists in inattentiveness. Someone wants to salt the soup, but he takes the sugar. If he accompanied this with the words “Now I am salting the soup”, then it would be absolutely in order if someone else said: “But look what you are doing”. What he is supposed to find out by looking is what it is that he is actually doing. Finally, if one folds one’s fingers in a certain way and then tries to move a particular finger, it may happen that one does not know what one does in the sense that one simply cannot do it without looking or grasping. But here looking and grasping are not directed at a discovery of what one is doing, but are rather themselves part of one’s doing by, for instance, being directed at the way of doing it. True, both can coincide insofar as, when doing something in this way, say, drawing a certain line, one would no longer call it “drawing a line”. Still, “He is, with all his senses, looking at what he is doing” does not mean “All his senses are directed at discovering what he is doing”.9

9. Compare Peter Hacker (2000, 240) who writes: “A person acting intentionally is aware of what it is that he is intentionally doing — if one does not know that one is Ving, then one is not Ving intentionally. However, it would be wrong to characterize this as knowledge without observation. For one cannot always say whether one is doing what one intends to do without perceiving what one is doing. (Try writing or drawing with closed eyes.)”

278

21. Before summarizing the whole and, thereafter, getting back to the beginning, a last side glance is worthwhile. The contrast between the Prime Minister’s plan and his smoking a cigar throws some light on cases in which it is a “living option”10 that something might still go wrong with what one is doing such that, in the end, it might not be true that one did it. This contrast resembles the one between doing and trying to do. Given that one is doing this or that, it does not follow that one is trying to do it. It is no contradiction to say that someone did this or that, although he did not try to do it. As far as this sounds strange, this is not due to a violation of a conversational implicature. One says that one tries to do it if there are recognizable difficulties. Often there are no such difficulties. If there are none, one does not try to do it. One simply does it. (It may belong to the recognizable difficulties that one knows that, so far, one has never succeeded in doing what one started to do or that one always changed one’s mind and so on.11) Although one cannot infer trying to do from doing, one way of not doing something consists in not (even, as we sometimes tend to put it in such cases) trying to do it. — “Did he do it?” — “Oh, he didn’t even try to do it.” — Likewise, that he did it may consist in his having tried to do it and having been successful. Successfully trying to do something is, as it were, a kind of doing something. But it is still one kind among others. It is not the kind par excellence. In countless cases, one just does not try it when doing it. As said before, one just does it. Thus, that one does something by successfully trying to do it means that the circumstances are unusual compared to how matters usually stand with respect to what one tried. Furthermore, there are things which usually, or so far, can only be tried to be done, like travelling around the earth in a plane without making any refuelling stop. That there are things which can only be tried is, in turn, due to the fact that there are countless things which can be done without trying to do them and, furthermore, to the resemblance of those questionable things with the latter ones. With regard to those questionable things, there is (so far) simply no “way in which they are usually done”. “He went around the earth without refuelling” is a kind of news, an achievement, an exception, in contrast to “He prepared a cup of tea”. In such a remarkable case like going around the earth without refuelling, to know what one does means to know that what one does is something extraordinary 10. This expression is formed following William James’s term “living hypothesis” (in James 1897, 2f.). 11. Cf. Hampshire and Hart 1958, 175; Hanfling 2000, 94ff.; and PI §§ 622f.

279

or, if it is something ordinary, something done under unusual circumstances. Wherever there is a usual way of doing this or that, one simply does it, given that there are no extraordinary circumstances. Here, knowing what one does simply means being able to do it. Saying what one does is, in such a case, not an expression of one’s having acquired, in this or that way, for instance, in that “direct, unmediated way”, knowledge of one’s own actions. It is rather a kind of commentary on one’s doing. It is like accompanying one’s doing with certain remarks in order to keep someone else who lacks visibility informed. There is no place here for talk about a way of knowing what one does. As far as one wants to talk about such a way, one might say that one knows what one does because one has learned to do it. And this kind of knowledge acquisition does of course not consist in looking after what one does. It is as different from the way one acquires one’s knowledge of one’s doing as knowing that p is different from knowing how to do this or that.12 22. At the same time, this throws some light on the idea according to which knowing what one does belongs with necessity to one’s doing it intentionally because one might be held responsible for what one does intentionally — something which would be out of place if one did not know it. But to the extent to which the acquisition of knowledge of what one does may consist in learning how to do it, “I know what I am doing” may simply mean as much as “I can do it”. And in that case “I do not know that I am doing it”, if it means anything at all, means that I do not do it since I cannot do it. By saying “I do not know I am doing that”, I do not indirectly deny that I do it, by denying a precondition of my doing it — namely my knowing what I am doing —, but I deny it directly by denying to be able to do it at all. 12. As is generally known, this difference has been implanted in his time by Ryle (1949, 175). At the place mentioned in note 2 he writes: “When a person is described as not being surprised when something takes place, he can also be described as having expected it or having been prepared for it.” Here the word “or” is important. One kind of being prepared, without expecting it, consists in having learned it. Wittgenstein writes in Philosophical Investigations § 628: “So one might say: voluntary movement is marked by the absence of surprise. And now I do not want you to ask ‘But why isn’t one surprised here?’” Wittgenstein does not say why he does not want this question to be asked. Although voluntary movement is not the same as intentional action, our consideration suggests an answer: Precisely for this reason! More explicitly, because to ask would mean to search for an explanation, and would come along with not seeing the phenomenon as something which has to be accepted as given, but taking it as something derivative (cf. von Savigny 2004, 433ff.).

280

23. To sum up, at the beginning we considered two examples, a natural one, as we put it, and a strange one. At first glance, the oddity seemed to point towards the conceptual necessity of knowing what one does once one is doing something. This kind of knowledge is not acquired in this or that way by the one who has it. This fact comes up to meet the necessity of knowing what one does insofar as it excludes one possibility of one actually not knowing what one is doing: by something going wrong during the course of knowledge acquisition. For how can something be necessary which, at the same time, may not be the case? And that something might go wrong is a notorious feature of processes of acquiring something. But if the knowledge is necessary, and if something may go wrong in the course of its acquisition, there can be no acquisition at all. But a knowledge of which one cannot say where it stems from or how it came about, seems to be no knowledge at all. Thus, the problem takes the form of a search for an answer to the question of how one could know what one does if, at the same time, one cannot help but know it. We considered one possible answer, and what speaks against it. But the crucial point of this consideration was to make clear, or to unveil, a deeper problem: a certain lack of being radical. This lack abounds by the very search for a solution instead of being suspicious of the problem. One is searching for a solution because one does not question what allows formulating the problem in the first place. Once one is distrustful, one easily recognizes a variety of cases in which nothing is more natural than to say that one does (not) know what one does. To this variety also belong cases in which it is normal to look for what one is doing. Insofar as this is normal, one can say that one knows what one does in that sense which one usually has in mind when considering looking at something as one kind of acquiring knowledge of something or as one form in which such knowledge might exist. If there are such cases, then it cannot be an oddity sans phrase to say “Let me look what I am doing … ah, I am preparing a cup of tea”. What is strange is, at best, to say this in a certain kind of context. In general, our mistake was to suppose that we were dealing with an oddity sans phrase. — This insight removes the starting point of our whole consideration, and thereby the problem. There remains the fact that cases like those on our list do not happen often. But oddity in that sense does not come along with conceptual necessity as it is usually understood. It only appears to do so as long as one has only certain cases in mind, which may be due to the frequency of their

281

occurence. But this does not change the fact that the conclusions we drew from that oddity were overhasty.13

REFERENCES Anscombe, G. E. M. 1963: Intention, Cornell University Press: Ithaca, New York, 2nd ed. Austin, John 1956/57: “A Plea for Excuses”, in: Proceedings of the Aristotelian Society 57, 1956/57, 1–30. Bratman, Michael 1999: Faces of Intention. Selected Essays on Intention and Agency, Cambridge University Press: Cambridge. Davidson, Donald 1984: “First Person Authority”, in: Davidson 2001, 3–14. –– 1991: “Three Varieties of Knowledge”, in: Davidson 2001, 205–220. –– 2001: Subjective, Intersubjective, Objective, Oxford: Clarendon Press. Dennett, Daniel 1976: “Conditions of Personhood”, in: A. O. Rorty (ed.), The Identities of Persons, University of California Press: Berkeley, Los Angeles. Evans, Gareth 1982: Varieties of Reference, ed. by John McDowell, Clarendon Press: Oxford. Hacker, Peter M. S. 2000: Wittgenstein. Mind and Will, Volume 4 of an Analytical Commentary on the ‘Philosophical Investigations’, Part I: Essays, Blackwell: Oxford & Cambridge/Ma. Hampshire, Stuart, H. L. A. Hart 1958: “Decision, Intention and Certainty”, in: Mind 67, 1–12. Hanfling, Oswald 2000: Philosophy and Ordinary Language: The Bent and Genius of Our Tongue, Routledge: London and New York. –– 2003: “Learning about Right and Wrong: Ethics and Language”, in: Philosophy 78, 2003. 13. Davidson (1991, 205) finds it disturbing that, although all knowledge refers to one single world, there are three kinds of knowledge. Compare this with someone saying that it is disturbing that there are three kinds of making tea, although the product of these activities is in all three cases exactly alike. In this latter case, one feels tempted to respond with saying “Disturbing? What did you expect?” Cf. also the beginning of Wright 1998 as well as his remark about the “cardinal problem of self-knowledge” (Wright 1998, 22). It seems that Wittgenstein has also seen the problem: “One person might say ‘A proposition is the most ordinary thing in the world’ and another: ‘A proposition—that’s something very queer!’—And the latter is unable simply to look and see how propositions really work. The forms that we use in expressing ourselves about propositions and thought stand in his way” (PI § 93). Of course, here one wonders why the one saying that a proposition is the most ordinary thing in the world should at all look and see how propositions really work. It seems as if, in philosophy, we had to choose between problem and solution.

282

James, William 1897: “The Will to Believe“, in: The Will to Believe and Other Essays in Popular Philosophy, Longman, Green & Co. Perry, John 1979: “The Problem of the Essential Indexical”, in: Noûs XIII, 3– 21. Ryle, Gilbert 1949: The Concept of Mind. London: Hutchinson. von Savigny, Eike 2004: “Zusammenhänge sehen: Bedeutung und Gebrauch, Seele und Benehmen”, in: Deutsche Zeitschrift für Philosophie 52, 2004. Wolgast, Elisabeth 1993: “Innocence”, in: Philosophy 68, no. 265, July 1993. Wright, Crispin: “Self-Knowledge: The Wittgensteinian Legacy”, in: B. C. Smith, C. MacDonalds (eds.): Knowing Our Own Minds, Clarendon: Oxford 1998, 13–45.

283